1 The LTTng Documentation
2 =======================
3 Philippe Proulx <pproulx@efficios.com>
7 include::../common/copyright.txt[]
10 include::../common/warning-not-maintained.txt[]
13 include::../common/welcome.txt[]
16 include::../common/audience.txt[]
20 === What's in this documentation?
22 The LTTng Documentation is divided into the following sections:
24 * **<<nuts-and-bolts,Nuts and bolts>>** explains the
25 rudiments of software tracing and the rationale behind the
28 You can skip this section if you’re familiar with software tracing and
29 with the LTTng project.
31 * **<<installing-lttng,Installation>>** describes the steps to
32 install the LTTng packages on common Linux distributions and from
35 You can skip this section if you already properly installed LTTng on
38 * **<<getting-started,Quick start>>** is a concise guide to
39 getting started quickly with LTTng kernel and user space tracing.
41 We recommend this section if you're new to LTTng or to software tracing
44 You can skip this section if you're not new to LTTng.
46 * **<<core-concepts,Core concepts>>** explains the concepts at
49 It's a good idea to become familiar with the core concepts
50 before attempting to use the toolkit.
52 * **<<plumbing,Components of LTTng>>** describes the various components
53 of the LTTng machinery, like the daemons, the libraries, and the
54 command-line interface.
55 * **<<instrumenting,Instrumentation>>** shows different ways to
56 instrument user applications and the Linux kernel.
58 Instrumenting source code is essential to provide a meaningful
61 You can skip this section if you do not have a programming background.
63 * **<<controlling-tracing,Tracing control>>** is divided into topics
64 which demonstrate how to use the vast array of features that
65 LTTng{nbsp}{revision} offers.
66 * **<<reference,Reference>>** contains reference tables.
67 * **<<glossary,Glossary>>** is a specialized dictionary of terms related
68 to LTTng or to the field of software tracing.
71 include::../common/convention.txt[]
74 include::../common/acknowledgements.txt[]
78 == What's new in LTTng {revision}?
80 LTTng{nbsp}{revision} bears the name _Isseki Nicho_. The result of a
81 collaboration between http://www.dieuduciel.com/[Dieu du Ciel!] and
82 Nagano-based Shiga Kogen,
83 https://www.beeradvocate.com/beer/profile/1141/53111/[_**Isseki
84 Nicho**_] is a strong Imperial Dark Saison offering a rich roasted malt
85 flavor combined with a complex fruity finish typical of Saison yeasts.
87 New features and changes in LTTng{nbsp}{revision}:
89 * **Tracing control**:
90 ** You can attach <<java-application-context,Java application-specific
91 context fields>> to a <<channel,channel>> with the
92 man:lttng-add-context(1) command:
97 $ lttng add-context --jul --type='$app.retriever:cur_msg_id'
101 Here, `$app` is the prefix of all application-specific context fields,
102 `retriever` names a _context information retriever_ defined at the
103 application level, and `cur_msg_id` names a context field read from this
106 Both the `java.util.logging` and Apache log4j <<domain,tracing domains>>
109 ** You can use Java application-specific <<adding-context,context>>
110 fields in the <<enabling-disabling-events,filter expression>> of an
111 <<event,event rule>>:
116 $ lttng enable-event --log4j my_logger \
117 --filter='$app.retriever:cur_msg_id == 23'
121 ** New `lttng status` command which is the equivalent of +lttng list
122 __CUR__+, where +__CUR__+ is the name of the current
123 <<tracing-session,tracing session>>.
125 See man:lttng-status(1).
127 ** New `lttng metadata regenerate` command to
128 <<metadata-regenerate,regenerate the metadata file of an LTTng
129 trace>> at any moment. This command is meant to be used to resample
130 the wall time following a major
131 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
132 so that a system which boots with an incorrect wall time can be
133 traced before its wall time is NTP-corrected.
135 See man:lttng-metadata(1).
137 ** New command-line interface warnings when <<event,event records>> or
138 whole sub-buffers are
139 <<channel-overwrite-mode-vs-discard-mode,lost>>. The warning messages
140 are printed when a <<tracing-session,tracing session>> is
141 <<basic-tracing-session-control,stopped>> (man:lttng-stop(1)
144 * **User space tracing**:
145 ** Shared object base address dump in order to map <<event,event
146 records>> to original source location (file and line number).
148 If you attach the `ip` and `vpid` <<adding-context,context fields>> to a
149 user space <<channel,channel>> and if you use the
150 <<liblttng-ust-dl,path:{liblttng-ust-dl.so} helper>>, you can retrieve
151 the source location where a given event record was generated.
153 The http://diamon.org/babeltrace/[Babeltrace] trace viewer supports this
154 state dump and those context fields since version 1.4 to print the
155 source location of a given event record. http://tracecompass.org/[Trace
156 Compass] also supports this since version 2.0.
158 ** A <<java-application,Java application>> which uses
159 `java.util.logging` now adds an LTTng-UST log handler to the desired
162 The previous workflow was to initialize the LTTng-UST Java agent
163 by calling `LTTngAgent.getLTTngAgent()`. This had the effect of adding
164 an LTTng-UST log handler to the root loggers.
166 ** A <<java-application,Java application>> which uses Apache log4j now
167 adds an LTTng-UST log appender to the desired log4j loggers.
169 The previous workflow was to initialize the LTTng-UST Java agent
170 by calling `LTTngAgent.getLTTngAgent()`. This had the effect of adding
171 an LTTng-UST appender to the root loggers.
173 ** Any <<java-application,Java application>> can provide
174 <<java-application-context,dynamic context fields>> while running
175 thanks to a new API provided by the <<lttng-ust-agents,LTTng-UST Java
176 agent>>. You can require LTTng to record specific context fields in
177 event records, and you can use them in the filter expression of
178 <<event,event rules>>.
180 * **Linux kernel tracing**:
181 ** The LTTng kernel modules can now be built into a Linux kernel image,
182 that is, not as loadable modules.
185 https://github.com/lttng/lttng-modules/blob/stable-{revision}/README.md#kernel-built-in-support[`README.md`]
188 ** New instrumentation:
189 *** ARM64 architecture support.
191 *** x86 `irq_vectors`.
192 ** New <<adding-context,context fields>>:
195 *** `need_reschedule`
196 *** `migratable` (specific to RT-Preempt)
197 ** Clock source plugin support for advanced cases where a custom source
198 of time is needed to timestamp LTTng event records.
200 See https://github.com/lttng/lttng-modules/blob/stable-{revision}/lttng-clock.h[`lttng-clock.h`]
201 for an overview of the small API.
204 ** The link:/man[man pages] of the man:lttng(1) command-line tool are
205 split into one man page per command (à la Git), for example:
210 $ man lttng-enable-event
214 You can also use the `--help` option of any man:lttng(1) command to
217 The content and formatting of all the LTTng man pages has improved
224 What is LTTng? As its name suggests, the _Linux Trace Toolkit: next
225 generation_ is a modern toolkit for tracing Linux systems and
226 applications. So your first question might be:
233 As the history of software engineering progressed and led to what
234 we now take for granted--complex, numerous and
235 interdependent software applications running in parallel on
236 sophisticated operating systems like Linux--the authors of such
237 components, software developers, began feeling a natural
238 urge to have tools that would ensure the robustness and good performance
239 of their masterpieces.
241 One major achievement in this field is, inarguably, the
242 https://www.gnu.org/software/gdb/[GNU debugger (GDB)],
243 an essential tool for developers to find and fix bugs. But even the best
244 debugger won't help make your software run faster, and nowadays, faster
245 software means either more work done by the same hardware, or cheaper
246 hardware for the same work.
248 A _profiler_ is often the tool of choice to identify performance
249 bottlenecks. Profiling is suitable to identify _where_ performance is
250 lost in a given software. The profiler outputs a profile, a statistical
251 summary of observed events, which you may use to discover which
252 functions took the most time to execute. However, a profiler won't
253 report _why_ some identified functions are the bottleneck. Bottlenecks
254 might only occur when specific conditions are met, conditions that are
255 sometimes impossible to capture by a statistical profiler, or impossible
256 to reproduce with an application altered by the overhead of an
257 event-based profiler. For a thorough investigation of software
258 performance issues, a history of execution is essential, with the
259 recorded values of variables and context fields you choose, and
260 with as little influence as possible on the instrumented software. This
261 is where tracing comes in handy.
263 _Tracing_ is a technique used to understand what goes on in a running
264 software system. The software used for tracing is called a _tracer_,
265 which is conceptually similar to a tape recorder. When recording,
266 specific instrumentation points placed in the software source code
267 generate events that are saved on a giant tape: a _trace_ file. You
268 can trace user applications and the operating system at the same time,
269 opening the possibility of resolving a wide range of problems that would
270 otherwise be extremely challenging.
272 Tracing is often compared to _logging_. However, tracers and loggers are
273 two different tools, serving two different purposes. Tracers are
274 designed to record much lower-level events that occur much more
275 frequently than log messages, often in the range of thousands per
276 second, with very little execution overhead. Logging is more appropriate
277 for a very high-level analysis of less frequent events: user accesses,
278 exceptional conditions (errors and warnings, for example), database
279 transactions, instant messaging communications, and such. Simply put,
280 logging is one of the many use cases that can be satisfied with tracing.
282 The list of recorded events inside a trace file can be read manually
283 like a log file for the maximum level of detail, but it is generally
284 much more interesting to perform application-specific analyses to
285 produce reduced statistics and graphs that are useful to resolve a
286 given problem. Trace viewers and analyzers are specialized tools
289 In the end, this is what LTTng is: a powerful, open source set of
290 tools to trace the Linux kernel and user applications at the same time.
291 LTTng is composed of several components actively maintained and
292 developed by its link:/community/#where[community].
295 [[lttng-alternatives]]
296 === Alternatives to noch:{LTTng}
298 Excluding proprietary solutions, a few competing software tracers
301 * https://github.com/dtrace4linux/linux[dtrace4linux] is a port of
302 Sun Microsystems's DTrace to Linux. The cmd:dtrace tool interprets
303 user scripts and is responsible for loading code into the
304 Linux kernel for further execution and collecting the outputted data.
305 * https://en.wikipedia.org/wiki/Berkeley_Packet_Filter[eBPF] is a
306 subsystem in the Linux kernel in which a virtual machine can execute
307 programs passed from the user space to the kernel. You can attach
308 such programs to tracepoints and KProbes thanks to a system call, and
309 they can output data to the user space when executed thanks to
310 different mechanisms (pipe, VM register values, and eBPF maps, to name
312 * https://www.kernel.org/doc/Documentation/trace/ftrace.txt[ftrace]
313 is the de facto function tracer of the Linux kernel. Its user
314 interface is a set of special files in sysfs.
315 * https://perf.wiki.kernel.org/[perf] is
316 a performance analyzing tool for Linux which supports hardware
317 performance counters, tracepoints, as well as other counters and
318 types of probes. perf's controlling utility is the cmd:perf command
320 * http://linux.die.net/man/1/strace[strace]
321 is a command-line utility which records system calls made by a
322 user process, as well as signal deliveries and changes of process
323 state. strace makes use of https://en.wikipedia.org/wiki/Ptrace[ptrace]
324 to fulfill its function.
325 * http://www.sysdig.org/[sysdig], like SystemTap, uses scripts to
326 analyze Linux kernel events. You write scripts, or _chisels_ in
327 sysdig's jargon, in Lua and sysdig executes them while the system is
328 being traced or afterwards. sysdig's interface is the cmd:sysdig
329 command-line tool as well as the curses-based cmd:csysdig tool.
330 * https://sourceware.org/systemtap/[SystemTap] is a Linux kernel and
331 user space tracer which uses custom user scripts to produce plain text
332 traces. SystemTap converts the scripts to the C language, and then
333 compiles them as Linux kernel modules which are loaded to produce
334 trace data. SystemTap's primary user interface is the cmd:stap
337 The main distinctive features of LTTng is that it produces correlated
338 kernel and user space traces, as well as doing so with the lowest
339 overhead amongst other solutions. It produces trace files in the
340 http://diamon.org/ctf[CTF] format, a file format optimized
341 for the production and analyses of multi-gigabyte data.
343 LTTng is the result of more than 10 years of active open source
344 development by a community of passionate developers.
345 LTTng{nbsp}{revision} is currently available on major desktop and server
348 The main interface for tracing control is a single command-line tool
349 named cmd:lttng. The latter can create several tracing sessions, enable
350 and disable events on the fly, filter events efficiently with custom
351 user expressions, start and stop tracing, and much more. LTTng can
352 record the traces on the file system or send them over the network, and
353 keep them totally or partially. You can view the traces once tracing
354 becomes inactive or in real-time.
356 <<installing-lttng,Install LTTng now>> and
357 <<getting-started,start tracing>>!
363 include::../common/warning-installation-outdated.txt[]
365 **LTTng** is a set of software <<plumbing,components>> which interact to
366 <<instrumenting,instrument>> the Linux kernel and user applications, and
367 to <<controlling-tracing,control tracing>> (start and stop
368 tracing, enable and disable event rules, and the rest). Those
369 components are bundled into the following packages:
371 * **LTTng-tools**: Libraries and command-line interface to
373 * **LTTng-modules**: Linux kernel modules to instrument and
375 * **LTTng-UST**: Libraries and Java/Python packages to instrument and
376 trace user applications.
378 Most distributions mark the LTTng-modules and LTTng-UST packages as
379 optional when installing LTTng-tools (which is always required). In the
380 following sections, we always provide the steps to install all three,
383 * You only need to install LTTng-modules if you intend to trace the
385 * You only need to install LTTng-UST if you intend to trace user
389 .Availability of LTTng{nbsp}{revision} for major Linux distributions as of 14 March 2017.
391 |Distribution |Available in releases |Alternatives
393 |https://www.ubuntu.com/[Ubuntu]
394 |<<ubuntu,Ubuntu{nbsp}16.10 _Yakkety Yak_>>.
395 |LTTng{nbsp}{revision} for Ubuntu{nbsp}14.04 _Trusty Tahr_
396 and Ubuntu{nbsp}16.04 _Xenial Xerus_:
397 <<ubuntu-ppa,use the LTTng Stable{nbsp}{revision} PPA>>.
399 LTTng{nbsp}2.9 for Ubuntu{nbsp}14.04 _Trusty Tahr_
400 and Ubuntu{nbsp}16.04 _Xenial Xerus_:
401 link:/docs/v2.9#doc-ubuntu-ppa[use the LTTng Stable{nbsp}2.9 PPA].
403 LTTng{nbsp}2.9 for Ubuntu{nbsp}17.04 _Zesty Zapus_ (not released
406 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
407 other Ubuntu releases.
409 |https://getfedora.org/[Fedora]
410 |<<fedora,Fedora{nbsp}25>>.
411 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
412 other Fedora releases.
414 |https://www.debian.org/[Debian]
416 |link:/docs/v2.9#doc-debian[LTTng{nbsp}2.9 for Debian "stretch" (testing)
417 and Debian "sid" (unstable)].
419 <<building-from-source,Build LTTng{nbsp}{revision} from source>>.
421 |https://www.opensuse.org/[openSUSE]
423 |<<building-from-source,Build LTTng{nbsp}{revision} from source>>.
425 |https://www.archlinux.org/[Arch Linux]
427 |link:/docs/v2.9#doc-arch-linux[LTTng{nbsp}2.9 from the AUR].
429 <<building-from-source,Build LTTng{nbsp}{revision} from source>>.
431 |https://alpinelinux.org/[Alpine Linux]
433 |link:/docs/v2.9#doc-alpine-linux[LTTng{nbsp}2.9 for Alpine Linux "edge"].
435 <<building-from-source,Build LTTng{nbsp}{revision} from source>>.
437 |https://www.redhat.com/[RHEL] and https://www.suse.com/[SLES]
438 |See http://packages.efficios.com/[EfficiOS Enterprise Packages].
441 |https://buildroot.org/[Buildroot]
442 |<<buildroot,Buildroot 2016.11>>.
443 |link:/docs/v2.9#doc-buildroot[LTTng{nbsp}2.9 for Buildroot 2017.02].
445 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
446 other Buildroot releases.
448 |http://www.openembedded.org/wiki/Main_Page[OpenEmbedded] and
449 https://www.yoctoproject.org/[Yocto]
450 |<<oe-yocto,Yocto Project{nbsp}2.2 _Morty_>> (`openembedded-core` layer).
451 |LTTng{nbsp}2.9 for Yocto Project{nbsp}2.3 _Pyro_
452 (`openembedded-core` layer) (not released yet).
454 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
455 other OpenEmbedded releases.
460 === [[ubuntu-official-repositories]]Ubuntu
462 LTTng{nbsp}{revision} is available on Ubuntu{nbsp}16.10 _Yakkety Yak_.
463 For previous releases of Ubuntu, <<ubuntu-ppa,use the LTTng
464 Stable{nbsp}{revision} PPA>>.
466 To install LTTng{nbsp}{revision} on Ubuntu{nbsp}16.10 _Yakkety Yak_:
468 . Install the main LTTng{nbsp}{revision} packages:
473 # apt-get install lttng-tools
474 # apt-get install lttng-modules-dkms
475 # apt-get install liblttng-ust-dev
479 . **If you need to instrument and trace
480 <<java-application,Java applications>>**, install the LTTng-UST
486 # apt-get install liblttng-ust-agent-java
490 . **If you need to instrument and trace
491 <<python-application,Python{nbsp}3 applications>>**, install the
492 LTTng-UST Python agent:
497 # apt-get install python3-lttngust
503 ==== noch:{LTTng} Stable {revision} PPA
505 The https://launchpad.net/~lttng/+archive/ubuntu/stable-{revision}[LTTng
506 Stable{nbsp}{revision} PPA] offers the latest stable
507 LTTng{nbsp}{revision} packages for:
509 * Ubuntu{nbsp}14.04 _Trusty Tahr_
510 * Ubuntu{nbsp}16.04 _Xenial Xerus_
512 To install LTTng{nbsp}{revision} from the LTTng Stable{nbsp}{revision} PPA:
514 . Add the LTTng Stable{nbsp}{revision} PPA repository and update the
520 # apt-add-repository ppa:lttng/stable-2.8
525 . Install the main LTTng{nbsp}{revision} packages:
530 # apt-get install lttng-tools
531 # apt-get install lttng-modules-dkms
532 # apt-get install liblttng-ust-dev
536 . **If you need to instrument and trace
537 <<java-application,Java applications>>**, install the LTTng-UST
543 # apt-get install liblttng-ust-agent-java
547 . **If you need to instrument and trace
548 <<python-application,Python{nbsp}3 applications>>**, install the
549 LTTng-UST Python agent:
554 # apt-get install python3-lttngust
562 To install LTTng{nbsp}{revision} on Fedora{nbsp}25:
564 . Install the LTTng-tools{nbsp}{revision} and LTTng-UST{nbsp}{revision}
570 # yum install lttng-tools
571 # yum install lttng-ust
575 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
581 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.8.tar.bz2 &&
582 tar -xf lttng-modules-latest-2.8.tar.bz2 &&
583 cd lttng-modules-2.8.* &&
585 sudo make modules_install &&
591 .Java and Python application instrumentation and tracing
593 If you need to instrument and trace <<java-application,Java
594 applications>> on Fedora, you need to build and install
595 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
596 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
597 `--enable-java-agent-all` options to the `configure` script, depending
598 on which Java logging framework you use.
600 If you need to instrument and trace <<python-application,Python
601 applications>> on Fedora, you need to build and install
602 LTTng-UST{nbsp}{revision} from source and pass the
603 `--enable-python-agent` option to the `configure` script.
607 [[enterprise-distributions]]
608 === RHEL, SUSE, and other enterprise distributions
610 To install LTTng on enterprise Linux distributions, such as Red Hat
611 Enterprise Linux (RHEL) and SUSE Linux Enterprise Server (SUSE), please
612 see http://packages.efficios.com/[EfficiOS Enterprise Packages].
618 To install LTTng{nbsp}{revision} on Buildroot{nbsp}2016.11:
620 . Launch the Buildroot configuration tool:
629 . In **Kernel**, check **Linux kernel**.
630 . In **Toolchain**, check **Enable WCHAR support**.
631 . In **Target packages**{nbsp}→ **Debugging, profiling and benchmark**,
632 check **lttng-modules** and **lttng-tools**.
633 . In **Target packages**{nbsp}→ **Libraries**{nbsp}→
634 **Other**, check **lttng-libust**.
638 === OpenEmbedded and Yocto
640 LTTng{nbsp}{revision} recipes are available in the
641 http://layers.openembedded.org/layerindex/branch/master/layer/openembedded-core/[`openembedded-core`]
642 layer for Yocto Project{nbsp}2.2 _Morty_ under the following names:
648 With BitBake, the simplest way to include LTTng recipes in your target
649 image is to add them to `IMAGE_INSTALL_append` in path:{conf/local.conf}:
652 IMAGE_INSTALL_append = " lttng-tools lttng-modules lttng-ust"
657 . Select a machine and an image recipe.
658 . Click **Edit image recipe**.
659 . Under the **All recipes** tab, search for **lttng**.
660 . Check the desired LTTng recipes.
663 .Java and Python application instrumentation and tracing
665 If you need to instrument and trace <<java-application,Java
666 applications>> on Yocto/OpenEmbedded, you need to build and install
667 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
668 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
669 `--enable-java-agent-all` options to the `configure` script, depending
670 on which Java logging framework you use.
672 If you need to instrument and trace <<python-application,Python
673 applications>> on Yocto/OpenEmbedded, you need to build and install
674 LTTng-UST{nbsp}{revision} from source and pass the
675 `--enable-python-agent` option to the `configure` script.
679 [[building-from-source]]
680 === Build from source
682 To build and install LTTng{nbsp}{revision} from source:
684 . Using your distribution's package manager, or from source, install
685 the following dependencies of LTTng-tools and LTTng-UST:
688 * https://sourceforge.net/projects/libuuid/[libuuid]
689 * http://directory.fsf.org/wiki/Popt[popt]
690 * http://liburcu.org/[Userspace RCU]
691 * http://www.xmlsoft.org/[libxml2]
694 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
700 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.8.tar.bz2 &&
701 tar -xf lttng-modules-latest-2.8.tar.bz2 &&
702 cd lttng-modules-2.8.* &&
704 sudo make modules_install &&
709 . Download, build, and install the latest LTTng-UST{nbsp}{revision}:
715 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.8.tar.bz2 &&
716 tar -xf lttng-ust-latest-2.8.tar.bz2 &&
717 cd lttng-ust-2.8.* &&
727 .Java and Python application tracing
729 If you need to instrument and trace <<java-application,Java
730 applications>>, pass the `--enable-java-agent-jul`,
731 `--enable-java-agent-log4j`, or `--enable-java-agent-all` options to the
732 `configure` script, depending on which Java logging framework you use.
734 If you need to instrument and trace <<python-application,Python
735 applications>>, pass the `--enable-python-agent` option to the
736 `configure` script. You can set the `PYTHON` environment variable to the
737 path to the Python interpreter for which to install the LTTng-UST Python
745 By default, LTTng-UST libraries are installed to
746 dir:{/usr/local/lib}, which is the de facto directory in which to
747 keep self-compiled and third-party libraries.
749 When <<building-tracepoint-providers-and-user-application,linking an
750 instrumented user application with `liblttng-ust`>>:
752 * Append `/usr/local/lib` to the env:LD_LIBRARY_PATH environment
754 * Pass the `-L/usr/local/lib` and `-Wl,-rpath,/usr/local/lib` options to
755 man:gcc(1), man:g++(1), or man:clang(1).
759 . Download, build, and install the latest LTTng-tools{nbsp}{revision}:
765 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.8.tar.bz2 &&
766 tar -xf lttng-tools-latest-2.8.tar.bz2 &&
767 cd lttng-tools-2.8.* &&
775 TIP: The https://github.com/eepp/vlttng[vlttng tool] can do all the
776 previous steps automatically for a given version of LTTng and confine
777 the installed files in a specific directory. This can be useful to test
778 LTTng without installing it on your system.
784 This is a short guide to get started quickly with LTTng kernel and user
787 Before you follow this guide, make sure to <<installing-lttng,install>>
790 This tutorial walks you through the steps to:
792 . <<tracing-the-linux-kernel,Trace the Linux kernel>>.
793 . <<tracing-your-own-user-application,Trace a user application>> written
795 . <<viewing-and-analyzing-your-traces,View and analyze the
799 [[tracing-the-linux-kernel]]
800 === Trace the Linux kernel
802 The following command lines start with the `#` prompt because you need
803 root privileges to trace the Linux kernel. You can also trace the kernel
804 as a regular user if your Unix user is a member of the
805 <<tracing-group,tracing group>>.
807 . Create a <<tracing-session,tracing session>> which writes its traces
808 to dir:{/tmp/my-kernel-trace}:
813 # lttng create my-kernel-session --output=/tmp/my-kernel-trace
817 . List the available kernel tracepoints and system calls:
822 # lttng list --kernel
823 # lttng list --kernel --syscall
827 . Create <<event,event rules>> which match the desired instrumentation
828 point names, for example the `sched_switch` and `sched_process_fork`
829 tracepoints, and the man:open(2) and man:close(2) system calls:
834 # lttng enable-event --kernel sched_switch,sched_process_fork
835 # lttng enable-event --kernel --syscall open,close
839 You can also create an event rule which matches _all_ the Linux kernel
840 tracepoints (this will generate a lot of data when tracing):
845 # lttng enable-event --kernel --all
849 . <<basic-tracing-session-control,Start tracing>>:
858 . Do some operation on your system for a few seconds. For example,
859 load a website, or list the files of a directory.
860 . <<basic-tracing-session-control,Stop tracing>> and destroy the
871 The man:lttng-destroy(1) command does not destroy the trace data; it
872 only destroys the state of the tracing session.
874 . For the sake of this example, make the recorded trace accessible to
880 sudo chown -R $(whoami) /tmp/my-kernel-trace
884 See <<viewing-and-analyzing-your-traces,View and analyze the
885 recorded events>> to view the recorded events.
888 [[tracing-your-own-user-application]]
889 === Trace a user application
891 This section steps you through a simple example to trace a
892 _Hello world_ program written in C.
894 To create the traceable user application:
896 . Create the tracepoint provider header file, which defines the
897 tracepoints and the events they can generate:
903 #undef TRACEPOINT_PROVIDER
904 #define TRACEPOINT_PROVIDER hello_world
906 #undef TRACEPOINT_INCLUDE
907 #define TRACEPOINT_INCLUDE "./hello-tp.h"
909 #if !defined(_HELLO_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
912 #include <lttng/tracepoint.h>
922 ctf_string(my_string_field, my_string_arg)
923 ctf_integer(int, my_integer_field, my_integer_arg)
927 #endif /* _HELLO_TP_H */
929 #include <lttng/tracepoint-event.h>
933 . Create the tracepoint provider package source file:
939 #define TRACEPOINT_CREATE_PROBES
940 #define TRACEPOINT_DEFINE
942 #include "hello-tp.h"
946 . Build the tracepoint provider package:
951 $ gcc -c -I. hello-tp.c
955 . Create the _Hello World_ application source file:
962 #include "hello-tp.h"
964 int main(int argc, char *argv[])
968 puts("Hello, World!\nPress Enter to continue...");
971 * The following getchar() call is only placed here for the purpose
972 * of this demonstration, to pause the application in order for
973 * you to have time to list its tracepoints. It is not
979 * A tracepoint() call.
981 * Arguments, as defined in hello-tp.h:
983 * 1. Tracepoint provider name (required)
984 * 2. Tracepoint name (required)
985 * 3. my_integer_arg (first user-defined argument)
986 * 4. my_string_arg (second user-defined argument)
988 * Notice the tracepoint provider and tracepoint names are
989 * NOT strings: they are in fact parts of variables that the
990 * macros in hello-tp.h create.
992 tracepoint(hello_world, my_first_tracepoint, 23, "hi there!");
994 for (x = 0; x < argc; ++x) {
995 tracepoint(hello_world, my_first_tracepoint, x, argv[x]);
998 puts("Quitting now!");
999 tracepoint(hello_world, my_first_tracepoint, x * x, "x^2");
1006 . Build the application:
1015 . Link the application with the tracepoint provider package,
1016 `liblttng-ust`, and `libdl`:
1021 $ gcc -o hello hello.o hello-tp.o -llttng-ust -ldl
1025 Here's the whole build process:
1028 .User space tracing tutorial's build steps.
1029 image::ust-flow.png[]
1031 To trace the user application:
1033 . Run the application with a few arguments:
1038 $ ./hello world and beyond
1047 Press Enter to continue...
1051 . Start an LTTng <<lttng-sessiond,session daemon>>:
1056 $ lttng-sessiond --daemonize
1060 Note that a session daemon might already be running, for example as
1061 a service that the distribution's service manager started.
1063 . List the available user space tracepoints:
1068 $ lttng list --userspace
1072 You see the `hello_world:my_first_tracepoint` tracepoint listed
1073 under the `./hello` process.
1075 . Create a <<tracing-session,tracing session>>:
1080 $ lttng create my-user-space-session
1084 . Create an <<event,event rule>> which matches the
1085 `hello_world:my_first_tracepoint` event name:
1090 $ lttng enable-event --userspace hello_world:my_first_tracepoint
1094 . <<basic-tracing-session-control,Start tracing>>:
1103 . Go back to the running `hello` application and press Enter. The
1104 program executes all `tracepoint()` instrumentation points and exits.
1105 . <<basic-tracing-session-control,Stop tracing>> and destroy the
1116 The man:lttng-destroy(1) command does not destroy the trace data; it
1117 only destroys the state of the tracing session.
1119 By default, LTTng saves the traces in
1120 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
1121 where +__name__+ is the tracing session name. The
1122 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
1124 See <<viewing-and-analyzing-your-traces,View and analyze the
1125 recorded events>> to view the recorded events.
1128 [[viewing-and-analyzing-your-traces]]
1129 === View and analyze the recorded events
1131 Once you have completed the <<tracing-the-linux-kernel,Trace the Linux
1132 kernel>> and <<tracing-your-own-user-application,Trace a user
1133 application>> tutorials, you can inspect the recorded events.
1135 Many tools are available to read LTTng traces:
1137 * **cmd:babeltrace** is a command-line utility which converts trace
1138 formats; it supports the format that LTTng produces, CTF, as well as a
1139 basic text output which can be ++grep++ed. The cmd:babeltrace command
1140 is part of the http://diamon.org/babeltrace[Babeltrace] project.
1141 * Babeltrace also includes
1142 **https://www.python.org/[Python] bindings** so
1143 that you can easily open and read an LTTng trace with your own script,
1144 benefiting from the power of Python.
1145 * http://tracecompass.org/[**Trace Compass**]
1146 is a graphical user interface for viewing and analyzing any type of
1147 logs or traces, including LTTng's.
1148 * https://github.com/lttng/lttng-analyses[**LTTng analyses**] is a
1149 project which includes many high-level analyses of LTTng kernel
1150 traces, like scheduling statistics, interrupt frequency distribution,
1151 top CPU usage, and more.
1153 NOTE: This section assumes that the traces recorded during the previous
1154 tutorials were saved to their default location, in the
1155 dir:{$LTTNG_HOME/lttng-traces} directory. The env:LTTNG_HOME
1156 environment variable defaults to `$HOME` if not set.
1159 [[viewing-and-analyzing-your-traces-bt]]
1160 ==== Use the cmd:babeltrace command-line tool
1162 The simplest way to list all the recorded events of a trace is to pass
1163 its path to cmd:babeltrace with no options:
1167 $ babeltrace ~/lttng-traces/my-user-space-session*
1170 cmd:babeltrace finds all traces recursively within the given path and
1171 prints all their events, merging them in chronological order.
1173 You can pipe the output of cmd:babeltrace into a tool like man:grep(1) for
1178 $ babeltrace /tmp/my-kernel-trace | grep _switch
1181 You can pipe the output of cmd:babeltrace into a tool like man:wc(1) to
1182 count the recorded events:
1186 $ babeltrace /tmp/my-kernel-trace | grep _open | wc --lines
1190 [[viewing-and-analyzing-your-traces-bt-python]]
1191 ==== Use the Babeltrace Python bindings
1193 The <<viewing-and-analyzing-your-traces-bt,text output of cmd:babeltrace>>
1194 is useful to isolate events by simple matching using man:grep(1) and
1195 similar utilities. However, more elaborate filters, such as keeping only
1196 event records with a field value falling within a specific range, are
1197 not trivial to write using a shell. Moreover, reductions and even the
1198 most basic computations involving multiple event records are virtually
1199 impossible to implement.
1201 Fortunately, Babeltrace ships with Python 3 bindings which makes it easy
1202 to read the event records of an LTTng trace sequentially and compute the
1203 desired information.
1205 The following script accepts an LTTng Linux kernel trace path as its
1206 first argument and prints the short names of the top 5 running processes
1207 on CPU 0 during the whole trace:
1212 from collections import Counter
1218 if len(sys.argv) != 2:
1219 msg = 'Usage: python3 {} TRACEPATH'.format(sys.argv[0])
1220 print(msg, file=sys.stderr)
1223 # A trace collection contains one or more traces
1224 col = babeltrace.TraceCollection()
1226 # Add the trace provided by the user (LTTng traces always have
1228 if col.add_trace(sys.argv[1], 'ctf') is None:
1229 raise RuntimeError('Cannot add trace')
1231 # This counter dict contains execution times:
1233 # task command name -> total execution time (ns)
1234 exec_times = Counter()
1236 # This contains the last `sched_switch` timestamp
1240 for event in col.events:
1241 # Keep only `sched_switch` events
1242 if event.name != 'sched_switch':
1245 # Keep only events which happened on CPU 0
1246 if event['cpu_id'] != 0:
1250 cur_ts = event.timestamp
1256 # Previous task command (short) name
1257 prev_comm = event['prev_comm']
1259 # Initialize entry in our dict if not yet done
1260 if prev_comm not in exec_times:
1261 exec_times[prev_comm] = 0
1263 # Compute previous command execution time
1264 diff = cur_ts - last_ts
1266 # Update execution time of this command
1267 exec_times[prev_comm] += diff
1269 # Update last timestamp
1273 for name, ns in exec_times.most_common(5):
1275 print('{:20}{} s'.format(name, s))
1280 if __name__ == '__main__':
1281 sys.exit(0 if top5proc() else 1)
1288 $ python3 top5proc.py /tmp/my-kernel-trace/kernel
1294 swapper/0 48.607245889 s
1295 chromium 7.192738188 s
1296 pavucontrol 0.709894415 s
1297 Compositor 0.660867933 s
1298 Xorg.bin 0.616753786 s
1301 Note that `swapper/0` is the "idle" process of CPU 0 on Linux; since we
1302 weren't using the CPU that much when tracing, its first position in the
1307 == [[understanding-lttng]]Core concepts
1309 From a user's perspective, the LTTng system is built on a few concepts,
1310 or objects, on which the <<lttng-cli,cmd:lttng command-line tool>>
1311 operates by sending commands to the <<lttng-sessiond,session daemon>>.
1312 Understanding how those objects relate to eachother is key in mastering
1315 The core concepts are:
1317 * <<tracing-session,Tracing session>>
1318 * <<domain,Tracing domain>>
1319 * <<channel,Channel and ring buffer>>
1320 * <<"event","Instrumentation point, event rule, event, and event record">>
1326 A _tracing session_ is a stateful dialogue between you and
1327 a <<lttng-sessiond,session daemon>>. You can
1328 <<creating-destroying-tracing-sessions,create a new tracing
1329 session>> with the `lttng create` command.
1331 Anything that you do when you control LTTng tracers happens within a
1332 tracing session. In particular, a tracing session:
1335 * Has its own set of trace files.
1336 * Has its own state of activity (started or stopped).
1337 * Has its own <<tracing-session-mode,mode>> (local, network streaming,
1339 * Has its own <<channel,channels>> which have their own
1340 <<event,event rules>>.
1343 .A _tracing session_ contains <<channel,channels>> that are members of <<domain,tracing domains>> and contain <<event,event rules>>.
1344 image::concepts.png[]
1346 Those attributes and objects are completely isolated between different
1349 A tracing session is analogous to a cash machine session:
1350 the operations you do on the banking system through the cash machine do
1351 not alter the data of other users of the same system. In the case of
1352 the cash machine, a session lasts as long as your bank card is inside.
1353 In the case of LTTng, a tracing session lasts from the `lttng create`
1354 command to the `lttng destroy` command.
1357 .Each Unix user has its own set of tracing sessions.
1358 image::many-sessions.png[]
1361 [[tracing-session-mode]]
1362 ==== Tracing session mode
1364 LTTng can send the generated trace data to different locations. The
1365 _tracing session mode_ dictates where to send it. The following modes
1366 are available in LTTng{nbsp}{revision}:
1369 LTTng writes the traces to the file system of the machine being traced
1372 Network streaming mode::
1373 LTTng sends the traces over the network to a
1374 <<lttng-relayd,relay daemon>> running on a remote system.
1377 LTTng does not write the traces by default. Instead, you can request
1378 LTTng to <<taking-a-snapshot,take a snapshot>>, that is, a copy of the
1379 current tracing buffers, and to write it to the target's file system
1380 or to send it over the network to a <<lttng-relayd,relay daemon>>
1381 running on a remote system.
1384 This mode is similar to the network streaming mode, but a live
1385 trace viewer can connect to the distant relay daemon to
1386 <<lttng-live,view event records as LTTng generates them>> by
1393 A _tracing domain_ is a namespace for event sources. A tracing domain
1394 has its own properties and features.
1396 There are currently five available tracing domains:
1400 * `java.util.logging` (JUL)
1404 You must specify a tracing domain when using some commands to avoid
1405 ambiguity. For example, since all the domains support named tracepoints
1406 as event sources (instrumentation points that you manually insert in the
1407 source code), you need to specify a tracing domain when
1408 <<enabling-disabling-events,creating an event rule>> because all the
1409 tracing domains could have tracepoints with the same names.
1411 Some features are reserved to specific tracing domains. Dynamic function
1412 entry and return instrumentation points, for example, are currently only
1413 supported in the Linux kernel tracing domain, but support for other
1414 tracing domains could be added in the future.
1416 You can create <<channel,channels>> in the Linux kernel and user space
1417 tracing domains. The other tracing domains have a single default
1422 === Channel and ring buffer
1424 A _channel_ is an object which is responsible for a set of ring buffers.
1425 Each ring buffer is divided into multiple sub-buffers. When an LTTng
1426 tracer emits an event, it can record it to one or more
1427 sub-buffers. The attributes of a channel determine what to do when
1428 there's no space left for a new event record because all sub-buffers
1429 are full, where to send a full sub-buffer, and other behaviours.
1431 A channel is always associated to a <<domain,tracing domain>>. The
1432 `java.util.logging` (JUL), log4j, and Python tracing domains each have
1433 a default channel which you cannot configure.
1435 A channel also owns <<event,event rules>>. When an LTTng tracer emits
1436 an event, it records it to the sub-buffers of all
1437 the enabled channels with a satisfied event rule, as long as those
1438 channels are part of active <<tracing-session,tracing sessions>>.
1441 [[channel-buffering-schemes]]
1442 ==== Per-user vs. per-process buffering schemes
1444 A channel has at least one ring buffer _per CPU_. LTTng always
1445 records an event to the ring buffer associated to the CPU on which it
1448 Two _buffering schemes_ are available when you
1449 <<enabling-disabling-channels,create a channel>> in the
1450 user space <<domain,tracing domain>>:
1452 Per-user buffering::
1453 Allocate one set of ring buffers--one per CPU--shared by all the
1454 instrumented processes of each Unix user.
1458 .Per-user buffering scheme.
1459 image::per-user-buffering.png[]
1462 Per-process buffering::
1463 Allocate one set of ring buffers--one per CPU--for each
1464 instrumented process.
1468 .Per-process buffering scheme.
1469 image::per-process-buffering.png[]
1472 The per-process buffering scheme tends to consume more memory than the
1473 per-user option because systems generally have more instrumented
1474 processes than Unix users running instrumented processes. However, the
1475 per-process buffering scheme ensures that one process having a high
1476 event throughput won't fill all the shared sub-buffers of the same
1479 The Linux kernel tracing domain has only one available buffering scheme
1480 which is to allocate a single set of ring buffers for the whole system.
1481 This scheme is similar to the per-user option, but with a single, global
1482 user "running" the kernel.
1485 [[channel-overwrite-mode-vs-discard-mode]]
1486 ==== Overwrite vs. discard event loss modes
1488 When an event occurs, LTTng records it to a specific sub-buffer (yellow
1489 arc in the following animation) of a specific channel's ring buffer.
1490 When there's no space left in a sub-buffer, the tracer marks it as
1491 consumable (red) and another, empty sub-buffer starts receiving the
1492 following event records. A <<lttng-consumerd,consumer daemon>>
1493 eventually consumes the marked sub-buffer (returns to white).
1496 [role="docsvg-channel-subbuf-anim"]
1501 In an ideal world, sub-buffers are consumed faster than they are filled,
1502 as is the case in the previous animation. In the real world,
1503 however, all sub-buffers can be full at some point, leaving no space to
1504 record the following events.
1506 By design, LTTng is a _non-blocking_ tracer: when no empty sub-buffer is
1507 available, it is acceptable to lose event records when the alternative
1508 would be to cause substantial delays in the instrumented application's
1509 execution. LTTng privileges performance over integrity; it aims at
1510 perturbing the traced system as little as possible in order to make
1511 tracing of subtle race conditions and rare interrupt cascades possible.
1513 When it comes to losing event records because no empty sub-buffer is
1514 available, the channel's _event loss mode_ determines what to do. The
1515 available event loss modes are:
1518 Drop the newest event records until a the tracer
1519 releases a sub-buffer.
1522 Clear the sub-buffer containing the oldest event records and start
1523 writing the newest event records there.
1525 This mode is sometimes called _flight recorder mode_ because it's
1527 https://en.wikipedia.org/wiki/Flight_recorder[flight recorder]:
1528 always keep a fixed amount of the latest data.
1530 Which mechanism you should choose depends on your context: prioritize
1531 the newest or the oldest event records in the ring buffer?
1533 Beware that, in overwrite mode, the tracer abandons a whole sub-buffer
1534 as soon as a there's no space left for a new event record, whereas in
1535 discard mode, the tracer only discards the event record that doesn't
1538 In discard mode, LTTng increments a count of lost event records when
1539 an event record is lost and saves this count to the trace. In
1540 overwrite mode, LTTng keeps no information when it overwrites a
1541 sub-buffer before consuming it.
1543 There are a few ways to decrease your probability of losing event
1545 <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>> shows
1546 how you can fine-une the sub-buffer count and size of a channel to
1547 virtually stop losing event records, though at the cost of greater
1551 [[channel-subbuf-size-vs-subbuf-count]]
1552 ==== Sub-buffer count and size
1554 When you <<enabling-disabling-channels,create a channel>>, you can
1555 set its number of sub-buffers and their size.
1557 Note that there is noticeable CPU overhead introduced when
1558 switching sub-buffers (marking a full one as consumable and switching
1559 to an empty one for the following events to be recorded). Knowing this,
1560 the following list presents a few practical situations along with how
1561 to configure the sub-buffer count and size for them:
1563 * **High event throughput**: In general, prefer bigger sub-buffers to
1564 lower the risk of losing event records.
1566 Having bigger sub-buffers also ensures a lower
1567 <<channel-switch-timer,sub-buffer switching frequency>>.
1569 The number of sub-buffers is only meaningful if you create the channel
1570 in overwrite mode: in this case, if a sub-buffer overwrite happens, the
1571 other sub-buffers are left unaltered.
1573 * **Low event throughput**: In general, prefer smaller sub-buffers
1574 since the risk of losing event records is low.
1576 Because events occur less frequently, the sub-buffer switching frequency
1577 should remain low and thus the tracer's overhead should not be a
1580 * **Low memory system**: If your target system has a low memory
1581 limit, prefer fewer first, then smaller sub-buffers.
1583 Even if the system is limited in memory, you want to keep the
1584 sub-buffers as big as possible to avoid a high sub-buffer switching
1587 Note that LTTng uses http://diamon.org/ctf/[CTF] as its trace format,
1588 which means event data is very compact. For example, the average
1589 LTTng kernel event record weights about 32{nbsp}bytes. Thus, a
1590 sub-buffer size of 1{nbsp}MiB is considered big.
1592 The previous situations highlight the major trade-off between a few big
1593 sub-buffers and more, smaller sub-buffers: sub-buffer switching
1594 frequency vs. how much data is lost in overwrite mode. Assuming a
1595 constant event throughput and using the overwrite mode, the two
1596 following configurations have the same ring buffer total size:
1599 [role="docsvg-channel-subbuf-size-vs-count-anim"]
1604 * **2 sub-buffers of 4{nbsp}MiB each**: Expect a very low sub-buffer
1605 switching frequency, but if a sub-buffer overwrite happens, half of
1606 the event records so far (4{nbsp}MiB) are definitely lost.
1607 * **8 sub-buffers of 1{nbsp}MiB each**: Expect 4{nbsp}times the tracer's
1608 overhead as the previous configuration, but if a sub-buffer
1609 overwrite happens, only the eighth of event records so far are
1612 In discard mode, the sub-buffers count parameter is pointless: use two
1613 sub-buffers and set their size according to the requirements of your
1617 [[channel-switch-timer]]
1618 ==== Switch timer period
1620 The _switch timer period_ is an important configurable attribute of
1621 a channel to ensure periodic sub-buffer flushing.
1623 When the _switch timer_ expires, a sub-buffer switch happens. You can
1624 set the switch timer period attribute when you
1625 <<enabling-disabling-channels,create a channel>> to ensure that event
1626 data is consumed and committed to trace files or to a distant relay
1627 daemon periodically in case of a low event throughput.
1630 [role="docsvg-channel-switch-timer"]
1635 This attribute is also convenient when you use big sub-buffers to cope
1636 with a sporadic high event throughput, even if the throughput is
1640 [[channel-read-timer]]
1641 ==== Read timer period
1643 By default, the LTTng tracers use a notification mechanism to signal a
1644 full sub-buffer so that a consumer daemon can consume it. When such
1645 notifications must be avoided, for example in real-time applications,
1646 you can use the channel's _read timer_ instead. When the read timer
1647 fires, the <<lttng-consumerd,consumer daemon>> checks for full,
1648 consumable sub-buffers.
1651 [[tracefile-rotation]]
1652 ==== Trace file count and size
1654 By default, trace files can grow as large as needed. You can set the
1655 maximum size of each trace file that a channel writes when you
1656 <<enabling-disabling-channels,create a channel>>. When the size of
1657 a trace file reaches the channel's fixed maximum size, LTTng creates
1658 another file to contain the next event records. LTTng appends a file
1659 count to each trace file name in this case.
1661 If you set the trace file size attribute when you create a channel, the
1662 maximum number of trace files that LTTng creates is _unlimited_ by
1663 default. To limit them, you can also set a maximum number of trace
1664 files. When the number of trace files reaches the channel's fixed
1665 maximum count, the oldest trace file is overwritten. This mechanism is
1666 called _trace file rotation_.
1670 === Instrumentation point, event rule, event, and event record
1672 An _event rule_ is a set of conditions which must be **all** satisfied
1673 for LTTng to record an occuring event.
1675 You set the conditions when you <<enabling-disabling-events,create
1678 You always attach an event rule to <<channel,channel>> when you create
1681 When an event passes the conditions of an event rule, LTTng records it
1682 in one of the attached channel's sub-buffers.
1684 The available conditions, as of LTTng{nbsp}{revision}, are:
1686 * The event rule _is enabled_.
1687 * The instrumentation point's type _is{nbsp}T_.
1688 * The instrumentation point's name (sometimes called _event name_)
1689 _matches{nbsp}N_, but _is not{nbsp}E_.
1690 * The instrumentation point's log level _is as severe as{nbsp}L_, or
1691 _is exactly{nbsp}L_.
1692 * The fields of the event's payload _satisfy_ a filter
1693 expression{nbsp}__F__.
1695 As you can see, all the conditions but the dynamic filter are related to
1696 the event rule's status or to the instrumentation point, not to the
1697 occurring events. This is why, without a filter, checking if an event
1698 passes an event rule is not a dynamic task: when you create or modify an
1699 event rule, all the tracers of its tracing domain enable or disable the
1700 instrumentation points themselves once. This is possible because the
1701 attributes of an instrumentation point (type, name, and log level) are
1702 defined statically. In other words, without a dynamic filter, the tracer
1703 _does not evaluate_ the arguments of an instrumentation point unless it
1704 matches an enabled event rule.
1706 Note that, for LTTng to record an event, the <<channel,channel>> to
1707 which a matching event rule is attached must also be enabled, and the
1708 tracing session owning this channel must be active.
1711 .Logical path from an instrumentation point to an event record.
1712 image::event-rule.png[]
1714 .Event, event record, or event rule?
1716 With so many similar terms, it's easy to get confused.
1718 An **event** is the consequence of the execution of an _instrumentation
1719 point_, like a tracepoint that you manually place in some source code,
1720 or a Linux kernel KProbe. An event is said to _occur_ at a specific
1721 time. Different actions can be taken upon the occurrence of an event,
1722 like record the event's payload to a buffer.
1724 An **event record** is the representation of an event in a sub-buffer. A
1725 tracer is responsible for capturing the payload of an event, current
1726 context variables, the event's ID, and the event's timestamp. LTTng
1727 can append this sub-buffer to a trace file.
1729 An **event rule** is a set of conditions which must all be satisfied for
1730 LTTng to record an occuring event. Events still occur without
1731 satisfying event rules, but LTTng does not record them.
1736 == Components of noch:{LTTng}
1738 The second _T_ in _LTTng_ stands for _toolkit_: it would be wrong
1739 to call LTTng a simple _tool_ since it is composed of multiple
1740 interacting components. This section describes those components,
1741 explains their respective roles, and shows how they connect together to
1742 form the LTTng ecosystem.
1744 The following diagram shows how the most important components of LTTng
1745 interact with user applications, the Linux kernel, and you:
1748 .Control and trace data paths between LTTng components.
1749 image::plumbing.png[]
1751 The LTTng project incorporates:
1753 * **LTTng-tools**: Libraries and command-line interface to
1754 control tracing sessions.
1755 ** <<lttng-sessiond,Session daemon>> (man:lttng-sessiond(8)).
1756 ** <<lttng-consumerd,Consumer daemon>> (man:lttng-consumerd(8)).
1757 ** <<lttng-relayd,Relay daemon>> (man:lttng-relayd(8)).
1758 ** <<liblttng-ctl-lttng,Tracing control library>> (`liblttng-ctl`).
1759 ** <<lttng-cli,Tracing control command-line tool>> (man:lttng(1)).
1760 * **LTTng-UST**: Libraries and Java/Python packages to trace user
1762 ** <<lttng-ust,User space tracing library>> (`liblttng-ust`) and its
1763 headers to instrument and trace any native user application.
1764 ** <<prebuilt-ust-helpers,Preloadable user space tracing helpers>>:
1765 *** `liblttng-ust-libc-wrapper`
1766 *** `liblttng-ust-pthread-wrapper`
1767 *** `liblttng-ust-cyg-profile`
1768 *** `liblttng-ust-cyg-profile-fast`
1769 *** `liblttng-ust-dl`
1770 ** User space tracepoint provider source files generator command-line
1771 tool (man:lttng-gen-tp(1)).
1772 ** <<lttng-ust-agents,LTTng-UST Java agent>> to instrument and trace
1773 Java applications using `java.util.logging` or
1774 Apache log4j 1.2 logging.
1775 ** <<lttng-ust-agents,LTTng-UST Python agent>> to instrument
1776 Python applications using the standard `logging` package.
1777 * **LTTng-modules**: <<lttng-modules,Linux kernel modules>> to trace
1779 ** LTTng kernel tracer module.
1780 ** Tracing ring buffer kernel modules.
1781 ** Probe kernel modules.
1782 ** LTTng logger kernel module.
1786 === Tracing control command-line interface
1789 .The tracing control command-line interface.
1790 image::plumbing-lttng-cli.png[]
1792 The _man:lttng(1) command-line tool_ is the standard user interface to
1793 control LTTng <<tracing-session,tracing sessions>>. The cmd:lttng tool
1794 is part of LTTng-tools.
1796 The cmd:lttng tool is linked with
1797 <<liblttng-ctl-lttng,`liblttng-ctl`>> to communicate with
1798 one or more <<lttng-sessiond,session daemons>> behind the scenes.
1800 The cmd:lttng tool has a Git-like interface:
1804 $ lttng <GENERAL OPTIONS> <COMMAND> <COMMAND OPTIONS>
1807 The <<controlling-tracing,Tracing control>> section explores the
1808 available features of LTTng using the cmd:lttng tool.
1811 [[liblttng-ctl-lttng]]
1812 === Tracing control library
1815 .The tracing control library.
1816 image::plumbing-liblttng-ctl.png[]
1818 The _LTTng control library_, `liblttng-ctl`, is used to communicate
1819 with a <<lttng-sessiond,session daemon>> using a C API that hides the
1820 underlying protocol's details. `liblttng-ctl` is part of LTTng-tools.
1822 The <<lttng-cli,cmd:lttng command-line tool>>
1823 is linked with `liblttng-ctl`.
1825 You can use `liblttng-ctl` in C or $$C++$$ source code by including its
1830 #include <lttng/lttng.h>
1833 Some objects are referenced by name (C string), such as tracing
1834 sessions, but most of them require to create a handle first using
1835 `lttng_create_handle()`.
1837 The best available developer documentation for `liblttng-ctl` is, as of
1838 LTTng{nbsp}{revision}, its installed header files. Every function and
1839 structure is thoroughly documented.
1843 === User space tracing library
1846 .The user space tracing library.
1847 image::plumbing-liblttng-ust.png[]
1849 The _user space tracing library_, `liblttng-ust` (see man:lttng-ust(3)),
1850 is the LTTng user space tracer. It receives commands from a
1851 <<lttng-sessiond,session daemon>>, for example to
1852 enable and disable specific instrumentation points, and writes event
1853 records to ring buffers shared with a
1854 <<lttng-consumerd,consumer daemon>>.
1855 `liblttng-ust` is part of LTTng-UST.
1857 Public C header files are installed beside `liblttng-ust` to
1858 instrument any <<c-application,C or $$C++$$ application>>.
1860 <<lttng-ust-agents,LTTng-UST agents>>, which are regular Java and Python
1861 packages, use their own library providing tracepoints which is
1862 linked with `liblttng-ust`.
1864 An application or library does not have to initialize `liblttng-ust`
1865 manually: its constructor does the necessary tasks to properly register
1866 to a session daemon. The initialization phase also enables the
1867 instrumentation points matching the <<event,event rules>> that you
1871 [[lttng-ust-agents]]
1872 === User space tracing agents
1875 .The user space tracing agents.
1876 image::plumbing-lttng-ust-agents.png[]
1878 The _LTTng-UST Java and Python agents_ are regular Java and Python
1879 packages which add LTTng tracing capabilities to the
1880 native logging frameworks. The LTTng-UST agents are part of LTTng-UST.
1882 In the case of Java, the
1883 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[`java.util.logging`
1884 core logging facilities] and
1885 https://logging.apache.org/log4j/1.2/[Apache log4j 1.2] are supported.
1886 Note that Apache Log4{nbsp}2 is not supported.
1888 In the case of Python, the standard
1889 https://docs.python.org/3/library/logging.html[`logging`] package
1890 is supported. Both Python 2 and Python 3 modules can import the
1891 LTTng-UST Python agent package.
1893 The applications using the LTTng-UST agents are in the
1894 `java.util.logging` (JUL),
1895 log4j, and Python <<domain,tracing domains>>.
1897 Both agents use the same mechanism to trace the log statements. When an
1898 agent is initialized, it creates a log handler that attaches to the root
1899 logger. The agent also registers to a <<lttng-sessiond,session daemon>>.
1900 When the application executes a log statement, it is passed to the
1901 agent's log handler by the root logger. The agent's log handler calls a
1902 native function in a tracepoint provider package shared library linked
1903 with <<lttng-ust,`liblttng-ust`>>, passing the formatted log message and
1904 other fields, like its logger name and its log level. This native
1905 function contains a user space instrumentation point, hence tracing the
1908 The log level condition of an
1909 <<event,event rule>> is considered when tracing
1910 a Java or a Python application, and it's compatible with the standard
1911 JUL, log4j, and Python log levels.
1915 === LTTng kernel modules
1918 .The LTTng kernel modules.
1919 image::plumbing-lttng-modules.png[]
1921 The _LTTng kernel modules_ are a set of Linux kernel modules
1922 which implement the kernel tracer of the LTTng project. The LTTng
1923 kernel modules are part of LTTng-modules.
1925 The LTTng kernel modules include:
1927 * A set of _probe_ modules.
1929 Each module attaches to a specific subsystem
1930 of the Linux kernel using its tracepoint instrument points. There are
1931 also modules to attach to the entry and return points of the Linux
1932 system call functions.
1934 * _Ring buffer_ modules.
1936 A ring buffer implementation is provided as kernel modules. The LTTng
1937 kernel tracer writes to the ring buffer; a
1938 <<lttng-consumerd,consumer daemon>> reads from the ring buffer.
1940 * The _LTTng kernel tracer_ module.
1941 * The _LTTng logger_ module.
1943 The LTTng logger module implements the special path:{/proc/lttng-logger}
1944 file so that any executable can generate LTTng events by opening and
1945 writing to this file.
1947 See <<proc-lttng-logger-abi,LTTng logger>>.
1949 Generally, you do not have to load the LTTng kernel modules manually
1950 (using man:modprobe(8), for example): a root <<lttng-sessiond,session
1951 daemon>> loads the necessary modules when starting. If you have extra
1952 probe modules, you can specify to load them to the session daemon on
1955 The LTTng kernel modules are installed in
1956 +/usr/lib/modules/__release__/extra+ by default, where +__release__+ is
1957 the kernel release (see `uname --kernel-release`).
1964 .The session daemon.
1965 image::plumbing-sessiond.png[]
1967 The _session daemon_, man:lttng-sessiond(8), is a daemon responsible for
1968 managing tracing sessions and for controlling the various components of
1969 LTTng. The session daemon is part of LTTng-tools.
1971 The session daemon sends control requests to and receives control
1974 * The <<lttng-ust,user space tracing library>>.
1976 Any instance of the user space tracing library first registers to
1977 a session daemon. Then, the session daemon can send requests to
1978 this instance, such as:
1981 ** Get the list of tracepoints.
1982 ** Share an <<event,event rule>> so that the user space tracing library
1983 can enable or disable tracepoints. Amongst the possible conditions
1984 of an event rule is a filter expression which `liblttng-ust` evalutes
1985 when an event occurs.
1986 ** Share <<channel,channel>> attributes and ring buffer locations.
1989 The session daemon and the user space tracing library use a Unix
1990 domain socket for their communication.
1992 * The <<lttng-ust-agents,user space tracing agents>>.
1994 Any instance of a user space tracing agent first registers to
1995 a session daemon. Then, the session daemon can send requests to
1996 this instance, such as:
1999 ** Get the list of loggers.
2000 ** Enable or disable a specific logger.
2003 The session daemon and the user space tracing agent use a TCP connection
2004 for their communication.
2006 * The <<lttng-modules,LTTng kernel tracer>>.
2007 * The <<lttng-consumerd,consumer daemon>>.
2009 The session daemon sends requests to the consumer daemon to instruct
2010 it where to send the trace data streams, amongst other information.
2012 * The <<lttng-relayd,relay daemon>>.
2014 The session daemon receives commands from the
2015 <<liblttng-ctl-lttng,tracing control library>>.
2017 The root session daemon loads the appropriate
2018 <<lttng-modules,LTTng kernel modules>> on startup. It also spawns
2019 a <<lttng-consumerd,consumer daemon>> as soon as you create
2020 an <<event,event rule>>.
2022 The session daemon does not send and receive trace data: this is the
2023 role of the <<lttng-consumerd,consumer daemon>> and
2024 <<lttng-relayd,relay daemon>>. It does, however, generate the
2025 http://diamon.org/ctf/[CTF] metadata stream.
2027 Each Unix user can have its own session daemon instance. The
2028 tracing sessions managed by different session daemons are completely
2031 The root user's session daemon is the only one which is
2032 allowed to control the LTTng kernel tracer, and its spawned consumer
2033 daemon is the only one which is allowed to consume trace data from the
2034 LTTng kernel tracer. Note, however, that any Unix user which is a member
2035 of the <<tracing-group,tracing group>> is allowed
2036 to create <<channel,channels>> in the
2037 Linux kernel <<domain,tracing domain>>, and thus to trace the Linux
2040 The <<lttng-cli,cmd:lttng command-line tool>> automatically starts a
2041 session daemon when using its `create` command if none is currently
2042 running. You can also start the session daemon manually.
2049 .The consumer daemon.
2050 image::plumbing-consumerd.png[]
2052 The _consumer daemon_, man:lttng-consumerd(8), is a daemon which shares
2053 ring buffers with user applications or with the LTTng kernel modules to
2054 collect trace data and send it to some location (on disk or to a
2055 <<lttng-relayd,relay daemon>> over the network). The consumer daemon
2056 is part of LTTng-tools.
2058 You do not start a consumer daemon manually: a consumer daemon is always
2059 spawned by a <<lttng-sessiond,session daemon>> as soon as you create an
2060 <<event,event rule>>, that is, before you start tracing. When you kill
2061 its owner session daemon, the consumer daemon also exits because it is
2062 the session daemon's child process. Command-line options of
2063 man:lttng-sessiond(8) target the consumer daemon process.
2065 There are up to two running consumer daemons per Unix user, whereas only
2066 one session daemon can run per user. This is because each process can be
2067 either 32-bit or 64-bit: if the target system runs a mixture of 32-bit
2068 and 64-bit processes, it is more efficient to have separate
2069 corresponding 32-bit and 64-bit consumer daemons. The root user is an
2070 exception: it can have up to _three_ running consumer daemons: 32-bit
2071 and 64-bit instances for its user applications, and one more
2072 reserved for collecting kernel trace data.
2080 image::plumbing-relayd.png[]
2082 The _relay daemon_, man:lttng-relayd(8), is a daemon acting as a bridge
2083 between remote session and consumer daemons, local trace files, and a
2084 remote live trace viewer. The relay daemon is part of LTTng-tools.
2086 The main purpose of the relay daemon is to implement a receiver of
2087 <<sending-trace-data-over-the-network,trace data over the network>>.
2088 This is useful when the target system does not have much file system
2089 space to record trace files locally.
2091 The relay daemon is also a server to which a
2092 <<lttng-live,live trace viewer>> can
2093 connect. The live trace viewer sends requests to the relay daemon to
2094 receive trace data as the target system emits events. The
2095 communication protocol is named _LTTng live_; it is used over TCP
2098 Note that you can start the relay daemon on the target system directly.
2099 This is the setup of choice when the use case is to view events as
2100 the target system emits them without the need of a remote system.
2104 == [[using-lttng]]Instrumentation
2106 There are many examples of tracing and monitoring in our everyday life:
2108 * You have access to real-time and historical weather reports and
2109 forecasts thanks to weather stations installed around the country.
2110 * You know your heart is safe thanks to an electrocardiogram.
2111 * You make sure not to drive your car too fast and to have enough fuel
2112 to reach your destination thanks to gauges visible on your dashboard.
2114 All the previous examples have something in common: they rely on
2115 **instruments**. Without the electrodes attached to the surface of your
2116 body's skin, cardiac monitoring is futile.
2118 LTTng, as a tracer, is no different from those real life examples. If
2119 you're about to trace a software system or, in other words, record its
2120 history of execution, you better have **instrumentation points** in the
2121 subject you're tracing, that is, the actual software.
2123 Various ways were developed to instrument a piece of software for LTTng
2124 tracing. The most straightforward one is to manually place
2125 instrumentation points, called _tracepoints_, in the software's source
2126 code. It is also possible to add instrumentation points dynamically in
2127 the Linux kernel <<domain,tracing domain>>.
2129 If you're only interested in tracing the Linux kernel, your
2130 instrumentation needs are probably already covered by LTTng's built-in
2131 <<lttng-modules,Linux kernel tracepoints>>. You may also wish to trace a
2132 user application which is already instrumented for LTTng tracing.
2133 In such cases, you can skip this whole section and read the topics of
2134 the <<controlling-tracing,Tracing control>> section.
2136 Many methods are available to instrument a piece of software for LTTng
2139 * <<c-application,User space instrumentation for C and $$C++$$
2141 * <<prebuilt-ust-helpers,Prebuilt user space tracing helpers>>.
2142 * <<java-application,User space Java agent>>.
2143 * <<python-application,User space Python agent>>.
2144 * <<proc-lttng-logger-abi,LTTng logger>>.
2145 * <<instrumenting-linux-kernel,LTTng kernel tracepoints>>.
2149 === [[cxx-application]]User space instrumentation for C and $$C++$$ applications
2151 The procedure to instrument a C or $$C++$$ user application with
2152 the <<lttng-ust,LTTng user space tracing library>>, `liblttng-ust`, is:
2154 . <<tracepoint-provider,Create the source files of a tracepoint provider
2156 . <<probing-the-application-source-code,Add tracepoints to
2157 the application's source code>>.
2158 . <<building-tracepoint-providers-and-user-application,Build and link
2159 a tracepoint provider package and the user application>>.
2161 If you need quick, man:printf(3)-like instrumentation, you can skip
2162 those steps and use <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>>
2165 IMPORTANT: You need to <<installing-lttng,install>> LTTng-UST to
2166 instrument a user application with `liblttng-ust`.
2169 [[tracepoint-provider]]
2170 ==== Create the source files of a tracepoint provider package
2172 A _tracepoint provider_ is a set of compiled functions which provide
2173 **tracepoints** to an application, the type of instrumentation point
2174 supported by LTTng-UST. Those functions can emit events with
2175 user-defined fields and serialize those events as event records to one
2176 or more LTTng-UST <<channel,channel>> sub-buffers. The `tracepoint()`
2177 macro, which you <<probing-the-application-source-code,insert in a user
2178 application's source code>>, calls those functions.
2180 A _tracepoint provider package_ is an object file (`.o`) or a shared
2181 library (`.so`) which contains one or more tracepoint providers.
2182 Its source files are:
2184 * One or more <<tpp-header,tracepoint provider header>> (`.h`).
2185 * A <<tpp-source,tracepoint provider package source>> (`.c`).
2187 A tracepoint provider package is dynamically linked with `liblttng-ust`,
2188 the LTTng user space tracer, at run time.
2191 .User application linked with `liblttng-ust` and containing a tracepoint provider.
2192 image::ust-app.png[]
2194 NOTE: If you need quick, man:printf(3)-like instrumentation, you can
2195 skip creating and using a tracepoint provider and use
2196 <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>> instead.
2200 ===== Create a tracepoint provider header file template
2202 A _tracepoint provider header file_ contains the tracepoint
2203 definitions of a tracepoint provider.
2205 To create a tracepoint provider header file:
2207 . Start from this template:
2211 .Tracepoint provider header file template (`.h` file extension).
2213 #undef TRACEPOINT_PROVIDER
2214 #define TRACEPOINT_PROVIDER provider_name
2216 #undef TRACEPOINT_INCLUDE
2217 #define TRACEPOINT_INCLUDE "./tp.h"
2219 #if !defined(_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
2222 #include <lttng/tracepoint.h>
2225 * Use TRACEPOINT_EVENT(), TRACEPOINT_EVENT_CLASS(),
2226 * TRACEPOINT_EVENT_INSTANCE(), and TRACEPOINT_LOGLEVEL() here.
2231 #include <lttng/tracepoint-event.h>
2237 * `provider_name` with the name of your tracepoint provider.
2238 * `"tp.h"` with the name of your tracepoint provider header file.
2240 . Below the `#include <lttng/tracepoint.h>` line, put your
2241 <<defining-tracepoints,tracepoint definitions>>.
2243 Your tracepoint provider name must be unique amongst all the possible
2244 tracepoint provider names used on the same target system. We
2245 suggest to include the name of your project or company in the name,
2246 for example, `org_lttng_my_project_tpp`.
2248 TIP: [[lttng-gen-tp]]You can use the man:lttng-gen-tp(1) tool to create
2249 this boilerplate for you. When using cmd:lttng-gen-tp, all you need to
2250 write are the <<defining-tracepoints,tracepoint definitions>>.
2253 [[defining-tracepoints]]
2254 ===== Create a tracepoint definition
2256 A _tracepoint definition_ defines, for a given tracepoint:
2258 * Its **input arguments**. They are the macro parameters that the
2259 `tracepoint()` macro accepts for this particular tracepoint
2260 in the user application's source code.
2261 * Its **output event fields**. They are the sources of event fields
2262 that form the payload of any event that the execution of the
2263 `tracepoint()` macro emits for this particular tracepoint.
2265 You can create a tracepoint definition by using the
2266 `TRACEPOINT_EVENT()` macro below the `#include <lttng/tracepoint.h>`
2268 <<tpp-header,tracepoint provider header file template>>.
2270 The syntax of the `TRACEPOINT_EVENT()` macro is:
2273 .`TRACEPOINT_EVENT()` macro syntax.
2276 /* Tracepoint provider name */
2279 /* Tracepoint name */
2282 /* Input arguments */
2287 /* Output event fields */
2296 * `provider_name` with your tracepoint provider name.
2297 * `tracepoint_name` with your tracepoint name.
2298 * `arguments` with the <<tpp-def-input-args,input arguments>>.
2299 * `fields` with the <<tpp-def-output-fields,output event field>>
2302 This tracepoint emits events named `provider_name:tracepoint_name`.
2305 .Event name's length limitation
2307 The concatenation of the tracepoint provider name and the
2308 tracepoint name must not exceed **254 characters**. If it does, the
2309 instrumented application compiles and runs, but LTTng throws multiple
2310 warnings and you could experience serious issues.
2313 [[tpp-def-input-args]]The syntax of the `TP_ARGS()` macro is:
2316 .`TP_ARGS()` macro syntax.
2325 * `type` with the C type of the argument.
2326 * `arg_name` with the argument name.
2328 You can repeat `type` and `arg_name` up to 10 times to have
2329 more than one argument.
2331 .`TP_ARGS()` usage with three arguments.
2343 The `TP_ARGS()` and `TP_ARGS(void)` forms are valid to create a
2344 tracepoint definition with no input arguments.
2346 [[tpp-def-output-fields]]The `TP_FIELDS()` macro contains a list of
2347 `ctf_*()` macros. Each `ctf_*()` macro defines one event field. See
2348 man:lttng-ust(3) for a complete description of the available `ctf_*()`
2349 macros. A `ctf_*()` macro specifies the type, size, and byte order of
2352 Each `ctf_*()` macro takes an _argument expression_ parameter. This is a
2353 C expression that the tracer evalutes at the `tracepoint()` macro site
2354 in the application's source code. This expression provides a field's
2355 source of data. The argument expression can include input argument names
2356 listed in the `TP_ARGS()` macro.
2358 Each `ctf_*()` macro also takes a _field name_ parameter. Field names
2359 must be unique within a given tracepoint definition.
2361 Here's a complete tracepoint definition example:
2363 .Tracepoint definition.
2365 The following tracepoint definition defines a tracepoint which takes
2366 three input arguments and has four output event fields.
2370 #include "my-custom-structure.h"
2376 const struct my_custom_structure*, my_custom_structure,
2381 ctf_string(query_field, query)
2382 ctf_float(double, ratio_field, ratio)
2383 ctf_integer(int, recv_size, my_custom_structure->recv_size)
2384 ctf_integer(int, send_size, my_custom_structure->send_size)
2389 You can refer to this tracepoint definition with the `tracepoint()`
2390 macro in your application's source code like this:
2394 tracepoint(my_provider, my_tracepoint,
2395 my_structure, some_ratio, the_query);
2399 NOTE: The LTTng tracer only evaluates tracepoint arguments at run time
2400 if they satisfy an enabled <<event,event rule>>.
2403 [[using-tracepoint-classes]]
2404 ===== Use a tracepoint class
2406 A _tracepoint class_ is a class of tracepoints which share the same
2407 output event field definitions. A _tracepoint instance_ is one
2408 instance of such a defined tracepoint class, with its own tracepoint
2411 The <<defining-tracepoints,`TRACEPOINT_EVENT()` macro>> is actually a
2412 shorthand which defines both a tracepoint class and a tracepoint
2413 instance at the same time.
2415 When you build a tracepoint provider package, the C or $$C++$$ compiler
2416 creates one serialization function for each **tracepoint class**. A
2417 serialization function is responsible for serializing the event fields
2418 of a tracepoint to a sub-buffer when tracing.
2420 For various performance reasons, when your situation requires multiple
2421 tracepoint definitions with different names, but with the same event
2422 fields, we recommend that you manually create a tracepoint class
2423 and instantiate as many tracepoint instances as needed. One positive
2424 effect of such a design, amongst other advantages, is that all
2425 tracepoint instances of the same tracepoint class reuse the same
2426 serialization function, thus reducing
2427 https://en.wikipedia.org/wiki/Cache_pollution[cache pollution].
2429 .Use a tracepoint class and tracepoint instances.
2431 Consider the following three tracepoint definitions:
2443 ctf_integer(int, userid, userid)
2444 ctf_integer(size_t, len, len)
2456 ctf_integer(int, userid, userid)
2457 ctf_integer(size_t, len, len)
2469 ctf_integer(int, userid, userid)
2470 ctf_integer(size_t, len, len)
2475 In this case, we create three tracepoint classes, with one implicit
2476 tracepoint instance for each of them: `get_account`, `get_settings`, and
2477 `get_transaction`. However, they all share the same event field names
2478 and types. Hence three identical, yet independent serialization
2479 functions are created when you build the tracepoint provider package.
2481 A better design choice is to define a single tracepoint class and three
2482 tracepoint instances:
2486 /* The tracepoint class */
2487 TRACEPOINT_EVENT_CLASS(
2488 /* Tracepoint provider name */
2491 /* Tracepoint class name */
2494 /* Input arguments */
2500 /* Output event fields */
2502 ctf_integer(int, userid, userid)
2503 ctf_integer(size_t, len, len)
2507 /* The tracepoint instances */
2508 TRACEPOINT_EVENT_INSTANCE(
2509 /* Tracepoint provider name */
2512 /* Tracepoint class name */
2515 /* Tracepoint name */
2518 /* Input arguments */
2524 TRACEPOINT_EVENT_INSTANCE(
2533 TRACEPOINT_EVENT_INSTANCE(
2546 [[assigning-log-levels]]
2547 ===== Assign a log level to a tracepoint definition
2549 You can assign an optional _log level_ to a
2550 <<defining-tracepoints,tracepoint definition>>.
2552 Assigning different levels of severity to tracepoint definitions can
2553 be useful: when you <<enabling-disabling-events,create an event rule>>,
2554 you can target tracepoints having a log level as severe as a specific
2557 The concept of LTTng-UST log levels is similar to the levels found
2558 in typical logging frameworks:
2560 * In a logging framework, the log level is given by the function
2561 or method name you use at the log statement site: `debug()`,
2562 `info()`, `warn()`, `error()`, and so on.
2563 * In LTTng-UST, you statically assign the log level to a tracepoint
2564 definition; any `tracepoint()` macro invocation which refers to
2565 this definition has this log level.
2567 You can assign a log level to a tracepoint definition with the
2568 `TRACEPOINT_LOGLEVEL()` macro. You must use this macro _after_ the
2569 <<defining-tracepoints,`TRACEPOINT_EVENT()`>> or
2570 <<using-tracepoint-classes,`TRACEPOINT_INSTANCE()`>> macro for a given
2573 The syntax of the `TRACEPOINT_LOGLEVEL()` macro is:
2576 .`TRACEPOINT_LOGLEVEL()` macro syntax.
2578 TRACEPOINT_LOGLEVEL(provider_name, tracepoint_name, log_level)
2583 * `provider_name` with the tracepoint provider name.
2584 * `tracepoint_name` with the tracepoint name.
2585 * `log_level` with the log level to assign to the tracepoint
2586 definition named `tracepoint_name` in the `provider_name`
2587 tracepoint provider.
2589 See man:lttng-ust(3) for a list of available log level names.
2591 .Assign the `TRACE_DEBUG_UNIT` log level to a tracepoint definition.
2595 /* Tracepoint definition */
2604 ctf_integer(int, userid, userid)
2605 ctf_integer(size_t, len, len)
2609 /* Log level assignment */
2610 TRACEPOINT_LOGLEVEL(my_app, get_transaction, TRACE_DEBUG_UNIT)
2616 ===== Create a tracepoint provider package source file
2618 A _tracepoint provider package source file_ is a C source file which
2619 includes a <<tpp-header,tracepoint provider header file>> to expand its
2620 macros into event serialization and other functions.
2622 You can always use the following tracepoint provider package source
2626 .Tracepoint provider package source file template.
2628 #define TRACEPOINT_CREATE_PROBES
2633 Replace `tp.h` with the name of your <<tpp-header,tracepoint provider
2634 header file>> name. You may also include more than one tracepoint
2635 provider header file here to create a tracepoint provider package
2636 holding more than one tracepoint providers.
2639 [[probing-the-application-source-code]]
2640 ==== Add tracepoints to an application's source code
2642 Once you <<tpp-header,create a tracepoint provider header file>>, you
2643 can use the `tracepoint()` macro in your application's
2644 source code to insert the tracepoints that this header
2645 <<defining-tracepoints,defines>>.
2647 The `tracepoint()` macro takes at least two parameters: the tracepoint
2648 provider name and the tracepoint name. The corresponding tracepoint
2649 definition defines the other parameters.
2651 .`tracepoint()` usage.
2653 The following <<defining-tracepoints,tracepoint definition>> defines a
2654 tracepoint which takes two input arguments and has two output event
2658 .Tracepoint provider header file.
2660 #include "my-custom-structure.h"
2667 const char*, cmd_name
2670 ctf_string(cmd_name, cmd_name)
2671 ctf_integer(int, number_of_args, argc)
2676 You can refer to this tracepoint definition with the `tracepoint()`
2677 macro in your application's source code like this:
2680 .Application's source file.
2684 int main(int argc, char* argv[])
2686 tracepoint(my_provider, my_tracepoint, argc, argv[0]);
2692 Note how the application's source code includes
2693 the tracepoint provider header file containing the tracepoint
2694 definitions to use, path:{tp.h}.
2697 .`tracepoint()` usage with a complex tracepoint definition.
2699 Consider this complex tracepoint definition, where multiple event
2700 fields refer to the same input arguments in their argument expression
2704 .Tracepoint provider header file.
2706 /* For `struct stat` */
2707 #include <sys/types.h>
2708 #include <sys/stat.h>
2720 ctf_integer(int, my_constant_field, 23 + 17)
2721 ctf_integer(int, my_int_arg_field, my_int_arg)
2722 ctf_integer(int, my_int_arg_field2, my_int_arg * my_int_arg)
2723 ctf_integer(int, sum4_field, my_str_arg[0] + my_str_arg[1] +
2724 my_str_arg[2] + my_str_arg[3])
2725 ctf_string(my_str_arg_field, my_str_arg)
2726 ctf_integer_hex(off_t, size_field, st->st_size)
2727 ctf_float(double, size_dbl_field, (double) st->st_size)
2728 ctf_sequence_text(char, half_my_str_arg_field, my_str_arg,
2729 size_t, strlen(my_str_arg) / 2)
2734 You can refer to this tracepoint definition with the `tracepoint()`
2735 macro in your application's source code like this:
2738 .Application's source file.
2740 #define TRACEPOINT_DEFINE
2747 stat("/etc/fstab", &s);
2748 tracepoint(my_provider, my_tracepoint, 23, "Hello, World!", &s);
2754 If you look at the event record that LTTng writes when tracing this
2755 program, assuming the file size of path:{/etc/fstab} is 301{nbsp}bytes,
2756 it should look like this:
2758 .Event record fields
2760 |Field's name |Field's value
2761 |`my_constant_field` |40
2762 |`my_int_arg_field` |23
2763 |`my_int_arg_field2` |529
2765 |`my_str_arg_field` |`Hello, World!`
2766 |`size_field` |0x12d
2767 |`size_dbl_field` |301.0
2768 |`half_my_str_arg_field` |`Hello,`
2772 Sometimes, the arguments you pass to `tracepoint()` are expensive to
2773 compute--they use the call stack, for example. To avoid this
2774 computation when the tracepoint is disabled, you can use the
2775 `tracepoint_enabled()` and `do_tracepoint()` macros.
2777 The syntax of the `tracepoint_enabled()` and `do_tracepoint()` macros
2781 .`tracepoint_enabled()` and `do_tracepoint()` macros syntax.
2783 tracepoint_enabled(provider_name, tracepoint_name)
2784 do_tracepoint(provider_name, tracepoint_name, ...)
2789 * `provider_name` with the tracepoint provider name.
2790 * `tracepoint_name` with the tracepoint name.
2792 `tracepoint_enabled()` returns a non-zero value if the tracepoint named
2793 `tracepoint_name` from the provider named `provider_name` is enabled
2796 `do_tracepoint()` is like `tracepoint()`, except that it doesn't check
2797 if the tracepoint is enabled. Using `tracepoint()` with
2798 `tracepoint_enabled()` is dangerous since `tracepoint()` also contains
2799 the `tracepoint_enabled()` check, thus a race condition is
2800 possible in this situation:
2803 .Possible race condition when using `tracepoint_enabled()` with `tracepoint()`.
2805 if (tracepoint_enabled(my_provider, my_tracepoint)) {
2806 stuff = prepare_stuff();
2809 tracepoint(my_provider, my_tracepoint, stuff);
2812 If the tracepoint is enabled after the condition, then `stuff` is not
2813 prepared: the emitted event will either contain wrong data, or the whole
2814 application could crash (segmentation fault, for example).
2816 NOTE: Neither `tracepoint_enabled()` nor `do_tracepoint()` have an
2817 `STAP_PROBEV()` call. If you need it, you must emit
2821 [[building-tracepoint-providers-and-user-application]]
2822 ==== Build and link a tracepoint provider package and an application
2824 Once you have one or more <<tpp-header,tracepoint provider header
2825 files>> and a <<tpp-source,tracepoint provider package source file>>,
2826 you can create the tracepoint provider package by compiling its source
2827 file. From here, multiple build and run scenarios are possible. The
2828 following table shows common application and library configurations
2829 along with the required command lines to achieve them.
2831 In the following diagrams, we use the following file names:
2834 Executable application.
2837 Application's object file.
2840 Tracepoint provider package object file.
2843 Tracepoint provider package archive file.
2846 Tracepoint provider package shared object file.
2849 User library object file.
2852 User library shared object file.
2854 We use the following symbols in the diagrams of table below:
2857 .Symbols used in the build scenario diagrams.
2858 image::ust-sit-symbols.png[]
2860 We assume that path:{.} is part of the env:LD_LIBRARY_PATH environment
2861 variable in the following instructions.
2863 [role="growable ust-scenarios",cols="asciidoc,asciidoc"]
2864 .Common tracepoint provider package scenarios.
2866 |Scenario |Instructions
2869 The instrumented application is statically linked with
2870 the tracepoint provider package object.
2872 image::ust-sit+app-linked-with-tp-o+app-instrumented.png[]
2875 include::../common/ust-sit-step-tp-o.txt[]
2877 To build the instrumented application:
2879 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2884 #define TRACEPOINT_DEFINE
2888 . Compile the application source file:
2897 . Build the application:
2902 $ gcc -o app app.o tpp.o -llttng-ust -ldl
2906 To run the instrumented application:
2908 * Start the application:
2918 The instrumented application is statically linked with the
2919 tracepoint provider package archive file.
2921 image::ust-sit+app-linked-with-tp-a+app-instrumented.png[]
2924 To create the tracepoint provider package archive file:
2926 . Compile the <<tpp-source,tracepoint provider package source file>>:
2935 . Create the tracepoint provider package archive file:
2940 $ ar rcs tpp.a tpp.o
2944 To build the instrumented application:
2946 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2951 #define TRACEPOINT_DEFINE
2955 . Compile the application source file:
2964 . Build the application:
2969 $ gcc -o app app.o tpp.a -llttng-ust -ldl
2973 To run the instrumented application:
2975 * Start the application:
2985 The instrumented application is linked with the tracepoint provider
2986 package shared object.
2988 image::ust-sit+app-linked-with-tp-so+app-instrumented.png[]
2991 include::../common/ust-sit-step-tp-so.txt[]
2993 To build the instrumented application:
2995 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3000 #define TRACEPOINT_DEFINE
3004 . Compile the application source file:
3013 . Build the application:
3018 $ gcc -o app app.o -ldl -L. -ltpp
3022 To run the instrumented application:
3024 * Start the application:
3034 The tracepoint provider package shared object is preloaded before the
3035 instrumented application starts.
3037 image::ust-sit+tp-so-preloaded+app-instrumented.png[]
3040 include::../common/ust-sit-step-tp-so.txt[]
3042 To build the instrumented application:
3044 . In path:{app.c}, before including path:{tpp.h}, add the
3050 #define TRACEPOINT_DEFINE
3051 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3055 . Compile the application source file:
3064 . Build the application:
3069 $ gcc -o app app.o -ldl
3073 To run the instrumented application with tracing support:
3075 * Preload the tracepoint provider package shared object and
3076 start the application:
3081 $ LD_PRELOAD=./libtpp.so ./app
3085 To run the instrumented application without tracing support:
3087 * Start the application:
3097 The instrumented application dynamically loads the tracepoint provider
3098 package shared object.
3100 See the <<dlclose-warning,warning about `dlclose()`>>.
3102 image::ust-sit+app-dlopens-tp-so+app-instrumented.png[]
3105 include::../common/ust-sit-step-tp-so.txt[]
3107 To build the instrumented application:
3109 . In path:{app.c}, before including path:{tpp.h}, add the
3115 #define TRACEPOINT_DEFINE
3116 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3120 . Compile the application source file:
3129 . Build the application:
3134 $ gcc -o app app.o -ldl
3138 To run the instrumented application:
3140 * Start the application:
3150 The application is linked with the instrumented user library.
3152 The instrumented user library is statically linked with the tracepoint
3153 provider package object file.
3155 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-o+lib-instrumented.png[]
3158 include::../common/ust-sit-step-tp-o-fpic.txt[]
3160 To build the instrumented user library:
3162 . In path:{emon.c}, before including path:{tpp.h}, add the
3168 #define TRACEPOINT_DEFINE
3172 . Compile the user library source file:
3177 $ gcc -I. -fpic -c emon.c
3181 . Build the user library shared object:
3186 $ gcc -shared -o libemon.so emon.o tpp.o -llttng-ust -ldl
3190 To build the application:
3192 . Compile the application source file:
3201 . Build the application:
3206 $ gcc -o app app.o -L. -lemon
3210 To run the application:
3212 * Start the application:
3222 The application is linked with the instrumented user library.
3224 The instrumented user library is linked with the tracepoint provider
3225 package shared object.
3227 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3230 include::../common/ust-sit-step-tp-so.txt[]
3232 To build the instrumented user library:
3234 . In path:{emon.c}, before including path:{tpp.h}, add the
3240 #define TRACEPOINT_DEFINE
3244 . Compile the user library source file:
3249 $ gcc -I. -fpic -c emon.c
3253 . Build the user library shared object:
3258 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3262 To build the application:
3264 . Compile the application source file:
3273 . Build the application:
3278 $ gcc -o app app.o -L. -lemon
3282 To run the application:
3284 * Start the application:
3294 The tracepoint provider package shared object is preloaded before the
3297 The application is linked with the instrumented user library.
3299 image::ust-sit+tp-so-preloaded+app-linked-with-lib+lib-instrumented.png[]
3302 include::../common/ust-sit-step-tp-so.txt[]
3304 To build the instrumented user library:
3306 . In path:{emon.c}, before including path:{tpp.h}, add the
3312 #define TRACEPOINT_DEFINE
3313 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3317 . Compile the user library source file:
3322 $ gcc -I. -fpic -c emon.c
3326 . Build the user library shared object:
3331 $ gcc -shared -o libemon.so emon.o -ldl
3335 To build the application:
3337 . Compile the application source file:
3346 . Build the application:
3351 $ gcc -o app app.o -L. -lemon
3355 To run the application with tracing support:
3357 * Preload the tracepoint provider package shared object and
3358 start the application:
3363 $ LD_PRELOAD=./libtpp.so ./app
3367 To run the application without tracing support:
3369 * Start the application:
3379 The application is linked with the instrumented user library.
3381 The instrumented user library dynamically loads the tracepoint provider
3382 package shared object.
3384 See the <<dlclose-warning,warning about `dlclose()`>>.
3386 image::ust-sit+app-linked-with-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3389 include::../common/ust-sit-step-tp-so.txt[]
3391 To build the instrumented user library:
3393 . In path:{emon.c}, before including path:{tpp.h}, add the
3399 #define TRACEPOINT_DEFINE
3400 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3404 . Compile the user library source file:
3409 $ gcc -I. -fpic -c emon.c
3413 . Build the user library shared object:
3418 $ gcc -shared -o libemon.so emon.o -ldl
3422 To build the application:
3424 . Compile the application source file:
3433 . Build the application:
3438 $ gcc -o app app.o -L. -lemon
3442 To run the application:
3444 * Start the application:
3454 The application dynamically loads the instrumented user library.
3456 The instrumented user library is linked with the tracepoint provider
3457 package shared object.
3459 See the <<dlclose-warning,warning about `dlclose()`>>.
3461 image::ust-sit+app-dlopens-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3464 include::../common/ust-sit-step-tp-so.txt[]
3466 To build the instrumented user library:
3468 . In path:{emon.c}, before including path:{tpp.h}, add the
3474 #define TRACEPOINT_DEFINE
3478 . Compile the user library source file:
3483 $ gcc -I. -fpic -c emon.c
3487 . Build the user library shared object:
3492 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3496 To build the application:
3498 . Compile the application source file:
3507 . Build the application:
3512 $ gcc -o app app.o -ldl -L. -lemon
3516 To run the application:
3518 * Start the application:
3528 The application dynamically loads the instrumented user library.
3530 The instrumented user library dynamically loads the tracepoint provider
3531 package shared object.
3533 See the <<dlclose-warning,warning about `dlclose()`>>.
3535 image::ust-sit+app-dlopens-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3538 include::../common/ust-sit-step-tp-so.txt[]
3540 To build the instrumented user library:
3542 . In path:{emon.c}, before including path:{tpp.h}, add the
3548 #define TRACEPOINT_DEFINE
3549 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3553 . Compile the user library source file:
3558 $ gcc -I. -fpic -c emon.c
3562 . Build the user library shared object:
3567 $ gcc -shared -o libemon.so emon.o -ldl
3571 To build the application:
3573 . Compile the application source file:
3582 . Build the application:
3587 $ gcc -o app app.o -ldl -L. -lemon
3591 To run the application:
3593 * Start the application:
3603 The tracepoint provider package shared object is preloaded before the
3606 The application dynamically loads the instrumented user library.
3608 image::ust-sit+tp-so-preloaded+app-dlopens-lib+lib-instrumented.png[]
3611 include::../common/ust-sit-step-tp-so.txt[]
3613 To build the instrumented user library:
3615 . In path:{emon.c}, before including path:{tpp.h}, add the
3621 #define TRACEPOINT_DEFINE
3622 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3626 . Compile the user library source file:
3631 $ gcc -I. -fpic -c emon.c
3635 . Build the user library shared object:
3640 $ gcc -shared -o libemon.so emon.o -ldl
3644 To build the application:
3646 . Compile the application source file:
3655 . Build the application:
3660 $ gcc -o app app.o -L. -lemon
3664 To run the application with tracing support:
3666 * Preload the tracepoint provider package shared object and
3667 start the application:
3672 $ LD_PRELOAD=./libtpp.so ./app
3676 To run the application without tracing support:
3678 * Start the application:
3688 The application is statically linked with the tracepoint provider
3689 package object file.
3691 The application is linked with the instrumented user library.
3693 image::ust-sit+app-linked-with-tp-o+app-linked-with-lib+lib-instrumented.png[]
3696 include::../common/ust-sit-step-tp-o.txt[]
3698 To build the instrumented user library:
3700 . In path:{emon.c}, before including path:{tpp.h}, add the
3706 #define TRACEPOINT_DEFINE
3710 . Compile the user library source file:
3715 $ gcc -I. -fpic -c emon.c
3719 . Build the user library shared object:
3724 $ gcc -shared -o libemon.so emon.o
3728 To build the application:
3730 . Compile the application source file:
3739 . Build the application:
3744 $ gcc -o app app.o tpp.o -llttng-ust -ldl -L. -lemon
3748 To run the instrumented application:
3750 * Start the application:
3760 The application is statically linked with the tracepoint provider
3761 package object file.
3763 The application dynamically loads the instrumented user library.
3765 image::ust-sit+app-linked-with-tp-o+app-dlopens-lib+lib-instrumented.png[]
3768 include::../common/ust-sit-step-tp-o.txt[]
3770 To build the application:
3772 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3777 #define TRACEPOINT_DEFINE
3781 . Compile the application source file:
3790 . Build the application:
3795 $ gcc -Wl,--export-dynamic -o app app.o tpp.o \
3800 The `--export-dynamic` option passed to the linker is necessary for the
3801 dynamically loaded library to ``see'' the tracepoint symbols defined in
3804 To build the instrumented user library:
3806 . Compile the user library source file:
3811 $ gcc -I. -fpic -c emon.c
3815 . Build the user library shared object:
3820 $ gcc -shared -o libemon.so emon.o
3824 To run the application:
3826 * Start the application:
3838 .Do not use man:dlclose(3) on a tracepoint provider package
3840 Never use man:dlclose(3) on any shared object which:
3842 * Is linked with, statically or dynamically, a tracepoint provider
3844 * Calls man:dlopen(3) itself to dynamically open a tracepoint provider
3845 package shared object.
3847 This is currently considered **unsafe** due to a lack of reference
3848 counting from LTTng-UST to the shared object.
3850 A known workaround (available since glibc 2.2) is to use the
3851 `RTLD_NODELETE` flag when calling man:dlopen(3) initially. This has the
3852 effect of not unloading the loaded shared object, even if man:dlclose(3)
3855 You can also preload the tracepoint provider package shared object with
3856 the env:LD_PRELOAD environment variable to overcome this limitation.
3860 [[using-lttng-ust-with-daemons]]
3861 ===== Use noch:{LTTng-UST} with daemons
3863 If your instrumented application calls man:fork(2), man:clone(2),
3864 or BSD's man:rfork(2), without a following man:exec(3)-family
3865 system call, you must preload the path:{liblttng-ust-fork.so} shared
3866 object when starting the application.
3870 $ LD_PRELOAD=liblttng-ust-fork.so ./my-app
3873 If your tracepoint provider package is
3874 a shared library which you also preload, you must put both
3875 shared objects in env:LD_PRELOAD:
3879 $ LD_PRELOAD=liblttng-ust-fork.so:/path/to/tp.so ./my-app
3883 [[lttng-ust-pkg-config]]
3884 ===== Use noch:{pkg-config}
3886 On some distributions, LTTng-UST ships with a
3887 https://www.freedesktop.org/wiki/Software/pkg-config/[pkg-config]
3888 metadata file. If this is your case, then you can use cmd:pkg-config to
3889 build an application on the command line:
3893 $ gcc -o my-app my-app.o tp.o $(pkg-config --cflags --libs lttng-ust)
3897 [[instrumenting-32-bit-app-on-64-bit-system]]
3898 ===== [[advanced-instrumenting-techniques]]Build a 32-bit instrumented application for a 64-bit target system
3900 In order to trace a 32-bit application running on a 64-bit system,
3901 LTTng must use a dedicated 32-bit
3902 <<lttng-consumerd,consumer daemon>>.
3904 The following steps show how to build and install a 32-bit consumer
3905 daemon, which is _not_ part of the default 64-bit LTTng build, how to
3906 build and install the 32-bit LTTng-UST libraries, and how to build and
3907 link an instrumented 32-bit application in that context.
3909 To build a 32-bit instrumented application for a 64-bit target system,
3910 assuming you have a fresh target system with no installed Userspace RCU
3913 . Download, build, and install a 32-bit version of Userspace RCU:
3918 $ cd $(mktemp -d) &&
3919 wget http://lttng.org/files/urcu/userspace-rcu-latest-0.9.tar.bz2 &&
3920 tar -xf userspace-rcu-latest-0.9.tar.bz2 &&
3921 cd userspace-rcu-0.9.* &&
3922 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 &&
3924 sudo make install &&
3929 . Using your distribution's package manager, or from source, install
3930 the following 32-bit versions of the following dependencies of
3931 LTTng-tools and LTTng-UST:
3934 * https://sourceforge.net/projects/libuuid/[libuuid]
3935 * http://directory.fsf.org/wiki/Popt[popt]
3936 * http://www.xmlsoft.org/[libxml2]
3939 . Download, build, and install a 32-bit version of the latest
3940 LTTng-UST{nbsp}{revision}:
3945 $ cd $(mktemp -d) &&
3946 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.8.tar.bz2 &&
3947 tar -xf lttng-ust-latest-2.8.tar.bz2 &&
3948 cd lttng-ust-2.8.* &&
3949 ./configure --libdir=/usr/local/lib32 \
3950 CFLAGS=-m32 CXXFLAGS=-m32 \
3951 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
3953 sudo make install &&
3960 Depending on your distribution,
3961 32-bit libraries could be installed at a different location than
3962 `/usr/lib32`. For example, Debian is known to install
3963 some 32-bit libraries in `/usr/lib/i386-linux-gnu`.
3965 In this case, make sure to set `LDFLAGS` to all the
3966 relevant 32-bit library paths, for example:
3970 $ LDFLAGS='-L/usr/lib/i386-linux-gnu -L/usr/lib32'
3974 . Download the latest LTTng-tools{nbsp}{revision}, build, and install
3975 the 32-bit consumer daemon:
3980 $ cd $(mktemp -d) &&
3981 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.8.tar.bz2 &&
3982 tar -xf lttng-tools-latest-2.8.tar.bz2 &&
3983 cd lttng-tools-2.8.* &&
3984 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 CXXFLAGS=-m32 \
3985 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' \
3986 --disable-bin-lttng --disable-bin-lttng-crash \
3987 --disable-bin-lttng-relayd --disable-bin-lttng-sessiond &&
3989 cd src/bin/lttng-consumerd &&
3990 sudo make install &&
3995 . From your distribution or from source,
3996 <<installing-lttng,install>> the 64-bit versions of
3997 LTTng-UST and Userspace RCU.
3998 . Download, build, and install the 64-bit version of the
3999 latest LTTng-tools{nbsp}{revision}:
4004 $ cd $(mktemp -d) &&
4005 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.8.tar.bz2 &&
4006 tar -xf lttng-tools-latest-2.8.tar.bz2 &&
4007 cd lttng-tools-2.8.* &&
4008 ./configure --with-consumerd32-libdir=/usr/local/lib32 \
4009 --with-consumerd32-bin=/usr/local/lib32/lttng/libexec/lttng-consumerd &&
4011 sudo make install &&
4016 . Pass the following options to man:gcc(1), man:g++(1), or man:clang(1)
4017 when linking your 32-bit application:
4020 -m32 -L/usr/lib32 -L/usr/local/lib32 \
4021 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32
4024 For example, let's rebuild the quick start example in
4025 <<tracing-your-own-user-application,Trace a user application>> as an
4026 instrumented 32-bit application:
4031 $ gcc -m32 -c -I. hello-tp.c
4032 $ gcc -m32 -c hello.c
4033 $ gcc -m32 -o hello hello.o hello-tp.o \
4034 -L/usr/lib32 -L/usr/local/lib32 \
4035 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32 \
4040 No special action is required to execute the 32-bit application and
4041 to trace it: use the command-line man:lttng(1) tool as usual.
4048 man:tracef(3) is a small LTTng-UST API designed for quick,
4049 man:printf(3)-like instrumentation without the burden of
4050 <<tracepoint-provider,creating>> and
4051 <<building-tracepoint-providers-and-user-application,building>>
4052 a tracepoint provider package.
4054 To use `tracef()` in your application:
4056 . In the C or C++ source files where you need to use `tracef()`,
4057 include `<lttng/tracef.h>`:
4062 #include <lttng/tracef.h>
4066 . In the application's source code, use `tracef()` like you would use
4074 tracef("my message: %d (%s)", my_integer, my_string);
4080 . Link your application with `liblttng-ust`:
4085 $ gcc -o app app.c -llttng-ust
4089 To trace the events that `tracef()` calls emit:
4091 * <<enabling-disabling-events,Create an event rule>> which matches the
4092 `lttng_ust_tracef:*` event name:
4097 $ lttng enable-event --userspace 'lttng_ust_tracef:*'
4102 .Limitations of `tracef()`
4104 The `tracef()` utility function was developed to make user space tracing
4105 super simple, albeit with notable disadvantages compared to
4106 <<defining-tracepoints,user-defined tracepoints>>:
4108 * All the emitted events have the same tracepoint provider and
4109 tracepoint names, respectively `lttng_ust_tracef` and `event`.
4110 * There is no static type checking.
4111 * The only event record field you actually get, named `msg`, is a string
4112 potentially containing the values you passed to `tracef()`
4113 using your own format string. This also means that you cannot filter
4114 events with a custom expression at run time because there are no
4116 * Since `tracef()` uses the C standard library's man:vasprintf(3)
4117 function behind the scenes to format the strings at run time, its
4118 expected performance is lower than with user-defined tracepoints,
4119 which do not require a conversion to a string.
4121 Taking this into consideration, `tracef()` is useful for some quick
4122 prototyping and debugging, but you should not consider it for any
4123 permanent and serious applicative instrumentation.
4129 ==== Use `tracelog()`
4131 The man:tracelog(3) API is very similar to <<tracef,`tracef()`>>, with
4132 the difference that it accepts an additional log level parameter.
4134 The goal of `tracelog()` is to ease the migration from logging to
4137 To use `tracelog()` in your application:
4139 . In the C or C++ source files where you need to use `tracelog()`,
4140 include `<lttng/tracelog.h>`:
4145 #include <lttng/tracelog.h>
4149 . In the application's source code, use `tracelog()` like you would use
4150 man:printf(3), except for the first parameter which is the log
4158 tracelog(TRACE_WARNING, "my message: %d (%s)",
4159 my_integer, my_string);
4165 See man:lttng-ust(3) for a list of available log level names.
4167 . Link your application with `liblttng-ust`:
4172 $ gcc -o app app.c -llttng-ust
4176 To trace the events that `tracelog()` calls emit with a log level
4177 _as severe as_ a specific log level:
4179 * <<enabling-disabling-events,Create an event rule>> which matches the
4180 `lttng_ust_tracelog:*` event name and a minimum level
4186 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4187 --loglevel=TRACE_WARNING
4191 To trace the events that `tracelog()` calls emit with a
4192 _specific log level_:
4194 * Create an event rule which matches the `lttng_ust_tracelog:*`
4195 event name and a specific log level:
4200 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4201 --loglevel-only=TRACE_INFO
4206 [[prebuilt-ust-helpers]]
4207 === Prebuilt user space tracing helpers
4209 The LTTng-UST package provides a few helpers in the form or preloadable
4210 shared objects which automatically instrument system functions and
4213 The helper shared objects are normally found in dir:{/usr/lib}. If you
4214 built LTTng-UST <<building-from-source,from source>>, they are probably
4215 located in dir:{/usr/local/lib}.
4217 The installed user space tracing helpers in LTTng-UST{nbsp}{revision}
4220 path:{liblttng-ust-libc-wrapper.so}::
4221 path:{liblttng-ust-pthread-wrapper.so}::
4222 <<liblttng-ust-libc-pthread-wrapper,C{nbsp}standard library
4223 memory and POSIX threads function tracing>>.
4225 path:{liblttng-ust-cyg-profile.so}::
4226 path:{liblttng-ust-cyg-profile-fast.so}::
4227 <<liblttng-ust-cyg-profile,Function entry and exit tracing>>.
4229 path:{liblttng-ust-dl.so}::
4230 <<liblttng-ust-dl,Dynamic linker tracing>>.
4232 To use a user space tracing helper with any user application:
4234 * Preload the helper shared object when you start the application:
4239 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so my-app
4243 You can preload more than one helper:
4248 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so:liblttng-ust-dl.so my-app
4254 [[liblttng-ust-libc-pthread-wrapper]]
4255 ==== Instrument C standard library memory and POSIX threads functions
4257 The path:{liblttng-ust-libc-wrapper.so} and
4258 path:{liblttng-ust-pthread-wrapper.so} helpers
4259 add instrumentation to some C standard library and POSIX
4263 .Functions instrumented by preloading path:{liblttng-ust-libc-wrapper.so}.
4265 |TP provider name |TP name |Instrumented function
4267 .6+|`lttng_ust_libc` |`malloc` |man:malloc(3)
4268 |`calloc` |man:calloc(3)
4269 |`realloc` |man:realloc(3)
4270 |`free` |man:free(3)
4271 |`memalign` |man:memalign(3)
4272 |`posix_memalign` |man:posix_memalign(3)
4276 .Functions instrumented by preloading path:{liblttng-ust-pthread-wrapper.so}.
4278 |TP provider name |TP name |Instrumented function
4280 .4+|`lttng_ust_pthread` |`pthread_mutex_lock_req` |man:pthread_mutex_lock(3p) (request time)
4281 |`pthread_mutex_lock_acq` |man:pthread_mutex_lock(3p) (acquire time)
4282 |`pthread_mutex_trylock` |man:pthread_mutex_trylock(3p)
4283 |`pthread_mutex_unlock` |man:pthread_mutex_unlock(3p)
4286 When you preload the shared object, it replaces the functions listed
4287 in the previous tables by wrappers which contain tracepoints and call
4288 the replaced functions.
4291 [[liblttng-ust-cyg-profile]]
4292 ==== Instrument function entry and exit
4294 The path:{liblttng-ust-cyg-profile*.so} helpers can add instrumentation
4295 to the entry and exit points of functions.
4297 man:gcc(1) and man:clang(1) have an option named
4298 https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html[`-finstrument-functions`]
4299 which generates instrumentation calls for entry and exit to functions.
4300 The LTTng-UST function tracing helpers,
4301 path:{liblttng-ust-cyg-profile.so} and
4302 path:{liblttng-ust-cyg-profile-fast.so}, take advantage of this feature
4303 to add tracepoints to the two generated functions (which contain
4304 `cyg_profile` in their names, hence the helper's name).
4306 To use the LTTng-UST function tracing helper, the source files to
4307 instrument must be built using the `-finstrument-functions` compiler
4310 There are two versions of the LTTng-UST function tracing helper:
4312 * **path:{liblttng-ust-cyg-profile-fast.so}** is a lightweight variant
4313 that you should only use when it can be _guaranteed_ that the
4314 complete event stream is recorded without any lost event record.
4315 Any kind of duplicate information is left out.
4317 Assuming no event record is lost, having only the function addresses on
4318 entry is enough to create a call graph, since an event record always
4319 contains the ID of the CPU that generated it.
4321 You can use a tool like man:addr2line(1) to convert function addresses
4322 back to source file names and line numbers.
4324 * **path:{liblttng-ust-cyg-profile.so}** is a more robust variant
4325 which also works in use cases where event records might get discarded or
4326 not recorded from application startup.
4327 In these cases, the trace analyzer needs more information to be
4328 able to reconstruct the program flow.
4330 See man:lttng-ust-cyg-profile(3) to learn more about the instrumentation
4331 points of this helper.
4333 All the tracepoints that this helper provides have the
4334 log level `TRACE_DEBUG_FUNCTION` (see man:lttng-ust(3)).
4336 TIP: It's sometimes a good idea to limit the number of source files that
4337 you compile with the `-finstrument-functions` option to prevent LTTng
4338 from writing an excessive amount of trace data at run time. When using
4339 man:gcc(1), you can use the
4340 `-finstrument-functions-exclude-function-list` option to avoid
4341 instrument entries and exits of specific function names.
4346 ==== Instrument the dynamic linker
4348 The path:{liblttng-ust-dl.so} helper adds instrumentation to the
4349 man:dlopen(3) and man:dlclose(3) function calls.
4351 See man:lttng-ust-dl(3) to learn more about the instrumentation points
4356 [[java-application]]
4357 === User space Java agent
4359 You can instrument any Java application which uses one of the following
4362 * The https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[**`java.util.logging`**]
4363 (JUL) core logging facilities.
4364 * http://logging.apache.org/log4j/1.2/[**Apache log4j 1.2**], since
4365 LTTng 2.6. Note that Apache Log4j{nbsp}2 is not supported.
4368 .LTTng-UST Java agent imported by a Java application.
4369 image::java-app.png[]
4371 Note that the methods described below are new in LTTng{nbsp}{revision}.
4372 Previous LTTng versions use another technique.
4374 NOTE: We use http://openjdk.java.net/[OpenJDK]{nbsp}8 for development
4375 and https://ci.lttng.org/[continuous integration], thus this version is
4376 directly supported. However, the LTTng-UST Java agent is also tested
4377 with OpenJDK{nbsp}7.
4382 ==== Use the LTTng-UST Java agent for `java.util.logging`
4384 To use the LTTng-UST Java agent in a Java application which uses
4385 `java.util.logging` (JUL):
4387 . In the Java application's source code, import the LTTng-UST
4388 log handler package for `java.util.logging`:
4393 import org.lttng.ust.agent.jul.LttngLogHandler;
4397 . Create an LTTng-UST JUL log handler:
4402 Handler lttngUstLogHandler = new LttngLogHandler();
4406 . Add this handler to the JUL loggers which should emit LTTng events:
4411 Logger myLogger = Logger.getLogger("some-logger");
4413 myLogger.addHandler(lttngUstLogHandler);
4417 . Use `java.util.logging` log statements and configuration as usual.
4418 The loggers with an attached LTTng-UST log handler can emit
4421 . Before exiting the application, remove the LTTng-UST log handler from
4422 the loggers attached to it and call its `close()` method:
4427 myLogger.removeHandler(lttngUstLogHandler);
4428 lttngUstLogHandler.close();
4432 This is not strictly necessary, but it is recommended for a clean
4433 disposal of the handler's resources.
4435 . Include the LTTng-UST Java agent's common and JUL-specific JAR files,
4436 path:{lttng-ust-agent-common.jar} and path:{lttng-ust-agent-jul.jar},
4438 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4439 path] when you build the Java application.
4441 The JAR files are typically located in dir:{/usr/share/java}.
4443 IMPORTANT: The LTTng-UST Java agent must be
4444 <<installing-lttng,installed>> for the logging framework your
4447 .Use the LTTng-UST Java agent for `java.util.logging`.
4452 import java.io.IOException;
4453 import java.util.logging.Handler;
4454 import java.util.logging.Logger;
4455 import org.lttng.ust.agent.jul.LttngLogHandler;
4459 private static final int answer = 42;
4461 public static void main(String[] argv) throws Exception
4464 Logger logger = Logger.getLogger("jello");
4466 // Create an LTTng-UST log handler
4467 Handler lttngUstLogHandler = new LttngLogHandler();
4469 // Add the LTTng-UST log handler to our logger
4470 logger.addHandler(lttngUstLogHandler);
4473 logger.info("some info");
4474 logger.warning("some warning");
4476 logger.finer("finer information; the answer is " + answer);
4478 logger.severe("error!");
4480 // Not mandatory, but cleaner
4481 logger.removeHandler(lttngUstLogHandler);
4482 lttngUstLogHandler.close();
4491 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4494 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4495 <<enabling-disabling-events,create an event rule>> matching the
4496 `jello` JUL logger, and <<basic-tracing-session-control,start tracing>>:
4501 $ lttng enable-event --jul jello
4505 Run the compiled class:
4509 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4512 <<basic-tracing-session-control,Stop tracing>> and inspect the
4522 In the resulting trace, an <<event,event record>> generated by a Java
4523 application using `java.util.logging` is named `lttng_jul:event` and
4524 has the following fields:
4527 Log record's message.
4533 Name of the class in which the log statement was executed.
4536 Name of the method in which the log statement was executed.
4539 Logging time (timestamp in milliseconds).
4542 Log level integer value.
4545 ID of the thread in which the log statement was executed.
4547 You can use the opt:lttng-enable-event(1):--loglevel or
4548 opt:lttng-enable-event(1):--loglevel-only option of the
4549 man:lttng-enable-event(1) command to target a range of JUL log levels
4550 or a specific JUL log level.
4555 ==== Use the LTTng-UST Java agent for Apache log4j
4557 To use the LTTng-UST Java agent in a Java application which uses
4560 . In the Java application's source code, import the LTTng-UST
4561 log appender package for Apache log4j:
4566 import org.lttng.ust.agent.log4j.LttngLogAppender;
4570 . Create an LTTng-UST log4j log appender:
4575 Appender lttngUstLogAppender = new LttngLogAppender();
4579 . Add this appender to the log4j loggers which should emit LTTng events:
4584 Logger myLogger = Logger.getLogger("some-logger");
4586 myLogger.addAppender(lttngUstLogAppender);
4590 . Use Apache log4j log statements and configuration as usual. The
4591 loggers with an attached LTTng-UST log appender can emit LTTng events.
4593 . Before exiting the application, remove the LTTng-UST log appender from
4594 the loggers attached to it and call its `close()` method:
4599 myLogger.removeAppender(lttngUstLogAppender);
4600 lttngUstLogAppender.close();
4604 This is not strictly necessary, but it is recommended for a clean
4605 disposal of the appender's resources.
4607 . Include the LTTng-UST Java agent's common and log4j-specific JAR
4608 files, path:{lttng-ust-agent-common.jar} and
4609 path:{lttng-ust-agent-log4j.jar}, in the
4610 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4611 path] when you build the Java application.
4613 The JAR files are typically located in dir:{/usr/share/java}.
4615 IMPORTANT: The LTTng-UST Java agent must be
4616 <<installing-lttng,installed>> for the logging framework your
4619 .Use the LTTng-UST Java agent for Apache log4j.
4624 import org.apache.log4j.Appender;
4625 import org.apache.log4j.Logger;
4626 import org.lttng.ust.agent.log4j.LttngLogAppender;
4630 private static final int answer = 42;
4632 public static void main(String[] argv) throws Exception
4635 Logger logger = Logger.getLogger("jello");
4637 // Create an LTTng-UST log appender
4638 Appender lttngUstLogAppender = new LttngLogAppender();
4640 // Add the LTTng-UST log appender to our logger
4641 logger.addAppender(lttngUstLogAppender);
4644 logger.info("some info");
4645 logger.warn("some warning");
4647 logger.debug("debug information; the answer is " + answer);
4649 logger.fatal("error!");
4651 // Not mandatory, but cleaner
4652 logger.removeAppender(lttngUstLogAppender);
4653 lttngUstLogAppender.close();
4659 Build this example (`$LOG4JPATH` is the path to the Apache log4j JAR
4664 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH Test.java
4667 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4668 <<enabling-disabling-events,create an event rule>> matching the
4669 `jello` log4j logger, and <<basic-tracing-session-control,start tracing>>:
4674 $ lttng enable-event --log4j jello
4678 Run the compiled class:
4682 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH:. Test
4685 <<basic-tracing-session-control,Stop tracing>> and inspect the
4695 In the resulting trace, an <<event,event record>> generated by a Java
4696 application using log4j is named `lttng_log4j:event` and
4697 has the following fields:
4700 Log record's message.
4706 Name of the class in which the log statement was executed.
4709 Name of the method in which the log statement was executed.
4712 Name of the file in which the executed log statement is located.
4715 Line number at which the log statement was executed.
4721 Log level integer value.
4724 Name of the Java thread in which the log statement was executed.
4726 You can use the opt:lttng-enable-event(1):--loglevel or
4727 opt:lttng-enable-event(1):--loglevel-only option of the
4728 man:lttng-enable-event(1) command to target a range of Apache log4j log levels
4729 or a specific log4j log level.
4733 [[java-application-context]]
4734 ==== Provide application-specific context fields in a Java application
4736 A Java application-specific context field is a piece of state provided
4737 by the application which <<adding-context,you can add>>, using the
4738 man:lttng-add-context(1) command, to each <<event,event record>>
4739 produced by the log statements of this application.
4741 For example, a given object might have a current request ID variable.
4742 You can create a context information retriever for this object and
4743 assign a name to this current request ID. You can then, using the
4744 man:lttng-add-context(1) command, add this context field by name to
4745 the JUL or log4j <<channel,channel>>.
4747 To provide application-specific context fields in a Java application:
4749 . In the Java application's source code, import the LTTng-UST
4750 Java agent context classes and interfaces:
4755 import org.lttng.ust.agent.context.ContextInfoManager;
4756 import org.lttng.ust.agent.context.IContextInfoRetriever;
4760 . Create a context information retriever class, that is, a class which
4761 implements the `IContextInfoRetriever` interface:
4766 class MyContextInfoRetriever implements IContextInfoRetriever
4769 public Object retrieveContextInfo(String key)
4771 if (key.equals("intCtx")) {
4773 } else if (key.equals("strContext")) {
4774 return "context value!";
4783 This `retrieveContextInfo()` method is the only member of the
4784 `IContextInfoRetriever` interface. Its role is to return the current
4785 value of a state by name to create a context field. The names of the
4786 context fields and which state variables they return depends on your
4789 All primitive types and objects are supported as context fields.
4790 When `retrieveContextInfo()` returns an object, the context field
4791 serializer calls its `toString()` method to add a string field to
4792 event records. The method can also return `null`, which means that
4793 no context field is available for the required name.
4795 . Register an instance of your context information retriever class to
4796 the context information manager singleton:
4801 IContextInfoRetriever cir = new MyContextInfoRetriever();
4802 ContextInfoManager cim = ContextInfoManager.getInstance();
4803 cim.registerContextInfoRetriever("retrieverName", cir);
4807 . Before exiting the application, remove your context information
4808 retriever from the context information manager singleton:
4813 ContextInfoManager cim = ContextInfoManager.getInstance();
4814 cim.unregisterContextInfoRetriever("retrieverName");
4818 This is not strictly necessary, but it is recommended for a clean
4819 disposal of some manager's resources.
4821 . Build your Java application with LTTng-UST Java agent support as
4822 usual, following the procedure for either the <<jul,JUL>> or
4823 <<log4j,Apache log4j>> framework.
4826 .Provide application-specific context fields in a Java application.
4831 import java.util.logging.Handler;
4832 import java.util.logging.Logger;
4833 import org.lttng.ust.agent.jul.LttngLogHandler;
4834 import org.lttng.ust.agent.context.ContextInfoManager;
4835 import org.lttng.ust.agent.context.IContextInfoRetriever;
4839 // Our context information retriever class
4840 private static class MyContextInfoRetriever
4841 implements IContextInfoRetriever
4844 public Object retrieveContextInfo(String key) {
4845 if (key.equals("intCtx")) {
4847 } else if (key.equals("strContext")) {
4848 return "context value!";
4855 private static final int answer = 42;
4857 public static void main(String args[]) throws Exception
4859 // Get the context information manager instance
4860 ContextInfoManager cim = ContextInfoManager.getInstance();
4862 // Create and register our context information retriever
4863 IContextInfoRetriever cir = new MyContextInfoRetriever();
4864 cim.registerContextInfoRetriever("myRetriever", cir);
4867 Logger logger = Logger.getLogger("jello");
4869 // Create an LTTng-UST log handler
4870 Handler lttngUstLogHandler = new LttngLogHandler();
4872 // Add the LTTng-UST log handler to our logger
4873 logger.addHandler(lttngUstLogHandler);
4876 logger.info("some info");
4877 logger.warning("some warning");
4879 logger.finer("finer information; the answer is " + answer);
4881 logger.severe("error!");
4883 // Not mandatory, but cleaner
4884 logger.removeHandler(lttngUstLogHandler);
4885 lttngUstLogHandler.close();
4886 cim.unregisterContextInfoRetriever("myRetriever");
4895 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4898 <<creating-destroying-tracing-sessions,Create a tracing session>>
4899 and <<enabling-disabling-events,create an event rule>> matching the
4905 $ lttng enable-event --jul jello
4908 <<adding-context,Add the application-specific context fields>> to the
4913 $ lttng add-context --jul --type='$app.myRetriever:intCtx'
4914 $ lttng add-context --jul --type='$app.myRetriever:strContext'
4917 <<basic-tracing-session-control,Start tracing>>:
4924 Run the compiled class:
4928 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4931 <<basic-tracing-session-control,Stop tracing>> and inspect the
4943 [[python-application]]
4944 === User space Python agent
4946 You can instrument a Python 2 or Python 3 application which uses the
4947 standard https://docs.python.org/3/library/logging.html[`logging`]
4950 Each log statement emits an LTTng event once the
4951 application module imports the
4952 <<lttng-ust-agents,LTTng-UST Python agent>> package.
4955 .A Python application importing the LTTng-UST Python agent.
4956 image::python-app.png[]
4958 To use the LTTng-UST Python agent:
4960 . In the Python application's source code, import the LTTng-UST Python
4970 The LTTng-UST Python agent automatically adds its logging handler to the
4971 root logger at import time.
4973 Any log statement that the application executes before this import does
4974 not emit an LTTng event.
4976 IMPORTANT: The LTTng-UST Python agent must be
4977 <<installing-lttng,installed>>.
4979 . Use log statements and logging configuration as usual.
4980 Since the LTTng-UST Python agent adds a handler to the _root_
4981 logger, you can trace any log statement from any logger.
4983 .Use the LTTng-UST Python agent.
4994 logging.basicConfig()
4995 logger = logging.getLogger('my-logger')
4998 logger.debug('debug message')
4999 logger.info('info message')
5000 logger.warn('warn message')
5001 logger.error('error message')
5002 logger.critical('critical message')
5006 if __name__ == '__main__':
5010 NOTE: `logging.basicConfig()`, which adds to the root logger a basic
5011 logging handler which prints to the standard error stream, is not
5012 strictly required for LTTng-UST tracing to work, but in versions of
5013 Python preceding 3.2, you could see a warning message which indicates
5014 that no handler exists for the logger `my-logger`.
5016 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5017 <<enabling-disabling-events,create an event rule>> matching the
5018 `my-logger` Python logger, and <<basic-tracing-session-control,start
5024 $ lttng enable-event --python my-logger
5028 Run the Python script:
5035 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5045 In the resulting trace, an <<event,event record>> generated by a Python
5046 application is named `lttng_python:event` and has the following fields:
5049 Logging time (string).
5052 Log record's message.
5058 Name of the function in which the log statement was executed.
5061 Line number at which the log statement was executed.
5064 Log level integer value.
5067 ID of the Python thread in which the log statement was executed.
5070 Name of the Python thread in which the log statement was executed.
5072 You can use the opt:lttng-enable-event(1):--loglevel or
5073 opt:lttng-enable-event(1):--loglevel-only option of the
5074 man:lttng-enable-event(1) command to target a range of Python log levels
5075 or a specific Python log level.
5077 When an application imports the LTTng-UST Python agent, the agent tries
5078 to register to a <<lttng-sessiond,session daemon>>. Note that you must
5079 <<start-sessiond,start the session daemon>> _before_ you run the Python
5080 application. If a session daemon is found, the agent tries to register
5081 to it during 5{nbsp}seconds, after which the application continues
5082 without LTTng tracing support. You can override this timeout value with
5083 the env:LTTNG_UST_PYTHON_REGISTER_TIMEOUT environment variable
5086 If the session daemon stops while a Python application with an imported
5087 LTTng-UST Python agent runs, the agent retries to connect and to
5088 register to a session daemon every 3{nbsp}seconds. You can override this
5089 delay with the env:LTTNG_UST_PYTHON_REGISTER_RETRY_DELAY environment
5094 [[proc-lttng-logger-abi]]
5097 The `lttng-tracer` Linux kernel module, part of
5098 <<lttng-modules,LTTng-modules>>, creates the special LTTng logger file
5099 path:{/proc/lttng-logger} when it's loaded. Any application can write
5100 text data to this file to emit an LTTng event.
5103 .An application writes to the LTTng logger file to emit an LTTng event.
5104 image::lttng-logger.png[]
5106 The LTTng logger is the quickest method--not the most efficient,
5107 however--to add instrumentation to an application. It is designed
5108 mostly to instrument shell scripts:
5112 $ echo "Some message, some $variable" > /proc/lttng-logger
5115 Any event that the LTTng logger emits is named `lttng_logger` and
5116 belongs to the Linux kernel <<domain,tracing domain>>. However, unlike
5117 other instrumentation points in the kernel tracing domain, **any Unix
5118 user** can <<enabling-disabling-events,create an event rule>> which
5119 matches its event name, not only the root user or users in the
5120 <<tracing-group,tracing group>>.
5122 To use the LTTng logger:
5124 * From any application, write text data to the path:{/proc/lttng-logger}
5127 The `msg` field of `lttng_logger` event records contains the
5130 NOTE: The maximum message length of an LTTng logger event is
5131 1024{nbsp}bytes. Writing more than this makes the LTTng logger emit more
5132 than one event to contain the remaining data.
5134 You should not use the LTTng logger to trace a user application which
5135 can be instrumented in a more efficient way, namely:
5137 * <<c-application,C and $$C++$$ applications>>.
5138 * <<java-application,Java applications>>.
5139 * <<python-application,Python applications>>.
5141 .Use the LTTng logger.
5146 echo 'Hello, World!' > /proc/lttng-logger
5148 df --human-readable --print-type / > /proc/lttng-logger
5151 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5152 <<enabling-disabling-events,create an event rule>> matching the
5153 `lttng_logger` Linux kernel tracepoint, and
5154 <<basic-tracing-session-control,start tracing>>:
5159 $ lttng enable-event --kernel lttng_logger
5163 Run the Bash script:
5170 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5181 [[instrumenting-linux-kernel]]
5182 === LTTng kernel tracepoints
5184 NOTE: This section shows how to _add_ instrumentation points to the
5185 Linux kernel. The kernel's subsystems are already thoroughly
5186 instrumented at strategic places for LTTng when you
5187 <<installing-lttng,install>> the <<lttng-modules,LTTng-modules>>
5191 There are two methods to instrument the Linux kernel:
5193 . <<linux-add-lttng-layer,Add an LTTng layer>> over an existing ftrace
5194 tracepoint which uses the `TRACE_EVENT()` API.
5196 Choose this if you want to instrumentation a Linux kernel tree with an
5197 instrumentation point compatible with ftrace, perf, and SystemTap.
5199 . Use an <<linux-lttng-tracepoint-event,LTTng-only approach>> to
5200 instrument an out-of-tree kernel module.
5202 Choose this if you don't need ftrace, perf, or SystemTap support.
5206 [[linux-add-lttng-layer]]
5207 ==== [[instrumenting-linux-kernel-itself]][[mainline-trace-event]][[lttng-adaptation-layer]]Add an LTTng layer to an existing ftrace tracepoint
5209 This section shows how to add an LTTng layer to existing ftrace
5210 instrumentation using the `TRACE_EVENT()` API.
5212 This section does not document the `TRACE_EVENT()` macro. You can
5213 read the following articles to learn more about this API:
5215 * http://lwn.net/Articles/379903/[Using the TRACE_EVENT() macro (Part 1)]
5216 * http://lwn.net/Articles/381064/[Using the TRACE_EVENT() macro (Part 2)]
5217 * http://lwn.net/Articles/383362/[Using the TRACE_EVENT() macro (Part 3)]
5219 The following procedure assumes that your ftrace tracepoints are
5220 correctly defined in their own header and that they are created in
5221 one source file using the `CREATE_TRACE_POINTS` definition.
5223 To add an LTTng layer over an existing ftrace tracepoint:
5225 . Make sure the following kernel configuration options are
5231 * `CONFIG_HIGH_RES_TIMERS`
5232 * `CONFIG_TRACEPOINTS`
5235 . Build the Linux source tree with your custom ftrace tracepoints.
5236 . Boot the resulting Linux image on your target system.
5238 Confirm that the tracepoints exist by looking for their names in the
5239 dir:{/sys/kernel/debug/tracing/events/subsys} directory, where `subsys`
5240 is your subsystem's name.
5242 . Get a copy of the latest LTTng-modules{nbsp}{revision}:
5247 $ cd $(mktemp -d) &&
5248 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.8.tar.bz2 &&
5249 tar -xf lttng-modules-latest-2.8.tar.bz2 &&
5250 cd lttng-modules-2.8.*
5254 . In dir:{instrumentation/events/lttng-module}, relative to the root
5255 of the LTTng-modules source tree, create a header file named
5256 +__subsys__.h+ for your custom subsystem +__subsys__+ and write your
5257 LTTng-modules tracepoint definitions using the LTTng-modules
5260 Start with this template:
5264 .path:{instrumentation/events/lttng-module/my_subsys.h}
5267 #define TRACE_SYSTEM my_subsys
5269 #if !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ)
5270 #define _LTTNG_MY_SUBSYS_H
5272 #include "../../../probes/lttng-tracepoint-event.h"
5273 #include <linux/tracepoint.h>
5275 LTTNG_TRACEPOINT_EVENT(
5277 * Format is identical to TRACE_EVENT()'s version for the three
5278 * following macro parameters:
5281 TP_PROTO(int my_int, const char *my_string),
5282 TP_ARGS(my_int, my_string),
5284 /* LTTng-modules specific macros */
5286 ctf_integer(int, my_int_field, my_int)
5287 ctf_string(my_bar_field, my_bar)
5291 #endif /* !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ) */
5293 #include "../../../probes/define_trace.h"
5297 The entries in the `TP_FIELDS()` section are the list of fields for the
5298 LTTng tracepoint. This is similar to the `TP_STRUCT__entry()` part of
5299 ftrace's `TRACE_EVENT()` macro.
5301 See <<lttng-modules-tp-fields,Tracepoint fields macros>> for a
5302 complete description of the available `ctf_*()` macros.
5304 . Create the LTTng-modules probe's kernel module C source file,
5305 +probes/lttng-probe-__subsys__.c+, where +__subsys__+ is your
5310 .path:{probes/lttng-probe-my-subsys.c}
5312 #include <linux/module.h>
5313 #include "../lttng-tracer.h"
5316 * Build-time verification of mismatch between mainline
5317 * TRACE_EVENT() arguments and the LTTng-modules adaptation
5318 * layer LTTNG_TRACEPOINT_EVENT() arguments.
5320 #include <trace/events/my_subsys.h>
5322 /* Create LTTng tracepoint probes */
5323 #define LTTNG_PACKAGE_BUILD
5324 #define CREATE_TRACE_POINTS
5325 #define TRACE_INCLUDE_PATH ../instrumentation/events/lttng-module
5327 #include "../instrumentation/events/lttng-module/my_subsys.h"
5329 MODULE_LICENSE("GPL and additional rights");
5330 MODULE_AUTHOR("Your name <your-email>");
5331 MODULE_DESCRIPTION("LTTng my_subsys probes");
5332 MODULE_VERSION(__stringify(LTTNG_MODULES_MAJOR_VERSION) "."
5333 __stringify(LTTNG_MODULES_MINOR_VERSION) "."
5334 __stringify(LTTNG_MODULES_PATCHLEVEL_VERSION)
5335 LTTNG_MODULES_EXTRAVERSION);
5339 . Edit path:{probes/KBuild} and add your new kernel module object
5340 next to the existing ones:
5344 .path:{probes/KBuild}
5348 obj-m += lttng-probe-module.o
5349 obj-m += lttng-probe-power.o
5351 obj-m += lttng-probe-my-subsys.o
5357 . Build and install the LTTng kernel modules:
5362 $ make KERNELDIR=/path/to/linux
5363 # make modules_install && depmod -a
5367 Replace `/path/to/linux` with the path to the Linux source tree where
5368 you defined and used tracepoints with ftrace's `TRACE_EVENT()` macro.
5370 Note that you can also use the
5371 <<lttng-tracepoint-event-code,`LTTNG_TRACEPOINT_EVENT_CODE()` macro>>
5372 instead of `LTTNG_TRACEPOINT_EVENT()` to use custom local variables and
5373 C code that need to be executed before the event fields are recorded.
5375 The best way to learn how to use the previous LTTng-modules macros is to
5376 inspect the existing LTTng-modules tracepoint definitions in the
5377 dir:{instrumentation/events/lttng-module} header files. Compare them
5378 with the Linux kernel mainline versions in the
5379 dir:{include/trace/events} directory of the Linux source tree.
5383 [[lttng-tracepoint-event-code]]
5384 ===== Use custom C code to access the data for tracepoint fields
5386 Although we recommended to always use the
5387 <<lttng-adaptation-layer,`LTTNG_TRACEPOINT_EVENT()`>> macro to describe
5388 the arguments and fields of an LTTng-modules tracepoint when possible,
5389 sometimes you need a more complex process to access the data that the
5390 tracer records as event record fields. In other words, you need local
5391 variables and multiple C{nbsp}statements instead of simple
5392 argument-based expressions that you pass to the
5393 <<lttng-modules-tp-fields,`ctf_*()` macros of `TP_FIELDS()`>>.
5395 You can use the `LTTNG_TRACEPOINT_EVENT_CODE()` macro instead of
5396 `LTTNG_TRACEPOINT_EVENT()` to declare custom local variables and define
5397 a block of C{nbsp}code to be executed before LTTng records the fields.
5398 The structure of this macro is:
5401 .`LTTNG_TRACEPOINT_EVENT_CODE()` macro syntax.
5403 LTTNG_TRACEPOINT_EVENT_CODE(
5405 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5406 * version for the following three macro parameters:
5409 TP_PROTO(int my_int, const char *my_string),
5410 TP_ARGS(my_int, my_string),
5412 /* Declarations of custom local variables */
5415 unsigned long b = 0;
5416 const char *name = "(undefined)";
5417 struct my_struct *my_struct;
5421 * Custom code which uses both tracepoint arguments
5422 * (in TP_ARGS()) and local variables (in TP_locvar()).
5424 * Local variables are actually members of a structure pointed
5425 * to by the special variable tp_locvar.
5429 tp_locvar->a = my_int + 17;
5430 tp_locvar->my_struct = get_my_struct_at(tp_locvar->a);
5431 tp_locvar->b = my_struct_compute_b(tp_locvar->my_struct);
5432 tp_locvar->name = my_struct_get_name(tp_locvar->my_struct);
5433 put_my_struct(tp_locvar->my_struct);
5442 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5443 * version for this, except that tp_locvar members can be
5444 * used in the argument expression parameters of
5445 * the ctf_*() macros.
5448 ctf_integer(unsigned long, my_struct_b, tp_locvar->b)
5449 ctf_integer(int, my_struct_a, tp_locvar->a)
5450 ctf_string(my_string_field, my_string)
5451 ctf_string(my_struct_name, tp_locvar->name)
5456 IMPORTANT: The C code defined in `TP_code()` must not have any side
5457 effects when executed. In particular, the code must not allocate
5458 memory or get resources without deallocating this memory or putting
5459 those resources afterwards.
5462 [[instrumenting-linux-kernel-tracing]]
5463 ==== Load and unload a custom probe kernel module
5465 You must load a <<lttng-adaptation-layer,created LTTng-modules probe
5466 kernel module>> in the kernel before it can emit LTTng events.
5468 To load the default probe kernel modules and a custom probe kernel
5471 * Use the opt:lttng-sessiond(8):--extra-kmod-probes option to give extra
5472 probe modules to load when starting a root <<lttng-sessiond,session
5476 .Load the `my_subsys`, `usb`, and the default probe modules.
5480 # lttng-sessiond --extra-kmod-probes=my_subsys,usb
5485 You only need to pass the subsystem name, not the whole kernel module
5488 To load _only_ a given custom probe kernel module:
5490 * Use the opt:lttng-sessiond(8):--kmod-probes option to give the probe
5491 modules to load when starting a root session daemon:
5494 .Load only the `my_subsys` and `usb` probe modules.
5498 # lttng-sessiond --kmod-probes=my_subsys,usb
5503 To confirm that a probe module is loaded:
5510 $ lsmod | grep lttng_probe_usb
5514 To unload the loaded probe modules:
5516 * Kill the session daemon with `SIGTERM`:
5521 # pkill lttng-sessiond
5525 You can also use man:modprobe(8)'s `--remove` option if the session
5526 daemon terminates abnormally.
5529 [[controlling-tracing]]
5532 Once an application or a Linux kernel is
5533 <<instrumenting,instrumented>> for LTTng tracing,
5536 This section is divided in topics on how to use the various
5537 <<plumbing,components of LTTng>>, in particular the <<lttng-cli,cmd:lttng
5538 command-line tool>>, to _control_ the LTTng daemons and tracers.
5540 NOTE: In the following subsections, we refer to an man:lttng(1) command
5541 using its man page name. For example, instead of _Run the `create`
5542 command to..._, we use _Run the man:lttng-create(1) command to..._.
5546 === Start a session daemon
5548 In some situations, you need to run a <<lttng-sessiond,session daemon>>
5549 (man:lttng-sessiond(8)) _before_ you can use the man:lttng(1)
5552 You will see the following error when you run a command while no session
5556 Error: No session daemon is available
5559 The only command that automatically runs a session daemon is
5560 man:lttng-create(1), which you use to
5561 <<creating-destroying-tracing-sessions,create a tracing session>>. While
5562 this is most of the time the first operation that you do, sometimes it's
5563 not. Some examples are:
5565 * <<list-instrumentation-points,List the available instrumentation points>>.
5566 * <<saving-loading-tracing-session,Load a tracing session configuration>>.
5568 [[tracing-group]] Each Unix user must have its own running session
5569 daemon to trace user applications. The session daemon that the root user
5570 starts is the only one allowed to control the LTTng kernel tracer. Users
5571 that are part of the _tracing group_ can control the root session
5572 daemon. The default tracing group name is `tracing`; you can set it to
5573 something else with the opt:lttng-sessiond(8):--group option when you
5574 start the root session daemon.
5576 To start a user session daemon:
5578 * Run man:lttng-sessiond(8):
5583 $ lttng-sessiond --daemonize
5587 To start the root session daemon:
5589 * Run man:lttng-sessiond(8) as the root user:
5594 # lttng-sessiond --daemonize
5598 In both cases, remove the opt:lttng-sessiond(8):--daemonize option to
5599 start the session daemon in foreground.
5601 To stop a session daemon, use man:kill(1) on its process ID (standard
5604 Note that some Linux distributions could manage the LTTng session daemon
5605 as a service. In this case, you should use the service manager to
5606 start, restart, and stop session daemons.
5609 [[creating-destroying-tracing-sessions]]
5610 === Create and destroy a tracing session
5612 Almost all the LTTng control operations happen in the scope of
5613 a <<tracing-session,tracing session>>, which is the dialogue between the
5614 <<lttng-sessiond,session daemon>> and you.
5616 To create a tracing session with a generated name:
5618 * Use the man:lttng-create(1) command:
5627 The created tracing session's name is `auto` followed by the
5630 To create a tracing session with a specific name:
5632 * Use the optional argument of the man:lttng-create(1) command:
5637 $ lttng create my-session
5641 Replace `my-session` with the specific tracing session name.
5643 LTTng appends the creation date to the created tracing session's name.
5645 LTTng writes the traces of a tracing session in
5646 +$LTTNG_HOME/lttng-trace/__name__+ by default, where +__name__+ is the
5647 name of the tracing session. Note that the env:LTTNG_HOME environment
5648 variable defaults to `$HOME` if not set.
5650 To output LTTng traces to a non-default location:
5652 * Use the opt:lttng-create(1):--output option of the man:lttng-create(1) command:
5657 $ lttng create my-session --output=/tmp/some-directory
5661 You may create as many tracing sessions as you wish.
5663 To list all the existing tracing sessions for your Unix user:
5665 * Use the man:lttng-list(1) command:
5674 When you create a tracing session, it is set as the _current tracing
5675 session_. The following man:lttng(1) commands operate on the current
5676 tracing session when you don't specify one:
5678 [role="list-3-cols"]
5694 To change the current tracing session:
5696 * Use the man:lttng-set-session(1) command:
5701 $ lttng set-session new-session
5705 Replace `new-session` by the name of the new current tracing session.
5707 When you are done tracing in a given tracing session, you can destroy
5708 it. This operation frees the resources taken by the tracing session
5709 to destroy; it does not destroy the trace data that LTTng wrote for
5710 this tracing session.
5712 To destroy the current tracing session:
5714 * Use the man:lttng-destroy(1) command:
5724 [[list-instrumentation-points]]
5725 === List the available instrumentation points
5727 The <<lttng-sessiond,session daemon>> can query the running instrumented
5728 user applications and the Linux kernel to get a list of available
5729 instrumentation points. For the Linux kernel <<domain,tracing domain>>,
5730 they are tracepoints and system calls. For the user space tracing
5731 domain, they are tracepoints. For the other tracing domains, they are
5734 To list the available instrumentation points:
5736 * Use the man:lttng-list(1) command with the requested tracing domain's
5740 * opt:lttng-list(1):--kernel: Linux kernel tracepoints (your Unix user
5741 must be a root user, or it must be a member of the
5742 <<tracing-group,tracing group>>).
5743 * opt:lttng-list(1):--kernel with opt:lttng-list(1):--syscall: Linux
5744 kernel system calls (your Unix user must be a root user, or it must be
5745 a member of the tracing group).
5746 * opt:lttng-list(1):--userspace: user space tracepoints.
5747 * opt:lttng-list(1):--jul: `java.util.logging` loggers.
5748 * opt:lttng-list(1):--log4j: Apache log4j loggers.
5749 * opt:lttng-list(1):--python: Python loggers.
5752 .List the available user space tracepoints.
5756 $ lttng list --userspace
5760 .List the available Linux kernel system call tracepoints.
5764 $ lttng list --kernel --syscall
5769 [[enabling-disabling-events]]
5770 === Create and enable an event rule
5772 Once you <<creating-destroying-tracing-sessions,create a tracing
5773 session>>, you can create <<event,event rules>> with the
5774 man:lttng-enable-event(1) command.
5776 You specify each condition with a command-line option. The available
5777 condition options are shown in the following table.
5779 [role="growable",cols="asciidoc,asciidoc,default"]
5780 .Condition command-line options for the man:lttng-enable-event(1) command.
5782 |Option |Description |Applicable tracing domains
5788 . +--probe=__ADDR__+
5789 . +--function=__ADDR__+
5792 Instead of using the default _tracepoint_ instrumentation type, use:
5794 . A Linux system call.
5795 . A Linux https://lwn.net/Articles/132196/[KProbe] (symbol or address).
5796 . The entry and return points of a Linux function (symbol or address).
5800 |First positional argument.
5803 Tracepoint or system call name. In the case of a Linux KProbe or
5804 function, this is a custom name given to the event rule. With the
5805 JUL, log4j, and Python domains, this is a logger name.
5807 With a tracepoint, logger, or system call name, the last character
5808 can be `*` to match anything that remains.
5815 . +--loglevel=__LEVEL__+
5816 . +--loglevel-only=__LEVEL__+
5819 . Match only tracepoints or log statements with a logging level at
5820 least as severe as +__LEVEL__+.
5821 . Match only tracepoints or log statements with a logging level
5822 equal to +__LEVEL__+.
5824 See man:lttng-enable-event(1) for the list of available logging level
5827 |User space, JUL, log4j, and Python.
5829 |+--exclude=__EXCLUSIONS__+
5832 When you use a `*` character at the end of the tracepoint or logger
5833 name (first positional argument), exclude the specific names in the
5834 comma-delimited list +__EXCLUSIONS__+.
5837 User space, JUL, log4j, and Python.
5839 |+--filter=__EXPR__+
5842 Match only events which satisfy the expression +__EXPR__+.
5844 See man:lttng-enable-event(1) to learn more about the syntax of a
5851 You attach an event rule to a <<channel,channel>> on creation. If you do
5852 not specify the channel with the opt:lttng-enable-event(1):--channel
5853 option, and if the event rule to create is the first in its
5854 <<domain,tracing domain>> for a given tracing session, then LTTng
5855 creates a _default channel_ for you. This default channel is reused in
5856 subsequent invocations of the man:lttng-enable-event(1) command for the
5857 same tracing domain.
5859 An event rule is always enabled at creation time.
5861 The following examples show how you can combine the previous
5862 command-line options to create simple to more complex event rules.
5864 .Create an event rule targetting a Linux kernel tracepoint (default channel).
5868 $ lttng enable-event --kernel sched_switch
5872 .Create an event rule matching four Linux kernel system calls (default channel).
5876 $ lttng enable-event --kernel --syscall open,write,read,close
5880 .Create event rules matching tracepoints with filter expressions (default channel).
5884 $ lttng enable-event --kernel sched_switch --filter='prev_comm == "bash"'
5889 $ lttng enable-event --kernel --all \
5890 --filter='$ctx.tid == 1988 || $ctx.tid == 1534'
5895 $ lttng enable-event --jul my_logger \
5896 --filter='$app.retriever:cur_msg_id > 3'
5899 IMPORTANT: Make sure to always quote the filter string when you
5900 use man:lttng(1) from a shell.
5903 .Create an event rule matching any user space tracepoint of a given tracepoint provider with a log level range (default channel).
5907 $ lttng enable-event --userspace my_app:'*' --loglevel=TRACE_INFO
5910 IMPORTANT: Make sure to always quote the wildcard character when you
5911 use man:lttng(1) from a shell.
5914 .Create an event rule matching multiple Python loggers with a wildcard and with exclusions (default channel).
5918 $ lttng enable-event --python my-app.'*' \
5919 --exclude='my-app.module,my-app.hello'
5923 .Create an event rule matching any Apache log4j logger with a specific log level (default channel).
5927 $ lttng enable-event --log4j --all --loglevel-only=LOG4J_WARN
5931 .Create an event rule attached to a specific channel matching a specific user space tracepoint provider and tracepoint.
5935 $ lttng enable-event --userspace my_app:my_tracepoint --channel=my-channel
5939 The event rules of a given channel form a whitelist: as soon as an
5940 emitted event passes one of them, LTTng can record the event. For
5941 example, an event named `my_app:my_tracepoint` emitted from a user space
5942 tracepoint with a `TRACE_ERROR` log level passes both of the following
5947 $ lttng enable-event --userspace my_app:my_tracepoint
5948 $ lttng enable-event --userspace my_app:my_tracepoint \
5949 --loglevel=TRACE_INFO
5952 The second event rule is redundant: the first one includes
5956 [[disable-event-rule]]
5957 === Disable an event rule
5959 To disable an event rule that you <<enabling-disabling-events,created>>
5960 previously, use the man:lttng-disable-event(1) command. This command
5961 disables _all_ the event rules (of a given tracing domain and channel)
5962 which match an instrumentation point. The other conditions are not
5963 supported as of LTTng{nbsp}{revision}.
5965 The LTTng tracer does not record an emitted event which passes
5966 a _disabled_ event rule.
5968 .Disable an event rule matching a Python logger (default channel).
5972 $ lttng disable-event --python my-logger
5976 .Disable an event rule matching all `java.util.logging` loggers (default channel).
5980 $ lttng disable-event --jul '*'
5984 .Disable _all_ the event rules of the default channel.
5986 The opt:lttng-disable-event(1):--all-events option is not, like the
5987 opt:lttng-enable-event(1):--all option of man:lttng-enable-event(1), the
5988 equivalent of the event name `*` (wildcard): it disables _all_ the event
5989 rules of a given channel.
5993 $ lttng disable-event --jul --all-events
5997 NOTE: You cannot delete an event rule once you create it.
6001 === Get the status of a tracing session
6003 To get the status of the current tracing session, that is, its
6004 parameters, its channels, event rules, and their attributes:
6006 * Use the man:lttng-status(1) command:
6016 To get the status of any tracing session:
6018 * Use the man:lttng-list(1) command with the tracing session's name:
6023 $ lttng list my-session
6027 Replace `my-session` with the desired tracing session's name.
6030 [[basic-tracing-session-control]]
6031 === Start and stop a tracing session
6033 Once you <<creating-destroying-tracing-sessions,create a tracing
6035 <<enabling-disabling-events,create one or more event rules>>,
6036 you can start and stop the tracers for this tracing session.
6038 To start tracing in the current tracing session:
6040 * Use the man:lttng-start(1) command:
6049 LTTng is very flexible: you can launch user applications before
6050 or after the you start the tracers. The tracers only record the events
6051 if they pass enabled event rules and if they occur while the tracers are
6054 To stop tracing in the current tracing session:
6056 * Use the man:lttng-stop(1) command:
6065 If there were <<channel-overwrite-mode-vs-discard-mode,lost event
6066 records>> or lost sub-buffers since the last time you ran
6067 man:lttng-start(1), warnings are printed when you run the
6068 man:lttng-stop(1) command.
6071 [[enabling-disabling-channels]]
6072 === Create a channel
6074 Once you create a tracing session, you can create a <<channel,channel>>
6075 with the man:lttng-enable-channel(1) command.
6077 Note that LTTng automatically creates a default channel when, for a
6078 given <<domain,tracing domain>>, no channels exist and you
6079 <<enabling-disabling-events,create>> the first event rule. This default
6080 channel is named `channel0` and its attributes are set to reasonable
6081 values. Therefore, you only need to create a channel when you need
6082 non-default attributes.
6084 You specify each non-default channel attribute with a command-line
6085 option when you use the man:lttng-enable-channel(1) command. The
6086 available command-line options are:
6088 [role="growable",cols="asciidoc,asciidoc"]
6089 .Command-line options for the man:lttng-enable-channel(1) command.
6091 |Option |Description
6097 <<channel-overwrite-mode-vs-discard-mode,event loss mode>> instead of
6098 the default _discard_ mode.
6100 |`--buffers-pid` (user space tracing domain only)
6103 Use the per-process <<channel-buffering-schemes,buffering scheme>>
6104 instead of the default per-user buffering scheme.
6106 |+--subbuf-size=__SIZE__+
6109 Allocate sub-buffers of +__SIZE__+ bytes (power of two), for each CPU,
6110 either for each Unix user (default), or for each instrumented process.
6112 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6114 |+--num-subbuf=__COUNT__+
6117 Allocate +__COUNT__+ sub-buffers (power of two), for each CPU, either
6118 for each Unix user (default), or for each instrumented process.
6120 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6122 |+--tracefile-size=__SIZE__+
6125 Set the maximum size of each trace file that this channel writes within
6126 a stream to +__SIZE__+ bytes instead of no maximum.
6128 See <<tracefile-rotation,Trace file count and size>>.
6130 |+--tracefile-count=__COUNT__+
6133 Limit the number of trace files that this channel creates to
6134 +__COUNT__+ channels instead of no limit.
6136 See <<tracefile-rotation,Trace file count and size>>.
6138 |+--switch-timer=__PERIODUS__+
6141 Set the <<channel-switch-timer,switch timer period>>
6142 to +__PERIODUS__+{nbsp}µs.
6144 |+--read-timer=__PERIODUS__+
6147 Set the <<channel-read-timer,read timer period>>
6148 to +__PERIODUS__+{nbsp}µs.
6150 |+--output=__TYPE__+ (Linux kernel tracing domain only)
6153 Set the channel's output type to +__TYPE__+, either `mmap` or `splice`.
6157 You can only create a channel in the Linux kernel and user space
6158 <<domain,tracing domains>>: other tracing domains have their own channel
6159 created on the fly when <<enabling-disabling-events,creating event
6164 Because of a current LTTng limitation, you must create all channels
6165 _before_ you <<basic-tracing-session-control,start tracing>> in a given
6166 tracing session, that is, before the first time you run
6169 Since LTTng automatically creates a default channel when you use the
6170 man:lttng-enable-event(1) command with a specific tracing domain, you
6171 cannot, for example, create a Linux kernel event rule, start tracing,
6172 and then create a user space event rule, because no user space channel
6173 exists yet and it's too late to create one.
6175 For this reason, make sure to configure your channels properly
6176 before starting the tracers for the first time!
6179 The following examples show how you can combine the previous
6180 command-line options to create simple to more complex channels.
6182 .Create a Linux kernel channel with default attributes.
6186 $ lttng enable-channel --kernel my-channel
6190 .Create a user space channel with 4 sub-buffers or 1{nbsp}MiB each, per CPU, per instrumented process.
6194 $ lttng enable-channel --userspace --num-subbuf=4 --subbuf-size=1M \
6195 --buffers-pid my-channel
6199 .Create a Linux kernel channel which rotates 8 trace files of 4{nbsp}MiB each for each stream
6203 $ lttng enable-channel --kernel --tracefile-count=8 \
6204 --tracefile-size=4194304 my-channel
6208 .Create a user space channel in overwrite (or _flight recorder_) mode.
6212 $ lttng enable-channel --userspace --overwrite my-channel
6216 You can <<enabling-disabling-events,create>> the same event rule in
6217 two different channels:
6221 $ lttng enable-event --userspace --channel=my-channel app:tp
6222 $ lttng enable-event --userspace --channel=other-channel app:tp
6225 If both channels are enabled, when a tracepoint named `app:tp` is
6226 reached, LTTng records two events, one for each channel.
6230 === Disable a channel
6232 To disable a specific channel that you <<enabling-disabling-channels,created>>
6233 previously, use the man:lttng-disable-channel(1) command.
6235 .Disable a specific Linux kernel channel.
6239 $ lttng disable-channel --kernel my-channel
6243 The state of a channel precedes the individual states of event rules
6244 attached to it: event rules which belong to a disabled channel, even if
6245 they are enabled, are also considered disabled.
6249 === Add context fields to a channel
6251 Event record fields in trace files provide important information about
6252 events that occured previously, but sometimes some external context may
6253 help you solve a problem faster. Examples of context fields are:
6255 * The **process ID**, **thread ID**, **process name**, and
6256 **process priority** of the thread in which the event occurs.
6257 * The **hostname** of the system on which the event occurs.
6258 * The current values of many possible **performance counters** using
6260 ** CPU cycles, stalled cycles, idle cycles, and the other cycle types.
6262 ** Branch instructions, misses, and loads.
6264 * Any context defined at the application level (supported for the
6265 JUL and log4j <<domain,tracing domains>>).
6267 To get the full list of available context fields, see
6268 `lttng add-context --list`. Some context fields are reserved for a
6269 specific <<domain,tracing domain>> (Linux kernel or user space).
6271 You add context fields to <<channel,channels>>. All the events
6272 that a channel with added context fields records contain those fields.
6274 To add context fields to one or all the channels of a given tracing
6277 * Use the man:lttng-add-context(1) command.
6279 .Add context fields to all the channels of the current tracing session.
6281 The following command line adds the virtual process identifier and
6282 the per-thread CPU cycles count fields to all the user space channels
6283 of the current tracing session.
6287 $ lttng add-context --userspace --type=vpid --type=perf:thread:cpu-cycles
6291 .Add a context field to a specific channel.
6293 The following command line adds the thread identifier context field
6294 to the Linux kernel channel named `my-channel` in the current
6299 $ lttng add-context --kernel --channel=my-channel --type=tid
6303 .Add an application-specific context field to a specific channel.
6305 The following command line adds the `cur_msg_id` context field of the
6306 `retriever` context retriever for all the instrumented
6307 <<java-application,Java applications>> recording <<event,event records>>
6308 in the channel named `my-channel`:
6312 $ lttng add-context --kernel --channel=my-channel \
6313 --type='$app:retriever:cur_msg_id'
6316 IMPORTANT: Make sure to always quote the `$` character when you
6317 use man:lttng-add-context(1) from a shell.
6320 NOTE: You cannot remove context fields from a channel once you add it.
6325 === Track process IDs
6327 It's often useful to allow only specific process IDs (PIDs) to emit
6328 events. For example, you may wish to record all the system calls made by
6329 a given process (à la http://linux.die.net/man/1/strace[strace]).
6331 The man:lttng-track(1) and man:lttng-untrack(1) commands serve this
6332 purpose. Both commands operate on a whitelist of process IDs. You _add_
6333 entries to this whitelist with the man:lttng-track(1) command and remove
6334 entries with the man:lttng-untrack(1) command. Any process which has one
6335 of the PIDs in the whitelist is allowed to emit LTTng events which pass
6336 an enabled <<event,event rule>>.
6338 NOTE: The PID tracker tracks the _numeric process IDs_. Should a
6339 process with a given tracked ID exit and another process be given this
6340 ID, then the latter would also be allowed to emit events.
6342 .Track and untrack process IDs.
6344 For the sake of the following example, assume the target system has 16
6348 <<creating-destroying-tracing-sessions,create a tracing session>>,
6349 the whitelist contains all the possible PIDs:
6352 .All PIDs are tracked.
6353 image::track-all.png[]
6355 When the whitelist is full and you use the man:lttng-track(1) command to
6356 specify some PIDs to track, LTTng first clears the whitelist, then it
6357 tracks the specific PIDs. After:
6361 $ lttng track --pid=3,4,7,10,13
6367 .PIDs 3, 4, 7, 10, and 13 are tracked.
6368 image::track-3-4-7-10-13.png[]
6370 You can add more PIDs to the whitelist afterwards:
6374 $ lttng track --pid=1,15,16
6380 .PIDs 1, 15, and 16 are added to the whitelist.
6381 image::track-1-3-4-7-10-13-15-16.png[]
6383 The man:lttng-untrack(1) command removes entries from the PID tracker's
6384 whitelist. Given the previous example, the following command:
6388 $ lttng untrack --pid=3,7,10,13
6391 leads to this whitelist:
6394 .PIDs 3, 7, 10, and 13 are removed from the whitelist.
6395 image::track-1-4-15-16.png[]
6397 LTTng can track all possible PIDs again using the opt:track(1):--all
6402 $ lttng track --pid --all
6405 The result is, again:
6408 .All PIDs are tracked.
6409 image::track-all.png[]
6412 .Track only specific PIDs
6414 A very typical use case with PID tracking is to start with an empty
6415 whitelist, then <<basic-tracing-session-control,start the tracers>>, and
6416 then add PIDs manually while tracers are active. You can accomplish this
6417 by using the opt:lttng-untrack(1):--all option of the
6418 man:lttng-untrack(1) command to clear the whitelist after you
6419 <<creating-destroying-tracing-sessions,create a tracing session>>:
6423 $ lttng untrack --pid --all
6429 .No PIDs are tracked.
6430 image::untrack-all.png[]
6432 If you trace with this whitelist configuration, the tracer records no
6433 events for this <<domain,tracing domain>> because no processes are
6434 tracked. You can use the man:lttng-track(1) command as usual to track
6435 specific PIDs, for example:
6439 $ lttng track --pid=6,11
6445 .PIDs 6 and 11 are tracked.
6446 image::track-6-11.png[]
6451 [[saving-loading-tracing-session]]
6452 === Save and load tracing session configurations
6454 Configuring a <<tracing-session,tracing session>> can be long. Some of
6455 the tasks involved are:
6457 * <<enabling-disabling-channels,Create channels>> with
6458 specific attributes.
6459 * <<adding-context,Add context fields>> to specific channels.
6460 * <<enabling-disabling-events,Create event rules>> with specific log
6461 level and filter conditions.
6463 If you use LTTng to solve real world problems, chances are you have to
6464 record events using the same tracing session setup over and over,
6465 modifying a few variables each time in your instrumented program
6466 or environment. To avoid constant tracing session reconfiguration,
6467 the man:lttng(1) command-line tool can save and load tracing session
6468 configurations to/from XML files.
6470 To save a given tracing session configuration:
6472 * Use the man:lttng-save(1) command:
6477 $ lttng save my-session
6481 Replace `my-session` with the name of the tracing session to save.
6483 LTTng saves tracing session configurations to
6484 dir:{$LTTNG_HOME/.lttng/sessions} by default. Note that the
6485 env:LTTNG_HOME environment variable defaults to `$HOME` if not set. Use
6486 the opt:lttng-save(1):--output-path option to change this destination
6489 LTTng saves all configuration parameters, for example:
6491 * The tracing session name.
6492 * The trace data output path.
6493 * The channels with their state and all their attributes.
6494 * The context fields you added to channels.
6495 * The event rules with their state, log level and filter conditions.
6497 To load a tracing session:
6499 * Use the man:lttng-load(1) command:
6504 $ lttng load my-session
6508 Replace `my-session` with the name of the tracing session to load.
6510 When LTTng loads a configuration, it restores your saved tracing session
6511 as if you just configured it manually.
6513 See man:lttng(1) for the complete list of command-line options. You
6514 can also save and load all many sessions at a time, and decide in which
6515 directory to output the XML files.
6518 [[sending-trace-data-over-the-network]]
6519 === Send trace data over the network
6521 LTTng can send the recorded trace data to a remote system over the
6522 network instead of writing it to the local file system.
6524 To send the trace data over the network:
6526 . On the _remote_ system (which can also be the target system),
6527 start an LTTng <<lttng-relayd,relay daemon>> (man:lttng-relayd(8)):
6536 . On the _target_ system, create a tracing session configured to
6537 send trace data over the network:
6542 $ lttng create my-session --set-url=net://remote-system
6546 Replace `remote-system` by the host name or IP address of the
6547 remote system. See man:lttng-create(1) for the exact URL format.
6549 . On the target system, use the man:lttng(1) command-line tool as usual.
6550 When tracing is active, the target's consumer daemon sends sub-buffers
6551 to the relay daemon running on the remote system instead of flushing
6552 them to the local file system. The relay daemon writes the received
6553 packets to the local file system.
6555 The relay daemon writes trace files to
6556 +$LTTNG_HOME/lttng-traces/__hostname__/__session__+ by default, where
6557 +__hostname__+ is the host name of the target system and +__session__+
6558 is the tracing session name. Note that the env:LTTNG_HOME environment
6559 variable defaults to `$HOME` if not set. Use the
6560 opt:lttng-relayd(8):--output option of man:lttng-relayd(8) to write
6561 trace files to another base directory.
6566 === View events as LTTng emits them (noch:{LTTng} live)
6568 LTTng live is a network protocol implemented by the <<lttng-relayd,relay
6569 daemon>> (man:lttng-relayd(8)) to allow compatible trace viewers to
6570 display events as LTTng emits them on the target system while tracing is
6573 The relay daemon creates a _tee_: it forwards the trace data to both
6574 the local file system and to connected live viewers:
6577 .The relay daemon creates a _tee_, forwarding the trace data to both trace files and a connected live viewer.
6582 . On the _target system_, create a <<tracing-session,tracing session>>
6588 $ lttng create my-session --live
6592 This spawns a local relay daemon.
6594 . Start the live viewer and configure it to connect to the relay
6595 daemon. For example, with http://diamon.org/babeltrace[Babeltrace]:
6600 $ babeltrace --input-format=lttng-live
6601 net://localhost/host/hostname/my-session
6608 * `hostname` with the host name of the target system.
6609 * `my-session` with the name of the tracing session to view.
6612 . Configure the tracing session as usual with the man:lttng(1)
6613 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6615 You can list the available live tracing sessions with Babeltrace:
6619 $ babeltrace --input-format=lttng-live net://localhost
6622 You can start the relay daemon on another system. In this case, you need
6623 to specify the relay daemon's URL when you create the tracing session
6624 with the opt:lttng-create(1):--set-url option. You also need to replace
6625 `localhost` in the procedure above with the host name of the system on
6626 which the relay daemon is running.
6628 See man:lttng-create(1) and man:lttng-relayd(8) for the complete list of
6629 command-line options.
6633 [[taking-a-snapshot]]
6634 === Take a snapshot of the current sub-buffers of a tracing session
6636 The normal behavior of LTTng is to append full sub-buffers to growing
6637 trace data files. This is ideal to keep a full history of the events
6638 that occurred on the target system, but it can
6639 represent too much data in some situations. For example, you may wish
6640 to trace your application continuously until some critical situation
6641 happens, in which case you only need the latest few recorded
6642 events to perform the desired analysis, not multi-gigabyte trace files.
6644 With the man:lttng-snapshot(1) command, you can take a snapshot of the
6645 current sub-buffers of a given <<tracing-session,tracing session>>.
6646 LTTng can write the snapshot to the local file system or send it over
6651 . Create a tracing session in _snapshot mode_:
6656 $ lttng create my-session --snapshot
6660 The <<channel-overwrite-mode-vs-discard-mode,event loss mode>> of
6661 <<channel,channels>> created in this mode is automatically set to
6662 _overwrite_ (flight recorder mode).
6664 . Configure the tracing session as usual with the man:lttng(1)
6665 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6667 . **Optional**: When you need to take a snapshot,
6668 <<basic-tracing-session-control,stop tracing>>.
6670 You can take a snapshot when the tracers are active, but if you stop
6671 them first, you are sure that the data in the sub-buffers does not
6672 change before you actually take the snapshot.
6679 $ lttng snapshot record --name=my-first-snapshot
6683 LTTng writes the current sub-buffers of all the current tracing
6684 session's channels to trace files on the local file system. Those trace
6685 files have `my-first-snapshot` in their name.
6687 There is no difference between the format of a normal trace file and the
6688 format of a snapshot: viewers of LTTng traces also support LTTng
6691 By default, LTTng writes snapshot files to the path shown by
6692 `lttng snapshot list-output`. You can change this path or decide to send
6693 snapshots over the network using either:
6695 . An output path or URL that you specify when you create the
6697 . An snapshot output path or URL that you add using
6698 `lttng snapshot add-output`
6699 . An output path or URL that you provide directly to the
6700 `lttng snapshot record` command.
6702 Method 3 overrides method 2, which overrides method 1. When you
6703 specify a URL, a relay daemon must listen on a remote system (see
6704 <<sending-trace-data-over-the-network,Send trace data over the network>>).
6709 === Use the machine interface
6711 With any command of the man:lttng(1) command-line tool, you can set the
6712 opt:lttng(1):--mi option to `xml` (before the command name) to get an
6713 XML machine interface output, for example:
6717 $ lttng --mi=xml enable-event --kernel --syscall open
6720 A schema definition (XSD) is
6721 https://github.com/lttng/lttng-tools/blob/stable-2.8/src/common/mi-lttng-3.0.xsd[available]
6722 to ease the integration with external tools as much as possible.
6726 [[metadata-regenerate]]
6727 === Regenerate the metadata of an LTTng trace
6729 An LTTng trace, which is a http://diamon.org/ctf[CTF] trace, has both
6730 data stream files and a metadata file. This metadata file contains,
6731 amongst other things, information about the offset of the clock sources
6732 used to timestamp <<event,event records>> when tracing.
6734 If, once a <<tracing-session,tracing session>> is
6735 <<basic-tracing-session-control,started>>, a major
6736 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
6737 happens, the trace's clock offset also needs to be updated. You
6738 can use the man:lttng-metadata(1) command to do so.
6740 The main use case of this command is to allow a system to boot with
6741 an incorrect wall time and trace it with LTTng before its wall time
6742 is corrected. Once the system is known to be in a state where its
6743 wall time is correct, it can run `lttng metadata regenerate`.
6745 To regenerate the metadata of an LTTng trace:
6747 * Use the `regenerate` action of the man:lttng-metadata(1) command:
6752 $ lttng metadata regenerate
6758 `lttng metadata regenerate` has the following limitations:
6760 * Tracing session <<creating-destroying-tracing-sessions,created>>
6762 * User space <<channel,channels>>, if any, using
6763 <<channel-buffering-schemes,per-user buffering>>.
6768 [[persistent-memory-file-systems]]
6769 === Record trace data on persistent memory file systems
6771 https://en.wikipedia.org/wiki/Non-volatile_random-access_memory[Non-volatile random-access memory]
6772 (NVRAM) is random-access memory that retains its information when power
6773 is turned off (non-volatile). Systems with such memory can store data
6774 structures in RAM and retrieve them after a reboot, without flushing
6775 to typical _storage_.
6777 Linux supports NVRAM file systems thanks to either
6778 http://pramfs.sourceforge.net/[PRAMFS] or
6779 https://www.kernel.org/doc/Documentation/filesystems/dax.txt[DAX]{nbsp}+{nbsp}http://lkml.iu.edu/hypermail/linux/kernel/1504.1/03463.html[pmem]
6780 (requires Linux 4.1+).
6782 This section does not describe how to operate such file systems;
6783 we assume that you have a working persistent memory file system.
6785 When you create a <<tracing-session,tracing session>>, you can specify
6786 the path of the shared memory holding the sub-buffers. If you specify a
6787 location on an NVRAM file system, then you can retrieve the latest
6788 recorded trace data when the system reboots after a crash.
6790 To record trace data on a persistent memory file system and retrieve the
6791 trace data after a system crash:
6793 . Create a tracing session with a sub-buffer shared memory path located
6794 on an NVRAM file system:
6799 $ lttng create my-session --shm-path=/path/to/shm
6803 . Configure the tracing session as usual with the man:lttng(1)
6804 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6806 . After a system crash, use the man:lttng-crash(1) command-line tool to
6807 view the trace data recorded on the NVRAM file system:
6812 $ lttng-crash /path/to/shm
6816 The binary layout of the ring buffer files is not exactly the same as
6817 the trace files layout. This is why you need to use man:lttng-crash(1)
6818 instead of your preferred trace viewer directly.
6820 To convert the ring buffer files to LTTng trace files:
6822 * Use the opt:lttng-crash(1):--extract option of man:lttng-crash(1):
6827 $ lttng-crash --extract=/path/to/trace /path/to/shm
6835 [[lttng-modules-ref]]
6836 === noch:{LTTng-modules}
6839 [[lttng-modules-tp-fields]]
6840 ==== Tracepoint fields macros (for `TP_FIELDS()`)
6842 [[tp-fast-assign]][[tp-struct-entry]]The available macros to define
6843 tracepoint fields, which must be listed within `TP_FIELDS()` in
6844 `LTTNG_TRACEPOINT_EVENT()`, are:
6846 [role="func-desc growable",cols="asciidoc,asciidoc"]
6847 .Available macros to define LTTng-modules tracepoint fields
6849 |Macro |Description and parameters
6852 +ctf_integer(__t__, __n__, __e__)+
6854 +ctf_integer_nowrite(__t__, __n__, __e__)+
6856 +ctf_user_integer(__t__, __n__, __e__)+
6858 +ctf_user_integer_nowrite(__t__, __n__, __e__)+
6860 Standard integer, displayed in base 10.
6863 Integer C type (`int`, `long`, `size_t`, ...).
6869 Argument expression.
6872 +ctf_integer_hex(__t__, __n__, __e__)+
6874 +ctf_user_integer_hex(__t__, __n__, __e__)+
6876 Standard integer, displayed in base 16.
6885 Argument expression.
6887 |+ctf_integer_oct(__t__, __n__, __e__)+
6889 Standard integer, displayed in base 8.
6898 Argument expression.
6901 +ctf_integer_network(__t__, __n__, __e__)+
6903 +ctf_user_integer_network(__t__, __n__, __e__)+
6905 Integer in network byte order (big-endian), displayed in base 10.
6914 Argument expression.
6917 +ctf_integer_network_hex(__t__, __n__, __e__)+
6919 +ctf_user_integer_network_hex(__t__, __n__, __e__)+
6921 Integer in network byte order, displayed in base 16.
6930 Argument expression.
6933 +ctf_string(__n__, __e__)+
6935 +ctf_string_nowrite(__n__, __e__)+
6937 +ctf_user_string(__n__, __e__)+
6939 +ctf_user_string_nowrite(__n__, __e__)+
6941 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
6947 Argument expression.
6950 +ctf_array(__t__, __n__, __e__, __s__)+
6952 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
6954 +ctf_user_array(__t__, __n__, __e__, __s__)+
6956 +ctf_user_array_nowrite(__t__, __n__, __e__, __s__)+
6958 Statically-sized array of integers.
6961 Array element C type.
6967 Argument expression.
6973 +ctf_array_bitfield(__t__, __n__, __e__, __s__)+
6975 +ctf_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
6977 +ctf_user_array_bitfield(__t__, __n__, __e__, __s__)+
6979 +ctf_user_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
6981 Statically-sized array of bits.
6983 The type of +__e__+ must be an integer type. +__s__+ is the number
6984 of elements of such type in +__e__+, not the number of bits.
6987 Array element C type.
6993 Argument expression.
6999 +ctf_array_text(__t__, __n__, __e__, __s__)+
7001 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
7003 +ctf_user_array_text(__t__, __n__, __e__, __s__)+
7005 +ctf_user_array_text_nowrite(__t__, __n__, __e__, __s__)+
7007 Statically-sized array, printed as text.
7009 The string does not need to be null-terminated.
7012 Array element C type (always `char`).
7018 Argument expression.
7024 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
7026 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7028 +ctf_user_sequence(__t__, __n__, __e__, __T__, __E__)+
7030 +ctf_user_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7032 Dynamically-sized array of integers.
7034 The type of +__E__+ must be unsigned.
7037 Array element C type.
7043 Argument expression.
7046 Length expression C type.
7052 +ctf_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7054 +ctf_user_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7056 Dynamically-sized array of integers, displayed in base 16.
7058 The type of +__E__+ must be unsigned.
7061 Array element C type.
7067 Argument expression.
7070 Length expression C type.
7075 |+ctf_sequence_network(__t__, __n__, __e__, __T__, __E__)+
7077 Dynamically-sized array of integers in network byte order (big-endian),
7078 displayed in base 10.
7080 The type of +__E__+ must be unsigned.
7083 Array element C type.
7089 Argument expression.
7092 Length expression C type.
7098 +ctf_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7100 +ctf_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7102 +ctf_user_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7104 +ctf_user_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7106 Dynamically-sized array of bits.
7108 The type of +__e__+ must be an integer type. +__s__+ is the number
7109 of elements of such type in +__e__+, not the number of bits.
7111 The type of +__E__+ must be unsigned.
7114 Array element C type.
7120 Argument expression.
7123 Length expression C type.
7129 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7131 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7133 +ctf_user_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7135 +ctf_user_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7137 Dynamically-sized array, displayed as text.
7139 The string does not need to be null-terminated.
7141 The type of +__E__+ must be unsigned.
7143 The behaviour is undefined if +__e__+ is `NULL`.
7146 Sequence element C type (always `char`).
7152 Argument expression.
7155 Length expression C type.
7161 Use the `_user` versions when the argument expression, `e`, is
7162 a user space address. In the cases of `ctf_user_integer*()` and
7163 `ctf_user_float*()`, `&e` must be a user space address, thus `e` must
7166 The `_nowrite` versions omit themselves from the session trace, but are
7167 otherwise identical. This means the `_nowrite` fields won't be written
7168 in the recorded trace. Their primary purpose is to make some
7169 of the event context available to the
7170 <<enabling-disabling-events,event filters>> without having to
7171 commit the data to sub-buffers.
7177 Terms related to LTTng and to tracing in general:
7180 The http://diamon.org/babeltrace[Babeltrace] project, which includes
7181 the cmd:babeltrace command, some libraries, and Python bindings.
7183 <<channel-buffering-schemes,buffering scheme>>::
7184 A layout of sub-buffers applied to a given channel.
7186 <<channel,channel>>::
7187 An entity which is responsible for a set of ring buffers.
7189 <<event,Event rules>> are always attached to a specific channel.
7192 A reference of time for a tracer.
7194 <<lttng-consumerd,consumer daemon>>::
7195 A process which is responsible for consuming the full sub-buffers
7196 and write them to a file system or send them over the network.
7198 <<channel-overwrite-mode-vs-discard-mode,discard mode>>:: The event loss
7199 mode in which the tracer _discards_ new event records when there's no
7200 sub-buffer space left to store them.
7203 The consequence of the execution of an instrumentation
7204 point, like a tracepoint that you manually place in some source code,
7205 or a Linux kernel KProbe.
7207 An event is said to _occur_ at a specific time. Different actions can
7208 be taken upon the occurrence of an event, like record the event's payload
7211 <<channel-overwrite-mode-vs-discard-mode,event loss mode>>::
7212 The mechanism by which event records of a given channel are lost
7213 (not recorded) when there is no sub-buffer space left to store them.
7215 [[def-event-name]]event name::
7216 The name of an event, which is also the name of the event record.
7217 This is also called the _instrumentation point name_.
7220 A record, in a trace, of the payload of an event which occured.
7222 <<event,event rule>>::
7223 Set of conditions which must be satisfied for one or more occuring
7224 events to be recorded.
7226 `java.util.logging`::
7228 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[core logging facilities].
7230 <<instrumenting,instrumentation>>::
7231 The use of LTTng probes to make a piece of software traceable.
7233 instrumentation point::
7234 A point in the execution path of a piece of software that, when
7235 reached by this execution, can emit an event.
7237 instrumentation point name::
7238 See _<<def-event-name,event name>>_.
7241 A http://logging.apache.org/log4j/1.2/[logging library] for Java
7242 developed by the Apache Software Foundation.
7245 Level of severity of a log statement or user space
7246 instrumentation point.
7249 The _Linux Trace Toolkit: next generation_ project.
7251 <<lttng-cli,cmd:lttng>>::
7252 A command-line tool provided by the LTTng-tools project which you
7253 can use to send and receive control messages to and from a
7257 The https://github.com/lttng/lttng-analyses[LTTng analyses] project,
7258 which is a set of analyzing programs that are used to obtain a
7259 higher level view of an LTTng trace.
7261 cmd:lttng-consumerd::
7262 The name of the consumer daemon program.
7265 A utility provided by the LTTng-tools project which can convert
7266 ring buffer files (usually
7267 <<persistent-memory-file-systems,saved on a persistent memory file system>>)
7270 LTTng Documentation::
7273 <<lttng-live,LTTng live>>::
7274 A communication protocol between the relay daemon and live viewers
7275 which makes it possible to see events "live", as they are received by
7278 <<lttng-modules,LTTng-modules>>::
7279 The https://github.com/lttng/lttng-modules[LTTng-modules] project,
7280 which contains the Linux kernel modules to make the Linux kernel
7281 instrumentation points available for LTTng tracing.
7284 The name of the relay daemon program.
7286 cmd:lttng-sessiond::
7287 The name of the session daemon program.
7290 The https://github.com/lttng/lttng-tools[LTTng-tools] project, which
7291 contains the various programs and libraries used to
7292 <<controlling-tracing,control tracing>>.
7294 <<lttng-ust,LTTng-UST>>::
7295 The https://github.com/lttng/lttng-ust[LTTng-UST] project, which
7296 contains libraries to instrument user applications.
7298 <<lttng-ust-agents,LTTng-UST Java agent>>::
7299 A Java package provided by the LTTng-UST project to allow the
7300 LTTng instrumentation of `java.util.logging` and Apache log4j 1.2
7303 <<lttng-ust-agents,LTTng-UST Python agent>>::
7304 A Python package provided by the LTTng-UST project to allow the
7305 LTTng instrumentation of Python logging statements.
7307 <<channel-overwrite-mode-vs-discard-mode,overwrite mode>>::
7308 The event loss mode in which new event records overwrite older
7309 event records when there's no sub-buffer space left to store them.
7311 <<channel-buffering-schemes,per-process buffering>>::
7312 A buffering scheme in which each instrumented process has its own
7313 sub-buffers for a given user space channel.
7315 <<channel-buffering-schemes,per-user buffering>>::
7316 A buffering scheme in which all the processes of a Unix user share the
7317 same sub-buffer for a given user space channel.
7319 <<lttng-relayd,relay daemon>>::
7320 A process which is responsible for receiving the trace data sent by
7321 a distant consumer daemon.
7324 A set of sub-buffers.
7326 <<lttng-sessiond,session daemon>>::
7327 A process which receives control commands from you and orchestrates
7328 the tracers and various LTTng daemons.
7330 <<taking-a-snapshot,snapshot>>::
7331 A copy of the current data of all the sub-buffers of a given tracing
7332 session, saved as trace files.
7335 One part of an LTTng ring buffer which contains event records.
7338 The time information attached to an event when it is emitted.
7341 A set of files which are the concatenations of one or more
7342 flushed sub-buffers.
7345 The action of recording the events emitted by an application
7346 or by a system, or to initiate such recording by controlling
7350 The http://tracecompass.org[Trace Compass] project and application.
7353 An instrumentation point using the tracepoint mechanism of the Linux
7354 kernel or of LTTng-UST.
7356 tracepoint definition::
7357 The definition of a single tracepoint.
7360 The name of a tracepoint.
7362 tracepoint provider::
7363 A set of functions providing tracepoints to an instrumented user
7366 Not to be confused with a _tracepoint provider package_: many tracepoint
7367 providers can exist within a tracepoint provider package.
7369 tracepoint provider package::
7370 One or more tracepoint providers compiled as an object file or as
7374 A software which records emitted events.
7376 <<domain,tracing domain>>::
7377 A namespace for event sources.
7379 <<tracing-group,tracing group>>::
7380 The Unix group in which a Unix user can be to be allowed to trace the
7383 <<tracing-session,tracing session>>::
7384 A stateful dialogue between you and a <<lttng-sessiond,session
7388 An application running in user space, as opposed to a Linux kernel
7389 module, for example.