1 The LTTng Documentation
2 =======================
3 Philippe Proulx <pproulx@efficios.com>
7 include::../common/copyright.txt[]
10 include::../common/welcome.txt[]
13 include::../common/audience.txt[]
17 === What's in this documentation?
19 The LTTng Documentation is divided into the following sections:
21 * **<<nuts-and-bolts,Nuts and bolts>>** explains the
22 rudiments of software tracing and the rationale behind the
25 You can skip this section if you’re familiar with software tracing and
26 with the LTTng project.
28 * **<<installing-lttng,Installation>>** describes the steps to
29 install the LTTng packages on common Linux distributions and from
32 You can skip this section if you already properly installed LTTng on
35 * **<<getting-started,Quick start>>** is a concise guide to
36 getting started quickly with LTTng kernel and user space tracing.
38 We recommend this section if you're new to LTTng or to software tracing
41 You can skip this section if you're not new to LTTng.
43 * **<<core-concepts,Core concepts>>** explains the concepts at
46 It's a good idea to become familiar with the core concepts
47 before attempting to use the toolkit.
49 * **<<plumbing,Components of LTTng>>** describes the various components
50 of the LTTng machinery, like the daemons, the libraries, and the
51 command-line interface.
52 * **<<instrumenting,Instrumentation>>** shows different ways to
53 instrument user applications and the Linux kernel.
55 Instrumenting source code is essential to provide a meaningful
58 You can skip this section if you do not have a programming background.
60 * **<<controlling-tracing,Tracing control>>** is divided into topics
61 which demonstrate how to use the vast array of features that
62 LTTng{nbsp}{revision} offers.
63 * **<<reference,Reference>>** contains reference tables.
64 * **<<glossary,Glossary>>** is a specialized dictionary of terms related
65 to LTTng or to the field of software tracing.
68 include::../common/convention.txt[]
71 include::../common/acknowledgements.txt[]
75 == What's new in LTTng {revision}?
77 LTTng{nbsp}{revision} bears the name _Joannès_. A Berliner Weisse style
78 beer from the http://letreflenoir.com/[Trèfle Noir] microbrewery in
79 https://en.wikipedia.org/wiki/Rouyn-Noranda[Rouyn-Noranda], the
80 https://www.beeradvocate.com/beer/profile/20537/238967/[_**Joannès**_]
81 is a tangy beer with a distinct pink dress and intense fruit flavor,
82 thanks to the presence of fresh blackcurrant grown in Témiscamingue.
84 New features and changes in LTTng{nbsp}{revision}:
86 * **Tracing control**:
87 ** You can override the name or the URL of a tracing session
88 configuration when you use man:lttng-load(1) thanks to the new
89 opt:lttng-load(1):--override-name and
90 opt:lttng-load(1):--override-url options.
91 ** The new `lttng regenerate` command replaces the now deprecated
92 `lttng metadata` command of LTTng 2.8. man:lttng-regenerate(1) can
93 also <<regenerate-statedump,generate the state dump event records>>
94 of a given tracing session on demand, a handy feature when
95 <<taking-a-snapshot,taking a snapshot>>.
96 ** You can add PMU counters by raw ID with man:lttng-add-context(1):
101 $ lttng add-context --kernel --type=perf:cpu:raw:r0013c:x86unhalted
105 The format of the raw ID is the same as used with man:perf-record(1).
106 See <<adding-context,Add context fields to a channel>> for more
109 ** The LTTng <<lttng-relayd,relay daemon>> is now supported on
110 OS{nbsp}X and macOS for a smoother integration within a trace
111 analysis workflow, regardless of the platform used.
113 * **User space tracing**:
114 ** Improved performance (tested on x86-64 and ARMv7-A
115 (https://en.wikipedia.org/wiki/Cubieboard[Cubieboard])
117 ** New helper library (`liblttng-ust-fd`) to help with
118 <<liblttng-ust-fd,applications which close file descriptors that
119 don't belong to them>>, for example, in a loop which closes file
120 descriptors after man:fork(2), or BSD's `closeall()`.
121 ** More accurate <<liblttng-ust-dl,dynamic linker instrumentation>> and
122 state dump event records, especially when a dynamically loaded
123 library manually loads its own dependencies.
124 ** New `ctf_*()` field definition macros (see man:lttng-ust(3)):
125 *** `ctf_array_hex()`
126 *** `ctf_array_network()`
127 *** `ctf_array_network_hex()`
128 *** `ctf_sequence_hex()`
129 *** `ctf_sequence_network()`
130 *** `ctf_sequence_network_hex()`
131 ** New `lttng_ust_loaded` weak symbol defined by `liblttng-ust` for
132 an application to know if the LTTng-UST shared library is loaded
140 int lttng_ust_loaded __attribute__((weak));
144 if (lttng_ust_loaded) {
145 puts("LTTng-UST is loaded!");
147 puts("LTTng-UST is not loaded!");
155 ** LTTng-UST thread names have the `-ust` suffix.
157 * **Linux kernel tracing**:
158 ** Improved performance (tested on x86-64 and ARMv7-A
159 (https://en.wikipedia.org/wiki/Cubieboard[Cubieboard])
161 ** New enumeration <<lttng-modules-tp-fields,field definition macros>>:
162 `ctf_enum()` and `ctf_user_enum()`.
163 ** IPv4, IPv6, and TCP header data is recorded in the event records
164 produced by tracepoints starting with `net_`.
165 ** Detailed system call event records: `select`, `pselect6`, `poll`,
166 `ppoll`, `epoll_wait`, `epoll_pwait`, and `epoll_ctl` on all
167 architectures supported by LTTng-modules, and `accept4` on x86-64.
168 ** New I²C instrumentation: the `extract_sensitive_payload` parameter
169 of the new `lttng-probe-i2c` LTTng module controls whether or not
170 the payloads of I²C messages are recorded in I²C event records, since
171 they may contain sensitive data (for example, keystrokes).
172 ** When the LTTng kernel modules are built into the Linux kernel image,
173 the `CONFIG_TRACEPOINTS` configuration option is automatically
180 What is LTTng? As its name suggests, the _Linux Trace Toolkit: next
181 generation_ is a modern toolkit for tracing Linux systems and
182 applications. So your first question might be:
189 As the history of software engineering progressed and led to what
190 we now take for granted--complex, numerous and
191 interdependent software applications running in parallel on
192 sophisticated operating systems like Linux--the authors of such
193 components, software developers, began feeling a natural
194 urge to have tools that would ensure the robustness and good performance
195 of their masterpieces.
197 One major achievement in this field is, inarguably, the
198 https://www.gnu.org/software/gdb/[GNU debugger (GDB)],
199 an essential tool for developers to find and fix bugs. But even the best
200 debugger won't help make your software run faster, and nowadays, faster
201 software means either more work done by the same hardware, or cheaper
202 hardware for the same work.
204 A _profiler_ is often the tool of choice to identify performance
205 bottlenecks. Profiling is suitable to identify _where_ performance is
206 lost in a given software. The profiler outputs a profile, a statistical
207 summary of observed events, which you may use to discover which
208 functions took the most time to execute. However, a profiler won't
209 report _why_ some identified functions are the bottleneck. Bottlenecks
210 might only occur when specific conditions are met, conditions that are
211 sometimes impossible to capture by a statistical profiler, or impossible
212 to reproduce with an application altered by the overhead of an
213 event-based profiler. For a thorough investigation of software
214 performance issues, a history of execution is essential, with the
215 recorded values of variables and context fields you choose, and
216 with as little influence as possible on the instrumented software. This
217 is where tracing comes in handy.
219 _Tracing_ is a technique used to understand what goes on in a running
220 software system. The software used for tracing is called a _tracer_,
221 which is conceptually similar to a tape recorder. When recording,
222 specific instrumentation points placed in the software source code
223 generate events that are saved on a giant tape: a _trace_ file. You
224 can trace user applications and the operating system at the same time,
225 opening the possibility of resolving a wide range of problems that would
226 otherwise be extremely challenging.
228 Tracing is often compared to _logging_. However, tracers and loggers are
229 two different tools, serving two different purposes. Tracers are
230 designed to record much lower-level events that occur much more
231 frequently than log messages, often in the range of thousands per
232 second, with very little execution overhead. Logging is more appropriate
233 for a very high-level analysis of less frequent events: user accesses,
234 exceptional conditions (errors and warnings, for example), database
235 transactions, instant messaging communications, and such. Simply put,
236 logging is one of the many use cases that can be satisfied with tracing.
238 The list of recorded events inside a trace file can be read manually
239 like a log file for the maximum level of detail, but it is generally
240 much more interesting to perform application-specific analyses to
241 produce reduced statistics and graphs that are useful to resolve a
242 given problem. Trace viewers and analyzers are specialized tools
245 In the end, this is what LTTng is: a powerful, open source set of
246 tools to trace the Linux kernel and user applications at the same time.
247 LTTng is composed of several components actively maintained and
248 developed by its link:/community/#where[community].
251 [[lttng-alternatives]]
252 === Alternatives to noch:{LTTng}
254 Excluding proprietary solutions, a few competing software tracers
257 * https://github.com/dtrace4linux/linux[dtrace4linux] is a port of
258 Sun Microsystems's DTrace to Linux. The cmd:dtrace tool interprets
259 user scripts and is responsible for loading code into the
260 Linux kernel for further execution and collecting the outputted data.
261 * https://en.wikipedia.org/wiki/Berkeley_Packet_Filter[eBPF] is a
262 subsystem in the Linux kernel in which a virtual machine can execute
263 programs passed from the user space to the kernel. You can attach
264 such programs to tracepoints and KProbes thanks to a system call, and
265 they can output data to the user space when executed thanks to
266 different mechanisms (pipe, VM register values, and eBPF maps, to name
268 * https://www.kernel.org/doc/Documentation/trace/ftrace.txt[ftrace]
269 is the de facto function tracer of the Linux kernel. Its user
270 interface is a set of special files in sysfs.
271 * https://perf.wiki.kernel.org/[perf] is
272 a performance analyzing tool for Linux which supports hardware
273 performance counters, tracepoints, as well as other counters and
274 types of probes. perf's controlling utility is the cmd:perf command
276 * http://linux.die.net/man/1/strace[strace]
277 is a command-line utility which records system calls made by a
278 user process, as well as signal deliveries and changes of process
279 state. strace makes use of https://en.wikipedia.org/wiki/Ptrace[ptrace]
280 to fulfill its function.
281 * http://www.sysdig.org/[sysdig], like SystemTap, uses scripts to
282 analyze Linux kernel events. You write scripts, or _chisels_ in
283 sysdig's jargon, in Lua and sysdig executes them while the system is
284 being traced or afterwards. sysdig's interface is the cmd:sysdig
285 command-line tool as well as the curses-based cmd:csysdig tool.
286 * https://sourceware.org/systemtap/[SystemTap] is a Linux kernel and
287 user space tracer which uses custom user scripts to produce plain text
288 traces. SystemTap converts the scripts to the C language, and then
289 compiles them as Linux kernel modules which are loaded to produce
290 trace data. SystemTap's primary user interface is the cmd:stap
293 The main distinctive features of LTTng is that it produces correlated
294 kernel and user space traces, as well as doing so with the lowest
295 overhead amongst other solutions. It produces trace files in the
296 http://diamon.org/ctf[CTF] format, a file format optimized
297 for the production and analyses of multi-gigabyte data.
299 LTTng is the result of more than 10 years of active open source
300 development by a community of passionate developers.
301 LTTng{nbsp}{revision} is currently available on major desktop and server
304 The main interface for tracing control is a single command-line tool
305 named cmd:lttng. The latter can create several tracing sessions, enable
306 and disable events on the fly, filter events efficiently with custom
307 user expressions, start and stop tracing, and much more. LTTng can
308 record the traces on the file system or send them over the network, and
309 keep them totally or partially. You can view the traces once tracing
310 becomes inactive or in real-time.
312 <<installing-lttng,Install LTTng now>> and
313 <<getting-started,start tracing>>!
319 **LTTng** is a set of software <<plumbing,components>> which interact to
320 <<instrumenting,instrument>> the Linux kernel and user applications, and
321 to <<controlling-tracing,control tracing>> (start and stop
322 tracing, enable and disable event rules, and the rest). Those
323 components are bundled into the following packages:
325 * **LTTng-tools**: Libraries and command-line interface to
327 * **LTTng-modules**: Linux kernel modules to instrument and
329 * **LTTng-UST**: Libraries and Java/Python packages to instrument and
330 trace user applications.
332 Most distributions mark the LTTng-modules and LTTng-UST packages as
333 optional when installing LTTng-tools (which is always required). In the
334 following sections, we always provide the steps to install all three,
337 * You only need to install LTTng-modules if you intend to trace the
339 * You only need to install LTTng-UST if you intend to trace user
343 .Availability of LTTng{nbsp}{revision} for major Linux distributions as of 25 July 2017.
345 |Distribution |Available in releases |Alternatives
347 |https://www.ubuntu.com/[Ubuntu]
348 |<<ubuntu,Ubuntu{nbsp}17.04 _Zesty Zapus_>>.
350 Ubuntu{nbsp}14.04 _Trusty Tahr_ and Ubuntu{nbsp}16.04 _Xenial Xerus_:
351 <<ubuntu-ppa,use the LTTng Stable{nbsp}{revision} PPA>>.
352 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
353 other Ubuntu releases.
355 |https://getfedora.org/[Fedora]
356 |<<fedora,Fedora{nbsp}26>>.
357 |link:/docs/v2.8#doc-fedora[LTTng{nbsp}2.8 for Fedora 25].
359 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
360 other Fedora releases.
362 |https://www.debian.org/[Debian]
363 |xref:debian[Debian "stretch" (stable), Debian "buster" (testing),
364 and Debian "sid" (unstable)].
365 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
366 other Debian releases.
368 |https://www.archlinux.org/[Arch Linux]
369 |<<arch-linux,Latest AUR packages>>.
372 |https://alpinelinux.org/[Alpine Linux]
373 |<<alpine-linux,Alpine Linux "edge">>.
374 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
375 other Alpine Linux releases.
377 |https://www.redhat.com/[RHEL] and https://www.suse.com/[SLES]
378 |See http://packages.efficios.com/[EfficiOS Enterprise Packages].
381 |https://buildroot.org/[Buildroot]
382 |<<buildroot,Buildroot{nbsp}2017.02 and Buildroot{nbsp}2017.05>>.
383 |link:/docs/v2.8#doc-buildroot[LTTng{nbsp}2.8 for Buildroot{nbsp}2016.11].
385 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
386 other Buildroot releases.
388 |http://www.openembedded.org/wiki/Main_Page[OpenEmbedded] and
389 https://www.yoctoproject.org/[Yocto]
390 |<<oe-yocto,Yocto Project{nbsp}2.3 _Pyro_>> (`openembedded-core` layer).
391 |link:/docs/v2.8#doc-oe-yocto[LTTng{nbsp}2.8 for Yocto Project{nbsp}2.2 _Morty_]
392 (`openembedded-core` layer).
394 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
395 other OpenEmbedded releases.
400 === [[ubuntu-official-repositories]]Ubuntu
402 LTTng{nbsp}{revision} is available on Ubuntu{nbsp}17.04 _Zesty Zapus_.
403 For previous releases of Ubuntu, <<ubuntu-ppa,use the LTTng
404 Stable{nbsp}{revision} PPA>>.
406 To install LTTng{nbsp}{revision} on Ubuntu{nbsp}17.04 _Zesty Zapus_:
408 . Install the main LTTng{nbsp}{revision} packages:
413 # apt-get install lttng-tools
414 # apt-get install lttng-modules-dkms
415 # apt-get install liblttng-ust-dev
419 . **If you need to instrument and trace
420 <<java-application,Java applications>>**, install the LTTng-UST
426 # apt-get install liblttng-ust-agent-java
430 . **If you need to instrument and trace
431 <<python-application,Python{nbsp}3 applications>>**, install the
432 LTTng-UST Python agent:
437 # apt-get install python3-lttngust
443 ==== noch:{LTTng} Stable {revision} PPA
445 The https://launchpad.net/~lttng/+archive/ubuntu/stable-{revision}[LTTng
446 Stable{nbsp}{revision} PPA] offers the latest stable
447 LTTng{nbsp}{revision} packages for:
449 * Ubuntu{nbsp}14.04 _Trusty Tahr_
450 * Ubuntu{nbsp}16.04 _Xenial Xerus_
452 To install LTTng{nbsp}{revision} from the LTTng Stable{nbsp}{revision} PPA:
454 . Add the LTTng Stable{nbsp}{revision} PPA repository and update the
460 # apt-add-repository ppa:lttng/stable-2.9
465 . Install the main LTTng{nbsp}{revision} packages:
470 # apt-get install lttng-tools
471 # apt-get install lttng-modules-dkms
472 # apt-get install liblttng-ust-dev
476 . **If you need to instrument and trace
477 <<java-application,Java applications>>**, install the LTTng-UST
483 # apt-get install liblttng-ust-agent-java
487 . **If you need to instrument and trace
488 <<python-application,Python{nbsp}3 applications>>**, install the
489 LTTng-UST Python agent:
494 # apt-get install python3-lttngust
502 To install LTTng{nbsp}{revision} on Fedora{nbsp}26:
504 . Install the LTTng-tools{nbsp}{revision} and LTTng-UST{nbsp}{revision}
510 # yum install lttng-tools
511 # yum install lttng-ust
515 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
521 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.9.tar.bz2 &&
522 tar -xf lttng-modules-latest-2.9.tar.bz2 &&
523 cd lttng-modules-2.9.* &&
525 sudo make modules_install &&
531 .Java and Python application instrumentation and tracing
533 If you need to instrument and trace <<java-application,Java
534 applications>> on Fedora, you need to build and install
535 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
536 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
537 `--enable-java-agent-all` options to the `configure` script, depending
538 on which Java logging framework you use.
540 If you need to instrument and trace <<python-application,Python
541 applications>> on Fedora, you need to build and install
542 LTTng-UST{nbsp}{revision} from source and pass the
543 `--enable-python-agent` option to the `configure` script.
550 To install LTTng{nbsp}{revision} on Debian "stretch" (stable),
551 Debian "buster" (testing), or Debian "sid" (unstable):
553 . Install the main LTTng{nbsp}{revision} packages:
558 # apt-get install lttng-modules-dkms
559 # apt-get install liblttng-ust-dev
560 # apt-get install lttng-tools
564 . **If you need to instrument and trace <<java-application,Java
565 applications>>**, install the LTTng-UST Java agent:
570 # apt-get install liblttng-ust-agent-java
574 . **If you need to instrument and trace <<python-application,Python
575 applications>>**, install the LTTng-UST Python agent:
580 # apt-get install python3-lttngust
588 To install LTTng{nbsp}{revision} on Arch Linux using
589 https://github.com/rmarquis/pacaur[pacaur]:
591 . Install the main LTTng{nbsp}{revision} packages:
596 $ pacaur -Sy lttng-tools
597 $ pacaur -Sy lttng-ust
598 $ pacaur -Sy lttng-modules
602 . **If you need to instrument and trace <<python-application,Python
603 applications>>**, install the LTTng-UST Python agent:
608 $ pacaur -Sy python-lttngust
609 $ pacaur -Sy python2-lttngust
617 To install LTTng-tools{nbsp}{revision} and LTTng-UST{nbsp}{revision} on
620 . Make sure your system is
621 https://wiki.alpinelinux.org/wiki/Edge[configured for "edge"].
622 . Enable the _testing_ repository by uncommenting the corresponding
623 line in path:{/etc/apk/repositories}.
624 . Add the LTTng packages:
629 # apk add lttng-tools
630 # apk add lttng-ust-dev
634 To install LTTng-modules{nbsp}{revision} (Linux kernel tracing support)
635 on Alpine Linux "edge":
637 . Add the vanilla Linux kernel:
642 # apk add linux-vanilla linux-vanilla-dev
646 . Reboot with the vanilla Linux kernel.
647 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
653 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.9.tar.bz2 &&
654 tar -xf lttng-modules-latest-2.9.tar.bz2 &&
655 cd lttng-modules-2.9.* &&
657 sudo make modules_install &&
663 [[enterprise-distributions]]
664 === RHEL, SUSE, and other enterprise distributions
666 To install LTTng on enterprise Linux distributions, such as Red Hat
667 Enterprise Linux (RHEL) and SUSE Linux Enterprise Server (SUSE), please
668 see http://packages.efficios.com/[EfficiOS Enterprise Packages].
674 To install LTTng{nbsp}{revision} on Buildroot{nbsp}2017.02 or
675 Buildroot{nbsp}2017.05:
677 . Launch the Buildroot configuration tool:
686 . In **Kernel**, check **Linux kernel**.
687 . In **Toolchain**, check **Enable WCHAR support**.
688 . In **Target packages**{nbsp}→ **Debugging, profiling and benchmark**,
689 check **lttng-modules** and **lttng-tools**.
690 . In **Target packages**{nbsp}→ **Libraries**{nbsp}→
691 **Other**, check **lttng-libust**.
695 === OpenEmbedded and Yocto
697 LTTng{nbsp}{revision} recipes are available in the
698 http://layers.openembedded.org/layerindex/branch/master/layer/openembedded-core/[`openembedded-core`]
699 layer for Yocto Project{nbsp}2.3 _Pyro_ under the following names:
705 With BitBake, the simplest way to include LTTng recipes in your target
706 image is to add them to `IMAGE_INSTALL_append` in path:{conf/local.conf}:
709 IMAGE_INSTALL_append = " lttng-tools lttng-modules lttng-ust"
714 . Select a machine and an image recipe.
715 . Click **Edit image recipe**.
716 . Under the **All recipes** tab, search for **lttng**.
717 . Check the desired LTTng recipes.
720 .Java and Python application instrumentation and tracing
722 If you need to instrument and trace <<java-application,Java
723 applications>> on Yocto/OpenEmbedded, you need to build and install
724 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
725 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
726 `--enable-java-agent-all` options to the `configure` script, depending
727 on which Java logging framework you use.
729 If you need to instrument and trace <<python-application,Python
730 applications>> on Yocto/OpenEmbedded, you need to build and install
731 LTTng-UST{nbsp}{revision} from source and pass the
732 `--enable-python-agent` option to the `configure` script.
736 [[building-from-source]]
737 === Build from source
739 To build and install LTTng{nbsp}{revision} from source:
741 . Using your distribution's package manager, or from source, install
742 the following dependencies of LTTng-tools and LTTng-UST:
745 * https://sourceforge.net/projects/libuuid/[libuuid]
746 * http://directory.fsf.org/wiki/Popt[popt]
747 * http://liburcu.org/[Userspace RCU]
748 * http://www.xmlsoft.org/[libxml2]
751 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
757 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.9.tar.bz2 &&
758 tar -xf lttng-modules-latest-2.9.tar.bz2 &&
759 cd lttng-modules-2.9.* &&
761 sudo make modules_install &&
766 . Download, build, and install the latest LTTng-UST{nbsp}{revision}:
772 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.9.tar.bz2 &&
773 tar -xf lttng-ust-latest-2.9.tar.bz2 &&
774 cd lttng-ust-2.9.* &&
784 .Java and Python application tracing
786 If you need to instrument and trace <<java-application,Java
787 applications>>, pass the `--enable-java-agent-jul`,
788 `--enable-java-agent-log4j`, or `--enable-java-agent-all` options to the
789 `configure` script, depending on which Java logging framework you use.
791 If you need to instrument and trace <<python-application,Python
792 applications>>, pass the `--enable-python-agent` option to the
793 `configure` script. You can set the `PYTHON` environment variable to the
794 path to the Python interpreter for which to install the LTTng-UST Python
802 By default, LTTng-UST libraries are installed to
803 dir:{/usr/local/lib}, which is the de facto directory in which to
804 keep self-compiled and third-party libraries.
806 When <<building-tracepoint-providers-and-user-application,linking an
807 instrumented user application with `liblttng-ust`>>:
809 * Append `/usr/local/lib` to the env:LD_LIBRARY_PATH environment
811 * Pass the `-L/usr/local/lib` and `-Wl,-rpath,/usr/local/lib` options to
812 man:gcc(1), man:g++(1), or man:clang(1).
816 . Download, build, and install the latest LTTng-tools{nbsp}{revision}:
822 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.9.tar.bz2 &&
823 tar -xf lttng-tools-latest-2.9.tar.bz2 &&
824 cd lttng-tools-2.9.* &&
832 TIP: The https://github.com/eepp/vlttng[vlttng tool] can do all the
833 previous steps automatically for a given version of LTTng and confine
834 the installed files in a specific directory. This can be useful to test
835 LTTng without installing it on your system.
841 This is a short guide to get started quickly with LTTng kernel and user
844 Before you follow this guide, make sure to <<installing-lttng,install>>
847 This tutorial walks you through the steps to:
849 . <<tracing-the-linux-kernel,Trace the Linux kernel>>.
850 . <<tracing-your-own-user-application,Trace a user application>> written
852 . <<viewing-and-analyzing-your-traces,View and analyze the
856 [[tracing-the-linux-kernel]]
857 === Trace the Linux kernel
859 The following command lines start with the `#` prompt because you need
860 root privileges to trace the Linux kernel. You can also trace the kernel
861 as a regular user if your Unix user is a member of the
862 <<tracing-group,tracing group>>.
864 . Create a <<tracing-session,tracing session>> which writes its traces
865 to dir:{/tmp/my-kernel-trace}:
870 # lttng create my-kernel-session --output=/tmp/my-kernel-trace
874 . List the available kernel tracepoints and system calls:
879 # lttng list --kernel
880 # lttng list --kernel --syscall
884 . Create <<event,event rules>> which match the desired instrumentation
885 point names, for example the `sched_switch` and `sched_process_fork`
886 tracepoints, and the man:open(2) and man:close(2) system calls:
891 # lttng enable-event --kernel sched_switch,sched_process_fork
892 # lttng enable-event --kernel --syscall open,close
896 You can also create an event rule which matches _all_ the Linux kernel
897 tracepoints (this will generate a lot of data when tracing):
902 # lttng enable-event --kernel --all
906 . <<basic-tracing-session-control,Start tracing>>:
915 . Do some operation on your system for a few seconds. For example,
916 load a website, or list the files of a directory.
917 . <<basic-tracing-session-control,Stop tracing>> and destroy the
928 The man:lttng-destroy(1) command does not destroy the trace data; it
929 only destroys the state of the tracing session.
931 . For the sake of this example, make the recorded trace accessible to
937 # chown -R $(whoami) /tmp/my-kernel-trace
941 See <<viewing-and-analyzing-your-traces,View and analyze the
942 recorded events>> to view the recorded events.
945 [[tracing-your-own-user-application]]
946 === Trace a user application
948 This section steps you through a simple example to trace a
949 _Hello world_ program written in C.
951 To create the traceable user application:
953 . Create the tracepoint provider header file, which defines the
954 tracepoints and the events they can generate:
960 #undef TRACEPOINT_PROVIDER
961 #define TRACEPOINT_PROVIDER hello_world
963 #undef TRACEPOINT_INCLUDE
964 #define TRACEPOINT_INCLUDE "./hello-tp.h"
966 #if !defined(_HELLO_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
969 #include <lttng/tracepoint.h>
979 ctf_string(my_string_field, my_string_arg)
980 ctf_integer(int, my_integer_field, my_integer_arg)
984 #endif /* _HELLO_TP_H */
986 #include <lttng/tracepoint-event.h>
990 . Create the tracepoint provider package source file:
996 #define TRACEPOINT_CREATE_PROBES
997 #define TRACEPOINT_DEFINE
999 #include "hello-tp.h"
1003 . Build the tracepoint provider package:
1008 $ gcc -c -I. hello-tp.c
1012 . Create the _Hello World_ application source file:
1019 #include "hello-tp.h"
1021 int main(int argc, char *argv[])
1025 puts("Hello, World!\nPress Enter to continue...");
1028 * The following getchar() call is only placed here for the purpose
1029 * of this demonstration, to pause the application in order for
1030 * you to have time to list its tracepoints. It is not
1036 * A tracepoint() call.
1038 * Arguments, as defined in hello-tp.h:
1040 * 1. Tracepoint provider name (required)
1041 * 2. Tracepoint name (required)
1042 * 3. my_integer_arg (first user-defined argument)
1043 * 4. my_string_arg (second user-defined argument)
1045 * Notice the tracepoint provider and tracepoint names are
1046 * NOT strings: they are in fact parts of variables that the
1047 * macros in hello-tp.h create.
1049 tracepoint(hello_world, my_first_tracepoint, 23, "hi there!");
1051 for (x = 0; x < argc; ++x) {
1052 tracepoint(hello_world, my_first_tracepoint, x, argv[x]);
1055 puts("Quitting now!");
1056 tracepoint(hello_world, my_first_tracepoint, x * x, "x^2");
1063 . Build the application:
1072 . Link the application with the tracepoint provider package,
1073 `liblttng-ust`, and `libdl`:
1078 $ gcc -o hello hello.o hello-tp.o -llttng-ust -ldl
1082 Here's the whole build process:
1085 .User space tracing tutorial's build steps.
1086 image::ust-flow.png[]
1088 To trace the user application:
1090 . Run the application with a few arguments:
1095 $ ./hello world and beyond
1104 Press Enter to continue...
1108 . Start an LTTng <<lttng-sessiond,session daemon>>:
1113 $ lttng-sessiond --daemonize
1117 Note that a session daemon might already be running, for example as
1118 a service that the distribution's service manager started.
1120 . List the available user space tracepoints:
1125 $ lttng list --userspace
1129 You see the `hello_world:my_first_tracepoint` tracepoint listed
1130 under the `./hello` process.
1132 . Create a <<tracing-session,tracing session>>:
1137 $ lttng create my-user-space-session
1141 . Create an <<event,event rule>> which matches the
1142 `hello_world:my_first_tracepoint` event name:
1147 $ lttng enable-event --userspace hello_world:my_first_tracepoint
1151 . <<basic-tracing-session-control,Start tracing>>:
1160 . Go back to the running `hello` application and press Enter. The
1161 program executes all `tracepoint()` instrumentation points and exits.
1162 . <<basic-tracing-session-control,Stop tracing>> and destroy the
1173 The man:lttng-destroy(1) command does not destroy the trace data; it
1174 only destroys the state of the tracing session.
1176 By default, LTTng saves the traces in
1177 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
1178 where +__name__+ is the tracing session name. The
1179 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
1181 See <<viewing-and-analyzing-your-traces,View and analyze the
1182 recorded events>> to view the recorded events.
1185 [[viewing-and-analyzing-your-traces]]
1186 === View and analyze the recorded events
1188 Once you have completed the <<tracing-the-linux-kernel,Trace the Linux
1189 kernel>> and <<tracing-your-own-user-application,Trace a user
1190 application>> tutorials, you can inspect the recorded events.
1192 Many tools are available to read LTTng traces:
1194 * **cmd:babeltrace** is a command-line utility which converts trace
1195 formats; it supports the format that LTTng produces, CTF, as well as a
1196 basic text output which can be ++grep++ed. The cmd:babeltrace command
1197 is part of the http://diamon.org/babeltrace[Babeltrace] project.
1198 * Babeltrace also includes
1199 **https://www.python.org/[Python] bindings** so
1200 that you can easily open and read an LTTng trace with your own script,
1201 benefiting from the power of Python.
1202 * http://tracecompass.org/[**Trace Compass**]
1203 is a graphical user interface for viewing and analyzing any type of
1204 logs or traces, including LTTng's.
1205 * https://github.com/lttng/lttng-analyses[**LTTng analyses**] is a
1206 project which includes many high-level analyses of LTTng kernel
1207 traces, like scheduling statistics, interrupt frequency distribution,
1208 top CPU usage, and more.
1210 NOTE: This section assumes that the traces recorded during the previous
1211 tutorials were saved to their default location, in the
1212 dir:{$LTTNG_HOME/lttng-traces} directory. The env:LTTNG_HOME
1213 environment variable defaults to `$HOME` if not set.
1216 [[viewing-and-analyzing-your-traces-bt]]
1217 ==== Use the cmd:babeltrace command-line tool
1219 The simplest way to list all the recorded events of a trace is to pass
1220 its path to cmd:babeltrace with no options:
1224 $ babeltrace ~/lttng-traces/my-user-space-session*
1227 cmd:babeltrace finds all traces recursively within the given path and
1228 prints all their events, merging them in chronological order.
1230 You can pipe the output of cmd:babeltrace into a tool like man:grep(1) for
1235 $ babeltrace /tmp/my-kernel-trace | grep _switch
1238 You can pipe the output of cmd:babeltrace into a tool like man:wc(1) to
1239 count the recorded events:
1243 $ babeltrace /tmp/my-kernel-trace | grep _open | wc --lines
1247 [[viewing-and-analyzing-your-traces-bt-python]]
1248 ==== Use the Babeltrace Python bindings
1250 The <<viewing-and-analyzing-your-traces-bt,text output of cmd:babeltrace>>
1251 is useful to isolate events by simple matching using man:grep(1) and
1252 similar utilities. However, more elaborate filters, such as keeping only
1253 event records with a field value falling within a specific range, are
1254 not trivial to write using a shell. Moreover, reductions and even the
1255 most basic computations involving multiple event records are virtually
1256 impossible to implement.
1258 Fortunately, Babeltrace ships with Python 3 bindings which makes it easy
1259 to read the event records of an LTTng trace sequentially and compute the
1260 desired information.
1262 The following script accepts an LTTng Linux kernel trace path as its
1263 first argument and prints the short names of the top 5 running processes
1264 on CPU 0 during the whole trace:
1269 from collections import Counter
1275 if len(sys.argv) != 2:
1276 msg = 'Usage: python3 {} TRACEPATH'.format(sys.argv[0])
1277 print(msg, file=sys.stderr)
1280 # A trace collection contains one or more traces
1281 col = babeltrace.TraceCollection()
1283 # Add the trace provided by the user (LTTng traces always have
1285 if col.add_trace(sys.argv[1], 'ctf') is None:
1286 raise RuntimeError('Cannot add trace')
1288 # This counter dict contains execution times:
1290 # task command name -> total execution time (ns)
1291 exec_times = Counter()
1293 # This contains the last `sched_switch` timestamp
1297 for event in col.events:
1298 # Keep only `sched_switch` events
1299 if event.name != 'sched_switch':
1302 # Keep only events which happened on CPU 0
1303 if event['cpu_id'] != 0:
1307 cur_ts = event.timestamp
1313 # Previous task command (short) name
1314 prev_comm = event['prev_comm']
1316 # Initialize entry in our dict if not yet done
1317 if prev_comm not in exec_times:
1318 exec_times[prev_comm] = 0
1320 # Compute previous command execution time
1321 diff = cur_ts - last_ts
1323 # Update execution time of this command
1324 exec_times[prev_comm] += diff
1326 # Update last timestamp
1330 for name, ns in exec_times.most_common(5):
1332 print('{:20}{} s'.format(name, s))
1337 if __name__ == '__main__':
1338 sys.exit(0 if top5proc() else 1)
1345 $ python3 top5proc.py /tmp/my-kernel-trace/kernel
1351 swapper/0 48.607245889 s
1352 chromium 7.192738188 s
1353 pavucontrol 0.709894415 s
1354 Compositor 0.660867933 s
1355 Xorg.bin 0.616753786 s
1358 Note that `swapper/0` is the "idle" process of CPU 0 on Linux; since we
1359 weren't using the CPU that much when tracing, its first position in the
1364 == [[understanding-lttng]]Core concepts
1366 From a user's perspective, the LTTng system is built on a few concepts,
1367 or objects, on which the <<lttng-cli,cmd:lttng command-line tool>>
1368 operates by sending commands to the <<lttng-sessiond,session daemon>>.
1369 Understanding how those objects relate to eachother is key in mastering
1372 The core concepts are:
1374 * <<tracing-session,Tracing session>>
1375 * <<domain,Tracing domain>>
1376 * <<channel,Channel and ring buffer>>
1377 * <<"event","Instrumentation point, event rule, event, and event record">>
1383 A _tracing session_ is a stateful dialogue between you and
1384 a <<lttng-sessiond,session daemon>>. You can
1385 <<creating-destroying-tracing-sessions,create a new tracing
1386 session>> with the `lttng create` command.
1388 Anything that you do when you control LTTng tracers happens within a
1389 tracing session. In particular, a tracing session:
1392 * Has its own set of trace files.
1393 * Has its own state of activity (started or stopped).
1394 * Has its own <<tracing-session-mode,mode>> (local, network streaming,
1396 * Has its own <<channel,channels>> which have their own
1397 <<event,event rules>>.
1400 .A _tracing session_ contains <<channel,channels>> that are members of <<domain,tracing domains>> and contain <<event,event rules>>.
1401 image::concepts.png[]
1403 Those attributes and objects are completely isolated between different
1406 A tracing session is analogous to a cash machine session:
1407 the operations you do on the banking system through the cash machine do
1408 not alter the data of other users of the same system. In the case of
1409 the cash machine, a session lasts as long as your bank card is inside.
1410 In the case of LTTng, a tracing session lasts from the `lttng create`
1411 command to the `lttng destroy` command.
1414 .Each Unix user has its own set of tracing sessions.
1415 image::many-sessions.png[]
1418 [[tracing-session-mode]]
1419 ==== Tracing session mode
1421 LTTng can send the generated trace data to different locations. The
1422 _tracing session mode_ dictates where to send it. The following modes
1423 are available in LTTng{nbsp}{revision}:
1426 LTTng writes the traces to the file system of the machine being traced
1429 Network streaming mode::
1430 LTTng sends the traces over the network to a
1431 <<lttng-relayd,relay daemon>> running on a remote system.
1434 LTTng does not write the traces by default. Instead, you can request
1435 LTTng to <<taking-a-snapshot,take a snapshot>>, that is, a copy of the
1436 current tracing buffers, and to write it to the target's file system
1437 or to send it over the network to a <<lttng-relayd,relay daemon>>
1438 running on a remote system.
1441 This mode is similar to the network streaming mode, but a live
1442 trace viewer can connect to the distant relay daemon to
1443 <<lttng-live,view event records as LTTng generates them>> by
1450 A _tracing domain_ is a namespace for event sources. A tracing domain
1451 has its own properties and features.
1453 There are currently five available tracing domains:
1457 * `java.util.logging` (JUL)
1461 You must specify a tracing domain when using some commands to avoid
1462 ambiguity. For example, since all the domains support named tracepoints
1463 as event sources (instrumentation points that you manually insert in the
1464 source code), you need to specify a tracing domain when
1465 <<enabling-disabling-events,creating an event rule>> because all the
1466 tracing domains could have tracepoints with the same names.
1468 Some features are reserved to specific tracing domains. Dynamic function
1469 entry and return instrumentation points, for example, are currently only
1470 supported in the Linux kernel tracing domain, but support for other
1471 tracing domains could be added in the future.
1473 You can create <<channel,channels>> in the Linux kernel and user space
1474 tracing domains. The other tracing domains have a single default
1479 === Channel and ring buffer
1481 A _channel_ is an object which is responsible for a set of ring buffers.
1482 Each ring buffer is divided into multiple sub-buffers. When an LTTng
1483 tracer emits an event, it can record it to one or more
1484 sub-buffers. The attributes of a channel determine what to do when
1485 there's no space left for a new event record because all sub-buffers
1486 are full, where to send a full sub-buffer, and other behaviours.
1488 A channel is always associated to a <<domain,tracing domain>>. The
1489 `java.util.logging` (JUL), log4j, and Python tracing domains each have
1490 a default channel which you cannot configure.
1492 A channel also owns <<event,event rules>>. When an LTTng tracer emits
1493 an event, it records it to the sub-buffers of all
1494 the enabled channels with a satisfied event rule, as long as those
1495 channels are part of active <<tracing-session,tracing sessions>>.
1498 [[channel-buffering-schemes]]
1499 ==== Per-user vs. per-process buffering schemes
1501 A channel has at least one ring buffer _per CPU_. LTTng always
1502 records an event to the ring buffer associated to the CPU on which it
1505 Two _buffering schemes_ are available when you
1506 <<enabling-disabling-channels,create a channel>> in the
1507 user space <<domain,tracing domain>>:
1509 Per-user buffering::
1510 Allocate one set of ring buffers--one per CPU--shared by all the
1511 instrumented processes of each Unix user.
1515 .Per-user buffering scheme.
1516 image::per-user-buffering.png[]
1519 Per-process buffering::
1520 Allocate one set of ring buffers--one per CPU--for each
1521 instrumented process.
1525 .Per-process buffering scheme.
1526 image::per-process-buffering.png[]
1529 The per-process buffering scheme tends to consume more memory than the
1530 per-user option because systems generally have more instrumented
1531 processes than Unix users running instrumented processes. However, the
1532 per-process buffering scheme ensures that one process having a high
1533 event throughput won't fill all the shared sub-buffers of the same
1536 The Linux kernel tracing domain has only one available buffering scheme
1537 which is to allocate a single set of ring buffers for the whole system.
1538 This scheme is similar to the per-user option, but with a single, global
1539 user "running" the kernel.
1542 [[channel-overwrite-mode-vs-discard-mode]]
1543 ==== Overwrite vs. discard event loss modes
1545 When an event occurs, LTTng records it to a specific sub-buffer (yellow
1546 arc in the following animation) of a specific channel's ring buffer.
1547 When there's no space left in a sub-buffer, the tracer marks it as
1548 consumable (red) and another, empty sub-buffer starts receiving the
1549 following event records. A <<lttng-consumerd,consumer daemon>>
1550 eventually consumes the marked sub-buffer (returns to white).
1553 [role="docsvg-channel-subbuf-anim"]
1558 In an ideal world, sub-buffers are consumed faster than they are filled,
1559 as is the case in the previous animation. In the real world,
1560 however, all sub-buffers can be full at some point, leaving no space to
1561 record the following events.
1563 By design, LTTng is a _non-blocking_ tracer: when no empty sub-buffer is
1564 available, it is acceptable to lose event records when the alternative
1565 would be to cause substantial delays in the instrumented application's
1566 execution. LTTng privileges performance over integrity; it aims at
1567 perturbing the traced system as little as possible in order to make
1568 tracing of subtle race conditions and rare interrupt cascades possible.
1570 When it comes to losing event records because no empty sub-buffer is
1571 available, the channel's _event loss mode_ determines what to do. The
1572 available event loss modes are:
1575 Drop the newest event records until a the tracer
1576 releases a sub-buffer.
1579 Clear the sub-buffer containing the oldest event records and start
1580 writing the newest event records there.
1582 This mode is sometimes called _flight recorder mode_ because it's
1584 https://en.wikipedia.org/wiki/Flight_recorder[flight recorder]:
1585 always keep a fixed amount of the latest data.
1587 Which mechanism you should choose depends on your context: prioritize
1588 the newest or the oldest event records in the ring buffer?
1590 Beware that, in overwrite mode, the tracer abandons a whole sub-buffer
1591 as soon as a there's no space left for a new event record, whereas in
1592 discard mode, the tracer only discards the event record that doesn't
1595 In discard mode, LTTng increments a count of lost event records when
1596 an event record is lost and saves this count to the trace. In
1597 overwrite mode, LTTng keeps no information when it overwrites a
1598 sub-buffer before consuming it.
1600 There are a few ways to decrease your probability of losing event
1602 <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>> shows
1603 how you can fine-une the sub-buffer count and size of a channel to
1604 virtually stop losing event records, though at the cost of greater
1608 [[channel-subbuf-size-vs-subbuf-count]]
1609 ==== Sub-buffer count and size
1611 When you <<enabling-disabling-channels,create a channel>>, you can
1612 set its number of sub-buffers and their size.
1614 Note that there is noticeable CPU overhead introduced when
1615 switching sub-buffers (marking a full one as consumable and switching
1616 to an empty one for the following events to be recorded). Knowing this,
1617 the following list presents a few practical situations along with how
1618 to configure the sub-buffer count and size for them:
1620 * **High event throughput**: In general, prefer bigger sub-buffers to
1621 lower the risk of losing event records.
1623 Having bigger sub-buffers also ensures a lower
1624 <<channel-switch-timer,sub-buffer switching frequency>>.
1626 The number of sub-buffers is only meaningful if you create the channel
1627 in overwrite mode: in this case, if a sub-buffer overwrite happens, the
1628 other sub-buffers are left unaltered.
1630 * **Low event throughput**: In general, prefer smaller sub-buffers
1631 since the risk of losing event records is low.
1633 Because events occur less frequently, the sub-buffer switching frequency
1634 should remain low and thus the tracer's overhead should not be a
1637 * **Low memory system**: If your target system has a low memory
1638 limit, prefer fewer first, then smaller sub-buffers.
1640 Even if the system is limited in memory, you want to keep the
1641 sub-buffers as big as possible to avoid a high sub-buffer switching
1644 Note that LTTng uses http://diamon.org/ctf/[CTF] as its trace format,
1645 which means event data is very compact. For example, the average
1646 LTTng kernel event record weights about 32{nbsp}bytes. Thus, a
1647 sub-buffer size of 1{nbsp}MiB is considered big.
1649 The previous situations highlight the major trade-off between a few big
1650 sub-buffers and more, smaller sub-buffers: sub-buffer switching
1651 frequency vs. how much data is lost in overwrite mode. Assuming a
1652 constant event throughput and using the overwrite mode, the two
1653 following configurations have the same ring buffer total size:
1656 [role="docsvg-channel-subbuf-size-vs-count-anim"]
1661 * **2 sub-buffers of 4{nbsp}MiB each**: Expect a very low sub-buffer
1662 switching frequency, but if a sub-buffer overwrite happens, half of
1663 the event records so far (4{nbsp}MiB) are definitely lost.
1664 * **8 sub-buffers of 1{nbsp}MiB each**: Expect 4{nbsp}times the tracer's
1665 overhead as the previous configuration, but if a sub-buffer
1666 overwrite happens, only the eighth of event records so far are
1669 In discard mode, the sub-buffers count parameter is pointless: use two
1670 sub-buffers and set their size according to the requirements of your
1674 [[channel-switch-timer]]
1675 ==== Switch timer period
1677 The _switch timer period_ is an important configurable attribute of
1678 a channel to ensure periodic sub-buffer flushing.
1680 When the _switch timer_ expires, a sub-buffer switch happens. You can
1681 set the switch timer period attribute when you
1682 <<enabling-disabling-channels,create a channel>> to ensure that event
1683 data is consumed and committed to trace files or to a distant relay
1684 daemon periodically in case of a low event throughput.
1687 [role="docsvg-channel-switch-timer"]
1692 This attribute is also convenient when you use big sub-buffers to cope
1693 with a sporadic high event throughput, even if the throughput is
1697 [[channel-read-timer]]
1698 ==== Read timer period
1700 By default, the LTTng tracers use a notification mechanism to signal a
1701 full sub-buffer so that a consumer daemon can consume it. When such
1702 notifications must be avoided, for example in real-time applications,
1703 you can use the channel's _read timer_ instead. When the read timer
1704 fires, the <<lttng-consumerd,consumer daemon>> checks for full,
1705 consumable sub-buffers.
1708 [[tracefile-rotation]]
1709 ==== Trace file count and size
1711 By default, trace files can grow as large as needed. You can set the
1712 maximum size of each trace file that a channel writes when you
1713 <<enabling-disabling-channels,create a channel>>. When the size of
1714 a trace file reaches the channel's fixed maximum size, LTTng creates
1715 another file to contain the next event records. LTTng appends a file
1716 count to each trace file name in this case.
1718 If you set the trace file size attribute when you create a channel, the
1719 maximum number of trace files that LTTng creates is _unlimited_ by
1720 default. To limit them, you can also set a maximum number of trace
1721 files. When the number of trace files reaches the channel's fixed
1722 maximum count, the oldest trace file is overwritten. This mechanism is
1723 called _trace file rotation_.
1727 === Instrumentation point, event rule, event, and event record
1729 An _event rule_ is a set of conditions which must be **all** satisfied
1730 for LTTng to record an occuring event.
1732 You set the conditions when you <<enabling-disabling-events,create
1735 You always attach an event rule to <<channel,channel>> when you create
1738 When an event passes the conditions of an event rule, LTTng records it
1739 in one of the attached channel's sub-buffers.
1741 The available conditions, as of LTTng{nbsp}{revision}, are:
1743 * The event rule _is enabled_.
1744 * The instrumentation point's type _is{nbsp}T_.
1745 * The instrumentation point's name (sometimes called _event name_)
1746 _matches{nbsp}N_, but _is not{nbsp}E_.
1747 * The instrumentation point's log level _is as severe as{nbsp}L_, or
1748 _is exactly{nbsp}L_.
1749 * The fields of the event's payload _satisfy_ a filter
1750 expression{nbsp}__F__.
1752 As you can see, all the conditions but the dynamic filter are related to
1753 the event rule's status or to the instrumentation point, not to the
1754 occurring events. This is why, without a filter, checking if an event
1755 passes an event rule is not a dynamic task: when you create or modify an
1756 event rule, all the tracers of its tracing domain enable or disable the
1757 instrumentation points themselves once. This is possible because the
1758 attributes of an instrumentation point (type, name, and log level) are
1759 defined statically. In other words, without a dynamic filter, the tracer
1760 _does not evaluate_ the arguments of an instrumentation point unless it
1761 matches an enabled event rule.
1763 Note that, for LTTng to record an event, the <<channel,channel>> to
1764 which a matching event rule is attached must also be enabled, and the
1765 tracing session owning this channel must be active.
1768 .Logical path from an instrumentation point to an event record.
1769 image::event-rule.png[]
1771 .Event, event record, or event rule?
1773 With so many similar terms, it's easy to get confused.
1775 An **event** is the consequence of the execution of an _instrumentation
1776 point_, like a tracepoint that you manually place in some source code,
1777 or a Linux kernel KProbe. An event is said to _occur_ at a specific
1778 time. Different actions can be taken upon the occurrence of an event,
1779 like record the event's payload to a buffer.
1781 An **event record** is the representation of an event in a sub-buffer. A
1782 tracer is responsible for capturing the payload of an event, current
1783 context variables, the event's ID, and the event's timestamp. LTTng
1784 can append this sub-buffer to a trace file.
1786 An **event rule** is a set of conditions which must all be satisfied for
1787 LTTng to record an occuring event. Events still occur without
1788 satisfying event rules, but LTTng does not record them.
1793 == Components of noch:{LTTng}
1795 The second _T_ in _LTTng_ stands for _toolkit_: it would be wrong
1796 to call LTTng a simple _tool_ since it is composed of multiple
1797 interacting components. This section describes those components,
1798 explains their respective roles, and shows how they connect together to
1799 form the LTTng ecosystem.
1801 The following diagram shows how the most important components of LTTng
1802 interact with user applications, the Linux kernel, and you:
1805 .Control and trace data paths between LTTng components.
1806 image::plumbing.png[]
1808 The LTTng project incorporates:
1810 * **LTTng-tools**: Libraries and command-line interface to
1811 control tracing sessions.
1812 ** <<lttng-sessiond,Session daemon>> (man:lttng-sessiond(8)).
1813 ** <<lttng-consumerd,Consumer daemon>> (man:lttng-consumerd(8)).
1814 ** <<lttng-relayd,Relay daemon>> (man:lttng-relayd(8)).
1815 ** <<liblttng-ctl-lttng,Tracing control library>> (`liblttng-ctl`).
1816 ** <<lttng-cli,Tracing control command-line tool>> (man:lttng(1)).
1817 * **LTTng-UST**: Libraries and Java/Python packages to trace user
1819 ** <<lttng-ust,User space tracing library>> (`liblttng-ust`) and its
1820 headers to instrument and trace any native user application.
1821 ** <<prebuilt-ust-helpers,Preloadable user space tracing helpers>>:
1822 *** `liblttng-ust-libc-wrapper`
1823 *** `liblttng-ust-pthread-wrapper`
1824 *** `liblttng-ust-cyg-profile`
1825 *** `liblttng-ust-cyg-profile-fast`
1826 *** `liblttng-ust-dl`
1827 ** User space tracepoint provider source files generator command-line
1828 tool (man:lttng-gen-tp(1)).
1829 ** <<lttng-ust-agents,LTTng-UST Java agent>> to instrument and trace
1830 Java applications using `java.util.logging` or
1831 Apache log4j 1.2 logging.
1832 ** <<lttng-ust-agents,LTTng-UST Python agent>> to instrument
1833 Python applications using the standard `logging` package.
1834 * **LTTng-modules**: <<lttng-modules,Linux kernel modules>> to trace
1836 ** LTTng kernel tracer module.
1837 ** Tracing ring buffer kernel modules.
1838 ** Probe kernel modules.
1839 ** LTTng logger kernel module.
1843 === Tracing control command-line interface
1846 .The tracing control command-line interface.
1847 image::plumbing-lttng-cli.png[]
1849 The _man:lttng(1) command-line tool_ is the standard user interface to
1850 control LTTng <<tracing-session,tracing sessions>>. The cmd:lttng tool
1851 is part of LTTng-tools.
1853 The cmd:lttng tool is linked with
1854 <<liblttng-ctl-lttng,`liblttng-ctl`>> to communicate with
1855 one or more <<lttng-sessiond,session daemons>> behind the scenes.
1857 The cmd:lttng tool has a Git-like interface:
1861 $ lttng <GENERAL OPTIONS> <COMMAND> <COMMAND OPTIONS>
1864 The <<controlling-tracing,Tracing control>> section explores the
1865 available features of LTTng using the cmd:lttng tool.
1868 [[liblttng-ctl-lttng]]
1869 === Tracing control library
1872 .The tracing control library.
1873 image::plumbing-liblttng-ctl.png[]
1875 The _LTTng control library_, `liblttng-ctl`, is used to communicate
1876 with a <<lttng-sessiond,session daemon>> using a C API that hides the
1877 underlying protocol's details. `liblttng-ctl` is part of LTTng-tools.
1879 The <<lttng-cli,cmd:lttng command-line tool>>
1880 is linked with `liblttng-ctl`.
1882 You can use `liblttng-ctl` in C or $$C++$$ source code by including its
1887 #include <lttng/lttng.h>
1890 Some objects are referenced by name (C string), such as tracing
1891 sessions, but most of them require to create a handle first using
1892 `lttng_create_handle()`.
1894 The best available developer documentation for `liblttng-ctl` is, as of
1895 LTTng{nbsp}{revision}, its installed header files. Every function and
1896 structure is thoroughly documented.
1900 === User space tracing library
1903 .The user space tracing library.
1904 image::plumbing-liblttng-ust.png[]
1906 The _user space tracing library_, `liblttng-ust` (see man:lttng-ust(3)),
1907 is the LTTng user space tracer. It receives commands from a
1908 <<lttng-sessiond,session daemon>>, for example to
1909 enable and disable specific instrumentation points, and writes event
1910 records to ring buffers shared with a
1911 <<lttng-consumerd,consumer daemon>>.
1912 `liblttng-ust` is part of LTTng-UST.
1914 Public C header files are installed beside `liblttng-ust` to
1915 instrument any <<c-application,C or $$C++$$ application>>.
1917 <<lttng-ust-agents,LTTng-UST agents>>, which are regular Java and Python
1918 packages, use their own library providing tracepoints which is
1919 linked with `liblttng-ust`.
1921 An application or library does not have to initialize `liblttng-ust`
1922 manually: its constructor does the necessary tasks to properly register
1923 to a session daemon. The initialization phase also enables the
1924 instrumentation points matching the <<event,event rules>> that you
1928 [[lttng-ust-agents]]
1929 === User space tracing agents
1932 .The user space tracing agents.
1933 image::plumbing-lttng-ust-agents.png[]
1935 The _LTTng-UST Java and Python agents_ are regular Java and Python
1936 packages which add LTTng tracing capabilities to the
1937 native logging frameworks. The LTTng-UST agents are part of LTTng-UST.
1939 In the case of Java, the
1940 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[`java.util.logging`
1941 core logging facilities] and
1942 https://logging.apache.org/log4j/1.2/[Apache log4j 1.2] are supported.
1943 Note that Apache Log4{nbsp}2 is not supported.
1945 In the case of Python, the standard
1946 https://docs.python.org/3/library/logging.html[`logging`] package
1947 is supported. Both Python 2 and Python 3 modules can import the
1948 LTTng-UST Python agent package.
1950 The applications using the LTTng-UST agents are in the
1951 `java.util.logging` (JUL),
1952 log4j, and Python <<domain,tracing domains>>.
1954 Both agents use the same mechanism to trace the log statements. When an
1955 agent is initialized, it creates a log handler that attaches to the root
1956 logger. The agent also registers to a <<lttng-sessiond,session daemon>>.
1957 When the application executes a log statement, it is passed to the
1958 agent's log handler by the root logger. The agent's log handler calls a
1959 native function in a tracepoint provider package shared library linked
1960 with <<lttng-ust,`liblttng-ust`>>, passing the formatted log message and
1961 other fields, like its logger name and its log level. This native
1962 function contains a user space instrumentation point, hence tracing the
1965 The log level condition of an
1966 <<event,event rule>> is considered when tracing
1967 a Java or a Python application, and it's compatible with the standard
1968 JUL, log4j, and Python log levels.
1972 === LTTng kernel modules
1975 .The LTTng kernel modules.
1976 image::plumbing-lttng-modules.png[]
1978 The _LTTng kernel modules_ are a set of Linux kernel modules
1979 which implement the kernel tracer of the LTTng project. The LTTng
1980 kernel modules are part of LTTng-modules.
1982 The LTTng kernel modules include:
1984 * A set of _probe_ modules.
1986 Each module attaches to a specific subsystem
1987 of the Linux kernel using its tracepoint instrument points. There are
1988 also modules to attach to the entry and return points of the Linux
1989 system call functions.
1991 * _Ring buffer_ modules.
1993 A ring buffer implementation is provided as kernel modules. The LTTng
1994 kernel tracer writes to the ring buffer; a
1995 <<lttng-consumerd,consumer daemon>> reads from the ring buffer.
1997 * The _LTTng kernel tracer_ module.
1998 * The _LTTng logger_ module.
2000 The LTTng logger module implements the special path:{/proc/lttng-logger}
2001 file so that any executable can generate LTTng events by opening and
2002 writing to this file.
2004 See <<proc-lttng-logger-abi,LTTng logger>>.
2006 Generally, you do not have to load the LTTng kernel modules manually
2007 (using man:modprobe(8), for example): a root <<lttng-sessiond,session
2008 daemon>> loads the necessary modules when starting. If you have extra
2009 probe modules, you can specify to load them to the session daemon on
2012 The LTTng kernel modules are installed in
2013 +/usr/lib/modules/__release__/extra+ by default, where +__release__+ is
2014 the kernel release (see `uname --kernel-release`).
2021 .The session daemon.
2022 image::plumbing-sessiond.png[]
2024 The _session daemon_, man:lttng-sessiond(8), is a daemon responsible for
2025 managing tracing sessions and for controlling the various components of
2026 LTTng. The session daemon is part of LTTng-tools.
2028 The session daemon sends control requests to and receives control
2031 * The <<lttng-ust,user space tracing library>>.
2033 Any instance of the user space tracing library first registers to
2034 a session daemon. Then, the session daemon can send requests to
2035 this instance, such as:
2038 ** Get the list of tracepoints.
2039 ** Share an <<event,event rule>> so that the user space tracing library
2040 can enable or disable tracepoints. Amongst the possible conditions
2041 of an event rule is a filter expression which `liblttng-ust` evalutes
2042 when an event occurs.
2043 ** Share <<channel,channel>> attributes and ring buffer locations.
2046 The session daemon and the user space tracing library use a Unix
2047 domain socket for their communication.
2049 * The <<lttng-ust-agents,user space tracing agents>>.
2051 Any instance of a user space tracing agent first registers to
2052 a session daemon. Then, the session daemon can send requests to
2053 this instance, such as:
2056 ** Get the list of loggers.
2057 ** Enable or disable a specific logger.
2060 The session daemon and the user space tracing agent use a TCP connection
2061 for their communication.
2063 * The <<lttng-modules,LTTng kernel tracer>>.
2064 * The <<lttng-consumerd,consumer daemon>>.
2066 The session daemon sends requests to the consumer daemon to instruct
2067 it where to send the trace data streams, amongst other information.
2069 * The <<lttng-relayd,relay daemon>>.
2071 The session daemon receives commands from the
2072 <<liblttng-ctl-lttng,tracing control library>>.
2074 The root session daemon loads the appropriate
2075 <<lttng-modules,LTTng kernel modules>> on startup. It also spawns
2076 a <<lttng-consumerd,consumer daemon>> as soon as you create
2077 an <<event,event rule>>.
2079 The session daemon does not send and receive trace data: this is the
2080 role of the <<lttng-consumerd,consumer daemon>> and
2081 <<lttng-relayd,relay daemon>>. It does, however, generate the
2082 http://diamon.org/ctf/[CTF] metadata stream.
2084 Each Unix user can have its own session daemon instance. The
2085 tracing sessions managed by different session daemons are completely
2088 The root user's session daemon is the only one which is
2089 allowed to control the LTTng kernel tracer, and its spawned consumer
2090 daemon is the only one which is allowed to consume trace data from the
2091 LTTng kernel tracer. Note, however, that any Unix user which is a member
2092 of the <<tracing-group,tracing group>> is allowed
2093 to create <<channel,channels>> in the
2094 Linux kernel <<domain,tracing domain>>, and thus to trace the Linux
2097 The <<lttng-cli,cmd:lttng command-line tool>> automatically starts a
2098 session daemon when using its `create` command if none is currently
2099 running. You can also start the session daemon manually.
2106 .The consumer daemon.
2107 image::plumbing-consumerd.png[]
2109 The _consumer daemon_, man:lttng-consumerd(8), is a daemon which shares
2110 ring buffers with user applications or with the LTTng kernel modules to
2111 collect trace data and send it to some location (on disk or to a
2112 <<lttng-relayd,relay daemon>> over the network). The consumer daemon
2113 is part of LTTng-tools.
2115 You do not start a consumer daemon manually: a consumer daemon is always
2116 spawned by a <<lttng-sessiond,session daemon>> as soon as you create an
2117 <<event,event rule>>, that is, before you start tracing. When you kill
2118 its owner session daemon, the consumer daemon also exits because it is
2119 the session daemon's child process. Command-line options of
2120 man:lttng-sessiond(8) target the consumer daemon process.
2122 There are up to two running consumer daemons per Unix user, whereas only
2123 one session daemon can run per user. This is because each process can be
2124 either 32-bit or 64-bit: if the target system runs a mixture of 32-bit
2125 and 64-bit processes, it is more efficient to have separate
2126 corresponding 32-bit and 64-bit consumer daemons. The root user is an
2127 exception: it can have up to _three_ running consumer daemons: 32-bit
2128 and 64-bit instances for its user applications, and one more
2129 reserved for collecting kernel trace data.
2137 image::plumbing-relayd.png[]
2139 The _relay daemon_, man:lttng-relayd(8), is a daemon acting as a bridge
2140 between remote session and consumer daemons, local trace files, and a
2141 remote live trace viewer. The relay daemon is part of LTTng-tools.
2143 The main purpose of the relay daemon is to implement a receiver of
2144 <<sending-trace-data-over-the-network,trace data over the network>>.
2145 This is useful when the target system does not have much file system
2146 space to record trace files locally.
2148 The relay daemon is also a server to which a
2149 <<lttng-live,live trace viewer>> can
2150 connect. The live trace viewer sends requests to the relay daemon to
2151 receive trace data as the target system emits events. The
2152 communication protocol is named _LTTng live_; it is used over TCP
2155 Note that you can start the relay daemon on the target system directly.
2156 This is the setup of choice when the use case is to view events as
2157 the target system emits them without the need of a remote system.
2161 == [[using-lttng]]Instrumentation
2163 There are many examples of tracing and monitoring in our everyday life:
2165 * You have access to real-time and historical weather reports and
2166 forecasts thanks to weather stations installed around the country.
2167 * You know your heart is safe thanks to an electrocardiogram.
2168 * You make sure not to drive your car too fast and to have enough fuel
2169 to reach your destination thanks to gauges visible on your dashboard.
2171 All the previous examples have something in common: they rely on
2172 **instruments**. Without the electrodes attached to the surface of your
2173 body's skin, cardiac monitoring is futile.
2175 LTTng, as a tracer, is no different from those real life examples. If
2176 you're about to trace a software system or, in other words, record its
2177 history of execution, you better have **instrumentation points** in the
2178 subject you're tracing, that is, the actual software.
2180 Various ways were developed to instrument a piece of software for LTTng
2181 tracing. The most straightforward one is to manually place
2182 instrumentation points, called _tracepoints_, in the software's source
2183 code. It is also possible to add instrumentation points dynamically in
2184 the Linux kernel <<domain,tracing domain>>.
2186 If you're only interested in tracing the Linux kernel, your
2187 instrumentation needs are probably already covered by LTTng's built-in
2188 <<lttng-modules,Linux kernel tracepoints>>. You may also wish to trace a
2189 user application which is already instrumented for LTTng tracing.
2190 In such cases, you can skip this whole section and read the topics of
2191 the <<controlling-tracing,Tracing control>> section.
2193 Many methods are available to instrument a piece of software for LTTng
2196 * <<c-application,User space instrumentation for C and $$C++$$
2198 * <<prebuilt-ust-helpers,Prebuilt user space tracing helpers>>.
2199 * <<java-application,User space Java agent>>.
2200 * <<python-application,User space Python agent>>.
2201 * <<proc-lttng-logger-abi,LTTng logger>>.
2202 * <<instrumenting-linux-kernel,LTTng kernel tracepoints>>.
2206 === [[cxx-application]]User space instrumentation for C and $$C++$$ applications
2208 The procedure to instrument a C or $$C++$$ user application with
2209 the <<lttng-ust,LTTng user space tracing library>>, `liblttng-ust`, is:
2211 . <<tracepoint-provider,Create the source files of a tracepoint provider
2213 . <<probing-the-application-source-code,Add tracepoints to
2214 the application's source code>>.
2215 . <<building-tracepoint-providers-and-user-application,Build and link
2216 a tracepoint provider package and the user application>>.
2218 If you need quick, man:printf(3)-like instrumentation, you can skip
2219 those steps and use <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>>
2222 IMPORTANT: You need to <<installing-lttng,install>> LTTng-UST to
2223 instrument a user application with `liblttng-ust`.
2226 [[tracepoint-provider]]
2227 ==== Create the source files of a tracepoint provider package
2229 A _tracepoint provider_ is a set of compiled functions which provide
2230 **tracepoints** to an application, the type of instrumentation point
2231 supported by LTTng-UST. Those functions can emit events with
2232 user-defined fields and serialize those events as event records to one
2233 or more LTTng-UST <<channel,channel>> sub-buffers. The `tracepoint()`
2234 macro, which you <<probing-the-application-source-code,insert in a user
2235 application's source code>>, calls those functions.
2237 A _tracepoint provider package_ is an object file (`.o`) or a shared
2238 library (`.so`) which contains one or more tracepoint providers.
2239 Its source files are:
2241 * One or more <<tpp-header,tracepoint provider header>> (`.h`).
2242 * A <<tpp-source,tracepoint provider package source>> (`.c`).
2244 A tracepoint provider package is dynamically linked with `liblttng-ust`,
2245 the LTTng user space tracer, at run time.
2248 .User application linked with `liblttng-ust` and containing a tracepoint provider.
2249 image::ust-app.png[]
2251 NOTE: If you need quick, man:printf(3)-like instrumentation, you can
2252 skip creating and using a tracepoint provider and use
2253 <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>> instead.
2257 ===== Create a tracepoint provider header file template
2259 A _tracepoint provider header file_ contains the tracepoint
2260 definitions of a tracepoint provider.
2262 To create a tracepoint provider header file:
2264 . Start from this template:
2268 .Tracepoint provider header file template (`.h` file extension).
2270 #undef TRACEPOINT_PROVIDER
2271 #define TRACEPOINT_PROVIDER provider_name
2273 #undef TRACEPOINT_INCLUDE
2274 #define TRACEPOINT_INCLUDE "./tp.h"
2276 #if !defined(_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
2279 #include <lttng/tracepoint.h>
2282 * Use TRACEPOINT_EVENT(), TRACEPOINT_EVENT_CLASS(),
2283 * TRACEPOINT_EVENT_INSTANCE(), and TRACEPOINT_LOGLEVEL() here.
2288 #include <lttng/tracepoint-event.h>
2294 * `provider_name` with the name of your tracepoint provider.
2295 * `"tp.h"` with the name of your tracepoint provider header file.
2297 . Below the `#include <lttng/tracepoint.h>` line, put your
2298 <<defining-tracepoints,tracepoint definitions>>.
2300 Your tracepoint provider name must be unique amongst all the possible
2301 tracepoint provider names used on the same target system. We
2302 suggest to include the name of your project or company in the name,
2303 for example, `org_lttng_my_project_tpp`.
2305 TIP: [[lttng-gen-tp]]You can use the man:lttng-gen-tp(1) tool to create
2306 this boilerplate for you. When using cmd:lttng-gen-tp, all you need to
2307 write are the <<defining-tracepoints,tracepoint definitions>>.
2310 [[defining-tracepoints]]
2311 ===== Create a tracepoint definition
2313 A _tracepoint definition_ defines, for a given tracepoint:
2315 * Its **input arguments**. They are the macro parameters that the
2316 `tracepoint()` macro accepts for this particular tracepoint
2317 in the user application's source code.
2318 * Its **output event fields**. They are the sources of event fields
2319 that form the payload of any event that the execution of the
2320 `tracepoint()` macro emits for this particular tracepoint.
2322 You can create a tracepoint definition by using the
2323 `TRACEPOINT_EVENT()` macro below the `#include <lttng/tracepoint.h>`
2325 <<tpp-header,tracepoint provider header file template>>.
2327 The syntax of the `TRACEPOINT_EVENT()` macro is:
2330 .`TRACEPOINT_EVENT()` macro syntax.
2333 /* Tracepoint provider name */
2336 /* Tracepoint name */
2339 /* Input arguments */
2344 /* Output event fields */
2353 * `provider_name` with your tracepoint provider name.
2354 * `tracepoint_name` with your tracepoint name.
2355 * `arguments` with the <<tpp-def-input-args,input arguments>>.
2356 * `fields` with the <<tpp-def-output-fields,output event field>>
2359 This tracepoint emits events named `provider_name:tracepoint_name`.
2362 .Event name's length limitation
2364 The concatenation of the tracepoint provider name and the
2365 tracepoint name must not exceed **254 characters**. If it does, the
2366 instrumented application compiles and runs, but LTTng throws multiple
2367 warnings and you could experience serious issues.
2370 [[tpp-def-input-args]]The syntax of the `TP_ARGS()` macro is:
2373 .`TP_ARGS()` macro syntax.
2382 * `type` with the C type of the argument.
2383 * `arg_name` with the argument name.
2385 You can repeat `type` and `arg_name` up to 10 times to have
2386 more than one argument.
2388 .`TP_ARGS()` usage with three arguments.
2400 The `TP_ARGS()` and `TP_ARGS(void)` forms are valid to create a
2401 tracepoint definition with no input arguments.
2403 [[tpp-def-output-fields]]The `TP_FIELDS()` macro contains a list of
2404 `ctf_*()` macros. Each `ctf_*()` macro defines one event field. See
2405 man:lttng-ust(3) for a complete description of the available `ctf_*()`
2406 macros. A `ctf_*()` macro specifies the type, size, and byte order of
2409 Each `ctf_*()` macro takes an _argument expression_ parameter. This is a
2410 C expression that the tracer evalutes at the `tracepoint()` macro site
2411 in the application's source code. This expression provides a field's
2412 source of data. The argument expression can include input argument names
2413 listed in the `TP_ARGS()` macro.
2415 Each `ctf_*()` macro also takes a _field name_ parameter. Field names
2416 must be unique within a given tracepoint definition.
2418 Here's a complete tracepoint definition example:
2420 .Tracepoint definition.
2422 The following tracepoint definition defines a tracepoint which takes
2423 three input arguments and has four output event fields.
2427 #include "my-custom-structure.h"
2433 const struct my_custom_structure*, my_custom_structure,
2438 ctf_string(query_field, query)
2439 ctf_float(double, ratio_field, ratio)
2440 ctf_integer(int, recv_size, my_custom_structure->recv_size)
2441 ctf_integer(int, send_size, my_custom_structure->send_size)
2446 You can refer to this tracepoint definition with the `tracepoint()`
2447 macro in your application's source code like this:
2451 tracepoint(my_provider, my_tracepoint,
2452 my_structure, some_ratio, the_query);
2456 NOTE: The LTTng tracer only evaluates tracepoint arguments at run time
2457 if they satisfy an enabled <<event,event rule>>.
2460 [[using-tracepoint-classes]]
2461 ===== Use a tracepoint class
2463 A _tracepoint class_ is a class of tracepoints which share the same
2464 output event field definitions. A _tracepoint instance_ is one
2465 instance of such a defined tracepoint class, with its own tracepoint
2468 The <<defining-tracepoints,`TRACEPOINT_EVENT()` macro>> is actually a
2469 shorthand which defines both a tracepoint class and a tracepoint
2470 instance at the same time.
2472 When you build a tracepoint provider package, the C or $$C++$$ compiler
2473 creates one serialization function for each **tracepoint class**. A
2474 serialization function is responsible for serializing the event fields
2475 of a tracepoint to a sub-buffer when tracing.
2477 For various performance reasons, when your situation requires multiple
2478 tracepoint definitions with different names, but with the same event
2479 fields, we recommend that you manually create a tracepoint class
2480 and instantiate as many tracepoint instances as needed. One positive
2481 effect of such a design, amongst other advantages, is that all
2482 tracepoint instances of the same tracepoint class reuse the same
2483 serialization function, thus reducing
2484 https://en.wikipedia.org/wiki/Cache_pollution[cache pollution].
2486 .Use a tracepoint class and tracepoint instances.
2488 Consider the following three tracepoint definitions:
2500 ctf_integer(int, userid, userid)
2501 ctf_integer(size_t, len, len)
2513 ctf_integer(int, userid, userid)
2514 ctf_integer(size_t, len, len)
2526 ctf_integer(int, userid, userid)
2527 ctf_integer(size_t, len, len)
2532 In this case, we create three tracepoint classes, with one implicit
2533 tracepoint instance for each of them: `get_account`, `get_settings`, and
2534 `get_transaction`. However, they all share the same event field names
2535 and types. Hence three identical, yet independent serialization
2536 functions are created when you build the tracepoint provider package.
2538 A better design choice is to define a single tracepoint class and three
2539 tracepoint instances:
2543 /* The tracepoint class */
2544 TRACEPOINT_EVENT_CLASS(
2545 /* Tracepoint provider name */
2548 /* Tracepoint class name */
2551 /* Input arguments */
2557 /* Output event fields */
2559 ctf_integer(int, userid, userid)
2560 ctf_integer(size_t, len, len)
2564 /* The tracepoint instances */
2565 TRACEPOINT_EVENT_INSTANCE(
2566 /* Tracepoint provider name */
2569 /* Tracepoint class name */
2572 /* Tracepoint name */
2575 /* Input arguments */
2581 TRACEPOINT_EVENT_INSTANCE(
2590 TRACEPOINT_EVENT_INSTANCE(
2603 [[assigning-log-levels]]
2604 ===== Assign a log level to a tracepoint definition
2606 You can assign an optional _log level_ to a
2607 <<defining-tracepoints,tracepoint definition>>.
2609 Assigning different levels of severity to tracepoint definitions can
2610 be useful: when you <<enabling-disabling-events,create an event rule>>,
2611 you can target tracepoints having a log level as severe as a specific
2614 The concept of LTTng-UST log levels is similar to the levels found
2615 in typical logging frameworks:
2617 * In a logging framework, the log level is given by the function
2618 or method name you use at the log statement site: `debug()`,
2619 `info()`, `warn()`, `error()`, and so on.
2620 * In LTTng-UST, you statically assign the log level to a tracepoint
2621 definition; any `tracepoint()` macro invocation which refers to
2622 this definition has this log level.
2624 You can assign a log level to a tracepoint definition with the
2625 `TRACEPOINT_LOGLEVEL()` macro. You must use this macro _after_ the
2626 <<defining-tracepoints,`TRACEPOINT_EVENT()`>> or
2627 <<using-tracepoint-classes,`TRACEPOINT_INSTANCE()`>> macro for a given
2630 The syntax of the `TRACEPOINT_LOGLEVEL()` macro is:
2633 .`TRACEPOINT_LOGLEVEL()` macro syntax.
2635 TRACEPOINT_LOGLEVEL(provider_name, tracepoint_name, log_level)
2640 * `provider_name` with the tracepoint provider name.
2641 * `tracepoint_name` with the tracepoint name.
2642 * `log_level` with the log level to assign to the tracepoint
2643 definition named `tracepoint_name` in the `provider_name`
2644 tracepoint provider.
2646 See man:lttng-ust(3) for a list of available log level names.
2648 .Assign the `TRACE_DEBUG_UNIT` log level to a tracepoint definition.
2652 /* Tracepoint definition */
2661 ctf_integer(int, userid, userid)
2662 ctf_integer(size_t, len, len)
2666 /* Log level assignment */
2667 TRACEPOINT_LOGLEVEL(my_app, get_transaction, TRACE_DEBUG_UNIT)
2673 ===== Create a tracepoint provider package source file
2675 A _tracepoint provider package source file_ is a C source file which
2676 includes a <<tpp-header,tracepoint provider header file>> to expand its
2677 macros into event serialization and other functions.
2679 You can always use the following tracepoint provider package source
2683 .Tracepoint provider package source file template.
2685 #define TRACEPOINT_CREATE_PROBES
2690 Replace `tp.h` with the name of your <<tpp-header,tracepoint provider
2691 header file>> name. You may also include more than one tracepoint
2692 provider header file here to create a tracepoint provider package
2693 holding more than one tracepoint providers.
2696 [[probing-the-application-source-code]]
2697 ==== Add tracepoints to an application's source code
2699 Once you <<tpp-header,create a tracepoint provider header file>>, you
2700 can use the `tracepoint()` macro in your application's
2701 source code to insert the tracepoints that this header
2702 <<defining-tracepoints,defines>>.
2704 The `tracepoint()` macro takes at least two parameters: the tracepoint
2705 provider name and the tracepoint name. The corresponding tracepoint
2706 definition defines the other parameters.
2708 .`tracepoint()` usage.
2710 The following <<defining-tracepoints,tracepoint definition>> defines a
2711 tracepoint which takes two input arguments and has two output event
2715 .Tracepoint provider header file.
2717 #include "my-custom-structure.h"
2724 const char*, cmd_name
2727 ctf_string(cmd_name, cmd_name)
2728 ctf_integer(int, number_of_args, argc)
2733 You can refer to this tracepoint definition with the `tracepoint()`
2734 macro in your application's source code like this:
2737 .Application's source file.
2741 int main(int argc, char* argv[])
2743 tracepoint(my_provider, my_tracepoint, argc, argv[0]);
2749 Note how the application's source code includes
2750 the tracepoint provider header file containing the tracepoint
2751 definitions to use, path:{tp.h}.
2754 .`tracepoint()` usage with a complex tracepoint definition.
2756 Consider this complex tracepoint definition, where multiple event
2757 fields refer to the same input arguments in their argument expression
2761 .Tracepoint provider header file.
2763 /* For `struct stat` */
2764 #include <sys/types.h>
2765 #include <sys/stat.h>
2777 ctf_integer(int, my_constant_field, 23 + 17)
2778 ctf_integer(int, my_int_arg_field, my_int_arg)
2779 ctf_integer(int, my_int_arg_field2, my_int_arg * my_int_arg)
2780 ctf_integer(int, sum4_field, my_str_arg[0] + my_str_arg[1] +
2781 my_str_arg[2] + my_str_arg[3])
2782 ctf_string(my_str_arg_field, my_str_arg)
2783 ctf_integer_hex(off_t, size_field, st->st_size)
2784 ctf_float(double, size_dbl_field, (double) st->st_size)
2785 ctf_sequence_text(char, half_my_str_arg_field, my_str_arg,
2786 size_t, strlen(my_str_arg) / 2)
2791 You can refer to this tracepoint definition with the `tracepoint()`
2792 macro in your application's source code like this:
2795 .Application's source file.
2797 #define TRACEPOINT_DEFINE
2804 stat("/etc/fstab", &s);
2805 tracepoint(my_provider, my_tracepoint, 23, "Hello, World!", &s);
2811 If you look at the event record that LTTng writes when tracing this
2812 program, assuming the file size of path:{/etc/fstab} is 301{nbsp}bytes,
2813 it should look like this:
2815 .Event record fields
2817 |Field's name |Field's value
2818 |`my_constant_field` |40
2819 |`my_int_arg_field` |23
2820 |`my_int_arg_field2` |529
2822 |`my_str_arg_field` |`Hello, World!`
2823 |`size_field` |0x12d
2824 |`size_dbl_field` |301.0
2825 |`half_my_str_arg_field` |`Hello,`
2829 Sometimes, the arguments you pass to `tracepoint()` are expensive to
2830 compute--they use the call stack, for example. To avoid this
2831 computation when the tracepoint is disabled, you can use the
2832 `tracepoint_enabled()` and `do_tracepoint()` macros.
2834 The syntax of the `tracepoint_enabled()` and `do_tracepoint()` macros
2838 .`tracepoint_enabled()` and `do_tracepoint()` macros syntax.
2840 tracepoint_enabled(provider_name, tracepoint_name)
2841 do_tracepoint(provider_name, tracepoint_name, ...)
2846 * `provider_name` with the tracepoint provider name.
2847 * `tracepoint_name` with the tracepoint name.
2849 `tracepoint_enabled()` returns a non-zero value if the tracepoint named
2850 `tracepoint_name` from the provider named `provider_name` is enabled
2853 `do_tracepoint()` is like `tracepoint()`, except that it doesn't check
2854 if the tracepoint is enabled. Using `tracepoint()` with
2855 `tracepoint_enabled()` is dangerous since `tracepoint()` also contains
2856 the `tracepoint_enabled()` check, thus a race condition is
2857 possible in this situation:
2860 .Possible race condition when using `tracepoint_enabled()` with `tracepoint()`.
2862 if (tracepoint_enabled(my_provider, my_tracepoint)) {
2863 stuff = prepare_stuff();
2866 tracepoint(my_provider, my_tracepoint, stuff);
2869 If the tracepoint is enabled after the condition, then `stuff` is not
2870 prepared: the emitted event will either contain wrong data, or the whole
2871 application could crash (segmentation fault, for example).
2873 NOTE: Neither `tracepoint_enabled()` nor `do_tracepoint()` have an
2874 `STAP_PROBEV()` call. If you need it, you must emit
2878 [[building-tracepoint-providers-and-user-application]]
2879 ==== Build and link a tracepoint provider package and an application
2881 Once you have one or more <<tpp-header,tracepoint provider header
2882 files>> and a <<tpp-source,tracepoint provider package source file>>,
2883 you can create the tracepoint provider package by compiling its source
2884 file. From here, multiple build and run scenarios are possible. The
2885 following table shows common application and library configurations
2886 along with the required command lines to achieve them.
2888 In the following diagrams, we use the following file names:
2891 Executable application.
2894 Application's object file.
2897 Tracepoint provider package object file.
2900 Tracepoint provider package archive file.
2903 Tracepoint provider package shared object file.
2906 User library object file.
2909 User library shared object file.
2911 We use the following symbols in the diagrams of table below:
2914 .Symbols used in the build scenario diagrams.
2915 image::ust-sit-symbols.png[]
2917 We assume that path:{.} is part of the env:LD_LIBRARY_PATH environment
2918 variable in the following instructions.
2920 [role="growable ust-scenarios",cols="asciidoc,asciidoc"]
2921 .Common tracepoint provider package scenarios.
2923 |Scenario |Instructions
2926 The instrumented application is statically linked with
2927 the tracepoint provider package object.
2929 image::ust-sit+app-linked-with-tp-o+app-instrumented.png[]
2932 include::../common/ust-sit-step-tp-o.txt[]
2934 To build the instrumented application:
2936 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2941 #define TRACEPOINT_DEFINE
2945 . Compile the application source file:
2954 . Build the application:
2959 $ gcc -o app app.o tpp.o -llttng-ust -ldl
2963 To run the instrumented application:
2965 * Start the application:
2975 The instrumented application is statically linked with the
2976 tracepoint provider package archive file.
2978 image::ust-sit+app-linked-with-tp-a+app-instrumented.png[]
2981 To create the tracepoint provider package archive file:
2983 . Compile the <<tpp-source,tracepoint provider package source file>>:
2992 . Create the tracepoint provider package archive file:
2997 $ ar rcs tpp.a tpp.o
3001 To build the instrumented application:
3003 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3008 #define TRACEPOINT_DEFINE
3012 . Compile the application source file:
3021 . Build the application:
3026 $ gcc -o app app.o tpp.a -llttng-ust -ldl
3030 To run the instrumented application:
3032 * Start the application:
3042 The instrumented application is linked with the tracepoint provider
3043 package shared object.
3045 image::ust-sit+app-linked-with-tp-so+app-instrumented.png[]
3048 include::../common/ust-sit-step-tp-so.txt[]
3050 To build the instrumented application:
3052 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3057 #define TRACEPOINT_DEFINE
3061 . Compile the application source file:
3070 . Build the application:
3075 $ gcc -o app app.o -ldl -L. -ltpp
3079 To run the instrumented application:
3081 * Start the application:
3091 The tracepoint provider package shared object is preloaded before the
3092 instrumented application starts.
3094 image::ust-sit+tp-so-preloaded+app-instrumented.png[]
3097 include::../common/ust-sit-step-tp-so.txt[]
3099 To build the instrumented application:
3101 . In path:{app.c}, before including path:{tpp.h}, add the
3107 #define TRACEPOINT_DEFINE
3108 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3112 . Compile the application source file:
3121 . Build the application:
3126 $ gcc -o app app.o -ldl
3130 To run the instrumented application with tracing support:
3132 * Preload the tracepoint provider package shared object and
3133 start the application:
3138 $ LD_PRELOAD=./libtpp.so ./app
3142 To run the instrumented application without tracing support:
3144 * Start the application:
3154 The instrumented application dynamically loads the tracepoint provider
3155 package shared object.
3157 See the <<dlclose-warning,warning about `dlclose()`>>.
3159 image::ust-sit+app-dlopens-tp-so+app-instrumented.png[]
3162 include::../common/ust-sit-step-tp-so.txt[]
3164 To build the instrumented application:
3166 . In path:{app.c}, before including path:{tpp.h}, add the
3172 #define TRACEPOINT_DEFINE
3173 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3177 . Compile the application source file:
3186 . Build the application:
3191 $ gcc -o app app.o -ldl
3195 To run the instrumented application:
3197 * Start the application:
3207 The application is linked with the instrumented user library.
3209 The instrumented user library is statically linked with the tracepoint
3210 provider package object file.
3212 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-o+lib-instrumented.png[]
3215 include::../common/ust-sit-step-tp-o-fpic.txt[]
3217 To build the instrumented user library:
3219 . In path:{emon.c}, before including path:{tpp.h}, add the
3225 #define TRACEPOINT_DEFINE
3229 . Compile the user library source file:
3234 $ gcc -I. -fpic -c emon.c
3238 . Build the user library shared object:
3243 $ gcc -shared -o libemon.so emon.o tpp.o -llttng-ust -ldl
3247 To build the application:
3249 . Compile the application source file:
3258 . Build the application:
3263 $ gcc -o app app.o -L. -lemon
3267 To run the application:
3269 * Start the application:
3279 The application is linked with the instrumented user library.
3281 The instrumented user library is linked with the tracepoint provider
3282 package shared object.
3284 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3287 include::../common/ust-sit-step-tp-so.txt[]
3289 To build the instrumented user library:
3291 . In path:{emon.c}, before including path:{tpp.h}, add the
3297 #define TRACEPOINT_DEFINE
3301 . Compile the user library source file:
3306 $ gcc -I. -fpic -c emon.c
3310 . Build the user library shared object:
3315 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3319 To build the application:
3321 . Compile the application source file:
3330 . Build the application:
3335 $ gcc -o app app.o -L. -lemon
3339 To run the application:
3341 * Start the application:
3351 The tracepoint provider package shared object is preloaded before the
3354 The application is linked with the instrumented user library.
3356 image::ust-sit+tp-so-preloaded+app-linked-with-lib+lib-instrumented.png[]
3359 include::../common/ust-sit-step-tp-so.txt[]
3361 To build the instrumented user library:
3363 . In path:{emon.c}, before including path:{tpp.h}, add the
3369 #define TRACEPOINT_DEFINE
3370 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3374 . Compile the user library source file:
3379 $ gcc -I. -fpic -c emon.c
3383 . Build the user library shared object:
3388 $ gcc -shared -o libemon.so emon.o -ldl
3392 To build the application:
3394 . Compile the application source file:
3403 . Build the application:
3408 $ gcc -o app app.o -L. -lemon
3412 To run the application with tracing support:
3414 * Preload the tracepoint provider package shared object and
3415 start the application:
3420 $ LD_PRELOAD=./libtpp.so ./app
3424 To run the application without tracing support:
3426 * Start the application:
3436 The application is linked with the instrumented user library.
3438 The instrumented user library dynamically loads the tracepoint provider
3439 package shared object.
3441 See the <<dlclose-warning,warning about `dlclose()`>>.
3443 image::ust-sit+app-linked-with-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3446 include::../common/ust-sit-step-tp-so.txt[]
3448 To build the instrumented user library:
3450 . In path:{emon.c}, before including path:{tpp.h}, add the
3456 #define TRACEPOINT_DEFINE
3457 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3461 . Compile the user library source file:
3466 $ gcc -I. -fpic -c emon.c
3470 . Build the user library shared object:
3475 $ gcc -shared -o libemon.so emon.o -ldl
3479 To build the application:
3481 . Compile the application source file:
3490 . Build the application:
3495 $ gcc -o app app.o -L. -lemon
3499 To run the application:
3501 * Start the application:
3511 The application dynamically loads the instrumented user library.
3513 The instrumented user library is linked with the tracepoint provider
3514 package shared object.
3516 See the <<dlclose-warning,warning about `dlclose()`>>.
3518 image::ust-sit+app-dlopens-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3521 include::../common/ust-sit-step-tp-so.txt[]
3523 To build the instrumented user library:
3525 . In path:{emon.c}, before including path:{tpp.h}, add the
3531 #define TRACEPOINT_DEFINE
3535 . Compile the user library source file:
3540 $ gcc -I. -fpic -c emon.c
3544 . Build the user library shared object:
3549 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3553 To build the application:
3555 . Compile the application source file:
3564 . Build the application:
3569 $ gcc -o app app.o -ldl -L. -lemon
3573 To run the application:
3575 * Start the application:
3585 The application dynamically loads the instrumented user library.
3587 The instrumented user library dynamically loads the tracepoint provider
3588 package shared object.
3590 See the <<dlclose-warning,warning about `dlclose()`>>.
3592 image::ust-sit+app-dlopens-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3595 include::../common/ust-sit-step-tp-so.txt[]
3597 To build the instrumented user library:
3599 . In path:{emon.c}, before including path:{tpp.h}, add the
3605 #define TRACEPOINT_DEFINE
3606 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3610 . Compile the user library source file:
3615 $ gcc -I. -fpic -c emon.c
3619 . Build the user library shared object:
3624 $ gcc -shared -o libemon.so emon.o -ldl
3628 To build the application:
3630 . Compile the application source file:
3639 . Build the application:
3644 $ gcc -o app app.o -ldl -L. -lemon
3648 To run the application:
3650 * Start the application:
3660 The tracepoint provider package shared object is preloaded before the
3663 The application dynamically loads the instrumented user library.
3665 image::ust-sit+tp-so-preloaded+app-dlopens-lib+lib-instrumented.png[]
3668 include::../common/ust-sit-step-tp-so.txt[]
3670 To build the instrumented user library:
3672 . In path:{emon.c}, before including path:{tpp.h}, add the
3678 #define TRACEPOINT_DEFINE
3679 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3683 . Compile the user library source file:
3688 $ gcc -I. -fpic -c emon.c
3692 . Build the user library shared object:
3697 $ gcc -shared -o libemon.so emon.o -ldl
3701 To build the application:
3703 . Compile the application source file:
3712 . Build the application:
3717 $ gcc -o app app.o -L. -lemon
3721 To run the application with tracing support:
3723 * Preload the tracepoint provider package shared object and
3724 start the application:
3729 $ LD_PRELOAD=./libtpp.so ./app
3733 To run the application without tracing support:
3735 * Start the application:
3745 The application is statically linked with the tracepoint provider
3746 package object file.
3748 The application is linked with the instrumented user library.
3750 image::ust-sit+app-linked-with-tp-o+app-linked-with-lib+lib-instrumented.png[]
3753 include::../common/ust-sit-step-tp-o.txt[]
3755 To build the instrumented user library:
3757 . In path:{emon.c}, before including path:{tpp.h}, add the
3763 #define TRACEPOINT_DEFINE
3767 . Compile the user library source file:
3772 $ gcc -I. -fpic -c emon.c
3776 . Build the user library shared object:
3781 $ gcc -shared -o libemon.so emon.o
3785 To build the application:
3787 . Compile the application source file:
3796 . Build the application:
3801 $ gcc -o app app.o tpp.o -llttng-ust -ldl -L. -lemon
3805 To run the instrumented application:
3807 * Start the application:
3817 The application is statically linked with the tracepoint provider
3818 package object file.
3820 The application dynamically loads the instrumented user library.
3822 image::ust-sit+app-linked-with-tp-o+app-dlopens-lib+lib-instrumented.png[]
3825 include::../common/ust-sit-step-tp-o.txt[]
3827 To build the application:
3829 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3834 #define TRACEPOINT_DEFINE
3838 . Compile the application source file:
3847 . Build the application:
3852 $ gcc -Wl,--export-dynamic -o app app.o tpp.o \
3857 The `--export-dynamic` option passed to the linker is necessary for the
3858 dynamically loaded library to ``see'' the tracepoint symbols defined in
3861 To build the instrumented user library:
3863 . Compile the user library source file:
3868 $ gcc -I. -fpic -c emon.c
3872 . Build the user library shared object:
3877 $ gcc -shared -o libemon.so emon.o
3881 To run the application:
3883 * Start the application:
3895 .Do not use man:dlclose(3) on a tracepoint provider package
3897 Never use man:dlclose(3) on any shared object which:
3899 * Is linked with, statically or dynamically, a tracepoint provider
3901 * Calls man:dlopen(3) itself to dynamically open a tracepoint provider
3902 package shared object.
3904 This is currently considered **unsafe** due to a lack of reference
3905 counting from LTTng-UST to the shared object.
3907 A known workaround (available since glibc 2.2) is to use the
3908 `RTLD_NODELETE` flag when calling man:dlopen(3) initially. This has the
3909 effect of not unloading the loaded shared object, even if man:dlclose(3)
3912 You can also preload the tracepoint provider package shared object with
3913 the env:LD_PRELOAD environment variable to overcome this limitation.
3917 [[using-lttng-ust-with-daemons]]
3918 ===== Use noch:{LTTng-UST} with daemons
3920 If your instrumented application calls man:fork(2), man:clone(2),
3921 or BSD's man:rfork(2), without a following man:exec(3)-family
3922 system call, you must preload the path:{liblttng-ust-fork.so} shared
3923 object when you start the application.
3927 $ LD_PRELOAD=liblttng-ust-fork.so ./my-app
3930 If your tracepoint provider package is
3931 a shared library which you also preload, you must put both
3932 shared objects in env:LD_PRELOAD:
3936 $ LD_PRELOAD=liblttng-ust-fork.so:/path/to/tp.so ./my-app
3942 ===== Use noch:{LTTng-UST} with applications which close file descriptors that don't belong to them
3944 If your instrumented application closes one or more file descriptors
3945 which it did not open itself, you must preload the
3946 path:{liblttng-ust-fd.so} shared object when you start the application:
3950 $ LD_PRELOAD=liblttng-ust-fd.so ./my-app
3953 Typical use cases include closing all the file descriptors after
3954 man:fork(2) or man:rfork(2) and buggy applications doing
3958 [[lttng-ust-pkg-config]]
3959 ===== Use noch:{pkg-config}
3961 On some distributions, LTTng-UST ships with a
3962 https://www.freedesktop.org/wiki/Software/pkg-config/[pkg-config]
3963 metadata file. If this is your case, then you can use cmd:pkg-config to
3964 build an application on the command line:
3968 $ gcc -o my-app my-app.o tp.o $(pkg-config --cflags --libs lttng-ust)
3972 [[instrumenting-32-bit-app-on-64-bit-system]]
3973 ===== [[advanced-instrumenting-techniques]]Build a 32-bit instrumented application for a 64-bit target system
3975 In order to trace a 32-bit application running on a 64-bit system,
3976 LTTng must use a dedicated 32-bit
3977 <<lttng-consumerd,consumer daemon>>.
3979 The following steps show how to build and install a 32-bit consumer
3980 daemon, which is _not_ part of the default 64-bit LTTng build, how to
3981 build and install the 32-bit LTTng-UST libraries, and how to build and
3982 link an instrumented 32-bit application in that context.
3984 To build a 32-bit instrumented application for a 64-bit target system,
3985 assuming you have a fresh target system with no installed Userspace RCU
3988 . Download, build, and install a 32-bit version of Userspace RCU:
3993 $ cd $(mktemp -d) &&
3994 wget http://lttng.org/files/urcu/userspace-rcu-latest-0.9.tar.bz2 &&
3995 tar -xf userspace-rcu-latest-0.9.tar.bz2 &&
3996 cd userspace-rcu-0.9.* &&
3997 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 &&
3999 sudo make install &&
4004 . Using your distribution's package manager, or from source, install
4005 the following 32-bit versions of the following dependencies of
4006 LTTng-tools and LTTng-UST:
4009 * https://sourceforge.net/projects/libuuid/[libuuid]
4010 * http://directory.fsf.org/wiki/Popt[popt]
4011 * http://www.xmlsoft.org/[libxml2]
4014 . Download, build, and install a 32-bit version of the latest
4015 LTTng-UST{nbsp}{revision}:
4020 $ cd $(mktemp -d) &&
4021 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.9.tar.bz2 &&
4022 tar -xf lttng-ust-latest-2.9.tar.bz2 &&
4023 cd lttng-ust-2.9.* &&
4024 ./configure --libdir=/usr/local/lib32 \
4025 CFLAGS=-m32 CXXFLAGS=-m32 \
4026 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
4028 sudo make install &&
4035 Depending on your distribution,
4036 32-bit libraries could be installed at a different location than
4037 `/usr/lib32`. For example, Debian is known to install
4038 some 32-bit libraries in `/usr/lib/i386-linux-gnu`.
4040 In this case, make sure to set `LDFLAGS` to all the
4041 relevant 32-bit library paths, for example:
4045 $ LDFLAGS='-L/usr/lib/i386-linux-gnu -L/usr/lib32'
4049 . Download the latest LTTng-tools{nbsp}{revision}, build, and install
4050 the 32-bit consumer daemon:
4055 $ cd $(mktemp -d) &&
4056 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.9.tar.bz2 &&
4057 tar -xf lttng-tools-latest-2.9.tar.bz2 &&
4058 cd lttng-tools-2.9.* &&
4059 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 CXXFLAGS=-m32 \
4060 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' \
4061 --disable-bin-lttng --disable-bin-lttng-crash \
4062 --disable-bin-lttng-relayd --disable-bin-lttng-sessiond &&
4064 cd src/bin/lttng-consumerd &&
4065 sudo make install &&
4070 . From your distribution or from source,
4071 <<installing-lttng,install>> the 64-bit versions of
4072 LTTng-UST and Userspace RCU.
4073 . Download, build, and install the 64-bit version of the
4074 latest LTTng-tools{nbsp}{revision}:
4079 $ cd $(mktemp -d) &&
4080 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.9.tar.bz2 &&
4081 tar -xf lttng-tools-latest-2.9.tar.bz2 &&
4082 cd lttng-tools-2.9.* &&
4083 ./configure --with-consumerd32-libdir=/usr/local/lib32 \
4084 --with-consumerd32-bin=/usr/local/lib32/lttng/libexec/lttng-consumerd &&
4086 sudo make install &&
4091 . Pass the following options to man:gcc(1), man:g++(1), or man:clang(1)
4092 when linking your 32-bit application:
4095 -m32 -L/usr/lib32 -L/usr/local/lib32 \
4096 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32
4099 For example, let's rebuild the quick start example in
4100 <<tracing-your-own-user-application,Trace a user application>> as an
4101 instrumented 32-bit application:
4106 $ gcc -m32 -c -I. hello-tp.c
4107 $ gcc -m32 -c hello.c
4108 $ gcc -m32 -o hello hello.o hello-tp.o \
4109 -L/usr/lib32 -L/usr/local/lib32 \
4110 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32 \
4115 No special action is required to execute the 32-bit application and
4116 to trace it: use the command-line man:lttng(1) tool as usual.
4123 man:tracef(3) is a small LTTng-UST API designed for quick,
4124 man:printf(3)-like instrumentation without the burden of
4125 <<tracepoint-provider,creating>> and
4126 <<building-tracepoint-providers-and-user-application,building>>
4127 a tracepoint provider package.
4129 To use `tracef()` in your application:
4131 . In the C or C++ source files where you need to use `tracef()`,
4132 include `<lttng/tracef.h>`:
4137 #include <lttng/tracef.h>
4141 . In the application's source code, use `tracef()` like you would use
4149 tracef("my message: %d (%s)", my_integer, my_string);
4155 . Link your application with `liblttng-ust`:
4160 $ gcc -o app app.c -llttng-ust
4164 To trace the events that `tracef()` calls emit:
4166 * <<enabling-disabling-events,Create an event rule>> which matches the
4167 `lttng_ust_tracef:*` event name:
4172 $ lttng enable-event --userspace 'lttng_ust_tracef:*'
4177 .Limitations of `tracef()`
4179 The `tracef()` utility function was developed to make user space tracing
4180 super simple, albeit with notable disadvantages compared to
4181 <<defining-tracepoints,user-defined tracepoints>>:
4183 * All the emitted events have the same tracepoint provider and
4184 tracepoint names, respectively `lttng_ust_tracef` and `event`.
4185 * There is no static type checking.
4186 * The only event record field you actually get, named `msg`, is a string
4187 potentially containing the values you passed to `tracef()`
4188 using your own format string. This also means that you cannot filter
4189 events with a custom expression at run time because there are no
4191 * Since `tracef()` uses the C standard library's man:vasprintf(3)
4192 function behind the scenes to format the strings at run time, its
4193 expected performance is lower than with user-defined tracepoints,
4194 which do not require a conversion to a string.
4196 Taking this into consideration, `tracef()` is useful for some quick
4197 prototyping and debugging, but you should not consider it for any
4198 permanent and serious applicative instrumentation.
4204 ==== Use `tracelog()`
4206 The man:tracelog(3) API is very similar to <<tracef,`tracef()`>>, with
4207 the difference that it accepts an additional log level parameter.
4209 The goal of `tracelog()` is to ease the migration from logging to
4212 To use `tracelog()` in your application:
4214 . In the C or C++ source files where you need to use `tracelog()`,
4215 include `<lttng/tracelog.h>`:
4220 #include <lttng/tracelog.h>
4224 . In the application's source code, use `tracelog()` like you would use
4225 man:printf(3), except for the first parameter which is the log
4233 tracelog(TRACE_WARNING, "my message: %d (%s)",
4234 my_integer, my_string);
4240 See man:lttng-ust(3) for a list of available log level names.
4242 . Link your application with `liblttng-ust`:
4247 $ gcc -o app app.c -llttng-ust
4251 To trace the events that `tracelog()` calls emit with a log level
4252 _as severe as_ a specific log level:
4254 * <<enabling-disabling-events,Create an event rule>> which matches the
4255 `lttng_ust_tracelog:*` event name and a minimum level
4261 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4262 --loglevel=TRACE_WARNING
4266 To trace the events that `tracelog()` calls emit with a
4267 _specific log level_:
4269 * Create an event rule which matches the `lttng_ust_tracelog:*`
4270 event name and a specific log level:
4275 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4276 --loglevel-only=TRACE_INFO
4281 [[prebuilt-ust-helpers]]
4282 === Prebuilt user space tracing helpers
4284 The LTTng-UST package provides a few helpers in the form or preloadable
4285 shared objects which automatically instrument system functions and
4288 The helper shared objects are normally found in dir:{/usr/lib}. If you
4289 built LTTng-UST <<building-from-source,from source>>, they are probably
4290 located in dir:{/usr/local/lib}.
4292 The installed user space tracing helpers in LTTng-UST{nbsp}{revision}
4295 path:{liblttng-ust-libc-wrapper.so}::
4296 path:{liblttng-ust-pthread-wrapper.so}::
4297 <<liblttng-ust-libc-pthread-wrapper,C{nbsp}standard library
4298 memory and POSIX threads function tracing>>.
4300 path:{liblttng-ust-cyg-profile.so}::
4301 path:{liblttng-ust-cyg-profile-fast.so}::
4302 <<liblttng-ust-cyg-profile,Function entry and exit tracing>>.
4304 path:{liblttng-ust-dl.so}::
4305 <<liblttng-ust-dl,Dynamic linker tracing>>.
4307 To use a user space tracing helper with any user application:
4309 * Preload the helper shared object when you start the application:
4314 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so my-app
4318 You can preload more than one helper:
4323 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so:liblttng-ust-dl.so my-app
4329 [[liblttng-ust-libc-pthread-wrapper]]
4330 ==== Instrument C standard library memory and POSIX threads functions
4332 The path:{liblttng-ust-libc-wrapper.so} and
4333 path:{liblttng-ust-pthread-wrapper.so} helpers
4334 add instrumentation to some C standard library and POSIX
4338 .Functions instrumented by preloading path:{liblttng-ust-libc-wrapper.so}.
4340 |TP provider name |TP name |Instrumented function
4342 .6+|`lttng_ust_libc` |`malloc` |man:malloc(3)
4343 |`calloc` |man:calloc(3)
4344 |`realloc` |man:realloc(3)
4345 |`free` |man:free(3)
4346 |`memalign` |man:memalign(3)
4347 |`posix_memalign` |man:posix_memalign(3)
4351 .Functions instrumented by preloading path:{liblttng-ust-pthread-wrapper.so}.
4353 |TP provider name |TP name |Instrumented function
4355 .4+|`lttng_ust_pthread` |`pthread_mutex_lock_req` |man:pthread_mutex_lock(3p) (request time)
4356 |`pthread_mutex_lock_acq` |man:pthread_mutex_lock(3p) (acquire time)
4357 |`pthread_mutex_trylock` |man:pthread_mutex_trylock(3p)
4358 |`pthread_mutex_unlock` |man:pthread_mutex_unlock(3p)
4361 When you preload the shared object, it replaces the functions listed
4362 in the previous tables by wrappers which contain tracepoints and call
4363 the replaced functions.
4366 [[liblttng-ust-cyg-profile]]
4367 ==== Instrument function entry and exit
4369 The path:{liblttng-ust-cyg-profile*.so} helpers can add instrumentation
4370 to the entry and exit points of functions.
4372 man:gcc(1) and man:clang(1) have an option named
4373 https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html[`-finstrument-functions`]
4374 which generates instrumentation calls for entry and exit to functions.
4375 The LTTng-UST function tracing helpers,
4376 path:{liblttng-ust-cyg-profile.so} and
4377 path:{liblttng-ust-cyg-profile-fast.so}, take advantage of this feature
4378 to add tracepoints to the two generated functions (which contain
4379 `cyg_profile` in their names, hence the helper's name).
4381 To use the LTTng-UST function tracing helper, the source files to
4382 instrument must be built using the `-finstrument-functions` compiler
4385 There are two versions of the LTTng-UST function tracing helper:
4387 * **path:{liblttng-ust-cyg-profile-fast.so}** is a lightweight variant
4388 that you should only use when it can be _guaranteed_ that the
4389 complete event stream is recorded without any lost event record.
4390 Any kind of duplicate information is left out.
4392 Assuming no event record is lost, having only the function addresses on
4393 entry is enough to create a call graph, since an event record always
4394 contains the ID of the CPU that generated it.
4396 You can use a tool like man:addr2line(1) to convert function addresses
4397 back to source file names and line numbers.
4399 * **path:{liblttng-ust-cyg-profile.so}** is a more robust variant
4400 which also works in use cases where event records might get discarded or
4401 not recorded from application startup.
4402 In these cases, the trace analyzer needs more information to be
4403 able to reconstruct the program flow.
4405 See man:lttng-ust-cyg-profile(3) to learn more about the instrumentation
4406 points of this helper.
4408 All the tracepoints that this helper provides have the
4409 log level `TRACE_DEBUG_FUNCTION` (see man:lttng-ust(3)).
4411 TIP: It's sometimes a good idea to limit the number of source files that
4412 you compile with the `-finstrument-functions` option to prevent LTTng
4413 from writing an excessive amount of trace data at run time. When using
4414 man:gcc(1), you can use the
4415 `-finstrument-functions-exclude-function-list` option to avoid
4416 instrument entries and exits of specific function names.
4421 ==== Instrument the dynamic linker
4423 The path:{liblttng-ust-dl.so} helper adds instrumentation to the
4424 man:dlopen(3) and man:dlclose(3) function calls.
4426 See man:lttng-ust-dl(3) to learn more about the instrumentation points
4431 [[java-application]]
4432 === User space Java agent
4434 You can instrument any Java application which uses one of the following
4437 * The https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[**`java.util.logging`**]
4438 (JUL) core logging facilities.
4439 * http://logging.apache.org/log4j/1.2/[**Apache log4j 1.2**], since
4440 LTTng 2.6. Note that Apache Log4j{nbsp}2 is not supported.
4443 .LTTng-UST Java agent imported by a Java application.
4444 image::java-app.png[]
4446 Note that the methods described below are new in LTTng{nbsp}{revision}.
4447 Previous LTTng versions use another technique.
4449 NOTE: We use http://openjdk.java.net/[OpenJDK]{nbsp}8 for development
4450 and https://ci.lttng.org/[continuous integration], thus this version is
4451 directly supported. However, the LTTng-UST Java agent is also tested
4452 with OpenJDK{nbsp}7.
4457 ==== Use the LTTng-UST Java agent for `java.util.logging`
4459 To use the LTTng-UST Java agent in a Java application which uses
4460 `java.util.logging` (JUL):
4462 . In the Java application's source code, import the LTTng-UST
4463 log handler package for `java.util.logging`:
4468 import org.lttng.ust.agent.jul.LttngLogHandler;
4472 . Create an LTTng-UST JUL log handler:
4477 Handler lttngUstLogHandler = new LttngLogHandler();
4481 . Add this handler to the JUL loggers which should emit LTTng events:
4486 Logger myLogger = Logger.getLogger("some-logger");
4488 myLogger.addHandler(lttngUstLogHandler);
4492 . Use `java.util.logging` log statements and configuration as usual.
4493 The loggers with an attached LTTng-UST log handler can emit
4496 . Before exiting the application, remove the LTTng-UST log handler from
4497 the loggers attached to it and call its `close()` method:
4502 myLogger.removeHandler(lttngUstLogHandler);
4503 lttngUstLogHandler.close();
4507 This is not strictly necessary, but it is recommended for a clean
4508 disposal of the handler's resources.
4510 . Include the LTTng-UST Java agent's common and JUL-specific JAR files,
4511 path:{lttng-ust-agent-common.jar} and path:{lttng-ust-agent-jul.jar},
4513 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4514 path] when you build the Java application.
4516 The JAR files are typically located in dir:{/usr/share/java}.
4518 IMPORTANT: The LTTng-UST Java agent must be
4519 <<installing-lttng,installed>> for the logging framework your
4522 .Use the LTTng-UST Java agent for `java.util.logging`.
4527 import java.io.IOException;
4528 import java.util.logging.Handler;
4529 import java.util.logging.Logger;
4530 import org.lttng.ust.agent.jul.LttngLogHandler;
4534 private static final int answer = 42;
4536 public static void main(String[] argv) throws Exception
4539 Logger logger = Logger.getLogger("jello");
4541 // Create an LTTng-UST log handler
4542 Handler lttngUstLogHandler = new LttngLogHandler();
4544 // Add the LTTng-UST log handler to our logger
4545 logger.addHandler(lttngUstLogHandler);
4548 logger.info("some info");
4549 logger.warning("some warning");
4551 logger.finer("finer information; the answer is " + answer);
4553 logger.severe("error!");
4555 // Not mandatory, but cleaner
4556 logger.removeHandler(lttngUstLogHandler);
4557 lttngUstLogHandler.close();
4566 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4569 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4570 <<enabling-disabling-events,create an event rule>> matching the
4571 `jello` JUL logger, and <<basic-tracing-session-control,start tracing>>:
4576 $ lttng enable-event --jul jello
4580 Run the compiled class:
4584 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4587 <<basic-tracing-session-control,Stop tracing>> and inspect the
4597 In the resulting trace, an <<event,event record>> generated by a Java
4598 application using `java.util.logging` is named `lttng_jul:event` and
4599 has the following fields:
4602 Log record's message.
4608 Name of the class in which the log statement was executed.
4611 Name of the method in which the log statement was executed.
4614 Logging time (timestamp in milliseconds).
4617 Log level integer value.
4620 ID of the thread in which the log statement was executed.
4622 You can use the opt:lttng-enable-event(1):--loglevel or
4623 opt:lttng-enable-event(1):--loglevel-only option of the
4624 man:lttng-enable-event(1) command to target a range of JUL log levels
4625 or a specific JUL log level.
4630 ==== Use the LTTng-UST Java agent for Apache log4j
4632 To use the LTTng-UST Java agent in a Java application which uses
4635 . In the Java application's source code, import the LTTng-UST
4636 log appender package for Apache log4j:
4641 import org.lttng.ust.agent.log4j.LttngLogAppender;
4645 . Create an LTTng-UST log4j log appender:
4650 Appender lttngUstLogAppender = new LttngLogAppender();
4654 . Add this appender to the log4j loggers which should emit LTTng events:
4659 Logger myLogger = Logger.getLogger("some-logger");
4661 myLogger.addAppender(lttngUstLogAppender);
4665 . Use Apache log4j log statements and configuration as usual. The
4666 loggers with an attached LTTng-UST log appender can emit LTTng events.
4668 . Before exiting the application, remove the LTTng-UST log appender from
4669 the loggers attached to it and call its `close()` method:
4674 myLogger.removeAppender(lttngUstLogAppender);
4675 lttngUstLogAppender.close();
4679 This is not strictly necessary, but it is recommended for a clean
4680 disposal of the appender's resources.
4682 . Include the LTTng-UST Java agent's common and log4j-specific JAR
4683 files, path:{lttng-ust-agent-common.jar} and
4684 path:{lttng-ust-agent-log4j.jar}, in the
4685 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4686 path] when you build the Java application.
4688 The JAR files are typically located in dir:{/usr/share/java}.
4690 IMPORTANT: The LTTng-UST Java agent must be
4691 <<installing-lttng,installed>> for the logging framework your
4694 .Use the LTTng-UST Java agent for Apache log4j.
4699 import org.apache.log4j.Appender;
4700 import org.apache.log4j.Logger;
4701 import org.lttng.ust.agent.log4j.LttngLogAppender;
4705 private static final int answer = 42;
4707 public static void main(String[] argv) throws Exception
4710 Logger logger = Logger.getLogger("jello");
4712 // Create an LTTng-UST log appender
4713 Appender lttngUstLogAppender = new LttngLogAppender();
4715 // Add the LTTng-UST log appender to our logger
4716 logger.addAppender(lttngUstLogAppender);
4719 logger.info("some info");
4720 logger.warn("some warning");
4722 logger.debug("debug information; the answer is " + answer);
4724 logger.fatal("error!");
4726 // Not mandatory, but cleaner
4727 logger.removeAppender(lttngUstLogAppender);
4728 lttngUstLogAppender.close();
4734 Build this example (`$LOG4JPATH` is the path to the Apache log4j JAR
4739 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH Test.java
4742 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4743 <<enabling-disabling-events,create an event rule>> matching the
4744 `jello` log4j logger, and <<basic-tracing-session-control,start tracing>>:
4749 $ lttng enable-event --log4j jello
4753 Run the compiled class:
4757 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH:. Test
4760 <<basic-tracing-session-control,Stop tracing>> and inspect the
4770 In the resulting trace, an <<event,event record>> generated by a Java
4771 application using log4j is named `lttng_log4j:event` and
4772 has the following fields:
4775 Log record's message.
4781 Name of the class in which the log statement was executed.
4784 Name of the method in which the log statement was executed.
4787 Name of the file in which the executed log statement is located.
4790 Line number at which the log statement was executed.
4796 Log level integer value.
4799 Name of the Java thread in which the log statement was executed.
4801 You can use the opt:lttng-enable-event(1):--loglevel or
4802 opt:lttng-enable-event(1):--loglevel-only option of the
4803 man:lttng-enable-event(1) command to target a range of Apache log4j log levels
4804 or a specific log4j log level.
4808 [[java-application-context]]
4809 ==== Provide application-specific context fields in a Java application
4811 A Java application-specific context field is a piece of state provided
4812 by the application which <<adding-context,you can add>>, using the
4813 man:lttng-add-context(1) command, to each <<event,event record>>
4814 produced by the log statements of this application.
4816 For example, a given object might have a current request ID variable.
4817 You can create a context information retriever for this object and
4818 assign a name to this current request ID. You can then, using the
4819 man:lttng-add-context(1) command, add this context field by name to
4820 the JUL or log4j <<channel,channel>>.
4822 To provide application-specific context fields in a Java application:
4824 . In the Java application's source code, import the LTTng-UST
4825 Java agent context classes and interfaces:
4830 import org.lttng.ust.agent.context.ContextInfoManager;
4831 import org.lttng.ust.agent.context.IContextInfoRetriever;
4835 . Create a context information retriever class, that is, a class which
4836 implements the `IContextInfoRetriever` interface:
4841 class MyContextInfoRetriever implements IContextInfoRetriever
4844 public Object retrieveContextInfo(String key)
4846 if (key.equals("intCtx")) {
4848 } else if (key.equals("strContext")) {
4849 return "context value!";
4858 This `retrieveContextInfo()` method is the only member of the
4859 `IContextInfoRetriever` interface. Its role is to return the current
4860 value of a state by name to create a context field. The names of the
4861 context fields and which state variables they return depends on your
4864 All primitive types and objects are supported as context fields.
4865 When `retrieveContextInfo()` returns an object, the context field
4866 serializer calls its `toString()` method to add a string field to
4867 event records. The method can also return `null`, which means that
4868 no context field is available for the required name.
4870 . Register an instance of your context information retriever class to
4871 the context information manager singleton:
4876 IContextInfoRetriever cir = new MyContextInfoRetriever();
4877 ContextInfoManager cim = ContextInfoManager.getInstance();
4878 cim.registerContextInfoRetriever("retrieverName", cir);
4882 . Before exiting the application, remove your context information
4883 retriever from the context information manager singleton:
4888 ContextInfoManager cim = ContextInfoManager.getInstance();
4889 cim.unregisterContextInfoRetriever("retrieverName");
4893 This is not strictly necessary, but it is recommended for a clean
4894 disposal of some manager's resources.
4896 . Build your Java application with LTTng-UST Java agent support as
4897 usual, following the procedure for either the <<jul,JUL>> or
4898 <<log4j,Apache log4j>> framework.
4901 .Provide application-specific context fields in a Java application.
4906 import java.util.logging.Handler;
4907 import java.util.logging.Logger;
4908 import org.lttng.ust.agent.jul.LttngLogHandler;
4909 import org.lttng.ust.agent.context.ContextInfoManager;
4910 import org.lttng.ust.agent.context.IContextInfoRetriever;
4914 // Our context information retriever class
4915 private static class MyContextInfoRetriever
4916 implements IContextInfoRetriever
4919 public Object retrieveContextInfo(String key) {
4920 if (key.equals("intCtx")) {
4922 } else if (key.equals("strContext")) {
4923 return "context value!";
4930 private static final int answer = 42;
4932 public static void main(String args[]) throws Exception
4934 // Get the context information manager instance
4935 ContextInfoManager cim = ContextInfoManager.getInstance();
4937 // Create and register our context information retriever
4938 IContextInfoRetriever cir = new MyContextInfoRetriever();
4939 cim.registerContextInfoRetriever("myRetriever", cir);
4942 Logger logger = Logger.getLogger("jello");
4944 // Create an LTTng-UST log handler
4945 Handler lttngUstLogHandler = new LttngLogHandler();
4947 // Add the LTTng-UST log handler to our logger
4948 logger.addHandler(lttngUstLogHandler);
4951 logger.info("some info");
4952 logger.warning("some warning");
4954 logger.finer("finer information; the answer is " + answer);
4956 logger.severe("error!");
4958 // Not mandatory, but cleaner
4959 logger.removeHandler(lttngUstLogHandler);
4960 lttngUstLogHandler.close();
4961 cim.unregisterContextInfoRetriever("myRetriever");
4970 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4973 <<creating-destroying-tracing-sessions,Create a tracing session>>
4974 and <<enabling-disabling-events,create an event rule>> matching the
4980 $ lttng enable-event --jul jello
4983 <<adding-context,Add the application-specific context fields>> to the
4988 $ lttng add-context --jul --type='$app.myRetriever:intCtx'
4989 $ lttng add-context --jul --type='$app.myRetriever:strContext'
4992 <<basic-tracing-session-control,Start tracing>>:
4999 Run the compiled class:
5003 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
5006 <<basic-tracing-session-control,Stop tracing>> and inspect the
5018 [[python-application]]
5019 === User space Python agent
5021 You can instrument a Python 2 or Python 3 application which uses the
5022 standard https://docs.python.org/3/library/logging.html[`logging`]
5025 Each log statement emits an LTTng event once the
5026 application module imports the
5027 <<lttng-ust-agents,LTTng-UST Python agent>> package.
5030 .A Python application importing the LTTng-UST Python agent.
5031 image::python-app.png[]
5033 To use the LTTng-UST Python agent:
5035 . In the Python application's source code, import the LTTng-UST Python
5045 The LTTng-UST Python agent automatically adds its logging handler to the
5046 root logger at import time.
5048 Any log statement that the application executes before this import does
5049 not emit an LTTng event.
5051 IMPORTANT: The LTTng-UST Python agent must be
5052 <<installing-lttng,installed>>.
5054 . Use log statements and logging configuration as usual.
5055 Since the LTTng-UST Python agent adds a handler to the _root_
5056 logger, you can trace any log statement from any logger.
5058 .Use the LTTng-UST Python agent.
5069 logging.basicConfig()
5070 logger = logging.getLogger('my-logger')
5073 logger.debug('debug message')
5074 logger.info('info message')
5075 logger.warn('warn message')
5076 logger.error('error message')
5077 logger.critical('critical message')
5081 if __name__ == '__main__':
5085 NOTE: `logging.basicConfig()`, which adds to the root logger a basic
5086 logging handler which prints to the standard error stream, is not
5087 strictly required for LTTng-UST tracing to work, but in versions of
5088 Python preceding 3.2, you could see a warning message which indicates
5089 that no handler exists for the logger `my-logger`.
5091 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5092 <<enabling-disabling-events,create an event rule>> matching the
5093 `my-logger` Python logger, and <<basic-tracing-session-control,start
5099 $ lttng enable-event --python my-logger
5103 Run the Python script:
5110 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5120 In the resulting trace, an <<event,event record>> generated by a Python
5121 application is named `lttng_python:event` and has the following fields:
5124 Logging time (string).
5127 Log record's message.
5133 Name of the function in which the log statement was executed.
5136 Line number at which the log statement was executed.
5139 Log level integer value.
5142 ID of the Python thread in which the log statement was executed.
5145 Name of the Python thread in which the log statement was executed.
5147 You can use the opt:lttng-enable-event(1):--loglevel or
5148 opt:lttng-enable-event(1):--loglevel-only option of the
5149 man:lttng-enable-event(1) command to target a range of Python log levels
5150 or a specific Python log level.
5152 When an application imports the LTTng-UST Python agent, the agent tries
5153 to register to a <<lttng-sessiond,session daemon>>. Note that you must
5154 <<start-sessiond,start the session daemon>> _before_ you run the Python
5155 application. If a session daemon is found, the agent tries to register
5156 to it during 5{nbsp}seconds, after which the application continues
5157 without LTTng tracing support. You can override this timeout value with
5158 the env:LTTNG_UST_PYTHON_REGISTER_TIMEOUT environment variable
5161 If the session daemon stops while a Python application with an imported
5162 LTTng-UST Python agent runs, the agent retries to connect and to
5163 register to a session daemon every 3{nbsp}seconds. You can override this
5164 delay with the env:LTTNG_UST_PYTHON_REGISTER_RETRY_DELAY environment
5169 [[proc-lttng-logger-abi]]
5172 The `lttng-tracer` Linux kernel module, part of
5173 <<lttng-modules,LTTng-modules>>, creates the special LTTng logger file
5174 path:{/proc/lttng-logger} when it's loaded. Any application can write
5175 text data to this file to emit an LTTng event.
5178 .An application writes to the LTTng logger file to emit an LTTng event.
5179 image::lttng-logger.png[]
5181 The LTTng logger is the quickest method--not the most efficient,
5182 however--to add instrumentation to an application. It is designed
5183 mostly to instrument shell scripts:
5187 $ echo "Some message, some $variable" > /proc/lttng-logger
5190 Any event that the LTTng logger emits is named `lttng_logger` and
5191 belongs to the Linux kernel <<domain,tracing domain>>. However, unlike
5192 other instrumentation points in the kernel tracing domain, **any Unix
5193 user** can <<enabling-disabling-events,create an event rule>> which
5194 matches its event name, not only the root user or users in the
5195 <<tracing-group,tracing group>>.
5197 To use the LTTng logger:
5199 * From any application, write text data to the path:{/proc/lttng-logger}
5202 The `msg` field of `lttng_logger` event records contains the
5205 NOTE: The maximum message length of an LTTng logger event is
5206 1024{nbsp}bytes. Writing more than this makes the LTTng logger emit more
5207 than one event to contain the remaining data.
5209 You should not use the LTTng logger to trace a user application which
5210 can be instrumented in a more efficient way, namely:
5212 * <<c-application,C and $$C++$$ applications>>.
5213 * <<java-application,Java applications>>.
5214 * <<python-application,Python applications>>.
5216 .Use the LTTng logger.
5221 echo 'Hello, World!' > /proc/lttng-logger
5223 df --human-readable --print-type / > /proc/lttng-logger
5226 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5227 <<enabling-disabling-events,create an event rule>> matching the
5228 `lttng_logger` Linux kernel tracepoint, and
5229 <<basic-tracing-session-control,start tracing>>:
5234 $ lttng enable-event --kernel lttng_logger
5238 Run the Bash script:
5245 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5256 [[instrumenting-linux-kernel]]
5257 === LTTng kernel tracepoints
5259 NOTE: This section shows how to _add_ instrumentation points to the
5260 Linux kernel. The kernel's subsystems are already thoroughly
5261 instrumented at strategic places for LTTng when you
5262 <<installing-lttng,install>> the <<lttng-modules,LTTng-modules>>
5266 There are two methods to instrument the Linux kernel:
5268 . <<linux-add-lttng-layer,Add an LTTng layer>> over an existing ftrace
5269 tracepoint which uses the `TRACE_EVENT()` API.
5271 Choose this if you want to instrumentation a Linux kernel tree with an
5272 instrumentation point compatible with ftrace, perf, and SystemTap.
5274 . Use an <<linux-lttng-tracepoint-event,LTTng-only approach>> to
5275 instrument an out-of-tree kernel module.
5277 Choose this if you don't need ftrace, perf, or SystemTap support.
5281 [[linux-add-lttng-layer]]
5282 ==== [[instrumenting-linux-kernel-itself]][[mainline-trace-event]][[lttng-adaptation-layer]]Add an LTTng layer to an existing ftrace tracepoint
5284 This section shows how to add an LTTng layer to existing ftrace
5285 instrumentation using the `TRACE_EVENT()` API.
5287 This section does not document the `TRACE_EVENT()` macro. You can
5288 read the following articles to learn more about this API:
5290 * http://lwn.net/Articles/379903/[Using the TRACE_EVENT() macro (Part 1)]
5291 * http://lwn.net/Articles/381064/[Using the TRACE_EVENT() macro (Part 2)]
5292 * http://lwn.net/Articles/383362/[Using the TRACE_EVENT() macro (Part 3)]
5294 The following procedure assumes that your ftrace tracepoints are
5295 correctly defined in their own header and that they are created in
5296 one source file using the `CREATE_TRACE_POINTS` definition.
5298 To add an LTTng layer over an existing ftrace tracepoint:
5300 . Make sure the following kernel configuration options are
5306 * `CONFIG_HIGH_RES_TIMERS`
5307 * `CONFIG_TRACEPOINTS`
5310 . Build the Linux source tree with your custom ftrace tracepoints.
5311 . Boot the resulting Linux image on your target system.
5313 Confirm that the tracepoints exist by looking for their names in the
5314 dir:{/sys/kernel/debug/tracing/events/subsys} directory, where `subsys`
5315 is your subsystem's name.
5317 . Get a copy of the latest LTTng-modules{nbsp}{revision}:
5322 $ cd $(mktemp -d) &&
5323 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.9.tar.bz2 &&
5324 tar -xf lttng-modules-latest-2.9.tar.bz2 &&
5325 cd lttng-modules-2.9.*
5329 . In dir:{instrumentation/events/lttng-module}, relative to the root
5330 of the LTTng-modules source tree, create a header file named
5331 +__subsys__.h+ for your custom subsystem +__subsys__+ and write your
5332 LTTng-modules tracepoint definitions using the LTTng-modules
5335 Start with this template:
5339 .path:{instrumentation/events/lttng-module/my_subsys.h}
5342 #define TRACE_SYSTEM my_subsys
5344 #if !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ)
5345 #define _LTTNG_MY_SUBSYS_H
5347 #include "../../../probes/lttng-tracepoint-event.h"
5348 #include <linux/tracepoint.h>
5350 LTTNG_TRACEPOINT_EVENT(
5352 * Format is identical to TRACE_EVENT()'s version for the three
5353 * following macro parameters:
5356 TP_PROTO(int my_int, const char *my_string),
5357 TP_ARGS(my_int, my_string),
5359 /* LTTng-modules specific macros */
5361 ctf_integer(int, my_int_field, my_int)
5362 ctf_string(my_bar_field, my_bar)
5366 #endif /* !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ) */
5368 #include "../../../probes/define_trace.h"
5372 The entries in the `TP_FIELDS()` section are the list of fields for the
5373 LTTng tracepoint. This is similar to the `TP_STRUCT__entry()` part of
5374 ftrace's `TRACE_EVENT()` macro.
5376 See <<lttng-modules-tp-fields,Tracepoint fields macros>> for a
5377 complete description of the available `ctf_*()` macros.
5379 . Create the LTTng-modules probe's kernel module C source file,
5380 +probes/lttng-probe-__subsys__.c+, where +__subsys__+ is your
5385 .path:{probes/lttng-probe-my-subsys.c}
5387 #include <linux/module.h>
5388 #include "../lttng-tracer.h"
5391 * Build-time verification of mismatch between mainline
5392 * TRACE_EVENT() arguments and the LTTng-modules adaptation
5393 * layer LTTNG_TRACEPOINT_EVENT() arguments.
5395 #include <trace/events/my_subsys.h>
5397 /* Create LTTng tracepoint probes */
5398 #define LTTNG_PACKAGE_BUILD
5399 #define CREATE_TRACE_POINTS
5400 #define TRACE_INCLUDE_PATH ../instrumentation/events/lttng-module
5402 #include "../instrumentation/events/lttng-module/my_subsys.h"
5404 MODULE_LICENSE("GPL and additional rights");
5405 MODULE_AUTHOR("Your name <your-email>");
5406 MODULE_DESCRIPTION("LTTng my_subsys probes");
5407 MODULE_VERSION(__stringify(LTTNG_MODULES_MAJOR_VERSION) "."
5408 __stringify(LTTNG_MODULES_MINOR_VERSION) "."
5409 __stringify(LTTNG_MODULES_PATCHLEVEL_VERSION)
5410 LTTNG_MODULES_EXTRAVERSION);
5414 . Edit path:{probes/KBuild} and add your new kernel module object
5415 next to the existing ones:
5419 .path:{probes/KBuild}
5423 obj-m += lttng-probe-module.o
5424 obj-m += lttng-probe-power.o
5426 obj-m += lttng-probe-my-subsys.o
5432 . Build and install the LTTng kernel modules:
5437 $ make KERNELDIR=/path/to/linux
5438 # make modules_install && depmod -a
5442 Replace `/path/to/linux` with the path to the Linux source tree where
5443 you defined and used tracepoints with ftrace's `TRACE_EVENT()` macro.
5445 Note that you can also use the
5446 <<lttng-tracepoint-event-code,`LTTNG_TRACEPOINT_EVENT_CODE()` macro>>
5447 instead of `LTTNG_TRACEPOINT_EVENT()` to use custom local variables and
5448 C code that need to be executed before the event fields are recorded.
5450 The best way to learn how to use the previous LTTng-modules macros is to
5451 inspect the existing LTTng-modules tracepoint definitions in the
5452 dir:{instrumentation/events/lttng-module} header files. Compare them
5453 with the Linux kernel mainline versions in the
5454 dir:{include/trace/events} directory of the Linux source tree.
5458 [[lttng-tracepoint-event-code]]
5459 ===== Use custom C code to access the data for tracepoint fields
5461 Although we recommended to always use the
5462 <<lttng-adaptation-layer,`LTTNG_TRACEPOINT_EVENT()`>> macro to describe
5463 the arguments and fields of an LTTng-modules tracepoint when possible,
5464 sometimes you need a more complex process to access the data that the
5465 tracer records as event record fields. In other words, you need local
5466 variables and multiple C{nbsp}statements instead of simple
5467 argument-based expressions that you pass to the
5468 <<lttng-modules-tp-fields,`ctf_*()` macros of `TP_FIELDS()`>>.
5470 You can use the `LTTNG_TRACEPOINT_EVENT_CODE()` macro instead of
5471 `LTTNG_TRACEPOINT_EVENT()` to declare custom local variables and define
5472 a block of C{nbsp}code to be executed before LTTng records the fields.
5473 The structure of this macro is:
5476 .`LTTNG_TRACEPOINT_EVENT_CODE()` macro syntax.
5478 LTTNG_TRACEPOINT_EVENT_CODE(
5480 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5481 * version for the following three macro parameters:
5484 TP_PROTO(int my_int, const char *my_string),
5485 TP_ARGS(my_int, my_string),
5487 /* Declarations of custom local variables */
5490 unsigned long b = 0;
5491 const char *name = "(undefined)";
5492 struct my_struct *my_struct;
5496 * Custom code which uses both tracepoint arguments
5497 * (in TP_ARGS()) and local variables (in TP_locvar()).
5499 * Local variables are actually members of a structure pointed
5500 * to by the special variable tp_locvar.
5504 tp_locvar->a = my_int + 17;
5505 tp_locvar->my_struct = get_my_struct_at(tp_locvar->a);
5506 tp_locvar->b = my_struct_compute_b(tp_locvar->my_struct);
5507 tp_locvar->name = my_struct_get_name(tp_locvar->my_struct);
5508 put_my_struct(tp_locvar->my_struct);
5517 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5518 * version for this, except that tp_locvar members can be
5519 * used in the argument expression parameters of
5520 * the ctf_*() macros.
5523 ctf_integer(unsigned long, my_struct_b, tp_locvar->b)
5524 ctf_integer(int, my_struct_a, tp_locvar->a)
5525 ctf_string(my_string_field, my_string)
5526 ctf_string(my_struct_name, tp_locvar->name)
5531 IMPORTANT: The C code defined in `TP_code()` must not have any side
5532 effects when executed. In particular, the code must not allocate
5533 memory or get resources without deallocating this memory or putting
5534 those resources afterwards.
5537 [[instrumenting-linux-kernel-tracing]]
5538 ==== Load and unload a custom probe kernel module
5540 You must load a <<lttng-adaptation-layer,created LTTng-modules probe
5541 kernel module>> in the kernel before it can emit LTTng events.
5543 To load the default probe kernel modules and a custom probe kernel
5546 * Use the opt:lttng-sessiond(8):--extra-kmod-probes option to give extra
5547 probe modules to load when starting a root <<lttng-sessiond,session
5551 .Load the `my_subsys`, `usb`, and the default probe modules.
5555 # lttng-sessiond --extra-kmod-probes=my_subsys,usb
5560 You only need to pass the subsystem name, not the whole kernel module
5563 To load _only_ a given custom probe kernel module:
5565 * Use the opt:lttng-sessiond(8):--kmod-probes option to give the probe
5566 modules to load when starting a root session daemon:
5569 .Load only the `my_subsys` and `usb` probe modules.
5573 # lttng-sessiond --kmod-probes=my_subsys,usb
5578 To confirm that a probe module is loaded:
5585 $ lsmod | grep lttng_probe_usb
5589 To unload the loaded probe modules:
5591 * Kill the session daemon with `SIGTERM`:
5596 # pkill lttng-sessiond
5600 You can also use man:modprobe(8)'s `--remove` option if the session
5601 daemon terminates abnormally.
5604 [[controlling-tracing]]
5607 Once an application or a Linux kernel is
5608 <<instrumenting,instrumented>> for LTTng tracing,
5611 This section is divided in topics on how to use the various
5612 <<plumbing,components of LTTng>>, in particular the <<lttng-cli,cmd:lttng
5613 command-line tool>>, to _control_ the LTTng daemons and tracers.
5615 NOTE: In the following subsections, we refer to an man:lttng(1) command
5616 using its man page name. For example, instead of _Run the `create`
5617 command to..._, we use _Run the man:lttng-create(1) command to..._.
5621 === Start a session daemon
5623 In some situations, you need to run a <<lttng-sessiond,session daemon>>
5624 (man:lttng-sessiond(8)) _before_ you can use the man:lttng(1)
5627 You will see the following error when you run a command while no session
5631 Error: No session daemon is available
5634 The only command that automatically runs a session daemon is
5635 man:lttng-create(1), which you use to
5636 <<creating-destroying-tracing-sessions,create a tracing session>>. While
5637 this is most of the time the first operation that you do, sometimes it's
5638 not. Some examples are:
5640 * <<list-instrumentation-points,List the available instrumentation points>>.
5641 * <<saving-loading-tracing-session,Load a tracing session configuration>>.
5643 [[tracing-group]] Each Unix user must have its own running session
5644 daemon to trace user applications. The session daemon that the root user
5645 starts is the only one allowed to control the LTTng kernel tracer. Users
5646 that are part of the _tracing group_ can control the root session
5647 daemon. The default tracing group name is `tracing`; you can set it to
5648 something else with the opt:lttng-sessiond(8):--group option when you
5649 start the root session daemon.
5651 To start a user session daemon:
5653 * Run man:lttng-sessiond(8):
5658 $ lttng-sessiond --daemonize
5662 To start the root session daemon:
5664 * Run man:lttng-sessiond(8) as the root user:
5669 # lttng-sessiond --daemonize
5673 In both cases, remove the opt:lttng-sessiond(8):--daemonize option to
5674 start the session daemon in foreground.
5676 To stop a session daemon, use man:kill(1) on its process ID (standard
5679 Note that some Linux distributions could manage the LTTng session daemon
5680 as a service. In this case, you should use the service manager to
5681 start, restart, and stop session daemons.
5684 [[creating-destroying-tracing-sessions]]
5685 === Create and destroy a tracing session
5687 Almost all the LTTng control operations happen in the scope of
5688 a <<tracing-session,tracing session>>, which is the dialogue between the
5689 <<lttng-sessiond,session daemon>> and you.
5691 To create a tracing session with a generated name:
5693 * Use the man:lttng-create(1) command:
5702 The created tracing session's name is `auto` followed by the
5705 To create a tracing session with a specific name:
5707 * Use the optional argument of the man:lttng-create(1) command:
5712 $ lttng create my-session
5716 Replace `my-session` with the specific tracing session name.
5718 LTTng appends the creation date to the created tracing session's name.
5720 LTTng writes the traces of a tracing session in
5721 +$LTTNG_HOME/lttng-trace/__name__+ by default, where +__name__+ is the
5722 name of the tracing session. Note that the env:LTTNG_HOME environment
5723 variable defaults to `$HOME` if not set.
5725 To output LTTng traces to a non-default location:
5727 * Use the opt:lttng-create(1):--output option of the man:lttng-create(1) command:
5732 $ lttng create my-session --output=/tmp/some-directory
5736 You may create as many tracing sessions as you wish.
5738 To list all the existing tracing sessions for your Unix user:
5740 * Use the man:lttng-list(1) command:
5749 When you create a tracing session, it is set as the _current tracing
5750 session_. The following man:lttng(1) commands operate on the current
5751 tracing session when you don't specify one:
5753 [role="list-3-cols"]
5770 To change the current tracing session:
5772 * Use the man:lttng-set-session(1) command:
5777 $ lttng set-session new-session
5781 Replace `new-session` by the name of the new current tracing session.
5783 When you are done tracing in a given tracing session, you can destroy
5784 it. This operation frees the resources taken by the tracing session
5785 to destroy; it does not destroy the trace data that LTTng wrote for
5786 this tracing session.
5788 To destroy the current tracing session:
5790 * Use the man:lttng-destroy(1) command:
5800 [[list-instrumentation-points]]
5801 === List the available instrumentation points
5803 The <<lttng-sessiond,session daemon>> can query the running instrumented
5804 user applications and the Linux kernel to get a list of available
5805 instrumentation points. For the Linux kernel <<domain,tracing domain>>,
5806 they are tracepoints and system calls. For the user space tracing
5807 domain, they are tracepoints. For the other tracing domains, they are
5810 To list the available instrumentation points:
5812 * Use the man:lttng-list(1) command with the requested tracing domain's
5816 * opt:lttng-list(1):--kernel: Linux kernel tracepoints (your Unix user
5817 must be a root user, or it must be a member of the
5818 <<tracing-group,tracing group>>).
5819 * opt:lttng-list(1):--kernel with opt:lttng-list(1):--syscall: Linux
5820 kernel system calls (your Unix user must be a root user, or it must be
5821 a member of the tracing group).
5822 * opt:lttng-list(1):--userspace: user space tracepoints.
5823 * opt:lttng-list(1):--jul: `java.util.logging` loggers.
5824 * opt:lttng-list(1):--log4j: Apache log4j loggers.
5825 * opt:lttng-list(1):--python: Python loggers.
5828 .List the available user space tracepoints.
5832 $ lttng list --userspace
5836 .List the available Linux kernel system call tracepoints.
5840 $ lttng list --kernel --syscall
5845 [[enabling-disabling-events]]
5846 === Create and enable an event rule
5848 Once you <<creating-destroying-tracing-sessions,create a tracing
5849 session>>, you can create <<event,event rules>> with the
5850 man:lttng-enable-event(1) command.
5852 You specify each condition with a command-line option. The available
5853 condition options are shown in the following table.
5855 [role="growable",cols="asciidoc,asciidoc,default"]
5856 .Condition command-line options for the man:lttng-enable-event(1) command.
5858 |Option |Description |Applicable tracing domains
5864 . +--probe=__ADDR__+
5865 . +--function=__ADDR__+
5868 Instead of using the default _tracepoint_ instrumentation type, use:
5870 . A Linux system call.
5871 . A Linux https://lwn.net/Articles/132196/[KProbe] (symbol or address).
5872 . The entry and return points of a Linux function (symbol or address).
5876 |First positional argument.
5879 Tracepoint or system call name. In the case of a Linux KProbe or
5880 function, this is a custom name given to the event rule. With the
5881 JUL, log4j, and Python domains, this is a logger name.
5883 With a tracepoint, logger, or system call name, the last character
5884 can be `*` to match anything that remains.
5891 . +--loglevel=__LEVEL__+
5892 . +--loglevel-only=__LEVEL__+
5895 . Match only tracepoints or log statements with a logging level at
5896 least as severe as +__LEVEL__+.
5897 . Match only tracepoints or log statements with a logging level
5898 equal to +__LEVEL__+.
5900 See man:lttng-enable-event(1) for the list of available logging level
5903 |User space, JUL, log4j, and Python.
5905 |+--exclude=__EXCLUSIONS__+
5908 When you use a `*` character at the end of the tracepoint or logger
5909 name (first positional argument), exclude the specific names in the
5910 comma-delimited list +__EXCLUSIONS__+.
5913 User space, JUL, log4j, and Python.
5915 |+--filter=__EXPR__+
5918 Match only events which satisfy the expression +__EXPR__+.
5920 See man:lttng-enable-event(1) to learn more about the syntax of a
5927 You attach an event rule to a <<channel,channel>> on creation. If you do
5928 not specify the channel with the opt:lttng-enable-event(1):--channel
5929 option, and if the event rule to create is the first in its
5930 <<domain,tracing domain>> for a given tracing session, then LTTng
5931 creates a _default channel_ for you. This default channel is reused in
5932 subsequent invocations of the man:lttng-enable-event(1) command for the
5933 same tracing domain.
5935 An event rule is always enabled at creation time.
5937 The following examples show how you can combine the previous
5938 command-line options to create simple to more complex event rules.
5940 .Create an event rule targetting a Linux kernel tracepoint (default channel).
5944 $ lttng enable-event --kernel sched_switch
5948 .Create an event rule matching four Linux kernel system calls (default channel).
5952 $ lttng enable-event --kernel --syscall open,write,read,close
5956 .Create event rules matching tracepoints with filter expressions (default channel).
5960 $ lttng enable-event --kernel sched_switch --filter='prev_comm == "bash"'
5965 $ lttng enable-event --kernel --all \
5966 --filter='$ctx.tid == 1988 || $ctx.tid == 1534'
5971 $ lttng enable-event --jul my_logger \
5972 --filter='$app.retriever:cur_msg_id > 3'
5975 IMPORTANT: Make sure to always quote the filter string when you
5976 use man:lttng(1) from a shell.
5979 .Create an event rule matching any user space tracepoint of a given tracepoint provider with a log level range (default channel).
5983 $ lttng enable-event --userspace my_app:'*' --loglevel=TRACE_INFO
5986 IMPORTANT: Make sure to always quote the wildcard character when you
5987 use man:lttng(1) from a shell.
5990 .Create an event rule matching multiple Python loggers with a wildcard and with exclusions (default channel).
5994 $ lttng enable-event --python my-app.'*' \
5995 --exclude='my-app.module,my-app.hello'
5999 .Create an event rule matching any Apache log4j logger with a specific log level (default channel).
6003 $ lttng enable-event --log4j --all --loglevel-only=LOG4J_WARN
6007 .Create an event rule attached to a specific channel matching a specific user space tracepoint provider and tracepoint.
6011 $ lttng enable-event --userspace my_app:my_tracepoint --channel=my-channel
6015 The event rules of a given channel form a whitelist: as soon as an
6016 emitted event passes one of them, LTTng can record the event. For
6017 example, an event named `my_app:my_tracepoint` emitted from a user space
6018 tracepoint with a `TRACE_ERROR` log level passes both of the following
6023 $ lttng enable-event --userspace my_app:my_tracepoint
6024 $ lttng enable-event --userspace my_app:my_tracepoint \
6025 --loglevel=TRACE_INFO
6028 The second event rule is redundant: the first one includes
6032 [[disable-event-rule]]
6033 === Disable an event rule
6035 To disable an event rule that you <<enabling-disabling-events,created>>
6036 previously, use the man:lttng-disable-event(1) command. This command
6037 disables _all_ the event rules (of a given tracing domain and channel)
6038 which match an instrumentation point. The other conditions are not
6039 supported as of LTTng{nbsp}{revision}.
6041 The LTTng tracer does not record an emitted event which passes
6042 a _disabled_ event rule.
6044 .Disable an event rule matching a Python logger (default channel).
6048 $ lttng disable-event --python my-logger
6052 .Disable an event rule matching all `java.util.logging` loggers (default channel).
6056 $ lttng disable-event --jul '*'
6060 .Disable _all_ the event rules of the default channel.
6062 The opt:lttng-disable-event(1):--all-events option is not, like the
6063 opt:lttng-enable-event(1):--all option of man:lttng-enable-event(1), the
6064 equivalent of the event name `*` (wildcard): it disables _all_ the event
6065 rules of a given channel.
6069 $ lttng disable-event --jul --all-events
6073 NOTE: You cannot delete an event rule once you create it.
6077 === Get the status of a tracing session
6079 To get the status of the current tracing session, that is, its
6080 parameters, its channels, event rules, and their attributes:
6082 * Use the man:lttng-status(1) command:
6092 To get the status of any tracing session:
6094 * Use the man:lttng-list(1) command with the tracing session's name:
6099 $ lttng list my-session
6103 Replace `my-session` with the desired tracing session's name.
6106 [[basic-tracing-session-control]]
6107 === Start and stop a tracing session
6109 Once you <<creating-destroying-tracing-sessions,create a tracing
6111 <<enabling-disabling-events,create one or more event rules>>,
6112 you can start and stop the tracers for this tracing session.
6114 To start tracing in the current tracing session:
6116 * Use the man:lttng-start(1) command:
6125 LTTng is very flexible: you can launch user applications before
6126 or after the you start the tracers. The tracers only record the events
6127 if they pass enabled event rules and if they occur while the tracers are
6130 To stop tracing in the current tracing session:
6132 * Use the man:lttng-stop(1) command:
6141 If there were <<channel-overwrite-mode-vs-discard-mode,lost event
6142 records>> or lost sub-buffers since the last time you ran
6143 man:lttng-start(1), warnings are printed when you run the
6144 man:lttng-stop(1) command.
6147 [[enabling-disabling-channels]]
6148 === Create a channel
6150 Once you create a tracing session, you can create a <<channel,channel>>
6151 with the man:lttng-enable-channel(1) command.
6153 Note that LTTng automatically creates a default channel when, for a
6154 given <<domain,tracing domain>>, no channels exist and you
6155 <<enabling-disabling-events,create>> the first event rule. This default
6156 channel is named `channel0` and its attributes are set to reasonable
6157 values. Therefore, you only need to create a channel when you need
6158 non-default attributes.
6160 You specify each non-default channel attribute with a command-line
6161 option when you use the man:lttng-enable-channel(1) command. The
6162 available command-line options are:
6164 [role="growable",cols="asciidoc,asciidoc"]
6165 .Command-line options for the man:lttng-enable-channel(1) command.
6167 |Option |Description
6173 <<channel-overwrite-mode-vs-discard-mode,event loss mode>> instead of
6174 the default _discard_ mode.
6176 |`--buffers-pid` (user space tracing domain only)
6179 Use the per-process <<channel-buffering-schemes,buffering scheme>>
6180 instead of the default per-user buffering scheme.
6182 |+--subbuf-size=__SIZE__+
6185 Allocate sub-buffers of +__SIZE__+ bytes (power of two), for each CPU,
6186 either for each Unix user (default), or for each instrumented process.
6188 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6190 |+--num-subbuf=__COUNT__+
6193 Allocate +__COUNT__+ sub-buffers (power of two), for each CPU, either
6194 for each Unix user (default), or for each instrumented process.
6196 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6198 |+--tracefile-size=__SIZE__+
6201 Set the maximum size of each trace file that this channel writes within
6202 a stream to +__SIZE__+ bytes instead of no maximum.
6204 See <<tracefile-rotation,Trace file count and size>>.
6206 |+--tracefile-count=__COUNT__+
6209 Limit the number of trace files that this channel creates to
6210 +__COUNT__+ channels instead of no limit.
6212 See <<tracefile-rotation,Trace file count and size>>.
6214 |+--switch-timer=__PERIODUS__+
6217 Set the <<channel-switch-timer,switch timer period>>
6218 to +__PERIODUS__+{nbsp}µs.
6220 |+--read-timer=__PERIODUS__+
6223 Set the <<channel-read-timer,read timer period>>
6224 to +__PERIODUS__+{nbsp}µs.
6226 |+--output=__TYPE__+ (Linux kernel tracing domain only)
6229 Set the channel's output type to +__TYPE__+, either `mmap` or `splice`.
6233 You can only create a channel in the Linux kernel and user space
6234 <<domain,tracing domains>>: other tracing domains have their own channel
6235 created on the fly when <<enabling-disabling-events,creating event
6240 Because of a current LTTng limitation, you must create all channels
6241 _before_ you <<basic-tracing-session-control,start tracing>> in a given
6242 tracing session, that is, before the first time you run
6245 Since LTTng automatically creates a default channel when you use the
6246 man:lttng-enable-event(1) command with a specific tracing domain, you
6247 cannot, for example, create a Linux kernel event rule, start tracing,
6248 and then create a user space event rule, because no user space channel
6249 exists yet and it's too late to create one.
6251 For this reason, make sure to configure your channels properly
6252 before starting the tracers for the first time!
6255 The following examples show how you can combine the previous
6256 command-line options to create simple to more complex channels.
6258 .Create a Linux kernel channel with default attributes.
6262 $ lttng enable-channel --kernel my-channel
6266 .Create a user space channel with 4 sub-buffers or 1{nbsp}MiB each, per CPU, per instrumented process.
6270 $ lttng enable-channel --userspace --num-subbuf=4 --subbuf-size=1M \
6271 --buffers-pid my-channel
6275 .Create a Linux kernel channel which rotates 8 trace files of 4{nbsp}MiB each for each stream
6279 $ lttng enable-channel --kernel --tracefile-count=8 \
6280 --tracefile-size=4194304 my-channel
6284 .Create a user space channel in overwrite (or _flight recorder_) mode.
6288 $ lttng enable-channel --userspace --overwrite my-channel
6292 You can <<enabling-disabling-events,create>> the same event rule in
6293 two different channels:
6297 $ lttng enable-event --userspace --channel=my-channel app:tp
6298 $ lttng enable-event --userspace --channel=other-channel app:tp
6301 If both channels are enabled, when a tracepoint named `app:tp` is
6302 reached, LTTng records two events, one for each channel.
6306 === Disable a channel
6308 To disable a specific channel that you <<enabling-disabling-channels,created>>
6309 previously, use the man:lttng-disable-channel(1) command.
6311 .Disable a specific Linux kernel channel.
6315 $ lttng disable-channel --kernel my-channel
6319 The state of a channel precedes the individual states of event rules
6320 attached to it: event rules which belong to a disabled channel, even if
6321 they are enabled, are also considered disabled.
6325 === Add context fields to a channel
6327 Event record fields in trace files provide important information about
6328 events that occured previously, but sometimes some external context may
6329 help you solve a problem faster. Examples of context fields are:
6331 * The **process ID**, **thread ID**, **process name**, and
6332 **process priority** of the thread in which the event occurs.
6333 * The **hostname** of the system on which the event occurs.
6334 * The current values of many possible **performance counters** using
6336 ** CPU cycles, stalled cycles, idle cycles, and the other cycle types.
6338 ** Branch instructions, misses, and loads.
6340 * Any context defined at the application level (supported for the
6341 JUL and log4j <<domain,tracing domains>>).
6343 To get the full list of available context fields, see
6344 `lttng add-context --list`. Some context fields are reserved for a
6345 specific <<domain,tracing domain>> (Linux kernel or user space).
6347 You add context fields to <<channel,channels>>. All the events
6348 that a channel with added context fields records contain those fields.
6350 To add context fields to one or all the channels of a given tracing
6353 * Use the man:lttng-add-context(1) command.
6355 .Add context fields to all the channels of the current tracing session.
6357 The following command line adds the virtual process identifier and
6358 the per-thread CPU cycles count fields to all the user space channels
6359 of the current tracing session.
6363 $ lttng add-context --userspace --type=vpid --type=perf:thread:cpu-cycles
6367 .Add performance counter context fields by raw ID
6369 See man:lttng-add-context(1) for the exact format of the context field
6370 type, which is partly compatible with the format used in
6375 $ lttng add-context --userspace --type=perf:thread:raw:r0110:test
6376 $ lttng add-context --kernel --type=perf:cpu:raw:r0013c:x86unhalted
6380 .Add a context field to a specific channel.
6382 The following command line adds the thread identifier context field
6383 to the Linux kernel channel named `my-channel` in the current
6388 $ lttng add-context --kernel --channel=my-channel --type=tid
6392 .Add an application-specific context field to a specific channel.
6394 The following command line adds the `cur_msg_id` context field of the
6395 `retriever` context retriever for all the instrumented
6396 <<java-application,Java applications>> recording <<event,event records>>
6397 in the channel named `my-channel`:
6401 $ lttng add-context --kernel --channel=my-channel \
6402 --type='$app:retriever:cur_msg_id'
6405 IMPORTANT: Make sure to always quote the `$` character when you
6406 use man:lttng-add-context(1) from a shell.
6409 NOTE: You cannot remove context fields from a channel once you add it.
6414 === Track process IDs
6416 It's often useful to allow only specific process IDs (PIDs) to emit
6417 events. For example, you may wish to record all the system calls made by
6418 a given process (à la http://linux.die.net/man/1/strace[strace]).
6420 The man:lttng-track(1) and man:lttng-untrack(1) commands serve this
6421 purpose. Both commands operate on a whitelist of process IDs. You _add_
6422 entries to this whitelist with the man:lttng-track(1) command and remove
6423 entries with the man:lttng-untrack(1) command. Any process which has one
6424 of the PIDs in the whitelist is allowed to emit LTTng events which pass
6425 an enabled <<event,event rule>>.
6427 NOTE: The PID tracker tracks the _numeric process IDs_. Should a
6428 process with a given tracked ID exit and another process be given this
6429 ID, then the latter would also be allowed to emit events.
6431 .Track and untrack process IDs.
6433 For the sake of the following example, assume the target system has 16
6437 <<creating-destroying-tracing-sessions,create a tracing session>>,
6438 the whitelist contains all the possible PIDs:
6441 .All PIDs are tracked.
6442 image::track-all.png[]
6444 When the whitelist is full and you use the man:lttng-track(1) command to
6445 specify some PIDs to track, LTTng first clears the whitelist, then it
6446 tracks the specific PIDs. After:
6450 $ lttng track --pid=3,4,7,10,13
6456 .PIDs 3, 4, 7, 10, and 13 are tracked.
6457 image::track-3-4-7-10-13.png[]
6459 You can add more PIDs to the whitelist afterwards:
6463 $ lttng track --pid=1,15,16
6469 .PIDs 1, 15, and 16 are added to the whitelist.
6470 image::track-1-3-4-7-10-13-15-16.png[]
6472 The man:lttng-untrack(1) command removes entries from the PID tracker's
6473 whitelist. Given the previous example, the following command:
6477 $ lttng untrack --pid=3,7,10,13
6480 leads to this whitelist:
6483 .PIDs 3, 7, 10, and 13 are removed from the whitelist.
6484 image::track-1-4-15-16.png[]
6486 LTTng can track all possible PIDs again using the opt:track(1):--all
6491 $ lttng track --pid --all
6494 The result is, again:
6497 .All PIDs are tracked.
6498 image::track-all.png[]
6501 .Track only specific PIDs
6503 A very typical use case with PID tracking is to start with an empty
6504 whitelist, then <<basic-tracing-session-control,start the tracers>>, and
6505 then add PIDs manually while tracers are active. You can accomplish this
6506 by using the opt:lttng-untrack(1):--all option of the
6507 man:lttng-untrack(1) command to clear the whitelist after you
6508 <<creating-destroying-tracing-sessions,create a tracing session>>:
6512 $ lttng untrack --pid --all
6518 .No PIDs are tracked.
6519 image::untrack-all.png[]
6521 If you trace with this whitelist configuration, the tracer records no
6522 events for this <<domain,tracing domain>> because no processes are
6523 tracked. You can use the man:lttng-track(1) command as usual to track
6524 specific PIDs, for example:
6528 $ lttng track --pid=6,11
6534 .PIDs 6 and 11 are tracked.
6535 image::track-6-11.png[]
6540 [[saving-loading-tracing-session]]
6541 === Save and load tracing session configurations
6543 Configuring a <<tracing-session,tracing session>> can be long. Some of
6544 the tasks involved are:
6546 * <<enabling-disabling-channels,Create channels>> with
6547 specific attributes.
6548 * <<adding-context,Add context fields>> to specific channels.
6549 * <<enabling-disabling-events,Create event rules>> with specific log
6550 level and filter conditions.
6552 If you use LTTng to solve real world problems, chances are you have to
6553 record events using the same tracing session setup over and over,
6554 modifying a few variables each time in your instrumented program
6555 or environment. To avoid constant tracing session reconfiguration,
6556 the man:lttng(1) command-line tool can save and load tracing session
6557 configurations to/from XML files.
6559 To save a given tracing session configuration:
6561 * Use the man:lttng-save(1) command:
6566 $ lttng save my-session
6570 Replace `my-session` with the name of the tracing session to save.
6572 LTTng saves tracing session configurations to
6573 dir:{$LTTNG_HOME/.lttng/sessions} by default. Note that the
6574 env:LTTNG_HOME environment variable defaults to `$HOME` if not set. Use
6575 the opt:lttng-save(1):--output-path option to change this destination
6578 LTTng saves all configuration parameters, for example:
6580 * The tracing session name.
6581 * The trace data output path.
6582 * The channels with their state and all their attributes.
6583 * The context fields you added to channels.
6584 * The event rules with their state, log level and filter conditions.
6586 To load a tracing session:
6588 * Use the man:lttng-load(1) command:
6593 $ lttng load my-session
6597 Replace `my-session` with the name of the tracing session to load.
6599 When LTTng loads a configuration, it restores your saved tracing session
6600 as if you just configured it manually.
6602 See man:lttng(1) for the complete list of command-line options. You
6603 can also save and load all many sessions at a time, and decide in which
6604 directory to output the XML files.
6607 [[sending-trace-data-over-the-network]]
6608 === Send trace data over the network
6610 LTTng can send the recorded trace data to a remote system over the
6611 network instead of writing it to the local file system.
6613 To send the trace data over the network:
6615 . On the _remote_ system (which can also be the target system),
6616 start an LTTng <<lttng-relayd,relay daemon>> (man:lttng-relayd(8)):
6625 . On the _target_ system, create a tracing session configured to
6626 send trace data over the network:
6631 $ lttng create my-session --set-url=net://remote-system
6635 Replace `remote-system` by the host name or IP address of the
6636 remote system. See man:lttng-create(1) for the exact URL format.
6638 . On the target system, use the man:lttng(1) command-line tool as usual.
6639 When tracing is active, the target's consumer daemon sends sub-buffers
6640 to the relay daemon running on the remote system instead of flushing
6641 them to the local file system. The relay daemon writes the received
6642 packets to the local file system.
6644 The relay daemon writes trace files to
6645 +$LTTNG_HOME/lttng-traces/__hostname__/__session__+ by default, where
6646 +__hostname__+ is the host name of the target system and +__session__+
6647 is the tracing session name. Note that the env:LTTNG_HOME environment
6648 variable defaults to `$HOME` if not set. Use the
6649 opt:lttng-relayd(8):--output option of man:lttng-relayd(8) to write
6650 trace files to another base directory.
6655 === View events as LTTng emits them (noch:{LTTng} live)
6657 LTTng live is a network protocol implemented by the <<lttng-relayd,relay
6658 daemon>> (man:lttng-relayd(8)) to allow compatible trace viewers to
6659 display events as LTTng emits them on the target system while tracing is
6662 The relay daemon creates a _tee_: it forwards the trace data to both
6663 the local file system and to connected live viewers:
6666 .The relay daemon creates a _tee_, forwarding the trace data to both trace files and a connected live viewer.
6671 . On the _target system_, create a <<tracing-session,tracing session>>
6677 $ lttng create my-session --live
6681 This spawns a local relay daemon.
6683 . Start the live viewer and configure it to connect to the relay
6684 daemon. For example, with http://diamon.org/babeltrace[Babeltrace]:
6689 $ babeltrace --input-format=lttng-live \
6690 net://localhost/host/hostname/my-session
6697 * `hostname` with the host name of the target system.
6698 * `my-session` with the name of the tracing session to view.
6701 . Configure the tracing session as usual with the man:lttng(1)
6702 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6704 You can list the available live tracing sessions with Babeltrace:
6708 $ babeltrace --input-format=lttng-live net://localhost
6711 You can start the relay daemon on another system. In this case, you need
6712 to specify the relay daemon's URL when you create the tracing session
6713 with the opt:lttng-create(1):--set-url option. You also need to replace
6714 `localhost` in the procedure above with the host name of the system on
6715 which the relay daemon is running.
6717 See man:lttng-create(1) and man:lttng-relayd(8) for the complete list of
6718 command-line options.
6722 [[taking-a-snapshot]]
6723 === Take a snapshot of the current sub-buffers of a tracing session
6725 The normal behavior of LTTng is to append full sub-buffers to growing
6726 trace data files. This is ideal to keep a full history of the events
6727 that occurred on the target system, but it can
6728 represent too much data in some situations. For example, you may wish
6729 to trace your application continuously until some critical situation
6730 happens, in which case you only need the latest few recorded
6731 events to perform the desired analysis, not multi-gigabyte trace files.
6733 With the man:lttng-snapshot(1) command, you can take a snapshot of the
6734 current sub-buffers of a given <<tracing-session,tracing session>>.
6735 LTTng can write the snapshot to the local file system or send it over
6740 . Create a tracing session in _snapshot mode_:
6745 $ lttng create my-session --snapshot
6749 The <<channel-overwrite-mode-vs-discard-mode,event loss mode>> of
6750 <<channel,channels>> created in this mode is automatically set to
6751 _overwrite_ (flight recorder mode).
6753 . Configure the tracing session as usual with the man:lttng(1)
6754 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6756 . **Optional**: When you need to take a snapshot,
6757 <<basic-tracing-session-control,stop tracing>>.
6759 You can take a snapshot when the tracers are active, but if you stop
6760 them first, you are sure that the data in the sub-buffers does not
6761 change before you actually take the snapshot.
6768 $ lttng snapshot record --name=my-first-snapshot
6772 LTTng writes the current sub-buffers of all the current tracing
6773 session's channels to trace files on the local file system. Those trace
6774 files have `my-first-snapshot` in their name.
6776 There is no difference between the format of a normal trace file and the
6777 format of a snapshot: viewers of LTTng traces also support LTTng
6780 By default, LTTng writes snapshot files to the path shown by
6781 `lttng snapshot list-output`. You can change this path or decide to send
6782 snapshots over the network using either:
6784 . An output path or URL that you specify when you create the
6786 . An snapshot output path or URL that you add using
6787 `lttng snapshot add-output`
6788 . An output path or URL that you provide directly to the
6789 `lttng snapshot record` command.
6791 Method 3 overrides method 2, which overrides method 1. When you
6792 specify a URL, a relay daemon must listen on a remote system (see
6793 <<sending-trace-data-over-the-network,Send trace data over the network>>).
6798 === Use the machine interface
6800 With any command of the man:lttng(1) command-line tool, you can set the
6801 opt:lttng(1):--mi option to `xml` (before the command name) to get an
6802 XML machine interface output, for example:
6806 $ lttng --mi=xml enable-event --kernel --syscall open
6809 A schema definition (XSD) is
6810 https://github.com/lttng/lttng-tools/blob/stable-2.9/src/common/mi-lttng-3.0.xsd[available]
6811 to ease the integration with external tools as much as possible.
6815 [[metadata-regenerate]]
6816 === Regenerate the metadata of an LTTng trace
6818 An LTTng trace, which is a http://diamon.org/ctf[CTF] trace, has both
6819 data stream files and a metadata file. This metadata file contains,
6820 amongst other things, information about the offset of the clock sources
6821 used to timestamp <<event,event records>> when tracing.
6823 If, once a <<tracing-session,tracing session>> is
6824 <<basic-tracing-session-control,started>>, a major
6825 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
6826 happens, the trace's clock offset also needs to be updated. You
6827 can use the `metadata` item of the man:lttng-regenerate(1) command
6830 The main use case of this command is to allow a system to boot with
6831 an incorrect wall time and trace it with LTTng before its wall time
6832 is corrected. Once the system is known to be in a state where its
6833 wall time is correct, it can run `lttng regenerate metadata`.
6835 To regenerate the metadata of an LTTng trace:
6837 * Use the `metadata` item of the man:lttng-regenerate(1) command:
6842 $ lttng regenerate metadata
6848 `lttng regenerate metadata` has the following limitations:
6850 * Tracing session <<creating-destroying-tracing-sessions,created>>
6852 * User space <<channel,channels>>, if any, are using
6853 <<channel-buffering-schemes,per-user buffering>>.
6858 [[regenerate-statedump]]
6859 === Regenerate the state dump of a tracing session
6861 The LTTng kernel and user space tracers generate state dump
6862 <<event,event records>> when the application starts or when you
6863 <<basic-tracing-session-control,start a tracing session>>. An analysis
6864 can use the state dump event records to set an initial state before it
6865 builds the rest of the state from the following event records.
6866 http://tracecompass.org/[Trace Compass] is a notable example of an
6867 application which uses the state dump of an LTTng trace.
6869 When you <<taking-a-snapshot,take a snapshot>>, it's possible that the
6870 state dump event records are not included in the snapshot because they
6871 were recorded to a sub-buffer that has been consumed or overwritten
6874 You can use the `lttng regenerate statedump` command to emit the state
6875 dump event records again.
6877 To regenerate the state dump of the current tracing session, provided
6878 create it in snapshot mode, before you take a snapshot:
6880 . Use the `statedump` item of the man:lttng-regenerate(1) command:
6885 $ lttng regenerate statedump
6889 . <<basic-tracing-session-control,Stop the tracing session>>:
6898 . <<taking-a-snapshot,Take a snapshot>>:
6903 $ lttng snapshot record --name=my-snapshot
6907 Depending on the event throughput, you should run steps 1 and 2
6908 as closely as possible.
6910 NOTE: To record the state dump events, you need to
6911 <<enabling-disabling-events,create event rules>> which enable them.
6912 LTTng-UST state dump tracepoints start with `lttng_ust_statedump:`.
6913 LTTng-modules state dump tracepoints start with `lttng_statedump_`.
6917 [[persistent-memory-file-systems]]
6918 === Record trace data on persistent memory file systems
6920 https://en.wikipedia.org/wiki/Non-volatile_random-access_memory[Non-volatile random-access memory]
6921 (NVRAM) is random-access memory that retains its information when power
6922 is turned off (non-volatile). Systems with such memory can store data
6923 structures in RAM and retrieve them after a reboot, without flushing
6924 to typical _storage_.
6926 Linux supports NVRAM file systems thanks to either
6927 http://pramfs.sourceforge.net/[PRAMFS] or
6928 https://www.kernel.org/doc/Documentation/filesystems/dax.txt[DAX]{nbsp}+{nbsp}http://lkml.iu.edu/hypermail/linux/kernel/1504.1/03463.html[pmem]
6929 (requires Linux 4.1+).
6931 This section does not describe how to operate such file systems;
6932 we assume that you have a working persistent memory file system.
6934 When you create a <<tracing-session,tracing session>>, you can specify
6935 the path of the shared memory holding the sub-buffers. If you specify a
6936 location on an NVRAM file system, then you can retrieve the latest
6937 recorded trace data when the system reboots after a crash.
6939 To record trace data on a persistent memory file system and retrieve the
6940 trace data after a system crash:
6942 . Create a tracing session with a sub-buffer shared memory path located
6943 on an NVRAM file system:
6948 $ lttng create my-session --shm-path=/path/to/shm
6952 . Configure the tracing session as usual with the man:lttng(1)
6953 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6955 . After a system crash, use the man:lttng-crash(1) command-line tool to
6956 view the trace data recorded on the NVRAM file system:
6961 $ lttng-crash /path/to/shm
6965 The binary layout of the ring buffer files is not exactly the same as
6966 the trace files layout. This is why you need to use man:lttng-crash(1)
6967 instead of your preferred trace viewer directly.
6969 To convert the ring buffer files to LTTng trace files:
6971 * Use the opt:lttng-crash(1):--extract option of man:lttng-crash(1):
6976 $ lttng-crash --extract=/path/to/trace /path/to/shm
6984 [[lttng-modules-ref]]
6985 === noch:{LTTng-modules}
6989 [[lttng-tracepoint-enum]]
6990 ==== `LTTNG_TRACEPOINT_ENUM()` usage
6992 Use the `LTTNG_TRACEPOINT_ENUM()` macro to define an enumeration:
6996 LTTNG_TRACEPOINT_ENUM(name, TP_ENUM_VALUES(entries))
7001 * `name` with the name of the enumeration (C identifier, unique
7002 amongst all the defined enumerations).
7003 * `entries` with a list of enumeration entries.
7005 The available enumeration entry macros are:
7007 +ctf_enum_value(__name__, __value__)+::
7008 Entry named +__name__+ mapped to the integral value +__value__+.
7010 +ctf_enum_range(__name__, __begin__, __end__)+::
7011 Entry named +__name__+ mapped to the range of integral values between
7012 +__begin__+ (included) and +__end__+ (included).
7014 +ctf_enum_auto(__name__)+::
7015 Entry named +__name__+ mapped to the integral value following the
7016 last mapping's value.
7018 The last value of a `ctf_enum_value()` entry is its +__value__+
7021 The last value of a `ctf_enum_range()` entry is its +__end__+ parameter.
7023 If `ctf_enum_auto()` is the first entry in the list, its integral
7026 Use the `ctf_enum()` <<lttng-modules-tp-fields,field definition macro>>
7027 to use a defined enumeration as a tracepoint field.
7029 .Define an enumeration with `LTTNG_TRACEPOINT_ENUM()`.
7033 LTTNG_TRACEPOINT_ENUM(
7036 ctf_enum_auto("AUTO: EXPECT 0")
7037 ctf_enum_value("VALUE: 23", 23)
7038 ctf_enum_value("VALUE: 27", 27)
7039 ctf_enum_auto("AUTO: EXPECT 28")
7040 ctf_enum_range("RANGE: 101 TO 303", 101, 303)
7041 ctf_enum_auto("AUTO: EXPECT 304")
7049 [[lttng-modules-tp-fields]]
7050 ==== Tracepoint fields macros (for `TP_FIELDS()`)
7052 [[tp-fast-assign]][[tp-struct-entry]]The available macros to define
7053 tracepoint fields, which must be listed within `TP_FIELDS()` in
7054 `LTTNG_TRACEPOINT_EVENT()`, are:
7056 [role="func-desc growable",cols="asciidoc,asciidoc"]
7057 .Available macros to define LTTng-modules tracepoint fields
7059 |Macro |Description and parameters
7062 +ctf_integer(__t__, __n__, __e__)+
7064 +ctf_integer_nowrite(__t__, __n__, __e__)+
7066 +ctf_user_integer(__t__, __n__, __e__)+
7068 +ctf_user_integer_nowrite(__t__, __n__, __e__)+
7070 Standard integer, displayed in base 10.
7073 Integer C type (`int`, `long`, `size_t`, ...).
7079 Argument expression.
7082 +ctf_integer_hex(__t__, __n__, __e__)+
7084 +ctf_user_integer_hex(__t__, __n__, __e__)+
7086 Standard integer, displayed in base 16.
7095 Argument expression.
7097 |+ctf_integer_oct(__t__, __n__, __e__)+
7099 Standard integer, displayed in base 8.
7108 Argument expression.
7111 +ctf_integer_network(__t__, __n__, __e__)+
7113 +ctf_user_integer_network(__t__, __n__, __e__)+
7115 Integer in network byte order (big-endian), displayed in base 10.
7124 Argument expression.
7127 +ctf_integer_network_hex(__t__, __n__, __e__)+
7129 +ctf_user_integer_network_hex(__t__, __n__, __e__)+
7131 Integer in network byte order, displayed in base 16.
7140 Argument expression.
7143 +ctf_enum(__N__, __t__, __n__, __e__)+
7145 +ctf_enum_nowrite(__N__, __t__, __n__, __e__)+
7147 +ctf_user_enum(__N__, __t__, __n__, __e__)+
7149 +ctf_user_enum_nowrite(__N__, __t__, __n__, __e__)+
7154 Name of a <<lttng-tracepoint-enum,previously defined enumeration>>.
7157 Integer C type (`int`, `long`, `size_t`, ...).
7163 Argument expression.
7166 +ctf_string(__n__, __e__)+
7168 +ctf_string_nowrite(__n__, __e__)+
7170 +ctf_user_string(__n__, __e__)+
7172 +ctf_user_string_nowrite(__n__, __e__)+
7174 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
7180 Argument expression.
7183 +ctf_array(__t__, __n__, __e__, __s__)+
7185 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
7187 +ctf_user_array(__t__, __n__, __e__, __s__)+
7189 +ctf_user_array_nowrite(__t__, __n__, __e__, __s__)+
7191 Statically-sized array of integers.
7194 Array element C type.
7200 Argument expression.
7206 +ctf_array_bitfield(__t__, __n__, __e__, __s__)+
7208 +ctf_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7210 +ctf_user_array_bitfield(__t__, __n__, __e__, __s__)+
7212 +ctf_user_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7214 Statically-sized array of bits.
7216 The type of +__e__+ must be an integer type. +__s__+ is the number
7217 of elements of such type in +__e__+, not the number of bits.
7220 Array element C type.
7226 Argument expression.
7232 +ctf_array_text(__t__, __n__, __e__, __s__)+
7234 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
7236 +ctf_user_array_text(__t__, __n__, __e__, __s__)+
7238 +ctf_user_array_text_nowrite(__t__, __n__, __e__, __s__)+
7240 Statically-sized array, printed as text.
7242 The string does not need to be null-terminated.
7245 Array element C type (always `char`).
7251 Argument expression.
7257 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
7259 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7261 +ctf_user_sequence(__t__, __n__, __e__, __T__, __E__)+
7263 +ctf_user_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7265 Dynamically-sized array of integers.
7267 The type of +__E__+ must be unsigned.
7270 Array element C type.
7276 Argument expression.
7279 Length expression C type.
7285 +ctf_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7287 +ctf_user_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7289 Dynamically-sized array of integers, displayed in base 16.
7291 The type of +__E__+ must be unsigned.
7294 Array element C type.
7300 Argument expression.
7303 Length expression C type.
7308 |+ctf_sequence_network(__t__, __n__, __e__, __T__, __E__)+
7310 Dynamically-sized array of integers in network byte order (big-endian),
7311 displayed in base 10.
7313 The type of +__E__+ must be unsigned.
7316 Array element C type.
7322 Argument expression.
7325 Length expression C type.
7331 +ctf_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7333 +ctf_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7335 +ctf_user_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7337 +ctf_user_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7339 Dynamically-sized array of bits.
7341 The type of +__e__+ must be an integer type. +__s__+ is the number
7342 of elements of such type in +__e__+, not the number of bits.
7344 The type of +__E__+ must be unsigned.
7347 Array element C type.
7353 Argument expression.
7356 Length expression C type.
7362 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7364 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7366 +ctf_user_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7368 +ctf_user_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7370 Dynamically-sized array, displayed as text.
7372 The string does not need to be null-terminated.
7374 The type of +__E__+ must be unsigned.
7376 The behaviour is undefined if +__e__+ is `NULL`.
7379 Sequence element C type (always `char`).
7385 Argument expression.
7388 Length expression C type.
7394 Use the `_user` versions when the argument expression, `e`, is
7395 a user space address. In the cases of `ctf_user_integer*()` and
7396 `ctf_user_float*()`, `&e` must be a user space address, thus `e` must
7399 The `_nowrite` versions omit themselves from the session trace, but are
7400 otherwise identical. This means the `_nowrite` fields won't be written
7401 in the recorded trace. Their primary purpose is to make some
7402 of the event context available to the
7403 <<enabling-disabling-events,event filters>> without having to
7404 commit the data to sub-buffers.
7410 Terms related to LTTng and to tracing in general:
7413 The http://diamon.org/babeltrace[Babeltrace] project, which includes
7414 the cmd:babeltrace command, some libraries, and Python bindings.
7416 <<channel-buffering-schemes,buffering scheme>>::
7417 A layout of sub-buffers applied to a given channel.
7419 <<channel,channel>>::
7420 An entity which is responsible for a set of ring buffers.
7422 <<event,Event rules>> are always attached to a specific channel.
7425 A reference of time for a tracer.
7427 <<lttng-consumerd,consumer daemon>>::
7428 A process which is responsible for consuming the full sub-buffers
7429 and write them to a file system or send them over the network.
7431 <<channel-overwrite-mode-vs-discard-mode,discard mode>>:: The event loss
7432 mode in which the tracer _discards_ new event records when there's no
7433 sub-buffer space left to store them.
7436 The consequence of the execution of an instrumentation
7437 point, like a tracepoint that you manually place in some source code,
7438 or a Linux kernel KProbe.
7440 An event is said to _occur_ at a specific time. Different actions can
7441 be taken upon the occurrence of an event, like record the event's payload
7444 <<channel-overwrite-mode-vs-discard-mode,event loss mode>>::
7445 The mechanism by which event records of a given channel are lost
7446 (not recorded) when there is no sub-buffer space left to store them.
7448 [[def-event-name]]event name::
7449 The name of an event, which is also the name of the event record.
7450 This is also called the _instrumentation point name_.
7453 A record, in a trace, of the payload of an event which occured.
7455 <<event,event rule>>::
7456 Set of conditions which must be satisfied for one or more occuring
7457 events to be recorded.
7459 `java.util.logging`::
7461 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[core logging facilities].
7463 <<instrumenting,instrumentation>>::
7464 The use of LTTng probes to make a piece of software traceable.
7466 instrumentation point::
7467 A point in the execution path of a piece of software that, when
7468 reached by this execution, can emit an event.
7470 instrumentation point name::
7471 See _<<def-event-name,event name>>_.
7474 A http://logging.apache.org/log4j/1.2/[logging library] for Java
7475 developed by the Apache Software Foundation.
7478 Level of severity of a log statement or user space
7479 instrumentation point.
7482 The _Linux Trace Toolkit: next generation_ project.
7484 <<lttng-cli,cmd:lttng>>::
7485 A command-line tool provided by the LTTng-tools project which you
7486 can use to send and receive control messages to and from a
7490 The https://github.com/lttng/lttng-analyses[LTTng analyses] project,
7491 which is a set of analyzing programs that are used to obtain a
7492 higher level view of an LTTng trace.
7494 cmd:lttng-consumerd::
7495 The name of the consumer daemon program.
7498 A utility provided by the LTTng-tools project which can convert
7499 ring buffer files (usually
7500 <<persistent-memory-file-systems,saved on a persistent memory file system>>)
7503 LTTng Documentation::
7506 <<lttng-live,LTTng live>>::
7507 A communication protocol between the relay daemon and live viewers
7508 which makes it possible to see events "live", as they are received by
7511 <<lttng-modules,LTTng-modules>>::
7512 The https://github.com/lttng/lttng-modules[LTTng-modules] project,
7513 which contains the Linux kernel modules to make the Linux kernel
7514 instrumentation points available for LTTng tracing.
7517 The name of the relay daemon program.
7519 cmd:lttng-sessiond::
7520 The name of the session daemon program.
7523 The https://github.com/lttng/lttng-tools[LTTng-tools] project, which
7524 contains the various programs and libraries used to
7525 <<controlling-tracing,control tracing>>.
7527 <<lttng-ust,LTTng-UST>>::
7528 The https://github.com/lttng/lttng-ust[LTTng-UST] project, which
7529 contains libraries to instrument user applications.
7531 <<lttng-ust-agents,LTTng-UST Java agent>>::
7532 A Java package provided by the LTTng-UST project to allow the
7533 LTTng instrumentation of `java.util.logging` and Apache log4j 1.2
7536 <<lttng-ust-agents,LTTng-UST Python agent>>::
7537 A Python package provided by the LTTng-UST project to allow the
7538 LTTng instrumentation of Python logging statements.
7540 <<channel-overwrite-mode-vs-discard-mode,overwrite mode>>::
7541 The event loss mode in which new event records overwrite older
7542 event records when there's no sub-buffer space left to store them.
7544 <<channel-buffering-schemes,per-process buffering>>::
7545 A buffering scheme in which each instrumented process has its own
7546 sub-buffers for a given user space channel.
7548 <<channel-buffering-schemes,per-user buffering>>::
7549 A buffering scheme in which all the processes of a Unix user share the
7550 same sub-buffer for a given user space channel.
7552 <<lttng-relayd,relay daemon>>::
7553 A process which is responsible for receiving the trace data sent by
7554 a distant consumer daemon.
7557 A set of sub-buffers.
7559 <<lttng-sessiond,session daemon>>::
7560 A process which receives control commands from you and orchestrates
7561 the tracers and various LTTng daemons.
7563 <<taking-a-snapshot,snapshot>>::
7564 A copy of the current data of all the sub-buffers of a given tracing
7565 session, saved as trace files.
7568 One part of an LTTng ring buffer which contains event records.
7571 The time information attached to an event when it is emitted.
7574 A set of files which are the concatenations of one or more
7575 flushed sub-buffers.
7578 The action of recording the events emitted by an application
7579 or by a system, or to initiate such recording by controlling
7583 The http://tracecompass.org[Trace Compass] project and application.
7586 An instrumentation point using the tracepoint mechanism of the Linux
7587 kernel or of LTTng-UST.
7589 tracepoint definition::
7590 The definition of a single tracepoint.
7593 The name of a tracepoint.
7595 tracepoint provider::
7596 A set of functions providing tracepoints to an instrumented user
7599 Not to be confused with a _tracepoint provider package_: many tracepoint
7600 providers can exist within a tracepoint provider package.
7602 tracepoint provider package::
7603 One or more tracepoint providers compiled as an object file or as
7607 A software which records emitted events.
7609 <<domain,tracing domain>>::
7610 A namespace for event sources.
7612 <<tracing-group,tracing group>>::
7613 The Unix group in which a Unix user can be to be allowed to trace the
7616 <<tracing-session,tracing session>>::
7617 A stateful dialogue between you and a <<lttng-sessiond,session
7621 An application running in user space, as opposed to a Linux kernel
7622 module, for example.