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 22 January 2018.
345 |Distribution |Available in releases |Alternatives
347 |https://www.ubuntu.com/[Ubuntu]
348 |<<ubuntu,Ubuntu{nbsp}17.04 _Zesty Zapus_ and Ubuntu{nbsp}17.10 _Artful Aardvark_>>.
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/v2.10#doc-fedora[LTTng{nbsp}2.10 for Fedora{nbsp}27].
359 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
360 other Fedora releases.
362 |https://www.debian.org/[Debian]
363 |<<debian,Debian "stretch" (stable)>>.
364 |link:/docs/v2.10#doc-debian[LTTng{nbsp}2.10 for Debian "buster" (testing)
365 and Debian "sid" (unstable)].
368 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
369 other Debian releases.
371 |https://www.archlinux.org/[Arch Linux]
373 |link:/docs/v2.10#doc-arch-linux[LTTng{nbsp}2.10 for the current Arch Linux build].
375 <<building-from-source,Build LTTng{nbsp}{revision} from source>>.
377 |https://alpinelinux.org/[Alpine Linux]
379 |link:/docs/v2.10#doc-alpine-linux[LTTng{nbsp}2.10 for Alpine Linux{nbsp}3.7
380 and Alpine Linux{nbsp}"edge"].
382 <<building-from-source,Build LTTng{nbsp}{revision} from source>>.
384 |https://www.redhat.com/[RHEL] and https://www.suse.com/[SLES]
385 |See http://packages.efficios.com/[EfficiOS Enterprise Packages].
388 |https://buildroot.org/[Buildroot]
389 |<<"buildroot", "Buildroot{nbsp}2017.02, Buildroot{nbsp}2017.05, Buildroot{nbsp}2017.08, and Buildroot{nbsp}2017.11">>.
390 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
391 other Buildroot releases.
393 |http://www.openembedded.org/wiki/Main_Page[OpenEmbedded] and
394 https://www.yoctoproject.org/[Yocto]
395 |<<oe-yocto,Yocto Project{nbsp}2.3 _Pyro_ and Yocto Project{nbsp}2.4 _Rocko_>>
396 (`openembedded-core` layer).
397 |<<building-from-source,Build LTTng{nbsp}{revision} from source>> for
398 other Yocto/OpenEmbedded releases.
403 === [[ubuntu-official-repositories]]Ubuntu
405 LTTng{nbsp}{revision} is available on Ubuntu{nbsp}17.04 _Zesty Zapus_
406 and Ubuntu{nbsp}17.10 _Artful Aardvark_. For previous releases of
407 Ubuntu, <<ubuntu-ppa,use the LTTng Stable{nbsp}{revision} PPA>>.
409 To install LTTng{nbsp}{revision} on Ubuntu{nbsp}17.04 _Zesty Zapus_:
411 . Install the main LTTng{nbsp}{revision} packages:
416 # apt-get install lttng-tools
417 # apt-get install lttng-modules-dkms
418 # apt-get install liblttng-ust-dev
422 . **If you need to instrument and trace
423 <<java-application,Java applications>>**, install the LTTng-UST
429 # apt-get install liblttng-ust-agent-java
433 . **If you need to instrument and trace
434 <<python-application,Python{nbsp}3 applications>>**, install the
435 LTTng-UST Python agent:
440 # apt-get install python3-lttngust
446 ==== noch:{LTTng} Stable {revision} PPA
448 The https://launchpad.net/~lttng/+archive/ubuntu/stable-{revision}[LTTng
449 Stable{nbsp}{revision} PPA] offers the latest stable
450 LTTng{nbsp}{revision} packages for:
452 * Ubuntu{nbsp}14.04 _Trusty Tahr_
453 * Ubuntu{nbsp}16.04 _Xenial Xerus_
455 To install LTTng{nbsp}{revision} from the LTTng Stable{nbsp}{revision} PPA:
457 . Add the LTTng Stable{nbsp}{revision} PPA repository and update the
463 # apt-add-repository ppa:lttng/stable-2.9
468 . Install the main LTTng{nbsp}{revision} packages:
473 # apt-get install lttng-tools
474 # apt-get install lttng-modules-dkms
475 # apt-get install liblttng-ust-dev
479 . **If you need to instrument and trace
480 <<java-application,Java applications>>**, install the LTTng-UST
486 # apt-get install liblttng-ust-agent-java
490 . **If you need to instrument and trace
491 <<python-application,Python{nbsp}3 applications>>**, install the
492 LTTng-UST Python agent:
497 # apt-get install python3-lttngust
505 To install LTTng{nbsp}{revision} on Fedora{nbsp}26:
507 . Install the LTTng-tools{nbsp}{revision} and LTTng-UST{nbsp}{revision}
513 # yum install lttng-tools
514 # yum install lttng-ust
518 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
524 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.9.tar.bz2 &&
525 tar -xf lttng-modules-latest-2.9.tar.bz2 &&
526 cd lttng-modules-2.9.* &&
528 sudo make modules_install &&
534 .Java and Python application instrumentation and tracing
536 If you need to instrument and trace <<java-application,Java
537 applications>> on Fedora, you need to build and install
538 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
539 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
540 `--enable-java-agent-all` options to the `configure` script, depending
541 on which Java logging framework you use.
543 If you need to instrument and trace <<python-application,Python
544 applications>> on Fedora, you need to build and install
545 LTTng-UST{nbsp}{revision} from source and pass the
546 `--enable-python-agent` option to the `configure` script.
553 To install LTTng{nbsp}{revision} on Debian "stretch" (stable):
555 . Install the main LTTng{nbsp}{revision} packages:
560 # apt-get install lttng-modules-dkms
561 # apt-get install liblttng-ust-dev
562 # apt-get install lttng-tools
566 . **If you need to instrument and trace <<java-application,Java
567 applications>>**, install the LTTng-UST Java agent:
572 # apt-get install liblttng-ust-agent-java
576 . **If you need to instrument and trace <<python-application,Python
577 applications>>**, install the LTTng-UST Python agent:
582 # apt-get install python3-lttngust
587 [[enterprise-distributions]]
588 === RHEL, SUSE, and other enterprise distributions
590 To install LTTng on enterprise Linux distributions, such as Red Hat
591 Enterprise Linux (RHEL) and SUSE Linux Enterprise Server (SUSE), please
592 see http://packages.efficios.com/[EfficiOS Enterprise Packages].
598 To install LTTng{nbsp}{revision} on Buildroot{nbsp}2017.02,
599 Buildroot{nbsp}2017.05, Buildroot{nbsp}2017.08, or
600 Buildroot{nbsp}2017.11:
602 . Launch the Buildroot configuration tool:
611 . In **Kernel**, check **Linux kernel**.
612 . In **Toolchain**, check **Enable WCHAR support**.
613 . In **Target packages**{nbsp}→ **Debugging, profiling and benchmark**,
614 check **lttng-modules** and **lttng-tools**.
615 . In **Target packages**{nbsp}→ **Libraries**{nbsp}→
616 **Other**, check **lttng-libust**.
620 === OpenEmbedded and Yocto
622 LTTng{nbsp}{revision} recipes are available in the
623 http://layers.openembedded.org/layerindex/branch/master/layer/openembedded-core/[`openembedded-core`]
624 layer for Yocto Project{nbsp}2.3 _Pyro_ and Yocto Project{nbsp}2.4 _Rocko_
625 under the following names:
631 With BitBake, the simplest way to include LTTng recipes in your target
632 image is to add them to `IMAGE_INSTALL_append` in path:{conf/local.conf}:
635 IMAGE_INSTALL_append = " lttng-tools lttng-modules lttng-ust"
640 . Select a machine and an image recipe.
641 . Click **Edit image recipe**.
642 . Under the **All recipes** tab, search for **lttng**.
643 . Check the desired LTTng recipes.
646 .Java and Python application instrumentation and tracing
648 If you need to instrument and trace <<java-application,Java
649 applications>> on Yocto/OpenEmbedded, you need to build and install
650 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
651 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
652 `--enable-java-agent-all` options to the `configure` script, depending
653 on which Java logging framework you use.
655 If you need to instrument and trace <<python-application,Python
656 applications>> on Yocto/OpenEmbedded, you need to build and install
657 LTTng-UST{nbsp}{revision} from source and pass the
658 `--enable-python-agent` option to the `configure` script.
662 [[building-from-source]]
663 === Build from source
665 To build and install LTTng{nbsp}{revision} from source:
667 . Using your distribution's package manager, or from source, install
668 the following dependencies of LTTng-tools and LTTng-UST:
671 * https://sourceforge.net/projects/libuuid/[libuuid]
672 * http://directory.fsf.org/wiki/Popt[popt]
673 * http://liburcu.org/[Userspace RCU]
674 * http://www.xmlsoft.org/[libxml2]
677 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
683 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.9.tar.bz2 &&
684 tar -xf lttng-modules-latest-2.9.tar.bz2 &&
685 cd lttng-modules-2.9.* &&
687 sudo make modules_install &&
692 . Download, build, and install the latest LTTng-UST{nbsp}{revision}:
698 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.9.tar.bz2 &&
699 tar -xf lttng-ust-latest-2.9.tar.bz2 &&
700 cd lttng-ust-2.9.* &&
710 .Java and Python application tracing
712 If you need to instrument and trace <<java-application,Java
713 applications>>, pass the `--enable-java-agent-jul`,
714 `--enable-java-agent-log4j`, or `--enable-java-agent-all` options to the
715 `configure` script, depending on which Java logging framework you use.
717 If you need to instrument and trace <<python-application,Python
718 applications>>, pass the `--enable-python-agent` option to the
719 `configure` script. You can set the `PYTHON` environment variable to the
720 path to the Python interpreter for which to install the LTTng-UST Python
728 By default, LTTng-UST libraries are installed to
729 dir:{/usr/local/lib}, which is the de facto directory in which to
730 keep self-compiled and third-party libraries.
732 When <<building-tracepoint-providers-and-user-application,linking an
733 instrumented user application with `liblttng-ust`>>:
735 * Append `/usr/local/lib` to the env:LD_LIBRARY_PATH environment
737 * Pass the `-L/usr/local/lib` and `-Wl,-rpath,/usr/local/lib` options to
738 man:gcc(1), man:g++(1), or man:clang(1).
742 . Download, build, and install the latest LTTng-tools{nbsp}{revision}:
748 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.9.tar.bz2 &&
749 tar -xf lttng-tools-latest-2.9.tar.bz2 &&
750 cd lttng-tools-2.9.* &&
758 TIP: The https://github.com/eepp/vlttng[vlttng tool] can do all the
759 previous steps automatically for a given version of LTTng and confine
760 the installed files in a specific directory. This can be useful to test
761 LTTng without installing it on your system.
767 This is a short guide to get started quickly with LTTng kernel and user
770 Before you follow this guide, make sure to <<installing-lttng,install>>
773 This tutorial walks you through the steps to:
775 . <<tracing-the-linux-kernel,Trace the Linux kernel>>.
776 . <<tracing-your-own-user-application,Trace a user application>> written
778 . <<viewing-and-analyzing-your-traces,View and analyze the
782 [[tracing-the-linux-kernel]]
783 === Trace the Linux kernel
785 The following command lines start with the `#` prompt because you need
786 root privileges to trace the Linux kernel. You can also trace the kernel
787 as a regular user if your Unix user is a member of the
788 <<tracing-group,tracing group>>.
790 . Create a <<tracing-session,tracing session>> which writes its traces
791 to dir:{/tmp/my-kernel-trace}:
796 # lttng create my-kernel-session --output=/tmp/my-kernel-trace
800 . List the available kernel tracepoints and system calls:
805 # lttng list --kernel
806 # lttng list --kernel --syscall
810 . Create <<event,event rules>> which match the desired instrumentation
811 point names, for example the `sched_switch` and `sched_process_fork`
812 tracepoints, and the man:open(2) and man:close(2) system calls:
817 # lttng enable-event --kernel sched_switch,sched_process_fork
818 # lttng enable-event --kernel --syscall open,close
822 You can also create an event rule which matches _all_ the Linux kernel
823 tracepoints (this will generate a lot of data when tracing):
828 # lttng enable-event --kernel --all
832 . <<basic-tracing-session-control,Start tracing>>:
841 . Do some operation on your system for a few seconds. For example,
842 load a website, or list the files of a directory.
843 . <<basic-tracing-session-control,Stop tracing>> and destroy the
854 The man:lttng-destroy(1) command does not destroy the trace data; it
855 only destroys the state of the tracing session.
857 . For the sake of this example, make the recorded trace accessible to
863 # chown -R $(whoami) /tmp/my-kernel-trace
867 See <<viewing-and-analyzing-your-traces,View and analyze the
868 recorded events>> to view the recorded events.
871 [[tracing-your-own-user-application]]
872 === Trace a user application
874 This section steps you through a simple example to trace a
875 _Hello world_ program written in C.
877 To create the traceable user application:
879 . Create the tracepoint provider header file, which defines the
880 tracepoints and the events they can generate:
886 #undef TRACEPOINT_PROVIDER
887 #define TRACEPOINT_PROVIDER hello_world
889 #undef TRACEPOINT_INCLUDE
890 #define TRACEPOINT_INCLUDE "./hello-tp.h"
892 #if !defined(_HELLO_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
895 #include <lttng/tracepoint.h>
905 ctf_string(my_string_field, my_string_arg)
906 ctf_integer(int, my_integer_field, my_integer_arg)
910 #endif /* _HELLO_TP_H */
912 #include <lttng/tracepoint-event.h>
916 . Create the tracepoint provider package source file:
922 #define TRACEPOINT_CREATE_PROBES
923 #define TRACEPOINT_DEFINE
925 #include "hello-tp.h"
929 . Build the tracepoint provider package:
934 $ gcc -c -I. hello-tp.c
938 . Create the _Hello World_ application source file:
945 #include "hello-tp.h"
947 int main(int argc, char *argv[])
951 puts("Hello, World!\nPress Enter to continue...");
954 * The following getchar() call is only placed here for the purpose
955 * of this demonstration, to pause the application in order for
956 * you to have time to list its tracepoints. It is not
962 * A tracepoint() call.
964 * Arguments, as defined in hello-tp.h:
966 * 1. Tracepoint provider name (required)
967 * 2. Tracepoint name (required)
968 * 3. my_integer_arg (first user-defined argument)
969 * 4. my_string_arg (second user-defined argument)
971 * Notice the tracepoint provider and tracepoint names are
972 * NOT strings: they are in fact parts of variables that the
973 * macros in hello-tp.h create.
975 tracepoint(hello_world, my_first_tracepoint, 23, "hi there!");
977 for (x = 0; x < argc; ++x) {
978 tracepoint(hello_world, my_first_tracepoint, x, argv[x]);
981 puts("Quitting now!");
982 tracepoint(hello_world, my_first_tracepoint, x * x, "x^2");
989 . Build the application:
998 . Link the application with the tracepoint provider package,
999 `liblttng-ust`, and `libdl`:
1004 $ gcc -o hello hello.o hello-tp.o -llttng-ust -ldl
1008 Here's the whole build process:
1011 .User space tracing tutorial's build steps.
1012 image::ust-flow.png[]
1014 To trace the user application:
1016 . Run the application with a few arguments:
1021 $ ./hello world and beyond
1030 Press Enter to continue...
1034 . Start an LTTng <<lttng-sessiond,session daemon>>:
1039 $ lttng-sessiond --daemonize
1043 Note that a session daemon might already be running, for example as
1044 a service that the distribution's service manager started.
1046 . List the available user space tracepoints:
1051 $ lttng list --userspace
1055 You see the `hello_world:my_first_tracepoint` tracepoint listed
1056 under the `./hello` process.
1058 . Create a <<tracing-session,tracing session>>:
1063 $ lttng create my-user-space-session
1067 . Create an <<event,event rule>> which matches the
1068 `hello_world:my_first_tracepoint` event name:
1073 $ lttng enable-event --userspace hello_world:my_first_tracepoint
1077 . <<basic-tracing-session-control,Start tracing>>:
1086 . Go back to the running `hello` application and press Enter. The
1087 program executes all `tracepoint()` instrumentation points and exits.
1088 . <<basic-tracing-session-control,Stop tracing>> and destroy the
1099 The man:lttng-destroy(1) command does not destroy the trace data; it
1100 only destroys the state of the tracing session.
1102 By default, LTTng saves the traces in
1103 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
1104 where +__name__+ is the tracing session name. The
1105 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
1107 See <<viewing-and-analyzing-your-traces,View and analyze the
1108 recorded events>> to view the recorded events.
1111 [[viewing-and-analyzing-your-traces]]
1112 === View and analyze the recorded events
1114 Once you have completed the <<tracing-the-linux-kernel,Trace the Linux
1115 kernel>> and <<tracing-your-own-user-application,Trace a user
1116 application>> tutorials, you can inspect the recorded events.
1118 Many tools are available to read LTTng traces:
1120 * **cmd:babeltrace** is a command-line utility which converts trace
1121 formats; it supports the format that LTTng produces, CTF, as well as a
1122 basic text output which can be ++grep++ed. The cmd:babeltrace command
1123 is part of the http://diamon.org/babeltrace[Babeltrace] project.
1124 * Babeltrace also includes
1125 **https://www.python.org/[Python] bindings** so
1126 that you can easily open and read an LTTng trace with your own script,
1127 benefiting from the power of Python.
1128 * http://tracecompass.org/[**Trace Compass**]
1129 is a graphical user interface for viewing and analyzing any type of
1130 logs or traces, including LTTng's.
1131 * https://github.com/lttng/lttng-analyses[**LTTng analyses**] is a
1132 project which includes many high-level analyses of LTTng kernel
1133 traces, like scheduling statistics, interrupt frequency distribution,
1134 top CPU usage, and more.
1136 NOTE: This section assumes that the traces recorded during the previous
1137 tutorials were saved to their default location, in the
1138 dir:{$LTTNG_HOME/lttng-traces} directory. The env:LTTNG_HOME
1139 environment variable defaults to `$HOME` if not set.
1142 [[viewing-and-analyzing-your-traces-bt]]
1143 ==== Use the cmd:babeltrace command-line tool
1145 The simplest way to list all the recorded events of a trace is to pass
1146 its path to cmd:babeltrace with no options:
1150 $ babeltrace ~/lttng-traces/my-user-space-session*
1153 cmd:babeltrace finds all traces recursively within the given path and
1154 prints all their events, merging them in chronological order.
1156 You can pipe the output of cmd:babeltrace into a tool like man:grep(1) for
1161 $ babeltrace /tmp/my-kernel-trace | grep _switch
1164 You can pipe the output of cmd:babeltrace into a tool like man:wc(1) to
1165 count the recorded events:
1169 $ babeltrace /tmp/my-kernel-trace | grep _open | wc --lines
1173 [[viewing-and-analyzing-your-traces-bt-python]]
1174 ==== Use the Babeltrace Python bindings
1176 The <<viewing-and-analyzing-your-traces-bt,text output of cmd:babeltrace>>
1177 is useful to isolate events by simple matching using man:grep(1) and
1178 similar utilities. However, more elaborate filters, such as keeping only
1179 event records with a field value falling within a specific range, are
1180 not trivial to write using a shell. Moreover, reductions and even the
1181 most basic computations involving multiple event records are virtually
1182 impossible to implement.
1184 Fortunately, Babeltrace ships with Python 3 bindings which makes it easy
1185 to read the event records of an LTTng trace sequentially and compute the
1186 desired information.
1188 The following script accepts an LTTng Linux kernel trace path as its
1189 first argument and prints the short names of the top 5 running processes
1190 on CPU 0 during the whole trace:
1195 from collections import Counter
1201 if len(sys.argv) != 2:
1202 msg = 'Usage: python3 {} TRACEPATH'.format(sys.argv[0])
1203 print(msg, file=sys.stderr)
1206 # A trace collection contains one or more traces
1207 col = babeltrace.TraceCollection()
1209 # Add the trace provided by the user (LTTng traces always have
1211 if col.add_trace(sys.argv[1], 'ctf') is None:
1212 raise RuntimeError('Cannot add trace')
1214 # This counter dict contains execution times:
1216 # task command name -> total execution time (ns)
1217 exec_times = Counter()
1219 # This contains the last `sched_switch` timestamp
1223 for event in col.events:
1224 # Keep only `sched_switch` events
1225 if event.name != 'sched_switch':
1228 # Keep only events which happened on CPU 0
1229 if event['cpu_id'] != 0:
1233 cur_ts = event.timestamp
1239 # Previous task command (short) name
1240 prev_comm = event['prev_comm']
1242 # Initialize entry in our dict if not yet done
1243 if prev_comm not in exec_times:
1244 exec_times[prev_comm] = 0
1246 # Compute previous command execution time
1247 diff = cur_ts - last_ts
1249 # Update execution time of this command
1250 exec_times[prev_comm] += diff
1252 # Update last timestamp
1256 for name, ns in exec_times.most_common(5):
1258 print('{:20}{} s'.format(name, s))
1263 if __name__ == '__main__':
1264 sys.exit(0 if top5proc() else 1)
1271 $ python3 top5proc.py /tmp/my-kernel-trace/kernel
1277 swapper/0 48.607245889 s
1278 chromium 7.192738188 s
1279 pavucontrol 0.709894415 s
1280 Compositor 0.660867933 s
1281 Xorg.bin 0.616753786 s
1284 Note that `swapper/0` is the "idle" process of CPU 0 on Linux; since we
1285 weren't using the CPU that much when tracing, its first position in the
1290 == [[understanding-lttng]]Core concepts
1292 From a user's perspective, the LTTng system is built on a few concepts,
1293 or objects, on which the <<lttng-cli,cmd:lttng command-line tool>>
1294 operates by sending commands to the <<lttng-sessiond,session daemon>>.
1295 Understanding how those objects relate to eachother is key in mastering
1298 The core concepts are:
1300 * <<tracing-session,Tracing session>>
1301 * <<domain,Tracing domain>>
1302 * <<channel,Channel and ring buffer>>
1303 * <<"event","Instrumentation point, event rule, event, and event record">>
1309 A _tracing session_ is a stateful dialogue between you and
1310 a <<lttng-sessiond,session daemon>>. You can
1311 <<creating-destroying-tracing-sessions,create a new tracing
1312 session>> with the `lttng create` command.
1314 Anything that you do when you control LTTng tracers happens within a
1315 tracing session. In particular, a tracing session:
1318 * Has its own set of trace files.
1319 * Has its own state of activity (started or stopped).
1320 * Has its own <<tracing-session-mode,mode>> (local, network streaming,
1322 * Has its own <<channel,channels>> which have their own
1323 <<event,event rules>>.
1326 .A _tracing session_ contains <<channel,channels>> that are members of <<domain,tracing domains>> and contain <<event,event rules>>.
1327 image::concepts.png[]
1329 Those attributes and objects are completely isolated between different
1332 A tracing session is analogous to a cash machine session:
1333 the operations you do on the banking system through the cash machine do
1334 not alter the data of other users of the same system. In the case of
1335 the cash machine, a session lasts as long as your bank card is inside.
1336 In the case of LTTng, a tracing session lasts from the `lttng create`
1337 command to the `lttng destroy` command.
1340 .Each Unix user has its own set of tracing sessions.
1341 image::many-sessions.png[]
1344 [[tracing-session-mode]]
1345 ==== Tracing session mode
1347 LTTng can send the generated trace data to different locations. The
1348 _tracing session mode_ dictates where to send it. The following modes
1349 are available in LTTng{nbsp}{revision}:
1352 LTTng writes the traces to the file system of the machine being traced
1355 Network streaming mode::
1356 LTTng sends the traces over the network to a
1357 <<lttng-relayd,relay daemon>> running on a remote system.
1360 LTTng does not write the traces by default. Instead, you can request
1361 LTTng to <<taking-a-snapshot,take a snapshot>>, that is, a copy of the
1362 current tracing buffers, and to write it to the target's file system
1363 or to send it over the network to a <<lttng-relayd,relay daemon>>
1364 running on a remote system.
1367 This mode is similar to the network streaming mode, but a live
1368 trace viewer can connect to the distant relay daemon to
1369 <<lttng-live,view event records as LTTng generates them>> by
1376 A _tracing domain_ is a namespace for event sources. A tracing domain
1377 has its own properties and features.
1379 There are currently five available tracing domains:
1383 * `java.util.logging` (JUL)
1387 You must specify a tracing domain when using some commands to avoid
1388 ambiguity. For example, since all the domains support named tracepoints
1389 as event sources (instrumentation points that you manually insert in the
1390 source code), you need to specify a tracing domain when
1391 <<enabling-disabling-events,creating an event rule>> because all the
1392 tracing domains could have tracepoints with the same names.
1394 Some features are reserved to specific tracing domains. Dynamic function
1395 entry and return instrumentation points, for example, are currently only
1396 supported in the Linux kernel tracing domain, but support for other
1397 tracing domains could be added in the future.
1399 You can create <<channel,channels>> in the Linux kernel and user space
1400 tracing domains. The other tracing domains have a single default
1405 === Channel and ring buffer
1407 A _channel_ is an object which is responsible for a set of ring buffers.
1408 Each ring buffer is divided into multiple sub-buffers. When an LTTng
1409 tracer emits an event, it can record it to one or more
1410 sub-buffers. The attributes of a channel determine what to do when
1411 there's no space left for a new event record because all sub-buffers
1412 are full, where to send a full sub-buffer, and other behaviours.
1414 A channel is always associated to a <<domain,tracing domain>>. The
1415 `java.util.logging` (JUL), log4j, and Python tracing domains each have
1416 a default channel which you cannot configure.
1418 A channel also owns <<event,event rules>>. When an LTTng tracer emits
1419 an event, it records it to the sub-buffers of all
1420 the enabled channels with a satisfied event rule, as long as those
1421 channels are part of active <<tracing-session,tracing sessions>>.
1424 [[channel-buffering-schemes]]
1425 ==== Per-user vs. per-process buffering schemes
1427 A channel has at least one ring buffer _per CPU_. LTTng always
1428 records an event to the ring buffer associated to the CPU on which it
1431 Two _buffering schemes_ are available when you
1432 <<enabling-disabling-channels,create a channel>> in the
1433 user space <<domain,tracing domain>>:
1435 Per-user buffering::
1436 Allocate one set of ring buffers--one per CPU--shared by all the
1437 instrumented processes of each Unix user.
1441 .Per-user buffering scheme.
1442 image::per-user-buffering.png[]
1445 Per-process buffering::
1446 Allocate one set of ring buffers--one per CPU--for each
1447 instrumented process.
1451 .Per-process buffering scheme.
1452 image::per-process-buffering.png[]
1455 The per-process buffering scheme tends to consume more memory than the
1456 per-user option because systems generally have more instrumented
1457 processes than Unix users running instrumented processes. However, the
1458 per-process buffering scheme ensures that one process having a high
1459 event throughput won't fill all the shared sub-buffers of the same
1462 The Linux kernel tracing domain has only one available buffering scheme
1463 which is to allocate a single set of ring buffers for the whole system.
1464 This scheme is similar to the per-user option, but with a single, global
1465 user "running" the kernel.
1468 [[channel-overwrite-mode-vs-discard-mode]]
1469 ==== Overwrite vs. discard event loss modes
1471 When an event occurs, LTTng records it to a specific sub-buffer (yellow
1472 arc in the following animation) of a specific channel's ring buffer.
1473 When there's no space left in a sub-buffer, the tracer marks it as
1474 consumable (red) and another, empty sub-buffer starts receiving the
1475 following event records. A <<lttng-consumerd,consumer daemon>>
1476 eventually consumes the marked sub-buffer (returns to white).
1479 [role="docsvg-channel-subbuf-anim"]
1484 In an ideal world, sub-buffers are consumed faster than they are filled,
1485 as is the case in the previous animation. In the real world,
1486 however, all sub-buffers can be full at some point, leaving no space to
1487 record the following events.
1489 By design, LTTng is a _non-blocking_ tracer: when no empty sub-buffer is
1490 available, it is acceptable to lose event records when the alternative
1491 would be to cause substantial delays in the instrumented application's
1492 execution. LTTng privileges performance over integrity; it aims at
1493 perturbing the traced system as little as possible in order to make
1494 tracing of subtle race conditions and rare interrupt cascades possible.
1496 When it comes to losing event records because no empty sub-buffer is
1497 available, the channel's _event loss mode_ determines what to do. The
1498 available event loss modes are:
1501 Drop the newest event records until a the tracer
1502 releases a sub-buffer.
1505 Clear the sub-buffer containing the oldest event records and start
1506 writing the newest event records there.
1508 This mode is sometimes called _flight recorder mode_ because it's
1510 https://en.wikipedia.org/wiki/Flight_recorder[flight recorder]:
1511 always keep a fixed amount of the latest data.
1513 Which mechanism you should choose depends on your context: prioritize
1514 the newest or the oldest event records in the ring buffer?
1516 Beware that, in overwrite mode, the tracer abandons a whole sub-buffer
1517 as soon as a there's no space left for a new event record, whereas in
1518 discard mode, the tracer only discards the event record that doesn't
1521 In discard mode, LTTng increments a count of lost event records when an
1522 event record is lost and saves this count to the trace. In overwrite
1523 mode, since LTTng 2.8, LTTng increments a count of lost sub-buffers when
1524 a sub-buffer is lost and saves this count to the trace. In this mode,
1525 the exact number of lost event records in those lost sub-buffers is not
1526 saved to the trace. Trace analyses can use the trace's saved discarded
1527 event record and sub-buffer counts to decide whether or not to perform
1528 the analyses even if trace data is known to be missing.
1530 There are a few ways to decrease your probability of losing event
1532 <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>> shows
1533 how you can fine-une the sub-buffer count and size of a channel to
1534 virtually stop losing event records, though at the cost of greater
1538 [[channel-subbuf-size-vs-subbuf-count]]
1539 ==== Sub-buffer count and size
1541 When you <<enabling-disabling-channels,create a channel>>, you can
1542 set its number of sub-buffers and their size.
1544 Note that there is noticeable CPU overhead introduced when
1545 switching sub-buffers (marking a full one as consumable and switching
1546 to an empty one for the following events to be recorded). Knowing this,
1547 the following list presents a few practical situations along with how
1548 to configure the sub-buffer count and size for them:
1550 * **High event throughput**: In general, prefer bigger sub-buffers to
1551 lower the risk of losing event records.
1553 Having bigger sub-buffers also ensures a lower
1554 <<channel-switch-timer,sub-buffer switching frequency>>.
1556 The number of sub-buffers is only meaningful if you create the channel
1557 in overwrite mode: in this case, if a sub-buffer overwrite happens, the
1558 other sub-buffers are left unaltered.
1560 * **Low event throughput**: In general, prefer smaller sub-buffers
1561 since the risk of losing event records is low.
1563 Because events occur less frequently, the sub-buffer switching frequency
1564 should remain low and thus the tracer's overhead should not be a
1567 * **Low memory system**: If your target system has a low memory
1568 limit, prefer fewer first, then smaller sub-buffers.
1570 Even if the system is limited in memory, you want to keep the
1571 sub-buffers as big as possible to avoid a high sub-buffer switching
1574 Note that LTTng uses http://diamon.org/ctf/[CTF] as its trace format,
1575 which means event data is very compact. For example, the average
1576 LTTng kernel event record weights about 32{nbsp}bytes. Thus, a
1577 sub-buffer size of 1{nbsp}MiB is considered big.
1579 The previous situations highlight the major trade-off between a few big
1580 sub-buffers and more, smaller sub-buffers: sub-buffer switching
1581 frequency vs. how much data is lost in overwrite mode. Assuming a
1582 constant event throughput and using the overwrite mode, the two
1583 following configurations have the same ring buffer total size:
1586 [role="docsvg-channel-subbuf-size-vs-count-anim"]
1591 * **2 sub-buffers of 4{nbsp}MiB each**: Expect a very low sub-buffer
1592 switching frequency, but if a sub-buffer overwrite happens, half of
1593 the event records so far (4{nbsp}MiB) are definitely lost.
1594 * **8 sub-buffers of 1{nbsp}MiB each**: Expect 4{nbsp}times the tracer's
1595 overhead as the previous configuration, but if a sub-buffer
1596 overwrite happens, only the eighth of event records so far are
1599 In discard mode, the sub-buffers count parameter is pointless: use two
1600 sub-buffers and set their size according to the requirements of your
1604 [[channel-switch-timer]]
1605 ==== Switch timer period
1607 The _switch timer period_ is an important configurable attribute of
1608 a channel to ensure periodic sub-buffer flushing.
1610 When the _switch timer_ expires, a sub-buffer switch happens. You can
1611 set the switch timer period attribute when you
1612 <<enabling-disabling-channels,create a channel>> to ensure that event
1613 data is consumed and committed to trace files or to a distant relay
1614 daemon periodically in case of a low event throughput.
1617 [role="docsvg-channel-switch-timer"]
1622 This attribute is also convenient when you use big sub-buffers to cope
1623 with a sporadic high event throughput, even if the throughput is
1627 [[channel-read-timer]]
1628 ==== Read timer period
1630 By default, the LTTng tracers use a notification mechanism to signal a
1631 full sub-buffer so that a consumer daemon can consume it. When such
1632 notifications must be avoided, for example in real-time applications,
1633 you can use the channel's _read timer_ instead. When the read timer
1634 fires, the <<lttng-consumerd,consumer daemon>> checks for full,
1635 consumable sub-buffers.
1638 [[tracefile-rotation]]
1639 ==== Trace file count and size
1641 By default, trace files can grow as large as needed. You can set the
1642 maximum size of each trace file that a channel writes when you
1643 <<enabling-disabling-channels,create a channel>>. When the size of
1644 a trace file reaches the channel's fixed maximum size, LTTng creates
1645 another file to contain the next event records. LTTng appends a file
1646 count to each trace file name in this case.
1648 If you set the trace file size attribute when you create a channel, the
1649 maximum number of trace files that LTTng creates is _unlimited_ by
1650 default. To limit them, you can also set a maximum number of trace
1651 files. When the number of trace files reaches the channel's fixed
1652 maximum count, the oldest trace file is overwritten. This mechanism is
1653 called _trace file rotation_.
1657 === Instrumentation point, event rule, event, and event record
1659 An _event rule_ is a set of conditions which must be **all** satisfied
1660 for LTTng to record an occuring event.
1662 You set the conditions when you <<enabling-disabling-events,create
1665 You always attach an event rule to <<channel,channel>> when you create
1668 When an event passes the conditions of an event rule, LTTng records it
1669 in one of the attached channel's sub-buffers.
1671 The available conditions, as of LTTng{nbsp}{revision}, are:
1673 * The event rule _is enabled_.
1674 * The instrumentation point's type _is{nbsp}T_.
1675 * The instrumentation point's name (sometimes called _event name_)
1676 _matches{nbsp}N_, but _is not{nbsp}E_.
1677 * The instrumentation point's log level _is as severe as{nbsp}L_, or
1678 _is exactly{nbsp}L_.
1679 * The fields of the event's payload _satisfy_ a filter
1680 expression{nbsp}__F__.
1682 As you can see, all the conditions but the dynamic filter are related to
1683 the event rule's status or to the instrumentation point, not to the
1684 occurring events. This is why, without a filter, checking if an event
1685 passes an event rule is not a dynamic task: when you create or modify an
1686 event rule, all the tracers of its tracing domain enable or disable the
1687 instrumentation points themselves once. This is possible because the
1688 attributes of an instrumentation point (type, name, and log level) are
1689 defined statically. In other words, without a dynamic filter, the tracer
1690 _does not evaluate_ the arguments of an instrumentation point unless it
1691 matches an enabled event rule.
1693 Note that, for LTTng to record an event, the <<channel,channel>> to
1694 which a matching event rule is attached must also be enabled, and the
1695 tracing session owning this channel must be active.
1698 .Logical path from an instrumentation point to an event record.
1699 image::event-rule.png[]
1701 .Event, event record, or event rule?
1703 With so many similar terms, it's easy to get confused.
1705 An **event** is the consequence of the execution of an _instrumentation
1706 point_, like a tracepoint that you manually place in some source code,
1707 or a Linux kernel KProbe. An event is said to _occur_ at a specific
1708 time. Different actions can be taken upon the occurrence of an event,
1709 like record the event's payload to a buffer.
1711 An **event record** is the representation of an event in a sub-buffer. A
1712 tracer is responsible for capturing the payload of an event, current
1713 context variables, the event's ID, and the event's timestamp. LTTng
1714 can append this sub-buffer to a trace file.
1716 An **event rule** is a set of conditions which must all be satisfied for
1717 LTTng to record an occuring event. Events still occur without
1718 satisfying event rules, but LTTng does not record them.
1723 == Components of noch:{LTTng}
1725 The second _T_ in _LTTng_ stands for _toolkit_: it would be wrong
1726 to call LTTng a simple _tool_ since it is composed of multiple
1727 interacting components. This section describes those components,
1728 explains their respective roles, and shows how they connect together to
1729 form the LTTng ecosystem.
1731 The following diagram shows how the most important components of LTTng
1732 interact with user applications, the Linux kernel, and you:
1735 .Control and trace data paths between LTTng components.
1736 image::plumbing.png[]
1738 The LTTng project incorporates:
1740 * **LTTng-tools**: Libraries and command-line interface to
1741 control tracing sessions.
1742 ** <<lttng-sessiond,Session daemon>> (man:lttng-sessiond(8)).
1743 ** <<lttng-consumerd,Consumer daemon>> (man:lttng-consumerd(8)).
1744 ** <<lttng-relayd,Relay daemon>> (man:lttng-relayd(8)).
1745 ** <<liblttng-ctl-lttng,Tracing control library>> (`liblttng-ctl`).
1746 ** <<lttng-cli,Tracing control command-line tool>> (man:lttng(1)).
1747 * **LTTng-UST**: Libraries and Java/Python packages to trace user
1749 ** <<lttng-ust,User space tracing library>> (`liblttng-ust`) and its
1750 headers to instrument and trace any native user application.
1751 ** <<prebuilt-ust-helpers,Preloadable user space tracing helpers>>:
1752 *** `liblttng-ust-libc-wrapper`
1753 *** `liblttng-ust-pthread-wrapper`
1754 *** `liblttng-ust-cyg-profile`
1755 *** `liblttng-ust-cyg-profile-fast`
1756 *** `liblttng-ust-dl`
1757 ** User space tracepoint provider source files generator command-line
1758 tool (man:lttng-gen-tp(1)).
1759 ** <<lttng-ust-agents,LTTng-UST Java agent>> to instrument and trace
1760 Java applications using `java.util.logging` or
1761 Apache log4j 1.2 logging.
1762 ** <<lttng-ust-agents,LTTng-UST Python agent>> to instrument
1763 Python applications using the standard `logging` package.
1764 * **LTTng-modules**: <<lttng-modules,Linux kernel modules>> to trace
1766 ** LTTng kernel tracer module.
1767 ** Tracing ring buffer kernel modules.
1768 ** Probe kernel modules.
1769 ** LTTng logger kernel module.
1773 === Tracing control command-line interface
1776 .The tracing control command-line interface.
1777 image::plumbing-lttng-cli.png[]
1779 The _man:lttng(1) command-line tool_ is the standard user interface to
1780 control LTTng <<tracing-session,tracing sessions>>. The cmd:lttng tool
1781 is part of LTTng-tools.
1783 The cmd:lttng tool is linked with
1784 <<liblttng-ctl-lttng,`liblttng-ctl`>> to communicate with
1785 one or more <<lttng-sessiond,session daemons>> behind the scenes.
1787 The cmd:lttng tool has a Git-like interface:
1791 $ lttng <GENERAL OPTIONS> <COMMAND> <COMMAND OPTIONS>
1794 The <<controlling-tracing,Tracing control>> section explores the
1795 available features of LTTng using the cmd:lttng tool.
1798 [[liblttng-ctl-lttng]]
1799 === Tracing control library
1802 .The tracing control library.
1803 image::plumbing-liblttng-ctl.png[]
1805 The _LTTng control library_, `liblttng-ctl`, is used to communicate
1806 with a <<lttng-sessiond,session daemon>> using a C API that hides the
1807 underlying protocol's details. `liblttng-ctl` is part of LTTng-tools.
1809 The <<lttng-cli,cmd:lttng command-line tool>>
1810 is linked with `liblttng-ctl`.
1812 You can use `liblttng-ctl` in C or $$C++$$ source code by including its
1817 #include <lttng/lttng.h>
1820 Some objects are referenced by name (C string), such as tracing
1821 sessions, but most of them require to create a handle first using
1822 `lttng_create_handle()`.
1824 The best available developer documentation for `liblttng-ctl` is, as of
1825 LTTng{nbsp}{revision}, its installed header files. Every function and
1826 structure is thoroughly documented.
1830 === User space tracing library
1833 .The user space tracing library.
1834 image::plumbing-liblttng-ust.png[]
1836 The _user space tracing library_, `liblttng-ust` (see man:lttng-ust(3)),
1837 is the LTTng user space tracer. It receives commands from a
1838 <<lttng-sessiond,session daemon>>, for example to
1839 enable and disable specific instrumentation points, and writes event
1840 records to ring buffers shared with a
1841 <<lttng-consumerd,consumer daemon>>.
1842 `liblttng-ust` is part of LTTng-UST.
1844 Public C header files are installed beside `liblttng-ust` to
1845 instrument any <<c-application,C or $$C++$$ application>>.
1847 <<lttng-ust-agents,LTTng-UST agents>>, which are regular Java and Python
1848 packages, use their own library providing tracepoints which is
1849 linked with `liblttng-ust`.
1851 An application or library does not have to initialize `liblttng-ust`
1852 manually: its constructor does the necessary tasks to properly register
1853 to a session daemon. The initialization phase also enables the
1854 instrumentation points matching the <<event,event rules>> that you
1858 [[lttng-ust-agents]]
1859 === User space tracing agents
1862 .The user space tracing agents.
1863 image::plumbing-lttng-ust-agents.png[]
1865 The _LTTng-UST Java and Python agents_ are regular Java and Python
1866 packages which add LTTng tracing capabilities to the
1867 native logging frameworks. The LTTng-UST agents are part of LTTng-UST.
1869 In the case of Java, the
1870 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[`java.util.logging`
1871 core logging facilities] and
1872 https://logging.apache.org/log4j/1.2/[Apache log4j 1.2] are supported.
1873 Note that Apache Log4{nbsp}2 is not supported.
1875 In the case of Python, the standard
1876 https://docs.python.org/3/library/logging.html[`logging`] package
1877 is supported. Both Python 2 and Python 3 modules can import the
1878 LTTng-UST Python agent package.
1880 The applications using the LTTng-UST agents are in the
1881 `java.util.logging` (JUL),
1882 log4j, and Python <<domain,tracing domains>>.
1884 Both agents use the same mechanism to trace the log statements. When an
1885 agent is initialized, it creates a log handler that attaches to the root
1886 logger. The agent also registers to a <<lttng-sessiond,session daemon>>.
1887 When the application executes a log statement, it is passed to the
1888 agent's log handler by the root logger. The agent's log handler calls a
1889 native function in a tracepoint provider package shared library linked
1890 with <<lttng-ust,`liblttng-ust`>>, passing the formatted log message and
1891 other fields, like its logger name and its log level. This native
1892 function contains a user space instrumentation point, hence tracing the
1895 The log level condition of an
1896 <<event,event rule>> is considered when tracing
1897 a Java or a Python application, and it's compatible with the standard
1898 JUL, log4j, and Python log levels.
1902 === LTTng kernel modules
1905 .The LTTng kernel modules.
1906 image::plumbing-lttng-modules.png[]
1908 The _LTTng kernel modules_ are a set of Linux kernel modules
1909 which implement the kernel tracer of the LTTng project. The LTTng
1910 kernel modules are part of LTTng-modules.
1912 The LTTng kernel modules include:
1914 * A set of _probe_ modules.
1916 Each module attaches to a specific subsystem
1917 of the Linux kernel using its tracepoint instrument points. There are
1918 also modules to attach to the entry and return points of the Linux
1919 system call functions.
1921 * _Ring buffer_ modules.
1923 A ring buffer implementation is provided as kernel modules. The LTTng
1924 kernel tracer writes to the ring buffer; a
1925 <<lttng-consumerd,consumer daemon>> reads from the ring buffer.
1927 * The _LTTng kernel tracer_ module.
1928 * The _LTTng logger_ module.
1930 The LTTng logger module implements the special path:{/proc/lttng-logger}
1931 file so that any executable can generate LTTng events by opening and
1932 writing to this file.
1934 See <<proc-lttng-logger-abi,LTTng logger>>.
1936 Generally, you do not have to load the LTTng kernel modules manually
1937 (using man:modprobe(8), for example): a root <<lttng-sessiond,session
1938 daemon>> loads the necessary modules when starting. If you have extra
1939 probe modules, you can specify to load them to the session daemon on
1942 The LTTng kernel modules are installed in
1943 +/usr/lib/modules/__release__/extra+ by default, where +__release__+ is
1944 the kernel release (see `uname --kernel-release`).
1951 .The session daemon.
1952 image::plumbing-sessiond.png[]
1954 The _session daemon_, man:lttng-sessiond(8), is a daemon responsible for
1955 managing tracing sessions and for controlling the various components of
1956 LTTng. The session daemon is part of LTTng-tools.
1958 The session daemon sends control requests to and receives control
1961 * The <<lttng-ust,user space tracing library>>.
1963 Any instance of the user space tracing library first registers to
1964 a session daemon. Then, the session daemon can send requests to
1965 this instance, such as:
1968 ** Get the list of tracepoints.
1969 ** Share an <<event,event rule>> so that the user space tracing library
1970 can enable or disable tracepoints. Amongst the possible conditions
1971 of an event rule is a filter expression which `liblttng-ust` evalutes
1972 when an event occurs.
1973 ** Share <<channel,channel>> attributes and ring buffer locations.
1976 The session daemon and the user space tracing library use a Unix
1977 domain socket for their communication.
1979 * The <<lttng-ust-agents,user space tracing agents>>.
1981 Any instance of a user space tracing agent first registers to
1982 a session daemon. Then, the session daemon can send requests to
1983 this instance, such as:
1986 ** Get the list of loggers.
1987 ** Enable or disable a specific logger.
1990 The session daemon and the user space tracing agent use a TCP connection
1991 for their communication.
1993 * The <<lttng-modules,LTTng kernel tracer>>.
1994 * The <<lttng-consumerd,consumer daemon>>.
1996 The session daemon sends requests to the consumer daemon to instruct
1997 it where to send the trace data streams, amongst other information.
1999 * The <<lttng-relayd,relay daemon>>.
2001 The session daemon receives commands from the
2002 <<liblttng-ctl-lttng,tracing control library>>.
2004 The root session daemon loads the appropriate
2005 <<lttng-modules,LTTng kernel modules>> on startup. It also spawns
2006 a <<lttng-consumerd,consumer daemon>> as soon as you create
2007 an <<event,event rule>>.
2009 The session daemon does not send and receive trace data: this is the
2010 role of the <<lttng-consumerd,consumer daemon>> and
2011 <<lttng-relayd,relay daemon>>. It does, however, generate the
2012 http://diamon.org/ctf/[CTF] metadata stream.
2014 Each Unix user can have its own session daemon instance. The
2015 tracing sessions managed by different session daemons are completely
2018 The root user's session daemon is the only one which is
2019 allowed to control the LTTng kernel tracer, and its spawned consumer
2020 daemon is the only one which is allowed to consume trace data from the
2021 LTTng kernel tracer. Note, however, that any Unix user which is a member
2022 of the <<tracing-group,tracing group>> is allowed
2023 to create <<channel,channels>> in the
2024 Linux kernel <<domain,tracing domain>>, and thus to trace the Linux
2027 The <<lttng-cli,cmd:lttng command-line tool>> automatically starts a
2028 session daemon when using its `create` command if none is currently
2029 running. You can also start the session daemon manually.
2036 .The consumer daemon.
2037 image::plumbing-consumerd.png[]
2039 The _consumer daemon_, man:lttng-consumerd(8), is a daemon which shares
2040 ring buffers with user applications or with the LTTng kernel modules to
2041 collect trace data and send it to some location (on disk or to a
2042 <<lttng-relayd,relay daemon>> over the network). The consumer daemon
2043 is part of LTTng-tools.
2045 You do not start a consumer daemon manually: a consumer daemon is always
2046 spawned by a <<lttng-sessiond,session daemon>> as soon as you create an
2047 <<event,event rule>>, that is, before you start tracing. When you kill
2048 its owner session daemon, the consumer daemon also exits because it is
2049 the session daemon's child process. Command-line options of
2050 man:lttng-sessiond(8) target the consumer daemon process.
2052 There are up to two running consumer daemons per Unix user, whereas only
2053 one session daemon can run per user. This is because each process can be
2054 either 32-bit or 64-bit: if the target system runs a mixture of 32-bit
2055 and 64-bit processes, it is more efficient to have separate
2056 corresponding 32-bit and 64-bit consumer daemons. The root user is an
2057 exception: it can have up to _three_ running consumer daemons: 32-bit
2058 and 64-bit instances for its user applications, and one more
2059 reserved for collecting kernel trace data.
2067 image::plumbing-relayd.png[]
2069 The _relay daemon_, man:lttng-relayd(8), is a daemon acting as a bridge
2070 between remote session and consumer daemons, local trace files, and a
2071 remote live trace viewer. The relay daemon is part of LTTng-tools.
2073 The main purpose of the relay daemon is to implement a receiver of
2074 <<sending-trace-data-over-the-network,trace data over the network>>.
2075 This is useful when the target system does not have much file system
2076 space to record trace files locally.
2078 The relay daemon is also a server to which a
2079 <<lttng-live,live trace viewer>> can
2080 connect. The live trace viewer sends requests to the relay daemon to
2081 receive trace data as the target system emits events. The
2082 communication protocol is named _LTTng live_; it is used over TCP
2085 Note that you can start the relay daemon on the target system directly.
2086 This is the setup of choice when the use case is to view events as
2087 the target system emits them without the need of a remote system.
2091 == [[using-lttng]]Instrumentation
2093 There are many examples of tracing and monitoring in our everyday life:
2095 * You have access to real-time and historical weather reports and
2096 forecasts thanks to weather stations installed around the country.
2097 * You know your heart is safe thanks to an electrocardiogram.
2098 * You make sure not to drive your car too fast and to have enough fuel
2099 to reach your destination thanks to gauges visible on your dashboard.
2101 All the previous examples have something in common: they rely on
2102 **instruments**. Without the electrodes attached to the surface of your
2103 body's skin, cardiac monitoring is futile.
2105 LTTng, as a tracer, is no different from those real life examples. If
2106 you're about to trace a software system or, in other words, record its
2107 history of execution, you better have **instrumentation points** in the
2108 subject you're tracing, that is, the actual software.
2110 Various ways were developed to instrument a piece of software for LTTng
2111 tracing. The most straightforward one is to manually place
2112 instrumentation points, called _tracepoints_, in the software's source
2113 code. It is also possible to add instrumentation points dynamically in
2114 the Linux kernel <<domain,tracing domain>>.
2116 If you're only interested in tracing the Linux kernel, your
2117 instrumentation needs are probably already covered by LTTng's built-in
2118 <<lttng-modules,Linux kernel tracepoints>>. You may also wish to trace a
2119 user application which is already instrumented for LTTng tracing.
2120 In such cases, you can skip this whole section and read the topics of
2121 the <<controlling-tracing,Tracing control>> section.
2123 Many methods are available to instrument a piece of software for LTTng
2126 * <<c-application,User space instrumentation for C and $$C++$$
2128 * <<prebuilt-ust-helpers,Prebuilt user space tracing helpers>>.
2129 * <<java-application,User space Java agent>>.
2130 * <<python-application,User space Python agent>>.
2131 * <<proc-lttng-logger-abi,LTTng logger>>.
2132 * <<instrumenting-linux-kernel,LTTng kernel tracepoints>>.
2136 === [[cxx-application]]User space instrumentation for C and $$C++$$ applications
2138 The procedure to instrument a C or $$C++$$ user application with
2139 the <<lttng-ust,LTTng user space tracing library>>, `liblttng-ust`, is:
2141 . <<tracepoint-provider,Create the source files of a tracepoint provider
2143 . <<probing-the-application-source-code,Add tracepoints to
2144 the application's source code>>.
2145 . <<building-tracepoint-providers-and-user-application,Build and link
2146 a tracepoint provider package and the user application>>.
2148 If you need quick, man:printf(3)-like instrumentation, you can skip
2149 those steps and use <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>>
2152 IMPORTANT: You need to <<installing-lttng,install>> LTTng-UST to
2153 instrument a user application with `liblttng-ust`.
2156 [[tracepoint-provider]]
2157 ==== Create the source files of a tracepoint provider package
2159 A _tracepoint provider_ is a set of compiled functions which provide
2160 **tracepoints** to an application, the type of instrumentation point
2161 supported by LTTng-UST. Those functions can emit events with
2162 user-defined fields and serialize those events as event records to one
2163 or more LTTng-UST <<channel,channel>> sub-buffers. The `tracepoint()`
2164 macro, which you <<probing-the-application-source-code,insert in a user
2165 application's source code>>, calls those functions.
2167 A _tracepoint provider package_ is an object file (`.o`) or a shared
2168 library (`.so`) which contains one or more tracepoint providers.
2169 Its source files are:
2171 * One or more <<tpp-header,tracepoint provider header>> (`.h`).
2172 * A <<tpp-source,tracepoint provider package source>> (`.c`).
2174 A tracepoint provider package is dynamically linked with `liblttng-ust`,
2175 the LTTng user space tracer, at run time.
2178 .User application linked with `liblttng-ust` and containing a tracepoint provider.
2179 image::ust-app.png[]
2181 NOTE: If you need quick, man:printf(3)-like instrumentation, you can
2182 skip creating and using a tracepoint provider and use
2183 <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>> instead.
2187 ===== Create a tracepoint provider header file template
2189 A _tracepoint provider header file_ contains the tracepoint
2190 definitions of a tracepoint provider.
2192 To create a tracepoint provider header file:
2194 . Start from this template:
2198 .Tracepoint provider header file template (`.h` file extension).
2200 #undef TRACEPOINT_PROVIDER
2201 #define TRACEPOINT_PROVIDER provider_name
2203 #undef TRACEPOINT_INCLUDE
2204 #define TRACEPOINT_INCLUDE "./tp.h"
2206 #if !defined(_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
2209 #include <lttng/tracepoint.h>
2212 * Use TRACEPOINT_EVENT(), TRACEPOINT_EVENT_CLASS(),
2213 * TRACEPOINT_EVENT_INSTANCE(), and TRACEPOINT_LOGLEVEL() here.
2218 #include <lttng/tracepoint-event.h>
2224 * `provider_name` with the name of your tracepoint provider.
2225 * `"tp.h"` with the name of your tracepoint provider header file.
2227 . Below the `#include <lttng/tracepoint.h>` line, put your
2228 <<defining-tracepoints,tracepoint definitions>>.
2230 Your tracepoint provider name must be unique amongst all the possible
2231 tracepoint provider names used on the same target system. We
2232 suggest to include the name of your project or company in the name,
2233 for example, `org_lttng_my_project_tpp`.
2235 TIP: [[lttng-gen-tp]]You can use the man:lttng-gen-tp(1) tool to create
2236 this boilerplate for you. When using cmd:lttng-gen-tp, all you need to
2237 write are the <<defining-tracepoints,tracepoint definitions>>.
2240 [[defining-tracepoints]]
2241 ===== Create a tracepoint definition
2243 A _tracepoint definition_ defines, for a given tracepoint:
2245 * Its **input arguments**. They are the macro parameters that the
2246 `tracepoint()` macro accepts for this particular tracepoint
2247 in the user application's source code.
2248 * Its **output event fields**. They are the sources of event fields
2249 that form the payload of any event that the execution of the
2250 `tracepoint()` macro emits for this particular tracepoint.
2252 You can create a tracepoint definition by using the
2253 `TRACEPOINT_EVENT()` macro below the `#include <lttng/tracepoint.h>`
2255 <<tpp-header,tracepoint provider header file template>>.
2257 The syntax of the `TRACEPOINT_EVENT()` macro is:
2260 .`TRACEPOINT_EVENT()` macro syntax.
2263 /* Tracepoint provider name */
2266 /* Tracepoint name */
2269 /* Input arguments */
2274 /* Output event fields */
2283 * `provider_name` with your tracepoint provider name.
2284 * `tracepoint_name` with your tracepoint name.
2285 * `arguments` with the <<tpp-def-input-args,input arguments>>.
2286 * `fields` with the <<tpp-def-output-fields,output event field>>
2289 This tracepoint emits events named `provider_name:tracepoint_name`.
2292 .Event name's length limitation
2294 The concatenation of the tracepoint provider name and the
2295 tracepoint name must not exceed **254 characters**. If it does, the
2296 instrumented application compiles and runs, but LTTng throws multiple
2297 warnings and you could experience serious issues.
2300 [[tpp-def-input-args]]The syntax of the `TP_ARGS()` macro is:
2303 .`TP_ARGS()` macro syntax.
2312 * `type` with the C type of the argument.
2313 * `arg_name` with the argument name.
2315 You can repeat `type` and `arg_name` up to 10 times to have
2316 more than one argument.
2318 .`TP_ARGS()` usage with three arguments.
2330 The `TP_ARGS()` and `TP_ARGS(void)` forms are valid to create a
2331 tracepoint definition with no input arguments.
2333 [[tpp-def-output-fields]]The `TP_FIELDS()` macro contains a list of
2334 `ctf_*()` macros. Each `ctf_*()` macro defines one event field. See
2335 man:lttng-ust(3) for a complete description of the available `ctf_*()`
2336 macros. A `ctf_*()` macro specifies the type, size, and byte order of
2339 Each `ctf_*()` macro takes an _argument expression_ parameter. This is a
2340 C expression that the tracer evalutes at the `tracepoint()` macro site
2341 in the application's source code. This expression provides a field's
2342 source of data. The argument expression can include input argument names
2343 listed in the `TP_ARGS()` macro.
2345 Each `ctf_*()` macro also takes a _field name_ parameter. Field names
2346 must be unique within a given tracepoint definition.
2348 Here's a complete tracepoint definition example:
2350 .Tracepoint definition.
2352 The following tracepoint definition defines a tracepoint which takes
2353 three input arguments and has four output event fields.
2357 #include "my-custom-structure.h"
2363 const struct my_custom_structure*, my_custom_structure,
2368 ctf_string(query_field, query)
2369 ctf_float(double, ratio_field, ratio)
2370 ctf_integer(int, recv_size, my_custom_structure->recv_size)
2371 ctf_integer(int, send_size, my_custom_structure->send_size)
2376 You can refer to this tracepoint definition with the `tracepoint()`
2377 macro in your application's source code like this:
2381 tracepoint(my_provider, my_tracepoint,
2382 my_structure, some_ratio, the_query);
2386 NOTE: The LTTng tracer only evaluates tracepoint arguments at run time
2387 if they satisfy an enabled <<event,event rule>>.
2390 [[using-tracepoint-classes]]
2391 ===== Use a tracepoint class
2393 A _tracepoint class_ is a class of tracepoints which share the same
2394 output event field definitions. A _tracepoint instance_ is one
2395 instance of such a defined tracepoint class, with its own tracepoint
2398 The <<defining-tracepoints,`TRACEPOINT_EVENT()` macro>> is actually a
2399 shorthand which defines both a tracepoint class and a tracepoint
2400 instance at the same time.
2402 When you build a tracepoint provider package, the C or $$C++$$ compiler
2403 creates one serialization function for each **tracepoint class**. A
2404 serialization function is responsible for serializing the event fields
2405 of a tracepoint to a sub-buffer when tracing.
2407 For various performance reasons, when your situation requires multiple
2408 tracepoint definitions with different names, but with the same event
2409 fields, we recommend that you manually create a tracepoint class
2410 and instantiate as many tracepoint instances as needed. One positive
2411 effect of such a design, amongst other advantages, is that all
2412 tracepoint instances of the same tracepoint class reuse the same
2413 serialization function, thus reducing
2414 https://en.wikipedia.org/wiki/Cache_pollution[cache pollution].
2416 .Use a tracepoint class and tracepoint instances.
2418 Consider the following three tracepoint definitions:
2430 ctf_integer(int, userid, userid)
2431 ctf_integer(size_t, len, len)
2443 ctf_integer(int, userid, userid)
2444 ctf_integer(size_t, len, len)
2456 ctf_integer(int, userid, userid)
2457 ctf_integer(size_t, len, len)
2462 In this case, we create three tracepoint classes, with one implicit
2463 tracepoint instance for each of them: `get_account`, `get_settings`, and
2464 `get_transaction`. However, they all share the same event field names
2465 and types. Hence three identical, yet independent serialization
2466 functions are created when you build the tracepoint provider package.
2468 A better design choice is to define a single tracepoint class and three
2469 tracepoint instances:
2473 /* The tracepoint class */
2474 TRACEPOINT_EVENT_CLASS(
2475 /* Tracepoint provider name */
2478 /* Tracepoint class name */
2481 /* Input arguments */
2487 /* Output event fields */
2489 ctf_integer(int, userid, userid)
2490 ctf_integer(size_t, len, len)
2494 /* The tracepoint instances */
2495 TRACEPOINT_EVENT_INSTANCE(
2496 /* Tracepoint provider name */
2499 /* Tracepoint class name */
2502 /* Tracepoint name */
2505 /* Input arguments */
2511 TRACEPOINT_EVENT_INSTANCE(
2520 TRACEPOINT_EVENT_INSTANCE(
2533 [[assigning-log-levels]]
2534 ===== Assign a log level to a tracepoint definition
2536 You can assign an optional _log level_ to a
2537 <<defining-tracepoints,tracepoint definition>>.
2539 Assigning different levels of severity to tracepoint definitions can
2540 be useful: when you <<enabling-disabling-events,create an event rule>>,
2541 you can target tracepoints having a log level as severe as a specific
2544 The concept of LTTng-UST log levels is similar to the levels found
2545 in typical logging frameworks:
2547 * In a logging framework, the log level is given by the function
2548 or method name you use at the log statement site: `debug()`,
2549 `info()`, `warn()`, `error()`, and so on.
2550 * In LTTng-UST, you statically assign the log level to a tracepoint
2551 definition; any `tracepoint()` macro invocation which refers to
2552 this definition has this log level.
2554 You can assign a log level to a tracepoint definition with the
2555 `TRACEPOINT_LOGLEVEL()` macro. You must use this macro _after_ the
2556 <<defining-tracepoints,`TRACEPOINT_EVENT()`>> or
2557 <<using-tracepoint-classes,`TRACEPOINT_INSTANCE()`>> macro for a given
2560 The syntax of the `TRACEPOINT_LOGLEVEL()` macro is:
2563 .`TRACEPOINT_LOGLEVEL()` macro syntax.
2565 TRACEPOINT_LOGLEVEL(provider_name, tracepoint_name, log_level)
2570 * `provider_name` with the tracepoint provider name.
2571 * `tracepoint_name` with the tracepoint name.
2572 * `log_level` with the log level to assign to the tracepoint
2573 definition named `tracepoint_name` in the `provider_name`
2574 tracepoint provider.
2576 See man:lttng-ust(3) for a list of available log level names.
2578 .Assign the `TRACE_DEBUG_UNIT` log level to a tracepoint definition.
2582 /* Tracepoint definition */
2591 ctf_integer(int, userid, userid)
2592 ctf_integer(size_t, len, len)
2596 /* Log level assignment */
2597 TRACEPOINT_LOGLEVEL(my_app, get_transaction, TRACE_DEBUG_UNIT)
2603 ===== Create a tracepoint provider package source file
2605 A _tracepoint provider package source file_ is a C source file which
2606 includes a <<tpp-header,tracepoint provider header file>> to expand its
2607 macros into event serialization and other functions.
2609 You can always use the following tracepoint provider package source
2613 .Tracepoint provider package source file template.
2615 #define TRACEPOINT_CREATE_PROBES
2620 Replace `tp.h` with the name of your <<tpp-header,tracepoint provider
2621 header file>> name. You may also include more than one tracepoint
2622 provider header file here to create a tracepoint provider package
2623 holding more than one tracepoint providers.
2626 [[probing-the-application-source-code]]
2627 ==== Add tracepoints to an application's source code
2629 Once you <<tpp-header,create a tracepoint provider header file>>, you
2630 can use the `tracepoint()` macro in your application's
2631 source code to insert the tracepoints that this header
2632 <<defining-tracepoints,defines>>.
2634 The `tracepoint()` macro takes at least two parameters: the tracepoint
2635 provider name and the tracepoint name. The corresponding tracepoint
2636 definition defines the other parameters.
2638 .`tracepoint()` usage.
2640 The following <<defining-tracepoints,tracepoint definition>> defines a
2641 tracepoint which takes two input arguments and has two output event
2645 .Tracepoint provider header file.
2647 #include "my-custom-structure.h"
2654 const char*, cmd_name
2657 ctf_string(cmd_name, cmd_name)
2658 ctf_integer(int, number_of_args, argc)
2663 You can refer to this tracepoint definition with the `tracepoint()`
2664 macro in your application's source code like this:
2667 .Application's source file.
2671 int main(int argc, char* argv[])
2673 tracepoint(my_provider, my_tracepoint, argc, argv[0]);
2679 Note how the application's source code includes
2680 the tracepoint provider header file containing the tracepoint
2681 definitions to use, path:{tp.h}.
2684 .`tracepoint()` usage with a complex tracepoint definition.
2686 Consider this complex tracepoint definition, where multiple event
2687 fields refer to the same input arguments in their argument expression
2691 .Tracepoint provider header file.
2693 /* For `struct stat` */
2694 #include <sys/types.h>
2695 #include <sys/stat.h>
2707 ctf_integer(int, my_constant_field, 23 + 17)
2708 ctf_integer(int, my_int_arg_field, my_int_arg)
2709 ctf_integer(int, my_int_arg_field2, my_int_arg * my_int_arg)
2710 ctf_integer(int, sum4_field, my_str_arg[0] + my_str_arg[1] +
2711 my_str_arg[2] + my_str_arg[3])
2712 ctf_string(my_str_arg_field, my_str_arg)
2713 ctf_integer_hex(off_t, size_field, st->st_size)
2714 ctf_float(double, size_dbl_field, (double) st->st_size)
2715 ctf_sequence_text(char, half_my_str_arg_field, my_str_arg,
2716 size_t, strlen(my_str_arg) / 2)
2721 You can refer to this tracepoint definition with the `tracepoint()`
2722 macro in your application's source code like this:
2725 .Application's source file.
2727 #define TRACEPOINT_DEFINE
2734 stat("/etc/fstab", &s);
2735 tracepoint(my_provider, my_tracepoint, 23, "Hello, World!", &s);
2741 If you look at the event record that LTTng writes when tracing this
2742 program, assuming the file size of path:{/etc/fstab} is 301{nbsp}bytes,
2743 it should look like this:
2745 .Event record fields
2747 |Field's name |Field's value
2748 |`my_constant_field` |40
2749 |`my_int_arg_field` |23
2750 |`my_int_arg_field2` |529
2752 |`my_str_arg_field` |`Hello, World!`
2753 |`size_field` |0x12d
2754 |`size_dbl_field` |301.0
2755 |`half_my_str_arg_field` |`Hello,`
2759 Sometimes, the arguments you pass to `tracepoint()` are expensive to
2760 compute--they use the call stack, for example. To avoid this
2761 computation when the tracepoint is disabled, you can use the
2762 `tracepoint_enabled()` and `do_tracepoint()` macros.
2764 The syntax of the `tracepoint_enabled()` and `do_tracepoint()` macros
2768 .`tracepoint_enabled()` and `do_tracepoint()` macros syntax.
2770 tracepoint_enabled(provider_name, tracepoint_name)
2771 do_tracepoint(provider_name, tracepoint_name, ...)
2776 * `provider_name` with the tracepoint provider name.
2777 * `tracepoint_name` with the tracepoint name.
2779 `tracepoint_enabled()` returns a non-zero value if the tracepoint named
2780 `tracepoint_name` from the provider named `provider_name` is enabled
2783 `do_tracepoint()` is like `tracepoint()`, except that it doesn't check
2784 if the tracepoint is enabled. Using `tracepoint()` with
2785 `tracepoint_enabled()` is dangerous since `tracepoint()` also contains
2786 the `tracepoint_enabled()` check, thus a race condition is
2787 possible in this situation:
2790 .Possible race condition when using `tracepoint_enabled()` with `tracepoint()`.
2792 if (tracepoint_enabled(my_provider, my_tracepoint)) {
2793 stuff = prepare_stuff();
2796 tracepoint(my_provider, my_tracepoint, stuff);
2799 If the tracepoint is enabled after the condition, then `stuff` is not
2800 prepared: the emitted event will either contain wrong data, or the whole
2801 application could crash (segmentation fault, for example).
2803 NOTE: Neither `tracepoint_enabled()` nor `do_tracepoint()` have an
2804 `STAP_PROBEV()` call. If you need it, you must emit
2808 [[building-tracepoint-providers-and-user-application]]
2809 ==== Build and link a tracepoint provider package and an application
2811 Once you have one or more <<tpp-header,tracepoint provider header
2812 files>> and a <<tpp-source,tracepoint provider package source file>>,
2813 you can create the tracepoint provider package by compiling its source
2814 file. From here, multiple build and run scenarios are possible. The
2815 following table shows common application and library configurations
2816 along with the required command lines to achieve them.
2818 In the following diagrams, we use the following file names:
2821 Executable application.
2824 Application's object file.
2827 Tracepoint provider package object file.
2830 Tracepoint provider package archive file.
2833 Tracepoint provider package shared object file.
2836 User library object file.
2839 User library shared object file.
2841 We use the following symbols in the diagrams of table below:
2844 .Symbols used in the build scenario diagrams.
2845 image::ust-sit-symbols.png[]
2847 We assume that path:{.} is part of the env:LD_LIBRARY_PATH environment
2848 variable in the following instructions.
2850 [role="growable ust-scenarios",cols="asciidoc,asciidoc"]
2851 .Common tracepoint provider package scenarios.
2853 |Scenario |Instructions
2856 The instrumented application is statically linked with
2857 the tracepoint provider package object.
2859 image::ust-sit+app-linked-with-tp-o+app-instrumented.png[]
2862 include::../common/ust-sit-step-tp-o.txt[]
2864 To build the instrumented application:
2866 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2871 #define TRACEPOINT_DEFINE
2875 . Compile the application source file:
2884 . Build the application:
2889 $ gcc -o app app.o tpp.o -llttng-ust -ldl
2893 To run the instrumented application:
2895 * Start the application:
2905 The instrumented application is statically linked with the
2906 tracepoint provider package archive file.
2908 image::ust-sit+app-linked-with-tp-a+app-instrumented.png[]
2911 To create the tracepoint provider package archive file:
2913 . Compile the <<tpp-source,tracepoint provider package source file>>:
2922 . Create the tracepoint provider package archive file:
2927 $ ar rcs tpp.a tpp.o
2931 To build the instrumented application:
2933 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2938 #define TRACEPOINT_DEFINE
2942 . Compile the application source file:
2951 . Build the application:
2956 $ gcc -o app app.o tpp.a -llttng-ust -ldl
2960 To run the instrumented application:
2962 * Start the application:
2972 The instrumented application is linked with the tracepoint provider
2973 package shared object.
2975 image::ust-sit+app-linked-with-tp-so+app-instrumented.png[]
2978 include::../common/ust-sit-step-tp-so.txt[]
2980 To build the instrumented application:
2982 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2987 #define TRACEPOINT_DEFINE
2991 . Compile the application source file:
3000 . Build the application:
3005 $ gcc -o app app.o -ldl -L. -ltpp
3009 To run the instrumented application:
3011 * Start the application:
3021 The tracepoint provider package shared object is preloaded before the
3022 instrumented application starts.
3024 image::ust-sit+tp-so-preloaded+app-instrumented.png[]
3027 include::../common/ust-sit-step-tp-so.txt[]
3029 To build the instrumented application:
3031 . In path:{app.c}, before including path:{tpp.h}, add the
3037 #define TRACEPOINT_DEFINE
3038 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3042 . Compile the application source file:
3051 . Build the application:
3056 $ gcc -o app app.o -ldl
3060 To run the instrumented application with tracing support:
3062 * Preload the tracepoint provider package shared object and
3063 start the application:
3068 $ LD_PRELOAD=./libtpp.so ./app
3072 To run the instrumented application without tracing support:
3074 * Start the application:
3084 The instrumented application dynamically loads the tracepoint provider
3085 package shared object.
3087 See the <<dlclose-warning,warning about `dlclose()`>>.
3089 image::ust-sit+app-dlopens-tp-so+app-instrumented.png[]
3092 include::../common/ust-sit-step-tp-so.txt[]
3094 To build the instrumented application:
3096 . In path:{app.c}, before including path:{tpp.h}, add the
3102 #define TRACEPOINT_DEFINE
3103 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3107 . Compile the application source file:
3116 . Build the application:
3121 $ gcc -o app app.o -ldl
3125 To run the instrumented application:
3127 * Start the application:
3137 The application is linked with the instrumented user library.
3139 The instrumented user library is statically linked with the tracepoint
3140 provider package object file.
3142 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-o+lib-instrumented.png[]
3145 include::../common/ust-sit-step-tp-o-fpic.txt[]
3147 To build the instrumented user library:
3149 . In path:{emon.c}, before including path:{tpp.h}, add the
3155 #define TRACEPOINT_DEFINE
3159 . Compile the user library source file:
3164 $ gcc -I. -fpic -c emon.c
3168 . Build the user library shared object:
3173 $ gcc -shared -o libemon.so emon.o tpp.o -llttng-ust -ldl
3177 To build the application:
3179 . Compile the application source file:
3188 . Build the application:
3193 $ gcc -o app app.o -L. -lemon
3197 To run the application:
3199 * Start the application:
3209 The application is linked with the instrumented user library.
3211 The instrumented user library is linked with the tracepoint provider
3212 package shared object.
3214 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3217 include::../common/ust-sit-step-tp-so.txt[]
3219 To build the instrumented user library:
3221 . In path:{emon.c}, before including path:{tpp.h}, add the
3227 #define TRACEPOINT_DEFINE
3231 . Compile the user library source file:
3236 $ gcc -I. -fpic -c emon.c
3240 . Build the user library shared object:
3245 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3249 To build the application:
3251 . Compile the application source file:
3260 . Build the application:
3265 $ gcc -o app app.o -L. -lemon
3269 To run the application:
3271 * Start the application:
3281 The tracepoint provider package shared object is preloaded before the
3284 The application is linked with the instrumented user library.
3286 image::ust-sit+tp-so-preloaded+app-linked-with-lib+lib-instrumented.png[]
3289 include::../common/ust-sit-step-tp-so.txt[]
3291 To build the instrumented user library:
3293 . In path:{emon.c}, before including path:{tpp.h}, add the
3299 #define TRACEPOINT_DEFINE
3300 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3304 . Compile the user library source file:
3309 $ gcc -I. -fpic -c emon.c
3313 . Build the user library shared object:
3318 $ gcc -shared -o libemon.so emon.o -ldl
3322 To build the application:
3324 . Compile the application source file:
3333 . Build the application:
3338 $ gcc -o app app.o -L. -lemon
3342 To run the application with tracing support:
3344 * Preload the tracepoint provider package shared object and
3345 start the application:
3350 $ LD_PRELOAD=./libtpp.so ./app
3354 To run the application without tracing support:
3356 * Start the application:
3366 The application is linked with the instrumented user library.
3368 The instrumented user library dynamically loads the tracepoint provider
3369 package shared object.
3371 See the <<dlclose-warning,warning about `dlclose()`>>.
3373 image::ust-sit+app-linked-with-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3376 include::../common/ust-sit-step-tp-so.txt[]
3378 To build the instrumented user library:
3380 . In path:{emon.c}, before including path:{tpp.h}, add the
3386 #define TRACEPOINT_DEFINE
3387 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3391 . Compile the user library source file:
3396 $ gcc -I. -fpic -c emon.c
3400 . Build the user library shared object:
3405 $ gcc -shared -o libemon.so emon.o -ldl
3409 To build the application:
3411 . Compile the application source file:
3420 . Build the application:
3425 $ gcc -o app app.o -L. -lemon
3429 To run the application:
3431 * Start the application:
3441 The application dynamically loads the instrumented user library.
3443 The instrumented user library is linked with the tracepoint provider
3444 package shared object.
3446 See the <<dlclose-warning,warning about `dlclose()`>>.
3448 image::ust-sit+app-dlopens-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3451 include::../common/ust-sit-step-tp-so.txt[]
3453 To build the instrumented user library:
3455 . In path:{emon.c}, before including path:{tpp.h}, add the
3461 #define TRACEPOINT_DEFINE
3465 . Compile the user library source file:
3470 $ gcc -I. -fpic -c emon.c
3474 . Build the user library shared object:
3479 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3483 To build the application:
3485 . Compile the application source file:
3494 . Build the application:
3499 $ gcc -o app app.o -ldl -L. -lemon
3503 To run the application:
3505 * Start the application:
3515 The application dynamically loads the instrumented user library.
3517 The instrumented user library dynamically loads the tracepoint provider
3518 package shared object.
3520 See the <<dlclose-warning,warning about `dlclose()`>>.
3522 image::ust-sit+app-dlopens-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3525 include::../common/ust-sit-step-tp-so.txt[]
3527 To build the instrumented user library:
3529 . In path:{emon.c}, before including path:{tpp.h}, add the
3535 #define TRACEPOINT_DEFINE
3536 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3540 . Compile the user library source file:
3545 $ gcc -I. -fpic -c emon.c
3549 . Build the user library shared object:
3554 $ gcc -shared -o libemon.so emon.o -ldl
3558 To build the application:
3560 . Compile the application source file:
3569 . Build the application:
3574 $ gcc -o app app.o -ldl -L. -lemon
3578 To run the application:
3580 * Start the application:
3590 The tracepoint provider package shared object is preloaded before the
3593 The application dynamically loads the instrumented user library.
3595 image::ust-sit+tp-so-preloaded+app-dlopens-lib+lib-instrumented.png[]
3598 include::../common/ust-sit-step-tp-so.txt[]
3600 To build the instrumented user library:
3602 . In path:{emon.c}, before including path:{tpp.h}, add the
3608 #define TRACEPOINT_DEFINE
3609 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3613 . Compile the user library source file:
3618 $ gcc -I. -fpic -c emon.c
3622 . Build the user library shared object:
3627 $ gcc -shared -o libemon.so emon.o -ldl
3631 To build the application:
3633 . Compile the application source file:
3642 . Build the application:
3647 $ gcc -o app app.o -L. -lemon
3651 To run the application with tracing support:
3653 * Preload the tracepoint provider package shared object and
3654 start the application:
3659 $ LD_PRELOAD=./libtpp.so ./app
3663 To run the application without tracing support:
3665 * Start the application:
3675 The application is statically linked with the tracepoint provider
3676 package object file.
3678 The application is linked with the instrumented user library.
3680 image::ust-sit+app-linked-with-tp-o+app-linked-with-lib+lib-instrumented.png[]
3683 include::../common/ust-sit-step-tp-o.txt[]
3685 To build the instrumented user library:
3687 . In path:{emon.c}, before including path:{tpp.h}, add the
3693 #define TRACEPOINT_DEFINE
3697 . Compile the user library source file:
3702 $ gcc -I. -fpic -c emon.c
3706 . Build the user library shared object:
3711 $ gcc -shared -o libemon.so emon.o
3715 To build the application:
3717 . Compile the application source file:
3726 . Build the application:
3731 $ gcc -o app app.o tpp.o -llttng-ust -ldl -L. -lemon
3735 To run the instrumented application:
3737 * Start the application:
3747 The application is statically linked with the tracepoint provider
3748 package object file.
3750 The application dynamically loads the instrumented user library.
3752 image::ust-sit+app-linked-with-tp-o+app-dlopens-lib+lib-instrumented.png[]
3755 include::../common/ust-sit-step-tp-o.txt[]
3757 To build the application:
3759 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3764 #define TRACEPOINT_DEFINE
3768 . Compile the application source file:
3777 . Build the application:
3782 $ gcc -Wl,--export-dynamic -o app app.o tpp.o \
3787 The `--export-dynamic` option passed to the linker is necessary for the
3788 dynamically loaded library to ``see'' the tracepoint symbols defined in
3791 To build the instrumented user library:
3793 . Compile the user library source file:
3798 $ gcc -I. -fpic -c emon.c
3802 . Build the user library shared object:
3807 $ gcc -shared -o libemon.so emon.o
3811 To run the application:
3813 * Start the application:
3825 .Do not use man:dlclose(3) on a tracepoint provider package
3827 Never use man:dlclose(3) on any shared object which:
3829 * Is linked with, statically or dynamically, a tracepoint provider
3831 * Calls man:dlopen(3) itself to dynamically open a tracepoint provider
3832 package shared object.
3834 This is currently considered **unsafe** due to a lack of reference
3835 counting from LTTng-UST to the shared object.
3837 A known workaround (available since glibc 2.2) is to use the
3838 `RTLD_NODELETE` flag when calling man:dlopen(3) initially. This has the
3839 effect of not unloading the loaded shared object, even if man:dlclose(3)
3842 You can also preload the tracepoint provider package shared object with
3843 the env:LD_PRELOAD environment variable to overcome this limitation.
3847 [[using-lttng-ust-with-daemons]]
3848 ===== Use noch:{LTTng-UST} with daemons
3850 If your instrumented application calls man:fork(2), man:clone(2),
3851 or BSD's man:rfork(2), without a following man:exec(3)-family
3852 system call, you must preload the path:{liblttng-ust-fork.so} shared
3853 object when you start the application.
3857 $ LD_PRELOAD=liblttng-ust-fork.so ./my-app
3860 If your tracepoint provider package is
3861 a shared library which you also preload, you must put both
3862 shared objects in env:LD_PRELOAD:
3866 $ LD_PRELOAD=liblttng-ust-fork.so:/path/to/tp.so ./my-app
3872 ===== Use noch:{LTTng-UST} with applications which close file descriptors that don't belong to them
3874 If your instrumented application closes one or more file descriptors
3875 which it did not open itself, you must preload the
3876 path:{liblttng-ust-fd.so} shared object when you start the application:
3880 $ LD_PRELOAD=liblttng-ust-fd.so ./my-app
3883 Typical use cases include closing all the file descriptors after
3884 man:fork(2) or man:rfork(2) and buggy applications doing
3888 [[lttng-ust-pkg-config]]
3889 ===== Use noch:{pkg-config}
3891 On some distributions, LTTng-UST ships with a
3892 https://www.freedesktop.org/wiki/Software/pkg-config/[pkg-config]
3893 metadata file. If this is your case, then you can use cmd:pkg-config to
3894 build an application on the command line:
3898 $ gcc -o my-app my-app.o tp.o $(pkg-config --cflags --libs lttng-ust)
3902 [[instrumenting-32-bit-app-on-64-bit-system]]
3903 ===== [[advanced-instrumenting-techniques]]Build a 32-bit instrumented application for a 64-bit target system
3905 In order to trace a 32-bit application running on a 64-bit system,
3906 LTTng must use a dedicated 32-bit
3907 <<lttng-consumerd,consumer daemon>>.
3909 The following steps show how to build and install a 32-bit consumer
3910 daemon, which is _not_ part of the default 64-bit LTTng build, how to
3911 build and install the 32-bit LTTng-UST libraries, and how to build and
3912 link an instrumented 32-bit application in that context.
3914 To build a 32-bit instrumented application for a 64-bit target system,
3915 assuming you have a fresh target system with no installed Userspace RCU
3918 . Download, build, and install a 32-bit version of Userspace RCU:
3923 $ cd $(mktemp -d) &&
3924 wget http://lttng.org/files/urcu/userspace-rcu-latest-0.9.tar.bz2 &&
3925 tar -xf userspace-rcu-latest-0.9.tar.bz2 &&
3926 cd userspace-rcu-0.9.* &&
3927 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 &&
3929 sudo make install &&
3934 . Using your distribution's package manager, or from source, install
3935 the following 32-bit versions of the following dependencies of
3936 LTTng-tools and LTTng-UST:
3939 * https://sourceforge.net/projects/libuuid/[libuuid]
3940 * http://directory.fsf.org/wiki/Popt[popt]
3941 * http://www.xmlsoft.org/[libxml2]
3944 . Download, build, and install a 32-bit version of the latest
3945 LTTng-UST{nbsp}{revision}:
3950 $ cd $(mktemp -d) &&
3951 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.9.tar.bz2 &&
3952 tar -xf lttng-ust-latest-2.9.tar.bz2 &&
3953 cd lttng-ust-2.9.* &&
3954 ./configure --libdir=/usr/local/lib32 \
3955 CFLAGS=-m32 CXXFLAGS=-m32 \
3956 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
3958 sudo make install &&
3965 Depending on your distribution,
3966 32-bit libraries could be installed at a different location than
3967 `/usr/lib32`. For example, Debian is known to install
3968 some 32-bit libraries in `/usr/lib/i386-linux-gnu`.
3970 In this case, make sure to set `LDFLAGS` to all the
3971 relevant 32-bit library paths, for example:
3975 $ LDFLAGS='-L/usr/lib/i386-linux-gnu -L/usr/lib32'
3979 . Download the latest LTTng-tools{nbsp}{revision}, build, and install
3980 the 32-bit consumer daemon:
3985 $ cd $(mktemp -d) &&
3986 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.9.tar.bz2 &&
3987 tar -xf lttng-tools-latest-2.9.tar.bz2 &&
3988 cd lttng-tools-2.9.* &&
3989 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 CXXFLAGS=-m32 \
3990 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' \
3991 --disable-bin-lttng --disable-bin-lttng-crash \
3992 --disable-bin-lttng-relayd --disable-bin-lttng-sessiond &&
3994 cd src/bin/lttng-consumerd &&
3995 sudo make install &&
4000 . From your distribution or from source,
4001 <<installing-lttng,install>> the 64-bit versions of
4002 LTTng-UST and Userspace RCU.
4003 . Download, build, and install the 64-bit version of the
4004 latest LTTng-tools{nbsp}{revision}:
4009 $ cd $(mktemp -d) &&
4010 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.9.tar.bz2 &&
4011 tar -xf lttng-tools-latest-2.9.tar.bz2 &&
4012 cd lttng-tools-2.9.* &&
4013 ./configure --with-consumerd32-libdir=/usr/local/lib32 \
4014 --with-consumerd32-bin=/usr/local/lib32/lttng/libexec/lttng-consumerd &&
4016 sudo make install &&
4021 . Pass the following options to man:gcc(1), man:g++(1), or man:clang(1)
4022 when linking your 32-bit application:
4025 -m32 -L/usr/lib32 -L/usr/local/lib32 \
4026 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32
4029 For example, let's rebuild the quick start example in
4030 <<tracing-your-own-user-application,Trace a user application>> as an
4031 instrumented 32-bit application:
4036 $ gcc -m32 -c -I. hello-tp.c
4037 $ gcc -m32 -c hello.c
4038 $ gcc -m32 -o hello hello.o hello-tp.o \
4039 -L/usr/lib32 -L/usr/local/lib32 \
4040 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32 \
4045 No special action is required to execute the 32-bit application and
4046 to trace it: use the command-line man:lttng(1) tool as usual.
4053 man:tracef(3) is a small LTTng-UST API designed for quick,
4054 man:printf(3)-like instrumentation without the burden of
4055 <<tracepoint-provider,creating>> and
4056 <<building-tracepoint-providers-and-user-application,building>>
4057 a tracepoint provider package.
4059 To use `tracef()` in your application:
4061 . In the C or C++ source files where you need to use `tracef()`,
4062 include `<lttng/tracef.h>`:
4067 #include <lttng/tracef.h>
4071 . In the application's source code, use `tracef()` like you would use
4079 tracef("my message: %d (%s)", my_integer, my_string);
4085 . Link your application with `liblttng-ust`:
4090 $ gcc -o app app.c -llttng-ust
4094 To trace the events that `tracef()` calls emit:
4096 * <<enabling-disabling-events,Create an event rule>> which matches the
4097 `lttng_ust_tracef:*` event name:
4102 $ lttng enable-event --userspace 'lttng_ust_tracef:*'
4107 .Limitations of `tracef()`
4109 The `tracef()` utility function was developed to make user space tracing
4110 super simple, albeit with notable disadvantages compared to
4111 <<defining-tracepoints,user-defined tracepoints>>:
4113 * All the emitted events have the same tracepoint provider and
4114 tracepoint names, respectively `lttng_ust_tracef` and `event`.
4115 * There is no static type checking.
4116 * The only event record field you actually get, named `msg`, is a string
4117 potentially containing the values you passed to `tracef()`
4118 using your own format string. This also means that you cannot filter
4119 events with a custom expression at run time because there are no
4121 * Since `tracef()` uses the C standard library's man:vasprintf(3)
4122 function behind the scenes to format the strings at run time, its
4123 expected performance is lower than with user-defined tracepoints,
4124 which do not require a conversion to a string.
4126 Taking this into consideration, `tracef()` is useful for some quick
4127 prototyping and debugging, but you should not consider it for any
4128 permanent and serious applicative instrumentation.
4134 ==== Use `tracelog()`
4136 The man:tracelog(3) API is very similar to <<tracef,`tracef()`>>, with
4137 the difference that it accepts an additional log level parameter.
4139 The goal of `tracelog()` is to ease the migration from logging to
4142 To use `tracelog()` in your application:
4144 . In the C or C++ source files where you need to use `tracelog()`,
4145 include `<lttng/tracelog.h>`:
4150 #include <lttng/tracelog.h>
4154 . In the application's source code, use `tracelog()` like you would use
4155 man:printf(3), except for the first parameter which is the log
4163 tracelog(TRACE_WARNING, "my message: %d (%s)",
4164 my_integer, my_string);
4170 See man:lttng-ust(3) for a list of available log level names.
4172 . Link your application with `liblttng-ust`:
4177 $ gcc -o app app.c -llttng-ust
4181 To trace the events that `tracelog()` calls emit with a log level
4182 _as severe as_ a specific log level:
4184 * <<enabling-disabling-events,Create an event rule>> which matches the
4185 `lttng_ust_tracelog:*` event name and a minimum level
4191 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4192 --loglevel=TRACE_WARNING
4196 To trace the events that `tracelog()` calls emit with a
4197 _specific log level_:
4199 * Create an event rule which matches the `lttng_ust_tracelog:*`
4200 event name and a specific log level:
4205 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4206 --loglevel-only=TRACE_INFO
4211 [[prebuilt-ust-helpers]]
4212 === Prebuilt user space tracing helpers
4214 The LTTng-UST package provides a few helpers in the form or preloadable
4215 shared objects which automatically instrument system functions and
4218 The helper shared objects are normally found in dir:{/usr/lib}. If you
4219 built LTTng-UST <<building-from-source,from source>>, they are probably
4220 located in dir:{/usr/local/lib}.
4222 The installed user space tracing helpers in LTTng-UST{nbsp}{revision}
4225 path:{liblttng-ust-libc-wrapper.so}::
4226 path:{liblttng-ust-pthread-wrapper.so}::
4227 <<liblttng-ust-libc-pthread-wrapper,C{nbsp}standard library
4228 memory and POSIX threads function tracing>>.
4230 path:{liblttng-ust-cyg-profile.so}::
4231 path:{liblttng-ust-cyg-profile-fast.so}::
4232 <<liblttng-ust-cyg-profile,Function entry and exit tracing>>.
4234 path:{liblttng-ust-dl.so}::
4235 <<liblttng-ust-dl,Dynamic linker tracing>>.
4237 To use a user space tracing helper with any user application:
4239 * Preload the helper shared object when you start the application:
4244 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so my-app
4248 You can preload more than one helper:
4253 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so:liblttng-ust-dl.so my-app
4259 [[liblttng-ust-libc-pthread-wrapper]]
4260 ==== Instrument C standard library memory and POSIX threads functions
4262 The path:{liblttng-ust-libc-wrapper.so} and
4263 path:{liblttng-ust-pthread-wrapper.so} helpers
4264 add instrumentation to some C standard library and POSIX
4268 .Functions instrumented by preloading path:{liblttng-ust-libc-wrapper.so}.
4270 |TP provider name |TP name |Instrumented function
4272 .6+|`lttng_ust_libc` |`malloc` |man:malloc(3)
4273 |`calloc` |man:calloc(3)
4274 |`realloc` |man:realloc(3)
4275 |`free` |man:free(3)
4276 |`memalign` |man:memalign(3)
4277 |`posix_memalign` |man:posix_memalign(3)
4281 .Functions instrumented by preloading path:{liblttng-ust-pthread-wrapper.so}.
4283 |TP provider name |TP name |Instrumented function
4285 .4+|`lttng_ust_pthread` |`pthread_mutex_lock_req` |man:pthread_mutex_lock(3p) (request time)
4286 |`pthread_mutex_lock_acq` |man:pthread_mutex_lock(3p) (acquire time)
4287 |`pthread_mutex_trylock` |man:pthread_mutex_trylock(3p)
4288 |`pthread_mutex_unlock` |man:pthread_mutex_unlock(3p)
4291 When you preload the shared object, it replaces the functions listed
4292 in the previous tables by wrappers which contain tracepoints and call
4293 the replaced functions.
4296 [[liblttng-ust-cyg-profile]]
4297 ==== Instrument function entry and exit
4299 The path:{liblttng-ust-cyg-profile*.so} helpers can add instrumentation
4300 to the entry and exit points of functions.
4302 man:gcc(1) and man:clang(1) have an option named
4303 https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html[`-finstrument-functions`]
4304 which generates instrumentation calls for entry and exit to functions.
4305 The LTTng-UST function tracing helpers,
4306 path:{liblttng-ust-cyg-profile.so} and
4307 path:{liblttng-ust-cyg-profile-fast.so}, take advantage of this feature
4308 to add tracepoints to the two generated functions (which contain
4309 `cyg_profile` in their names, hence the helper's name).
4311 To use the LTTng-UST function tracing helper, the source files to
4312 instrument must be built using the `-finstrument-functions` compiler
4315 There are two versions of the LTTng-UST function tracing helper:
4317 * **path:{liblttng-ust-cyg-profile-fast.so}** is a lightweight variant
4318 that you should only use when it can be _guaranteed_ that the
4319 complete event stream is recorded without any lost event record.
4320 Any kind of duplicate information is left out.
4322 Assuming no event record is lost, having only the function addresses on
4323 entry is enough to create a call graph, since an event record always
4324 contains the ID of the CPU that generated it.
4326 You can use a tool like man:addr2line(1) to convert function addresses
4327 back to source file names and line numbers.
4329 * **path:{liblttng-ust-cyg-profile.so}** is a more robust variant
4330 which also works in use cases where event records might get discarded or
4331 not recorded from application startup.
4332 In these cases, the trace analyzer needs more information to be
4333 able to reconstruct the program flow.
4335 See man:lttng-ust-cyg-profile(3) to learn more about the instrumentation
4336 points of this helper.
4338 All the tracepoints that this helper provides have the
4339 log level `TRACE_DEBUG_FUNCTION` (see man:lttng-ust(3)).
4341 TIP: It's sometimes a good idea to limit the number of source files that
4342 you compile with the `-finstrument-functions` option to prevent LTTng
4343 from writing an excessive amount of trace data at run time. When using
4344 man:gcc(1), you can use the
4345 `-finstrument-functions-exclude-function-list` option to avoid
4346 instrument entries and exits of specific function names.
4351 ==== Instrument the dynamic linker
4353 The path:{liblttng-ust-dl.so} helper adds instrumentation to the
4354 man:dlopen(3) and man:dlclose(3) function calls.
4356 See man:lttng-ust-dl(3) to learn more about the instrumentation points
4361 [[java-application]]
4362 === User space Java agent
4364 You can instrument any Java application which uses one of the following
4367 * The https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[**`java.util.logging`**]
4368 (JUL) core logging facilities.
4369 * http://logging.apache.org/log4j/1.2/[**Apache log4j 1.2**], since
4370 LTTng 2.6. Note that Apache Log4j{nbsp}2 is not supported.
4373 .LTTng-UST Java agent imported by a Java application.
4374 image::java-app.png[]
4376 Note that the methods described below are new in LTTng{nbsp}{revision}.
4377 Previous LTTng versions use another technique.
4379 NOTE: We use http://openjdk.java.net/[OpenJDK]{nbsp}8 for development
4380 and https://ci.lttng.org/[continuous integration], thus this version is
4381 directly supported. However, the LTTng-UST Java agent is also tested
4382 with OpenJDK{nbsp}7.
4387 ==== Use the LTTng-UST Java agent for `java.util.logging`
4389 To use the LTTng-UST Java agent in a Java application which uses
4390 `java.util.logging` (JUL):
4392 . In the Java application's source code, import the LTTng-UST
4393 log handler package for `java.util.logging`:
4398 import org.lttng.ust.agent.jul.LttngLogHandler;
4402 . Create an LTTng-UST JUL log handler:
4407 Handler lttngUstLogHandler = new LttngLogHandler();
4411 . Add this handler to the JUL loggers which should emit LTTng events:
4416 Logger myLogger = Logger.getLogger("some-logger");
4418 myLogger.addHandler(lttngUstLogHandler);
4422 . Use `java.util.logging` log statements and configuration as usual.
4423 The loggers with an attached LTTng-UST log handler can emit
4426 . Before exiting the application, remove the LTTng-UST log handler from
4427 the loggers attached to it and call its `close()` method:
4432 myLogger.removeHandler(lttngUstLogHandler);
4433 lttngUstLogHandler.close();
4437 This is not strictly necessary, but it is recommended for a clean
4438 disposal of the handler's resources.
4440 . Include the LTTng-UST Java agent's common and JUL-specific JAR files,
4441 path:{lttng-ust-agent-common.jar} and path:{lttng-ust-agent-jul.jar},
4443 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4444 path] when you build the Java application.
4446 The JAR files are typically located in dir:{/usr/share/java}.
4448 IMPORTANT: The LTTng-UST Java agent must be
4449 <<installing-lttng,installed>> for the logging framework your
4452 .Use the LTTng-UST Java agent for `java.util.logging`.
4457 import java.io.IOException;
4458 import java.util.logging.Handler;
4459 import java.util.logging.Logger;
4460 import org.lttng.ust.agent.jul.LttngLogHandler;
4464 private static final int answer = 42;
4466 public static void main(String[] argv) throws Exception
4469 Logger logger = Logger.getLogger("jello");
4471 // Create an LTTng-UST log handler
4472 Handler lttngUstLogHandler = new LttngLogHandler();
4474 // Add the LTTng-UST log handler to our logger
4475 logger.addHandler(lttngUstLogHandler);
4478 logger.info("some info");
4479 logger.warning("some warning");
4481 logger.finer("finer information; the answer is " + answer);
4483 logger.severe("error!");
4485 // Not mandatory, but cleaner
4486 logger.removeHandler(lttngUstLogHandler);
4487 lttngUstLogHandler.close();
4496 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4499 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4500 <<enabling-disabling-events,create an event rule>> matching the
4501 `jello` JUL logger, and <<basic-tracing-session-control,start tracing>>:
4506 $ lttng enable-event --jul jello
4510 Run the compiled class:
4514 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4517 <<basic-tracing-session-control,Stop tracing>> and inspect the
4527 In the resulting trace, an <<event,event record>> generated by a Java
4528 application using `java.util.logging` is named `lttng_jul:event` and
4529 has the following fields:
4532 Log record's message.
4538 Name of the class in which the log statement was executed.
4541 Name of the method in which the log statement was executed.
4544 Logging time (timestamp in milliseconds).
4547 Log level integer value.
4550 ID of the thread in which the log statement was executed.
4552 You can use the opt:lttng-enable-event(1):--loglevel or
4553 opt:lttng-enable-event(1):--loglevel-only option of the
4554 man:lttng-enable-event(1) command to target a range of JUL log levels
4555 or a specific JUL log level.
4560 ==== Use the LTTng-UST Java agent for Apache log4j
4562 To use the LTTng-UST Java agent in a Java application which uses
4565 . In the Java application's source code, import the LTTng-UST
4566 log appender package for Apache log4j:
4571 import org.lttng.ust.agent.log4j.LttngLogAppender;
4575 . Create an LTTng-UST log4j log appender:
4580 Appender lttngUstLogAppender = new LttngLogAppender();
4584 . Add this appender to the log4j loggers which should emit LTTng events:
4589 Logger myLogger = Logger.getLogger("some-logger");
4591 myLogger.addAppender(lttngUstLogAppender);
4595 . Use Apache log4j log statements and configuration as usual. The
4596 loggers with an attached LTTng-UST log appender can emit LTTng events.
4598 . Before exiting the application, remove the LTTng-UST log appender from
4599 the loggers attached to it and call its `close()` method:
4604 myLogger.removeAppender(lttngUstLogAppender);
4605 lttngUstLogAppender.close();
4609 This is not strictly necessary, but it is recommended for a clean
4610 disposal of the appender's resources.
4612 . Include the LTTng-UST Java agent's common and log4j-specific JAR
4613 files, path:{lttng-ust-agent-common.jar} and
4614 path:{lttng-ust-agent-log4j.jar}, in the
4615 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4616 path] when you build the Java application.
4618 The JAR files are typically located in dir:{/usr/share/java}.
4620 IMPORTANT: The LTTng-UST Java agent must be
4621 <<installing-lttng,installed>> for the logging framework your
4624 .Use the LTTng-UST Java agent for Apache log4j.
4629 import org.apache.log4j.Appender;
4630 import org.apache.log4j.Logger;
4631 import org.lttng.ust.agent.log4j.LttngLogAppender;
4635 private static final int answer = 42;
4637 public static void main(String[] argv) throws Exception
4640 Logger logger = Logger.getLogger("jello");
4642 // Create an LTTng-UST log appender
4643 Appender lttngUstLogAppender = new LttngLogAppender();
4645 // Add the LTTng-UST log appender to our logger
4646 logger.addAppender(lttngUstLogAppender);
4649 logger.info("some info");
4650 logger.warn("some warning");
4652 logger.debug("debug information; the answer is " + answer);
4654 logger.fatal("error!");
4656 // Not mandatory, but cleaner
4657 logger.removeAppender(lttngUstLogAppender);
4658 lttngUstLogAppender.close();
4664 Build this example (`$LOG4JPATH` is the path to the Apache log4j JAR
4669 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH Test.java
4672 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4673 <<enabling-disabling-events,create an event rule>> matching the
4674 `jello` log4j logger, and <<basic-tracing-session-control,start tracing>>:
4679 $ lttng enable-event --log4j jello
4683 Run the compiled class:
4687 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH:. Test
4690 <<basic-tracing-session-control,Stop tracing>> and inspect the
4700 In the resulting trace, an <<event,event record>> generated by a Java
4701 application using log4j is named `lttng_log4j:event` and
4702 has the following fields:
4705 Log record's message.
4711 Name of the class in which the log statement was executed.
4714 Name of the method in which the log statement was executed.
4717 Name of the file in which the executed log statement is located.
4720 Line number at which the log statement was executed.
4726 Log level integer value.
4729 Name of the Java thread in which the log statement was executed.
4731 You can use the opt:lttng-enable-event(1):--loglevel or
4732 opt:lttng-enable-event(1):--loglevel-only option of the
4733 man:lttng-enable-event(1) command to target a range of Apache log4j log levels
4734 or a specific log4j log level.
4738 [[java-application-context]]
4739 ==== Provide application-specific context fields in a Java application
4741 A Java application-specific context field is a piece of state provided
4742 by the application which <<adding-context,you can add>>, using the
4743 man:lttng-add-context(1) command, to each <<event,event record>>
4744 produced by the log statements of this application.
4746 For example, a given object might have a current request ID variable.
4747 You can create a context information retriever for this object and
4748 assign a name to this current request ID. You can then, using the
4749 man:lttng-add-context(1) command, add this context field by name to
4750 the JUL or log4j <<channel,channel>>.
4752 To provide application-specific context fields in a Java application:
4754 . In the Java application's source code, import the LTTng-UST
4755 Java agent context classes and interfaces:
4760 import org.lttng.ust.agent.context.ContextInfoManager;
4761 import org.lttng.ust.agent.context.IContextInfoRetriever;
4765 . Create a context information retriever class, that is, a class which
4766 implements the `IContextInfoRetriever` interface:
4771 class MyContextInfoRetriever implements IContextInfoRetriever
4774 public Object retrieveContextInfo(String key)
4776 if (key.equals("intCtx")) {
4778 } else if (key.equals("strContext")) {
4779 return "context value!";
4788 This `retrieveContextInfo()` method is the only member of the
4789 `IContextInfoRetriever` interface. Its role is to return the current
4790 value of a state by name to create a context field. The names of the
4791 context fields and which state variables they return depends on your
4794 All primitive types and objects are supported as context fields.
4795 When `retrieveContextInfo()` returns an object, the context field
4796 serializer calls its `toString()` method to add a string field to
4797 event records. The method can also return `null`, which means that
4798 no context field is available for the required name.
4800 . Register an instance of your context information retriever class to
4801 the context information manager singleton:
4806 IContextInfoRetriever cir = new MyContextInfoRetriever();
4807 ContextInfoManager cim = ContextInfoManager.getInstance();
4808 cim.registerContextInfoRetriever("retrieverName", cir);
4812 . Before exiting the application, remove your context information
4813 retriever from the context information manager singleton:
4818 ContextInfoManager cim = ContextInfoManager.getInstance();
4819 cim.unregisterContextInfoRetriever("retrieverName");
4823 This is not strictly necessary, but it is recommended for a clean
4824 disposal of some manager's resources.
4826 . Build your Java application with LTTng-UST Java agent support as
4827 usual, following the procedure for either the <<jul,JUL>> or
4828 <<log4j,Apache log4j>> framework.
4831 .Provide application-specific context fields in a Java application.
4836 import java.util.logging.Handler;
4837 import java.util.logging.Logger;
4838 import org.lttng.ust.agent.jul.LttngLogHandler;
4839 import org.lttng.ust.agent.context.ContextInfoManager;
4840 import org.lttng.ust.agent.context.IContextInfoRetriever;
4844 // Our context information retriever class
4845 private static class MyContextInfoRetriever
4846 implements IContextInfoRetriever
4849 public Object retrieveContextInfo(String key) {
4850 if (key.equals("intCtx")) {
4852 } else if (key.equals("strContext")) {
4853 return "context value!";
4860 private static final int answer = 42;
4862 public static void main(String args[]) throws Exception
4864 // Get the context information manager instance
4865 ContextInfoManager cim = ContextInfoManager.getInstance();
4867 // Create and register our context information retriever
4868 IContextInfoRetriever cir = new MyContextInfoRetriever();
4869 cim.registerContextInfoRetriever("myRetriever", cir);
4872 Logger logger = Logger.getLogger("jello");
4874 // Create an LTTng-UST log handler
4875 Handler lttngUstLogHandler = new LttngLogHandler();
4877 // Add the LTTng-UST log handler to our logger
4878 logger.addHandler(lttngUstLogHandler);
4881 logger.info("some info");
4882 logger.warning("some warning");
4884 logger.finer("finer information; the answer is " + answer);
4886 logger.severe("error!");
4888 // Not mandatory, but cleaner
4889 logger.removeHandler(lttngUstLogHandler);
4890 lttngUstLogHandler.close();
4891 cim.unregisterContextInfoRetriever("myRetriever");
4900 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4903 <<creating-destroying-tracing-sessions,Create a tracing session>>
4904 and <<enabling-disabling-events,create an event rule>> matching the
4910 $ lttng enable-event --jul jello
4913 <<adding-context,Add the application-specific context fields>> to the
4918 $ lttng add-context --jul --type='$app.myRetriever:intCtx'
4919 $ lttng add-context --jul --type='$app.myRetriever:strContext'
4922 <<basic-tracing-session-control,Start tracing>>:
4929 Run the compiled class:
4933 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4936 <<basic-tracing-session-control,Stop tracing>> and inspect the
4948 [[python-application]]
4949 === User space Python agent
4951 You can instrument a Python 2 or Python 3 application which uses the
4952 standard https://docs.python.org/3/library/logging.html[`logging`]
4955 Each log statement emits an LTTng event once the
4956 application module imports the
4957 <<lttng-ust-agents,LTTng-UST Python agent>> package.
4960 .A Python application importing the LTTng-UST Python agent.
4961 image::python-app.png[]
4963 To use the LTTng-UST Python agent:
4965 . In the Python application's source code, import the LTTng-UST Python
4975 The LTTng-UST Python agent automatically adds its logging handler to the
4976 root logger at import time.
4978 Any log statement that the application executes before this import does
4979 not emit an LTTng event.
4981 IMPORTANT: The LTTng-UST Python agent must be
4982 <<installing-lttng,installed>>.
4984 . Use log statements and logging configuration as usual.
4985 Since the LTTng-UST Python agent adds a handler to the _root_
4986 logger, you can trace any log statement from any logger.
4988 .Use the LTTng-UST Python agent.
4999 logging.basicConfig()
5000 logger = logging.getLogger('my-logger')
5003 logger.debug('debug message')
5004 logger.info('info message')
5005 logger.warn('warn message')
5006 logger.error('error message')
5007 logger.critical('critical message')
5011 if __name__ == '__main__':
5015 NOTE: `logging.basicConfig()`, which adds to the root logger a basic
5016 logging handler which prints to the standard error stream, is not
5017 strictly required for LTTng-UST tracing to work, but in versions of
5018 Python preceding 3.2, you could see a warning message which indicates
5019 that no handler exists for the logger `my-logger`.
5021 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5022 <<enabling-disabling-events,create an event rule>> matching the
5023 `my-logger` Python logger, and <<basic-tracing-session-control,start
5029 $ lttng enable-event --python my-logger
5033 Run the Python script:
5040 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5050 In the resulting trace, an <<event,event record>> generated by a Python
5051 application is named `lttng_python:event` and has the following fields:
5054 Logging time (string).
5057 Log record's message.
5063 Name of the function in which the log statement was executed.
5066 Line number at which the log statement was executed.
5069 Log level integer value.
5072 ID of the Python thread in which the log statement was executed.
5075 Name of the Python thread in which the log statement was executed.
5077 You can use the opt:lttng-enable-event(1):--loglevel or
5078 opt:lttng-enable-event(1):--loglevel-only option of the
5079 man:lttng-enable-event(1) command to target a range of Python log levels
5080 or a specific Python log level.
5082 When an application imports the LTTng-UST Python agent, the agent tries
5083 to register to a <<lttng-sessiond,session daemon>>. Note that you must
5084 <<start-sessiond,start the session daemon>> _before_ you run the Python
5085 application. If a session daemon is found, the agent tries to register
5086 to it during 5{nbsp}seconds, after which the application continues
5087 without LTTng tracing support. You can override this timeout value with
5088 the env:LTTNG_UST_PYTHON_REGISTER_TIMEOUT environment variable
5091 If the session daemon stops while a Python application with an imported
5092 LTTng-UST Python agent runs, the agent retries to connect and to
5093 register to a session daemon every 3{nbsp}seconds. You can override this
5094 delay with the env:LTTNG_UST_PYTHON_REGISTER_RETRY_DELAY environment
5099 [[proc-lttng-logger-abi]]
5102 The `lttng-tracer` Linux kernel module, part of
5103 <<lttng-modules,LTTng-modules>>, creates the special LTTng logger file
5104 path:{/proc/lttng-logger} when it's loaded. Any application can write
5105 text data to this file to emit an LTTng event.
5108 .An application writes to the LTTng logger file to emit an LTTng event.
5109 image::lttng-logger.png[]
5111 The LTTng logger is the quickest method--not the most efficient,
5112 however--to add instrumentation to an application. It is designed
5113 mostly to instrument shell scripts:
5117 $ echo "Some message, some $variable" > /proc/lttng-logger
5120 Any event that the LTTng logger emits is named `lttng_logger` and
5121 belongs to the Linux kernel <<domain,tracing domain>>. However, unlike
5122 other instrumentation points in the kernel tracing domain, **any Unix
5123 user** can <<enabling-disabling-events,create an event rule>> which
5124 matches its event name, not only the root user or users in the
5125 <<tracing-group,tracing group>>.
5127 To use the LTTng logger:
5129 * From any application, write text data to the path:{/proc/lttng-logger}
5132 The `msg` field of `lttng_logger` event records contains the
5135 NOTE: The maximum message length of an LTTng logger event is
5136 1024{nbsp}bytes. Writing more than this makes the LTTng logger emit more
5137 than one event to contain the remaining data.
5139 You should not use the LTTng logger to trace a user application which
5140 can be instrumented in a more efficient way, namely:
5142 * <<c-application,C and $$C++$$ applications>>.
5143 * <<java-application,Java applications>>.
5144 * <<python-application,Python applications>>.
5146 .Use the LTTng logger.
5151 echo 'Hello, World!' > /proc/lttng-logger
5153 df --human-readable --print-type / > /proc/lttng-logger
5156 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5157 <<enabling-disabling-events,create an event rule>> matching the
5158 `lttng_logger` Linux kernel tracepoint, and
5159 <<basic-tracing-session-control,start tracing>>:
5164 $ lttng enable-event --kernel lttng_logger
5168 Run the Bash script:
5175 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5186 [[instrumenting-linux-kernel]]
5187 === LTTng kernel tracepoints
5189 NOTE: This section shows how to _add_ instrumentation points to the
5190 Linux kernel. The kernel's subsystems are already thoroughly
5191 instrumented at strategic places for LTTng when you
5192 <<installing-lttng,install>> the <<lttng-modules,LTTng-modules>>
5196 There are two methods to instrument the Linux kernel:
5198 . <<linux-add-lttng-layer,Add an LTTng layer>> over an existing ftrace
5199 tracepoint which uses the `TRACE_EVENT()` API.
5201 Choose this if you want to instrumentation a Linux kernel tree with an
5202 instrumentation point compatible with ftrace, perf, and SystemTap.
5204 . Use an <<linux-lttng-tracepoint-event,LTTng-only approach>> to
5205 instrument an out-of-tree kernel module.
5207 Choose this if you don't need ftrace, perf, or SystemTap support.
5211 [[linux-add-lttng-layer]]
5212 ==== [[instrumenting-linux-kernel-itself]][[mainline-trace-event]][[lttng-adaptation-layer]]Add an LTTng layer to an existing ftrace tracepoint
5214 This section shows how to add an LTTng layer to existing ftrace
5215 instrumentation using the `TRACE_EVENT()` API.
5217 This section does not document the `TRACE_EVENT()` macro. You can
5218 read the following articles to learn more about this API:
5220 * http://lwn.net/Articles/379903/[Using the TRACE_EVENT() macro (Part 1)]
5221 * http://lwn.net/Articles/381064/[Using the TRACE_EVENT() macro (Part 2)]
5222 * http://lwn.net/Articles/383362/[Using the TRACE_EVENT() macro (Part 3)]
5224 The following procedure assumes that your ftrace tracepoints are
5225 correctly defined in their own header and that they are created in
5226 one source file using the `CREATE_TRACE_POINTS` definition.
5228 To add an LTTng layer over an existing ftrace tracepoint:
5230 . Make sure the following kernel configuration options are
5236 * `CONFIG_HIGH_RES_TIMERS`
5237 * `CONFIG_TRACEPOINTS`
5240 . Build the Linux source tree with your custom ftrace tracepoints.
5241 . Boot the resulting Linux image on your target system.
5243 Confirm that the tracepoints exist by looking for their names in the
5244 dir:{/sys/kernel/debug/tracing/events/subsys} directory, where `subsys`
5245 is your subsystem's name.
5247 . Get a copy of the latest LTTng-modules{nbsp}{revision}:
5252 $ cd $(mktemp -d) &&
5253 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.9.tar.bz2 &&
5254 tar -xf lttng-modules-latest-2.9.tar.bz2 &&
5255 cd lttng-modules-2.9.*
5259 . In dir:{instrumentation/events/lttng-module}, relative to the root
5260 of the LTTng-modules source tree, create a header file named
5261 +__subsys__.h+ for your custom subsystem +__subsys__+ and write your
5262 LTTng-modules tracepoint definitions using the LTTng-modules
5265 Start with this template:
5269 .path:{instrumentation/events/lttng-module/my_subsys.h}
5272 #define TRACE_SYSTEM my_subsys
5274 #if !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ)
5275 #define _LTTNG_MY_SUBSYS_H
5277 #include "../../../probes/lttng-tracepoint-event.h"
5278 #include <linux/tracepoint.h>
5280 LTTNG_TRACEPOINT_EVENT(
5282 * Format is identical to TRACE_EVENT()'s version for the three
5283 * following macro parameters:
5286 TP_PROTO(int my_int, const char *my_string),
5287 TP_ARGS(my_int, my_string),
5289 /* LTTng-modules specific macros */
5291 ctf_integer(int, my_int_field, my_int)
5292 ctf_string(my_bar_field, my_bar)
5296 #endif /* !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ) */
5298 #include "../../../probes/define_trace.h"
5302 The entries in the `TP_FIELDS()` section are the list of fields for the
5303 LTTng tracepoint. This is similar to the `TP_STRUCT__entry()` part of
5304 ftrace's `TRACE_EVENT()` macro.
5306 See <<lttng-modules-tp-fields,Tracepoint fields macros>> for a
5307 complete description of the available `ctf_*()` macros.
5309 . Create the LTTng-modules probe's kernel module C source file,
5310 +probes/lttng-probe-__subsys__.c+, where +__subsys__+ is your
5315 .path:{probes/lttng-probe-my-subsys.c}
5317 #include <linux/module.h>
5318 #include "../lttng-tracer.h"
5321 * Build-time verification of mismatch between mainline
5322 * TRACE_EVENT() arguments and the LTTng-modules adaptation
5323 * layer LTTNG_TRACEPOINT_EVENT() arguments.
5325 #include <trace/events/my_subsys.h>
5327 /* Create LTTng tracepoint probes */
5328 #define LTTNG_PACKAGE_BUILD
5329 #define CREATE_TRACE_POINTS
5330 #define TRACE_INCLUDE_PATH ../instrumentation/events/lttng-module
5332 #include "../instrumentation/events/lttng-module/my_subsys.h"
5334 MODULE_LICENSE("GPL and additional rights");
5335 MODULE_AUTHOR("Your name <your-email>");
5336 MODULE_DESCRIPTION("LTTng my_subsys probes");
5337 MODULE_VERSION(__stringify(LTTNG_MODULES_MAJOR_VERSION) "."
5338 __stringify(LTTNG_MODULES_MINOR_VERSION) "."
5339 __stringify(LTTNG_MODULES_PATCHLEVEL_VERSION)
5340 LTTNG_MODULES_EXTRAVERSION);
5344 . Edit path:{probes/KBuild} and add your new kernel module object
5345 next to the existing ones:
5349 .path:{probes/KBuild}
5353 obj-m += lttng-probe-module.o
5354 obj-m += lttng-probe-power.o
5356 obj-m += lttng-probe-my-subsys.o
5362 . Build and install the LTTng kernel modules:
5367 $ make KERNELDIR=/path/to/linux
5368 # make modules_install && depmod -a
5372 Replace `/path/to/linux` with the path to the Linux source tree where
5373 you defined and used tracepoints with ftrace's `TRACE_EVENT()` macro.
5375 Note that you can also use the
5376 <<lttng-tracepoint-event-code,`LTTNG_TRACEPOINT_EVENT_CODE()` macro>>
5377 instead of `LTTNG_TRACEPOINT_EVENT()` to use custom local variables and
5378 C code that need to be executed before the event fields are recorded.
5380 The best way to learn how to use the previous LTTng-modules macros is to
5381 inspect the existing LTTng-modules tracepoint definitions in the
5382 dir:{instrumentation/events/lttng-module} header files. Compare them
5383 with the Linux kernel mainline versions in the
5384 dir:{include/trace/events} directory of the Linux source tree.
5388 [[lttng-tracepoint-event-code]]
5389 ===== Use custom C code to access the data for tracepoint fields
5391 Although we recommended to always use the
5392 <<lttng-adaptation-layer,`LTTNG_TRACEPOINT_EVENT()`>> macro to describe
5393 the arguments and fields of an LTTng-modules tracepoint when possible,
5394 sometimes you need a more complex process to access the data that the
5395 tracer records as event record fields. In other words, you need local
5396 variables and multiple C{nbsp}statements instead of simple
5397 argument-based expressions that you pass to the
5398 <<lttng-modules-tp-fields,`ctf_*()` macros of `TP_FIELDS()`>>.
5400 You can use the `LTTNG_TRACEPOINT_EVENT_CODE()` macro instead of
5401 `LTTNG_TRACEPOINT_EVENT()` to declare custom local variables and define
5402 a block of C{nbsp}code to be executed before LTTng records the fields.
5403 The structure of this macro is:
5406 .`LTTNG_TRACEPOINT_EVENT_CODE()` macro syntax.
5408 LTTNG_TRACEPOINT_EVENT_CODE(
5410 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5411 * version for the following three macro parameters:
5414 TP_PROTO(int my_int, const char *my_string),
5415 TP_ARGS(my_int, my_string),
5417 /* Declarations of custom local variables */
5420 unsigned long b = 0;
5421 const char *name = "(undefined)";
5422 struct my_struct *my_struct;
5426 * Custom code which uses both tracepoint arguments
5427 * (in TP_ARGS()) and local variables (in TP_locvar()).
5429 * Local variables are actually members of a structure pointed
5430 * to by the special variable tp_locvar.
5434 tp_locvar->a = my_int + 17;
5435 tp_locvar->my_struct = get_my_struct_at(tp_locvar->a);
5436 tp_locvar->b = my_struct_compute_b(tp_locvar->my_struct);
5437 tp_locvar->name = my_struct_get_name(tp_locvar->my_struct);
5438 put_my_struct(tp_locvar->my_struct);
5447 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5448 * version for this, except that tp_locvar members can be
5449 * used in the argument expression parameters of
5450 * the ctf_*() macros.
5453 ctf_integer(unsigned long, my_struct_b, tp_locvar->b)
5454 ctf_integer(int, my_struct_a, tp_locvar->a)
5455 ctf_string(my_string_field, my_string)
5456 ctf_string(my_struct_name, tp_locvar->name)
5461 IMPORTANT: The C code defined in `TP_code()` must not have any side
5462 effects when executed. In particular, the code must not allocate
5463 memory or get resources without deallocating this memory or putting
5464 those resources afterwards.
5467 [[instrumenting-linux-kernel-tracing]]
5468 ==== Load and unload a custom probe kernel module
5470 You must load a <<lttng-adaptation-layer,created LTTng-modules probe
5471 kernel module>> in the kernel before it can emit LTTng events.
5473 To load the default probe kernel modules and a custom probe kernel
5476 * Use the opt:lttng-sessiond(8):--extra-kmod-probes option to give extra
5477 probe modules to load when starting a root <<lttng-sessiond,session
5481 .Load the `my_subsys`, `usb`, and the default probe modules.
5485 # lttng-sessiond --extra-kmod-probes=my_subsys,usb
5490 You only need to pass the subsystem name, not the whole kernel module
5493 To load _only_ a given custom probe kernel module:
5495 * Use the opt:lttng-sessiond(8):--kmod-probes option to give the probe
5496 modules to load when starting a root session daemon:
5499 .Load only the `my_subsys` and `usb` probe modules.
5503 # lttng-sessiond --kmod-probes=my_subsys,usb
5508 To confirm that a probe module is loaded:
5515 $ lsmod | grep lttng_probe_usb
5519 To unload the loaded probe modules:
5521 * Kill the session daemon with `SIGTERM`:
5526 # pkill lttng-sessiond
5530 You can also use man:modprobe(8)'s `--remove` option if the session
5531 daemon terminates abnormally.
5534 [[controlling-tracing]]
5537 Once an application or a Linux kernel is
5538 <<instrumenting,instrumented>> for LTTng tracing,
5541 This section is divided in topics on how to use the various
5542 <<plumbing,components of LTTng>>, in particular the <<lttng-cli,cmd:lttng
5543 command-line tool>>, to _control_ the LTTng daemons and tracers.
5545 NOTE: In the following subsections, we refer to an man:lttng(1) command
5546 using its man page name. For example, instead of _Run the `create`
5547 command to..._, we use _Run the man:lttng-create(1) command to..._.
5551 === Start a session daemon
5553 In some situations, you need to run a <<lttng-sessiond,session daemon>>
5554 (man:lttng-sessiond(8)) _before_ you can use the man:lttng(1)
5557 You will see the following error when you run a command while no session
5561 Error: No session daemon is available
5564 The only command that automatically runs a session daemon is
5565 man:lttng-create(1), which you use to
5566 <<creating-destroying-tracing-sessions,create a tracing session>>. While
5567 this is most of the time the first operation that you do, sometimes it's
5568 not. Some examples are:
5570 * <<list-instrumentation-points,List the available instrumentation points>>.
5571 * <<saving-loading-tracing-session,Load a tracing session configuration>>.
5573 [[tracing-group]] Each Unix user must have its own running session
5574 daemon to trace user applications. The session daemon that the root user
5575 starts is the only one allowed to control the LTTng kernel tracer. Users
5576 that are part of the _tracing group_ can control the root session
5577 daemon. The default tracing group name is `tracing`; you can set it to
5578 something else with the opt:lttng-sessiond(8):--group option when you
5579 start the root session daemon.
5581 To start a user session daemon:
5583 * Run man:lttng-sessiond(8):
5588 $ lttng-sessiond --daemonize
5592 To start the root session daemon:
5594 * Run man:lttng-sessiond(8) as the root user:
5599 # lttng-sessiond --daemonize
5603 In both cases, remove the opt:lttng-sessiond(8):--daemonize option to
5604 start the session daemon in foreground.
5606 To stop a session daemon, use man:kill(1) on its process ID (standard
5609 Note that some Linux distributions could manage the LTTng session daemon
5610 as a service. In this case, you should use the service manager to
5611 start, restart, and stop session daemons.
5614 [[creating-destroying-tracing-sessions]]
5615 === Create and destroy a tracing session
5617 Almost all the LTTng control operations happen in the scope of
5618 a <<tracing-session,tracing session>>, which is the dialogue between the
5619 <<lttng-sessiond,session daemon>> and you.
5621 To create a tracing session with a generated name:
5623 * Use the man:lttng-create(1) command:
5632 The created tracing session's name is `auto` followed by the
5635 To create a tracing session with a specific name:
5637 * Use the optional argument of the man:lttng-create(1) command:
5642 $ lttng create my-session
5646 Replace `my-session` with the specific tracing session name.
5648 LTTng appends the creation date to the created tracing session's name.
5650 LTTng writes the traces of a tracing session in
5651 +$LTTNG_HOME/lttng-trace/__name__+ by default, where +__name__+ is the
5652 name of the tracing session. Note that the env:LTTNG_HOME environment
5653 variable defaults to `$HOME` if not set.
5655 To output LTTng traces to a non-default location:
5657 * Use the opt:lttng-create(1):--output option of the man:lttng-create(1) command:
5662 $ lttng create my-session --output=/tmp/some-directory
5666 You may create as many tracing sessions as you wish.
5668 To list all the existing tracing sessions for your Unix user:
5670 * Use the man:lttng-list(1) command:
5679 When you create a tracing session, it is set as the _current tracing
5680 session_. The following man:lttng(1) commands operate on the current
5681 tracing session when you don't specify one:
5683 [role="list-3-cols"]
5700 To change the current tracing session:
5702 * Use the man:lttng-set-session(1) command:
5707 $ lttng set-session new-session
5711 Replace `new-session` by the name of the new current tracing session.
5713 When you are done tracing in a given tracing session, you can destroy
5714 it. This operation frees the resources taken by the tracing session
5715 to destroy; it does not destroy the trace data that LTTng wrote for
5716 this tracing session.
5718 To destroy the current tracing session:
5720 * Use the man:lttng-destroy(1) command:
5730 [[list-instrumentation-points]]
5731 === List the available instrumentation points
5733 The <<lttng-sessiond,session daemon>> can query the running instrumented
5734 user applications and the Linux kernel to get a list of available
5735 instrumentation points. For the Linux kernel <<domain,tracing domain>>,
5736 they are tracepoints and system calls. For the user space tracing
5737 domain, they are tracepoints. For the other tracing domains, they are
5740 To list the available instrumentation points:
5742 * Use the man:lttng-list(1) command with the requested tracing domain's
5746 * opt:lttng-list(1):--kernel: Linux kernel tracepoints (your Unix user
5747 must be a root user, or it must be a member of the
5748 <<tracing-group,tracing group>>).
5749 * opt:lttng-list(1):--kernel with opt:lttng-list(1):--syscall: Linux
5750 kernel system calls (your Unix user must be a root user, or it must be
5751 a member of the tracing group).
5752 * opt:lttng-list(1):--userspace: user space tracepoints.
5753 * opt:lttng-list(1):--jul: `java.util.logging` loggers.
5754 * opt:lttng-list(1):--log4j: Apache log4j loggers.
5755 * opt:lttng-list(1):--python: Python loggers.
5758 .List the available user space tracepoints.
5762 $ lttng list --userspace
5766 .List the available Linux kernel system call tracepoints.
5770 $ lttng list --kernel --syscall
5775 [[enabling-disabling-events]]
5776 === Create and enable an event rule
5778 Once you <<creating-destroying-tracing-sessions,create a tracing
5779 session>>, you can create <<event,event rules>> with the
5780 man:lttng-enable-event(1) command.
5782 You specify each condition with a command-line option. The available
5783 condition options are shown in the following table.
5785 [role="growable",cols="asciidoc,asciidoc,default"]
5786 .Condition command-line options for the man:lttng-enable-event(1) command.
5788 |Option |Description |Applicable tracing domains
5794 . +--probe=__ADDR__+
5795 . +--function=__ADDR__+
5798 Instead of using the default _tracepoint_ instrumentation type, use:
5800 . A Linux system call.
5801 . A Linux https://lwn.net/Articles/132196/[KProbe] (symbol or address).
5802 . The entry and return points of a Linux function (symbol or address).
5806 |First positional argument.
5809 Tracepoint or system call name. In the case of a Linux KProbe or
5810 function, this is a custom name given to the event rule. With the
5811 JUL, log4j, and Python domains, this is a logger name.
5813 With a tracepoint, logger, or system call name, the last character
5814 can be `*` to match anything that remains.
5821 . +--loglevel=__LEVEL__+
5822 . +--loglevel-only=__LEVEL__+
5825 . Match only tracepoints or log statements with a logging level at
5826 least as severe as +__LEVEL__+.
5827 . Match only tracepoints or log statements with a logging level
5828 equal to +__LEVEL__+.
5830 See man:lttng-enable-event(1) for the list of available logging level
5833 |User space, JUL, log4j, and Python.
5835 |+--exclude=__EXCLUSIONS__+
5838 When you use a `*` character at the end of the tracepoint or logger
5839 name (first positional argument), exclude the specific names in the
5840 comma-delimited list +__EXCLUSIONS__+.
5843 User space, JUL, log4j, and Python.
5845 |+--filter=__EXPR__+
5848 Match only events which satisfy the expression +__EXPR__+.
5850 See man:lttng-enable-event(1) to learn more about the syntax of a
5857 You attach an event rule to a <<channel,channel>> on creation. If you do
5858 not specify the channel with the opt:lttng-enable-event(1):--channel
5859 option, and if the event rule to create is the first in its
5860 <<domain,tracing domain>> for a given tracing session, then LTTng
5861 creates a _default channel_ for you. This default channel is reused in
5862 subsequent invocations of the man:lttng-enable-event(1) command for the
5863 same tracing domain.
5865 An event rule is always enabled at creation time.
5867 The following examples show how you can combine the previous
5868 command-line options to create simple to more complex event rules.
5870 .Create an event rule targetting a Linux kernel tracepoint (default channel).
5874 $ lttng enable-event --kernel sched_switch
5878 .Create an event rule matching four Linux kernel system calls (default channel).
5882 $ lttng enable-event --kernel --syscall open,write,read,close
5886 .Create event rules matching tracepoints with filter expressions (default channel).
5890 $ lttng enable-event --kernel sched_switch --filter='prev_comm == "bash"'
5895 $ lttng enable-event --kernel --all \
5896 --filter='$ctx.tid == 1988 || $ctx.tid == 1534'
5901 $ lttng enable-event --jul my_logger \
5902 --filter='$app.retriever:cur_msg_id > 3'
5905 IMPORTANT: Make sure to always quote the filter string when you
5906 use man:lttng(1) from a shell.
5909 .Create an event rule matching any user space tracepoint of a given tracepoint provider with a log level range (default channel).
5913 $ lttng enable-event --userspace my_app:'*' --loglevel=TRACE_INFO
5916 IMPORTANT: Make sure to always quote the wildcard character when you
5917 use man:lttng(1) from a shell.
5920 .Create an event rule matching multiple Python loggers with a wildcard and with exclusions (default channel).
5924 $ lttng enable-event --python my-app.'*' \
5925 --exclude='my-app.module,my-app.hello'
5929 .Create an event rule matching any Apache log4j logger with a specific log level (default channel).
5933 $ lttng enable-event --log4j --all --loglevel-only=LOG4J_WARN
5937 .Create an event rule attached to a specific channel matching a specific user space tracepoint provider and tracepoint.
5941 $ lttng enable-event --userspace my_app:my_tracepoint --channel=my-channel
5945 The event rules of a given channel form a whitelist: as soon as an
5946 emitted event passes one of them, LTTng can record the event. For
5947 example, an event named `my_app:my_tracepoint` emitted from a user space
5948 tracepoint with a `TRACE_ERROR` log level passes both of the following
5953 $ lttng enable-event --userspace my_app:my_tracepoint
5954 $ lttng enable-event --userspace my_app:my_tracepoint \
5955 --loglevel=TRACE_INFO
5958 The second event rule is redundant: the first one includes
5962 [[disable-event-rule]]
5963 === Disable an event rule
5965 To disable an event rule that you <<enabling-disabling-events,created>>
5966 previously, use the man:lttng-disable-event(1) command. This command
5967 disables _all_ the event rules (of a given tracing domain and channel)
5968 which match an instrumentation point. The other conditions are not
5969 supported as of LTTng{nbsp}{revision}.
5971 The LTTng tracer does not record an emitted event which passes
5972 a _disabled_ event rule.
5974 .Disable an event rule matching a Python logger (default channel).
5978 $ lttng disable-event --python my-logger
5982 .Disable an event rule matching all `java.util.logging` loggers (default channel).
5986 $ lttng disable-event --jul '*'
5990 .Disable _all_ the event rules of the default channel.
5992 The opt:lttng-disable-event(1):--all-events option is not, like the
5993 opt:lttng-enable-event(1):--all option of man:lttng-enable-event(1), the
5994 equivalent of the event name `*` (wildcard): it disables _all_ the event
5995 rules of a given channel.
5999 $ lttng disable-event --jul --all-events
6003 NOTE: You cannot delete an event rule once you create it.
6007 === Get the status of a tracing session
6009 To get the status of the current tracing session, that is, its
6010 parameters, its channels, event rules, and their attributes:
6012 * Use the man:lttng-status(1) command:
6022 To get the status of any tracing session:
6024 * Use the man:lttng-list(1) command with the tracing session's name:
6029 $ lttng list my-session
6033 Replace `my-session` with the desired tracing session's name.
6036 [[basic-tracing-session-control]]
6037 === Start and stop a tracing session
6039 Once you <<creating-destroying-tracing-sessions,create a tracing
6041 <<enabling-disabling-events,create one or more event rules>>,
6042 you can start and stop the tracers for this tracing session.
6044 To start tracing in the current tracing session:
6046 * Use the man:lttng-start(1) command:
6055 LTTng is very flexible: you can launch user applications before
6056 or after the you start the tracers. The tracers only record the events
6057 if they pass enabled event rules and if they occur while the tracers are
6060 To stop tracing in the current tracing session:
6062 * Use the man:lttng-stop(1) command:
6071 If there were <<channel-overwrite-mode-vs-discard-mode,lost event
6072 records>> or lost sub-buffers since the last time you ran
6073 man:lttng-start(1), warnings are printed when you run the
6074 man:lttng-stop(1) command.
6077 [[enabling-disabling-channels]]
6078 === Create a channel
6080 Once you create a tracing session, you can create a <<channel,channel>>
6081 with the man:lttng-enable-channel(1) command.
6083 Note that LTTng automatically creates a default channel when, for a
6084 given <<domain,tracing domain>>, no channels exist and you
6085 <<enabling-disabling-events,create>> the first event rule. This default
6086 channel is named `channel0` and its attributes are set to reasonable
6087 values. Therefore, you only need to create a channel when you need
6088 non-default attributes.
6090 You specify each non-default channel attribute with a command-line
6091 option when you use the man:lttng-enable-channel(1) command. The
6092 available command-line options are:
6094 [role="growable",cols="asciidoc,asciidoc"]
6095 .Command-line options for the man:lttng-enable-channel(1) command.
6097 |Option |Description
6103 <<channel-overwrite-mode-vs-discard-mode,event loss mode>> instead of
6104 the default _discard_ mode.
6106 |`--buffers-pid` (user space tracing domain only)
6109 Use the per-process <<channel-buffering-schemes,buffering scheme>>
6110 instead of the default per-user buffering scheme.
6112 |+--subbuf-size=__SIZE__+
6115 Allocate sub-buffers of +__SIZE__+ bytes (power of two), for each CPU,
6116 either for each Unix user (default), or for each instrumented process.
6118 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6120 |+--num-subbuf=__COUNT__+
6123 Allocate +__COUNT__+ sub-buffers (power of two), for each CPU, either
6124 for each Unix user (default), or for each instrumented process.
6126 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6128 |+--tracefile-size=__SIZE__+
6131 Set the maximum size of each trace file that this channel writes within
6132 a stream to +__SIZE__+ bytes instead of no maximum.
6134 See <<tracefile-rotation,Trace file count and size>>.
6136 |+--tracefile-count=__COUNT__+
6139 Limit the number of trace files that this channel creates to
6140 +__COUNT__+ channels instead of no limit.
6142 See <<tracefile-rotation,Trace file count and size>>.
6144 |+--switch-timer=__PERIODUS__+
6147 Set the <<channel-switch-timer,switch timer period>>
6148 to +__PERIODUS__+{nbsp}µs.
6150 |+--read-timer=__PERIODUS__+
6153 Set the <<channel-read-timer,read timer period>>
6154 to +__PERIODUS__+{nbsp}µs.
6156 |+--output=__TYPE__+ (Linux kernel tracing domain only)
6159 Set the channel's output type to +__TYPE__+, either `mmap` or `splice`.
6163 You can only create a channel in the Linux kernel and user space
6164 <<domain,tracing domains>>: other tracing domains have their own channel
6165 created on the fly when <<enabling-disabling-events,creating event
6170 Because of a current LTTng limitation, you must create all channels
6171 _before_ you <<basic-tracing-session-control,start tracing>> in a given
6172 tracing session, that is, before the first time you run
6175 Since LTTng automatically creates a default channel when you use the
6176 man:lttng-enable-event(1) command with a specific tracing domain, you
6177 cannot, for example, create a Linux kernel event rule, start tracing,
6178 and then create a user space event rule, because no user space channel
6179 exists yet and it's too late to create one.
6181 For this reason, make sure to configure your channels properly
6182 before starting the tracers for the first time!
6185 The following examples show how you can combine the previous
6186 command-line options to create simple to more complex channels.
6188 .Create a Linux kernel channel with default attributes.
6192 $ lttng enable-channel --kernel my-channel
6196 .Create a user space channel with 4 sub-buffers or 1{nbsp}MiB each, per CPU, per instrumented process.
6200 $ lttng enable-channel --userspace --num-subbuf=4 --subbuf-size=1M \
6201 --buffers-pid my-channel
6205 .Create a Linux kernel channel which rotates 8 trace files of 4{nbsp}MiB each for each stream
6209 $ lttng enable-channel --kernel --tracefile-count=8 \
6210 --tracefile-size=4194304 my-channel
6214 .Create a user space channel in overwrite (or _flight recorder_) mode.
6218 $ lttng enable-channel --userspace --overwrite my-channel
6222 You can <<enabling-disabling-events,create>> the same event rule in
6223 two different channels:
6227 $ lttng enable-event --userspace --channel=my-channel app:tp
6228 $ lttng enable-event --userspace --channel=other-channel app:tp
6231 If both channels are enabled, when a tracepoint named `app:tp` is
6232 reached, LTTng records two events, one for each channel.
6236 === Disable a channel
6238 To disable a specific channel that you <<enabling-disabling-channels,created>>
6239 previously, use the man:lttng-disable-channel(1) command.
6241 .Disable a specific Linux kernel channel.
6245 $ lttng disable-channel --kernel my-channel
6249 The state of a channel precedes the individual states of event rules
6250 attached to it: event rules which belong to a disabled channel, even if
6251 they are enabled, are also considered disabled.
6255 === Add context fields to a channel
6257 Event record fields in trace files provide important information about
6258 events that occured previously, but sometimes some external context may
6259 help you solve a problem faster. Examples of context fields are:
6261 * The **process ID**, **thread ID**, **process name**, and
6262 **process priority** of the thread in which the event occurs.
6263 * The **hostname** of the system on which the event occurs.
6264 * The current values of many possible **performance counters** using
6266 ** CPU cycles, stalled cycles, idle cycles, and the other cycle types.
6268 ** Branch instructions, misses, and loads.
6270 * Any context defined at the application level (supported for the
6271 JUL and log4j <<domain,tracing domains>>).
6273 To get the full list of available context fields, see
6274 `lttng add-context --list`. Some context fields are reserved for a
6275 specific <<domain,tracing domain>> (Linux kernel or user space).
6277 You add context fields to <<channel,channels>>. All the events
6278 that a channel with added context fields records contain those fields.
6280 To add context fields to one or all the channels of a given tracing
6283 * Use the man:lttng-add-context(1) command.
6285 .Add context fields to all the channels of the current tracing session.
6287 The following command line adds the virtual process identifier and
6288 the per-thread CPU cycles count fields to all the user space channels
6289 of the current tracing session.
6293 $ lttng add-context --userspace --type=vpid --type=perf:thread:cpu-cycles
6297 .Add performance counter context fields by raw ID
6299 See man:lttng-add-context(1) for the exact format of the context field
6300 type, which is partly compatible with the format used in
6305 $ lttng add-context --userspace --type=perf:thread:raw:r0110:test
6306 $ lttng add-context --kernel --type=perf:cpu:raw:r0013c:x86unhalted
6310 .Add a context field to a specific channel.
6312 The following command line adds the thread identifier context field
6313 to the Linux kernel channel named `my-channel` in the current
6318 $ lttng add-context --kernel --channel=my-channel --type=tid
6322 .Add an application-specific context field to a specific channel.
6324 The following command line adds the `cur_msg_id` context field of the
6325 `retriever` context retriever for all the instrumented
6326 <<java-application,Java applications>> recording <<event,event records>>
6327 in the channel named `my-channel`:
6331 $ lttng add-context --kernel --channel=my-channel \
6332 --type='$app:retriever:cur_msg_id'
6335 IMPORTANT: Make sure to always quote the `$` character when you
6336 use man:lttng-add-context(1) from a shell.
6339 NOTE: You cannot remove context fields from a channel once you add it.
6344 === Track process IDs
6346 It's often useful to allow only specific process IDs (PIDs) to emit
6347 events. For example, you may wish to record all the system calls made by
6348 a given process (à la http://linux.die.net/man/1/strace[strace]).
6350 The man:lttng-track(1) and man:lttng-untrack(1) commands serve this
6351 purpose. Both commands operate on a whitelist of process IDs. You _add_
6352 entries to this whitelist with the man:lttng-track(1) command and remove
6353 entries with the man:lttng-untrack(1) command. Any process which has one
6354 of the PIDs in the whitelist is allowed to emit LTTng events which pass
6355 an enabled <<event,event rule>>.
6357 NOTE: The PID tracker tracks the _numeric process IDs_. Should a
6358 process with a given tracked ID exit and another process be given this
6359 ID, then the latter would also be allowed to emit events.
6361 .Track and untrack process IDs.
6363 For the sake of the following example, assume the target system has 16
6367 <<creating-destroying-tracing-sessions,create a tracing session>>,
6368 the whitelist contains all the possible PIDs:
6371 .All PIDs are tracked.
6372 image::track-all.png[]
6374 When the whitelist is full and you use the man:lttng-track(1) command to
6375 specify some PIDs to track, LTTng first clears the whitelist, then it
6376 tracks the specific PIDs. After:
6380 $ lttng track --pid=3,4,7,10,13
6386 .PIDs 3, 4, 7, 10, and 13 are tracked.
6387 image::track-3-4-7-10-13.png[]
6389 You can add more PIDs to the whitelist afterwards:
6393 $ lttng track --pid=1,15,16
6399 .PIDs 1, 15, and 16 are added to the whitelist.
6400 image::track-1-3-4-7-10-13-15-16.png[]
6402 The man:lttng-untrack(1) command removes entries from the PID tracker's
6403 whitelist. Given the previous example, the following command:
6407 $ lttng untrack --pid=3,7,10,13
6410 leads to this whitelist:
6413 .PIDs 3, 7, 10, and 13 are removed from the whitelist.
6414 image::track-1-4-15-16.png[]
6416 LTTng can track all possible PIDs again using the opt:track(1):--all
6421 $ lttng track --pid --all
6424 The result is, again:
6427 .All PIDs are tracked.
6428 image::track-all.png[]
6431 .Track only specific PIDs
6433 A very typical use case with PID tracking is to start with an empty
6434 whitelist, then <<basic-tracing-session-control,start the tracers>>, and
6435 then add PIDs manually while tracers are active. You can accomplish this
6436 by using the opt:lttng-untrack(1):--all option of the
6437 man:lttng-untrack(1) command to clear the whitelist after you
6438 <<creating-destroying-tracing-sessions,create a tracing session>>:
6442 $ lttng untrack --pid --all
6448 .No PIDs are tracked.
6449 image::untrack-all.png[]
6451 If you trace with this whitelist configuration, the tracer records no
6452 events for this <<domain,tracing domain>> because no processes are
6453 tracked. You can use the man:lttng-track(1) command as usual to track
6454 specific PIDs, for example:
6458 $ lttng track --pid=6,11
6464 .PIDs 6 and 11 are tracked.
6465 image::track-6-11.png[]
6470 [[saving-loading-tracing-session]]
6471 === Save and load tracing session configurations
6473 Configuring a <<tracing-session,tracing session>> can be long. Some of
6474 the tasks involved are:
6476 * <<enabling-disabling-channels,Create channels>> with
6477 specific attributes.
6478 * <<adding-context,Add context fields>> to specific channels.
6479 * <<enabling-disabling-events,Create event rules>> with specific log
6480 level and filter conditions.
6482 If you use LTTng to solve real world problems, chances are you have to
6483 record events using the same tracing session setup over and over,
6484 modifying a few variables each time in your instrumented program
6485 or environment. To avoid constant tracing session reconfiguration,
6486 the man:lttng(1) command-line tool can save and load tracing session
6487 configurations to/from XML files.
6489 To save a given tracing session configuration:
6491 * Use the man:lttng-save(1) command:
6496 $ lttng save my-session
6500 Replace `my-session` with the name of the tracing session to save.
6502 LTTng saves tracing session configurations to
6503 dir:{$LTTNG_HOME/.lttng/sessions} by default. Note that the
6504 env:LTTNG_HOME environment variable defaults to `$HOME` if not set. Use
6505 the opt:lttng-save(1):--output-path option to change this destination
6508 LTTng saves all configuration parameters, for example:
6510 * The tracing session name.
6511 * The trace data output path.
6512 * The channels with their state and all their attributes.
6513 * The context fields you added to channels.
6514 * The event rules with their state, log level and filter conditions.
6516 To load a tracing session:
6518 * Use the man:lttng-load(1) command:
6523 $ lttng load my-session
6527 Replace `my-session` with the name of the tracing session to load.
6529 When LTTng loads a configuration, it restores your saved tracing session
6530 as if you just configured it manually.
6532 See man:lttng(1) for the complete list of command-line options. You
6533 can also save and load all many sessions at a time, and decide in which
6534 directory to output the XML files.
6537 [[sending-trace-data-over-the-network]]
6538 === Send trace data over the network
6540 LTTng can send the recorded trace data to a remote system over the
6541 network instead of writing it to the local file system.
6543 To send the trace data over the network:
6545 . On the _remote_ system (which can also be the target system),
6546 start an LTTng <<lttng-relayd,relay daemon>> (man:lttng-relayd(8)):
6555 . On the _target_ system, create a tracing session configured to
6556 send trace data over the network:
6561 $ lttng create my-session --set-url=net://remote-system
6565 Replace `remote-system` by the host name or IP address of the
6566 remote system. See man:lttng-create(1) for the exact URL format.
6568 . On the target system, use the man:lttng(1) command-line tool as usual.
6569 When tracing is active, the target's consumer daemon sends sub-buffers
6570 to the relay daemon running on the remote system instead of flushing
6571 them to the local file system. The relay daemon writes the received
6572 packets to the local file system.
6574 The relay daemon writes trace files to
6575 +$LTTNG_HOME/lttng-traces/__hostname__/__session__+ by default, where
6576 +__hostname__+ is the host name of the target system and +__session__+
6577 is the tracing session name. Note that the env:LTTNG_HOME environment
6578 variable defaults to `$HOME` if not set. Use the
6579 opt:lttng-relayd(8):--output option of man:lttng-relayd(8) to write
6580 trace files to another base directory.
6585 === View events as LTTng emits them (noch:{LTTng} live)
6587 LTTng live is a network protocol implemented by the <<lttng-relayd,relay
6588 daemon>> (man:lttng-relayd(8)) to allow compatible trace viewers to
6589 display events as LTTng emits them on the target system while tracing is
6592 The relay daemon creates a _tee_: it forwards the trace data to both
6593 the local file system and to connected live viewers:
6596 .The relay daemon creates a _tee_, forwarding the trace data to both trace files and a connected live viewer.
6601 . On the _target system_, create a <<tracing-session,tracing session>>
6607 $ lttng create my-session --live
6611 This spawns a local relay daemon.
6613 . Start the live viewer and configure it to connect to the relay
6614 daemon. For example, with http://diamon.org/babeltrace[Babeltrace]:
6619 $ babeltrace --input-format=lttng-live \
6620 net://localhost/host/hostname/my-session
6627 * `hostname` with the host name of the target system.
6628 * `my-session` with the name of the tracing session to view.
6631 . Configure the tracing session as usual with the man:lttng(1)
6632 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6634 You can list the available live tracing sessions with Babeltrace:
6638 $ babeltrace --input-format=lttng-live net://localhost
6641 You can start the relay daemon on another system. In this case, you need
6642 to specify the relay daemon's URL when you create the tracing session
6643 with the opt:lttng-create(1):--set-url option. You also need to replace
6644 `localhost` in the procedure above with the host name of the system on
6645 which the relay daemon is running.
6647 See man:lttng-create(1) and man:lttng-relayd(8) for the complete list of
6648 command-line options.
6652 [[taking-a-snapshot]]
6653 === Take a snapshot of the current sub-buffers of a tracing session
6655 The normal behavior of LTTng is to append full sub-buffers to growing
6656 trace data files. This is ideal to keep a full history of the events
6657 that occurred on the target system, but it can
6658 represent too much data in some situations. For example, you may wish
6659 to trace your application continuously until some critical situation
6660 happens, in which case you only need the latest few recorded
6661 events to perform the desired analysis, not multi-gigabyte trace files.
6663 With the man:lttng-snapshot(1) command, you can take a snapshot of the
6664 current sub-buffers of a given <<tracing-session,tracing session>>.
6665 LTTng can write the snapshot to the local file system or send it over
6670 . Create a tracing session in _snapshot mode_:
6675 $ lttng create my-session --snapshot
6679 The <<channel-overwrite-mode-vs-discard-mode,event loss mode>> of
6680 <<channel,channels>> created in this mode is automatically set to
6681 _overwrite_ (flight recorder mode).
6683 . Configure the tracing session as usual with the man:lttng(1)
6684 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6686 . **Optional**: When you need to take a snapshot,
6687 <<basic-tracing-session-control,stop tracing>>.
6689 You can take a snapshot when the tracers are active, but if you stop
6690 them first, you are sure that the data in the sub-buffers does not
6691 change before you actually take the snapshot.
6698 $ lttng snapshot record --name=my-first-snapshot
6702 LTTng writes the current sub-buffers of all the current tracing
6703 session's channels to trace files on the local file system. Those trace
6704 files have `my-first-snapshot` in their name.
6706 There is no difference between the format of a normal trace file and the
6707 format of a snapshot: viewers of LTTng traces also support LTTng
6710 By default, LTTng writes snapshot files to the path shown by
6711 `lttng snapshot list-output`. You can change this path or decide to send
6712 snapshots over the network using either:
6714 . An output path or URL that you specify when you create the
6716 . An snapshot output path or URL that you add using
6717 `lttng snapshot add-output`
6718 . An output path or URL that you provide directly to the
6719 `lttng snapshot record` command.
6721 Method 3 overrides method 2, which overrides method 1. When you
6722 specify a URL, a relay daemon must listen on a remote system (see
6723 <<sending-trace-data-over-the-network,Send trace data over the network>>).
6728 === Use the machine interface
6730 With any command of the man:lttng(1) command-line tool, you can set the
6731 opt:lttng(1):--mi option to `xml` (before the command name) to get an
6732 XML machine interface output, for example:
6736 $ lttng --mi=xml enable-event --kernel --syscall open
6739 A schema definition (XSD) is
6740 https://github.com/lttng/lttng-tools/blob/stable-2.9/src/common/mi-lttng-3.0.xsd[available]
6741 to ease the integration with external tools as much as possible.
6745 [[metadata-regenerate]]
6746 === Regenerate the metadata of an LTTng trace
6748 An LTTng trace, which is a http://diamon.org/ctf[CTF] trace, has both
6749 data stream files and a metadata file. This metadata file contains,
6750 amongst other things, information about the offset of the clock sources
6751 used to timestamp <<event,event records>> when tracing.
6753 If, once a <<tracing-session,tracing session>> is
6754 <<basic-tracing-session-control,started>>, a major
6755 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
6756 happens, the trace's clock offset also needs to be updated. You
6757 can use the `metadata` item of the man:lttng-regenerate(1) command
6760 The main use case of this command is to allow a system to boot with
6761 an incorrect wall time and trace it with LTTng before its wall time
6762 is corrected. Once the system is known to be in a state where its
6763 wall time is correct, it can run `lttng regenerate metadata`.
6765 To regenerate the metadata of an LTTng trace:
6767 * Use the `metadata` item of the man:lttng-regenerate(1) command:
6772 $ lttng regenerate metadata
6778 `lttng regenerate metadata` has the following limitations:
6780 * Tracing session <<creating-destroying-tracing-sessions,created>>
6782 * User space <<channel,channels>>, if any, are using
6783 <<channel-buffering-schemes,per-user buffering>>.
6788 [[regenerate-statedump]]
6789 === Regenerate the state dump of a tracing session
6791 The LTTng kernel and user space tracers generate state dump
6792 <<event,event records>> when the application starts or when you
6793 <<basic-tracing-session-control,start a tracing session>>. An analysis
6794 can use the state dump event records to set an initial state before it
6795 builds the rest of the state from the following event records.
6796 http://tracecompass.org/[Trace Compass] is a notable example of an
6797 application which uses the state dump of an LTTng trace.
6799 When you <<taking-a-snapshot,take a snapshot>>, it's possible that the
6800 state dump event records are not included in the snapshot because they
6801 were recorded to a sub-buffer that has been consumed or overwritten
6804 You can use the `lttng regenerate statedump` command to emit the state
6805 dump event records again.
6807 To regenerate the state dump of the current tracing session, provided
6808 create it in snapshot mode, before you take a snapshot:
6810 . Use the `statedump` item of the man:lttng-regenerate(1) command:
6815 $ lttng regenerate statedump
6819 . <<basic-tracing-session-control,Stop the tracing session>>:
6828 . <<taking-a-snapshot,Take a snapshot>>:
6833 $ lttng snapshot record --name=my-snapshot
6837 Depending on the event throughput, you should run steps 1 and 2
6838 as closely as possible.
6840 NOTE: To record the state dump events, you need to
6841 <<enabling-disabling-events,create event rules>> which enable them.
6842 LTTng-UST state dump tracepoints start with `lttng_ust_statedump:`.
6843 LTTng-modules state dump tracepoints start with `lttng_statedump_`.
6847 [[persistent-memory-file-systems]]
6848 === Record trace data on persistent memory file systems
6850 https://en.wikipedia.org/wiki/Non-volatile_random-access_memory[Non-volatile random-access memory]
6851 (NVRAM) is random-access memory that retains its information when power
6852 is turned off (non-volatile). Systems with such memory can store data
6853 structures in RAM and retrieve them after a reboot, without flushing
6854 to typical _storage_.
6856 Linux supports NVRAM file systems thanks to either
6857 http://pramfs.sourceforge.net/[PRAMFS] or
6858 https://www.kernel.org/doc/Documentation/filesystems/dax.txt[DAX]{nbsp}+{nbsp}http://lkml.iu.edu/hypermail/linux/kernel/1504.1/03463.html[pmem]
6859 (requires Linux 4.1+).
6861 This section does not describe how to operate such file systems;
6862 we assume that you have a working persistent memory file system.
6864 When you create a <<tracing-session,tracing session>>, you can specify
6865 the path of the shared memory holding the sub-buffers. If you specify a
6866 location on an NVRAM file system, then you can retrieve the latest
6867 recorded trace data when the system reboots after a crash.
6869 To record trace data on a persistent memory file system and retrieve the
6870 trace data after a system crash:
6872 . Create a tracing session with a sub-buffer shared memory path located
6873 on an NVRAM file system:
6878 $ lttng create my-session --shm-path=/path/to/shm
6882 . Configure the tracing session as usual with the man:lttng(1)
6883 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6885 . After a system crash, use the man:lttng-crash(1) command-line tool to
6886 view the trace data recorded on the NVRAM file system:
6891 $ lttng-crash /path/to/shm
6895 The binary layout of the ring buffer files is not exactly the same as
6896 the trace files layout. This is why you need to use man:lttng-crash(1)
6897 instead of your preferred trace viewer directly.
6899 To convert the ring buffer files to LTTng trace files:
6901 * Use the opt:lttng-crash(1):--extract option of man:lttng-crash(1):
6906 $ lttng-crash --extract=/path/to/trace /path/to/shm
6914 [[lttng-modules-ref]]
6915 === noch:{LTTng-modules}
6919 [[lttng-tracepoint-enum]]
6920 ==== `LTTNG_TRACEPOINT_ENUM()` usage
6922 Use the `LTTNG_TRACEPOINT_ENUM()` macro to define an enumeration:
6926 LTTNG_TRACEPOINT_ENUM(name, TP_ENUM_VALUES(entries))
6931 * `name` with the name of the enumeration (C identifier, unique
6932 amongst all the defined enumerations).
6933 * `entries` with a list of enumeration entries.
6935 The available enumeration entry macros are:
6937 +ctf_enum_value(__name__, __value__)+::
6938 Entry named +__name__+ mapped to the integral value +__value__+.
6940 +ctf_enum_range(__name__, __begin__, __end__)+::
6941 Entry named +__name__+ mapped to the range of integral values between
6942 +__begin__+ (included) and +__end__+ (included).
6944 +ctf_enum_auto(__name__)+::
6945 Entry named +__name__+ mapped to the integral value following the
6946 last mapping's value.
6948 The last value of a `ctf_enum_value()` entry is its +__value__+
6951 The last value of a `ctf_enum_range()` entry is its +__end__+ parameter.
6953 If `ctf_enum_auto()` is the first entry in the list, its integral
6956 Use the `ctf_enum()` <<lttng-modules-tp-fields,field definition macro>>
6957 to use a defined enumeration as a tracepoint field.
6959 .Define an enumeration with `LTTNG_TRACEPOINT_ENUM()`.
6963 LTTNG_TRACEPOINT_ENUM(
6966 ctf_enum_auto("AUTO: EXPECT 0")
6967 ctf_enum_value("VALUE: 23", 23)
6968 ctf_enum_value("VALUE: 27", 27)
6969 ctf_enum_auto("AUTO: EXPECT 28")
6970 ctf_enum_range("RANGE: 101 TO 303", 101, 303)
6971 ctf_enum_auto("AUTO: EXPECT 304")
6979 [[lttng-modules-tp-fields]]
6980 ==== Tracepoint fields macros (for `TP_FIELDS()`)
6982 [[tp-fast-assign]][[tp-struct-entry]]The available macros to define
6983 tracepoint fields, which must be listed within `TP_FIELDS()` in
6984 `LTTNG_TRACEPOINT_EVENT()`, are:
6986 [role="func-desc growable",cols="asciidoc,asciidoc"]
6987 .Available macros to define LTTng-modules tracepoint fields
6989 |Macro |Description and parameters
6992 +ctf_integer(__t__, __n__, __e__)+
6994 +ctf_integer_nowrite(__t__, __n__, __e__)+
6996 +ctf_user_integer(__t__, __n__, __e__)+
6998 +ctf_user_integer_nowrite(__t__, __n__, __e__)+
7000 Standard integer, displayed in base 10.
7003 Integer C type (`int`, `long`, `size_t`, ...).
7009 Argument expression.
7012 +ctf_integer_hex(__t__, __n__, __e__)+
7014 +ctf_user_integer_hex(__t__, __n__, __e__)+
7016 Standard integer, displayed in base 16.
7025 Argument expression.
7027 |+ctf_integer_oct(__t__, __n__, __e__)+
7029 Standard integer, displayed in base 8.
7038 Argument expression.
7041 +ctf_integer_network(__t__, __n__, __e__)+
7043 +ctf_user_integer_network(__t__, __n__, __e__)+
7045 Integer in network byte order (big-endian), displayed in base 10.
7054 Argument expression.
7057 +ctf_integer_network_hex(__t__, __n__, __e__)+
7059 +ctf_user_integer_network_hex(__t__, __n__, __e__)+
7061 Integer in network byte order, displayed in base 16.
7070 Argument expression.
7073 +ctf_enum(__N__, __t__, __n__, __e__)+
7075 +ctf_enum_nowrite(__N__, __t__, __n__, __e__)+
7077 +ctf_user_enum(__N__, __t__, __n__, __e__)+
7079 +ctf_user_enum_nowrite(__N__, __t__, __n__, __e__)+
7084 Name of a <<lttng-tracepoint-enum,previously defined enumeration>>.
7087 Integer C type (`int`, `long`, `size_t`, ...).
7093 Argument expression.
7096 +ctf_string(__n__, __e__)+
7098 +ctf_string_nowrite(__n__, __e__)+
7100 +ctf_user_string(__n__, __e__)+
7102 +ctf_user_string_nowrite(__n__, __e__)+
7104 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
7110 Argument expression.
7113 +ctf_array(__t__, __n__, __e__, __s__)+
7115 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
7117 +ctf_user_array(__t__, __n__, __e__, __s__)+
7119 +ctf_user_array_nowrite(__t__, __n__, __e__, __s__)+
7121 Statically-sized array of integers.
7124 Array element C type.
7130 Argument expression.
7136 +ctf_array_bitfield(__t__, __n__, __e__, __s__)+
7138 +ctf_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7140 +ctf_user_array_bitfield(__t__, __n__, __e__, __s__)+
7142 +ctf_user_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7144 Statically-sized array of bits.
7146 The type of +__e__+ must be an integer type. +__s__+ is the number
7147 of elements of such type in +__e__+, not the number of bits.
7150 Array element C type.
7156 Argument expression.
7162 +ctf_array_text(__t__, __n__, __e__, __s__)+
7164 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
7166 +ctf_user_array_text(__t__, __n__, __e__, __s__)+
7168 +ctf_user_array_text_nowrite(__t__, __n__, __e__, __s__)+
7170 Statically-sized array, printed as text.
7172 The string does not need to be null-terminated.
7175 Array element C type (always `char`).
7181 Argument expression.
7187 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
7189 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7191 +ctf_user_sequence(__t__, __n__, __e__, __T__, __E__)+
7193 +ctf_user_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7195 Dynamically-sized array of integers.
7197 The type of +__E__+ must be unsigned.
7200 Array element C type.
7206 Argument expression.
7209 Length expression C type.
7215 +ctf_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7217 +ctf_user_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7219 Dynamically-sized array of integers, displayed in base 16.
7221 The type of +__E__+ must be unsigned.
7224 Array element C type.
7230 Argument expression.
7233 Length expression C type.
7238 |+ctf_sequence_network(__t__, __n__, __e__, __T__, __E__)+
7240 Dynamically-sized array of integers in network byte order (big-endian),
7241 displayed in base 10.
7243 The type of +__E__+ must be unsigned.
7246 Array element C type.
7252 Argument expression.
7255 Length expression C type.
7261 +ctf_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7263 +ctf_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7265 +ctf_user_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7267 +ctf_user_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7269 Dynamically-sized array of bits.
7271 The type of +__e__+ must be an integer type. +__s__+ is the number
7272 of elements of such type in +__e__+, not the number of bits.
7274 The type of +__E__+ must be unsigned.
7277 Array element C type.
7283 Argument expression.
7286 Length expression C type.
7292 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7294 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7296 +ctf_user_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7298 +ctf_user_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7300 Dynamically-sized array, displayed as text.
7302 The string does not need to be null-terminated.
7304 The type of +__E__+ must be unsigned.
7306 The behaviour is undefined if +__e__+ is `NULL`.
7309 Sequence element C type (always `char`).
7315 Argument expression.
7318 Length expression C type.
7324 Use the `_user` versions when the argument expression, `e`, is
7325 a user space address. In the cases of `ctf_user_integer*()` and
7326 `ctf_user_float*()`, `&e` must be a user space address, thus `e` must
7329 The `_nowrite` versions omit themselves from the session trace, but are
7330 otherwise identical. This means the `_nowrite` fields won't be written
7331 in the recorded trace. Their primary purpose is to make some
7332 of the event context available to the
7333 <<enabling-disabling-events,event filters>> without having to
7334 commit the data to sub-buffers.
7340 Terms related to LTTng and to tracing in general:
7343 The http://diamon.org/babeltrace[Babeltrace] project, which includes
7344 the cmd:babeltrace command, some libraries, and Python bindings.
7346 <<channel-buffering-schemes,buffering scheme>>::
7347 A layout of sub-buffers applied to a given channel.
7349 <<channel,channel>>::
7350 An entity which is responsible for a set of ring buffers.
7352 <<event,Event rules>> are always attached to a specific channel.
7355 A reference of time for a tracer.
7357 <<lttng-consumerd,consumer daemon>>::
7358 A process which is responsible for consuming the full sub-buffers
7359 and write them to a file system or send them over the network.
7361 <<channel-overwrite-mode-vs-discard-mode,discard mode>>:: The event loss
7362 mode in which the tracer _discards_ new event records when there's no
7363 sub-buffer space left to store them.
7366 The consequence of the execution of an instrumentation
7367 point, like a tracepoint that you manually place in some source code,
7368 or a Linux kernel KProbe.
7370 An event is said to _occur_ at a specific time. Different actions can
7371 be taken upon the occurrence of an event, like record the event's payload
7374 <<channel-overwrite-mode-vs-discard-mode,event loss mode>>::
7375 The mechanism by which event records of a given channel are lost
7376 (not recorded) when there is no sub-buffer space left to store them.
7378 [[def-event-name]]event name::
7379 The name of an event, which is also the name of the event record.
7380 This is also called the _instrumentation point name_.
7383 A record, in a trace, of the payload of an event which occured.
7385 <<event,event rule>>::
7386 Set of conditions which must be satisfied for one or more occuring
7387 events to be recorded.
7389 `java.util.logging`::
7391 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[core logging facilities].
7393 <<instrumenting,instrumentation>>::
7394 The use of LTTng probes to make a piece of software traceable.
7396 instrumentation point::
7397 A point in the execution path of a piece of software that, when
7398 reached by this execution, can emit an event.
7400 instrumentation point name::
7401 See _<<def-event-name,event name>>_.
7404 A http://logging.apache.org/log4j/1.2/[logging library] for Java
7405 developed by the Apache Software Foundation.
7408 Level of severity of a log statement or user space
7409 instrumentation point.
7412 The _Linux Trace Toolkit: next generation_ project.
7414 <<lttng-cli,cmd:lttng>>::
7415 A command-line tool provided by the LTTng-tools project which you
7416 can use to send and receive control messages to and from a
7420 The https://github.com/lttng/lttng-analyses[LTTng analyses] project,
7421 which is a set of analyzing programs that are used to obtain a
7422 higher level view of an LTTng trace.
7424 cmd:lttng-consumerd::
7425 The name of the consumer daemon program.
7428 A utility provided by the LTTng-tools project which can convert
7429 ring buffer files (usually
7430 <<persistent-memory-file-systems,saved on a persistent memory file system>>)
7433 LTTng Documentation::
7436 <<lttng-live,LTTng live>>::
7437 A communication protocol between the relay daemon and live viewers
7438 which makes it possible to see events "live", as they are received by
7441 <<lttng-modules,LTTng-modules>>::
7442 The https://github.com/lttng/lttng-modules[LTTng-modules] project,
7443 which contains the Linux kernel modules to make the Linux kernel
7444 instrumentation points available for LTTng tracing.
7447 The name of the relay daemon program.
7449 cmd:lttng-sessiond::
7450 The name of the session daemon program.
7453 The https://github.com/lttng/lttng-tools[LTTng-tools] project, which
7454 contains the various programs and libraries used to
7455 <<controlling-tracing,control tracing>>.
7457 <<lttng-ust,LTTng-UST>>::
7458 The https://github.com/lttng/lttng-ust[LTTng-UST] project, which
7459 contains libraries to instrument user applications.
7461 <<lttng-ust-agents,LTTng-UST Java agent>>::
7462 A Java package provided by the LTTng-UST project to allow the
7463 LTTng instrumentation of `java.util.logging` and Apache log4j 1.2
7466 <<lttng-ust-agents,LTTng-UST Python agent>>::
7467 A Python package provided by the LTTng-UST project to allow the
7468 LTTng instrumentation of Python logging statements.
7470 <<channel-overwrite-mode-vs-discard-mode,overwrite mode>>::
7471 The event loss mode in which new event records overwrite older
7472 event records when there's no sub-buffer space left to store them.
7474 <<channel-buffering-schemes,per-process buffering>>::
7475 A buffering scheme in which each instrumented process has its own
7476 sub-buffers for a given user space channel.
7478 <<channel-buffering-schemes,per-user buffering>>::
7479 A buffering scheme in which all the processes of a Unix user share the
7480 same sub-buffer for a given user space channel.
7482 <<lttng-relayd,relay daemon>>::
7483 A process which is responsible for receiving the trace data sent by
7484 a distant consumer daemon.
7487 A set of sub-buffers.
7489 <<lttng-sessiond,session daemon>>::
7490 A process which receives control commands from you and orchestrates
7491 the tracers and various LTTng daemons.
7493 <<taking-a-snapshot,snapshot>>::
7494 A copy of the current data of all the sub-buffers of a given tracing
7495 session, saved as trace files.
7498 One part of an LTTng ring buffer which contains event records.
7501 The time information attached to an event when it is emitted.
7504 A set of files which are the concatenations of one or more
7505 flushed sub-buffers.
7508 The action of recording the events emitted by an application
7509 or by a system, or to initiate such recording by controlling
7513 The http://tracecompass.org[Trace Compass] project and application.
7516 An instrumentation point using the tracepoint mechanism of the Linux
7517 kernel or of LTTng-UST.
7519 tracepoint definition::
7520 The definition of a single tracepoint.
7523 The name of a tracepoint.
7525 tracepoint provider::
7526 A set of functions providing tracepoints to an instrumented user
7529 Not to be confused with a _tracepoint provider package_: many tracepoint
7530 providers can exist within a tracepoint provider package.
7532 tracepoint provider package::
7533 One or more tracepoint providers compiled as an object file or as
7537 A software which records emitted events.
7539 <<domain,tracing domain>>::
7540 A namespace for event sources.
7542 <<tracing-group,tracing group>>::
7543 The Unix group in which a Unix user can be to be allowed to trace the
7546 <<tracing-session,tracing session>>::
7547 A stateful dialogue between you and a <<lttng-sessiond,session
7551 An application running in user space, as opposed to a Linux kernel
7552 module, for example.