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 _KeKriek_. From
78 http://brasseriedunham.com/[Brasserie Dunham], the _**KeKriek**_ is a
79 sour mashed golden wheat ale fermented with local sour cherries from
80 Tougas orchards. Fresh sweet cherry notes with some tartness, lively
81 carbonation with a dry finish.
83 New features and changes in LTTng{nbsp}{revision}:
85 * **Tracing control**:
86 ** You can put more than one wildcard special character (`*`), and not
87 only at the end, when you <<enabling-disabling-events,create an event
88 rule>>, in both the instrumentation point name and the literal
90 link:/man/1/lttng-enable-event/v{revision}/#doc-filter-syntax[filter expressions]:
95 # lttng enable-event --kernel 'x86_*_local_timer_*' \
96 --filter='name == "*a*b*c*d*e" && count >= 23'
103 $ lttng enable-event --userspace '*_my_org:*msg*'
107 ** New trigger and notification API for
108 <<liblttng-ctl-lttng,`liblttng-ctl`>>. This new subsystem allows you
109 to register triggers which emit a notification when a given
110 condition is satisfied. As of LTTng{nbsp}{revision}, only
111 <<channel,channel>> buffer usage conditions are available.
112 Documentation is available in the
113 https://github.com/lttng/lttng-tools/tree/stable-{revision}/include/lttng[`liblttng-ctl`
115 <<notif-trigger-api,Get notified when a channel's buffer usage is too
118 ** You can now embed the whole textual LTTng-tools man pages into the
119 executables at build time with the `--enable-embedded-help`
120 configuration option. Thanks to this option, you don't need the
121 http://www.methods.co.nz/asciidoc/[AsciiDoc] and
122 https://directory.fsf.org/wiki/Xmlto[xmlto] tools at build time, and
123 a manual pager at run time, to get access to this documentation.
125 * **User space tracing**:
126 ** New blocking mode: an LTTng-UST tracepoint can now block until
127 <<channel,sub-buffer>> space is available instead of discarding event
128 records in <<channel-overwrite-mode-vs-discard-mode,discard mode>>.
129 With this feature, you can be sure that no event records are
130 discarded during your application's execution at the expense of
133 For example, the following command lines create a user space tracing
134 channel with an infinite blocking timeout and run an application
135 instrumented with LTTng-UST which is explicitly allowed to block:
141 $ lttng enable-channel --userspace --blocking-timeout=inf blocking-channel
142 $ lttng enable-event --userspace --channel=blocking-channel --all
144 $ LTTNG_UST_ALLOW_BLOCKING=1 my-app
148 See the complete <<blocking-timeout-example,blocking timeout example>>.
150 * **Linux kernel tracing**:
151 ** Linux 4.10, 4.11, and 4.12 support.
152 ** The thread state dump events recorded by LTTng-modules now contain
153 the task's CPU identifier. This improves the precision of the
154 scheduler model for analyses.
155 ** Extended man:socketpair(2) system call tracing data.
161 What is LTTng? As its name suggests, the _Linux Trace Toolkit: next
162 generation_ is a modern toolkit for tracing Linux systems and
163 applications. So your first question might be:
170 As the history of software engineering progressed and led to what
171 we now take for granted--complex, numerous and
172 interdependent software applications running in parallel on
173 sophisticated operating systems like Linux--the authors of such
174 components, software developers, began feeling a natural
175 urge to have tools that would ensure the robustness and good performance
176 of their masterpieces.
178 One major achievement in this field is, inarguably, the
179 https://www.gnu.org/software/gdb/[GNU debugger (GDB)],
180 an essential tool for developers to find and fix bugs. But even the best
181 debugger won't help make your software run faster, and nowadays, faster
182 software means either more work done by the same hardware, or cheaper
183 hardware for the same work.
185 A _profiler_ is often the tool of choice to identify performance
186 bottlenecks. Profiling is suitable to identify _where_ performance is
187 lost in a given software. The profiler outputs a profile, a statistical
188 summary of observed events, which you may use to discover which
189 functions took the most time to execute. However, a profiler won't
190 report _why_ some identified functions are the bottleneck. Bottlenecks
191 might only occur when specific conditions are met, conditions that are
192 sometimes impossible to capture by a statistical profiler, or impossible
193 to reproduce with an application altered by the overhead of an
194 event-based profiler. For a thorough investigation of software
195 performance issues, a history of execution is essential, with the
196 recorded values of variables and context fields you choose, and
197 with as little influence as possible on the instrumented software. This
198 is where tracing comes in handy.
200 _Tracing_ is a technique used to understand what goes on in a running
201 software system. The software used for tracing is called a _tracer_,
202 which is conceptually similar to a tape recorder. When recording,
203 specific instrumentation points placed in the software source code
204 generate events that are saved on a giant tape: a _trace_ file. You
205 can trace user applications and the operating system at the same time,
206 opening the possibility of resolving a wide range of problems that would
207 otherwise be extremely challenging.
209 Tracing is often compared to _logging_. However, tracers and loggers are
210 two different tools, serving two different purposes. Tracers are
211 designed to record much lower-level events that occur much more
212 frequently than log messages, often in the range of thousands per
213 second, with very little execution overhead. Logging is more appropriate
214 for a very high-level analysis of less frequent events: user accesses,
215 exceptional conditions (errors and warnings, for example), database
216 transactions, instant messaging communications, and such. Simply put,
217 logging is one of the many use cases that can be satisfied with tracing.
219 The list of recorded events inside a trace file can be read manually
220 like a log file for the maximum level of detail, but it is generally
221 much more interesting to perform application-specific analyses to
222 produce reduced statistics and graphs that are useful to resolve a
223 given problem. Trace viewers and analyzers are specialized tools
226 In the end, this is what LTTng is: a powerful, open source set of
227 tools to trace the Linux kernel and user applications at the same time.
228 LTTng is composed of several components actively maintained and
229 developed by its link:/community/#where[community].
232 [[lttng-alternatives]]
233 === Alternatives to noch:{LTTng}
235 Excluding proprietary solutions, a few competing software tracers
238 * https://github.com/dtrace4linux/linux[dtrace4linux] is a port of
239 Sun Microsystems's DTrace to Linux. The cmd:dtrace tool interprets
240 user scripts and is responsible for loading code into the
241 Linux kernel for further execution and collecting the outputted data.
242 * https://en.wikipedia.org/wiki/Berkeley_Packet_Filter[eBPF] is a
243 subsystem in the Linux kernel in which a virtual machine can execute
244 programs passed from the user space to the kernel. You can attach
245 such programs to tracepoints and KProbes thanks to a system call, and
246 they can output data to the user space when executed thanks to
247 different mechanisms (pipe, VM register values, and eBPF maps, to name
249 * https://www.kernel.org/doc/Documentation/trace/ftrace.txt[ftrace]
250 is the de facto function tracer of the Linux kernel. Its user
251 interface is a set of special files in sysfs.
252 * https://perf.wiki.kernel.org/[perf] is
253 a performance analyzing tool for Linux which supports hardware
254 performance counters, tracepoints, as well as other counters and
255 types of probes. perf's controlling utility is the cmd:perf command
257 * http://linux.die.net/man/1/strace[strace]
258 is a command-line utility which records system calls made by a
259 user process, as well as signal deliveries and changes of process
260 state. strace makes use of https://en.wikipedia.org/wiki/Ptrace[ptrace]
261 to fulfill its function.
262 * http://www.sysdig.org/[sysdig], like SystemTap, uses scripts to
263 analyze Linux kernel events. You write scripts, or _chisels_ in
264 sysdig's jargon, in Lua and sysdig executes them while the system is
265 being traced or afterwards. sysdig's interface is the cmd:sysdig
266 command-line tool as well as the curses-based cmd:csysdig tool.
267 * https://sourceware.org/systemtap/[SystemTap] is a Linux kernel and
268 user space tracer which uses custom user scripts to produce plain text
269 traces. SystemTap converts the scripts to the C language, and then
270 compiles them as Linux kernel modules which are loaded to produce
271 trace data. SystemTap's primary user interface is the cmd:stap
274 The main distinctive features of LTTng is that it produces correlated
275 kernel and user space traces, as well as doing so with the lowest
276 overhead amongst other solutions. It produces trace files in the
277 http://diamon.org/ctf[CTF] format, a file format optimized
278 for the production and analyses of multi-gigabyte data.
280 LTTng is the result of more than 10 years of active open source
281 development by a community of passionate developers.
282 LTTng{nbsp}{revision} is currently available on major desktop and server
285 The main interface for tracing control is a single command-line tool
286 named cmd:lttng. The latter can create several tracing sessions, enable
287 and disable events on the fly, filter events efficiently with custom
288 user expressions, start and stop tracing, and much more. LTTng can
289 record the traces on the file system or send them over the network, and
290 keep them totally or partially. You can view the traces once tracing
291 becomes inactive or in real-time.
293 <<installing-lttng,Install LTTng now>> and
294 <<getting-started,start tracing>>!
300 **LTTng** is a set of software <<plumbing,components>> which interact to
301 <<instrumenting,instrument>> the Linux kernel and user applications, and
302 to <<controlling-tracing,control tracing>> (start and stop
303 tracing, enable and disable event rules, and the rest). Those
304 components are bundled into the following packages:
306 * **LTTng-tools**: Libraries and command-line interface to
308 * **LTTng-modules**: Linux kernel modules to instrument and
310 * **LTTng-UST**: Libraries and Java/Python packages to instrument and
311 trace user applications.
313 Most distributions mark the LTTng-modules and LTTng-UST packages as
314 optional when installing LTTng-tools (which is always required). In the
315 following sections, we always provide the steps to install all three,
318 * You only need to install LTTng-modules if you intend to trace the
320 * You only need to install LTTng-UST if you intend to trace user
324 .Availability of LTTng{nbsp}{revision} for major Linux distributions as of 3 October 2017.
326 |Distribution |Available in releases |Alternatives
328 |https://www.ubuntu.com/[Ubuntu]
329 |Ubuntu{nbsp}14.04 _Trusty Tahr_ and Ubuntu{nbsp}16.04 _Xenial Xerus_:
330 <<ubuntu-ppa,use the LTTng Stable{nbsp}{revision} PPA>>.
331 |link:/docs/v2.9#doc-ubuntu[LTTng{nbsp}2.9 for Ubuntu{nbsp}17.04 _Zesty Zapus_].
333 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
334 other Ubuntu releases.
336 |https://getfedora.org/[Fedora]
337 |<<fedora,Fedora{nbsp}27>>.
338 |link:/docs/v2.9#doc-fedora[LTTng{nbsp}2.9 for Fedora 26].
340 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
341 other Fedora releases.
343 |https://www.debian.org/[Debian]
344 |<<debian,Debian "buster" (testing) and Debian "sid" (unstable)>>.
345 |link:/docs/v2.9#doc-debian[LTTng{nbsp}2.9 for Debian "stretch" (stable)].
347 <<building-from-source,Build LTTng{nbsp}{revision} from source>>.
349 |https://www.archlinux.org/[Arch Linux]
350 |<<arch-linux,Current Arch Linux build>>.
351 |<<building-from-source,Build LTTng{nbsp}{revision} from source>>.
353 |https://alpinelinux.org/[Alpine Linux]
355 |link:/docs/v2.9#doc-alpine-linux[LTTng{nbsp}2.9 for Alpine Linux "edge"].
357 <<building-from-source,Build LTTng{nbsp}{revision} from source>>.
359 |https://www.redhat.com/[RHEL] and https://www.suse.com/[SLES]
360 |See http://packages.efficios.com/[EfficiOS Enterprise Packages].
363 |https://buildroot.org/[Buildroot]
365 |link:/docs/v2.9#doc-buildroot[LTTng{nbsp}2.9 for Buildroot{nbsp}2017.02 and
366 Buildroot{nbsp}2017.05].
368 <<building-from-source,Build LTTng{nbsp}{revision} from source>>.
370 |http://www.openembedded.org/wiki/Main_Page[OpenEmbedded] and
371 https://www.yoctoproject.org/[Yocto]
373 |link:/docs/v2.9#doc-oe-yocto[LTTng{nbsp}2.9 for Yocto Project{nbsp}2.3 _Pyro_]
374 (`openembedded-core` layer).
376 <<building-from-source,Build LTTng{nbsp}{revision} from source>>.
381 === [[ubuntu-official-repositories]]Ubuntu
384 ==== noch:{LTTng} Stable {revision} PPA
386 The https://launchpad.net/~lttng/+archive/ubuntu/stable-{revision}[LTTng
387 Stable{nbsp}{revision} PPA] offers the latest stable
388 LTTng{nbsp}{revision} packages for:
390 * Ubuntu{nbsp}14.04 _Trusty Tahr_
391 * Ubuntu{nbsp}16.04 _Xenial Xerus_
393 To install LTTng{nbsp}{revision} from the LTTng Stable{nbsp}{revision} PPA:
395 . Add the LTTng Stable{nbsp}{revision} PPA repository and update the
401 # apt-add-repository ppa:lttng/stable-2.10
406 . Install the main LTTng{nbsp}{revision} packages:
411 # apt-get install lttng-tools
412 # apt-get install lttng-modules-dkms
413 # apt-get install liblttng-ust-dev
417 . **If you need to instrument and trace
418 <<java-application,Java applications>>**, install the LTTng-UST
424 # apt-get install liblttng-ust-agent-java
428 . **If you need to instrument and trace
429 <<python-application,Python{nbsp}3 applications>>**, install the
430 LTTng-UST Python agent:
435 # apt-get install python3-lttngust
443 To install LTTng{nbsp}{revision} on Fedora{nbsp}27:
445 . Install the LTTng-tools{nbsp}{revision} and LTTng-UST{nbsp}{revision}
451 # yum install lttng-tools
452 # yum install lttng-ust
456 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
462 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.10.tar.bz2 &&
463 tar -xf lttng-modules-latest-2.10.tar.bz2 &&
464 cd lttng-modules-2.10.* &&
466 sudo make modules_install &&
472 .Java and Python application instrumentation and tracing
474 If you need to instrument and trace <<java-application,Java
475 applications>> on Fedora, you need to build and install
476 LTTng-UST{nbsp}{revision} <<building-from-source,from source>> and pass
477 the `--enable-java-agent-jul`, `--enable-java-agent-log4j`, or
478 `--enable-java-agent-all` options to the `configure` script, depending
479 on which Java logging framework you use.
481 If you need to instrument and trace <<python-application,Python
482 applications>> on Fedora, you need to build and install
483 LTTng-UST{nbsp}{revision} from source and pass the
484 `--enable-python-agent` option to the `configure` script.
491 To install LTTng{nbsp}{revision} on Debian "buster" (testing)
492 or Debian "sid" (unstable):
494 . Install the main LTTng{nbsp}{revision} packages:
499 # apt-get install lttng-modules-dkms
500 # apt-get install liblttng-ust-dev
501 # apt-get install lttng-tools
505 . **If you need to instrument and trace <<java-application,Java
506 applications>>**, install the LTTng-UST Java agent:
511 # apt-get install liblttng-ust-agent-java
515 . **If you need to instrument and trace <<python-application,Python
516 applications>>**, install the LTTng-UST Python agent:
521 # apt-get install python3-lttngust
529 LTTng-UST{nbsp}{revision} is available in Arch Linux's _Community_
530 repository, while LTTng-tools{nbsp}{revision} and
531 LTTng-modules{nbsp}{revision} are available in the
532 https://aur.archlinux.org/[AUR].
534 To install LTTng{nbsp}{revision} on Arch Linux, using
535 https://github.com/rmarquis/pacaur[pacaur] for the AUR packages:
537 . Install the main LTTng{nbsp}{revision} packages:
542 # pacman -Sy lttng-ust
543 $ pacaur -Sy lttng-tools
544 $ pacaur -Sy lttng-modules
548 . **If you need to instrument and trace <<python-application,Python
549 applications>>**, install the LTTng-UST Python agent:
554 # pacman -Sy python-lttngust
555 # pacman -Sy python2-lttngust
560 [[enterprise-distributions]]
561 === RHEL, SUSE, and other enterprise distributions
563 To install LTTng on enterprise Linux distributions, such as Red Hat
564 Enterprise Linux (RHEL) and SUSE Linux Enterprise Server (SUSE), please
565 see http://packages.efficios.com/[EfficiOS Enterprise Packages].
568 [[building-from-source]]
569 === Build from source
571 To build and install LTTng{nbsp}{revision} from source:
573 . Using your distribution's package manager, or from source, install
574 the following dependencies of LTTng-tools and LTTng-UST:
577 * https://sourceforge.net/projects/libuuid/[libuuid]
578 * http://directory.fsf.org/wiki/Popt[popt]
579 * http://liburcu.org/[Userspace RCU]
580 * http://www.xmlsoft.org/[libxml2]
583 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
589 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.10.tar.bz2 &&
590 tar -xf lttng-modules-latest-2.10.tar.bz2 &&
591 cd lttng-modules-2.10.* &&
593 sudo make modules_install &&
598 . Download, build, and install the latest LTTng-UST{nbsp}{revision}:
604 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.10.tar.bz2 &&
605 tar -xf lttng-ust-latest-2.10.tar.bz2 &&
606 cd lttng-ust-2.10.* &&
616 .Java and Python application tracing
618 If you need to instrument and trace <<java-application,Java
619 applications>>, pass the `--enable-java-agent-jul`,
620 `--enable-java-agent-log4j`, or `--enable-java-agent-all` options to the
621 `configure` script, depending on which Java logging framework you use.
623 If you need to instrument and trace <<python-application,Python
624 applications>>, pass the `--enable-python-agent` option to the
625 `configure` script. You can set the `PYTHON` environment variable to the
626 path to the Python interpreter for which to install the LTTng-UST Python
634 By default, LTTng-UST libraries are installed to
635 dir:{/usr/local/lib}, which is the de facto directory in which to
636 keep self-compiled and third-party libraries.
638 When <<building-tracepoint-providers-and-user-application,linking an
639 instrumented user application with `liblttng-ust`>>:
641 * Append `/usr/local/lib` to the env:LD_LIBRARY_PATH environment
643 * Pass the `-L/usr/local/lib` and `-Wl,-rpath,/usr/local/lib` options to
644 man:gcc(1), man:g++(1), or man:clang(1).
648 . Download, build, and install the latest LTTng-tools{nbsp}{revision}:
654 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.10.tar.bz2 &&
655 tar -xf lttng-tools-latest-2.10.tar.bz2 &&
656 cd lttng-tools-2.10.* &&
664 TIP: The https://github.com/eepp/vlttng[vlttng tool] can do all the
665 previous steps automatically for a given version of LTTng and confine
666 the installed files in a specific directory. This can be useful to test
667 LTTng without installing it on your system.
673 This is a short guide to get started quickly with LTTng kernel and user
676 Before you follow this guide, make sure to <<installing-lttng,install>>
679 This tutorial walks you through the steps to:
681 . <<tracing-the-linux-kernel,Trace the Linux kernel>>.
682 . <<tracing-your-own-user-application,Trace a user application>> written
684 . <<viewing-and-analyzing-your-traces,View and analyze the
688 [[tracing-the-linux-kernel]]
689 === Trace the Linux kernel
691 The following command lines start with the `#` prompt because you need
692 root privileges to trace the Linux kernel. You can also trace the kernel
693 as a regular user if your Unix user is a member of the
694 <<tracing-group,tracing group>>.
696 . Create a <<tracing-session,tracing session>> which writes its traces
697 to dir:{/tmp/my-kernel-trace}:
702 # lttng create my-kernel-session --output=/tmp/my-kernel-trace
706 . List the available kernel tracepoints and system calls:
711 # lttng list --kernel
712 # lttng list --kernel --syscall
716 . Create <<event,event rules>> which match the desired instrumentation
717 point names, for example the `sched_switch` and `sched_process_fork`
718 tracepoints, and the man:open(2) and man:close(2) system calls:
723 # lttng enable-event --kernel sched_switch,sched_process_fork
724 # lttng enable-event --kernel --syscall open,close
728 You can also create an event rule which matches _all_ the Linux kernel
729 tracepoints (this will generate a lot of data when tracing):
734 # lttng enable-event --kernel --all
738 . <<basic-tracing-session-control,Start tracing>>:
747 . Do some operation on your system for a few seconds. For example,
748 load a website, or list the files of a directory.
749 . <<basic-tracing-session-control,Stop tracing>> and destroy the
760 The man:lttng-destroy(1) command does not destroy the trace data; it
761 only destroys the state of the tracing session.
763 . For the sake of this example, make the recorded trace accessible to
769 # chown -R $(whoami) /tmp/my-kernel-trace
773 See <<viewing-and-analyzing-your-traces,View and analyze the
774 recorded events>> to view the recorded events.
777 [[tracing-your-own-user-application]]
778 === Trace a user application
780 This section steps you through a simple example to trace a
781 _Hello world_ program written in C.
783 To create the traceable user application:
785 . Create the tracepoint provider header file, which defines the
786 tracepoints and the events they can generate:
792 #undef TRACEPOINT_PROVIDER
793 #define TRACEPOINT_PROVIDER hello_world
795 #undef TRACEPOINT_INCLUDE
796 #define TRACEPOINT_INCLUDE "./hello-tp.h"
798 #if !defined(_HELLO_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
801 #include <lttng/tracepoint.h>
811 ctf_string(my_string_field, my_string_arg)
812 ctf_integer(int, my_integer_field, my_integer_arg)
816 #endif /* _HELLO_TP_H */
818 #include <lttng/tracepoint-event.h>
822 . Create the tracepoint provider package source file:
828 #define TRACEPOINT_CREATE_PROBES
829 #define TRACEPOINT_DEFINE
831 #include "hello-tp.h"
835 . Build the tracepoint provider package:
840 $ gcc -c -I. hello-tp.c
844 . Create the _Hello World_ application source file:
851 #include "hello-tp.h"
853 int main(int argc, char *argv[])
857 puts("Hello, World!\nPress Enter to continue...");
860 * The following getchar() call is only placed here for the purpose
861 * of this demonstration, to pause the application in order for
862 * you to have time to list its tracepoints. It is not
868 * A tracepoint() call.
870 * Arguments, as defined in hello-tp.h:
872 * 1. Tracepoint provider name (required)
873 * 2. Tracepoint name (required)
874 * 3. my_integer_arg (first user-defined argument)
875 * 4. my_string_arg (second user-defined argument)
877 * Notice the tracepoint provider and tracepoint names are
878 * NOT strings: they are in fact parts of variables that the
879 * macros in hello-tp.h create.
881 tracepoint(hello_world, my_first_tracepoint, 23, "hi there!");
883 for (x = 0; x < argc; ++x) {
884 tracepoint(hello_world, my_first_tracepoint, x, argv[x]);
887 puts("Quitting now!");
888 tracepoint(hello_world, my_first_tracepoint, x * x, "x^2");
895 . Build the application:
904 . Link the application with the tracepoint provider package,
905 `liblttng-ust`, and `libdl`:
910 $ gcc -o hello hello.o hello-tp.o -llttng-ust -ldl
914 Here's the whole build process:
917 .User space tracing tutorial's build steps.
918 image::ust-flow.png[]
920 To trace the user application:
922 . Run the application with a few arguments:
927 $ ./hello world and beyond
936 Press Enter to continue...
940 . Start an LTTng <<lttng-sessiond,session daemon>>:
945 $ lttng-sessiond --daemonize
949 Note that a session daemon might already be running, for example as
950 a service that the distribution's service manager started.
952 . List the available user space tracepoints:
957 $ lttng list --userspace
961 You see the `hello_world:my_first_tracepoint` tracepoint listed
962 under the `./hello` process.
964 . Create a <<tracing-session,tracing session>>:
969 $ lttng create my-user-space-session
973 . Create an <<event,event rule>> which matches the
974 `hello_world:my_first_tracepoint` event name:
979 $ lttng enable-event --userspace hello_world:my_first_tracepoint
983 . <<basic-tracing-session-control,Start tracing>>:
992 . Go back to the running `hello` application and press Enter. The
993 program executes all `tracepoint()` instrumentation points and exits.
994 . <<basic-tracing-session-control,Stop tracing>> and destroy the
1005 The man:lttng-destroy(1) command does not destroy the trace data; it
1006 only destroys the state of the tracing session.
1008 By default, LTTng saves the traces in
1009 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
1010 where +__name__+ is the tracing session name. The
1011 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
1013 See <<viewing-and-analyzing-your-traces,View and analyze the
1014 recorded events>> to view the recorded events.
1017 [[viewing-and-analyzing-your-traces]]
1018 === View and analyze the recorded events
1020 Once you have completed the <<tracing-the-linux-kernel,Trace the Linux
1021 kernel>> and <<tracing-your-own-user-application,Trace a user
1022 application>> tutorials, you can inspect the recorded events.
1024 Many tools are available to read LTTng traces:
1026 * **cmd:babeltrace** is a command-line utility which converts trace
1027 formats; it supports the format that LTTng produces, CTF, as well as a
1028 basic text output which can be ++grep++ed. The cmd:babeltrace command
1029 is part of the http://diamon.org/babeltrace[Babeltrace] project.
1030 * Babeltrace also includes
1031 **https://www.python.org/[Python] bindings** so
1032 that you can easily open and read an LTTng trace with your own script,
1033 benefiting from the power of Python.
1034 * http://tracecompass.org/[**Trace Compass**]
1035 is a graphical user interface for viewing and analyzing any type of
1036 logs or traces, including LTTng's.
1037 * https://github.com/lttng/lttng-analyses[**LTTng analyses**] is a
1038 project which includes many high-level analyses of LTTng kernel
1039 traces, like scheduling statistics, interrupt frequency distribution,
1040 top CPU usage, and more.
1042 NOTE: This section assumes that the traces recorded during the previous
1043 tutorials were saved to their default location, in the
1044 dir:{$LTTNG_HOME/lttng-traces} directory. The env:LTTNG_HOME
1045 environment variable defaults to `$HOME` if not set.
1048 [[viewing-and-analyzing-your-traces-bt]]
1049 ==== Use the cmd:babeltrace command-line tool
1051 The simplest way to list all the recorded events of a trace is to pass
1052 its path to cmd:babeltrace with no options:
1056 $ babeltrace ~/lttng-traces/my-user-space-session*
1059 cmd:babeltrace finds all traces recursively within the given path and
1060 prints all their events, merging them in chronological order.
1062 You can pipe the output of cmd:babeltrace into a tool like man:grep(1) for
1067 $ babeltrace /tmp/my-kernel-trace | grep _switch
1070 You can pipe the output of cmd:babeltrace into a tool like man:wc(1) to
1071 count the recorded events:
1075 $ babeltrace /tmp/my-kernel-trace | grep _open | wc --lines
1079 [[viewing-and-analyzing-your-traces-bt-python]]
1080 ==== Use the Babeltrace Python bindings
1082 The <<viewing-and-analyzing-your-traces-bt,text output of cmd:babeltrace>>
1083 is useful to isolate events by simple matching using man:grep(1) and
1084 similar utilities. However, more elaborate filters, such as keeping only
1085 event records with a field value falling within a specific range, are
1086 not trivial to write using a shell. Moreover, reductions and even the
1087 most basic computations involving multiple event records are virtually
1088 impossible to implement.
1090 Fortunately, Babeltrace ships with Python 3 bindings which makes it easy
1091 to read the event records of an LTTng trace sequentially and compute the
1092 desired information.
1094 The following script accepts an LTTng Linux kernel trace path as its
1095 first argument and prints the short names of the top 5 running processes
1096 on CPU 0 during the whole trace:
1101 from collections import Counter
1107 if len(sys.argv) != 2:
1108 msg = 'Usage: python3 {} TRACEPATH'.format(sys.argv[0])
1109 print(msg, file=sys.stderr)
1112 # A trace collection contains one or more traces
1113 col = babeltrace.TraceCollection()
1115 # Add the trace provided by the user (LTTng traces always have
1117 if col.add_trace(sys.argv[1], 'ctf') is None:
1118 raise RuntimeError('Cannot add trace')
1120 # This counter dict contains execution times:
1122 # task command name -> total execution time (ns)
1123 exec_times = Counter()
1125 # This contains the last `sched_switch` timestamp
1129 for event in col.events:
1130 # Keep only `sched_switch` events
1131 if event.name != 'sched_switch':
1134 # Keep only events which happened on CPU 0
1135 if event['cpu_id'] != 0:
1139 cur_ts = event.timestamp
1145 # Previous task command (short) name
1146 prev_comm = event['prev_comm']
1148 # Initialize entry in our dict if not yet done
1149 if prev_comm not in exec_times:
1150 exec_times[prev_comm] = 0
1152 # Compute previous command execution time
1153 diff = cur_ts - last_ts
1155 # Update execution time of this command
1156 exec_times[prev_comm] += diff
1158 # Update last timestamp
1162 for name, ns in exec_times.most_common(5):
1164 print('{:20}{} s'.format(name, s))
1169 if __name__ == '__main__':
1170 sys.exit(0 if top5proc() else 1)
1177 $ python3 top5proc.py /tmp/my-kernel-trace/kernel
1183 swapper/0 48.607245889 s
1184 chromium 7.192738188 s
1185 pavucontrol 0.709894415 s
1186 Compositor 0.660867933 s
1187 Xorg.bin 0.616753786 s
1190 Note that `swapper/0` is the "idle" process of CPU 0 on Linux; since we
1191 weren't using the CPU that much when tracing, its first position in the
1196 == [[understanding-lttng]]Core concepts
1198 From a user's perspective, the LTTng system is built on a few concepts,
1199 or objects, on which the <<lttng-cli,cmd:lttng command-line tool>>
1200 operates by sending commands to the <<lttng-sessiond,session daemon>>.
1201 Understanding how those objects relate to eachother is key in mastering
1204 The core concepts are:
1206 * <<tracing-session,Tracing session>>
1207 * <<domain,Tracing domain>>
1208 * <<channel,Channel and ring buffer>>
1209 * <<"event","Instrumentation point, event rule, event, and event record">>
1215 A _tracing session_ is a stateful dialogue between you and
1216 a <<lttng-sessiond,session daemon>>. You can
1217 <<creating-destroying-tracing-sessions,create a new tracing
1218 session>> with the `lttng create` command.
1220 Anything that you do when you control LTTng tracers happens within a
1221 tracing session. In particular, a tracing session:
1224 * Has its own set of trace files.
1225 * Has its own state of activity (started or stopped).
1226 * Has its own <<tracing-session-mode,mode>> (local, network streaming,
1228 * Has its own <<channel,channels>> which have their own
1229 <<event,event rules>>.
1232 .A _tracing session_ contains <<channel,channels>> that are members of <<domain,tracing domains>> and contain <<event,event rules>>.
1233 image::concepts.png[]
1235 Those attributes and objects are completely isolated between different
1238 A tracing session is analogous to a cash machine session:
1239 the operations you do on the banking system through the cash machine do
1240 not alter the data of other users of the same system. In the case of
1241 the cash machine, a session lasts as long as your bank card is inside.
1242 In the case of LTTng, a tracing session lasts from the `lttng create`
1243 command to the `lttng destroy` command.
1246 .Each Unix user has its own set of tracing sessions.
1247 image::many-sessions.png[]
1250 [[tracing-session-mode]]
1251 ==== Tracing session mode
1253 LTTng can send the generated trace data to different locations. The
1254 _tracing session mode_ dictates where to send it. The following modes
1255 are available in LTTng{nbsp}{revision}:
1258 LTTng writes the traces to the file system of the machine being traced
1261 Network streaming mode::
1262 LTTng sends the traces over the network to a
1263 <<lttng-relayd,relay daemon>> running on a remote system.
1266 LTTng does not write the traces by default. Instead, you can request
1267 LTTng to <<taking-a-snapshot,take a snapshot>>, that is, a copy of the
1268 current tracing buffers, and to write it to the target's file system
1269 or to send it over the network to a <<lttng-relayd,relay daemon>>
1270 running on a remote system.
1273 This mode is similar to the network streaming mode, but a live
1274 trace viewer can connect to the distant relay daemon to
1275 <<lttng-live,view event records as LTTng generates them>> by
1282 A _tracing domain_ is a namespace for event sources. A tracing domain
1283 has its own properties and features.
1285 There are currently five available tracing domains:
1289 * `java.util.logging` (JUL)
1293 You must specify a tracing domain when using some commands to avoid
1294 ambiguity. For example, since all the domains support named tracepoints
1295 as event sources (instrumentation points that you manually insert in the
1296 source code), you need to specify a tracing domain when
1297 <<enabling-disabling-events,creating an event rule>> because all the
1298 tracing domains could have tracepoints with the same names.
1300 Some features are reserved to specific tracing domains. Dynamic function
1301 entry and return instrumentation points, for example, are currently only
1302 supported in the Linux kernel tracing domain, but support for other
1303 tracing domains could be added in the future.
1305 You can create <<channel,channels>> in the Linux kernel and user space
1306 tracing domains. The other tracing domains have a single default
1311 === Channel and ring buffer
1313 A _channel_ is an object which is responsible for a set of ring buffers.
1314 Each ring buffer is divided into multiple sub-buffers. When an LTTng
1315 tracer emits an event, it can record it to one or more
1316 sub-buffers. The attributes of a channel determine what to do when
1317 there's no space left for a new event record because all sub-buffers
1318 are full, where to send a full sub-buffer, and other behaviours.
1320 A channel is always associated to a <<domain,tracing domain>>. The
1321 `java.util.logging` (JUL), log4j, and Python tracing domains each have
1322 a default channel which you cannot configure.
1324 A channel also owns <<event,event rules>>. When an LTTng tracer emits
1325 an event, it records it to the sub-buffers of all
1326 the enabled channels with a satisfied event rule, as long as those
1327 channels are part of active <<tracing-session,tracing sessions>>.
1330 [[channel-buffering-schemes]]
1331 ==== Per-user vs. per-process buffering schemes
1333 A channel has at least one ring buffer _per CPU_. LTTng always
1334 records an event to the ring buffer associated to the CPU on which it
1337 Two _buffering schemes_ are available when you
1338 <<enabling-disabling-channels,create a channel>> in the
1339 user space <<domain,tracing domain>>:
1341 Per-user buffering::
1342 Allocate one set of ring buffers--one per CPU--shared by all the
1343 instrumented processes of each Unix user.
1347 .Per-user buffering scheme.
1348 image::per-user-buffering.png[]
1351 Per-process buffering::
1352 Allocate one set of ring buffers--one per CPU--for each
1353 instrumented process.
1357 .Per-process buffering scheme.
1358 image::per-process-buffering.png[]
1361 The per-process buffering scheme tends to consume more memory than the
1362 per-user option because systems generally have more instrumented
1363 processes than Unix users running instrumented processes. However, the
1364 per-process buffering scheme ensures that one process having a high
1365 event throughput won't fill all the shared sub-buffers of the same
1368 The Linux kernel tracing domain has only one available buffering scheme
1369 which is to allocate a single set of ring buffers for the whole system.
1370 This scheme is similar to the per-user option, but with a single, global
1371 user "running" the kernel.
1374 [[channel-overwrite-mode-vs-discard-mode]]
1375 ==== Overwrite vs. discard event loss modes
1377 When an event occurs, LTTng records it to a specific sub-buffer (yellow
1378 arc in the following animation) of a specific channel's ring buffer.
1379 When there's no space left in a sub-buffer, the tracer marks it as
1380 consumable (red) and another, empty sub-buffer starts receiving the
1381 following event records. A <<lttng-consumerd,consumer daemon>>
1382 eventually consumes the marked sub-buffer (returns to white).
1385 [role="docsvg-channel-subbuf-anim"]
1390 In an ideal world, sub-buffers are consumed faster than they are filled,
1391 as is the case in the previous animation. In the real world,
1392 however, all sub-buffers can be full at some point, leaving no space to
1393 record the following events.
1395 By default, LTTng-modules and LTTng-UST are _non-blocking_ tracers: when
1396 no empty sub-buffer is available, it is acceptable to lose event records
1397 when the alternative would be to cause substantial delays in the
1398 instrumented application's execution. LTTng privileges performance over
1399 integrity; it aims at perturbing the traced system as little as possible
1400 in order to make tracing of subtle race conditions and rare interrupt
1403 Starting from LTTng{nbsp}2.10, the LTTng user space tracer, LTTng-UST,
1404 supports a _blocking mode_. See the <<blocking-timeout-example,blocking
1405 timeout example>> to learn how to use the blocking mode.
1407 When it comes to losing event records because no empty sub-buffer is
1408 available, or because the <<opt-blocking-timeout,blocking timeout>> is
1409 reached, the channel's _event loss mode_ determines what to do. The
1410 available event loss modes are:
1413 Drop the newest event records until a the tracer releases a
1416 This is the only available mode when you specify a
1417 <<opt-blocking-timeout,blocking timeout>>.
1420 Clear the sub-buffer containing the oldest event records and start
1421 writing the newest event records there.
1423 This mode is sometimes called _flight recorder mode_ because it's
1425 https://en.wikipedia.org/wiki/Flight_recorder[flight recorder]:
1426 always keep a fixed amount of the latest data.
1428 Which mechanism you should choose depends on your context: prioritize
1429 the newest or the oldest event records in the ring buffer?
1431 Beware that, in overwrite mode, the tracer abandons a _whole sub-buffer_
1432 as soon as a there's no space left for a new event record, whereas in
1433 discard mode, the tracer only discards the event record that doesn't
1436 In discard mode, LTTng increments a count of lost event records when an
1437 event record is lost and saves this count to the trace. In overwrite
1438 mode, since LTTng 2.8, LTTng increments a count of lost sub-buffers when
1439 a sub-buffer is lost and saves this count to the trace. In this mode,
1440 the exact number of lost event records in those lost sub-buffers is not
1441 saved to the trace. Trace analyses can use the trace's saved discarded
1442 event record and sub-buffer counts to decide whether or not to perform
1443 the analyses even if trace data is known to be missing.
1445 There are a few ways to decrease your probability of losing event
1447 <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>> shows
1448 how you can fine-une the sub-buffer count and size of a channel to
1449 virtually stop losing event records, though at the cost of greater
1453 [[channel-subbuf-size-vs-subbuf-count]]
1454 ==== Sub-buffer count and size
1456 When you <<enabling-disabling-channels,create a channel>>, you can
1457 set its number of sub-buffers and their size.
1459 Note that there is noticeable CPU overhead introduced when
1460 switching sub-buffers (marking a full one as consumable and switching
1461 to an empty one for the following events to be recorded). Knowing this,
1462 the following list presents a few practical situations along with how
1463 to configure the sub-buffer count and size for them:
1465 * **High event throughput**: In general, prefer bigger sub-buffers to
1466 lower the risk of losing event records.
1468 Having bigger sub-buffers also ensures a lower
1469 <<channel-switch-timer,sub-buffer switching frequency>>.
1471 The number of sub-buffers is only meaningful if you create the channel
1472 in overwrite mode: in this case, if a sub-buffer overwrite happens, the
1473 other sub-buffers are left unaltered.
1475 * **Low event throughput**: In general, prefer smaller sub-buffers
1476 since the risk of losing event records is low.
1478 Because events occur less frequently, the sub-buffer switching frequency
1479 should remain low and thus the tracer's overhead should not be a
1482 * **Low memory system**: If your target system has a low memory
1483 limit, prefer fewer first, then smaller sub-buffers.
1485 Even if the system is limited in memory, you want to keep the
1486 sub-buffers as big as possible to avoid a high sub-buffer switching
1489 Note that LTTng uses http://diamon.org/ctf/[CTF] as its trace format,
1490 which means event data is very compact. For example, the average
1491 LTTng kernel event record weights about 32{nbsp}bytes. Thus, a
1492 sub-buffer size of 1{nbsp}MiB is considered big.
1494 The previous situations highlight the major trade-off between a few big
1495 sub-buffers and more, smaller sub-buffers: sub-buffer switching
1496 frequency vs. how much data is lost in overwrite mode. Assuming a
1497 constant event throughput and using the overwrite mode, the two
1498 following configurations have the same ring buffer total size:
1501 [role="docsvg-channel-subbuf-size-vs-count-anim"]
1506 * **2 sub-buffers of 4{nbsp}MiB each**: Expect a very low sub-buffer
1507 switching frequency, but if a sub-buffer overwrite happens, half of
1508 the event records so far (4{nbsp}MiB) are definitely lost.
1509 * **8 sub-buffers of 1{nbsp}MiB each**: Expect 4{nbsp}times the tracer's
1510 overhead as the previous configuration, but if a sub-buffer
1511 overwrite happens, only the eighth of event records so far are
1514 In discard mode, the sub-buffers count parameter is pointless: use two
1515 sub-buffers and set their size according to the requirements of your
1519 [[channel-switch-timer]]
1520 ==== Switch timer period
1522 The _switch timer period_ is an important configurable attribute of
1523 a channel to ensure periodic sub-buffer flushing.
1525 When the _switch timer_ expires, a sub-buffer switch happens. You can
1526 set the switch timer period attribute when you
1527 <<enabling-disabling-channels,create a channel>> to ensure that event
1528 data is consumed and committed to trace files or to a distant relay
1529 daemon periodically in case of a low event throughput.
1532 [role="docsvg-channel-switch-timer"]
1537 This attribute is also convenient when you use big sub-buffers to cope
1538 with a sporadic high event throughput, even if the throughput is
1542 [[channel-read-timer]]
1543 ==== Read timer period
1545 By default, the LTTng tracers use a notification mechanism to signal a
1546 full sub-buffer so that a consumer daemon can consume it. When such
1547 notifications must be avoided, for example in real-time applications,
1548 you can use the channel's _read timer_ instead. When the read timer
1549 fires, the <<lttng-consumerd,consumer daemon>> checks for full,
1550 consumable sub-buffers.
1553 [[tracefile-rotation]]
1554 ==== Trace file count and size
1556 By default, trace files can grow as large as needed. You can set the
1557 maximum size of each trace file that a channel writes when you
1558 <<enabling-disabling-channels,create a channel>>. When the size of
1559 a trace file reaches the channel's fixed maximum size, LTTng creates
1560 another file to contain the next event records. LTTng appends a file
1561 count to each trace file name in this case.
1563 If you set the trace file size attribute when you create a channel, the
1564 maximum number of trace files that LTTng creates is _unlimited_ by
1565 default. To limit them, you can also set a maximum number of trace
1566 files. When the number of trace files reaches the channel's fixed
1567 maximum count, the oldest trace file is overwritten. This mechanism is
1568 called _trace file rotation_.
1572 === Instrumentation point, event rule, event, and event record
1574 An _event rule_ is a set of conditions which must be **all** satisfied
1575 for LTTng to record an occuring event.
1577 You set the conditions when you <<enabling-disabling-events,create
1580 You always attach an event rule to <<channel,channel>> when you create
1583 When an event passes the conditions of an event rule, LTTng records it
1584 in one of the attached channel's sub-buffers.
1586 The available conditions, as of LTTng{nbsp}{revision}, are:
1588 * The event rule _is enabled_.
1589 * The instrumentation point's type _is{nbsp}T_.
1590 * The instrumentation point's name (sometimes called _event name_)
1591 _matches{nbsp}N_, but _is not{nbsp}E_.
1592 * The instrumentation point's log level _is as severe as{nbsp}L_, or
1593 _is exactly{nbsp}L_.
1594 * The fields of the event's payload _satisfy_ a filter
1595 expression{nbsp}__F__.
1597 As you can see, all the conditions but the dynamic filter are related to
1598 the event rule's status or to the instrumentation point, not to the
1599 occurring events. This is why, without a filter, checking if an event
1600 passes an event rule is not a dynamic task: when you create or modify an
1601 event rule, all the tracers of its tracing domain enable or disable the
1602 instrumentation points themselves once. This is possible because the
1603 attributes of an instrumentation point (type, name, and log level) are
1604 defined statically. In other words, without a dynamic filter, the tracer
1605 _does not evaluate_ the arguments of an instrumentation point unless it
1606 matches an enabled event rule.
1608 Note that, for LTTng to record an event, the <<channel,channel>> to
1609 which a matching event rule is attached must also be enabled, and the
1610 tracing session owning this channel must be active.
1613 .Logical path from an instrumentation point to an event record.
1614 image::event-rule.png[]
1616 .Event, event record, or event rule?
1618 With so many similar terms, it's easy to get confused.
1620 An **event** is the consequence of the execution of an _instrumentation
1621 point_, like a tracepoint that you manually place in some source code,
1622 or a Linux kernel KProbe. An event is said to _occur_ at a specific
1623 time. Different actions can be taken upon the occurrence of an event,
1624 like record the event's payload to a buffer.
1626 An **event record** is the representation of an event in a sub-buffer. A
1627 tracer is responsible for capturing the payload of an event, current
1628 context variables, the event's ID, and the event's timestamp. LTTng
1629 can append this sub-buffer to a trace file.
1631 An **event rule** is a set of conditions which must all be satisfied for
1632 LTTng to record an occuring event. Events still occur without
1633 satisfying event rules, but LTTng does not record them.
1638 == Components of noch:{LTTng}
1640 The second _T_ in _LTTng_ stands for _toolkit_: it would be wrong
1641 to call LTTng a simple _tool_ since it is composed of multiple
1642 interacting components. This section describes those components,
1643 explains their respective roles, and shows how they connect together to
1644 form the LTTng ecosystem.
1646 The following diagram shows how the most important components of LTTng
1647 interact with user applications, the Linux kernel, and you:
1650 .Control and trace data paths between LTTng components.
1651 image::plumbing.png[]
1653 The LTTng project incorporates:
1655 * **LTTng-tools**: Libraries and command-line interface to
1656 control tracing sessions.
1657 ** <<lttng-sessiond,Session daemon>> (man:lttng-sessiond(8)).
1658 ** <<lttng-consumerd,Consumer daemon>> (man:lttng-consumerd(8)).
1659 ** <<lttng-relayd,Relay daemon>> (man:lttng-relayd(8)).
1660 ** <<liblttng-ctl-lttng,Tracing control library>> (`liblttng-ctl`).
1661 ** <<lttng-cli,Tracing control command-line tool>> (man:lttng(1)).
1662 * **LTTng-UST**: Libraries and Java/Python packages to trace user
1664 ** <<lttng-ust,User space tracing library>> (`liblttng-ust`) and its
1665 headers to instrument and trace any native user application.
1666 ** <<prebuilt-ust-helpers,Preloadable user space tracing helpers>>:
1667 *** `liblttng-ust-libc-wrapper`
1668 *** `liblttng-ust-pthread-wrapper`
1669 *** `liblttng-ust-cyg-profile`
1670 *** `liblttng-ust-cyg-profile-fast`
1671 *** `liblttng-ust-dl`
1672 ** User space tracepoint provider source files generator command-line
1673 tool (man:lttng-gen-tp(1)).
1674 ** <<lttng-ust-agents,LTTng-UST Java agent>> to instrument and trace
1675 Java applications using `java.util.logging` or
1676 Apache log4j 1.2 logging.
1677 ** <<lttng-ust-agents,LTTng-UST Python agent>> to instrument
1678 Python applications using the standard `logging` package.
1679 * **LTTng-modules**: <<lttng-modules,Linux kernel modules>> to trace
1681 ** LTTng kernel tracer module.
1682 ** Tracing ring buffer kernel modules.
1683 ** Probe kernel modules.
1684 ** LTTng logger kernel module.
1688 === Tracing control command-line interface
1691 .The tracing control command-line interface.
1692 image::plumbing-lttng-cli.png[]
1694 The _man:lttng(1) command-line tool_ is the standard user interface to
1695 control LTTng <<tracing-session,tracing sessions>>. The cmd:lttng tool
1696 is part of LTTng-tools.
1698 The cmd:lttng tool is linked with
1699 <<liblttng-ctl-lttng,`liblttng-ctl`>> to communicate with
1700 one or more <<lttng-sessiond,session daemons>> behind the scenes.
1702 The cmd:lttng tool has a Git-like interface:
1706 $ lttng <GENERAL OPTIONS> <COMMAND> <COMMAND OPTIONS>
1709 The <<controlling-tracing,Tracing control>> section explores the
1710 available features of LTTng using the cmd:lttng tool.
1713 [[liblttng-ctl-lttng]]
1714 === Tracing control library
1717 .The tracing control library.
1718 image::plumbing-liblttng-ctl.png[]
1720 The _LTTng control library_, `liblttng-ctl`, is used to communicate
1721 with a <<lttng-sessiond,session daemon>> using a C API that hides the
1722 underlying protocol's details. `liblttng-ctl` is part of LTTng-tools.
1724 The <<lttng-cli,cmd:lttng command-line tool>>
1725 is linked with `liblttng-ctl`.
1727 You can use `liblttng-ctl` in C or $$C++$$ source code by including its
1732 #include <lttng/lttng.h>
1735 Some objects are referenced by name (C string), such as tracing
1736 sessions, but most of them require to create a handle first using
1737 `lttng_create_handle()`.
1739 The best available developer documentation for `liblttng-ctl` is, as of
1740 LTTng{nbsp}{revision}, its installed header files. Every function and
1741 structure is thoroughly documented.
1745 === User space tracing library
1748 .The user space tracing library.
1749 image::plumbing-liblttng-ust.png[]
1751 The _user space tracing library_, `liblttng-ust` (see man:lttng-ust(3)),
1752 is the LTTng user space tracer. It receives commands from a
1753 <<lttng-sessiond,session daemon>>, for example to
1754 enable and disable specific instrumentation points, and writes event
1755 records to ring buffers shared with a
1756 <<lttng-consumerd,consumer daemon>>.
1757 `liblttng-ust` is part of LTTng-UST.
1759 Public C header files are installed beside `liblttng-ust` to
1760 instrument any <<c-application,C or $$C++$$ application>>.
1762 <<lttng-ust-agents,LTTng-UST agents>>, which are regular Java and Python
1763 packages, use their own library providing tracepoints which is
1764 linked with `liblttng-ust`.
1766 An application or library does not have to initialize `liblttng-ust`
1767 manually: its constructor does the necessary tasks to properly register
1768 to a session daemon. The initialization phase also enables the
1769 instrumentation points matching the <<event,event rules>> that you
1773 [[lttng-ust-agents]]
1774 === User space tracing agents
1777 .The user space tracing agents.
1778 image::plumbing-lttng-ust-agents.png[]
1780 The _LTTng-UST Java and Python agents_ are regular Java and Python
1781 packages which add LTTng tracing capabilities to the
1782 native logging frameworks. The LTTng-UST agents are part of LTTng-UST.
1784 In the case of Java, the
1785 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[`java.util.logging`
1786 core logging facilities] and
1787 https://logging.apache.org/log4j/1.2/[Apache log4j 1.2] are supported.
1788 Note that Apache Log4{nbsp}2 is not supported.
1790 In the case of Python, the standard
1791 https://docs.python.org/3/library/logging.html[`logging`] package
1792 is supported. Both Python 2 and Python 3 modules can import the
1793 LTTng-UST Python agent package.
1795 The applications using the LTTng-UST agents are in the
1796 `java.util.logging` (JUL),
1797 log4j, and Python <<domain,tracing domains>>.
1799 Both agents use the same mechanism to trace the log statements. When an
1800 agent is initialized, it creates a log handler that attaches to the root
1801 logger. The agent also registers to a <<lttng-sessiond,session daemon>>.
1802 When the application executes a log statement, it is passed to the
1803 agent's log handler by the root logger. The agent's log handler calls a
1804 native function in a tracepoint provider package shared library linked
1805 with <<lttng-ust,`liblttng-ust`>>, passing the formatted log message and
1806 other fields, like its logger name and its log level. This native
1807 function contains a user space instrumentation point, hence tracing the
1810 The log level condition of an
1811 <<event,event rule>> is considered when tracing
1812 a Java or a Python application, and it's compatible with the standard
1813 JUL, log4j, and Python log levels.
1817 === LTTng kernel modules
1820 .The LTTng kernel modules.
1821 image::plumbing-lttng-modules.png[]
1823 The _LTTng kernel modules_ are a set of Linux kernel modules
1824 which implement the kernel tracer of the LTTng project. The LTTng
1825 kernel modules are part of LTTng-modules.
1827 The LTTng kernel modules include:
1829 * A set of _probe_ modules.
1831 Each module attaches to a specific subsystem
1832 of the Linux kernel using its tracepoint instrument points. There are
1833 also modules to attach to the entry and return points of the Linux
1834 system call functions.
1836 * _Ring buffer_ modules.
1838 A ring buffer implementation is provided as kernel modules. The LTTng
1839 kernel tracer writes to the ring buffer; a
1840 <<lttng-consumerd,consumer daemon>> reads from the ring buffer.
1842 * The _LTTng kernel tracer_ module.
1843 * The _LTTng logger_ module.
1845 The LTTng logger module implements the special path:{/proc/lttng-logger}
1846 file so that any executable can generate LTTng events by opening and
1847 writing to this file.
1849 See <<proc-lttng-logger-abi,LTTng logger>>.
1851 Generally, you do not have to load the LTTng kernel modules manually
1852 (using man:modprobe(8), for example): a root <<lttng-sessiond,session
1853 daemon>> loads the necessary modules when starting. If you have extra
1854 probe modules, you can specify to load them to the session daemon on
1857 The LTTng kernel modules are installed in
1858 +/usr/lib/modules/__release__/extra+ by default, where +__release__+ is
1859 the kernel release (see `uname --kernel-release`).
1866 .The session daemon.
1867 image::plumbing-sessiond.png[]
1869 The _session daemon_, man:lttng-sessiond(8), is a daemon responsible for
1870 managing tracing sessions and for controlling the various components of
1871 LTTng. The session daemon is part of LTTng-tools.
1873 The session daemon sends control requests to and receives control
1876 * The <<lttng-ust,user space tracing library>>.
1878 Any instance of the user space tracing library first registers to
1879 a session daemon. Then, the session daemon can send requests to
1880 this instance, such as:
1883 ** Get the list of tracepoints.
1884 ** Share an <<event,event rule>> so that the user space tracing library
1885 can enable or disable tracepoints. Amongst the possible conditions
1886 of an event rule is a filter expression which `liblttng-ust` evalutes
1887 when an event occurs.
1888 ** Share <<channel,channel>> attributes and ring buffer locations.
1891 The session daemon and the user space tracing library use a Unix
1892 domain socket for their communication.
1894 * The <<lttng-ust-agents,user space tracing agents>>.
1896 Any instance of a user space tracing agent first registers to
1897 a session daemon. Then, the session daemon can send requests to
1898 this instance, such as:
1901 ** Get the list of loggers.
1902 ** Enable or disable a specific logger.
1905 The session daemon and the user space tracing agent use a TCP connection
1906 for their communication.
1908 * The <<lttng-modules,LTTng kernel tracer>>.
1909 * The <<lttng-consumerd,consumer daemon>>.
1911 The session daemon sends requests to the consumer daemon to instruct
1912 it where to send the trace data streams, amongst other information.
1914 * The <<lttng-relayd,relay daemon>>.
1916 The session daemon receives commands from the
1917 <<liblttng-ctl-lttng,tracing control library>>.
1919 The root session daemon loads the appropriate
1920 <<lttng-modules,LTTng kernel modules>> on startup. It also spawns
1921 a <<lttng-consumerd,consumer daemon>> as soon as you create
1922 an <<event,event rule>>.
1924 The session daemon does not send and receive trace data: this is the
1925 role of the <<lttng-consumerd,consumer daemon>> and
1926 <<lttng-relayd,relay daemon>>. It does, however, generate the
1927 http://diamon.org/ctf/[CTF] metadata stream.
1929 Each Unix user can have its own session daemon instance. The
1930 tracing sessions managed by different session daemons are completely
1933 The root user's session daemon is the only one which is
1934 allowed to control the LTTng kernel tracer, and its spawned consumer
1935 daemon is the only one which is allowed to consume trace data from the
1936 LTTng kernel tracer. Note, however, that any Unix user which is a member
1937 of the <<tracing-group,tracing group>> is allowed
1938 to create <<channel,channels>> in the
1939 Linux kernel <<domain,tracing domain>>, and thus to trace the Linux
1942 The <<lttng-cli,cmd:lttng command-line tool>> automatically starts a
1943 session daemon when using its `create` command if none is currently
1944 running. You can also start the session daemon manually.
1951 .The consumer daemon.
1952 image::plumbing-consumerd.png[]
1954 The _consumer daemon_, man:lttng-consumerd(8), is a daemon which shares
1955 ring buffers with user applications or with the LTTng kernel modules to
1956 collect trace data and send it to some location (on disk or to a
1957 <<lttng-relayd,relay daemon>> over the network). The consumer daemon
1958 is part of LTTng-tools.
1960 You do not start a consumer daemon manually: a consumer daemon is always
1961 spawned by a <<lttng-sessiond,session daemon>> as soon as you create an
1962 <<event,event rule>>, that is, before you start tracing. When you kill
1963 its owner session daemon, the consumer daemon also exits because it is
1964 the session daemon's child process. Command-line options of
1965 man:lttng-sessiond(8) target the consumer daemon process.
1967 There are up to two running consumer daemons per Unix user, whereas only
1968 one session daemon can run per user. This is because each process can be
1969 either 32-bit or 64-bit: if the target system runs a mixture of 32-bit
1970 and 64-bit processes, it is more efficient to have separate
1971 corresponding 32-bit and 64-bit consumer daemons. The root user is an
1972 exception: it can have up to _three_ running consumer daemons: 32-bit
1973 and 64-bit instances for its user applications, and one more
1974 reserved for collecting kernel trace data.
1982 image::plumbing-relayd.png[]
1984 The _relay daemon_, man:lttng-relayd(8), is a daemon acting as a bridge
1985 between remote session and consumer daemons, local trace files, and a
1986 remote live trace viewer. The relay daemon is part of LTTng-tools.
1988 The main purpose of the relay daemon is to implement a receiver of
1989 <<sending-trace-data-over-the-network,trace data over the network>>.
1990 This is useful when the target system does not have much file system
1991 space to record trace files locally.
1993 The relay daemon is also a server to which a
1994 <<lttng-live,live trace viewer>> can
1995 connect. The live trace viewer sends requests to the relay daemon to
1996 receive trace data as the target system emits events. The
1997 communication protocol is named _LTTng live_; it is used over TCP
2000 Note that you can start the relay daemon on the target system directly.
2001 This is the setup of choice when the use case is to view events as
2002 the target system emits them without the need of a remote system.
2006 == [[using-lttng]]Instrumentation
2008 There are many examples of tracing and monitoring in our everyday life:
2010 * You have access to real-time and historical weather reports and
2011 forecasts thanks to weather stations installed around the country.
2012 * You know your heart is safe thanks to an electrocardiogram.
2013 * You make sure not to drive your car too fast and to have enough fuel
2014 to reach your destination thanks to gauges visible on your dashboard.
2016 All the previous examples have something in common: they rely on
2017 **instruments**. Without the electrodes attached to the surface of your
2018 body's skin, cardiac monitoring is futile.
2020 LTTng, as a tracer, is no different from those real life examples. If
2021 you're about to trace a software system or, in other words, record its
2022 history of execution, you better have **instrumentation points** in the
2023 subject you're tracing, that is, the actual software.
2025 Various ways were developed to instrument a piece of software for LTTng
2026 tracing. The most straightforward one is to manually place
2027 instrumentation points, called _tracepoints_, in the software's source
2028 code. It is also possible to add instrumentation points dynamically in
2029 the Linux kernel <<domain,tracing domain>>.
2031 If you're only interested in tracing the Linux kernel, your
2032 instrumentation needs are probably already covered by LTTng's built-in
2033 <<lttng-modules,Linux kernel tracepoints>>. You may also wish to trace a
2034 user application which is already instrumented for LTTng tracing.
2035 In such cases, you can skip this whole section and read the topics of
2036 the <<controlling-tracing,Tracing control>> section.
2038 Many methods are available to instrument a piece of software for LTTng
2041 * <<c-application,User space instrumentation for C and $$C++$$
2043 * <<prebuilt-ust-helpers,Prebuilt user space tracing helpers>>.
2044 * <<java-application,User space Java agent>>.
2045 * <<python-application,User space Python agent>>.
2046 * <<proc-lttng-logger-abi,LTTng logger>>.
2047 * <<instrumenting-linux-kernel,LTTng kernel tracepoints>>.
2051 === [[cxx-application]]User space instrumentation for C and $$C++$$ applications
2053 The procedure to instrument a C or $$C++$$ user application with
2054 the <<lttng-ust,LTTng user space tracing library>>, `liblttng-ust`, is:
2056 . <<tracepoint-provider,Create the source files of a tracepoint provider
2058 . <<probing-the-application-source-code,Add tracepoints to
2059 the application's source code>>.
2060 . <<building-tracepoint-providers-and-user-application,Build and link
2061 a tracepoint provider package and the user application>>.
2063 If you need quick, man:printf(3)-like instrumentation, you can skip
2064 those steps and use <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>>
2067 IMPORTANT: You need to <<installing-lttng,install>> LTTng-UST to
2068 instrument a user application with `liblttng-ust`.
2071 [[tracepoint-provider]]
2072 ==== Create the source files of a tracepoint provider package
2074 A _tracepoint provider_ is a set of compiled functions which provide
2075 **tracepoints** to an application, the type of instrumentation point
2076 supported by LTTng-UST. Those functions can emit events with
2077 user-defined fields and serialize those events as event records to one
2078 or more LTTng-UST <<channel,channel>> sub-buffers. The `tracepoint()`
2079 macro, which you <<probing-the-application-source-code,insert in a user
2080 application's source code>>, calls those functions.
2082 A _tracepoint provider package_ is an object file (`.o`) or a shared
2083 library (`.so`) which contains one or more tracepoint providers.
2084 Its source files are:
2086 * One or more <<tpp-header,tracepoint provider header>> (`.h`).
2087 * A <<tpp-source,tracepoint provider package source>> (`.c`).
2089 A tracepoint provider package is dynamically linked with `liblttng-ust`,
2090 the LTTng user space tracer, at run time.
2093 .User application linked with `liblttng-ust` and containing a tracepoint provider.
2094 image::ust-app.png[]
2096 NOTE: If you need quick, man:printf(3)-like instrumentation, you can
2097 skip creating and using a tracepoint provider and use
2098 <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>> instead.
2102 ===== Create a tracepoint provider header file template
2104 A _tracepoint provider header file_ contains the tracepoint
2105 definitions of a tracepoint provider.
2107 To create a tracepoint provider header file:
2109 . Start from this template:
2113 .Tracepoint provider header file template (`.h` file extension).
2115 #undef TRACEPOINT_PROVIDER
2116 #define TRACEPOINT_PROVIDER provider_name
2118 #undef TRACEPOINT_INCLUDE
2119 #define TRACEPOINT_INCLUDE "./tp.h"
2121 #if !defined(_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
2124 #include <lttng/tracepoint.h>
2127 * Use TRACEPOINT_EVENT(), TRACEPOINT_EVENT_CLASS(),
2128 * TRACEPOINT_EVENT_INSTANCE(), and TRACEPOINT_LOGLEVEL() here.
2133 #include <lttng/tracepoint-event.h>
2139 * `provider_name` with the name of your tracepoint provider.
2140 * `"tp.h"` with the name of your tracepoint provider header file.
2142 . Below the `#include <lttng/tracepoint.h>` line, put your
2143 <<defining-tracepoints,tracepoint definitions>>.
2145 Your tracepoint provider name must be unique amongst all the possible
2146 tracepoint provider names used on the same target system. We
2147 suggest to include the name of your project or company in the name,
2148 for example, `org_lttng_my_project_tpp`.
2150 TIP: [[lttng-gen-tp]]You can use the man:lttng-gen-tp(1) tool to create
2151 this boilerplate for you. When using cmd:lttng-gen-tp, all you need to
2152 write are the <<defining-tracepoints,tracepoint definitions>>.
2155 [[defining-tracepoints]]
2156 ===== Create a tracepoint definition
2158 A _tracepoint definition_ defines, for a given tracepoint:
2160 * Its **input arguments**. They are the macro parameters that the
2161 `tracepoint()` macro accepts for this particular tracepoint
2162 in the user application's source code.
2163 * Its **output event fields**. They are the sources of event fields
2164 that form the payload of any event that the execution of the
2165 `tracepoint()` macro emits for this particular tracepoint.
2167 You can create a tracepoint definition by using the
2168 `TRACEPOINT_EVENT()` macro below the `#include <lttng/tracepoint.h>`
2170 <<tpp-header,tracepoint provider header file template>>.
2172 The syntax of the `TRACEPOINT_EVENT()` macro is:
2175 .`TRACEPOINT_EVENT()` macro syntax.
2178 /* Tracepoint provider name */
2181 /* Tracepoint name */
2184 /* Input arguments */
2189 /* Output event fields */
2198 * `provider_name` with your tracepoint provider name.
2199 * `tracepoint_name` with your tracepoint name.
2200 * `arguments` with the <<tpp-def-input-args,input arguments>>.
2201 * `fields` with the <<tpp-def-output-fields,output event field>>
2204 This tracepoint emits events named `provider_name:tracepoint_name`.
2207 .Event name's length limitation
2209 The concatenation of the tracepoint provider name and the
2210 tracepoint name must not exceed **254 characters**. If it does, the
2211 instrumented application compiles and runs, but LTTng throws multiple
2212 warnings and you could experience serious issues.
2215 [[tpp-def-input-args]]The syntax of the `TP_ARGS()` macro is:
2218 .`TP_ARGS()` macro syntax.
2227 * `type` with the C type of the argument.
2228 * `arg_name` with the argument name.
2230 You can repeat `type` and `arg_name` up to 10 times to have
2231 more than one argument.
2233 .`TP_ARGS()` usage with three arguments.
2245 The `TP_ARGS()` and `TP_ARGS(void)` forms are valid to create a
2246 tracepoint definition with no input arguments.
2248 [[tpp-def-output-fields]]The `TP_FIELDS()` macro contains a list of
2249 `ctf_*()` macros. Each `ctf_*()` macro defines one event field. See
2250 man:lttng-ust(3) for a complete description of the available `ctf_*()`
2251 macros. A `ctf_*()` macro specifies the type, size, and byte order of
2254 Each `ctf_*()` macro takes an _argument expression_ parameter. This is a
2255 C expression that the tracer evalutes at the `tracepoint()` macro site
2256 in the application's source code. This expression provides a field's
2257 source of data. The argument expression can include input argument names
2258 listed in the `TP_ARGS()` macro.
2260 Each `ctf_*()` macro also takes a _field name_ parameter. Field names
2261 must be unique within a given tracepoint definition.
2263 Here's a complete tracepoint definition example:
2265 .Tracepoint definition.
2267 The following tracepoint definition defines a tracepoint which takes
2268 three input arguments and has four output event fields.
2272 #include "my-custom-structure.h"
2278 const struct my_custom_structure*, my_custom_structure,
2283 ctf_string(query_field, query)
2284 ctf_float(double, ratio_field, ratio)
2285 ctf_integer(int, recv_size, my_custom_structure->recv_size)
2286 ctf_integer(int, send_size, my_custom_structure->send_size)
2291 You can refer to this tracepoint definition with the `tracepoint()`
2292 macro in your application's source code like this:
2296 tracepoint(my_provider, my_tracepoint,
2297 my_structure, some_ratio, the_query);
2301 NOTE: The LTTng tracer only evaluates tracepoint arguments at run time
2302 if they satisfy an enabled <<event,event rule>>.
2305 [[using-tracepoint-classes]]
2306 ===== Use a tracepoint class
2308 A _tracepoint class_ is a class of tracepoints which share the same
2309 output event field definitions. A _tracepoint instance_ is one
2310 instance of such a defined tracepoint class, with its own tracepoint
2313 The <<defining-tracepoints,`TRACEPOINT_EVENT()` macro>> is actually a
2314 shorthand which defines both a tracepoint class and a tracepoint
2315 instance at the same time.
2317 When you build a tracepoint provider package, the C or $$C++$$ compiler
2318 creates one serialization function for each **tracepoint class**. A
2319 serialization function is responsible for serializing the event fields
2320 of a tracepoint to a sub-buffer when tracing.
2322 For various performance reasons, when your situation requires multiple
2323 tracepoint definitions with different names, but with the same event
2324 fields, we recommend that you manually create a tracepoint class
2325 and instantiate as many tracepoint instances as needed. One positive
2326 effect of such a design, amongst other advantages, is that all
2327 tracepoint instances of the same tracepoint class reuse the same
2328 serialization function, thus reducing
2329 https://en.wikipedia.org/wiki/Cache_pollution[cache pollution].
2331 .Use a tracepoint class and tracepoint instances.
2333 Consider the following three tracepoint definitions:
2345 ctf_integer(int, userid, userid)
2346 ctf_integer(size_t, len, len)
2358 ctf_integer(int, userid, userid)
2359 ctf_integer(size_t, len, len)
2371 ctf_integer(int, userid, userid)
2372 ctf_integer(size_t, len, len)
2377 In this case, we create three tracepoint classes, with one implicit
2378 tracepoint instance for each of them: `get_account`, `get_settings`, and
2379 `get_transaction`. However, they all share the same event field names
2380 and types. Hence three identical, yet independent serialization
2381 functions are created when you build the tracepoint provider package.
2383 A better design choice is to define a single tracepoint class and three
2384 tracepoint instances:
2388 /* The tracepoint class */
2389 TRACEPOINT_EVENT_CLASS(
2390 /* Tracepoint provider name */
2393 /* Tracepoint class name */
2396 /* Input arguments */
2402 /* Output event fields */
2404 ctf_integer(int, userid, userid)
2405 ctf_integer(size_t, len, len)
2409 /* The tracepoint instances */
2410 TRACEPOINT_EVENT_INSTANCE(
2411 /* Tracepoint provider name */
2414 /* Tracepoint class name */
2417 /* Tracepoint name */
2420 /* Input arguments */
2426 TRACEPOINT_EVENT_INSTANCE(
2435 TRACEPOINT_EVENT_INSTANCE(
2448 [[assigning-log-levels]]
2449 ===== Assign a log level to a tracepoint definition
2451 You can assign an optional _log level_ to a
2452 <<defining-tracepoints,tracepoint definition>>.
2454 Assigning different levels of severity to tracepoint definitions can
2455 be useful: when you <<enabling-disabling-events,create an event rule>>,
2456 you can target tracepoints having a log level as severe as a specific
2459 The concept of LTTng-UST log levels is similar to the levels found
2460 in typical logging frameworks:
2462 * In a logging framework, the log level is given by the function
2463 or method name you use at the log statement site: `debug()`,
2464 `info()`, `warn()`, `error()`, and so on.
2465 * In LTTng-UST, you statically assign the log level to a tracepoint
2466 definition; any `tracepoint()` macro invocation which refers to
2467 this definition has this log level.
2469 You can assign a log level to a tracepoint definition with the
2470 `TRACEPOINT_LOGLEVEL()` macro. You must use this macro _after_ the
2471 <<defining-tracepoints,`TRACEPOINT_EVENT()`>> or
2472 <<using-tracepoint-classes,`TRACEPOINT_INSTANCE()`>> macro for a given
2475 The syntax of the `TRACEPOINT_LOGLEVEL()` macro is:
2478 .`TRACEPOINT_LOGLEVEL()` macro syntax.
2480 TRACEPOINT_LOGLEVEL(provider_name, tracepoint_name, log_level)
2485 * `provider_name` with the tracepoint provider name.
2486 * `tracepoint_name` with the tracepoint name.
2487 * `log_level` with the log level to assign to the tracepoint
2488 definition named `tracepoint_name` in the `provider_name`
2489 tracepoint provider.
2491 See man:lttng-ust(3) for a list of available log level names.
2493 .Assign the `TRACE_DEBUG_UNIT` log level to a tracepoint definition.
2497 /* Tracepoint definition */
2506 ctf_integer(int, userid, userid)
2507 ctf_integer(size_t, len, len)
2511 /* Log level assignment */
2512 TRACEPOINT_LOGLEVEL(my_app, get_transaction, TRACE_DEBUG_UNIT)
2518 ===== Create a tracepoint provider package source file
2520 A _tracepoint provider package source file_ is a C source file which
2521 includes a <<tpp-header,tracepoint provider header file>> to expand its
2522 macros into event serialization and other functions.
2524 You can always use the following tracepoint provider package source
2528 .Tracepoint provider package source file template.
2530 #define TRACEPOINT_CREATE_PROBES
2535 Replace `tp.h` with the name of your <<tpp-header,tracepoint provider
2536 header file>> name. You may also include more than one tracepoint
2537 provider header file here to create a tracepoint provider package
2538 holding more than one tracepoint providers.
2541 [[probing-the-application-source-code]]
2542 ==== Add tracepoints to an application's source code
2544 Once you <<tpp-header,create a tracepoint provider header file>>, you
2545 can use the `tracepoint()` macro in your application's
2546 source code to insert the tracepoints that this header
2547 <<defining-tracepoints,defines>>.
2549 The `tracepoint()` macro takes at least two parameters: the tracepoint
2550 provider name and the tracepoint name. The corresponding tracepoint
2551 definition defines the other parameters.
2553 .`tracepoint()` usage.
2555 The following <<defining-tracepoints,tracepoint definition>> defines a
2556 tracepoint which takes two input arguments and has two output event
2560 .Tracepoint provider header file.
2562 #include "my-custom-structure.h"
2569 const char*, cmd_name
2572 ctf_string(cmd_name, cmd_name)
2573 ctf_integer(int, number_of_args, argc)
2578 You can refer to this tracepoint definition with the `tracepoint()`
2579 macro in your application's source code like this:
2582 .Application's source file.
2586 int main(int argc, char* argv[])
2588 tracepoint(my_provider, my_tracepoint, argc, argv[0]);
2594 Note how the application's source code includes
2595 the tracepoint provider header file containing the tracepoint
2596 definitions to use, path:{tp.h}.
2599 .`tracepoint()` usage with a complex tracepoint definition.
2601 Consider this complex tracepoint definition, where multiple event
2602 fields refer to the same input arguments in their argument expression
2606 .Tracepoint provider header file.
2608 /* For `struct stat` */
2609 #include <sys/types.h>
2610 #include <sys/stat.h>
2622 ctf_integer(int, my_constant_field, 23 + 17)
2623 ctf_integer(int, my_int_arg_field, my_int_arg)
2624 ctf_integer(int, my_int_arg_field2, my_int_arg * my_int_arg)
2625 ctf_integer(int, sum4_field, my_str_arg[0] + my_str_arg[1] +
2626 my_str_arg[2] + my_str_arg[3])
2627 ctf_string(my_str_arg_field, my_str_arg)
2628 ctf_integer_hex(off_t, size_field, st->st_size)
2629 ctf_float(double, size_dbl_field, (double) st->st_size)
2630 ctf_sequence_text(char, half_my_str_arg_field, my_str_arg,
2631 size_t, strlen(my_str_arg) / 2)
2636 You can refer to this tracepoint definition with the `tracepoint()`
2637 macro in your application's source code like this:
2640 .Application's source file.
2642 #define TRACEPOINT_DEFINE
2649 stat("/etc/fstab", &s);
2650 tracepoint(my_provider, my_tracepoint, 23, "Hello, World!", &s);
2656 If you look at the event record that LTTng writes when tracing this
2657 program, assuming the file size of path:{/etc/fstab} is 301{nbsp}bytes,
2658 it should look like this:
2660 .Event record fields
2662 |Field's name |Field's value
2663 |`my_constant_field` |40
2664 |`my_int_arg_field` |23
2665 |`my_int_arg_field2` |529
2667 |`my_str_arg_field` |`Hello, World!`
2668 |`size_field` |0x12d
2669 |`size_dbl_field` |301.0
2670 |`half_my_str_arg_field` |`Hello,`
2674 Sometimes, the arguments you pass to `tracepoint()` are expensive to
2675 compute--they use the call stack, for example. To avoid this
2676 computation when the tracepoint is disabled, you can use the
2677 `tracepoint_enabled()` and `do_tracepoint()` macros.
2679 The syntax of the `tracepoint_enabled()` and `do_tracepoint()` macros
2683 .`tracepoint_enabled()` and `do_tracepoint()` macros syntax.
2685 tracepoint_enabled(provider_name, tracepoint_name)
2686 do_tracepoint(provider_name, tracepoint_name, ...)
2691 * `provider_name` with the tracepoint provider name.
2692 * `tracepoint_name` with the tracepoint name.
2694 `tracepoint_enabled()` returns a non-zero value if the tracepoint named
2695 `tracepoint_name` from the provider named `provider_name` is enabled
2698 `do_tracepoint()` is like `tracepoint()`, except that it doesn't check
2699 if the tracepoint is enabled. Using `tracepoint()` with
2700 `tracepoint_enabled()` is dangerous since `tracepoint()` also contains
2701 the `tracepoint_enabled()` check, thus a race condition is
2702 possible in this situation:
2705 .Possible race condition when using `tracepoint_enabled()` with `tracepoint()`.
2707 if (tracepoint_enabled(my_provider, my_tracepoint)) {
2708 stuff = prepare_stuff();
2711 tracepoint(my_provider, my_tracepoint, stuff);
2714 If the tracepoint is enabled after the condition, then `stuff` is not
2715 prepared: the emitted event will either contain wrong data, or the whole
2716 application could crash (segmentation fault, for example).
2718 NOTE: Neither `tracepoint_enabled()` nor `do_tracepoint()` have an
2719 `STAP_PROBEV()` call. If you need it, you must emit
2723 [[building-tracepoint-providers-and-user-application]]
2724 ==== Build and link a tracepoint provider package and an application
2726 Once you have one or more <<tpp-header,tracepoint provider header
2727 files>> and a <<tpp-source,tracepoint provider package source file>>,
2728 you can create the tracepoint provider package by compiling its source
2729 file. From here, multiple build and run scenarios are possible. The
2730 following table shows common application and library configurations
2731 along with the required command lines to achieve them.
2733 In the following diagrams, we use the following file names:
2736 Executable application.
2739 Application's object file.
2742 Tracepoint provider package object file.
2745 Tracepoint provider package archive file.
2748 Tracepoint provider package shared object file.
2751 User library object file.
2754 User library shared object file.
2756 We use the following symbols in the diagrams of table below:
2759 .Symbols used in the build scenario diagrams.
2760 image::ust-sit-symbols.png[]
2762 We assume that path:{.} is part of the env:LD_LIBRARY_PATH environment
2763 variable in the following instructions.
2765 [role="growable ust-scenarios",cols="asciidoc,asciidoc"]
2766 .Common tracepoint provider package scenarios.
2768 |Scenario |Instructions
2771 The instrumented application is statically linked with
2772 the tracepoint provider package object.
2774 image::ust-sit+app-linked-with-tp-o+app-instrumented.png[]
2777 include::../common/ust-sit-step-tp-o.txt[]
2779 To build the instrumented application:
2781 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2786 #define TRACEPOINT_DEFINE
2790 . Compile the application source file:
2799 . Build the application:
2804 $ gcc -o app app.o tpp.o -llttng-ust -ldl
2808 To run the instrumented application:
2810 * Start the application:
2820 The instrumented application is statically linked with the
2821 tracepoint provider package archive file.
2823 image::ust-sit+app-linked-with-tp-a+app-instrumented.png[]
2826 To create the tracepoint provider package archive file:
2828 . Compile the <<tpp-source,tracepoint provider package source file>>:
2837 . Create the tracepoint provider package archive file:
2842 $ ar rcs tpp.a tpp.o
2846 To build the instrumented application:
2848 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2853 #define TRACEPOINT_DEFINE
2857 . Compile the application source file:
2866 . Build the application:
2871 $ gcc -o app app.o tpp.a -llttng-ust -ldl
2875 To run the instrumented application:
2877 * Start the application:
2887 The instrumented application is linked with the tracepoint provider
2888 package shared object.
2890 image::ust-sit+app-linked-with-tp-so+app-instrumented.png[]
2893 include::../common/ust-sit-step-tp-so.txt[]
2895 To build the instrumented application:
2897 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2902 #define TRACEPOINT_DEFINE
2906 . Compile the application source file:
2915 . Build the application:
2920 $ gcc -o app app.o -ldl -L. -ltpp
2924 To run the instrumented application:
2926 * Start the application:
2936 The tracepoint provider package shared object is preloaded before the
2937 instrumented application starts.
2939 image::ust-sit+tp-so-preloaded+app-instrumented.png[]
2942 include::../common/ust-sit-step-tp-so.txt[]
2944 To build the instrumented application:
2946 . In path:{app.c}, before including path:{tpp.h}, add the
2952 #define TRACEPOINT_DEFINE
2953 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
2957 . Compile the application source file:
2966 . Build the application:
2971 $ gcc -o app app.o -ldl
2975 To run the instrumented application with tracing support:
2977 * Preload the tracepoint provider package shared object and
2978 start the application:
2983 $ LD_PRELOAD=./libtpp.so ./app
2987 To run the instrumented application without tracing support:
2989 * Start the application:
2999 The instrumented application dynamically loads the tracepoint provider
3000 package shared object.
3002 See the <<dlclose-warning,warning about `dlclose()`>>.
3004 image::ust-sit+app-dlopens-tp-so+app-instrumented.png[]
3007 include::../common/ust-sit-step-tp-so.txt[]
3009 To build the instrumented application:
3011 . In path:{app.c}, before including path:{tpp.h}, add the
3017 #define TRACEPOINT_DEFINE
3018 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3022 . Compile the application source file:
3031 . Build the application:
3036 $ gcc -o app app.o -ldl
3040 To run the instrumented application:
3042 * Start the application:
3052 The application is linked with the instrumented user library.
3054 The instrumented user library is statically linked with the tracepoint
3055 provider package object file.
3057 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-o+lib-instrumented.png[]
3060 include::../common/ust-sit-step-tp-o-fpic.txt[]
3062 To build the instrumented user library:
3064 . In path:{emon.c}, before including path:{tpp.h}, add the
3070 #define TRACEPOINT_DEFINE
3074 . Compile the user library source file:
3079 $ gcc -I. -fpic -c emon.c
3083 . Build the user library shared object:
3088 $ gcc -shared -o libemon.so emon.o tpp.o -llttng-ust -ldl
3092 To build the application:
3094 . Compile the application source file:
3103 . Build the application:
3108 $ gcc -o app app.o -L. -lemon
3112 To run the application:
3114 * Start the application:
3124 The application is linked with the instrumented user library.
3126 The instrumented user library is linked with the tracepoint provider
3127 package shared object.
3129 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3132 include::../common/ust-sit-step-tp-so.txt[]
3134 To build the instrumented user library:
3136 . In path:{emon.c}, before including path:{tpp.h}, add the
3142 #define TRACEPOINT_DEFINE
3146 . Compile the user library source file:
3151 $ gcc -I. -fpic -c emon.c
3155 . Build the user library shared object:
3160 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3164 To build the application:
3166 . Compile the application source file:
3175 . Build the application:
3180 $ gcc -o app app.o -L. -lemon
3184 To run the application:
3186 * Start the application:
3196 The tracepoint provider package shared object is preloaded before the
3199 The application is linked with the instrumented user library.
3201 image::ust-sit+tp-so-preloaded+app-linked-with-lib+lib-instrumented.png[]
3204 include::../common/ust-sit-step-tp-so.txt[]
3206 To build the instrumented user library:
3208 . In path:{emon.c}, before including path:{tpp.h}, add the
3214 #define TRACEPOINT_DEFINE
3215 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3219 . Compile the user library source file:
3224 $ gcc -I. -fpic -c emon.c
3228 . Build the user library shared object:
3233 $ gcc -shared -o libemon.so emon.o -ldl
3237 To build the application:
3239 . Compile the application source file:
3248 . Build the application:
3253 $ gcc -o app app.o -L. -lemon
3257 To run the application with tracing support:
3259 * Preload the tracepoint provider package shared object and
3260 start the application:
3265 $ LD_PRELOAD=./libtpp.so ./app
3269 To run the application without tracing support:
3271 * Start the application:
3281 The application is linked with the instrumented user library.
3283 The instrumented user library dynamically loads the tracepoint provider
3284 package shared object.
3286 See the <<dlclose-warning,warning about `dlclose()`>>.
3288 image::ust-sit+app-linked-with-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3291 include::../common/ust-sit-step-tp-so.txt[]
3293 To build the instrumented user library:
3295 . In path:{emon.c}, before including path:{tpp.h}, add the
3301 #define TRACEPOINT_DEFINE
3302 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3306 . Compile the user library source file:
3311 $ gcc -I. -fpic -c emon.c
3315 . Build the user library shared object:
3320 $ gcc -shared -o libemon.so emon.o -ldl
3324 To build the application:
3326 . Compile the application source file:
3335 . Build the application:
3340 $ gcc -o app app.o -L. -lemon
3344 To run the application:
3346 * Start the application:
3356 The application dynamically loads the instrumented user library.
3358 The instrumented user library is linked with the tracepoint provider
3359 package shared object.
3361 See the <<dlclose-warning,warning about `dlclose()`>>.
3363 image::ust-sit+app-dlopens-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3366 include::../common/ust-sit-step-tp-so.txt[]
3368 To build the instrumented user library:
3370 . In path:{emon.c}, before including path:{tpp.h}, add the
3376 #define TRACEPOINT_DEFINE
3380 . Compile the user library source file:
3385 $ gcc -I. -fpic -c emon.c
3389 . Build the user library shared object:
3394 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3398 To build the application:
3400 . Compile the application source file:
3409 . Build the application:
3414 $ gcc -o app app.o -ldl -L. -lemon
3418 To run the application:
3420 * Start the application:
3430 The application dynamically loads the instrumented user library.
3432 The instrumented user library dynamically loads the tracepoint provider
3433 package shared object.
3435 See the <<dlclose-warning,warning about `dlclose()`>>.
3437 image::ust-sit+app-dlopens-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3440 include::../common/ust-sit-step-tp-so.txt[]
3442 To build the instrumented user library:
3444 . In path:{emon.c}, before including path:{tpp.h}, add the
3450 #define TRACEPOINT_DEFINE
3451 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3455 . Compile the user library source file:
3460 $ gcc -I. -fpic -c emon.c
3464 . Build the user library shared object:
3469 $ gcc -shared -o libemon.so emon.o -ldl
3473 To build the application:
3475 . Compile the application source file:
3484 . Build the application:
3489 $ gcc -o app app.o -ldl -L. -lemon
3493 To run the application:
3495 * Start the application:
3505 The tracepoint provider package shared object is preloaded before the
3508 The application dynamically loads the instrumented user library.
3510 image::ust-sit+tp-so-preloaded+app-dlopens-lib+lib-instrumented.png[]
3513 include::../common/ust-sit-step-tp-so.txt[]
3515 To build the instrumented user library:
3517 . In path:{emon.c}, before including path:{tpp.h}, add the
3523 #define TRACEPOINT_DEFINE
3524 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3528 . Compile the user library source file:
3533 $ gcc -I. -fpic -c emon.c
3537 . Build the user library shared object:
3542 $ gcc -shared -o libemon.so emon.o -ldl
3546 To build the application:
3548 . Compile the application source file:
3557 . Build the application:
3562 $ gcc -o app app.o -L. -lemon
3566 To run the application with tracing support:
3568 * Preload the tracepoint provider package shared object and
3569 start the application:
3574 $ LD_PRELOAD=./libtpp.so ./app
3578 To run the application without tracing support:
3580 * Start the application:
3590 The application is statically linked with the tracepoint provider
3591 package object file.
3593 The application is linked with the instrumented user library.
3595 image::ust-sit+app-linked-with-tp-o+app-linked-with-lib+lib-instrumented.png[]
3598 include::../common/ust-sit-step-tp-o.txt[]
3600 To build the instrumented user library:
3602 . In path:{emon.c}, before including path:{tpp.h}, add the
3608 #define TRACEPOINT_DEFINE
3612 . Compile the user library source file:
3617 $ gcc -I. -fpic -c emon.c
3621 . Build the user library shared object:
3626 $ gcc -shared -o libemon.so emon.o
3630 To build the application:
3632 . Compile the application source file:
3641 . Build the application:
3646 $ gcc -o app app.o tpp.o -llttng-ust -ldl -L. -lemon
3650 To run the instrumented application:
3652 * Start the application:
3662 The application is statically linked with the tracepoint provider
3663 package object file.
3665 The application dynamically loads the instrumented user library.
3667 image::ust-sit+app-linked-with-tp-o+app-dlopens-lib+lib-instrumented.png[]
3670 include::../common/ust-sit-step-tp-o.txt[]
3672 To build the application:
3674 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3679 #define TRACEPOINT_DEFINE
3683 . Compile the application source file:
3692 . Build the application:
3697 $ gcc -Wl,--export-dynamic -o app app.o tpp.o \
3702 The `--export-dynamic` option passed to the linker is necessary for the
3703 dynamically loaded library to ``see'' the tracepoint symbols defined in
3706 To build the instrumented user library:
3708 . Compile the user library source file:
3713 $ gcc -I. -fpic -c emon.c
3717 . Build the user library shared object:
3722 $ gcc -shared -o libemon.so emon.o
3726 To run the application:
3728 * Start the application:
3740 .Do not use man:dlclose(3) on a tracepoint provider package
3742 Never use man:dlclose(3) on any shared object which:
3744 * Is linked with, statically or dynamically, a tracepoint provider
3746 * Calls man:dlopen(3) itself to dynamically open a tracepoint provider
3747 package shared object.
3749 This is currently considered **unsafe** due to a lack of reference
3750 counting from LTTng-UST to the shared object.
3752 A known workaround (available since glibc 2.2) is to use the
3753 `RTLD_NODELETE` flag when calling man:dlopen(3) initially. This has the
3754 effect of not unloading the loaded shared object, even if man:dlclose(3)
3757 You can also preload the tracepoint provider package shared object with
3758 the env:LD_PRELOAD environment variable to overcome this limitation.
3762 [[using-lttng-ust-with-daemons]]
3763 ===== Use noch:{LTTng-UST} with daemons
3765 If your instrumented application calls man:fork(2), man:clone(2),
3766 or BSD's man:rfork(2), without a following man:exec(3)-family
3767 system call, you must preload the path:{liblttng-ust-fork.so} shared
3768 object when you start the application.
3772 $ LD_PRELOAD=liblttng-ust-fork.so ./my-app
3775 If your tracepoint provider package is
3776 a shared library which you also preload, you must put both
3777 shared objects in env:LD_PRELOAD:
3781 $ LD_PRELOAD=liblttng-ust-fork.so:/path/to/tp.so ./my-app
3787 ===== Use noch:{LTTng-UST} with applications which close file descriptors that don't belong to them
3789 If your instrumented application closes one or more file descriptors
3790 which it did not open itself, you must preload the
3791 path:{liblttng-ust-fd.so} shared object when you start the application:
3795 $ LD_PRELOAD=liblttng-ust-fd.so ./my-app
3798 Typical use cases include closing all the file descriptors after
3799 man:fork(2) or man:rfork(2) and buggy applications doing
3803 [[lttng-ust-pkg-config]]
3804 ===== Use noch:{pkg-config}
3806 On some distributions, LTTng-UST ships with a
3807 https://www.freedesktop.org/wiki/Software/pkg-config/[pkg-config]
3808 metadata file. If this is your case, then you can use cmd:pkg-config to
3809 build an application on the command line:
3813 $ gcc -o my-app my-app.o tp.o $(pkg-config --cflags --libs lttng-ust)
3817 [[instrumenting-32-bit-app-on-64-bit-system]]
3818 ===== [[advanced-instrumenting-techniques]]Build a 32-bit instrumented application for a 64-bit target system
3820 In order to trace a 32-bit application running on a 64-bit system,
3821 LTTng must use a dedicated 32-bit
3822 <<lttng-consumerd,consumer daemon>>.
3824 The following steps show how to build and install a 32-bit consumer
3825 daemon, which is _not_ part of the default 64-bit LTTng build, how to
3826 build and install the 32-bit LTTng-UST libraries, and how to build and
3827 link an instrumented 32-bit application in that context.
3829 To build a 32-bit instrumented application for a 64-bit target system,
3830 assuming you have a fresh target system with no installed Userspace RCU
3833 . Download, build, and install a 32-bit version of Userspace RCU:
3838 $ cd $(mktemp -d) &&
3839 wget http://lttng.org/files/urcu/userspace-rcu-latest-0.9.tar.bz2 &&
3840 tar -xf userspace-rcu-latest-0.9.tar.bz2 &&
3841 cd userspace-rcu-0.9.* &&
3842 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 &&
3844 sudo make install &&
3849 . Using your distribution's package manager, or from source, install
3850 the following 32-bit versions of the following dependencies of
3851 LTTng-tools and LTTng-UST:
3854 * https://sourceforge.net/projects/libuuid/[libuuid]
3855 * http://directory.fsf.org/wiki/Popt[popt]
3856 * http://www.xmlsoft.org/[libxml2]
3859 . Download, build, and install a 32-bit version of the latest
3860 LTTng-UST{nbsp}{revision}:
3865 $ cd $(mktemp -d) &&
3866 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.10.tar.bz2 &&
3867 tar -xf lttng-ust-latest-2.10.tar.bz2 &&
3868 cd lttng-ust-2.10.* &&
3869 ./configure --libdir=/usr/local/lib32 \
3870 CFLAGS=-m32 CXXFLAGS=-m32 \
3871 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
3873 sudo make install &&
3880 Depending on your distribution,
3881 32-bit libraries could be installed at a different location than
3882 `/usr/lib32`. For example, Debian is known to install
3883 some 32-bit libraries in `/usr/lib/i386-linux-gnu`.
3885 In this case, make sure to set `LDFLAGS` to all the
3886 relevant 32-bit library paths, for example:
3890 $ LDFLAGS='-L/usr/lib/i386-linux-gnu -L/usr/lib32'
3894 . Download the latest LTTng-tools{nbsp}{revision}, build, and install
3895 the 32-bit consumer daemon:
3900 $ cd $(mktemp -d) &&
3901 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.10.tar.bz2 &&
3902 tar -xf lttng-tools-latest-2.10.tar.bz2 &&
3903 cd lttng-tools-2.10.* &&
3904 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 CXXFLAGS=-m32 \
3905 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' \
3906 --disable-bin-lttng --disable-bin-lttng-crash \
3907 --disable-bin-lttng-relayd --disable-bin-lttng-sessiond &&
3909 cd src/bin/lttng-consumerd &&
3910 sudo make install &&
3915 . From your distribution or from source,
3916 <<installing-lttng,install>> the 64-bit versions of
3917 LTTng-UST and Userspace RCU.
3918 . Download, build, and install the 64-bit version of the
3919 latest LTTng-tools{nbsp}{revision}:
3924 $ cd $(mktemp -d) &&
3925 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.10.tar.bz2 &&
3926 tar -xf lttng-tools-latest-2.10.tar.bz2 &&
3927 cd lttng-tools-2.10.* &&
3928 ./configure --with-consumerd32-libdir=/usr/local/lib32 \
3929 --with-consumerd32-bin=/usr/local/lib32/lttng/libexec/lttng-consumerd &&
3931 sudo make install &&
3936 . Pass the following options to man:gcc(1), man:g++(1), or man:clang(1)
3937 when linking your 32-bit application:
3940 -m32 -L/usr/lib32 -L/usr/local/lib32 \
3941 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32
3944 For example, let's rebuild the quick start example in
3945 <<tracing-your-own-user-application,Trace a user application>> as an
3946 instrumented 32-bit application:
3951 $ gcc -m32 -c -I. hello-tp.c
3952 $ gcc -m32 -c hello.c
3953 $ gcc -m32 -o hello hello.o hello-tp.o \
3954 -L/usr/lib32 -L/usr/local/lib32 \
3955 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32 \
3960 No special action is required to execute the 32-bit application and
3961 to trace it: use the command-line man:lttng(1) tool as usual.
3968 man:tracef(3) is a small LTTng-UST API designed for quick,
3969 man:printf(3)-like instrumentation without the burden of
3970 <<tracepoint-provider,creating>> and
3971 <<building-tracepoint-providers-and-user-application,building>>
3972 a tracepoint provider package.
3974 To use `tracef()` in your application:
3976 . In the C or C++ source files where you need to use `tracef()`,
3977 include `<lttng/tracef.h>`:
3982 #include <lttng/tracef.h>
3986 . In the application's source code, use `tracef()` like you would use
3994 tracef("my message: %d (%s)", my_integer, my_string);
4000 . Link your application with `liblttng-ust`:
4005 $ gcc -o app app.c -llttng-ust
4009 To trace the events that `tracef()` calls emit:
4011 * <<enabling-disabling-events,Create an event rule>> which matches the
4012 `lttng_ust_tracef:*` event name:
4017 $ lttng enable-event --userspace 'lttng_ust_tracef:*'
4022 .Limitations of `tracef()`
4024 The `tracef()` utility function was developed to make user space tracing
4025 super simple, albeit with notable disadvantages compared to
4026 <<defining-tracepoints,user-defined tracepoints>>:
4028 * All the emitted events have the same tracepoint provider and
4029 tracepoint names, respectively `lttng_ust_tracef` and `event`.
4030 * There is no static type checking.
4031 * The only event record field you actually get, named `msg`, is a string
4032 potentially containing the values you passed to `tracef()`
4033 using your own format string. This also means that you cannot filter
4034 events with a custom expression at run time because there are no
4036 * Since `tracef()` uses the C standard library's man:vasprintf(3)
4037 function behind the scenes to format the strings at run time, its
4038 expected performance is lower than with user-defined tracepoints,
4039 which do not require a conversion to a string.
4041 Taking this into consideration, `tracef()` is useful for some quick
4042 prototyping and debugging, but you should not consider it for any
4043 permanent and serious applicative instrumentation.
4049 ==== Use `tracelog()`
4051 The man:tracelog(3) API is very similar to <<tracef,`tracef()`>>, with
4052 the difference that it accepts an additional log level parameter.
4054 The goal of `tracelog()` is to ease the migration from logging to
4057 To use `tracelog()` in your application:
4059 . In the C or C++ source files where you need to use `tracelog()`,
4060 include `<lttng/tracelog.h>`:
4065 #include <lttng/tracelog.h>
4069 . In the application's source code, use `tracelog()` like you would use
4070 man:printf(3), except for the first parameter which is the log
4078 tracelog(TRACE_WARNING, "my message: %d (%s)",
4079 my_integer, my_string);
4085 See man:lttng-ust(3) for a list of available log level names.
4087 . Link your application with `liblttng-ust`:
4092 $ gcc -o app app.c -llttng-ust
4096 To trace the events that `tracelog()` calls emit with a log level
4097 _as severe as_ a specific log level:
4099 * <<enabling-disabling-events,Create an event rule>> which matches the
4100 `lttng_ust_tracelog:*` event name and a minimum level
4106 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4107 --loglevel=TRACE_WARNING
4111 To trace the events that `tracelog()` calls emit with a
4112 _specific log level_:
4114 * Create an event rule which matches the `lttng_ust_tracelog:*`
4115 event name and a specific log level:
4120 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4121 --loglevel-only=TRACE_INFO
4126 [[prebuilt-ust-helpers]]
4127 === Prebuilt user space tracing helpers
4129 The LTTng-UST package provides a few helpers in the form or preloadable
4130 shared objects which automatically instrument system functions and
4133 The helper shared objects are normally found in dir:{/usr/lib}. If you
4134 built LTTng-UST <<building-from-source,from source>>, they are probably
4135 located in dir:{/usr/local/lib}.
4137 The installed user space tracing helpers in LTTng-UST{nbsp}{revision}
4140 path:{liblttng-ust-libc-wrapper.so}::
4141 path:{liblttng-ust-pthread-wrapper.so}::
4142 <<liblttng-ust-libc-pthread-wrapper,C{nbsp}standard library
4143 memory and POSIX threads function tracing>>.
4145 path:{liblttng-ust-cyg-profile.so}::
4146 path:{liblttng-ust-cyg-profile-fast.so}::
4147 <<liblttng-ust-cyg-profile,Function entry and exit tracing>>.
4149 path:{liblttng-ust-dl.so}::
4150 <<liblttng-ust-dl,Dynamic linker tracing>>.
4152 To use a user space tracing helper with any user application:
4154 * Preload the helper shared object when you start the application:
4159 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so my-app
4163 You can preload more than one helper:
4168 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so:liblttng-ust-dl.so my-app
4174 [[liblttng-ust-libc-pthread-wrapper]]
4175 ==== Instrument C standard library memory and POSIX threads functions
4177 The path:{liblttng-ust-libc-wrapper.so} and
4178 path:{liblttng-ust-pthread-wrapper.so} helpers
4179 add instrumentation to some C standard library and POSIX
4183 .Functions instrumented by preloading path:{liblttng-ust-libc-wrapper.so}.
4185 |TP provider name |TP name |Instrumented function
4187 .6+|`lttng_ust_libc` |`malloc` |man:malloc(3)
4188 |`calloc` |man:calloc(3)
4189 |`realloc` |man:realloc(3)
4190 |`free` |man:free(3)
4191 |`memalign` |man:memalign(3)
4192 |`posix_memalign` |man:posix_memalign(3)
4196 .Functions instrumented by preloading path:{liblttng-ust-pthread-wrapper.so}.
4198 |TP provider name |TP name |Instrumented function
4200 .4+|`lttng_ust_pthread` |`pthread_mutex_lock_req` |man:pthread_mutex_lock(3p) (request time)
4201 |`pthread_mutex_lock_acq` |man:pthread_mutex_lock(3p) (acquire time)
4202 |`pthread_mutex_trylock` |man:pthread_mutex_trylock(3p)
4203 |`pthread_mutex_unlock` |man:pthread_mutex_unlock(3p)
4206 When you preload the shared object, it replaces the functions listed
4207 in the previous tables by wrappers which contain tracepoints and call
4208 the replaced functions.
4211 [[liblttng-ust-cyg-profile]]
4212 ==== Instrument function entry and exit
4214 The path:{liblttng-ust-cyg-profile*.so} helpers can add instrumentation
4215 to the entry and exit points of functions.
4217 man:gcc(1) and man:clang(1) have an option named
4218 https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html[`-finstrument-functions`]
4219 which generates instrumentation calls for entry and exit to functions.
4220 The LTTng-UST function tracing helpers,
4221 path:{liblttng-ust-cyg-profile.so} and
4222 path:{liblttng-ust-cyg-profile-fast.so}, take advantage of this feature
4223 to add tracepoints to the two generated functions (which contain
4224 `cyg_profile` in their names, hence the helper's name).
4226 To use the LTTng-UST function tracing helper, the source files to
4227 instrument must be built using the `-finstrument-functions` compiler
4230 There are two versions of the LTTng-UST function tracing helper:
4232 * **path:{liblttng-ust-cyg-profile-fast.so}** is a lightweight variant
4233 that you should only use when it can be _guaranteed_ that the
4234 complete event stream is recorded without any lost event record.
4235 Any kind of duplicate information is left out.
4237 Assuming no event record is lost, having only the function addresses on
4238 entry is enough to create a call graph, since an event record always
4239 contains the ID of the CPU that generated it.
4241 You can use a tool like man:addr2line(1) to convert function addresses
4242 back to source file names and line numbers.
4244 * **path:{liblttng-ust-cyg-profile.so}** is a more robust variant
4245 which also works in use cases where event records might get discarded or
4246 not recorded from application startup.
4247 In these cases, the trace analyzer needs more information to be
4248 able to reconstruct the program flow.
4250 See man:lttng-ust-cyg-profile(3) to learn more about the instrumentation
4251 points of this helper.
4253 All the tracepoints that this helper provides have the
4254 log level `TRACE_DEBUG_FUNCTION` (see man:lttng-ust(3)).
4256 TIP: It's sometimes a good idea to limit the number of source files that
4257 you compile with the `-finstrument-functions` option to prevent LTTng
4258 from writing an excessive amount of trace data at run time. When using
4259 man:gcc(1), you can use the
4260 `-finstrument-functions-exclude-function-list` option to avoid
4261 instrument entries and exits of specific function names.
4266 ==== Instrument the dynamic linker
4268 The path:{liblttng-ust-dl.so} helper adds instrumentation to the
4269 man:dlopen(3) and man:dlclose(3) function calls.
4271 See man:lttng-ust-dl(3) to learn more about the instrumentation points
4276 [[java-application]]
4277 === User space Java agent
4279 You can instrument any Java application which uses one of the following
4282 * The https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[**`java.util.logging`**]
4283 (JUL) core logging facilities.
4284 * http://logging.apache.org/log4j/1.2/[**Apache log4j 1.2**], since
4285 LTTng 2.6. Note that Apache Log4j{nbsp}2 is not supported.
4288 .LTTng-UST Java agent imported by a Java application.
4289 image::java-app.png[]
4291 Note that the methods described below are new in LTTng{nbsp}{revision}.
4292 Previous LTTng versions use another technique.
4294 NOTE: We use http://openjdk.java.net/[OpenJDK]{nbsp}8 for development
4295 and https://ci.lttng.org/[continuous integration], thus this version is
4296 directly supported. However, the LTTng-UST Java agent is also tested
4297 with OpenJDK{nbsp}7.
4302 ==== Use the LTTng-UST Java agent for `java.util.logging`
4304 To use the LTTng-UST Java agent in a Java application which uses
4305 `java.util.logging` (JUL):
4307 . In the Java application's source code, import the LTTng-UST
4308 log handler package for `java.util.logging`:
4313 import org.lttng.ust.agent.jul.LttngLogHandler;
4317 . Create an LTTng-UST JUL log handler:
4322 Handler lttngUstLogHandler = new LttngLogHandler();
4326 . Add this handler to the JUL loggers which should emit LTTng events:
4331 Logger myLogger = Logger.getLogger("some-logger");
4333 myLogger.addHandler(lttngUstLogHandler);
4337 . Use `java.util.logging` log statements and configuration as usual.
4338 The loggers with an attached LTTng-UST log handler can emit
4341 . Before exiting the application, remove the LTTng-UST log handler from
4342 the loggers attached to it and call its `close()` method:
4347 myLogger.removeHandler(lttngUstLogHandler);
4348 lttngUstLogHandler.close();
4352 This is not strictly necessary, but it is recommended for a clean
4353 disposal of the handler's resources.
4355 . Include the LTTng-UST Java agent's common and JUL-specific JAR files,
4356 path:{lttng-ust-agent-common.jar} and path:{lttng-ust-agent-jul.jar},
4358 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4359 path] when you build the Java application.
4361 The JAR files are typically located in dir:{/usr/share/java}.
4363 IMPORTANT: The LTTng-UST Java agent must be
4364 <<installing-lttng,installed>> for the logging framework your
4367 .Use the LTTng-UST Java agent for `java.util.logging`.
4372 import java.io.IOException;
4373 import java.util.logging.Handler;
4374 import java.util.logging.Logger;
4375 import org.lttng.ust.agent.jul.LttngLogHandler;
4379 private static final int answer = 42;
4381 public static void main(String[] argv) throws Exception
4384 Logger logger = Logger.getLogger("jello");
4386 // Create an LTTng-UST log handler
4387 Handler lttngUstLogHandler = new LttngLogHandler();
4389 // Add the LTTng-UST log handler to our logger
4390 logger.addHandler(lttngUstLogHandler);
4393 logger.info("some info");
4394 logger.warning("some warning");
4396 logger.finer("finer information; the answer is " + answer);
4398 logger.severe("error!");
4400 // Not mandatory, but cleaner
4401 logger.removeHandler(lttngUstLogHandler);
4402 lttngUstLogHandler.close();
4411 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4414 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4415 <<enabling-disabling-events,create an event rule>> matching the
4416 `jello` JUL logger, and <<basic-tracing-session-control,start tracing>>:
4421 $ lttng enable-event --jul jello
4425 Run the compiled class:
4429 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4432 <<basic-tracing-session-control,Stop tracing>> and inspect the
4442 In the resulting trace, an <<event,event record>> generated by a Java
4443 application using `java.util.logging` is named `lttng_jul:event` and
4444 has the following fields:
4447 Log record's message.
4453 Name of the class in which the log statement was executed.
4456 Name of the method in which the log statement was executed.
4459 Logging time (timestamp in milliseconds).
4462 Log level integer value.
4465 ID of the thread in which the log statement was executed.
4467 You can use the opt:lttng-enable-event(1):--loglevel or
4468 opt:lttng-enable-event(1):--loglevel-only option of the
4469 man:lttng-enable-event(1) command to target a range of JUL log levels
4470 or a specific JUL log level.
4475 ==== Use the LTTng-UST Java agent for Apache log4j
4477 To use the LTTng-UST Java agent in a Java application which uses
4480 . In the Java application's source code, import the LTTng-UST
4481 log appender package for Apache log4j:
4486 import org.lttng.ust.agent.log4j.LttngLogAppender;
4490 . Create an LTTng-UST log4j log appender:
4495 Appender lttngUstLogAppender = new LttngLogAppender();
4499 . Add this appender to the log4j loggers which should emit LTTng events:
4504 Logger myLogger = Logger.getLogger("some-logger");
4506 myLogger.addAppender(lttngUstLogAppender);
4510 . Use Apache log4j log statements and configuration as usual. The
4511 loggers with an attached LTTng-UST log appender can emit LTTng events.
4513 . Before exiting the application, remove the LTTng-UST log appender from
4514 the loggers attached to it and call its `close()` method:
4519 myLogger.removeAppender(lttngUstLogAppender);
4520 lttngUstLogAppender.close();
4524 This is not strictly necessary, but it is recommended for a clean
4525 disposal of the appender's resources.
4527 . Include the LTTng-UST Java agent's common and log4j-specific JAR
4528 files, path:{lttng-ust-agent-common.jar} and
4529 path:{lttng-ust-agent-log4j.jar}, in the
4530 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4531 path] when you build the Java application.
4533 The JAR files are typically located in dir:{/usr/share/java}.
4535 IMPORTANT: The LTTng-UST Java agent must be
4536 <<installing-lttng,installed>> for the logging framework your
4539 .Use the LTTng-UST Java agent for Apache log4j.
4544 import org.apache.log4j.Appender;
4545 import org.apache.log4j.Logger;
4546 import org.lttng.ust.agent.log4j.LttngLogAppender;
4550 private static final int answer = 42;
4552 public static void main(String[] argv) throws Exception
4555 Logger logger = Logger.getLogger("jello");
4557 // Create an LTTng-UST log appender
4558 Appender lttngUstLogAppender = new LttngLogAppender();
4560 // Add the LTTng-UST log appender to our logger
4561 logger.addAppender(lttngUstLogAppender);
4564 logger.info("some info");
4565 logger.warn("some warning");
4567 logger.debug("debug information; the answer is " + answer);
4569 logger.fatal("error!");
4571 // Not mandatory, but cleaner
4572 logger.removeAppender(lttngUstLogAppender);
4573 lttngUstLogAppender.close();
4579 Build this example (`$LOG4JPATH` is the path to the Apache log4j JAR
4584 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH Test.java
4587 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4588 <<enabling-disabling-events,create an event rule>> matching the
4589 `jello` log4j logger, and <<basic-tracing-session-control,start tracing>>:
4594 $ lttng enable-event --log4j jello
4598 Run the compiled class:
4602 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH:. Test
4605 <<basic-tracing-session-control,Stop tracing>> and inspect the
4615 In the resulting trace, an <<event,event record>> generated by a Java
4616 application using log4j is named `lttng_log4j:event` and
4617 has the following fields:
4620 Log record's message.
4626 Name of the class in which the log statement was executed.
4629 Name of the method in which the log statement was executed.
4632 Name of the file in which the executed log statement is located.
4635 Line number at which the log statement was executed.
4641 Log level integer value.
4644 Name of the Java thread in which the log statement was executed.
4646 You can use the opt:lttng-enable-event(1):--loglevel or
4647 opt:lttng-enable-event(1):--loglevel-only option of the
4648 man:lttng-enable-event(1) command to target a range of Apache log4j log levels
4649 or a specific log4j log level.
4653 [[java-application-context]]
4654 ==== Provide application-specific context fields in a Java application
4656 A Java application-specific context field is a piece of state provided
4657 by the application which <<adding-context,you can add>>, using the
4658 man:lttng-add-context(1) command, to each <<event,event record>>
4659 produced by the log statements of this application.
4661 For example, a given object might have a current request ID variable.
4662 You can create a context information retriever for this object and
4663 assign a name to this current request ID. You can then, using the
4664 man:lttng-add-context(1) command, add this context field by name to
4665 the JUL or log4j <<channel,channel>>.
4667 To provide application-specific context fields in a Java application:
4669 . In the Java application's source code, import the LTTng-UST
4670 Java agent context classes and interfaces:
4675 import org.lttng.ust.agent.context.ContextInfoManager;
4676 import org.lttng.ust.agent.context.IContextInfoRetriever;
4680 . Create a context information retriever class, that is, a class which
4681 implements the `IContextInfoRetriever` interface:
4686 class MyContextInfoRetriever implements IContextInfoRetriever
4689 public Object retrieveContextInfo(String key)
4691 if (key.equals("intCtx")) {
4693 } else if (key.equals("strContext")) {
4694 return "context value!";
4703 This `retrieveContextInfo()` method is the only member of the
4704 `IContextInfoRetriever` interface. Its role is to return the current
4705 value of a state by name to create a context field. The names of the
4706 context fields and which state variables they return depends on your
4709 All primitive types and objects are supported as context fields.
4710 When `retrieveContextInfo()` returns an object, the context field
4711 serializer calls its `toString()` method to add a string field to
4712 event records. The method can also return `null`, which means that
4713 no context field is available for the required name.
4715 . Register an instance of your context information retriever class to
4716 the context information manager singleton:
4721 IContextInfoRetriever cir = new MyContextInfoRetriever();
4722 ContextInfoManager cim = ContextInfoManager.getInstance();
4723 cim.registerContextInfoRetriever("retrieverName", cir);
4727 . Before exiting the application, remove your context information
4728 retriever from the context information manager singleton:
4733 ContextInfoManager cim = ContextInfoManager.getInstance();
4734 cim.unregisterContextInfoRetriever("retrieverName");
4738 This is not strictly necessary, but it is recommended for a clean
4739 disposal of some manager's resources.
4741 . Build your Java application with LTTng-UST Java agent support as
4742 usual, following the procedure for either the <<jul,JUL>> or
4743 <<log4j,Apache log4j>> framework.
4746 .Provide application-specific context fields in a Java application.
4751 import java.util.logging.Handler;
4752 import java.util.logging.Logger;
4753 import org.lttng.ust.agent.jul.LttngLogHandler;
4754 import org.lttng.ust.agent.context.ContextInfoManager;
4755 import org.lttng.ust.agent.context.IContextInfoRetriever;
4759 // Our context information retriever class
4760 private static class MyContextInfoRetriever
4761 implements IContextInfoRetriever
4764 public Object retrieveContextInfo(String key) {
4765 if (key.equals("intCtx")) {
4767 } else if (key.equals("strContext")) {
4768 return "context value!";
4775 private static final int answer = 42;
4777 public static void main(String args[]) throws Exception
4779 // Get the context information manager instance
4780 ContextInfoManager cim = ContextInfoManager.getInstance();
4782 // Create and register our context information retriever
4783 IContextInfoRetriever cir = new MyContextInfoRetriever();
4784 cim.registerContextInfoRetriever("myRetriever", cir);
4787 Logger logger = Logger.getLogger("jello");
4789 // Create an LTTng-UST log handler
4790 Handler lttngUstLogHandler = new LttngLogHandler();
4792 // Add the LTTng-UST log handler to our logger
4793 logger.addHandler(lttngUstLogHandler);
4796 logger.info("some info");
4797 logger.warning("some warning");
4799 logger.finer("finer information; the answer is " + answer);
4801 logger.severe("error!");
4803 // Not mandatory, but cleaner
4804 logger.removeHandler(lttngUstLogHandler);
4805 lttngUstLogHandler.close();
4806 cim.unregisterContextInfoRetriever("myRetriever");
4815 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4818 <<creating-destroying-tracing-sessions,Create a tracing session>>
4819 and <<enabling-disabling-events,create an event rule>> matching the
4825 $ lttng enable-event --jul jello
4828 <<adding-context,Add the application-specific context fields>> to the
4833 $ lttng add-context --jul --type='$app.myRetriever:intCtx'
4834 $ lttng add-context --jul --type='$app.myRetriever:strContext'
4837 <<basic-tracing-session-control,Start tracing>>:
4844 Run the compiled class:
4848 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4851 <<basic-tracing-session-control,Stop tracing>> and inspect the
4863 [[python-application]]
4864 === User space Python agent
4866 You can instrument a Python 2 or Python 3 application which uses the
4867 standard https://docs.python.org/3/library/logging.html[`logging`]
4870 Each log statement emits an LTTng event once the
4871 application module imports the
4872 <<lttng-ust-agents,LTTng-UST Python agent>> package.
4875 .A Python application importing the LTTng-UST Python agent.
4876 image::python-app.png[]
4878 To use the LTTng-UST Python agent:
4880 . In the Python application's source code, import the LTTng-UST Python
4890 The LTTng-UST Python agent automatically adds its logging handler to the
4891 root logger at import time.
4893 Any log statement that the application executes before this import does
4894 not emit an LTTng event.
4896 IMPORTANT: The LTTng-UST Python agent must be
4897 <<installing-lttng,installed>>.
4899 . Use log statements and logging configuration as usual.
4900 Since the LTTng-UST Python agent adds a handler to the _root_
4901 logger, you can trace any log statement from any logger.
4903 .Use the LTTng-UST Python agent.
4914 logging.basicConfig()
4915 logger = logging.getLogger('my-logger')
4918 logger.debug('debug message')
4919 logger.info('info message')
4920 logger.warn('warn message')
4921 logger.error('error message')
4922 logger.critical('critical message')
4926 if __name__ == '__main__':
4930 NOTE: `logging.basicConfig()`, which adds to the root logger a basic
4931 logging handler which prints to the standard error stream, is not
4932 strictly required for LTTng-UST tracing to work, but in versions of
4933 Python preceding 3.2, you could see a warning message which indicates
4934 that no handler exists for the logger `my-logger`.
4936 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4937 <<enabling-disabling-events,create an event rule>> matching the
4938 `my-logger` Python logger, and <<basic-tracing-session-control,start
4944 $ lttng enable-event --python my-logger
4948 Run the Python script:
4955 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
4965 In the resulting trace, an <<event,event record>> generated by a Python
4966 application is named `lttng_python:event` and has the following fields:
4969 Logging time (string).
4972 Log record's message.
4978 Name of the function in which the log statement was executed.
4981 Line number at which the log statement was executed.
4984 Log level integer value.
4987 ID of the Python thread in which the log statement was executed.
4990 Name of the Python thread in which the log statement was executed.
4992 You can use the opt:lttng-enable-event(1):--loglevel or
4993 opt:lttng-enable-event(1):--loglevel-only option of the
4994 man:lttng-enable-event(1) command to target a range of Python log levels
4995 or a specific Python log level.
4997 When an application imports the LTTng-UST Python agent, the agent tries
4998 to register to a <<lttng-sessiond,session daemon>>. Note that you must
4999 <<start-sessiond,start the session daemon>> _before_ you run the Python
5000 application. If a session daemon is found, the agent tries to register
5001 to it during 5{nbsp}seconds, after which the application continues
5002 without LTTng tracing support. You can override this timeout value with
5003 the env:LTTNG_UST_PYTHON_REGISTER_TIMEOUT environment variable
5006 If the session daemon stops while a Python application with an imported
5007 LTTng-UST Python agent runs, the agent retries to connect and to
5008 register to a session daemon every 3{nbsp}seconds. You can override this
5009 delay with the env:LTTNG_UST_PYTHON_REGISTER_RETRY_DELAY environment
5014 [[proc-lttng-logger-abi]]
5017 The `lttng-tracer` Linux kernel module, part of
5018 <<lttng-modules,LTTng-modules>>, creates the special LTTng logger file
5019 path:{/proc/lttng-logger} when it's loaded. Any application can write
5020 text data to this file to emit an LTTng event.
5023 .An application writes to the LTTng logger file to emit an LTTng event.
5024 image::lttng-logger.png[]
5026 The LTTng logger is the quickest method--not the most efficient,
5027 however--to add instrumentation to an application. It is designed
5028 mostly to instrument shell scripts:
5032 $ echo "Some message, some $variable" > /proc/lttng-logger
5035 Any event that the LTTng logger emits is named `lttng_logger` and
5036 belongs to the Linux kernel <<domain,tracing domain>>. However, unlike
5037 other instrumentation points in the kernel tracing domain, **any Unix
5038 user** can <<enabling-disabling-events,create an event rule>> which
5039 matches its event name, not only the root user or users in the
5040 <<tracing-group,tracing group>>.
5042 To use the LTTng logger:
5044 * From any application, write text data to the path:{/proc/lttng-logger}
5047 The `msg` field of `lttng_logger` event records contains the
5050 NOTE: The maximum message length of an LTTng logger event is
5051 1024{nbsp}bytes. Writing more than this makes the LTTng logger emit more
5052 than one event to contain the remaining data.
5054 You should not use the LTTng logger to trace a user application which
5055 can be instrumented in a more efficient way, namely:
5057 * <<c-application,C and $$C++$$ applications>>.
5058 * <<java-application,Java applications>>.
5059 * <<python-application,Python applications>>.
5061 .Use the LTTng logger.
5066 echo 'Hello, World!' > /proc/lttng-logger
5068 df --human-readable --print-type / > /proc/lttng-logger
5071 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5072 <<enabling-disabling-events,create an event rule>> matching the
5073 `lttng_logger` Linux kernel tracepoint, and
5074 <<basic-tracing-session-control,start tracing>>:
5079 $ lttng enable-event --kernel lttng_logger
5083 Run the Bash script:
5090 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5101 [[instrumenting-linux-kernel]]
5102 === LTTng kernel tracepoints
5104 NOTE: This section shows how to _add_ instrumentation points to the
5105 Linux kernel. The kernel's subsystems are already thoroughly
5106 instrumented at strategic places for LTTng when you
5107 <<installing-lttng,install>> the <<lttng-modules,LTTng-modules>>
5111 There are two methods to instrument the Linux kernel:
5113 . <<linux-add-lttng-layer,Add an LTTng layer>> over an existing ftrace
5114 tracepoint which uses the `TRACE_EVENT()` API.
5116 Choose this if you want to instrumentation a Linux kernel tree with an
5117 instrumentation point compatible with ftrace, perf, and SystemTap.
5119 . Use an <<linux-lttng-tracepoint-event,LTTng-only approach>> to
5120 instrument an out-of-tree kernel module.
5122 Choose this if you don't need ftrace, perf, or SystemTap support.
5126 [[linux-add-lttng-layer]]
5127 ==== [[instrumenting-linux-kernel-itself]][[mainline-trace-event]][[lttng-adaptation-layer]]Add an LTTng layer to an existing ftrace tracepoint
5129 This section shows how to add an LTTng layer to existing ftrace
5130 instrumentation using the `TRACE_EVENT()` API.
5132 This section does not document the `TRACE_EVENT()` macro. You can
5133 read the following articles to learn more about this API:
5135 * http://lwn.net/Articles/379903/[Using the TRACE_EVENT() macro (Part 1)]
5136 * http://lwn.net/Articles/381064/[Using the TRACE_EVENT() macro (Part 2)]
5137 * http://lwn.net/Articles/383362/[Using the TRACE_EVENT() macro (Part 3)]
5139 The following procedure assumes that your ftrace tracepoints are
5140 correctly defined in their own header and that they are created in
5141 one source file using the `CREATE_TRACE_POINTS` definition.
5143 To add an LTTng layer over an existing ftrace tracepoint:
5145 . Make sure the following kernel configuration options are
5151 * `CONFIG_HIGH_RES_TIMERS`
5152 * `CONFIG_TRACEPOINTS`
5155 . Build the Linux source tree with your custom ftrace tracepoints.
5156 . Boot the resulting Linux image on your target system.
5158 Confirm that the tracepoints exist by looking for their names in the
5159 dir:{/sys/kernel/debug/tracing/events/subsys} directory, where `subsys`
5160 is your subsystem's name.
5162 . Get a copy of the latest LTTng-modules{nbsp}{revision}:
5167 $ cd $(mktemp -d) &&
5168 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.10.tar.bz2 &&
5169 tar -xf lttng-modules-latest-2.10.tar.bz2 &&
5170 cd lttng-modules-2.10.*
5174 . In dir:{instrumentation/events/lttng-module}, relative to the root
5175 of the LTTng-modules source tree, create a header file named
5176 +__subsys__.h+ for your custom subsystem +__subsys__+ and write your
5177 LTTng-modules tracepoint definitions using the LTTng-modules
5180 Start with this template:
5184 .path:{instrumentation/events/lttng-module/my_subsys.h}
5187 #define TRACE_SYSTEM my_subsys
5189 #if !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ)
5190 #define _LTTNG_MY_SUBSYS_H
5192 #include "../../../probes/lttng-tracepoint-event.h"
5193 #include <linux/tracepoint.h>
5195 LTTNG_TRACEPOINT_EVENT(
5197 * Format is identical to TRACE_EVENT()'s version for the three
5198 * following macro parameters:
5201 TP_PROTO(int my_int, const char *my_string),
5202 TP_ARGS(my_int, my_string),
5204 /* LTTng-modules specific macros */
5206 ctf_integer(int, my_int_field, my_int)
5207 ctf_string(my_bar_field, my_bar)
5211 #endif /* !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ) */
5213 #include "../../../probes/define_trace.h"
5217 The entries in the `TP_FIELDS()` section are the list of fields for the
5218 LTTng tracepoint. This is similar to the `TP_STRUCT__entry()` part of
5219 ftrace's `TRACE_EVENT()` macro.
5221 See <<lttng-modules-tp-fields,Tracepoint fields macros>> for a
5222 complete description of the available `ctf_*()` macros.
5224 . Create the LTTng-modules probe's kernel module C source file,
5225 +probes/lttng-probe-__subsys__.c+, where +__subsys__+ is your
5230 .path:{probes/lttng-probe-my-subsys.c}
5232 #include <linux/module.h>
5233 #include "../lttng-tracer.h"
5236 * Build-time verification of mismatch between mainline
5237 * TRACE_EVENT() arguments and the LTTng-modules adaptation
5238 * layer LTTNG_TRACEPOINT_EVENT() arguments.
5240 #include <trace/events/my_subsys.h>
5242 /* Create LTTng tracepoint probes */
5243 #define LTTNG_PACKAGE_BUILD
5244 #define CREATE_TRACE_POINTS
5245 #define TRACE_INCLUDE_PATH ../instrumentation/events/lttng-module
5247 #include "../instrumentation/events/lttng-module/my_subsys.h"
5249 MODULE_LICENSE("GPL and additional rights");
5250 MODULE_AUTHOR("Your name <your-email>");
5251 MODULE_DESCRIPTION("LTTng my_subsys probes");
5252 MODULE_VERSION(__stringify(LTTNG_MODULES_MAJOR_VERSION) "."
5253 __stringify(LTTNG_MODULES_MINOR_VERSION) "."
5254 __stringify(LTTNG_MODULES_PATCHLEVEL_VERSION)
5255 LTTNG_MODULES_EXTRAVERSION);
5259 . Edit path:{probes/KBuild} and add your new kernel module object
5260 next to the existing ones:
5264 .path:{probes/KBuild}
5268 obj-m += lttng-probe-module.o
5269 obj-m += lttng-probe-power.o
5271 obj-m += lttng-probe-my-subsys.o
5277 . Build and install the LTTng kernel modules:
5282 $ make KERNELDIR=/path/to/linux
5283 # make modules_install && depmod -a
5287 Replace `/path/to/linux` with the path to the Linux source tree where
5288 you defined and used tracepoints with ftrace's `TRACE_EVENT()` macro.
5290 Note that you can also use the
5291 <<lttng-tracepoint-event-code,`LTTNG_TRACEPOINT_EVENT_CODE()` macro>>
5292 instead of `LTTNG_TRACEPOINT_EVENT()` to use custom local variables and
5293 C code that need to be executed before the event fields are recorded.
5295 The best way to learn how to use the previous LTTng-modules macros is to
5296 inspect the existing LTTng-modules tracepoint definitions in the
5297 dir:{instrumentation/events/lttng-module} header files. Compare them
5298 with the Linux kernel mainline versions in the
5299 dir:{include/trace/events} directory of the Linux source tree.
5303 [[lttng-tracepoint-event-code]]
5304 ===== Use custom C code to access the data for tracepoint fields
5306 Although we recommended to always use the
5307 <<lttng-adaptation-layer,`LTTNG_TRACEPOINT_EVENT()`>> macro to describe
5308 the arguments and fields of an LTTng-modules tracepoint when possible,
5309 sometimes you need a more complex process to access the data that the
5310 tracer records as event record fields. In other words, you need local
5311 variables and multiple C{nbsp}statements instead of simple
5312 argument-based expressions that you pass to the
5313 <<lttng-modules-tp-fields,`ctf_*()` macros of `TP_FIELDS()`>>.
5315 You can use the `LTTNG_TRACEPOINT_EVENT_CODE()` macro instead of
5316 `LTTNG_TRACEPOINT_EVENT()` to declare custom local variables and define
5317 a block of C{nbsp}code to be executed before LTTng records the fields.
5318 The structure of this macro is:
5321 .`LTTNG_TRACEPOINT_EVENT_CODE()` macro syntax.
5323 LTTNG_TRACEPOINT_EVENT_CODE(
5325 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5326 * version for the following three macro parameters:
5329 TP_PROTO(int my_int, const char *my_string),
5330 TP_ARGS(my_int, my_string),
5332 /* Declarations of custom local variables */
5335 unsigned long b = 0;
5336 const char *name = "(undefined)";
5337 struct my_struct *my_struct;
5341 * Custom code which uses both tracepoint arguments
5342 * (in TP_ARGS()) and local variables (in TP_locvar()).
5344 * Local variables are actually members of a structure pointed
5345 * to by the special variable tp_locvar.
5349 tp_locvar->a = my_int + 17;
5350 tp_locvar->my_struct = get_my_struct_at(tp_locvar->a);
5351 tp_locvar->b = my_struct_compute_b(tp_locvar->my_struct);
5352 tp_locvar->name = my_struct_get_name(tp_locvar->my_struct);
5353 put_my_struct(tp_locvar->my_struct);
5362 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5363 * version for this, except that tp_locvar members can be
5364 * used in the argument expression parameters of
5365 * the ctf_*() macros.
5368 ctf_integer(unsigned long, my_struct_b, tp_locvar->b)
5369 ctf_integer(int, my_struct_a, tp_locvar->a)
5370 ctf_string(my_string_field, my_string)
5371 ctf_string(my_struct_name, tp_locvar->name)
5376 IMPORTANT: The C code defined in `TP_code()` must not have any side
5377 effects when executed. In particular, the code must not allocate
5378 memory or get resources without deallocating this memory or putting
5379 those resources afterwards.
5382 [[instrumenting-linux-kernel-tracing]]
5383 ==== Load and unload a custom probe kernel module
5385 You must load a <<lttng-adaptation-layer,created LTTng-modules probe
5386 kernel module>> in the kernel before it can emit LTTng events.
5388 To load the default probe kernel modules and a custom probe kernel
5391 * Use the opt:lttng-sessiond(8):--extra-kmod-probes option to give extra
5392 probe modules to load when starting a root <<lttng-sessiond,session
5396 .Load the `my_subsys`, `usb`, and the default probe modules.
5400 # lttng-sessiond --extra-kmod-probes=my_subsys,usb
5405 You only need to pass the subsystem name, not the whole kernel module
5408 To load _only_ a given custom probe kernel module:
5410 * Use the opt:lttng-sessiond(8):--kmod-probes option to give the probe
5411 modules to load when starting a root session daemon:
5414 .Load only the `my_subsys` and `usb` probe modules.
5418 # lttng-sessiond --kmod-probes=my_subsys,usb
5423 To confirm that a probe module is loaded:
5430 $ lsmod | grep lttng_probe_usb
5434 To unload the loaded probe modules:
5436 * Kill the session daemon with `SIGTERM`:
5441 # pkill lttng-sessiond
5445 You can also use man:modprobe(8)'s `--remove` option if the session
5446 daemon terminates abnormally.
5449 [[controlling-tracing]]
5452 Once an application or a Linux kernel is
5453 <<instrumenting,instrumented>> for LTTng tracing,
5456 This section is divided in topics on how to use the various
5457 <<plumbing,components of LTTng>>, in particular the <<lttng-cli,cmd:lttng
5458 command-line tool>>, to _control_ the LTTng daemons and tracers.
5460 NOTE: In the following subsections, we refer to an man:lttng(1) command
5461 using its man page name. For example, instead of _Run the `create`
5462 command to..._, we use _Run the man:lttng-create(1) command to..._.
5466 === Start a session daemon
5468 In some situations, you need to run a <<lttng-sessiond,session daemon>>
5469 (man:lttng-sessiond(8)) _before_ you can use the man:lttng(1)
5472 You will see the following error when you run a command while no session
5476 Error: No session daemon is available
5479 The only command that automatically runs a session daemon is
5480 man:lttng-create(1), which you use to
5481 <<creating-destroying-tracing-sessions,create a tracing session>>. While
5482 this is most of the time the first operation that you do, sometimes it's
5483 not. Some examples are:
5485 * <<list-instrumentation-points,List the available instrumentation points>>.
5486 * <<saving-loading-tracing-session,Load a tracing session configuration>>.
5488 [[tracing-group]] Each Unix user must have its own running session
5489 daemon to trace user applications. The session daemon that the root user
5490 starts is the only one allowed to control the LTTng kernel tracer. Users
5491 that are part of the _tracing group_ can control the root session
5492 daemon. The default tracing group name is `tracing`; you can set it to
5493 something else with the opt:lttng-sessiond(8):--group option when you
5494 start the root session daemon.
5496 To start a user session daemon:
5498 * Run man:lttng-sessiond(8):
5503 $ lttng-sessiond --daemonize
5507 To start the root session daemon:
5509 * Run man:lttng-sessiond(8) as the root user:
5514 # lttng-sessiond --daemonize
5518 In both cases, remove the opt:lttng-sessiond(8):--daemonize option to
5519 start the session daemon in foreground.
5521 To stop a session daemon, use man:kill(1) on its process ID (standard
5524 Note that some Linux distributions could manage the LTTng session daemon
5525 as a service. In this case, you should use the service manager to
5526 start, restart, and stop session daemons.
5529 [[creating-destroying-tracing-sessions]]
5530 === Create and destroy a tracing session
5532 Almost all the LTTng control operations happen in the scope of
5533 a <<tracing-session,tracing session>>, which is the dialogue between the
5534 <<lttng-sessiond,session daemon>> and you.
5536 To create a tracing session with a generated name:
5538 * Use the man:lttng-create(1) command:
5547 The created tracing session's name is `auto` followed by the
5550 To create a tracing session with a specific name:
5552 * Use the optional argument of the man:lttng-create(1) command:
5557 $ lttng create my-session
5561 Replace `my-session` with the specific tracing session name.
5563 LTTng appends the creation date to the created tracing session's name.
5565 LTTng writes the traces of a tracing session in
5566 +$LTTNG_HOME/lttng-trace/__name__+ by default, where +__name__+ is the
5567 name of the tracing session. Note that the env:LTTNG_HOME environment
5568 variable defaults to `$HOME` if not set.
5570 To output LTTng traces to a non-default location:
5572 * Use the opt:lttng-create(1):--output option of the man:lttng-create(1) command:
5577 $ lttng create my-session --output=/tmp/some-directory
5581 You may create as many tracing sessions as you wish.
5583 To list all the existing tracing sessions for your Unix user:
5585 * Use the man:lttng-list(1) command:
5594 When you create a tracing session, it is set as the _current tracing
5595 session_. The following man:lttng(1) commands operate on the current
5596 tracing session when you don't specify one:
5598 [role="list-3-cols"]
5615 To change the current tracing session:
5617 * Use the man:lttng-set-session(1) command:
5622 $ lttng set-session new-session
5626 Replace `new-session` by the name of the new current tracing session.
5628 When you are done tracing in a given tracing session, you can destroy
5629 it. This operation frees the resources taken by the tracing session
5630 to destroy; it does not destroy the trace data that LTTng wrote for
5631 this tracing session.
5633 To destroy the current tracing session:
5635 * Use the man:lttng-destroy(1) command:
5645 [[list-instrumentation-points]]
5646 === List the available instrumentation points
5648 The <<lttng-sessiond,session daemon>> can query the running instrumented
5649 user applications and the Linux kernel to get a list of available
5650 instrumentation points. For the Linux kernel <<domain,tracing domain>>,
5651 they are tracepoints and system calls. For the user space tracing
5652 domain, they are tracepoints. For the other tracing domains, they are
5655 To list the available instrumentation points:
5657 * Use the man:lttng-list(1) command with the requested tracing domain's
5661 * opt:lttng-list(1):--kernel: Linux kernel tracepoints (your Unix user
5662 must be a root user, or it must be a member of the
5663 <<tracing-group,tracing group>>).
5664 * opt:lttng-list(1):--kernel with opt:lttng-list(1):--syscall: Linux
5665 kernel system calls (your Unix user must be a root user, or it must be
5666 a member of the tracing group).
5667 * opt:lttng-list(1):--userspace: user space tracepoints.
5668 * opt:lttng-list(1):--jul: `java.util.logging` loggers.
5669 * opt:lttng-list(1):--log4j: Apache log4j loggers.
5670 * opt:lttng-list(1):--python: Python loggers.
5673 .List the available user space tracepoints.
5677 $ lttng list --userspace
5681 .List the available Linux kernel system call tracepoints.
5685 $ lttng list --kernel --syscall
5690 [[enabling-disabling-events]]
5691 === Create and enable an event rule
5693 Once you <<creating-destroying-tracing-sessions,create a tracing
5694 session>>, you can create <<event,event rules>> with the
5695 man:lttng-enable-event(1) command.
5697 You specify each condition with a command-line option. The available
5698 condition options are shown in the following table.
5700 [role="growable",cols="asciidoc,asciidoc,default"]
5701 .Condition command-line options for the man:lttng-enable-event(1) command.
5703 |Option |Description |Applicable tracing domains
5709 . +--probe=__ADDR__+
5710 . +--function=__ADDR__+
5713 Instead of using the default _tracepoint_ instrumentation type, use:
5715 . A Linux system call.
5716 . A Linux https://lwn.net/Articles/132196/[KProbe] (symbol or address).
5717 . The entry and return points of a Linux function (symbol or address).
5721 |First positional argument.
5724 Tracepoint or system call name. In the case of a Linux KProbe or
5725 function, this is a custom name given to the event rule. With the
5726 JUL, log4j, and Python domains, this is a logger name.
5728 With a tracepoint, logger, or system call name, the last character
5729 can be `*` to match anything that remains.
5736 . +--loglevel=__LEVEL__+
5737 . +--loglevel-only=__LEVEL__+
5740 . Match only tracepoints or log statements with a logging level at
5741 least as severe as +__LEVEL__+.
5742 . Match only tracepoints or log statements with a logging level
5743 equal to +__LEVEL__+.
5745 See man:lttng-enable-event(1) for the list of available logging level
5748 |User space, JUL, log4j, and Python.
5750 |+--exclude=__EXCLUSIONS__+
5753 When you use a `*` character at the end of the tracepoint or logger
5754 name (first positional argument), exclude the specific names in the
5755 comma-delimited list +__EXCLUSIONS__+.
5758 User space, JUL, log4j, and Python.
5760 |+--filter=__EXPR__+
5763 Match only events which satisfy the expression +__EXPR__+.
5765 See man:lttng-enable-event(1) to learn more about the syntax of a
5772 You attach an event rule to a <<channel,channel>> on creation. If you do
5773 not specify the channel with the opt:lttng-enable-event(1):--channel
5774 option, and if the event rule to create is the first in its
5775 <<domain,tracing domain>> for a given tracing session, then LTTng
5776 creates a _default channel_ for you. This default channel is reused in
5777 subsequent invocations of the man:lttng-enable-event(1) command for the
5778 same tracing domain.
5780 An event rule is always enabled at creation time.
5782 The following examples show how you can combine the previous
5783 command-line options to create simple to more complex event rules.
5785 .Create an event rule targetting a Linux kernel tracepoint (default channel).
5789 $ lttng enable-event --kernel sched_switch
5793 .Create an event rule matching four Linux kernel system calls (default channel).
5797 $ lttng enable-event --kernel --syscall open,write,read,close
5801 .Create event rules matching tracepoints with filter expressions (default channel).
5805 $ lttng enable-event --kernel sched_switch --filter='prev_comm == "bash"'
5810 $ lttng enable-event --kernel --all \
5811 --filter='$ctx.tid == 1988 || $ctx.tid == 1534'
5816 $ lttng enable-event --jul my_logger \
5817 --filter='$app.retriever:cur_msg_id > 3'
5820 IMPORTANT: Make sure to always quote the filter string when you
5821 use man:lttng(1) from a shell.
5824 .Create an event rule matching any user space tracepoint of a given tracepoint provider with a log level range (default channel).
5828 $ lttng enable-event --userspace my_app:'*' --loglevel=TRACE_INFO
5831 IMPORTANT: Make sure to always quote the wildcard character when you
5832 use man:lttng(1) from a shell.
5835 .Create an event rule matching multiple Python loggers with a wildcard and with exclusions (default channel).
5839 $ lttng enable-event --python my-app.'*' \
5840 --exclude='my-app.module,my-app.hello'
5844 .Create an event rule matching any Apache log4j logger with a specific log level (default channel).
5848 $ lttng enable-event --log4j --all --loglevel-only=LOG4J_WARN
5852 .Create an event rule attached to a specific channel matching a specific user space tracepoint provider and tracepoint.
5856 $ lttng enable-event --userspace my_app:my_tracepoint --channel=my-channel
5860 The event rules of a given channel form a whitelist: as soon as an
5861 emitted event passes one of them, LTTng can record the event. For
5862 example, an event named `my_app:my_tracepoint` emitted from a user space
5863 tracepoint with a `TRACE_ERROR` log level passes both of the following
5868 $ lttng enable-event --userspace my_app:my_tracepoint
5869 $ lttng enable-event --userspace my_app:my_tracepoint \
5870 --loglevel=TRACE_INFO
5873 The second event rule is redundant: the first one includes
5877 [[disable-event-rule]]
5878 === Disable an event rule
5880 To disable an event rule that you <<enabling-disabling-events,created>>
5881 previously, use the man:lttng-disable-event(1) command. This command
5882 disables _all_ the event rules (of a given tracing domain and channel)
5883 which match an instrumentation point. The other conditions are not
5884 supported as of LTTng{nbsp}{revision}.
5886 The LTTng tracer does not record an emitted event which passes
5887 a _disabled_ event rule.
5889 .Disable an event rule matching a Python logger (default channel).
5893 $ lttng disable-event --python my-logger
5897 .Disable an event rule matching all `java.util.logging` loggers (default channel).
5901 $ lttng disable-event --jul '*'
5905 .Disable _all_ the event rules of the default channel.
5907 The opt:lttng-disable-event(1):--all-events option is not, like the
5908 opt:lttng-enable-event(1):--all option of man:lttng-enable-event(1), the
5909 equivalent of the event name `*` (wildcard): it disables _all_ the event
5910 rules of a given channel.
5914 $ lttng disable-event --jul --all-events
5918 NOTE: You cannot delete an event rule once you create it.
5922 === Get the status of a tracing session
5924 To get the status of the current tracing session, that is, its
5925 parameters, its channels, event rules, and their attributes:
5927 * Use the man:lttng-status(1) command:
5937 To get the status of any tracing session:
5939 * Use the man:lttng-list(1) command with the tracing session's name:
5944 $ lttng list my-session
5948 Replace `my-session` with the desired tracing session's name.
5951 [[basic-tracing-session-control]]
5952 === Start and stop a tracing session
5954 Once you <<creating-destroying-tracing-sessions,create a tracing
5956 <<enabling-disabling-events,create one or more event rules>>,
5957 you can start and stop the tracers for this tracing session.
5959 To start tracing in the current tracing session:
5961 * Use the man:lttng-start(1) command:
5970 LTTng is very flexible: you can launch user applications before
5971 or after the you start the tracers. The tracers only record the events
5972 if they pass enabled event rules and if they occur while the tracers are
5975 To stop tracing in the current tracing session:
5977 * Use the man:lttng-stop(1) command:
5986 If there were <<channel-overwrite-mode-vs-discard-mode,lost event
5987 records>> or lost sub-buffers since the last time you ran
5988 man:lttng-start(1), warnings are printed when you run the
5989 man:lttng-stop(1) command.
5992 [[enabling-disabling-channels]]
5993 === Create a channel
5995 Once you create a tracing session, you can create a <<channel,channel>>
5996 with the man:lttng-enable-channel(1) command.
5998 Note that LTTng automatically creates a default channel when, for a
5999 given <<domain,tracing domain>>, no channels exist and you
6000 <<enabling-disabling-events,create>> the first event rule. This default
6001 channel is named `channel0` and its attributes are set to reasonable
6002 values. Therefore, you only need to create a channel when you need
6003 non-default attributes.
6005 You specify each non-default channel attribute with a command-line
6006 option when you use the man:lttng-enable-channel(1) command. The
6007 available command-line options are:
6009 [role="growable",cols="asciidoc,asciidoc"]
6010 .Command-line options for the man:lttng-enable-channel(1) command.
6012 |Option |Description
6018 <<channel-overwrite-mode-vs-discard-mode,event loss mode>> instead of
6019 the default _discard_ mode.
6021 |`--buffers-pid` (user space tracing domain only)
6024 Use the per-process <<channel-buffering-schemes,buffering scheme>>
6025 instead of the default per-user buffering scheme.
6027 |+--subbuf-size=__SIZE__+
6030 Allocate sub-buffers of +__SIZE__+ bytes (power of two), for each CPU,
6031 either for each Unix user (default), or for each instrumented process.
6033 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6035 |+--num-subbuf=__COUNT__+
6038 Allocate +__COUNT__+ sub-buffers (power of two), for each CPU, either
6039 for each Unix user (default), or for each instrumented process.
6041 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6043 |+--tracefile-size=__SIZE__+
6046 Set the maximum size of each trace file that this channel writes within
6047 a stream to +__SIZE__+ bytes instead of no maximum.
6049 See <<tracefile-rotation,Trace file count and size>>.
6051 |+--tracefile-count=__COUNT__+
6054 Limit the number of trace files that this channel creates to
6055 +__COUNT__+ channels instead of no limit.
6057 See <<tracefile-rotation,Trace file count and size>>.
6059 |+--switch-timer=__PERIODUS__+
6062 Set the <<channel-switch-timer,switch timer period>>
6063 to +__PERIODUS__+{nbsp}µs.
6065 |+--read-timer=__PERIODUS__+
6068 Set the <<channel-read-timer,read timer period>>
6069 to +__PERIODUS__+{nbsp}µs.
6071 |[[opt-blocking-timeout]]+--blocking-timeout=__TIMEOUTUS__+
6074 Set the timeout of user space applications which load LTTng-UST
6075 in blocking mode to +__TIMEOUTUS__+:
6078 Never block (non-blocking mode).
6081 Block forever until space is available in a sub-buffer to record
6084 __n__, a positive value::
6085 Wait for at most __n__ µs when trying to write into a sub-buffer.
6087 Note that, for this option to have any effect on an instrumented
6088 user space application, you need to run the application with a set
6089 env:LTTNG_UST_ALLOW_BLOCKING environment variable.
6091 |+--output=__TYPE__+ (Linux kernel tracing domain only)
6094 Set the channel's output type to +__TYPE__+, either `mmap` or `splice`.
6098 You can only create a channel in the Linux kernel and user space
6099 <<domain,tracing domains>>: other tracing domains have their own channel
6100 created on the fly when <<enabling-disabling-events,creating event
6105 Because of a current LTTng limitation, you must create all channels
6106 _before_ you <<basic-tracing-session-control,start tracing>> in a given
6107 tracing session, that is, before the first time you run
6110 Since LTTng automatically creates a default channel when you use the
6111 man:lttng-enable-event(1) command with a specific tracing domain, you
6112 cannot, for example, create a Linux kernel event rule, start tracing,
6113 and then create a user space event rule, because no user space channel
6114 exists yet and it's too late to create one.
6116 For this reason, make sure to configure your channels properly
6117 before starting the tracers for the first time!
6120 The following examples show how you can combine the previous
6121 command-line options to create simple to more complex channels.
6123 .Create a Linux kernel channel with default attributes.
6127 $ lttng enable-channel --kernel my-channel
6131 .Create a user space channel with 4 sub-buffers or 1{nbsp}MiB each, per CPU, per instrumented process.
6135 $ lttng enable-channel --userspace --num-subbuf=4 --subbuf-size=1M \
6136 --buffers-pid my-channel
6140 .[[blocking-timeout-example]]Create a default user space channel with an infinite blocking timeout.
6142 <<creating-destroying-tracing-sessions,Create a tracing-session>>,
6143 create the channel, <<enabling-disabling-events,create an event rule>>,
6144 and <<basic-tracing-session-control,start tracing>>:
6149 $ lttng enable-channel --userspace --blocking-timeout=inf blocking-channel
6150 $ lttng enable-event --userspace --channel=blocking-channel --all
6154 Run an application instrumented with LTTng-UST and allow it to block:
6158 $ LTTNG_UST_ALLOW_BLOCKING=1 my-app
6162 .Create a Linux kernel channel which rotates 8 trace files of 4{nbsp}MiB each for each stream
6166 $ lttng enable-channel --kernel --tracefile-count=8 \
6167 --tracefile-size=4194304 my-channel
6171 .Create a user space channel in overwrite (or _flight recorder_) mode.
6175 $ lttng enable-channel --userspace --overwrite my-channel
6179 You can <<enabling-disabling-events,create>> the same event rule in
6180 two different channels:
6184 $ lttng enable-event --userspace --channel=my-channel app:tp
6185 $ lttng enable-event --userspace --channel=other-channel app:tp
6188 If both channels are enabled, when a tracepoint named `app:tp` is
6189 reached, LTTng records two events, one for each channel.
6193 === Disable a channel
6195 To disable a specific channel that you <<enabling-disabling-channels,created>>
6196 previously, use the man:lttng-disable-channel(1) command.
6198 .Disable a specific Linux kernel channel.
6202 $ lttng disable-channel --kernel my-channel
6206 The state of a channel precedes the individual states of event rules
6207 attached to it: event rules which belong to a disabled channel, even if
6208 they are enabled, are also considered disabled.
6212 === Add context fields to a channel
6214 Event record fields in trace files provide important information about
6215 events that occured previously, but sometimes some external context may
6216 help you solve a problem faster. Examples of context fields are:
6218 * The **process ID**, **thread ID**, **process name**, and
6219 **process priority** of the thread in which the event occurs.
6220 * The **hostname** of the system on which the event occurs.
6221 * The current values of many possible **performance counters** using
6223 ** CPU cycles, stalled cycles, idle cycles, and the other cycle types.
6225 ** Branch instructions, misses, and loads.
6227 * Any context defined at the application level (supported for the
6228 JUL and log4j <<domain,tracing domains>>).
6230 To get the full list of available context fields, see
6231 `lttng add-context --list`. Some context fields are reserved for a
6232 specific <<domain,tracing domain>> (Linux kernel or user space).
6234 You add context fields to <<channel,channels>>. All the events
6235 that a channel with added context fields records contain those fields.
6237 To add context fields to one or all the channels of a given tracing
6240 * Use the man:lttng-add-context(1) command.
6242 .Add context fields to all the channels of the current tracing session.
6244 The following command line adds the virtual process identifier and
6245 the per-thread CPU cycles count fields to all the user space channels
6246 of the current tracing session.
6250 $ lttng add-context --userspace --type=vpid --type=perf:thread:cpu-cycles
6254 .Add performance counter context fields by raw ID
6256 See man:lttng-add-context(1) for the exact format of the context field
6257 type, which is partly compatible with the format used in
6262 $ lttng add-context --userspace --type=perf:thread:raw:r0110:test
6263 $ lttng add-context --kernel --type=perf:cpu:raw:r0013c:x86unhalted
6267 .Add a context field to a specific channel.
6269 The following command line adds the thread identifier context field
6270 to the Linux kernel channel named `my-channel` in the current
6275 $ lttng add-context --kernel --channel=my-channel --type=tid
6279 .Add an application-specific context field to a specific channel.
6281 The following command line adds the `cur_msg_id` context field of the
6282 `retriever` context retriever for all the instrumented
6283 <<java-application,Java applications>> recording <<event,event records>>
6284 in the channel named `my-channel`:
6288 $ lttng add-context --kernel --channel=my-channel \
6289 --type='$app:retriever:cur_msg_id'
6292 IMPORTANT: Make sure to always quote the `$` character when you
6293 use man:lttng-add-context(1) from a shell.
6296 NOTE: You cannot remove context fields from a channel once you add it.
6301 === Track process IDs
6303 It's often useful to allow only specific process IDs (PIDs) to emit
6304 events. For example, you may wish to record all the system calls made by
6305 a given process (Ă la http://linux.die.net/man/1/strace[strace]).
6307 The man:lttng-track(1) and man:lttng-untrack(1) commands serve this
6308 purpose. Both commands operate on a whitelist of process IDs. You _add_
6309 entries to this whitelist with the man:lttng-track(1) command and remove
6310 entries with the man:lttng-untrack(1) command. Any process which has one
6311 of the PIDs in the whitelist is allowed to emit LTTng events which pass
6312 an enabled <<event,event rule>>.
6314 NOTE: The PID tracker tracks the _numeric process IDs_. Should a
6315 process with a given tracked ID exit and another process be given this
6316 ID, then the latter would also be allowed to emit events.
6318 .Track and untrack process IDs.
6320 For the sake of the following example, assume the target system has 16
6324 <<creating-destroying-tracing-sessions,create a tracing session>>,
6325 the whitelist contains all the possible PIDs:
6328 .All PIDs are tracked.
6329 image::track-all.png[]
6331 When the whitelist is full and you use the man:lttng-track(1) command to
6332 specify some PIDs to track, LTTng first clears the whitelist, then it
6333 tracks the specific PIDs. After:
6337 $ lttng track --pid=3,4,7,10,13
6343 .PIDs 3, 4, 7, 10, and 13 are tracked.
6344 image::track-3-4-7-10-13.png[]
6346 You can add more PIDs to the whitelist afterwards:
6350 $ lttng track --pid=1,15,16
6356 .PIDs 1, 15, and 16 are added to the whitelist.
6357 image::track-1-3-4-7-10-13-15-16.png[]
6359 The man:lttng-untrack(1) command removes entries from the PID tracker's
6360 whitelist. Given the previous example, the following command:
6364 $ lttng untrack --pid=3,7,10,13
6367 leads to this whitelist:
6370 .PIDs 3, 7, 10, and 13 are removed from the whitelist.
6371 image::track-1-4-15-16.png[]
6373 LTTng can track all possible PIDs again using the opt:track(1):--all
6378 $ lttng track --pid --all
6381 The result is, again:
6384 .All PIDs are tracked.
6385 image::track-all.png[]
6388 .Track only specific PIDs
6390 A very typical use case with PID tracking is to start with an empty
6391 whitelist, then <<basic-tracing-session-control,start the tracers>>, and
6392 then add PIDs manually while tracers are active. You can accomplish this
6393 by using the opt:lttng-untrack(1):--all option of the
6394 man:lttng-untrack(1) command to clear the whitelist after you
6395 <<creating-destroying-tracing-sessions,create a tracing session>>:
6399 $ lttng untrack --pid --all
6405 .No PIDs are tracked.
6406 image::untrack-all.png[]
6408 If you trace with this whitelist configuration, the tracer records no
6409 events for this <<domain,tracing domain>> because no processes are
6410 tracked. You can use the man:lttng-track(1) command as usual to track
6411 specific PIDs, for example:
6415 $ lttng track --pid=6,11
6421 .PIDs 6 and 11 are tracked.
6422 image::track-6-11.png[]
6427 [[saving-loading-tracing-session]]
6428 === Save and load tracing session configurations
6430 Configuring a <<tracing-session,tracing session>> can be long. Some of
6431 the tasks involved are:
6433 * <<enabling-disabling-channels,Create channels>> with
6434 specific attributes.
6435 * <<adding-context,Add context fields>> to specific channels.
6436 * <<enabling-disabling-events,Create event rules>> with specific log
6437 level and filter conditions.
6439 If you use LTTng to solve real world problems, chances are you have to
6440 record events using the same tracing session setup over and over,
6441 modifying a few variables each time in your instrumented program
6442 or environment. To avoid constant tracing session reconfiguration,
6443 the man:lttng(1) command-line tool can save and load tracing session
6444 configurations to/from XML files.
6446 To save a given tracing session configuration:
6448 * Use the man:lttng-save(1) command:
6453 $ lttng save my-session
6457 Replace `my-session` with the name of the tracing session to save.
6459 LTTng saves tracing session configurations to
6460 dir:{$LTTNG_HOME/.lttng/sessions} by default. Note that the
6461 env:LTTNG_HOME environment variable defaults to `$HOME` if not set. Use
6462 the opt:lttng-save(1):--output-path option to change this destination
6465 LTTng saves all configuration parameters, for example:
6467 * The tracing session name.
6468 * The trace data output path.
6469 * The channels with their state and all their attributes.
6470 * The context fields you added to channels.
6471 * The event rules with their state, log level and filter conditions.
6473 To load a tracing session:
6475 * Use the man:lttng-load(1) command:
6480 $ lttng load my-session
6484 Replace `my-session` with the name of the tracing session to load.
6486 When LTTng loads a configuration, it restores your saved tracing session
6487 as if you just configured it manually.
6489 See man:lttng(1) for the complete list of command-line options. You
6490 can also save and load all many sessions at a time, and decide in which
6491 directory to output the XML files.
6494 [[sending-trace-data-over-the-network]]
6495 === Send trace data over the network
6497 LTTng can send the recorded trace data to a remote system over the
6498 network instead of writing it to the local file system.
6500 To send the trace data over the network:
6502 . On the _remote_ system (which can also be the target system),
6503 start an LTTng <<lttng-relayd,relay daemon>> (man:lttng-relayd(8)):
6512 . On the _target_ system, create a tracing session configured to
6513 send trace data over the network:
6518 $ lttng create my-session --set-url=net://remote-system
6522 Replace `remote-system` by the host name or IP address of the
6523 remote system. See man:lttng-create(1) for the exact URL format.
6525 . On the target system, use the man:lttng(1) command-line tool as usual.
6526 When tracing is active, the target's consumer daemon sends sub-buffers
6527 to the relay daemon running on the remote system instead of flushing
6528 them to the local file system. The relay daemon writes the received
6529 packets to the local file system.
6531 The relay daemon writes trace files to
6532 +$LTTNG_HOME/lttng-traces/__hostname__/__session__+ by default, where
6533 +__hostname__+ is the host name of the target system and +__session__+
6534 is the tracing session name. Note that the env:LTTNG_HOME environment
6535 variable defaults to `$HOME` if not set. Use the
6536 opt:lttng-relayd(8):--output option of man:lttng-relayd(8) to write
6537 trace files to another base directory.
6542 === View events as LTTng emits them (noch:{LTTng} live)
6544 LTTng live is a network protocol implemented by the <<lttng-relayd,relay
6545 daemon>> (man:lttng-relayd(8)) to allow compatible trace viewers to
6546 display events as LTTng emits them on the target system while tracing is
6549 The relay daemon creates a _tee_: it forwards the trace data to both
6550 the local file system and to connected live viewers:
6553 .The relay daemon creates a _tee_, forwarding the trace data to both trace files and a connected live viewer.
6558 . On the _target system_, create a <<tracing-session,tracing session>>
6564 $ lttng create my-session --live
6568 This spawns a local relay daemon.
6570 . Start the live viewer and configure it to connect to the relay
6571 daemon. For example, with http://diamon.org/babeltrace[Babeltrace]:
6576 $ babeltrace --input-format=lttng-live \
6577 net://localhost/host/hostname/my-session
6584 * `hostname` with the host name of the target system.
6585 * `my-session` with the name of the tracing session to view.
6588 . Configure the tracing session as usual with the man:lttng(1)
6589 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6591 You can list the available live tracing sessions with Babeltrace:
6595 $ babeltrace --input-format=lttng-live net://localhost
6598 You can start the relay daemon on another system. In this case, you need
6599 to specify the relay daemon's URL when you create the tracing session
6600 with the opt:lttng-create(1):--set-url option. You also need to replace
6601 `localhost` in the procedure above with the host name of the system on
6602 which the relay daemon is running.
6604 See man:lttng-create(1) and man:lttng-relayd(8) for the complete list of
6605 command-line options.
6609 [[taking-a-snapshot]]
6610 === Take a snapshot of the current sub-buffers of a tracing session
6612 The normal behavior of LTTng is to append full sub-buffers to growing
6613 trace data files. This is ideal to keep a full history of the events
6614 that occurred on the target system, but it can
6615 represent too much data in some situations. For example, you may wish
6616 to trace your application continuously until some critical situation
6617 happens, in which case you only need the latest few recorded
6618 events to perform the desired analysis, not multi-gigabyte trace files.
6620 With the man:lttng-snapshot(1) command, you can take a snapshot of the
6621 current sub-buffers of a given <<tracing-session,tracing session>>.
6622 LTTng can write the snapshot to the local file system or send it over
6627 . Create a tracing session in _snapshot mode_:
6632 $ lttng create my-session --snapshot
6636 The <<channel-overwrite-mode-vs-discard-mode,event loss mode>> of
6637 <<channel,channels>> created in this mode is automatically set to
6638 _overwrite_ (flight recorder mode).
6640 . Configure the tracing session as usual with the man:lttng(1)
6641 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6643 . **Optional**: When you need to take a snapshot,
6644 <<basic-tracing-session-control,stop tracing>>.
6646 You can take a snapshot when the tracers are active, but if you stop
6647 them first, you are sure that the data in the sub-buffers does not
6648 change before you actually take the snapshot.
6655 $ lttng snapshot record --name=my-first-snapshot
6659 LTTng writes the current sub-buffers of all the current tracing
6660 session's channels to trace files on the local file system. Those trace
6661 files have `my-first-snapshot` in their name.
6663 There is no difference between the format of a normal trace file and the
6664 format of a snapshot: viewers of LTTng traces also support LTTng
6667 By default, LTTng writes snapshot files to the path shown by
6668 `lttng snapshot list-output`. You can change this path or decide to send
6669 snapshots over the network using either:
6671 . An output path or URL that you specify when you create the
6673 . An snapshot output path or URL that you add using
6674 `lttng snapshot add-output`
6675 . An output path or URL that you provide directly to the
6676 `lttng snapshot record` command.
6678 Method 3 overrides method 2, which overrides method 1. When you
6679 specify a URL, a relay daemon must listen on a remote system (see
6680 <<sending-trace-data-over-the-network,Send trace data over the network>>).
6685 === Use the machine interface
6687 With any command of the man:lttng(1) command-line tool, you can set the
6688 opt:lttng(1):--mi option to `xml` (before the command name) to get an
6689 XML machine interface output, for example:
6693 $ lttng --mi=xml enable-event --kernel --syscall open
6696 A schema definition (XSD) is
6697 https://github.com/lttng/lttng-tools/blob/stable-2.10/src/common/mi-lttng-3.0.xsd[available]
6698 to ease the integration with external tools as much as possible.
6702 [[metadata-regenerate]]
6703 === Regenerate the metadata of an LTTng trace
6705 An LTTng trace, which is a http://diamon.org/ctf[CTF] trace, has both
6706 data stream files and a metadata file. This metadata file contains,
6707 amongst other things, information about the offset of the clock sources
6708 used to timestamp <<event,event records>> when tracing.
6710 If, once a <<tracing-session,tracing session>> is
6711 <<basic-tracing-session-control,started>>, a major
6712 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
6713 happens, the trace's clock offset also needs to be updated. You
6714 can use the `metadata` item of the man:lttng-regenerate(1) command
6717 The main use case of this command is to allow a system to boot with
6718 an incorrect wall time and trace it with LTTng before its wall time
6719 is corrected. Once the system is known to be in a state where its
6720 wall time is correct, it can run `lttng regenerate metadata`.
6722 To regenerate the metadata of an LTTng trace:
6724 * Use the `metadata` item of the man:lttng-regenerate(1) command:
6729 $ lttng regenerate metadata
6735 `lttng regenerate metadata` has the following limitations:
6737 * Tracing session <<creating-destroying-tracing-sessions,created>>
6739 * User space <<channel,channels>>, if any, are using
6740 <<channel-buffering-schemes,per-user buffering>>.
6745 [[regenerate-statedump]]
6746 === Regenerate the state dump of a tracing session
6748 The LTTng kernel and user space tracers generate state dump
6749 <<event,event records>> when the application starts or when you
6750 <<basic-tracing-session-control,start a tracing session>>. An analysis
6751 can use the state dump event records to set an initial state before it
6752 builds the rest of the state from the following event records.
6753 http://tracecompass.org/[Trace Compass] is a notable example of an
6754 application which uses the state dump of an LTTng trace.
6756 When you <<taking-a-snapshot,take a snapshot>>, it's possible that the
6757 state dump event records are not included in the snapshot because they
6758 were recorded to a sub-buffer that has been consumed or overwritten
6761 You can use the `lttng regenerate statedump` command to emit the state
6762 dump event records again.
6764 To regenerate the state dump of the current tracing session, provided
6765 create it in snapshot mode, before you take a snapshot:
6767 . Use the `statedump` item of the man:lttng-regenerate(1) command:
6772 $ lttng regenerate statedump
6776 . <<basic-tracing-session-control,Stop the tracing session>>:
6785 . <<taking-a-snapshot,Take a snapshot>>:
6790 $ lttng snapshot record --name=my-snapshot
6794 Depending on the event throughput, you should run steps 1 and 2
6795 as closely as possible.
6797 NOTE: To record the state dump events, you need to
6798 <<enabling-disabling-events,create event rules>> which enable them.
6799 LTTng-UST state dump tracepoints start with `lttng_ust_statedump:`.
6800 LTTng-modules state dump tracepoints start with `lttng_statedump_`.
6804 [[persistent-memory-file-systems]]
6805 === Record trace data on persistent memory file systems
6807 https://en.wikipedia.org/wiki/Non-volatile_random-access_memory[Non-volatile random-access memory]
6808 (NVRAM) is random-access memory that retains its information when power
6809 is turned off (non-volatile). Systems with such memory can store data
6810 structures in RAM and retrieve them after a reboot, without flushing
6811 to typical _storage_.
6813 Linux supports NVRAM file systems thanks to either
6814 http://pramfs.sourceforge.net/[PRAMFS] or
6815 https://www.kernel.org/doc/Documentation/filesystems/dax.txt[DAX]{nbsp}+{nbsp}http://lkml.iu.edu/hypermail/linux/kernel/1504.1/03463.html[pmem]
6816 (requires Linux 4.1+).
6818 This section does not describe how to operate such file systems;
6819 we assume that you have a working persistent memory file system.
6821 When you create a <<tracing-session,tracing session>>, you can specify
6822 the path of the shared memory holding the sub-buffers. If you specify a
6823 location on an NVRAM file system, then you can retrieve the latest
6824 recorded trace data when the system reboots after a crash.
6826 To record trace data on a persistent memory file system and retrieve the
6827 trace data after a system crash:
6829 . Create a tracing session with a sub-buffer shared memory path located
6830 on an NVRAM file system:
6835 $ lttng create my-session --shm-path=/path/to/shm
6839 . Configure the tracing session as usual with the man:lttng(1)
6840 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6842 . After a system crash, use the man:lttng-crash(1) command-line tool to
6843 view the trace data recorded on the NVRAM file system:
6848 $ lttng-crash /path/to/shm
6852 The binary layout of the ring buffer files is not exactly the same as
6853 the trace files layout. This is why you need to use man:lttng-crash(1)
6854 instead of your preferred trace viewer directly.
6856 To convert the ring buffer files to LTTng trace files:
6858 * Use the opt:lttng-crash(1):--extract option of man:lttng-crash(1):
6863 $ lttng-crash --extract=/path/to/trace /path/to/shm
6869 [[notif-trigger-api]]
6870 === Get notified when a channel's buffer usage is too high or too low
6872 With LTTng's $$C/C++$$ notification and trigger API, your user
6873 application can get notified when the buffer usage of one or more
6874 <<channel,channels>> becomes too low or too high. You can use this API
6875 and enable or disable <<event,event rules>> during tracing to avoid
6876 <<channel-overwrite-mode-vs-discard-mode,discarded event records>>.
6878 .Have a user application get notified when an LTTng channel's buffer usage is too high.
6880 In this example, we create and build an application which gets notified
6881 when the buffer usage of a specific LTTng channel is higher than
6882 75{nbsp}%. We only print that it is the case in the example, but we
6883 could as well use the API of <<liblttng-ctl-lttng,`liblttng-ctl`>> to
6884 disable event rules when this happens.
6886 . Create the application's C source file:
6894 #include <lttng/domain.h>
6895 #include <lttng/action/action.h>
6896 #include <lttng/action/notify.h>
6897 #include <lttng/condition/condition.h>
6898 #include <lttng/condition/buffer-usage.h>
6899 #include <lttng/condition/evaluation.h>
6900 #include <lttng/notification/channel.h>
6901 #include <lttng/notification/notification.h>
6902 #include <lttng/trigger/trigger.h>
6903 #include <lttng/endpoint.h>
6905 int main(int argc, char *argv[])
6907 int exit_status = 0;
6908 struct lttng_notification_channel *notification_channel;
6909 struct lttng_condition *condition;
6910 struct lttng_action *action;
6911 struct lttng_trigger *trigger;
6912 const char *tracing_session_name;
6913 const char *channel_name;
6916 tracing_session_name = argv[1];
6917 channel_name = argv[2];
6920 * Create a notification channel. A notification channel
6921 * connects the user application to the LTTng session daemon.
6922 * This notification channel can be used to listen to various
6923 * types of notifications.
6925 notification_channel = lttng_notification_channel_create(
6926 lttng_session_daemon_notification_endpoint);
6929 * Create a "high buffer usage" condition. In this case, the
6930 * condition is reached when the buffer usage is greater than or
6931 * equal to 75 %. We create the condition for a specific tracing
6932 * session name, channel name, and for the user space tracing
6935 * The "low buffer usage" condition type also exists.
6937 condition = lttng_condition_buffer_usage_high_create();
6938 lttng_condition_buffer_usage_set_threshold_ratio(condition, .75);
6939 lttng_condition_buffer_usage_set_session_name(
6940 condition, tracing_session_name);
6941 lttng_condition_buffer_usage_set_channel_name(condition,
6943 lttng_condition_buffer_usage_set_domain_type(condition,
6947 * Create an action (get a notification) to take when the
6948 * condition created above is reached.
6950 action = lttng_action_notify_create();
6953 * Create a trigger. A trigger associates a condition to an
6954 * action: the action is executed when the condition is reached.
6956 trigger = lttng_trigger_create(condition, action);
6958 /* Register the trigger to LTTng. */
6959 lttng_register_trigger(trigger);
6962 * Now that we have registered a trigger, a notification will be
6963 * emitted everytime its condition is met. To receive this
6964 * notification, we must subscribe to notifications that match
6965 * the same condition.
6967 lttng_notification_channel_subscribe(notification_channel,
6971 * Notification loop. You can put this in a dedicated thread to
6972 * avoid blocking the main thread.
6975 struct lttng_notification *notification;
6976 enum lttng_notification_channel_status status;
6977 const struct lttng_evaluation *notification_evaluation;
6978 const struct lttng_condition *notification_condition;
6979 double buffer_usage;
6981 /* Receive the next notification. */
6982 status = lttng_notification_channel_get_next_notification(
6983 notification_channel, ¬ification);
6986 case LTTNG_NOTIFICATION_CHANNEL_STATUS_OK:
6988 case LTTNG_NOTIFICATION_CHANNEL_STATUS_NOTIFICATIONS_DROPPED:
6990 * The session daemon can drop notifications if
6991 * a monitoring application is not consuming the
6992 * notifications fast enough.
6995 case LTTNG_NOTIFICATION_CHANNEL_STATUS_CLOSED:
6997 * The notification channel has been closed by the
6998 * session daemon. This is typically caused by a session
6999 * daemon shutting down.
7003 /* Unhandled conditions or errors. */
7009 * A notification provides, amongst other things:
7011 * * The condition that caused this notification to be
7013 * * The condition evaluation, which provides more
7014 * specific information on the evaluation of the
7017 * The condition evaluation provides the buffer usage
7018 * value at the moment the condition was reached.
7020 notification_condition = lttng_notification_get_condition(
7022 notification_evaluation = lttng_notification_get_evaluation(
7025 /* We're subscribed to only one condition. */
7026 assert(lttng_condition_get_type(notification_condition) ==
7027 LTTNG_CONDITION_TYPE_BUFFER_USAGE_HIGH);
7030 * Get the exact sampled buffer usage from the
7031 * condition evaluation.
7033 lttng_evaluation_buffer_usage_get_usage_ratio(
7034 notification_evaluation, &buffer_usage);
7037 * At this point, instead of printing a message, we
7038 * could do something to reduce the channel's buffer
7039 * usage, like disable specific events.
7041 printf("Buffer usage is %f %% in tracing session \"%s\", "
7042 "user space channel \"%s\".\n", buffer_usage * 100,
7043 tracing_session_name, channel_name);
7044 lttng_notification_destroy(notification);
7048 lttng_action_destroy(action);
7049 lttng_condition_destroy(condition);
7050 lttng_trigger_destroy(trigger);
7051 lttng_notification_channel_destroy(notification_channel);
7057 . Build the `notif-app` application, linking it to `liblttng-ctl`:
7062 $ gcc -o notif-app notif-app.c -llttng-ctl
7066 . <<creating-destroying-tracing-sessions,Create a tracing session>>,
7067 <<enabling-disabling-events,create an event rule>> matching all the
7068 user space tracepoints, and
7069 <<basic-tracing-session-control,start tracing>>:
7074 $ lttng create my-session
7075 $ lttng enable-event --userspace --all
7080 If you create the channel manually with the man:lttng-enable-channel(1)
7081 command, you can control how frequently are the current values of the
7082 channel's properties sampled to evaluate user conditions with the
7083 opt:lttng-enable-channel(1):--monitor-timer option.
7085 . Run the `notif-app` application. This program accepts the
7086 <<tracing-session,tracing session>> name and the user space channel
7087 name as its two first arguments. The channel which LTTng automatically
7088 creates with the man:lttng-enable-event(1) command above is named
7094 $ ./notif-app my-session channel0
7098 . In another terminal, run an application with a very high event
7099 throughput so that the 75{nbsp}% buffer usage condition is reached.
7101 In the first terminal, the application should print lines like this:
7104 Buffer usage is 81.45197 % in tracing session "my-session", user space
7108 If you don't see anything, try modifying the condition in
7109 path:{notif-app.c} to a lower value (0.1, for example), rebuilding it
7110 (step 2) and running it again (step 4).
7117 [[lttng-modules-ref]]
7118 === noch:{LTTng-modules}
7122 [[lttng-tracepoint-enum]]
7123 ==== `LTTNG_TRACEPOINT_ENUM()` usage
7125 Use the `LTTNG_TRACEPOINT_ENUM()` macro to define an enumeration:
7129 LTTNG_TRACEPOINT_ENUM(name, TP_ENUM_VALUES(entries))
7134 * `name` with the name of the enumeration (C identifier, unique
7135 amongst all the defined enumerations).
7136 * `entries` with a list of enumeration entries.
7138 The available enumeration entry macros are:
7140 +ctf_enum_value(__name__, __value__)+::
7141 Entry named +__name__+ mapped to the integral value +__value__+.
7143 +ctf_enum_range(__name__, __begin__, __end__)+::
7144 Entry named +__name__+ mapped to the range of integral values between
7145 +__begin__+ (included) and +__end__+ (included).
7147 +ctf_enum_auto(__name__)+::
7148 Entry named +__name__+ mapped to the integral value following the
7149 last mapping's value.
7151 The last value of a `ctf_enum_value()` entry is its +__value__+
7154 The last value of a `ctf_enum_range()` entry is its +__end__+ parameter.
7156 If `ctf_enum_auto()` is the first entry in the list, its integral
7159 Use the `ctf_enum()` <<lttng-modules-tp-fields,field definition macro>>
7160 to use a defined enumeration as a tracepoint field.
7162 .Define an enumeration with `LTTNG_TRACEPOINT_ENUM()`.
7166 LTTNG_TRACEPOINT_ENUM(
7169 ctf_enum_auto("AUTO: EXPECT 0")
7170 ctf_enum_value("VALUE: 23", 23)
7171 ctf_enum_value("VALUE: 27", 27)
7172 ctf_enum_auto("AUTO: EXPECT 28")
7173 ctf_enum_range("RANGE: 101 TO 303", 101, 303)
7174 ctf_enum_auto("AUTO: EXPECT 304")
7182 [[lttng-modules-tp-fields]]
7183 ==== Tracepoint fields macros (for `TP_FIELDS()`)
7185 [[tp-fast-assign]][[tp-struct-entry]]The available macros to define
7186 tracepoint fields, which must be listed within `TP_FIELDS()` in
7187 `LTTNG_TRACEPOINT_EVENT()`, are:
7189 [role="func-desc growable",cols="asciidoc,asciidoc"]
7190 .Available macros to define LTTng-modules tracepoint fields
7192 |Macro |Description and parameters
7195 +ctf_integer(__t__, __n__, __e__)+
7197 +ctf_integer_nowrite(__t__, __n__, __e__)+
7199 +ctf_user_integer(__t__, __n__, __e__)+
7201 +ctf_user_integer_nowrite(__t__, __n__, __e__)+
7203 Standard integer, displayed in base 10.
7206 Integer C type (`int`, `long`, `size_t`, ...).
7212 Argument expression.
7215 +ctf_integer_hex(__t__, __n__, __e__)+
7217 +ctf_user_integer_hex(__t__, __n__, __e__)+
7219 Standard integer, displayed in base 16.
7228 Argument expression.
7230 |+ctf_integer_oct(__t__, __n__, __e__)+
7232 Standard integer, displayed in base 8.
7241 Argument expression.
7244 +ctf_integer_network(__t__, __n__, __e__)+
7246 +ctf_user_integer_network(__t__, __n__, __e__)+
7248 Integer in network byte order (big-endian), displayed in base 10.
7257 Argument expression.
7260 +ctf_integer_network_hex(__t__, __n__, __e__)+
7262 +ctf_user_integer_network_hex(__t__, __n__, __e__)+
7264 Integer in network byte order, displayed in base 16.
7273 Argument expression.
7276 +ctf_enum(__N__, __t__, __n__, __e__)+
7278 +ctf_enum_nowrite(__N__, __t__, __n__, __e__)+
7280 +ctf_user_enum(__N__, __t__, __n__, __e__)+
7282 +ctf_user_enum_nowrite(__N__, __t__, __n__, __e__)+
7287 Name of a <<lttng-tracepoint-enum,previously defined enumeration>>.
7290 Integer C type (`int`, `long`, `size_t`, ...).
7296 Argument expression.
7299 +ctf_string(__n__, __e__)+
7301 +ctf_string_nowrite(__n__, __e__)+
7303 +ctf_user_string(__n__, __e__)+
7305 +ctf_user_string_nowrite(__n__, __e__)+
7307 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
7313 Argument expression.
7316 +ctf_array(__t__, __n__, __e__, __s__)+
7318 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
7320 +ctf_user_array(__t__, __n__, __e__, __s__)+
7322 +ctf_user_array_nowrite(__t__, __n__, __e__, __s__)+
7324 Statically-sized array of integers.
7327 Array element C type.
7333 Argument expression.
7339 +ctf_array_bitfield(__t__, __n__, __e__, __s__)+
7341 +ctf_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7343 +ctf_user_array_bitfield(__t__, __n__, __e__, __s__)+
7345 +ctf_user_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7347 Statically-sized array of bits.
7349 The type of +__e__+ must be an integer type. +__s__+ is the number
7350 of elements of such type in +__e__+, not the number of bits.
7353 Array element C type.
7359 Argument expression.
7365 +ctf_array_text(__t__, __n__, __e__, __s__)+
7367 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
7369 +ctf_user_array_text(__t__, __n__, __e__, __s__)+
7371 +ctf_user_array_text_nowrite(__t__, __n__, __e__, __s__)+
7373 Statically-sized array, printed as text.
7375 The string does not need to be null-terminated.
7378 Array element C type (always `char`).
7384 Argument expression.
7390 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
7392 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7394 +ctf_user_sequence(__t__, __n__, __e__, __T__, __E__)+
7396 +ctf_user_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7398 Dynamically-sized array of integers.
7400 The type of +__E__+ must be unsigned.
7403 Array element C type.
7409 Argument expression.
7412 Length expression C type.
7418 +ctf_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7420 +ctf_user_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7422 Dynamically-sized array of integers, displayed in base 16.
7424 The type of +__E__+ must be unsigned.
7427 Array element C type.
7433 Argument expression.
7436 Length expression C type.
7441 |+ctf_sequence_network(__t__, __n__, __e__, __T__, __E__)+
7443 Dynamically-sized array of integers in network byte order (big-endian),
7444 displayed in base 10.
7446 The type of +__E__+ must be unsigned.
7449 Array element C type.
7455 Argument expression.
7458 Length expression C type.
7464 +ctf_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7466 +ctf_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7468 +ctf_user_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7470 +ctf_user_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7472 Dynamically-sized array of bits.
7474 The type of +__e__+ must be an integer type. +__s__+ is the number
7475 of elements of such type in +__e__+, not the number of bits.
7477 The type of +__E__+ must be unsigned.
7480 Array element C type.
7486 Argument expression.
7489 Length expression C type.
7495 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7497 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7499 +ctf_user_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7501 +ctf_user_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7503 Dynamically-sized array, displayed as text.
7505 The string does not need to be null-terminated.
7507 The type of +__E__+ must be unsigned.
7509 The behaviour is undefined if +__e__+ is `NULL`.
7512 Sequence element C type (always `char`).
7518 Argument expression.
7521 Length expression C type.
7527 Use the `_user` versions when the argument expression, `e`, is
7528 a user space address. In the cases of `ctf_user_integer*()` and
7529 `ctf_user_float*()`, `&e` must be a user space address, thus `e` must
7532 The `_nowrite` versions omit themselves from the session trace, but are
7533 otherwise identical. This means the `_nowrite` fields won't be written
7534 in the recorded trace. Their primary purpose is to make some
7535 of the event context available to the
7536 <<enabling-disabling-events,event filters>> without having to
7537 commit the data to sub-buffers.
7543 Terms related to LTTng and to tracing in general:
7546 The http://diamon.org/babeltrace[Babeltrace] project, which includes
7547 the cmd:babeltrace command, some libraries, and Python bindings.
7549 <<channel-buffering-schemes,buffering scheme>>::
7550 A layout of sub-buffers applied to a given channel.
7552 <<channel,channel>>::
7553 An entity which is responsible for a set of ring buffers.
7555 <<event,Event rules>> are always attached to a specific channel.
7558 A reference of time for a tracer.
7560 <<lttng-consumerd,consumer daemon>>::
7561 A process which is responsible for consuming the full sub-buffers
7562 and write them to a file system or send them over the network.
7564 <<channel-overwrite-mode-vs-discard-mode,discard mode>>:: The event loss
7565 mode in which the tracer _discards_ new event records when there's no
7566 sub-buffer space left to store them.
7569 The consequence of the execution of an instrumentation
7570 point, like a tracepoint that you manually place in some source code,
7571 or a Linux kernel KProbe.
7573 An event is said to _occur_ at a specific time. Different actions can
7574 be taken upon the occurrence of an event, like record the event's payload
7577 <<channel-overwrite-mode-vs-discard-mode,event loss mode>>::
7578 The mechanism by which event records of a given channel are lost
7579 (not recorded) when there is no sub-buffer space left to store them.
7581 [[def-event-name]]event name::
7582 The name of an event, which is also the name of the event record.
7583 This is also called the _instrumentation point name_.
7586 A record, in a trace, of the payload of an event which occured.
7588 <<event,event rule>>::
7589 Set of conditions which must be satisfied for one or more occuring
7590 events to be recorded.
7592 `java.util.logging`::
7594 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[core logging facilities].
7596 <<instrumenting,instrumentation>>::
7597 The use of LTTng probes to make a piece of software traceable.
7599 instrumentation point::
7600 A point in the execution path of a piece of software that, when
7601 reached by this execution, can emit an event.
7603 instrumentation point name::
7604 See _<<def-event-name,event name>>_.
7607 A http://logging.apache.org/log4j/1.2/[logging library] for Java
7608 developed by the Apache Software Foundation.
7611 Level of severity of a log statement or user space
7612 instrumentation point.
7615 The _Linux Trace Toolkit: next generation_ project.
7617 <<lttng-cli,cmd:lttng>>::
7618 A command-line tool provided by the LTTng-tools project which you
7619 can use to send and receive control messages to and from a
7623 The https://github.com/lttng/lttng-analyses[LTTng analyses] project,
7624 which is a set of analyzing programs that are used to obtain a
7625 higher level view of an LTTng trace.
7627 cmd:lttng-consumerd::
7628 The name of the consumer daemon program.
7631 A utility provided by the LTTng-tools project which can convert
7632 ring buffer files (usually
7633 <<persistent-memory-file-systems,saved on a persistent memory file system>>)
7636 LTTng Documentation::
7639 <<lttng-live,LTTng live>>::
7640 A communication protocol between the relay daemon and live viewers
7641 which makes it possible to see events "live", as they are received by
7644 <<lttng-modules,LTTng-modules>>::
7645 The https://github.com/lttng/lttng-modules[LTTng-modules] project,
7646 which contains the Linux kernel modules to make the Linux kernel
7647 instrumentation points available for LTTng tracing.
7650 The name of the relay daemon program.
7652 cmd:lttng-sessiond::
7653 The name of the session daemon program.
7656 The https://github.com/lttng/lttng-tools[LTTng-tools] project, which
7657 contains the various programs and libraries used to
7658 <<controlling-tracing,control tracing>>.
7660 <<lttng-ust,LTTng-UST>>::
7661 The https://github.com/lttng/lttng-ust[LTTng-UST] project, which
7662 contains libraries to instrument user applications.
7664 <<lttng-ust-agents,LTTng-UST Java agent>>::
7665 A Java package provided by the LTTng-UST project to allow the
7666 LTTng instrumentation of `java.util.logging` and Apache log4j 1.2
7669 <<lttng-ust-agents,LTTng-UST Python agent>>::
7670 A Python package provided by the LTTng-UST project to allow the
7671 LTTng instrumentation of Python logging statements.
7673 <<channel-overwrite-mode-vs-discard-mode,overwrite mode>>::
7674 The event loss mode in which new event records overwrite older
7675 event records when there's no sub-buffer space left to store them.
7677 <<channel-buffering-schemes,per-process buffering>>::
7678 A buffering scheme in which each instrumented process has its own
7679 sub-buffers for a given user space channel.
7681 <<channel-buffering-schemes,per-user buffering>>::
7682 A buffering scheme in which all the processes of a Unix user share the
7683 same sub-buffer for a given user space channel.
7685 <<lttng-relayd,relay daemon>>::
7686 A process which is responsible for receiving the trace data sent by
7687 a distant consumer daemon.
7690 A set of sub-buffers.
7692 <<lttng-sessiond,session daemon>>::
7693 A process which receives control commands from you and orchestrates
7694 the tracers and various LTTng daemons.
7696 <<taking-a-snapshot,snapshot>>::
7697 A copy of the current data of all the sub-buffers of a given tracing
7698 session, saved as trace files.
7701 One part of an LTTng ring buffer which contains event records.
7704 The time information attached to an event when it is emitted.
7707 A set of files which are the concatenations of one or more
7708 flushed sub-buffers.
7711 The action of recording the events emitted by an application
7712 or by a system, or to initiate such recording by controlling
7716 The http://tracecompass.org[Trace Compass] project and application.
7719 An instrumentation point using the tracepoint mechanism of the Linux
7720 kernel or of LTTng-UST.
7722 tracepoint definition::
7723 The definition of a single tracepoint.
7726 The name of a tracepoint.
7728 tracepoint provider::
7729 A set of functions providing tracepoints to an instrumented user
7732 Not to be confused with a _tracepoint provider package_: many tracepoint
7733 providers can exist within a tracepoint provider package.
7735 tracepoint provider package::
7736 One or more tracepoint providers compiled as an object file or as
7740 A software which records emitted events.
7742 <<domain,tracing domain>>::
7743 A namespace for event sources.
7745 <<tracing-group,tracing group>>::
7746 The Unix group in which a Unix user can be to be allowed to trace the
7749 <<tracing-session,tracing session>>::
7750 A stateful dialogue between you and a <<lttng-sessiond,session
7754 An application running in user space, as opposed to a Linux kernel
7755 module, for example.