1 The LTTng Documentation
2 =======================
3 Philippe Proulx <pproulx@efficios.com>
7 include::../common/copyright.txt[]
10 include::../common/welcome.txt[]
13 include::../common/audience.txt[]
17 === What's in this documentation?
19 The LTTng Documentation is divided into the following sections:
21 * **<<nuts-and-bolts,Nuts and bolts>>** explains the
22 rudiments of software tracing and the rationale behind the
25 You can skip this section if you’re familiar with software tracing and
26 with the LTTng project.
28 * **<<installing-lttng,Installation>>** describes the steps to
29 install the LTTng packages on common Linux distributions and from
32 You can skip this section if you already properly installed LTTng on
35 * **<<getting-started,Quick start>>** is a concise guide to
36 getting started quickly with LTTng kernel and user space tracing.
38 We recommend this section if you're new to LTTng or to software tracing
41 You can skip this section if you're not new to LTTng.
43 * **<<core-concepts,Core concepts>>** explains the concepts at
46 It's a good idea to become familiar with the core concepts
47 before attempting to use the toolkit.
49 * **<<plumbing,Components of LTTng>>** describes the various components
50 of the LTTng machinery, like the daemons, the libraries, and the
51 command-line interface.
52 * **<<instrumenting,Instrumentation>>** shows different ways to
53 instrument user applications and the Linux kernel.
55 Instrumenting source code is essential to provide a meaningful
58 You can skip this section if you do not have a programming background.
60 * **<<controlling-tracing,Tracing control>>** is divided into topics
61 which demonstrate how to use the vast array of features that
62 LTTng{nbsp}{revision} offers.
63 * **<<reference,Reference>>** contains reference tables.
64 * **<<glossary,Glossary>>** is a specialized dictionary of terms related
65 to LTTng or to the field of software tracing.
68 include::../common/convention.txt[]
71 include::../common/acknowledgements.txt[]
75 == What's new in LTTng {revision}?
77 * **Tracing control**:
78 ** You can override the name or the URL of a tracing session
79 configuration when you use man:lttng-load(1) thanks to the new
80 opt:lttng-load(1):--override-name and
81 opt:lttng-load(1):--override-url options.
82 ** The new `lttng regenerate` command replaces the now deprecated
83 `lttng metadata` command of LTTng 2.8. man:lttng-regenerate(1) can
84 also <<regenerate-statedump,generate the state dump event records>>
85 of a given tracing session on demand, a handy feature when
86 <<taking-a-snapshot,taking a snapshot>>.
87 ** You can add PMU counters by raw ID with man:lttng-add-context(1):
92 lttng add-context --kernel --type=perf:cpu:raw:r0013c:x86unhalted
96 The format of the raw ID is the same as used with man:perf-record(1).
97 See <<adding-context,Add context fields to a channel>> for more
100 ** The LTTng <<lttng-relayd,relay daemon>> is now supported on
101 OS{nbsp}X and macOS for a smoother integration within a trace
102 analysis workflow, regardless of the platform used.
104 * **User space tracing**:
105 ** Improved performance (tested on x86-64 and ARMv7-A
106 (https://en.wikipedia.org/wiki/Cubieboard[Cubieboard])
108 ** New helper library (`liblttng-ust-fd`) to help with
109 <<liblttng-ust-fd,applications which close file descriptors that
110 don't belong to them>>, for example, in a loop which closes file
111 descriptors after man:fork(2), or BSD's `closeall()`.
112 ** More accurate <<liblttng-ust-dl,dynamic linker instrumentation>> and
113 state dump event records, especially when a dynamically loaded
114 library manually loads its own dependencies.
115 ** New `ctf_*()` field definition macros (see man:lttng-ust(3)):
116 *** `ctf_array_hex()`
117 *** `ctf_array_network()`
118 *** `ctf_array_network_hex()`
119 *** `ctf_sequence_hex()`
120 *** `ctf_sequence_network()`
121 *** `ctf_sequence_network_hex()`
122 ** New `lttng_ust_loaded` weak symbol defined by `liblttng-ust` for
123 an application to know if the LTTng-UST shared library is loaded
131 int lttng_ust_loaded __attribute__((weak));
135 if (lttng_ust_loaded) {
136 puts("LTTng-UST is loaded!");
138 puts("LTTng-UST is not loaded!");
146 ** LTTng-UST thread names have the `-ust` suffix.
148 * **Linux kernel tracing**:
149 ** Improved performance (tested on x86-64 and ARMv7-A
150 (https://en.wikipedia.org/wiki/Cubieboard[Cubieboard])
152 ** New enumeration <<lttng-modules-tp-fields,field definition macros>>:
153 `ctf_enum()` and `ctf_user_enum()`.
154 ** IPv4, IPv6, and TCP header data is recorded in the event records
155 produced by tracepoints starting with `net_`.
156 ** Detailed system call event records: `select`, `pselect6`, `poll`,
157 `ppoll`, `epoll_wait`, `epoll_pwait`, and `epoll_ctl` on all
158 architectures supported by LTTng-modules, and `accept4` on x86-64.
159 ** New I²C instrumentation: the `extract_sensitive_payload` parameter
160 of the new `lttng-probe-i2c` LTTng module controls whether or not
161 the payloads of I²C messages are recorded in I²C event records, since
162 they may contain sensitive data (for example, keystrokes).
163 ** When the LTTng kernel modules are built into the Linux kernel image,
164 the `CONFIG_TRACEPOINTS` configuration option is automatically
171 What is LTTng? As its name suggests, the _Linux Trace Toolkit: next
172 generation_ is a modern toolkit for tracing Linux systems and
173 applications. So your first question might be:
180 As the history of software engineering progressed and led to what
181 we now take for granted--complex, numerous and
182 interdependent software applications running in parallel on
183 sophisticated operating systems like Linux--the authors of such
184 components, software developers, began feeling a natural
185 urge to have tools that would ensure the robustness and good performance
186 of their masterpieces.
188 One major achievement in this field is, inarguably, the
189 https://www.gnu.org/software/gdb/[GNU debugger (GDB)],
190 an essential tool for developers to find and fix bugs. But even the best
191 debugger won't help make your software run faster, and nowadays, faster
192 software means either more work done by the same hardware, or cheaper
193 hardware for the same work.
195 A _profiler_ is often the tool of choice to identify performance
196 bottlenecks. Profiling is suitable to identify _where_ performance is
197 lost in a given software. The profiler outputs a profile, a statistical
198 summary of observed events, which you may use to discover which
199 functions took the most time to execute. However, a profiler won't
200 report _why_ some identified functions are the bottleneck. Bottlenecks
201 might only occur when specific conditions are met, conditions that are
202 sometimes impossible to capture by a statistical profiler, or impossible
203 to reproduce with an application altered by the overhead of an
204 event-based profiler. For a thorough investigation of software
205 performance issues, a history of execution is essential, with the
206 recorded values of variables and context fields you choose, and
207 with as little influence as possible on the instrumented software. This
208 is where tracing comes in handy.
210 _Tracing_ is a technique used to understand what goes on in a running
211 software system. The software used for tracing is called a _tracer_,
212 which is conceptually similar to a tape recorder. When recording,
213 specific instrumentation points placed in the software source code
214 generate events that are saved on a giant tape: a _trace_ file. You
215 can trace user applications and the operating system at the same time,
216 opening the possibility of resolving a wide range of problems that would
217 otherwise be extremely challenging.
219 Tracing is often compared to _logging_. However, tracers and loggers are
220 two different tools, serving two different purposes. Tracers are
221 designed to record much lower-level events that occur much more
222 frequently than log messages, often in the range of thousands per
223 second, with very little execution overhead. Logging is more appropriate
224 for a very high-level analysis of less frequent events: user accesses,
225 exceptional conditions (errors and warnings, for example), database
226 transactions, instant messaging communications, and such. Simply put,
227 logging is one of the many use cases that can be satisfied with tracing.
229 The list of recorded events inside a trace file can be read manually
230 like a log file for the maximum level of detail, but it is generally
231 much more interesting to perform application-specific analyses to
232 produce reduced statistics and graphs that are useful to resolve a
233 given problem. Trace viewers and analyzers are specialized tools
236 In the end, this is what LTTng is: a powerful, open source set of
237 tools to trace the Linux kernel and user applications at the same time.
238 LTTng is composed of several components actively maintained and
239 developed by its link:/community/#where[community].
242 [[lttng-alternatives]]
243 === Alternatives to noch:{LTTng}
245 Excluding proprietary solutions, a few competing software tracers
248 * https://github.com/dtrace4linux/linux[dtrace4linux] is a port of
249 Sun Microsystems's DTrace to Linux. The cmd:dtrace tool interprets
250 user scripts and is responsible for loading code into the
251 Linux kernel for further execution and collecting the outputted data.
252 * https://en.wikipedia.org/wiki/Berkeley_Packet_Filter[eBPF] is a
253 subsystem in the Linux kernel in which a virtual machine can execute
254 programs passed from the user space to the kernel. You can attach
255 such programs to tracepoints and KProbes thanks to a system call, and
256 they can output data to the user space when executed thanks to
257 different mechanisms (pipe, VM register values, and eBPF maps, to name
259 * https://www.kernel.org/doc/Documentation/trace/ftrace.txt[ftrace]
260 is the de facto function tracer of the Linux kernel. Its user
261 interface is a set of special files in sysfs.
262 * https://perf.wiki.kernel.org/[perf] is
263 a performance analyzing tool for Linux which supports hardware
264 performance counters, tracepoints, as well as other counters and
265 types of probes. perf's controlling utility is the cmd:perf command
267 * http://linux.die.net/man/1/strace[strace]
268 is a command-line utility which records system calls made by a
269 user process, as well as signal deliveries and changes of process
270 state. strace makes use of https://en.wikipedia.org/wiki/Ptrace[ptrace]
271 to fulfill its function.
272 * http://www.sysdig.org/[sysdig], like SystemTap, uses scripts to
273 analyze Linux kernel events. You write scripts, or _chisels_ in
274 sysdig's jargon, in Lua and sysdig executes them while the system is
275 being traced or afterwards. sysdig's interface is the cmd:sysdig
276 command-line tool as well as the curses-based cmd:csysdig tool.
277 * https://sourceware.org/systemtap/[SystemTap] is a Linux kernel and
278 user space tracer which uses custom user scripts to produce plain text
279 traces. SystemTap converts the scripts to the C language, and then
280 compiles them as Linux kernel modules which are loaded to produce
281 trace data. SystemTap's primary user interface is the cmd:stap
284 The main distinctive features of LTTng is that it produces correlated
285 kernel and user space traces, as well as doing so with the lowest
286 overhead amongst other solutions. It produces trace files in the
287 http://diamon.org/ctf[CTF] format, a file format optimized
288 for the production and analyses of multi-gigabyte data.
290 LTTng is the result of more than 10 years of active open source
291 development by a community of passionate developers.
292 LTTng{nbsp}{revision} is currently available on major desktop and server
295 The main interface for tracing control is a single command-line tool
296 named cmd:lttng. The latter can create several tracing sessions, enable
297 and disable events on the fly, filter events efficiently with custom
298 user expressions, start and stop tracing, and much more. LTTng can
299 record the traces on the file system or send them over the network, and
300 keep them totally or partially. You can view the traces once tracing
301 becomes inactive or in real-time.
303 <<installing-lttng,Install LTTng now>> and
304 <<getting-started,start tracing>>!
310 **LTTng** is a set of software <<plumbing,components>> which interact to
311 <<instrumenting,instrument>> the Linux kernel and user applications, and
312 to <<controlling-tracing,control tracing>> (start and stop
313 tracing, enable and disable event rules, and the rest). Those
314 components are bundled into the following packages:
316 * **LTTng-tools**: Libraries and command-line interface to
318 * **LTTng-modules**: Linux kernel modules to instrument and
320 * **LTTng-UST**: Libraries and Java/Python packages to instrument and
321 trace user applications.
323 Most distributions mark the LTTng-modules and LTTng-UST packages as
324 optional when installing LTTng-tools (which is always required). In the
325 following sections, we always provide the steps to install all three,
328 * You only need to install LTTng-modules if you intend to trace the
330 * You only need to install LTTng-UST if you intend to trace user
334 .Availability of LTTng{nbsp}{revision} for major Linux distributions as of 2 December 2016.
336 |Distribution |Available in releases |Alternatives
338 |https://www.ubuntu.com/[Ubuntu]
339 |Ubuntu{nbsp}14.04 _Trusty Tahr_ and Ubuntu{nbsp}16.04 _Xenial Xerus_:
340 <<ubuntu-ppa,use the LTTng Stable{nbsp}{revision} PPA>>.
341 |link:/docs/v2.8#doc-ubuntu[LTTng{nbsp}2.8 for Ubuntu{nbsp}16.10 _Yakkety Yak_].
343 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
344 other Ubuntu releases.
346 |https://getfedora.org/[Fedora]
348 |<<building-from-source,Build LTTng-modules{nbsp}{revision} from
351 |https://www.debian.org/[Debian]
353 |<<building-from-source,Build LTTng-modules{nbsp}{revision} from
356 |https://www.opensuse.org/[openSUSE]
358 |<<building-from-source,Build LTTng-modules{nbsp}{revision} from
361 |https://www.archlinux.org/[Arch Linux]
362 |<<arch-linux,Latest AUR packages>>.
365 |https://alpinelinux.org/[Alpine Linux]
367 |link:/docs/v2.8#doc-alpine-linux[LTTng{nbsp}2.8 for Alpine Linux{nbsp}"edge"].
369 LTTng{nbsp}2.8 for Alpine Linux{nbsp}3.5 (not released yet).
371 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
372 other Alpine Linux releases.
374 |https://www.redhat.com/[RHEL] and https://www.suse.com/[SLES]
375 |See http://packages.efficios.com/[EfficiOS Enterprise Packages].
378 |https://buildroot.org/[Buildroot]
380 |link:/docs/v2.8#doc-buildroot[LTTng{nbsp}2.8 for Buildroot{nbsp}2016.11].
382 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
383 other Buildroot releases.
385 |http://www.openembedded.org/wiki/Main_Page[OpenEmbedded] and
386 https://www.yoctoproject.org/[Yocto]
388 |link:/docs/v2.8#doc-oe-yocto[LTTng{nbsp}2.8 for Yocto Project{nbsp}2.2 _Morty_]
389 (`openembedded-core` layer).
391 <<building-from-source,Build LTTng{nbsp}{revision} from source>> for
392 other OpenEmbedded releases.
397 === [[ubuntu-official-repositories]]Ubuntu
400 ==== noch:{LTTng} Stable {revision} PPA
402 The https://launchpad.net/~lttng/+archive/ubuntu/stable-{revision}[LTTng
403 Stable{nbsp}{revision} PPA] offers the latest stable
404 LTTng{nbsp}{revision} packages for:
406 * Ubuntu{nbsp}14.04 _Trusty Tahr_
407 * Ubuntu{nbsp}16.04 _Xenial Xerus_
409 To install LTTng{nbsp}{revision} from the LTTng Stable{nbsp}{revision} PPA:
411 . Add the LTTng Stable{nbsp}{revision} PPA repository and update the
417 sudo apt-add-repository ppa:lttng/stable-2.9
422 . Install the main LTTng{nbsp}{revision} packages:
427 sudo apt-get install lttng-tools
428 sudo apt-get install lttng-modules-dkms
429 sudo apt-get install liblttng-ust-dev
433 . **If you need to instrument and trace
434 <<java-application,Java applications>>**, install the LTTng-UST
440 sudo apt-get install liblttng-ust-agent-java
444 . **If you need to instrument and trace
445 <<python-application,Python{nbsp}3 applications>>**, install the
446 LTTng-UST Python agent:
451 sudo apt-get install python3-lttngust
459 To install LTTng{nbsp}{revision} on Arch Linux using
460 https://archlinux.fr/yaourt-en[Yaourt]:
462 . Install the main LTTng{nbsp}{revision} packages:
467 yaourt -S lttng-tools
469 yaourt -S lttng-modules
473 . **If you need to instrument and trace <<python-application,Python
474 applications>>**, install the LTTng-UST Python agent:
479 yaourt -S python-lttngust
480 yaourt -S python2-lttngust
485 [[enterprise-distributions]]
486 === RHEL, SUSE, and other enterprise distributions
488 To install LTTng on enterprise Linux distributions, such as Red Hat
489 Enterprise Linux (RHEL) and SUSE Linux Enterprise Server (SUSE), please
490 see http://packages.efficios.com/[EfficiOS Enterprise Packages].
493 [[building-from-source]]
494 === Build from source
496 To build and install LTTng{nbsp}{revision} from source:
498 . Using your distribution's package manager, or from source, install
499 the following dependencies of LTTng-tools and LTTng-UST:
502 * https://sourceforge.net/projects/libuuid/[libuuid]
503 * http://directory.fsf.org/wiki/Popt[popt]
504 * http://liburcu.org/[Userspace RCU]
505 * http://www.xmlsoft.org/[libxml2]
508 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
514 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.9.tar.bz2 &&
515 tar -xf lttng-modules-latest-2.9.tar.bz2 &&
516 cd lttng-modules-2.9.* &&
518 sudo make modules_install &&
523 . Download, build, and install the latest LTTng-UST{nbsp}{revision}:
529 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.9.tar.bz2 &&
530 tar -xf lttng-ust-latest-2.9.tar.bz2 &&
531 cd lttng-ust-2.9.* &&
541 .Java and Python application tracing
543 If you need to instrument and trace <<java-application,Java
544 applications>>, pass the `--enable-java-agent-jul`,
545 `--enable-java-agent-log4j`, or `--enable-java-agent-all` options to the
546 `configure` script, depending on which Java logging framework you use.
548 If you need to instrument and trace <<python-application,Python
549 applications>>, pass the `--enable-python-agent` option to the
550 `configure` script. You can set the `PYTHON` environment variable to the
551 path to the Python interpreter for which to install the LTTng-UST Python
559 By default, LTTng-UST libraries are installed to
560 dir:{/usr/local/lib}, which is the de facto directory in which to
561 keep self-compiled and third-party libraries.
563 When <<building-tracepoint-providers-and-user-application,linking an
564 instrumented user application with `liblttng-ust`>>:
566 * Append `/usr/local/lib` to the env:LD_LIBRARY_PATH environment
568 * Pass the `-L/usr/local/lib` and `-Wl,-rpath,/usr/local/lib` options to
569 man:gcc(1), man:g++(1), or man:clang(1).
573 . Download, build, and install the latest LTTng-tools{nbsp}{revision}:
579 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.9.tar.bz2 &&
580 tar -xf lttng-tools-latest-2.9.tar.bz2 &&
581 cd lttng-tools-2.9.* &&
589 TIP: The https://github.com/eepp/vlttng[vlttng tool] can do all the
590 previous steps automatically for a given version of LTTng and confine
591 the installed files in a specific directory. This can be useful to test
592 LTTng without installing it on your system.
598 This is a short guide to get started quickly with LTTng kernel and user
601 Before you follow this guide, make sure to <<installing-lttng,install>>
604 This tutorial walks you through the steps to:
606 . <<tracing-the-linux-kernel,Trace the Linux kernel>>.
607 . <<tracing-your-own-user-application,Trace a user application>> written
609 . <<viewing-and-analyzing-your-traces,View and analyze the
613 [[tracing-the-linux-kernel]]
614 === Trace the Linux kernel
616 The following command lines start with cmd:sudo because you need root
617 privileges to trace the Linux kernel. You can avoid using cmd:sudo if
618 your Unix user is a member of the <<lttng-sessiond,tracing group>>.
620 . Create a <<tracing-session,tracing session>>:
625 sudo lttng create my-kernel-session
629 . List the available kernel tracepoints and system calls:
638 . Create an <<event,event rule>> which matches the desired event names,
639 for example `sched_switch` and `sched_process_fork`:
644 sudo lttng enable-event --kernel sched_switch,sched_process_fork
648 You can also create an event rule which _matches_ all the Linux kernel
649 tracepoints (this will generate a lot of data when tracing):
654 sudo lttng enable-event --kernel --all
667 . Do some operation on your system for a few seconds. For example,
668 load a website, or list the files of a directory.
669 . Stop tracing and destroy the tracing session:
679 The man:lttng-destroy(1) command does not destroy the trace data; it
680 only destroys the state of the tracing session.
682 By default, LTTng saves the traces in
683 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
684 where +__name__+ is the tracing session name. Note that the
685 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
687 See <<viewing-and-analyzing-your-traces,View and analyze the
688 recorded events>> to view the recorded events.
691 [[tracing-your-own-user-application]]
692 === Trace a user application
694 This section steps you through a simple example to trace a
695 _Hello world_ program written in C.
697 To create the traceable user application:
699 . Create the tracepoint provider header file, which defines the
700 tracepoints and the events they can generate:
706 #undef TRACEPOINT_PROVIDER
707 #define TRACEPOINT_PROVIDER hello_world
709 #undef TRACEPOINT_INCLUDE
710 #define TRACEPOINT_INCLUDE "./hello-tp.h"
712 #if !defined(_HELLO_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
715 #include <lttng/tracepoint.h>
725 ctf_string(my_string_field, my_string_arg)
726 ctf_integer(int, my_integer_field, my_integer_arg)
730 #endif /* _HELLO_TP_H */
732 #include <lttng/tracepoint-event.h>
736 . Create the tracepoint provider package source file:
742 #define TRACEPOINT_CREATE_PROBES
743 #define TRACEPOINT_DEFINE
745 #include "hello-tp.h"
749 . Build the tracepoint provider package:
754 gcc -c -I. hello-tp.c
758 . Create the _Hello World_ application source file:
765 #include "hello-tp.h"
767 int main(int argc, char *argv[])
771 puts("Hello, World!\nPress Enter to continue...");
774 * The following getchar() call is only placed here for the purpose
775 * of this demonstration, to pause the application in order for
776 * you to have time to list its tracepoints. It is not
782 * A tracepoint() call.
784 * Arguments, as defined in hello-tp.h:
786 * 1. Tracepoint provider name (required)
787 * 2. Tracepoint name (required)
788 * 3. my_integer_arg (first user-defined argument)
789 * 4. my_string_arg (second user-defined argument)
791 * Notice the tracepoint provider and tracepoint names are
792 * NOT strings: they are in fact parts of variables that the
793 * macros in hello-tp.h create.
795 tracepoint(hello_world, my_first_tracepoint, 23, "hi there!");
797 for (x = 0; x < argc; ++x) {
798 tracepoint(hello_world, my_first_tracepoint, x, argv[x]);
801 puts("Quitting now!");
802 tracepoint(hello_world, my_first_tracepoint, x * x, "x^2");
809 . Build the application:
818 . Link the application with the tracepoint provider package,
819 `liblttng-ust`, and `libdl`:
824 gcc -o hello hello.o hello-tp.o -llttng-ust -ldl
828 Here's the whole build process:
831 .User space tracing tutorial's build steps.
832 image::ust-flow.png[]
834 To trace the user application:
836 . Run the application with a few arguments:
841 ./hello world and beyond
850 Press Enter to continue...
854 . Start an LTTng <<lttng-sessiond,session daemon>>:
859 lttng-sessiond --daemonize
863 Note that a session daemon might already be running, for example as
864 a service that the distribution's service manager started.
866 . List the available user space tracepoints:
871 lttng list --userspace
875 You see the `hello_world:my_first_tracepoint` tracepoint listed
876 under the `./hello` process.
878 . Create a <<tracing-session,tracing session>>:
883 lttng create my-user-space-session
887 . Create an <<event,event rule>> which matches the
888 `hello_world:my_first_tracepoint` event name:
893 lttng enable-event --userspace hello_world:my_first_tracepoint
906 . Go back to the running `hello` application and press Enter. The
907 program executes all `tracepoint()` instrumentation points and exits.
908 . Stop tracing and destroy the tracing session:
918 The man:lttng-destroy(1) command does not destroy the trace data; it
919 only destroys the state of the tracing session.
921 By default, LTTng saves the traces in
922 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
923 where +__name__+ is the tracing session name. Note that the
924 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
926 See <<viewing-and-analyzing-your-traces,View and analyze the
927 recorded events>> to view the recorded events.
930 [[viewing-and-analyzing-your-traces]]
931 === View and analyze the recorded events
933 Once you have completed the <<tracing-the-linux-kernel,Trace the Linux
934 kernel>> and <<tracing-your-own-user-application,Trace a user
935 application>> tutorials, you can inspect the recorded events.
937 Many tools are available to read LTTng traces:
939 * **cmd:babeltrace** is a command-line utility which converts trace
940 formats; it supports the format that LTTng produces, CTF, as well as a
941 basic text output which can be ++grep++ed. The cmd:babeltrace command
942 is part of the http://diamon.org/babeltrace[Babeltrace] project.
943 * Babeltrace also includes
944 **https://www.python.org/[Python] bindings** so
945 that you can easily open and read an LTTng trace with your own script,
946 benefiting from the power of Python.
947 * http://tracecompass.org/[**Trace Compass**]
948 is a graphical user interface for viewing and analyzing any type of
949 logs or traces, including LTTng's.
950 * https://github.com/lttng/lttng-analyses[**LTTng analyses**] is a
951 project which includes many high-level analyses of LTTng kernel
952 traces, like scheduling statistics, interrupt frequency distribution,
953 top CPU usage, and more.
955 NOTE: This section assumes that the traces recorded during the previous
956 tutorials were saved to their default location, in the
957 dir:{$LTTNG_HOME/lttng-traces} directory. Note that the env:LTTNG_HOME
958 environment variable defaults to `$HOME` if not set.
961 [[viewing-and-analyzing-your-traces-bt]]
962 ==== Use the cmd:babeltrace command-line tool
964 The simplest way to list all the recorded events of a trace is to pass
965 its path to cmd:babeltrace with no options:
969 babeltrace ~/lttng-traces/my-user-space-session*
972 cmd:babeltrace finds all traces recursively within the given path and
973 prints all their events, merging them in chronological order.
975 You can pipe the output of cmd:babeltrace into a tool like man:grep(1) for
980 babeltrace ~/lttng-traces/my-kernel-session* | grep sys_
983 You can pipe the output of cmd:babeltrace into a tool like man:wc(1) to
984 count the recorded events:
988 babeltrace ~/lttng-traces/my-kernel-session* | grep sys_read | wc --lines
992 [[viewing-and-analyzing-your-traces-bt-python]]
993 ==== Use the Babeltrace Python bindings
995 The <<viewing-and-analyzing-your-traces-bt,text output of cmd:babeltrace>>
996 is useful to isolate events by simple matching using man:grep(1) and
997 similar utilities. However, more elaborate filters, such as keeping only
998 event records with a field value falling within a specific range, are
999 not trivial to write using a shell. Moreover, reductions and even the
1000 most basic computations involving multiple event records are virtually
1001 impossible to implement.
1003 Fortunately, Babeltrace ships with Python 3 bindings which makes it easy
1004 to read the event records of an LTTng trace sequentially and compute the
1005 desired information.
1007 The following script accepts an LTTng Linux kernel trace path as its
1008 first argument and prints the short names of the top 5 running processes
1009 on CPU 0 during the whole trace:
1014 from collections import Counter
1020 if len(sys.argv) != 2:
1021 msg = 'Usage: python3 {} TRACEPATH'.format(sys.argv[0])
1022 print(msg, file=sys.stderr)
1025 # A trace collection contains one or more traces
1026 col = babeltrace.TraceCollection()
1028 # Add the trace provided by the user (LTTng traces always have
1030 if col.add_trace(sys.argv[1], 'ctf') is None:
1031 raise RuntimeError('Cannot add trace')
1033 # This counter dict contains execution times:
1035 # task command name -> total execution time (ns)
1036 exec_times = Counter()
1038 # This contains the last `sched_switch` timestamp
1042 for event in col.events:
1043 # Keep only `sched_switch` events
1044 if event.name != 'sched_switch':
1047 # Keep only events which happened on CPU 0
1048 if event['cpu_id'] != 0:
1052 cur_ts = event.timestamp
1058 # Previous task command (short) name
1059 prev_comm = event['prev_comm']
1061 # Initialize entry in our dict if not yet done
1062 if prev_comm not in exec_times:
1063 exec_times[prev_comm] = 0
1065 # Compute previous command execution time
1066 diff = cur_ts - last_ts
1068 # Update execution time of this command
1069 exec_times[prev_comm] += diff
1071 # Update last timestamp
1075 for name, ns in exec_times.most_common(5):
1077 print('{:20}{} s'.format(name, s))
1082 if __name__ == '__main__':
1083 sys.exit(0 if top5proc() else 1)
1090 python3 top5proc.py ~/lttng-traces/my-kernel-session*/kernel
1096 swapper/0 48.607245889 s
1097 chromium 7.192738188 s
1098 pavucontrol 0.709894415 s
1099 Compositor 0.660867933 s
1100 Xorg.bin 0.616753786 s
1103 Note that `swapper/0` is the "idle" process of CPU 0 on Linux; since we
1104 weren't using the CPU that much when tracing, its first position in the
1109 == [[understanding-lttng]]Core concepts
1111 From a user's perspective, the LTTng system is built on a few concepts,
1112 or objects, on which the <<lttng-cli,cmd:lttng command-line tool>>
1113 operates by sending commands to the <<lttng-sessiond,session daemon>>.
1114 Understanding how those objects relate to eachother is key in mastering
1117 The core concepts are:
1119 * <<tracing-session,Tracing session>>
1120 * <<domain,Tracing domain>>
1121 * <<channel,Channel and ring buffer>>
1122 * <<"event","Instrumentation point, event rule, event, and event record">>
1128 A _tracing session_ is a stateful dialogue between you and
1129 a <<lttng-sessiond,session daemon>>. You can
1130 <<creating-destroying-tracing-sessions,create a new tracing
1131 session>> with the `lttng create` command.
1133 Anything that you do when you control LTTng tracers happens within a
1134 tracing session. In particular, a tracing session:
1137 * Has its own set of trace files.
1138 * Has its own state of activity (started or stopped).
1139 * Has its own <<tracing-session-mode,mode>> (local, network streaming,
1141 * Has its own <<channel,channels>> which have their own
1142 <<event,event rules>>.
1145 .A _tracing session_ contains <<channel,channels>> that are members of <<domain,tracing domains>> and contain <<event,event rules>>.
1146 image::concepts.png[]
1148 Those attributes and objects are completely isolated between different
1151 A tracing session is analogous to a cash machine session:
1152 the operations you do on the banking system through the cash machine do
1153 not alter the data of other users of the same system. In the case of
1154 the cash machine, a session lasts as long as your bank card is inside.
1155 In the case of LTTng, a tracing session lasts from the `lttng create`
1156 command to the `lttng destroy` command.
1159 .Each Unix user has its own set of tracing sessions.
1160 image::many-sessions.png[]
1163 [[tracing-session-mode]]
1164 ==== Tracing session mode
1166 LTTng can send the generated trace data to different locations. The
1167 _tracing session mode_ dictates where to send it. The following modes
1168 are available in LTTng{nbsp}{revision}:
1171 LTTng writes the traces to the file system of the machine being traced
1174 Network streaming mode::
1175 LTTng sends the traces over the network to a
1176 <<lttng-relayd,relay daemon>> running on a remote system.
1179 LTTng does not write the traces by default. Instead, you can request
1180 LTTng to <<taking-a-snapshot,take a snapshot>>, that is, a copy of the
1181 current tracing buffers, and to write it to the target's file system
1182 or to send it over the network to a <<lttng-relayd,relay daemon>>
1183 running on a remote system.
1186 This mode is similar to the network streaming mode, but a live
1187 trace viewer can connect to the distant relay daemon to
1188 <<lttng-live,view event records as LTTng generates them>> by
1195 A _tracing domain_ is a namespace for event sources. A tracing domain
1196 has its own properties and features.
1198 There are currently five available tracing domains:
1202 * `java.util.logging` (JUL)
1206 You must specify a tracing domain when using some commands to avoid
1207 ambiguity. For example, since all the domains support named tracepoints
1208 as event sources (instrumentation points that you manually insert in the
1209 source code), you need to specify a tracing domain when
1210 <<enabling-disabling-events,creating an event rule>> because all the
1211 tracing domains could have tracepoints with the same names.
1213 Some features are reserved to specific tracing domains. Dynamic function
1214 entry and return instrumentation points, for example, are currently only
1215 supported in the Linux kernel tracing domain, but support for other
1216 tracing domains could be added in the future.
1218 You can create <<channel,channels>> in the Linux kernel and user space
1219 tracing domains. The other tracing domains have a single default
1224 === Channel and ring buffer
1226 A _channel_ is an object which is responsible for a set of ring buffers.
1227 Each ring buffer is divided into multiple sub-buffers. When an LTTng
1228 tracer emits an event, it can record it to one or more
1229 sub-buffers. The attributes of a channel determine what to do when
1230 there's no space left for a new event record because all sub-buffers
1231 are full, where to send a full sub-buffer, and other behaviours.
1233 A channel is always associated to a <<domain,tracing domain>>. The
1234 `java.util.logging` (JUL), log4j, and Python tracing domains each have
1235 a default channel which you cannot configure.
1237 A channel also owns <<event,event rules>>. When an LTTng tracer emits
1238 an event, it records it to the sub-buffers of all
1239 the enabled channels with a satisfied event rule, as long as those
1240 channels are part of active <<tracing-session,tracing sessions>>.
1243 [[channel-buffering-schemes]]
1244 ==== Per-user vs. per-process buffering schemes
1246 A channel has at least one ring buffer _per CPU_. LTTng always
1247 records an event to the ring buffer associated to the CPU on which it
1250 Two _buffering schemes_ are available when you
1251 <<enabling-disabling-channels,create a channel>> in the
1252 user space <<domain,tracing domain>>:
1254 Per-user buffering::
1255 Allocate one set of ring buffers--one per CPU--shared by all the
1256 instrumented processes of each Unix user.
1260 .Per-user buffering scheme.
1261 image::per-user-buffering.png[]
1264 Per-process buffering::
1265 Allocate one set of ring buffers--one per CPU--for each
1266 instrumented process.
1270 .Per-process buffering scheme.
1271 image::per-process-buffering.png[]
1274 The per-process buffering scheme tends to consume more memory than the
1275 per-user option because systems generally have more instrumented
1276 processes than Unix users running instrumented processes. However, the
1277 per-process buffering scheme ensures that one process having a high
1278 event throughput won't fill all the shared sub-buffers of the same
1281 The Linux kernel tracing domain has only one available buffering scheme
1282 which is to allocate a single set of ring buffers for the whole system.
1283 This scheme is similar to the per-user option, but with a single, global
1284 user "running" the kernel.
1287 [[channel-overwrite-mode-vs-discard-mode]]
1288 ==== Overwrite vs. discard event loss modes
1290 When an event occurs, LTTng records it to a specific sub-buffer (yellow
1291 arc in the following animation) of a specific channel's ring buffer.
1292 When there's no space left in a sub-buffer, the tracer marks it as
1293 consumable (red) and another, empty sub-buffer starts receiving the
1294 following event records. A <<lttng-consumerd,consumer daemon>>
1295 eventually consumes the marked sub-buffer (returns to white).
1298 [role="docsvg-channel-subbuf-anim"]
1303 In an ideal world, sub-buffers are consumed faster than they are filled,
1304 as is the case in the previous animation. In the real world,
1305 however, all sub-buffers can be full at some point, leaving no space to
1306 record the following events.
1308 By design, LTTng is a _non-blocking_ tracer: when no empty sub-buffer is
1309 available, it is acceptable to lose event records when the alternative
1310 would be to cause substantial delays in the instrumented application's
1311 execution. LTTng privileges performance over integrity; it aims at
1312 perturbing the traced system as little as possible in order to make
1313 tracing of subtle race conditions and rare interrupt cascades possible.
1315 When it comes to losing event records because no empty sub-buffer is
1316 available, the channel's _event loss mode_ determines what to do. The
1317 available event loss modes are:
1320 Drop the newest event records until a the tracer
1321 releases a sub-buffer.
1324 Clear the sub-buffer containing the oldest event records and start
1325 writing the newest event records there.
1327 This mode is sometimes called _flight recorder mode_ because it's
1329 https://en.wikipedia.org/wiki/Flight_recorder[flight recorder]:
1330 always keep a fixed amount of the latest data.
1332 Which mechanism you should choose depends on your context: prioritize
1333 the newest or the oldest event records in the ring buffer?
1335 Beware that, in overwrite mode, the tracer abandons a whole sub-buffer
1336 as soon as a there's no space left for a new event record, whereas in
1337 discard mode, the tracer only discards the event record that doesn't
1340 In discard mode, LTTng increments a count of lost event records when
1341 an event record is lost and saves this count to the trace. In
1342 overwrite mode, LTTng keeps no information when it overwrites a
1343 sub-buffer before consuming it.
1345 There are a few ways to decrease your probability of losing event
1347 <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>> shows
1348 how you can fine-une the sub-buffer count and size of a channel to
1349 virtually stop losing event records, though at the cost of greater
1353 [[channel-subbuf-size-vs-subbuf-count]]
1354 ==== Sub-buffer count and size
1356 When you <<enabling-disabling-channels,create a channel>>, you can
1357 set its number of sub-buffers and their size.
1359 Note that there is noticeable CPU overhead introduced when
1360 switching sub-buffers (marking a full one as consumable and switching
1361 to an empty one for the following events to be recorded). Knowing this,
1362 the following list presents a few practical situations along with how
1363 to configure the sub-buffer count and size for them:
1365 * **High event throughput**: In general, prefer bigger sub-buffers to
1366 lower the risk of losing event records.
1368 Having bigger sub-buffers also ensures a lower sub-buffer switching
1371 The number of sub-buffers is only meaningful if you create the channel
1372 in overwrite mode: in this case, if a sub-buffer overwrite happens, the
1373 other sub-buffers are left unaltered.
1375 * **Low event throughput**: In general, prefer smaller sub-buffers
1376 since the risk of losing event records is low.
1378 Because events occur less frequently, the sub-buffer switching frequency
1379 should remain low and thus the tracer's overhead should not be a
1382 * **Low memory system**: If your target system has a low memory
1383 limit, prefer fewer first, then smaller sub-buffers.
1385 Even if the system is limited in memory, you want to keep the
1386 sub-buffers as big as possible to avoid a high sub-buffer switching
1389 Note that LTTng uses http://diamon.org/ctf/[CTF] as its trace format,
1390 which means event data is very compact. For example, the average
1391 LTTng kernel event record weights about 32{nbsp}bytes. Thus, a
1392 sub-buffer size of 1{nbsp}MiB is considered big.
1394 The previous situations highlight the major trade-off between a few big
1395 sub-buffers and more, smaller sub-buffers: sub-buffer switching
1396 frequency vs. how much data is lost in overwrite mode. Assuming a
1397 constant event throughput and using the overwrite mode, the two
1398 following configurations have the same ring buffer total size:
1401 [role="docsvg-channel-subbuf-size-vs-count-anim"]
1406 * **2 sub-buffers of 4{nbsp}MiB each**: Expect a very low sub-buffer
1407 switching frequency, but if a sub-buffer overwrite happens, half of
1408 the event records so far (4{nbsp}MiB) are definitely lost.
1409 * **8 sub-buffers of 1{nbsp}MiB each**: Expect 4{nbsp}times the tracer's
1410 overhead as the previous configuration, but if a sub-buffer
1411 overwrite happens, only the eighth of event records so far are
1414 In discard mode, the sub-buffers count parameter is pointless: use two
1415 sub-buffers and set their size according to the requirements of your
1419 [[channel-switch-timer]]
1420 ==== Switch timer period
1422 The _switch timer period_ is an important configurable attribute of
1423 a channel to ensure periodic sub-buffer flushing.
1425 When the _switch timer_ expires, a sub-buffer switch happens. You can
1426 set the switch timer period attribute when you
1427 <<enabling-disabling-channels,create a channel>> to ensure that event
1428 data is consumed and committed to trace files or to a distant relay
1429 daemon periodically in case of a low event throughput.
1432 [role="docsvg-channel-switch-timer"]
1437 This attribute is also convenient when you use big sub-buffers to cope
1438 with a sporadic high event throughput, even if the throughput is
1442 [[channel-read-timer]]
1443 ==== Read timer period
1445 By default, the LTTng tracers use a notification mechanism to signal a
1446 full sub-buffer so that a consumer daemon can consume it. When such
1447 notifications must be avoided, for example in real-time applications,
1448 you can use the channel's _read timer_ instead. When the read timer
1449 fires, the <<lttng-consumerd,consumer daemon>> checks for full,
1450 consumable sub-buffers.
1453 [[tracefile-rotation]]
1454 ==== Trace file count and size
1456 By default, trace files can grow as large as needed. You can set the
1457 maximum size of each trace file that a channel writes when you
1458 <<enabling-disabling-channels,create a channel>>. When the size of
1459 a trace file reaches the channel's fixed maximum size, LTTng creates
1460 another file to contain the next event records. LTTng appends a file
1461 count to each trace file name in this case.
1463 If you set the trace file size attribute when you create a channel, the
1464 maximum number of trace files that LTTng creates is _unlimited_ by
1465 default. To limit them, you can also set a maximum number of trace
1466 files. When the number of trace files reaches the channel's fixed
1467 maximum count, the oldest trace file is overwritten. This mechanism is
1468 called _trace file rotation_.
1472 === Instrumentation point, event rule, event, and event record
1474 An _event rule_ is a set of conditions which must be **all** satisfied
1475 for LTTng to record an occuring event.
1477 You set the conditions when you <<enabling-disabling-events,create
1480 You always attach an event rule to <<channel,channel>> when you create
1483 When an event passes the conditions of an event rule, LTTng records it
1484 in one of the attached channel's sub-buffers.
1486 The available conditions, as of LTTng{nbsp}{revision}, are:
1488 * The event rule _is enabled_.
1489 * The instrumentation point's type _is{nbsp}T_.
1490 * The instrumentation point's name (sometimes called _event name_)
1491 _matches{nbsp}N_, but _is not{nbsp}E_.
1492 * The instrumentation point's log level _is as severe as{nbsp}L_, or
1493 _is exactly{nbsp}L_.
1494 * The fields of the event's payload _satisfy_ a filter
1495 expression{nbsp}__F__.
1497 As you can see, all the conditions but the dynamic filter are related to
1498 the event rule's status or to the instrumentation point, not to the
1499 occurring events. This is why, without a filter, checking if an event
1500 passes an event rule is not a dynamic task: when you create or modify an
1501 event rule, all the tracers of its tracing domain enable or disable the
1502 instrumentation points themselves once. This is possible because the
1503 attributes of an instrumentation point (type, name, and log level) are
1504 defined statically. In other words, without a dynamic filter, the tracer
1505 _does not evaluate_ the arguments of an instrumentation point unless it
1506 matches an enabled event rule.
1508 Note that, for LTTng to record an event, the <<channel,channel>> to
1509 which a matching event rule is attached must also be enabled, and the
1510 tracing session owning this channel must be active.
1513 .Logical path from an instrumentation point to an event record.
1514 image::event-rule.png[]
1516 .Event, event record, or event rule?
1518 With so many similar terms, it's easy to get confused.
1520 An **event** is the consequence of the execution of an _instrumentation
1521 point_, like a tracepoint that you manually place in some source code,
1522 or a Linux kernel KProbe. An event is said to _occur_ at a specific
1523 time. Different actions can be taken upon the occurance of an event,
1524 like record the event's payload to a buffer.
1526 An **event record** is the representation of an event in a sub-buffer. A
1527 tracer is responsible for capturing the payload of an event, current
1528 context variables, the event's ID, and the event's timestamp. LTTng
1529 can append this sub-buffer to a trace file.
1531 An **event rule** is a set of conditions which must all be satisfied for
1532 LTTng to record an occuring event. Events still occur without
1533 satisfying event rules, but LTTng does not record them.
1538 == Components of noch:{LTTng}
1540 The second _T_ in _LTTng_ stands for _toolkit_: it would be wrong
1541 to call LTTng a simple _tool_ since it is composed of multiple
1542 interacting components. This section describes those components,
1543 explains their respective roles, and shows how they connect together to
1544 form the LTTng ecosystem.
1546 The following diagram shows how the most important components of LTTng
1547 interact with user applications, the Linux kernel, and you:
1550 .Control and trace data paths between LTTng components.
1551 image::plumbing.png[]
1553 The LTTng project incorporates:
1555 * **LTTng-tools**: Libraries and command-line interface to
1556 control tracing sessions.
1557 ** <<lttng-sessiond,Session daemon>> (man:lttng-sessiond(8)).
1558 ** <<lttng-consumerd,Consumer daemon>> (man:lttng-consumerd(8)).
1559 ** <<lttng-relayd,Relay daemon>> (man:lttng-relayd(8)).
1560 ** <<liblttng-ctl-lttng,Tracing control library>> (`liblttng-ctl`).
1561 ** <<lttng-cli,Tracing control command-line tool>> (man:lttng(1)).
1562 * **LTTng-UST**: Libraries and Java/Python packages to trace user
1564 ** <<lttng-ust,User space tracing library>> (`liblttng-ust`) and its
1565 headers to instrument and trace any native user application.
1566 ** <<prebuilt-ust-helpers,Preloadable user space tracing helpers>>:
1567 *** `liblttng-ust-libc-wrapper`
1568 *** `liblttng-ust-pthread-wrapper`
1569 *** `liblttng-ust-cyg-profile`
1570 *** `liblttng-ust-cyg-profile-fast`
1571 *** `liblttng-ust-dl`
1572 ** User space tracepoint provider source files generator command-line
1573 tool (man:lttng-gen-tp(1)).
1574 ** <<lttng-ust-agents,LTTng-UST Java agent>> to instrument and trace
1575 Java applications using `java.util.logging` or
1576 Apache log4j 1.2 logging.
1577 ** <<lttng-ust-agents,LTTng-UST Python agent>> to instrument
1578 Python applications using the standard `logging` package.
1579 * **LTTng-modules**: <<lttng-modules,Linux kernel modules>> to trace
1581 ** LTTng kernel tracer module.
1582 ** Tracing ring buffer kernel modules.
1583 ** Probe kernel modules.
1584 ** LTTng logger kernel module.
1588 === Tracing control command-line interface
1591 .The tracing control command-line interface.
1592 image::plumbing-lttng-cli.png[]
1594 The _man:lttng(1) command-line tool_ is the standard user interface to
1595 control LTTng <<tracing-session,tracing sessions>>. The cmd:lttng tool
1596 is part of LTTng-tools.
1598 The cmd:lttng tool is linked with
1599 <<liblttng-ctl-lttng,`liblttng-ctl`>> to communicate with
1600 one or more <<lttng-sessiond,session daemons>> behind the scenes.
1602 The cmd:lttng tool has a Git-like interface:
1606 lttng <general options> <command> <command options>
1609 The <<controlling-tracing,Tracing control>> section explores the
1610 available features of LTTng using the cmd:lttng tool.
1613 [[liblttng-ctl-lttng]]
1614 === Tracing control library
1617 .The tracing control library.
1618 image::plumbing-liblttng-ctl.png[]
1620 The _LTTng control library_, `liblttng-ctl`, is used to communicate
1621 with a <<lttng-sessiond,session daemon>> using a C API that hides the
1622 underlying protocol's details. `liblttng-ctl` is part of LTTng-tools.
1624 The <<lttng-cli,cmd:lttng command-line tool>>
1625 is linked with `liblttng-ctl`.
1627 You can use `liblttng-ctl` in C or $$C++$$ source code by including its
1632 #include <lttng/lttng.h>
1635 Some objects are referenced by name (C string), such as tracing
1636 sessions, but most of them require to create a handle first using
1637 `lttng_create_handle()`.
1639 The best available developer documentation for `liblttng-ctl` is, as of
1640 LTTng{nbsp}{revision}, its installed header files. Every function and
1641 structure is thoroughly documented.
1645 === User space tracing library
1648 .The user space tracing library.
1649 image::plumbing-liblttng-ust.png[]
1651 The _user space tracing library_, `liblttng-ust` (see man:lttng-ust(3)),
1652 is the LTTng user space tracer. It receives commands from a
1653 <<lttng-sessiond,session daemon>>, for example to
1654 enable and disable specific instrumentation points, and writes event
1655 records to ring buffers shared with a
1656 <<lttng-consumerd,consumer daemon>>.
1657 `liblttng-ust` is part of LTTng-UST.
1659 Public C header files are installed beside `liblttng-ust` to
1660 instrument any <<c-application,C or $$C++$$ application>>.
1662 <<lttng-ust-agents,LTTng-UST agents>>, which are regular Java and Python
1663 packages, use their own library providing tracepoints which is
1664 linked with `liblttng-ust`.
1666 An application or library does not have to initialize `liblttng-ust`
1667 manually: its constructor does the necessary tasks to properly register
1668 to a session daemon. The initialization phase also enables the
1669 instrumentation points matching the <<event,event rules>> that you
1673 [[lttng-ust-agents]]
1674 === User space tracing agents
1677 .The user space tracing agents.
1678 image::plumbing-lttng-ust-agents.png[]
1680 The _LTTng-UST Java and Python agents_ are regular Java and Python
1681 packages which add LTTng tracing capabilities to the
1682 native logging frameworks. The LTTng-UST agents are part of LTTng-UST.
1684 In the case of Java, the
1685 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[`java.util.logging`
1686 core logging facilities] and
1687 https://logging.apache.org/log4j/1.2/[Apache log4j 1.2] are supported.
1688 Note that Apache Log4{nbsp}2 is not supported.
1690 In the case of Python, the standard
1691 https://docs.python.org/3/library/logging.html[`logging`] package
1692 is supported. Both Python 2 and Python 3 modules can import the
1693 LTTng-UST Python agent package.
1695 The applications using the LTTng-UST agents are in the
1696 `java.util.logging` (JUL),
1697 log4j, and Python <<domain,tracing domains>>.
1699 Both agents use the same mechanism to trace the log statements. When an
1700 agent is initialized, it creates a log handler that attaches to the root
1701 logger. The agent also registers to a <<lttng-sessiond,session daemon>>.
1702 When the application executes a log statement, it is passed to the
1703 agent's log handler by the root logger. The agent's log handler calls a
1704 native function in a tracepoint provider package shared library linked
1705 with <<lttng-ust,`liblttng-ust`>>, passing the formatted log message and
1706 other fields, like its logger name and its log level. This native
1707 function contains a user space instrumentation point, hence tracing the
1710 The log level condition of an
1711 <<event,event rule>> is considered when tracing
1712 a Java or a Python application, and it's compatible with the standard
1713 JUL, log4j, and Python log levels.
1717 === LTTng kernel modules
1720 .The LTTng kernel modules.
1721 image::plumbing-lttng-modules.png[]
1723 The _LTTng kernel modules_ are a set of Linux kernel modules
1724 which implement the kernel tracer of the LTTng project. The LTTng
1725 kernel modules are part of LTTng-modules.
1727 The LTTng kernel modules include:
1729 * A set of _probe_ modules.
1731 Each module attaches to a specific subsystem
1732 of the Linux kernel using its tracepoint instrument points. There are
1733 also modules to attach to the entry and return points of the Linux
1734 system call functions.
1736 * _Ring buffer_ modules.
1738 A ring buffer implementation is provided as kernel modules. The LTTng
1739 kernel tracer writes to the ring buffer; a
1740 <<lttng-consumerd,consumer daemon>> reads from the ring buffer.
1742 * The _LTTng kernel tracer_ module.
1743 * The _LTTng logger_ module.
1745 The LTTng logger module implements the special path:{/proc/lttng-logger}
1746 file so that any executable can generate LTTng events by opening and
1747 writing to this file.
1749 See <<proc-lttng-logger-abi,LTTng logger>>.
1751 Generally, you do not have to load the LTTng kernel modules manually
1752 (using man:modprobe(8), for example): a root <<lttng-sessiond,session
1753 daemon>> loads the necessary modules when starting. If you have extra
1754 probe modules, you can specify to load them to the session daemon on
1757 The LTTng kernel modules are installed in
1758 +/usr/lib/modules/__release__/extra+ by default, where +__release__+ is
1759 the kernel release (see `uname --kernel-release`).
1766 .The session daemon.
1767 image::plumbing-sessiond.png[]
1769 The _session daemon_, man:lttng-sessiond(8), is a daemon responsible for
1770 managing tracing sessions and for controlling the various components of
1771 LTTng. The session daemon is part of LTTng-tools.
1773 The session daemon sends control requests to and receives control
1776 * The <<lttng-ust,user space tracing library>>.
1778 Any instance of the user space tracing library first registers to
1779 a session daemon. Then, the session daemon can send requests to
1780 this instance, such as:
1783 ** Get the list of tracepoints.
1784 ** Share an <<event,event rule>> so that the user space tracing library
1785 can enable or disable tracepoints. Amongst the possible conditions
1786 of an event rule is a filter expression which `liblttng-ust` evalutes
1787 when an event occurs.
1788 ** Share <<channel,channel>> attributes and ring buffer locations.
1791 The session daemon and the user space tracing library use a Unix
1792 domain socket for their communication.
1794 * The <<lttng-ust-agents,user space tracing agents>>.
1796 Any instance of a user space tracing agent first registers to
1797 a session daemon. Then, the session daemon can send requests to
1798 this instance, such as:
1801 ** Get the list of loggers.
1802 ** Enable or disable a specific logger.
1805 The session daemon and the user space tracing agent use a TCP connection
1806 for their communication.
1808 * The <<lttng-modules,LTTng kernel tracer>>.
1809 * The <<lttng-consumerd,consumer daemon>>.
1811 The session daemon sends requests to the consumer daemon to instruct
1812 it where to send the trace data streams, amongst other information.
1814 * The <<lttng-relayd,relay daemon>>.
1816 The session daemon receives commands from the
1817 <<liblttng-ctl-lttng,tracing control library>>.
1819 The root session daemon loads the appropriate
1820 <<lttng-modules,LTTng kernel modules>> on startup. It also spawns
1821 a <<lttng-consumerd,consumer daemon>> as soon as you create
1822 an <<event,event rule>>.
1824 The session daemon does not send and receive trace data: this is the
1825 role of the <<lttng-consumerd,consumer daemon>> and
1826 <<lttng-relayd,relay daemon>>. It does, however, generate the
1827 http://diamon.org/ctf/[CTF] metadata stream.
1829 Each Unix user can have its own session daemon instance. The
1830 tracing sessions managed by different session daemons are completely
1833 The root user's session daemon is the only one which is
1834 allowed to control the LTTng kernel tracer, and its spawned consumer
1835 daemon is the only one which is allowed to consume trace data from the
1836 LTTng kernel tracer. Note, however, that any Unix user which is a member
1837 of the <<tracing-group,tracing group>> is allowed
1838 to create <<channel,channels>> in the
1839 Linux kernel <<domain,tracing domain>>, and thus to trace the Linux
1842 The <<lttng-cli,cmd:lttng command-line tool>> automatically starts a
1843 session daemon when using its `create` command if none is currently
1844 running. You can also start the session daemon manually.
1851 .The consumer daemon.
1852 image::plumbing-consumerd.png[]
1854 The _consumer daemon_, man:lttng-consumerd(8), is a daemon which shares
1855 ring buffers with user applications or with the LTTng kernel modules to
1856 collect trace data and send it to some location (on disk or to a
1857 <<lttng-relayd,relay daemon>> over the network). The consumer daemon
1858 is part of LTTng-tools.
1860 You do not start a consumer daemon manually: a consumer daemon is always
1861 spawned by a <<lttng-sessiond,session daemon>> as soon as you create an
1862 <<event,event rule>>, that is, before you start tracing. When you kill
1863 its owner session daemon, the consumer daemon also exits because it is
1864 the session daemon's child process. Command-line options of
1865 man:lttng-sessiond(8) target the consumer daemon process.
1867 There are up to two running consumer daemons per Unix user, whereas only
1868 one session daemon can run per user. This is because each process can be
1869 either 32-bit or 64-bit: if the target system runs a mixture of 32-bit
1870 and 64-bit processes, it is more efficient to have separate
1871 corresponding 32-bit and 64-bit consumer daemons. The root user is an
1872 exception: it can have up to _three_ running consumer daemons: 32-bit
1873 and 64-bit instances for its user applications, and one more
1874 reserved for collecting kernel trace data.
1882 image::plumbing-relayd.png[]
1884 The _relay daemon_, man:lttng-relayd(8), is a daemon acting as a bridge
1885 between remote session and consumer daemons, local trace files, and a
1886 remote live trace viewer. The relay daemon is part of LTTng-tools.
1888 The main purpose of the relay daemon is to implement a receiver of
1889 <<sending-trace-data-over-the-network,trace data over the network>>.
1890 This is useful when the target system does not have much file system
1891 space to record trace files locally.
1893 The relay daemon is also a server to which a
1894 <<lttng-live,live trace viewer>> can
1895 connect. The live trace viewer sends requests to the relay daemon to
1896 receive trace data as the target system emits events. The
1897 communication protocol is named _LTTng live_; it is used over TCP
1900 Note that you can start the relay daemon on the target system directly.
1901 This is the setup of choice when the use case is to view events as
1902 the target system emits them without the need of a remote system.
1906 == [[using-lttng]]Instrumentation
1908 There are many examples of tracing and monitoring in our everyday life:
1910 * You have access to real-time and historical weather reports and
1911 forecasts thanks to weather stations installed around the country.
1912 * You know your heart is safe thanks to an electrocardiogram.
1913 * You make sure not to drive your car too fast and to have enough fuel
1914 to reach your destination thanks to gauges visible on your dashboard.
1916 All the previous examples have something in common: they rely on
1917 **instruments**. Without the electrodes attached to the surface of your
1918 body's skin, cardiac monitoring is futile.
1920 LTTng, as a tracer, is no different from those real life examples. If
1921 you're about to trace a software system or, in other words, record its
1922 history of execution, you better have **instrumentation points** in the
1923 subject you're tracing, that is, the actual software.
1925 Various ways were developed to instrument a piece of software for LTTng
1926 tracing. The most straightforward one is to manually place
1927 instrumentation points, called _tracepoints_, in the software's source
1928 code. It is also possible to add instrumentation points dynamically in
1929 the Linux kernel <<domain,tracing domain>>.
1931 If you're only interested in tracing the Linux kernel, your
1932 instrumentation needs are probably already covered by LTTng's built-in
1933 <<lttng-modules,Linux kernel tracepoints>>. You may also wish to trace a
1934 user application which is already instrumented for LTTng tracing.
1935 In such cases, you can skip this whole section and read the topics of
1936 the <<controlling-tracing,Tracing control>> section.
1938 Many methods are available to instrument a piece of software for LTTng
1941 * <<c-application,User space instrumentation for C and $$C++$$
1943 * <<prebuilt-ust-helpers,Prebuilt user space tracing helpers>>.
1944 * <<java-application,User space Java agent>>.
1945 * <<python-application,User space Python agent>>.
1946 * <<proc-lttng-logger-abi,LTTng logger>>.
1947 * <<instrumenting-linux-kernel,LTTng kernel tracepoints>>.
1951 === [[cxx-application]]User space instrumentation for C and $$C++$$ applications
1953 The procedure to instrument a C or $$C++$$ user application with
1954 the <<lttng-ust,LTTng user space tracing library>>, `liblttng-ust`, is:
1956 . <<tracepoint-provider,Create the source files of a tracepoint provider
1958 . <<probing-the-application-source-code,Add tracepoints to
1959 the application's source code>>.
1960 . <<building-tracepoint-providers-and-user-application,Build and link
1961 a tracepoint provider package and the user application>>.
1963 If you need quick, man:printf(3)-like instrumentation, you can skip
1964 those steps and use <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>>
1967 IMPORTANT: You need to <<installing-lttng,install>> LTTng-UST to
1968 instrument a user application with `liblttng-ust`.
1971 [[tracepoint-provider]]
1972 ==== Create the source files of a tracepoint provider package
1974 A _tracepoint provider_ is a set of compiled functions which provide
1975 **tracepoints** to an application, the type of instrumentation point
1976 supported by LTTng-UST. Those functions can emit events with
1977 user-defined fields and serialize those events as event records to one
1978 or more LTTng-UST <<channel,channel>> sub-buffers. The `tracepoint()`
1979 macro, which you <<probing-the-application-source-code,insert in a user
1980 application's source code>>, calls those functions.
1982 A _tracepoint provider package_ is an object file (`.o`) or a shared
1983 library (`.so`) which contains one or more tracepoint providers.
1984 Its source files are:
1986 * One or more <<tpp-header,tracepoint provider header>> (`.h`).
1987 * A <<tpp-source,tracepoint provider package source>> (`.c`).
1989 A tracepoint provider package is dynamically linked with `liblttng-ust`,
1990 the LTTng user space tracer, at run time.
1993 .User application linked with `liblttng-ust` and containing a tracepoint provider.
1994 image::ust-app.png[]
1996 NOTE: If you need quick, man:printf(3)-like instrumentation, you can
1997 skip creating and using a tracepoint provider and use
1998 <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>> instead.
2002 ===== Create a tracepoint provider header file template
2004 A _tracepoint provider header file_ contains the tracepoint
2005 definitions of a tracepoint provider.
2007 To create a tracepoint provider header file:
2009 . Start from this template:
2013 .Tracepoint provider header file template (`.h` file extension).
2015 #undef TRACEPOINT_PROVIDER
2016 #define TRACEPOINT_PROVIDER provider_name
2018 #undef TRACEPOINT_INCLUDE
2019 #define TRACEPOINT_INCLUDE "./tp.h"
2021 #if !defined(_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
2024 #include <lttng/tracepoint.h>
2027 * Use TRACEPOINT_EVENT(), TRACEPOINT_EVENT_CLASS(),
2028 * TRACEPOINT_EVENT_INSTANCE(), and TRACEPOINT_LOGLEVEL() here.
2033 #include <lttng/tracepoint-event.h>
2039 * `provider_name` with the name of your tracepoint provider.
2040 * `"tp.h"` with the name of your tracepoint provider header file.
2042 . Below the `#include <lttng/tracepoint.h>` line, put your
2043 <<defining-tracepoints,tracepoint definitions>>.
2045 Your tracepoint provider name must be unique amongst all the possible
2046 tracepoint provider names used on the same target system. We
2047 suggest to include the name of your project or company in the name,
2048 for example, `org_lttng_my_project_tpp`.
2050 TIP: [[lttng-gen-tp]]You can use the man:lttng-gen-tp(1) tool to create
2051 this boilerplate for you. When using cmd:lttng-gen-tp, all you need to
2052 write are the <<defining-tracepoints,tracepoint definitions>>.
2055 [[defining-tracepoints]]
2056 ===== Create a tracepoint definition
2058 A _tracepoint definition_ defines, for a given tracepoint:
2060 * Its **input arguments**. They are the macro parameters that the
2061 `tracepoint()` macro accepts for this particular tracepoint
2062 in the user application's source code.
2063 * Its **output event fields**. They are the sources of event fields
2064 that form the payload of any event that the execution of the
2065 `tracepoint()` macro emits for this particular tracepoint.
2067 You can create a tracepoint definition by using the
2068 `TRACEPOINT_EVENT()` macro below the `#include <lttng/tracepoint.h>`
2070 <<tpp-header,tracepoint provider header file template>>.
2072 The syntax of the `TRACEPOINT_EVENT()` macro is:
2075 .`TRACEPOINT_EVENT()` macro syntax.
2078 /* Tracepoint provider name */
2081 /* Tracepoint name */
2084 /* Input arguments */
2089 /* Output event fields */
2098 * `provider_name` with your tracepoint provider name.
2099 * `tracepoint_name` with your tracepoint name.
2100 * `arguments` with the <<tpp-def-input-args,input arguments>>.
2101 * `fields` with the <<tpp-def-output-fields,output event field>>
2104 This tracepoint emits events named `provider_name:tracepoint_name`.
2107 .Event name's length limitation
2109 The concatenation of the tracepoint provider name and the
2110 tracepoint name must not exceed **254 characters**. If it does, the
2111 instrumented application compiles and runs, but LTTng throws multiple
2112 warnings and you could experience serious issues.
2115 [[tpp-def-input-args]]The syntax of the `TP_ARGS()` macro is:
2118 .`TP_ARGS()` macro syntax.
2127 * `type` with the C type of the argument.
2128 * `arg_name` with the argument name.
2130 You can repeat `type` and `arg_name` up to 10 times to have
2131 more than one argument.
2133 .`TP_ARGS()` usage with three arguments.
2145 The `TP_ARGS()` and `TP_ARGS(void)` forms are valid to create a
2146 tracepoint definition with no input arguments.
2148 [[tpp-def-output-fields]]The `TP_FIELDS()` macro contains a list of
2149 `ctf_*()` macros. Each `ctf_*()` macro defines one event field. See
2150 man:lttng-ust(3) for a complete description of the available `ctf_*()`
2151 macros. A `ctf_*()` macro specifies the type, size, and byte order of
2154 Each `ctf_*()` macro takes an _argument expression_ parameter. This is a
2155 C expression that the tracer evalutes at the `tracepoint()` macro site
2156 in the application's source code. This expression provides a field's
2157 source of data. The argument expression can include input argument names
2158 listed in the `TP_ARGS()` macro.
2160 Each `ctf_*()` macro also takes a _field name_ parameter. Field names
2161 must be unique within a given tracepoint definition.
2163 Here's a complete tracepoint definition example:
2165 .Tracepoint definition.
2167 The following tracepoint definition defines a tracepoint which takes
2168 three input arguments and has four output event fields.
2172 #include "my-custom-structure.h"
2178 const struct my_custom_structure*, my_custom_structure,
2183 ctf_string(query_field, query)
2184 ctf_float(double, ratio_field, ratio)
2185 ctf_integer(int, recv_size, my_custom_structure->recv_size)
2186 ctf_integer(int, send_size, my_custom_structure->send_size)
2191 You can refer to this tracepoint definition with the `tracepoint()`
2192 macro in your application's source code like this:
2196 tracepoint(my_provider, my_tracepoint,
2197 my_structure, some_ratio, the_query);
2201 NOTE: The LTTng tracer only evaluates tracepoint arguments at run time
2202 if they satisfy an enabled <<event,event rule>>.
2205 [[using-tracepoint-classes]]
2206 ===== Use a tracepoint class
2208 A _tracepoint class_ is a class of tracepoints which share the same
2209 output event field definitions. A _tracepoint instance_ is one
2210 instance of such a defined tracepoint class, with its own tracepoint
2213 The <<defining-tracepoints,`TRACEPOINT_EVENT()` macro>> is actually a
2214 shorthand which defines both a tracepoint class and a tracepoint
2215 instance at the same time.
2217 When you build a tracepoint provider package, the C or $$C++$$ compiler
2218 creates one serialization function for each **tracepoint class**. A
2219 serialization function is responsible for serializing the event fields
2220 of a tracepoint to a sub-buffer when tracing.
2222 For various performance reasons, when your situation requires multiple
2223 tracepoint definitions with different names, but with the same event
2224 fields, we recommend that you manually create a tracepoint class
2225 and instantiate as many tracepoint instances as needed. One positive
2226 effect of such a design, amongst other advantages, is that all
2227 tracepoint instances of the same tracepoint class reuse the same
2228 serialization function, thus reducing
2229 https://en.wikipedia.org/wiki/Cache_pollution[cache pollution].
2231 .Use a tracepoint class and tracepoint instances.
2233 Consider the following three tracepoint definitions:
2245 ctf_integer(int, userid, userid)
2246 ctf_integer(size_t, len, len)
2258 ctf_integer(int, userid, userid)
2259 ctf_integer(size_t, len, len)
2271 ctf_integer(int, userid, userid)
2272 ctf_integer(size_t, len, len)
2277 In this case, we create three tracepoint classes, with one implicit
2278 tracepoint instance for each of them: `get_account`, `get_settings`, and
2279 `get_transaction`. However, they all share the same event field names
2280 and types. Hence three identical, yet independent serialization
2281 functions are created when you build the tracepoint provider package.
2283 A better design choice is to define a single tracepoint class and three
2284 tracepoint instances:
2288 /* The tracepoint class */
2289 TRACEPOINT_EVENT_CLASS(
2290 /* Tracepoint provider name */
2293 /* Tracepoint class name */
2296 /* Input arguments */
2302 /* Output event fields */
2304 ctf_integer(int, userid, userid)
2305 ctf_integer(size_t, len, len)
2309 /* The tracepoint instances */
2310 TRACEPOINT_EVENT_INSTANCE(
2311 /* Tracepoint provider name */
2314 /* Tracepoint class name */
2317 /* Tracepoint name */
2320 /* Input arguments */
2326 TRACEPOINT_EVENT_INSTANCE(
2335 TRACEPOINT_EVENT_INSTANCE(
2348 [[assigning-log-levels]]
2349 ===== Assign a log level to a tracepoint definition
2351 You can assign an optional _log level_ to a
2352 <<defining-tracepoints,tracepoint definition>>.
2354 Assigning different levels of severity to tracepoint definitions can
2355 be useful: when you <<enabling-disabling-events,create an event rule>>,
2356 you can target tracepoints having a log level as severe as a specific
2359 The concept of LTTng-UST log levels is similar to the levels found
2360 in typical logging frameworks:
2362 * In a logging framework, the log level is given by the function
2363 or method name you use at the log statement site: `debug()`,
2364 `info()`, `warn()`, `error()`, and so on.
2365 * In LTTng-UST, you statically assign the log level to a tracepoint
2366 definition; any `tracepoint()` macro invocation which refers to
2367 this definition has this log level.
2369 You can assign a log level to a tracepoint definition with the
2370 `TRACEPOINT_LOGLEVEL()` macro. You must use this macro _after_ the
2371 <<defining-tracepoints,`TRACEPOINT_EVENT()`>> or
2372 <<using-tracepoint-classes,`TRACEPOINT_INSTANCE()`>> macro for a given
2375 The syntax of the `TRACEPOINT_LOGLEVEL()` macro is:
2378 .`TRACEPOINT_LOGLEVEL()` macro syntax.
2380 TRACEPOINT_LOGLEVEL(provider_name, tracepoint_name, log_level)
2385 * `provider_name` with the tracepoint provider name.
2386 * `tracepoint_name` with the tracepoint name.
2387 * `log_level` with the log level to assign to the tracepoint
2388 definition named `tracepoint_name` in the `provider_name`
2389 tracepoint provider.
2391 See man:lttng-ust(3) for a list of available log level names.
2393 .Assign the `TRACE_DEBUG_UNIT` log level to a tracepoint definition.
2397 /* Tracepoint definition */
2406 ctf_integer(int, userid, userid)
2407 ctf_integer(size_t, len, len)
2411 /* Log level assignment */
2412 TRACEPOINT_LOGLEVEL(my_app, get_transaction, TRACE_DEBUG_UNIT)
2418 ===== Create a tracepoint provider package source file
2420 A _tracepoint provider package source file_ is a C source file which
2421 includes a <<tpp-header,tracepoint provider header file>> to expand its
2422 macros into event serialization and other functions.
2424 You can always use the following tracepoint provider package source
2428 .Tracepoint provider package source file template.
2430 #define TRACEPOINT_CREATE_PROBES
2435 Replace `tp.h` with the name of your <<tpp-header,tracepoint provider
2436 header file>> name. You may also include more than one tracepoint
2437 provider header file here to create a tracepoint provider package
2438 holding more than one tracepoint providers.
2441 [[probing-the-application-source-code]]
2442 ==== Add tracepoints to an application's source code
2444 Once you <<tpp-header,create a tracepoint provider header file>>, you
2445 can use the `tracepoint()` macro in your application's
2446 source code to insert the tracepoints that this header
2447 <<defining-tracepoints,defines>>.
2449 The `tracepoint()` macro takes at least two parameters: the tracepoint
2450 provider name and the tracepoint name. The corresponding tracepoint
2451 definition defines the other parameters.
2453 .`tracepoint()` usage.
2455 The following <<defining-tracepoints,tracepoint definition>> defines a
2456 tracepoint which takes two input arguments and has two output event
2460 .Tracepoint provider header file.
2462 #include "my-custom-structure.h"
2469 const char*, cmd_name
2472 ctf_string(cmd_name, cmd_name)
2473 ctf_integer(int, number_of_args, argc)
2478 You can refer to this tracepoint definition with the `tracepoint()`
2479 macro in your application's source code like this:
2482 .Application's source file.
2486 int main(int argc, char* argv[])
2488 tracepoint(my_provider, my_tracepoint, argc, argv[0]);
2494 Note how the application's source code includes
2495 the tracepoint provider header file containing the tracepoint
2496 definitions to use, path:{tp.h}.
2499 .`tracepoint()` usage with a complex tracepoint definition.
2501 Consider this complex tracepoint definition, where multiple event
2502 fields refer to the same input arguments in their argument expression
2506 .Tracepoint provider header file.
2508 /* For `struct stat` */
2509 #include <sys/types.h>
2510 #include <sys/stat.h>
2522 ctf_integer(int, my_constant_field, 23 + 17)
2523 ctf_integer(int, my_int_arg_field, my_int_arg)
2524 ctf_integer(int, my_int_arg_field2, my_int_arg * my_int_arg)
2525 ctf_integer(int, sum4_field, my_str_arg[0] + my_str_arg[1] +
2526 my_str_arg[2] + my_str_arg[3])
2527 ctf_string(my_str_arg_field, my_str_arg)
2528 ctf_integer_hex(off_t, size_field, st->st_size)
2529 ctf_float(double, size_dbl_field, (double) st->st_size)
2530 ctf_sequence_text(char, half_my_str_arg_field, my_str_arg,
2531 size_t, strlen(my_str_arg) / 2)
2536 You can refer to this tracepoint definition with the `tracepoint()`
2537 macro in your application's source code like this:
2540 .Application's source file.
2542 #define TRACEPOINT_DEFINE
2549 stat("/etc/fstab", &s);
2550 tracepoint(my_provider, my_tracepoint, 23, "Hello, World!", &s);
2556 If you look at the event record that LTTng writes when tracing this
2557 program, assuming the file size of path:{/etc/fstab} is 301{nbsp}bytes,
2558 it should look like this:
2560 .Event record fields
2562 |Field's name |Field's value
2563 |`my_constant_field` |40
2564 |`my_int_arg_field` |23
2565 |`my_int_arg_field2` |529
2567 |`my_str_arg_field` |`Hello, World!`
2568 |`size_field` |0x12d
2569 |`size_dbl_field` |301.0
2570 |`half_my_str_arg_field` |`Hello,`
2574 Sometimes, the arguments you pass to `tracepoint()` are expensive to
2575 compute--they use the call stack, for example. To avoid this
2576 computation when the tracepoint is disabled, you can use the
2577 `tracepoint_enabled()` and `do_tracepoint()` macros.
2579 The syntax of the `tracepoint_enabled()` and `do_tracepoint()` macros
2583 .`tracepoint_enabled()` and `do_tracepoint()` macros syntax.
2585 tracepoint_enabled(provider_name, tracepoint_name)
2586 do_tracepoint(provider_name, tracepoint_name, ...)
2591 * `provider_name` with the tracepoint provider name.
2592 * `tracepoint_name` with the tracepoint name.
2594 `tracepoint_enabled()` returns a non-zero value if the tracepoint named
2595 `tracepoint_name` from the provider named `provider_name` is enabled
2598 `do_tracepoint()` is like `tracepoint()`, except that it doesn't check
2599 if the tracepoint is enabled. Using `tracepoint()` with
2600 `tracepoint_enabled()` is dangerous since `tracepoint()` also contains
2601 the `tracepoint_enabled()` check, thus a race condition is
2602 possible in this situation:
2605 .Possible race condition when using `tracepoint_enabled()` with `tracepoint()`.
2607 if (tracepoint_enabled(my_provider, my_tracepoint)) {
2608 stuff = prepare_stuff();
2611 tracepoint(my_provider, my_tracepoint, stuff);
2614 If the tracepoint is enabled after the condition, then `stuff` is not
2615 prepared: the emitted event will either contain wrong data, or the whole
2616 application could crash (segmentation fault, for example).
2618 NOTE: Neither `tracepoint_enabled()` nor `do_tracepoint()` have an
2619 `STAP_PROBEV()` call. If you need it, you must emit
2623 [[building-tracepoint-providers-and-user-application]]
2624 ==== Build and link a tracepoint provider package and an application
2626 Once you have one or more <<tpp-header,tracepoint provider header
2627 files>> and a <<tpp-source,tracepoint provider package source file>>,
2628 you can create the tracepoint provider package by compiling its source
2629 file. From here, multiple build and run scenarios are possible. The
2630 following table shows common application and library configurations
2631 along with the required command lines to achieve them.
2633 In the following diagrams, we use the following file names:
2636 Executable application.
2639 Application's object file.
2642 Tracepoint provider package object file.
2645 Tracepoint provider package archive file.
2648 Tracepoint provider package shared object file.
2651 User library object file.
2654 User library shared object file.
2656 We use the following symbols in the diagrams of table below:
2659 .Symbols used in the build scenario diagrams.
2660 image::ust-sit-symbols.png[]
2662 We assume that path:{.} is part of the env:LD_LIBRARY_PATH environment
2663 variable in the following instructions.
2665 [role="growable ust-scenarios",cols="asciidoc,asciidoc"]
2666 .Common tracepoint provider package scenarios.
2668 |Scenario |Instructions
2671 The instrumented application is statically linked with
2672 the tracepoint provider package object.
2674 image::ust-sit+app-linked-with-tp-o+app-instrumented.png[]
2677 include::../common/ust-sit-step-tp-o.txt[]
2679 To build the instrumented application:
2681 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2686 #define TRACEPOINT_DEFINE
2690 . Compile the application source file:
2699 . Build the application:
2704 gcc -o app app.o tpp.o -llttng-ust -ldl
2708 To run the instrumented application:
2710 * Start the application:
2720 The instrumented application is statically linked with the
2721 tracepoint provider package archive file.
2723 image::ust-sit+app-linked-with-tp-a+app-instrumented.png[]
2726 To create the tracepoint provider package archive file:
2728 . Compile the <<tpp-source,tracepoint provider package source file>>:
2737 . Create the tracepoint provider package archive file:
2746 To build the instrumented application:
2748 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2753 #define TRACEPOINT_DEFINE
2757 . Compile the application source file:
2766 . Build the application:
2771 gcc -o app app.o tpp.a -llttng-ust -ldl
2775 To run the instrumented application:
2777 * Start the application:
2787 The instrumented application is linked with the tracepoint provider
2788 package shared object.
2790 image::ust-sit+app-linked-with-tp-so+app-instrumented.png[]
2793 include::../common/ust-sit-step-tp-so.txt[]
2795 To build the instrumented application:
2797 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2802 #define TRACEPOINT_DEFINE
2806 . Compile the application source file:
2815 . Build the application:
2820 gcc -o app app.o -ldl -L. -ltpp
2824 To run the instrumented application:
2826 * Start the application:
2836 The tracepoint provider package shared object is preloaded before the
2837 instrumented application starts.
2839 image::ust-sit+tp-so-preloaded+app-instrumented.png[]
2842 include::../common/ust-sit-step-tp-so.txt[]
2844 To build the instrumented application:
2846 . In path:{app.c}, before including path:{tpp.h}, add the
2852 #define TRACEPOINT_DEFINE
2853 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
2857 . Compile the application source file:
2866 . Build the application:
2871 gcc -o app app.o -ldl
2875 To run the instrumented application with tracing support:
2877 * Preload the tracepoint provider package shared object and
2878 start the application:
2883 LD_PRELOAD=./libtpp.so ./app
2887 To run the instrumented application without tracing support:
2889 * Start the application:
2899 The instrumented application dynamically loads the tracepoint provider
2900 package shared object.
2902 See the <<dlclose-warning,warning about `dlclose()`>>.
2904 image::ust-sit+app-dlopens-tp-so+app-instrumented.png[]
2907 include::../common/ust-sit-step-tp-so.txt[]
2909 To build the instrumented application:
2911 . In path:{app.c}, before including path:{tpp.h}, add the
2917 #define TRACEPOINT_DEFINE
2918 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
2922 . Compile the application source file:
2931 . Build the application:
2936 gcc -o app app.o -ldl
2940 To run the instrumented application:
2942 * Start the application:
2952 The application is linked with the instrumented user library.
2954 The instrumented user library is statically linked with the tracepoint
2955 provider package object file.
2957 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-o+lib-instrumented.png[]
2960 include::../common/ust-sit-step-tp-o-fpic.txt[]
2962 To build the instrumented user library:
2964 . In path:{emon.c}, before including path:{tpp.h}, add the
2970 #define TRACEPOINT_DEFINE
2974 . Compile the user library source file:
2979 gcc -I. -fpic -c emon.c
2983 . Build the user library shared object:
2988 gcc -shared -o libemon.so emon.o tpp.o -llttng-ust -ldl
2992 To build the application:
2994 . Compile the application source file:
3003 . Build the application:
3008 gcc -o app app.o -L. -lemon
3012 To run the application:
3014 * Start the application:
3024 The application is linked with the instrumented user library.
3026 The instrumented user library is linked with the tracepoint provider
3027 package shared object.
3029 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3032 include::../common/ust-sit-step-tp-so.txt[]
3034 To build the instrumented user library:
3036 . In path:{emon.c}, before including path:{tpp.h}, add the
3042 #define TRACEPOINT_DEFINE
3046 . Compile the user library source file:
3051 gcc -I. -fpic -c emon.c
3055 . Build the user library shared object:
3060 gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3064 To build the application:
3066 . Compile the application source file:
3075 . Build the application:
3080 gcc -o app app.o -L. -lemon
3084 To run the application:
3086 * Start the application:
3096 The tracepoint provider package shared object is preloaded before the
3099 The application is linked with the instrumented user library.
3101 image::ust-sit+tp-so-preloaded+app-linked-with-lib+lib-instrumented.png[]
3104 include::../common/ust-sit-step-tp-so.txt[]
3106 To build the instrumented user library:
3108 . In path:{emon.c}, before including path:{tpp.h}, add the
3114 #define TRACEPOINT_DEFINE
3115 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3119 . Compile the user library source file:
3124 gcc -I. -fpic -c emon.c
3128 . Build the user library shared object:
3133 gcc -shared -o libemon.so emon.o -ldl
3137 To build the application:
3139 . Compile the application source file:
3148 . Build the application:
3153 gcc -o app app.o -L. -lemon
3157 To run the application with tracing support:
3159 * Preload the tracepoint provider package shared object and
3160 start the application:
3165 LD_PRELOAD=./libtpp.so ./app
3169 To run the application without tracing support:
3171 * Start the application:
3181 The application is linked with the instrumented user library.
3183 The instrumented user library dynamically loads the tracepoint provider
3184 package shared object.
3186 See the <<dlclose-warning,warning about `dlclose()`>>.
3188 image::ust-sit+app-linked-with-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3191 include::../common/ust-sit-step-tp-so.txt[]
3193 To build the instrumented user library:
3195 . In path:{emon.c}, before including path:{tpp.h}, add the
3201 #define TRACEPOINT_DEFINE
3202 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3206 . Compile the user library source file:
3211 gcc -I. -fpic -c emon.c
3215 . Build the user library shared object:
3220 gcc -shared -o libemon.so emon.o -ldl
3224 To build the application:
3226 . Compile the application source file:
3235 . Build the application:
3240 gcc -o app app.o -L. -lemon
3244 To run the application:
3246 * Start the application:
3256 The application dynamically loads the instrumented user library.
3258 The instrumented user library is linked with the tracepoint provider
3259 package shared object.
3261 See the <<dlclose-warning,warning about `dlclose()`>>.
3263 image::ust-sit+app-dlopens-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3266 include::../common/ust-sit-step-tp-so.txt[]
3268 To build the instrumented user library:
3270 . In path:{emon.c}, before including path:{tpp.h}, add the
3276 #define TRACEPOINT_DEFINE
3280 . Compile the user library source file:
3285 gcc -I. -fpic -c emon.c
3289 . Build the user library shared object:
3294 gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3298 To build the application:
3300 . Compile the application source file:
3309 . Build the application:
3314 gcc -o app app.o -ldl -L. -lemon
3318 To run the application:
3320 * Start the application:
3330 The application dynamically loads the instrumented user library.
3332 The instrumented user library dynamically loads the tracepoint provider
3333 package shared object.
3335 See the <<dlclose-warning,warning about `dlclose()`>>.
3337 image::ust-sit+app-dlopens-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3340 include::../common/ust-sit-step-tp-so.txt[]
3342 To build the instrumented user library:
3344 . In path:{emon.c}, before including path:{tpp.h}, add the
3350 #define TRACEPOINT_DEFINE
3351 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3355 . Compile the user library source file:
3360 gcc -I. -fpic -c emon.c
3364 . Build the user library shared object:
3369 gcc -shared -o libemon.so emon.o -ldl
3373 To build the application:
3375 . Compile the application source file:
3384 . Build the application:
3389 gcc -o app app.o -ldl -L. -lemon
3393 To run the application:
3395 * Start the application:
3405 The tracepoint provider package shared object is preloaded before the
3408 The application dynamically loads the instrumented user library.
3410 image::ust-sit+tp-so-preloaded+app-dlopens-lib+lib-instrumented.png[]
3413 include::../common/ust-sit-step-tp-so.txt[]
3415 To build the instrumented user library:
3417 . In path:{emon.c}, before including path:{tpp.h}, add the
3423 #define TRACEPOINT_DEFINE
3424 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3428 . Compile the user library source file:
3433 gcc -I. -fpic -c emon.c
3437 . Build the user library shared object:
3442 gcc -shared -o libemon.so emon.o -ldl
3446 To build the application:
3448 . Compile the application source file:
3457 . Build the application:
3462 gcc -o app app.o -L. -lemon
3466 To run the application with tracing support:
3468 * Preload the tracepoint provider package shared object and
3469 start the application:
3474 LD_PRELOAD=./libtpp.so ./app
3478 To run the application without tracing support:
3480 * Start the application:
3490 The application is statically linked with the tracepoint provider
3491 package object file.
3493 The application is linked with the instrumented user library.
3495 image::ust-sit+app-linked-with-tp-o+app-linked-with-lib+lib-instrumented.png[]
3498 include::../common/ust-sit-step-tp-o.txt[]
3500 To build the instrumented user library:
3502 . In path:{emon.c}, before including path:{tpp.h}, add the
3508 #define TRACEPOINT_DEFINE
3512 . Compile the user library source file:
3517 gcc -I. -fpic -c emon.c
3521 . Build the user library shared object:
3526 gcc -shared -o libemon.so emon.o
3530 To build the application:
3532 . Compile the application source file:
3541 . Build the application:
3546 gcc -o app app.o tpp.o -llttng-ust -ldl -L. -lemon
3550 To run the instrumented application:
3552 * Start the application:
3562 The application is statically linked with the tracepoint provider
3563 package object file.
3565 The application dynamically loads the instrumented user library.
3567 image::ust-sit+app-linked-with-tp-o+app-dlopens-lib+lib-instrumented.png[]
3570 include::../common/ust-sit-step-tp-o.txt[]
3572 To build the application:
3574 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3579 #define TRACEPOINT_DEFINE
3583 . Compile the application source file:
3592 . Build the application:
3597 gcc -Wl,--export-dynamic -o app app.o tpp.o \
3602 The `--export-dynamic` option passed to the linker is necessary for the
3603 dynamically loaded library to ``see'' the tracepoint symbols defined in
3606 To build the instrumented user library:
3608 . Compile the user library source file:
3613 gcc -I. -fpic -c emon.c
3617 . Build the user library shared object:
3622 gcc -shared -o libemon.so emon.o
3626 To run the application:
3628 * Start the application:
3640 .Do not use man:dlclose(3) on a tracepoint provider package
3642 Never use man:dlclose(3) on any shared object which:
3644 * Is linked with, statically or dynamically, a tracepoint provider
3646 * Calls man:dlopen(3) itself to dynamically open a tracepoint provider
3647 package shared object.
3649 This is currently considered **unsafe** due to a lack of reference
3650 counting from LTTng-UST to the shared object.
3652 A known workaround (available since glibc 2.2) is to use the
3653 `RTLD_NODELETE` flag when calling man:dlopen(3) initially. This has the
3654 effect of not unloading the loaded shared object, even if man:dlclose(3)
3657 You can also preload the tracepoint provider package shared object with
3658 the env:LD_PRELOAD environment variable to overcome this limitation.
3662 [[using-lttng-ust-with-daemons]]
3663 ===== Use noch:{LTTng-UST} with daemons
3665 If your instrumented application calls man:fork(2), man:clone(2),
3666 or BSD's man:rfork(2), without a following man:exec(3)-family
3667 system call, you must preload the path:{liblttng-ust-fork.so} shared
3668 object when you start the application.
3672 LD_PRELOAD=liblttng-ust-fork.so ./my-app
3675 If your tracepoint provider package is
3676 a shared library which you also preload, you must put both
3677 shared objects in env:LD_PRELOAD:
3681 LD_PRELOAD=liblttng-ust-fork.so:/path/to/tp.so ./my-app
3687 ===== Use noch:{LTTng-UST} with applications which close file descriptors that don't belong to them
3689 If your instrumented application closes one or more file descriptors
3690 which it did not open itself, you must preload the
3691 path:{liblttng-ust-fd.so} shared object when you start the application:
3695 LD_PRELOAD=liblttng-ust-fd.so ./my-app
3698 Typical use cases include closing all the file descriptors after
3699 man:fork(2) or man:rfork(2) and buggy applications doing
3703 [[lttng-ust-pkg-config]]
3704 ===== Use noch:{pkg-config}
3706 On some distributions, LTTng-UST ships with a
3707 https://www.freedesktop.org/wiki/Software/pkg-config/[pkg-config]
3708 metadata file. If this is your case, then you can use cmd:pkg-config to
3709 build an application on the command line:
3713 gcc -o my-app my-app.o tp.o $(pkg-config --cflags --libs lttng-ust)
3717 [[instrumenting-32-bit-app-on-64-bit-system]]
3718 ===== [[advanced-instrumenting-techniques]]Build a 32-bit instrumented application for a 64-bit target system
3720 In order to trace a 32-bit application running on a 64-bit system,
3721 LTTng must use a dedicated 32-bit
3722 <<lttng-consumerd,consumer daemon>>.
3724 The following steps show how to build and install a 32-bit consumer
3725 daemon, which is _not_ part of the default 64-bit LTTng build, how to
3726 build and install the 32-bit LTTng-UST libraries, and how to build and
3727 link an instrumented 32-bit application in that context.
3729 To build a 32-bit instrumented application for a 64-bit target system,
3730 assuming you have a fresh target system with no installed Userspace RCU
3733 . Download, build, and install a 32-bit version of Userspace RCU:
3739 wget http://lttng.org/files/urcu/userspace-rcu-latest-0.9.tar.bz2 &&
3740 tar -xf userspace-rcu-latest-0.9.tar.bz2 &&
3741 cd userspace-rcu-0.9.* &&
3742 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 &&
3744 sudo make install &&
3749 . Using your distribution's package manager, or from source, install
3750 the following 32-bit versions of the following dependencies of
3751 LTTng-tools and LTTng-UST:
3754 * https://sourceforge.net/projects/libuuid/[libuuid]
3755 * http://directory.fsf.org/wiki/Popt[popt]
3756 * http://www.xmlsoft.org/[libxml2]
3759 . Download, build, and install a 32-bit version of the latest
3760 LTTng-UST{nbsp}{revision}:
3766 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.9.tar.bz2 &&
3767 tar -xf lttng-ust-latest-2.9.tar.bz2 &&
3768 cd lttng-ust-2.9.* &&
3769 ./configure --libdir=/usr/local/lib32 \
3770 CFLAGS=-m32 CXXFLAGS=-m32 \
3771 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
3773 sudo make install &&
3780 Depending on your distribution,
3781 32-bit libraries could be installed at a different location than
3782 `/usr/lib32`. For example, Debian is known to install
3783 some 32-bit libraries in `/usr/lib/i386-linux-gnu`.
3785 In this case, make sure to set `LDFLAGS` to all the
3786 relevant 32-bit library paths, for example:
3790 LDFLAGS='-L/usr/lib/i386-linux-gnu -L/usr/lib32'
3794 . Download the latest LTTng-tools{nbsp}{revision}, build, and install
3795 the 32-bit consumer daemon:
3801 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.9.tar.bz2 &&
3802 tar -xf lttng-tools-latest-2.9.tar.bz2 &&
3803 cd lttng-tools-2.9.* &&
3804 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 CXXFLAGS=-m32 \
3805 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
3807 cd src/bin/lttng-consumerd &&
3808 sudo make install &&
3813 . From your distribution or from source,
3814 <<installing-lttng,install>> the 64-bit versions of
3815 LTTng-UST and Userspace RCU.
3816 . Download, build, and install the 64-bit version of the
3817 latest LTTng-tools{nbsp}{revision}:
3823 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.9.tar.bz2 &&
3824 tar -xf lttng-tools-latest-2.9.tar.bz2 &&
3825 cd lttng-tools-2.9.* &&
3826 ./configure --with-consumerd32-libdir=/usr/local/lib32 \
3827 --with-consumerd32-bin=/usr/local/lib32/lttng/libexec/lttng-consumerd &&
3829 sudo make install &&
3834 . Pass the following options to man:gcc(1), man:g++(1), or man:clang(1)
3835 when linking your 32-bit application:
3838 -m32 -L/usr/lib32 -L/usr/local/lib32 \
3839 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32
3842 For example, let's rebuild the quick start example in
3843 <<tracing-your-own-user-application,Trace a user application>> as an
3844 instrumented 32-bit application:
3849 gcc -m32 -c -I. hello-tp.c
3851 gcc -m32 -o hello hello.o hello-tp.o \
3852 -L/usr/lib32 -L/usr/local/lib32 \
3853 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32 \
3858 No special action is required to execute the 32-bit application and
3859 to trace it: use the command-line man:lttng(1) tool as usual.
3866 man:tracef(3) is a small LTTng-UST API designed for quick,
3867 man:printf(3)-like instrumentation without the burden of
3868 <<tracepoint-provider,creating>> and
3869 <<building-tracepoint-providers-and-user-application,building>>
3870 a tracepoint provider package.
3872 To use `tracef()` in your application:
3874 . In the C or C++ source files where you need to use `tracef()`,
3875 include `<lttng/tracef.h>`:
3880 #include <lttng/tracef.h>
3884 . In the application's source code, use `tracef()` like you would use
3892 tracef("my message: %d (%s)", my_integer, my_string);
3898 . Link your application with `liblttng-ust`:
3903 gcc -o app app.c -llttng-ust
3907 To trace the events that `tracef()` calls emit:
3909 * <<enabling-disabling-events,Create an event rule>> which matches the
3910 `lttng_ust_tracef:*` event name:
3915 lttng enable-event --userspace 'lttng_ust_tracef:*'
3920 .Limitations of `tracef()`
3922 The `tracef()` utility function was developed to make user space tracing
3923 super simple, albeit with notable disadvantages compared to
3924 <<defining-tracepoints,user-defined tracepoints>>:
3926 * All the emitted events have the same tracepoint provider and
3927 tracepoint names, respectively `lttng_ust_tracef` and `event`.
3928 * There is no static type checking.
3929 * The only event record field you actually get, named `msg`, is a string
3930 potentially containing the values you passed to `tracef()`
3931 using your own format string. This also means that you cannot filter
3932 events with a custom expression at run time because there are no
3934 * Since `tracef()` uses the C standard library's man:vasprintf(3)
3935 function behind the scenes to format the strings at run time, its
3936 expected performance is lower than with user-defined tracepoints,
3937 which do not require a conversion to a string.
3939 Taking this into consideration, `tracef()` is useful for some quick
3940 prototyping and debugging, but you should not consider it for any
3941 permanent and serious applicative instrumentation.
3947 ==== Use `tracelog()`
3949 The man:tracelog(3) API is very similar to <<tracef,`tracef()`>>, with
3950 the difference that it accepts an additional log level parameter.
3952 The goal of `tracelog()` is to ease the migration from logging to
3955 To use `tracelog()` in your application:
3957 . In the C or C++ source files where you need to use `tracelog()`,
3958 include `<lttng/tracelog.h>`:
3963 #include <lttng/tracelog.h>
3967 . In the application's source code, use `tracelog()` like you would use
3968 man:printf(3), except for the first parameter which is the log
3976 tracelog(TRACE_WARNING, "my message: %d (%s)",
3977 my_integer, my_string);
3983 See man:lttng-ust(3) for a list of available log level names.
3985 . Link your application with `liblttng-ust`:
3990 gcc -o app app.c -llttng-ust
3994 To trace the events that `tracelog()` calls emit with a log level
3995 _as severe as_ a specific log level:
3997 * <<enabling-disabling-events,Create an event rule>> which matches the
3998 `lttng_ust_tracelog:*` event name and a minimum level
4004 lttng enable-event --userspace 'lttng_ust_tracelog:*'
4005 --loglevel=TRACE_WARNING
4009 To trace the events that `tracelog()` calls emit with a
4010 _specific log level_:
4012 * Create an event rule which matches the `lttng_ust_tracelog:*`
4013 event name and a specific log level:
4018 lttng enable-event --userspace 'lttng_ust_tracelog:*'
4019 --loglevel-only=TRACE_INFO
4024 [[prebuilt-ust-helpers]]
4025 === Prebuilt user space tracing helpers
4027 The LTTng-UST package provides a few helpers in the form or preloadable
4028 shared objects which automatically instrument system functions and
4031 The helper shared objects are normally found in dir:{/usr/lib}. If you
4032 built LTTng-UST <<building-from-source,from source>>, they are probably
4033 located in dir:{/usr/local/lib}.
4035 The installed user space tracing helpers in LTTng-UST{nbsp}{revision}
4038 path:{liblttng-ust-libc-wrapper.so}::
4039 path:{liblttng-ust-pthread-wrapper.so}::
4040 <<liblttng-ust-libc-pthread-wrapper,C{nbsp}standard library
4041 memory and POSIX threads function tracing>>.
4043 path:{liblttng-ust-cyg-profile.so}::
4044 path:{liblttng-ust-cyg-profile-fast.so}::
4045 <<liblttng-ust-cyg-profile,Function entry and exit tracing>>.
4047 path:{liblttng-ust-dl.so}::
4048 <<liblttng-ust-dl,Dynamic linker tracing>>.
4050 To use a user space tracing helper with any user application:
4052 * Preload the helper shared object when you start the application:
4057 LD_PRELOAD=liblttng-ust-libc-wrapper.so my-app
4061 You can preload more than one helper:
4066 LD_PRELOAD=liblttng-ust-libc-wrapper.so:liblttng-ust-dl.so my-app
4072 [[liblttng-ust-libc-pthread-wrapper]]
4073 ==== Instrument C standard library memory and POSIX threads functions
4075 The path:{liblttng-ust-libc-wrapper.so} and
4076 path:{liblttng-ust-pthread-wrapper.so} helpers
4077 add instrumentation to some C standard library and POSIX
4081 .Functions instrumented by preloading path:{liblttng-ust-libc-wrapper.so}.
4083 |TP provider name |TP name |Instrumented function
4085 .6+|`lttng_ust_libc` |`malloc` |man:malloc(3)
4086 |`calloc` |man:calloc(3)
4087 |`realloc` |man:realloc(3)
4088 |`free` |man:free(3)
4089 |`memalign` |man:memalign(3)
4090 |`posix_memalign` |man:posix_memalign(3)
4094 .Functions instrumented by preloading path:{liblttng-ust-pthread-wrapper.so}.
4096 |TP provider name |TP name |Instrumented function
4098 .4+|`lttng_ust_pthread` |`pthread_mutex_lock_req` |man:pthread_mutex_lock(3p) (request time)
4099 |`pthread_mutex_lock_acq` |man:pthread_mutex_lock(3p) (acquire time)
4100 |`pthread_mutex_trylock` |man:pthread_mutex_trylock(3p)
4101 |`pthread_mutex_unlock` |man:pthread_mutex_unlock(3p)
4104 When you preload the shared object, it replaces the functions listed
4105 in the previous tables by wrappers which contain tracepoints and call
4106 the replaced functions.
4109 [[liblttng-ust-cyg-profile]]
4110 ==== Instrument function entry and exit
4112 The path:{liblttng-ust-cyg-profile*.so} helpers can add instrumentation
4113 to the entry and exit points of functions.
4115 man:gcc(1) and man:clang(1) have an option named
4116 https://gcc.gnu.org/onlinedocs/gcc/Code-Gen-Options.html[`-finstrument-functions`]
4117 which generates instrumentation calls for entry and exit to functions.
4118 The LTTng-UST function tracing helpers,
4119 path:{liblttng-ust-cyg-profile.so} and
4120 path:{liblttng-ust-cyg-profile-fast.so}, take advantage of this feature
4121 to add tracepoints to the two generated functions (which contain
4122 `cyg_profile` in their names, hence the helper's name).
4124 To use the LTTng-UST function tracing helper, the source files to
4125 instrument must be built using the `-finstrument-functions` compiler
4128 There are two versions of the LTTng-UST function tracing helper:
4130 * **path:{liblttng-ust-cyg-profile-fast.so}** is a lightweight variant
4131 that you should only use when it can be _guaranteed_ that the
4132 complete event stream is recorded without any lost event record.
4133 Any kind of duplicate information is left out.
4135 Assuming no event record is lost, having only the function addresses on
4136 entry is enough to create a call graph, since an event record always
4137 contains the ID of the CPU that generated it.
4139 You can use a tool like man:addr2line(1) to convert function addresses
4140 back to source file names and line numbers.
4142 * **path:{liblttng-ust-cyg-profile.so}** is a more robust variant
4143 which also works in use cases where event records might get discarded or
4144 not recorded from application startup.
4145 In these cases, the trace analyzer needs more information to be
4146 able to reconstruct the program flow.
4148 See man:lttng-ust-cyg-profile(3) to learn more about the instrumentation
4149 points of this helper.
4151 All the tracepoints that this helper provides have the
4152 log level `TRACE_DEBUG_FUNCTION` (see man:lttng-ust(3)).
4154 TIP: It's sometimes a good idea to limit the number of source files that
4155 you compile with the `-finstrument-functions` option to prevent LTTng
4156 from writing an excessive amount of trace data at run time. When using
4157 man:gcc(1), you can use the
4158 `-finstrument-functions-exclude-function-list` option to avoid
4159 instrument entries and exits of specific function names.
4164 ==== Instrument the dynamic linker
4166 The path:{liblttng-ust-dl.so} helper adds instrumentation to the
4167 man:dlopen(3) and man:dlclose(3) function calls.
4169 See man:lttng-ust-dl(3) to learn more about the instrumentation points
4174 [[java-application]]
4175 === User space Java agent
4177 You can instrument any Java application which uses one of the following
4180 * The https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[**`java.util.logging`**]
4181 (JUL) core logging facilities.
4182 * http://logging.apache.org/log4j/1.2/[**Apache log4j 1.2**], since
4183 LTTng 2.6. Note that Apache Log4j{nbsp}2 is not supported.
4186 .LTTng-UST Java agent imported by a Java application.
4187 image::java-app.png[]
4189 Note that the methods described below are new in LTTng{nbsp}{revision}.
4190 Previous LTTng versions use another technique.
4192 NOTE: We use http://openjdk.java.net/[OpenJDK]{nbsp}8 for development
4193 and https://ci.lttng.org/[continuous integration], thus this version is
4194 directly supported. However, the LTTng-UST Java agent is also tested
4195 with OpenJDK{nbsp}7.
4200 ==== Use the LTTng-UST Java agent for `java.util.logging`
4202 To use the LTTng-UST Java agent in a Java application which uses
4203 `java.util.logging` (JUL):
4205 . In the Java application's source code, import the LTTng-UST
4206 log handler package for `java.util.logging`:
4211 import org.lttng.ust.agent.jul.LttngLogHandler;
4215 . Create an LTTng-UST JUL log handler:
4220 Handler lttngUstLogHandler = new LttngLogHandler();
4224 . Add this handler to the JUL loggers which should emit LTTng events:
4229 Logger myLogger = Logger.getLogger("some-logger");
4231 myLogger.addHandler(lttngUstLogHandler);
4235 . Use `java.util.logging` log statements and configuration as usual.
4236 The loggers with an attached LTTng-UST log handler can emit
4239 . Before exiting the application, remove the LTTng-UST log handler from
4240 the loggers attached to it and call its `close()` method:
4245 myLogger.removeHandler(lttngUstLogHandler);
4246 lttngUstLogHandler.close();
4250 This is not strictly necessary, but it is recommended for a clean
4251 disposal of the handler's resources.
4253 . Include the LTTng-UST Java agent's common and JUL-specific JAR files,
4254 path:{lttng-ust-agent-common.jar} and path:{lttng-ust-agent-jul.jar},
4256 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4257 path] when you build the Java application.
4259 The JAR files are typically located in dir:{/usr/share/java}.
4261 IMPORTANT: The LTTng-UST Java agent must be
4262 <<installing-lttng,installed>> for the logging framework your
4265 .Use the LTTng-UST Java agent for `java.util.logging`.
4270 import java.io.IOException;
4271 import java.util.logging.Handler;
4272 import java.util.logging.Logger;
4273 import org.lttng.ust.agent.jul.LttngLogHandler;
4277 private static final int answer = 42;
4279 public static void main(String[] argv) throws Exception
4282 Logger logger = Logger.getLogger("jello");
4284 // Create an LTTng-UST log handler
4285 Handler lttngUstLogHandler = new LttngLogHandler();
4287 // Add the LTTng-UST log handler to our logger
4288 logger.addHandler(lttngUstLogHandler);
4291 logger.info("some info");
4292 logger.warning("some warning");
4294 logger.finer("finer information; the answer is " + answer);
4296 logger.severe("error!");
4298 // Not mandatory, but cleaner
4299 logger.removeHandler(lttngUstLogHandler);
4300 lttngUstLogHandler.close();
4309 javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4312 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4313 <<enabling-disabling-events,create an event rule>> matching the
4314 `jello` JUL logger, and <<basic-tracing-session-control,start tracing>>:
4319 lttng enable-event --jul jello
4323 Run the compiled class:
4327 java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4330 <<basic-tracing-session-control,Stop tracing>> and inspect the
4340 You can use the opt:lttng-enable-event(1):--loglevel or
4341 opt:lttng-enable-event(1):--loglevel-only option of the
4342 man:lttng-enable-event(1) command to target a range of JUL log levels
4343 or a specific JUL log level.
4348 ==== Use the LTTng-UST Java agent for Apache log4j
4350 To use the LTTng-UST Java agent in a Java application which uses
4353 . In the Java application's source code, import the LTTng-UST
4354 log appender package for Apache log4j:
4359 import org.lttng.ust.agent.log4j.LttngLogAppender;
4363 . Create an LTTng-UST log4j log appender:
4368 Appender lttngUstLogAppender = new LttngLogAppender();
4372 . Add this appender to the log4j loggers which should emit LTTng events:
4377 Logger myLogger = Logger.getLogger("some-logger");
4379 myLogger.addAppender(lttngUstLogAppender);
4383 . Use Apache log4j log statements and configuration as usual. The
4384 loggers with an attached LTTng-UST log appender can emit LTTng events.
4386 . Before exiting the application, remove the LTTng-UST log appender from
4387 the loggers attached to it and call its `close()` method:
4392 myLogger.removeAppender(lttngUstLogAppender);
4393 lttngUstLogAppender.close();
4397 This is not strictly necessary, but it is recommended for a clean
4398 disposal of the appender's resources.
4400 . Include the LTTng-UST Java agent's common and log4j-specific JAR
4401 files, path:{lttng-ust-agent-common.jar} and
4402 path:{lttng-ust-agent-log4j.jar}, in the
4403 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4404 path] when you build the Java application.
4406 The JAR files are typically located in dir:{/usr/share/java}.
4408 IMPORTANT: The LTTng-UST Java agent must be
4409 <<installing-lttng,installed>> for the logging framework your
4412 .Use the LTTng-UST Java agent for Apache log4j.
4417 import org.apache.log4j.Appender;
4418 import org.apache.log4j.Logger;
4419 import org.lttng.ust.agent.log4j.LttngLogAppender;
4423 private static final int answer = 42;
4425 public static void main(String[] argv) throws Exception
4428 Logger logger = Logger.getLogger("jello");
4430 // Create an LTTng-UST log appender
4431 Appender lttngUstLogAppender = new LttngLogAppender();
4433 // Add the LTTng-UST log appender to our logger
4434 logger.addAppender(lttngUstLogAppender);
4437 logger.info("some info");
4438 logger.warn("some warning");
4440 logger.debug("debug information; the answer is " + answer);
4442 logger.fatal("error!");
4444 // Not mandatory, but cleaner
4445 logger.removeAppender(lttngUstLogAppender);
4446 lttngUstLogAppender.close();
4452 Build this example (`$LOG4JPATH` is the path to the Apache log4j JAR
4457 javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH Test.java
4460 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4461 <<enabling-disabling-events,create an event rule>> matching the
4462 `jello` log4j logger, and <<basic-tracing-session-control,start tracing>>:
4467 lttng enable-event --log4j jello
4471 Run the compiled class:
4475 java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH:. Test
4478 <<basic-tracing-session-control,Stop tracing>> and inspect the
4488 You can use the opt:lttng-enable-event(1):--loglevel or
4489 opt:lttng-enable-event(1):--loglevel-only option of the
4490 man:lttng-enable-event(1) command to target a range of Apache log4j log levels
4491 or a specific log4j log level.
4495 [[java-application-context]]
4496 ==== Provide application-specific context fields in a Java application
4498 A Java application-specific context field is a piece of state provided
4499 by the application which <<adding-context,you can add>>, using the
4500 man:lttng-add-context(1) command, to each <<event,event record>>
4501 produced by the log statements of this application.
4503 For example, a given object might have a current request ID variable.
4504 You can create a context information retriever for this object and
4505 assign a name to this current request ID. You can then, using the
4506 man:lttng-add-context(1) command, add this context field by name to
4507 the JUL or log4j <<channel,channel>>.
4509 To provide application-specific context fields in a Java application:
4511 . In the Java application's source code, import the LTTng-UST
4512 Java agent context classes and interfaces:
4517 import org.lttng.ust.agent.context.ContextInfoManager;
4518 import org.lttng.ust.agent.context.IContextInfoRetriever;
4522 . Create a context information retriever class, that is, a class which
4523 implements the `IContextInfoRetriever` interface:
4528 class MyContextInfoRetriever implements IContextInfoRetriever
4531 public Object retrieveContextInfo(String key)
4533 if (key.equals("intCtx")) {
4535 } else if (key.equals("strContext")) {
4536 return "context value!";
4545 This `retrieveContextInfo()` method is the only member of the
4546 `IContextInfoRetriever` interface. Its role is to return the current
4547 value of a state by name to create a context field. The names of the
4548 context fields and which state variables they return depends on your
4551 All primitive types and objects are supported as context fields.
4552 When `retrieveContextInfo()` returns an object, the context field
4553 serializer calls its `toString()` method to add a string field to
4554 event records. The method can also return `null`, which means that
4555 no context field is available for the required name.
4557 . Register an instance of your context information retriever class to
4558 the context information manager singleton:
4563 IContextInfoRetriever cir = new MyContextInfoRetriever();
4564 ContextInfoManager cim = ContextInfoManager.getInstance();
4565 cim.registerContextInfoRetriever("retrieverName", cir);
4569 . Before exiting the application, remove your context information
4570 retriever from the context information manager singleton:
4575 ContextInfoManager cim = ContextInfoManager.getInstance();
4576 cim.unregisterContextInfoRetriever("retrieverName");
4580 This is not strictly necessary, but it is recommended for a clean
4581 disposal of some manager's resources.
4583 . Build your Java application with LTTng-UST Java agent support as
4584 usual, following the procedure for either the <<jul,JUL>> or
4585 <<log4j,Apache log4j>> framework.
4588 .Provide application-specific context fields in a Java application.
4593 import java.util.logging.Handler;
4594 import java.util.logging.Logger;
4595 import org.lttng.ust.agent.jul.LttngLogHandler;
4596 import org.lttng.ust.agent.context.ContextInfoManager;
4597 import org.lttng.ust.agent.context.IContextInfoRetriever;
4601 // Our context information retriever class
4602 private static class MyContextInfoRetriever
4603 implements IContextInfoRetriever
4606 public Object retrieveContextInfo(String key) {
4607 if (key.equals("intCtx")) {
4609 } else if (key.equals("strContext")) {
4610 return "context value!";
4617 private static final int answer = 42;
4619 public static void main(String args[]) throws Exception
4621 // Get the context information manager instance
4622 ContextInfoManager cim = ContextInfoManager.getInstance();
4624 // Create and register our context information retriever
4625 IContextInfoRetriever cir = new MyContextInfoRetriever();
4626 cim.registerContextInfoRetriever("myRetriever", cir);
4629 Logger logger = Logger.getLogger("jello");
4631 // Create an LTTng-UST log handler
4632 Handler lttngUstLogHandler = new LttngLogHandler();
4634 // Add the LTTng-UST log handler to our logger
4635 logger.addHandler(lttngUstLogHandler);
4638 logger.info("some info");
4639 logger.warning("some warning");
4641 logger.finer("finer information; the answer is " + answer);
4643 logger.severe("error!");
4645 // Not mandatory, but cleaner
4646 logger.removeHandler(lttngUstLogHandler);
4647 lttngUstLogHandler.close();
4648 cim.unregisterContextInfoRetriever("myRetriever");
4657 javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4660 <<creating-destroying-tracing-sessions,Create a tracing session>>
4661 and <<enabling-disabling-events,create an event rule>> matching the
4667 lttng enable-event --jul jello
4670 <<adding-context,Add the application-specific context fields>> to the
4675 lttng add-context --jul --type='$app.myRetriever:intCtx'
4676 lttng add-context --jul --type='$app.myRetriever:strContext'
4679 <<basic-tracing-session-control,Start tracing>>:
4686 Run the compiled class:
4690 java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4693 <<basic-tracing-session-control,Stop tracing>> and inspect the
4705 [[python-application]]
4706 === User space Python agent
4708 You can instrument a Python 2 or Python 3 application which uses the
4709 standard https://docs.python.org/3/library/logging.html[`logging`]
4712 Each log statement emits an LTTng event once the
4713 application module imports the
4714 <<lttng-ust-agents,LTTng-UST Python agent>> package.
4717 .A Python application importing the LTTng-UST Python agent.
4718 image::python-app.png[]
4720 To use the LTTng-UST Python agent:
4722 . In the Python application's source code, import the LTTng-UST Python
4732 The LTTng-UST Python agent automatically adds its logging handler to the
4733 root logger at import time.
4735 Any log statement that the application executes before this import does
4736 not emit an LTTng event.
4738 IMPORTANT: The LTTng-UST Python agent must be
4739 <<installing-lttng,installed>>.
4741 . Use log statements and logging configuration as usual.
4742 Since the LTTng-UST Python agent adds a handler to the _root_
4743 logger, you can trace any log statement from any logger.
4745 .Use the LTTng-UST Python agent.
4756 logging.basicConfig()
4757 logger = logging.getLogger('my-logger')
4760 logger.debug('debug message')
4761 logger.info('info message')
4762 logger.warn('warn message')
4763 logger.error('error message')
4764 logger.critical('critical message')
4768 if __name__ == '__main__':
4772 NOTE: `logging.basicConfig()`, which adds to the root logger a basic
4773 logging handler which prints to the standard error stream, is not
4774 strictly required for LTTng-UST tracing to work, but in versions of
4775 Python preceding 3.2, you could see a warning message which indicates
4776 that no handler exists for the logger `my-logger`.
4778 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4779 <<enabling-disabling-events,create an event rule>> matching the
4780 `my-logger` Python logger, and <<basic-tracing-session-control,start
4786 lttng enable-event --python my-logger
4790 Run the Python script:
4797 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
4807 You can use the opt:lttng-enable-event(1):--loglevel or
4808 opt:lttng-enable-event(1):--loglevel-only option of the
4809 man:lttng-enable-event(1) command to target a range of Python log levels
4810 or a specific Python log level.
4812 When an application imports the LTTng-UST Python agent, the agent tries
4813 to register to a <<lttng-sessiond,session daemon>>. Note that you must
4814 <<start-sessiond,start the session daemon>> _before_ you run the Python
4815 application. If a session daemon is found, the agent tries to register
4816 to it during 5{nbsp}seconds, after which the application continues
4817 without LTTng tracing support. You can override this timeout value with
4818 the env:LTTNG_UST_PYTHON_REGISTER_TIMEOUT environment variable
4821 If the session daemon stops while a Python application with an imported
4822 LTTng-UST Python agent runs, the agent retries to connect and to
4823 register to a session daemon every 3{nbsp}seconds. You can override this
4824 delay with the env:LTTNG_UST_PYTHON_REGISTER_RETRY_DELAY environment
4829 [[proc-lttng-logger-abi]]
4832 The `lttng-tracer` Linux kernel module, part of
4833 <<lttng-modules,LTTng-modules>>, creates the special LTTng logger file
4834 path:{/proc/lttng-logger} when it's loaded. Any application can write
4835 text data to this file to emit an LTTng event.
4838 .An application writes to the LTTng logger file to emit an LTTng event.
4839 image::lttng-logger.png[]
4841 The LTTng logger is the quickest method--not the most efficient,
4842 however--to add instrumentation to an application. It is designed
4843 mostly to instrument shell scripts:
4847 echo "Some message, some $variable" > /proc/lttng-logger
4850 Any event that the LTTng logger emits is named `lttng_logger` and
4851 belongs to the Linux kernel <<domain,tracing domain>>. However, unlike
4852 other instrumentation points in the kernel tracing domain, **any Unix
4853 user** can <<enabling-disabling-events,create an event rule>> which
4854 matches its event name, not only the root user or users in the tracing
4857 To use the LTTng logger:
4859 * From any application, write text data to the path:{/proc/lttng-logger}
4862 The `msg` field of `lttng_logger` event records contains the
4865 NOTE: The maximum message length of an LTTng logger event is
4866 1024{nbsp}bytes. Writing more than this makes the LTTng logger emit more
4867 than one event to contain the remaining data.
4869 You should not use the LTTng logger to trace a user application which
4870 can be instrumented in a more efficient way, namely:
4872 * <<c-application,C and $$C++$$ applications>>.
4873 * <<java-application,Java applications>>.
4874 * <<python-application,Python applications>>.
4876 .Use the LTTng logger.
4881 echo 'Hello, World!' > /proc/lttng-logger
4883 df --human-readable --print-type / > /proc/lttng-logger
4886 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4887 <<enabling-disabling-events,create an event rule>> matching the
4888 `lttng_logger` Linux kernel tracepoint, and
4889 <<basic-tracing-session-control,start tracing>>:
4894 lttng enable-event --kernel lttng_logger
4898 Run the Bash script:
4905 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
4916 [[instrumenting-linux-kernel]]
4917 === LTTng kernel tracepoints
4919 NOTE: This section shows how to _add_ instrumentation points to the
4920 Linux kernel. The kernel's subsystems are already thoroughly
4921 instrumented at strategic places for LTTng when you
4922 <<installing-lttng,install>> the <<lttng-modules,LTTng-modules>>
4926 There are two methods to instrument the Linux kernel:
4928 . <<linux-add-lttng-layer,Add an LTTng layer>> over an existing ftrace
4929 tracepoint which uses the `TRACE_EVENT()` API.
4931 Choose this if you want to instrumentation a Linux kernel tree with an
4932 instrumentation point compatible with ftrace, perf, and SystemTap.
4934 . Use an <<linux-lttng-tracepoint-event,LTTng-only approach>> to
4935 instrument an out-of-tree kernel module.
4937 Choose this if you don't need ftrace, perf, or SystemTap support.
4941 [[linux-add-lttng-layer]]
4942 ==== [[instrumenting-linux-kernel-itself]][[mainline-trace-event]][[lttng-adaptation-layer]]Add an LTTng layer to an existing ftrace tracepoint
4944 This section shows how to add an LTTng layer to existing ftrace
4945 instrumentation using the `TRACE_EVENT()` API.
4947 This section does not document the `TRACE_EVENT()` macro. You can
4948 read the following articles to learn more about this API:
4950 * http://lwn.net/Articles/379903/[Using the TRACE_EVENT() macro (Part 1)]
4951 * http://lwn.net/Articles/381064/[Using the TRACE_EVENT() macro (Part 2)]
4952 * http://lwn.net/Articles/383362/[Using the TRACE_EVENT() macro (Part 3)]
4954 The following procedure assumes that your ftrace tracepoints are
4955 correctly defined in their own header and that they are created in
4956 one source file using the `CREATE_TRACE_POINTS` definition.
4958 To add an LTTng layer over an existing ftrace tracepoint:
4960 . Make sure the following kernel configuration options are
4966 * `CONFIG_HIGH_RES_TIMERS`
4967 * `CONFIG_TRACEPOINTS`
4970 . Build the Linux source tree with your custom ftrace tracepoints.
4971 . Boot the resulting Linux image on your target system.
4973 Confirm that the tracepoints exist by looking for their names in the
4974 dir:{/sys/kernel/debug/tracing/events/subsys} directory, where `subsys`
4975 is your subsystem's name.
4977 . Get a copy of the latest LTTng-modules{nbsp}{revision}:
4983 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.9.tar.bz2 &&
4984 tar -xf lttng-modules-latest-2.9.tar.bz2 &&
4985 cd lttng-modules-2.9.*
4989 . In dir:{instrumentation/events/lttng-module}, relative to the root
4990 of the LTTng-modules source tree, create a header file named
4991 +__subsys__.h+ for your custom subsystem +__subsys__+ and write your
4992 LTTng-modules tracepoint definitions using the LTTng-modules
4995 Start with this template:
4999 .path:{instrumentation/events/lttng-module/my_subsys.h}
5002 #define TRACE_SYSTEM my_subsys
5004 #if !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ)
5005 #define _LTTNG_MY_SUBSYS_H
5007 #include "../../../probes/lttng-tracepoint-event.h"
5008 #include <linux/tracepoint.h>
5010 LTTNG_TRACEPOINT_EVENT(
5012 * Format is identical to TRACE_EVENT()'s version for the three
5013 * following macro parameters:
5016 TP_PROTO(int my_int, const char *my_string),
5017 TP_ARGS(my_int, my_string),
5019 /* LTTng-modules specific macros */
5021 ctf_integer(int, my_int_field, my_int)
5022 ctf_string(my_bar_field, my_bar)
5026 #endif /* !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ) */
5028 #include "../../../probes/define_trace.h"
5032 The entries in the `TP_FIELDS()` section are the list of fields for the
5033 LTTng tracepoint. This is similar to the `TP_STRUCT__entry()` part of
5034 ftrace's `TRACE_EVENT()` macro.
5036 See <<lttng-modules-tp-fields,Tracepoint fields macros>> for a
5037 complete description of the available `ctf_*()` macros.
5039 . Create the LTTng-modules probe's kernel module C source file,
5040 +probes/lttng-probe-__subsys__.c+, where +__subsys__+ is your
5045 .path:{probes/lttng-probe-my-subsys.c}
5047 #include <linux/module.h>
5048 #include "../lttng-tracer.h"
5051 * Build-time verification of mismatch between mainline
5052 * TRACE_EVENT() arguments and the LTTng-modules adaptation
5053 * layer LTTNG_TRACEPOINT_EVENT() arguments.
5055 #include <trace/events/my_subsys.h>
5057 /* Create LTTng tracepoint probes */
5058 #define LTTNG_PACKAGE_BUILD
5059 #define CREATE_TRACE_POINTS
5060 #define TRACE_INCLUDE_PATH ../instrumentation/events/lttng-module
5062 #include "../instrumentation/events/lttng-module/my_subsys.h"
5064 MODULE_LICENSE("GPL and additional rights");
5065 MODULE_AUTHOR("Your name <your-email>");
5066 MODULE_DESCRIPTION("LTTng my_subsys probes");
5067 MODULE_VERSION(__stringify(LTTNG_MODULES_MAJOR_VERSION) "."
5068 __stringify(LTTNG_MODULES_MINOR_VERSION) "."
5069 __stringify(LTTNG_MODULES_PATCHLEVEL_VERSION)
5070 LTTNG_MODULES_EXTRAVERSION);
5074 . Edit path:{probes/Makefile} and add your new kernel module object
5075 next to the existing ones:
5079 .path:{probes/Makefile}
5083 obj-m += lttng-probe-module.o
5084 obj-m += lttng-probe-power.o
5086 obj-m += lttng-probe-my-subsys.o
5092 . Build and install the LTTng kernel modules:
5097 make KERNELDIR=/path/to/linux
5098 sudo make modules_install
5102 Replace `/path/to/linux` with the path to the Linux source tree where
5103 you defined and used tracepoints with ftrace's `TRACE_EVENT()` macro.
5105 Note that you can also use the
5106 <<lttng-tracepoint-event-code,`LTTNG_TRACEPOINT_EVENT_CODE()` macro>>
5107 instead of `LTTNG_TRACEPOINT_EVENT()` to use custom local variables and
5108 C code that need to be executed before the event fields are recorded.
5110 The best way to learn how to use the previous LTTng-modules macros is to
5111 inspect the existing LTTng-modules tracepoint definitions in the
5112 dir:{instrumentation/events/lttng-module} header files. Compare them
5113 with the Linux kernel mainline versions in the
5114 dir:{include/trace/events} directory of the Linux source tree.
5118 [[lttng-tracepoint-event-code]]
5119 ===== Use custom C code to access the data for tracepoint fields
5121 Although we recommended to always use the
5122 <<lttng-adaptation-layer,`LTTNG_TRACEPOINT_EVENT()`>> macro to describe
5123 the arguments and fields of an LTTng-modules tracepoint when possible,
5124 sometimes you need a more complex process to access the data that the
5125 tracer records as event record fields. In other words, you need local
5126 variables and multiple C{nbsp}statements instead of simple
5127 argument-based expressions that you pass to the
5128 <<lttng-modules-tp-fields,`ctf_*()` macros of `TP_FIELDS()`>>.
5130 You can use the `LTTNG_TRACEPOINT_EVENT_CODE()` macro instead of
5131 `LTTNG_TRACEPOINT_EVENT()` to declare custom local variables and define
5132 a block of C{nbsp}code to be executed before LTTng records the fields.
5133 The structure of this macro is:
5136 .`LTTNG_TRACEPOINT_EVENT_CODE()` macro syntax.
5138 LTTNG_TRACEPOINT_EVENT_CODE(
5140 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5141 * version for the following three macro parameters:
5144 TP_PROTO(int my_int, const char *my_string),
5145 TP_ARGS(my_int, my_string),
5147 /* Declarations of custom local variables */
5150 unsigned long b = 0;
5151 const char *name = "(undefined)";
5152 struct my_struct *my_struct;
5156 * Custom code which uses both tracepoint arguments
5157 * (in TP_ARGS()) and local variables (in TP_locvar()).
5159 * Local variables are actually members of a structure pointed
5160 * to by the special variable tp_locvar.
5164 tp_locvar->a = my_int + 17;
5165 tp_locvar->my_struct = get_my_struct_at(tp_locvar->a);
5166 tp_locvar->b = my_struct_compute_b(tp_locvar->my_struct);
5167 tp_locvar->name = my_struct_get_name(tp_locvar->my_struct);
5168 put_my_struct(tp_locvar->my_struct);
5177 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5178 * version for this, except that tp_locvar members can be
5179 * used in the argument expression parameters of
5180 * the ctf_*() macros.
5183 ctf_integer(unsigned long, my_struct_b, tp_locvar->b)
5184 ctf_integer(int, my_struct_a, tp_locvar->a)
5185 ctf_string(my_string_field, my_string)
5186 ctf_string(my_struct_name, tp_locvar->name)
5191 IMPORTANT: The C code defined in `TP_code()` must not have any side
5192 effects when executed. In particular, the code must not allocate
5193 memory or get resources without deallocating this memory or putting
5194 those resources afterwards.
5197 [[instrumenting-linux-kernel-tracing]]
5198 ==== Load and unload a custom probe kernel module
5200 You must load a <<lttng-adaptation-layer,created LTTng-modules probe
5201 kernel module>> in the kernel before it can emit LTTng events.
5203 To load the default probe kernel modules and a custom probe kernel
5206 * Use the opt:lttng-sessiond(8):--extra-kmod-probes option to give extra
5207 probe modules to load when starting a root <<lttng-sessiond,session
5211 .Load the `my_subsys`, `usb`, and the default probe modules.
5215 sudo lttng-sessiond --extra-kmod-probes=my_subsys,usb
5220 You only need to pass the subsystem name, not the whole kernel module
5223 To load _only_ a given custom probe kernel module:
5225 * Use the opt:lttng-sessiond(8):--kmod-probes option to give the probe
5226 modules to load when starting a root session daemon:
5229 .Load only the `my_subsys` and `usb` probe modules.
5233 sudo lttng-sessiond --kmod-probes=my_subsys,usb
5238 To confirm that a probe module is loaded:
5245 lsmod | grep lttng_probe_usb
5249 To unload the loaded probe modules:
5251 * Kill the session daemon with `SIGTERM`:
5256 sudo pkill lttng-sessiond
5260 You can also use man:modprobe(8)'s `--remove` option if the session
5261 daemon terminates abnormally.
5264 [[controlling-tracing]]
5267 Once an application or a Linux kernel is
5268 <<instrumenting,instrumented>> for LTTng tracing,
5271 This section is divided in topics on how to use the various
5272 <<plumbing,components of LTTng>>, in particular the <<lttng-cli,cmd:lttng
5273 command-line tool>>, to _control_ the LTTng daemons and tracers.
5275 NOTE: In the following subsections, we refer to an man:lttng(1) command
5276 using its man page name. For example, instead of _Run the `create`
5277 command to..._, we use _Run the man:lttng-create(1) command to..._.
5281 === Start a session daemon
5283 In some situations, you need to run a <<lttng-sessiond,session daemon>>
5284 (man:lttng-sessiond(8)) _before_ you can use the man:lttng(1)
5287 You will see the following error when you run a command while no session
5291 Error: No session daemon is available
5294 The only command that automatically runs a session daemon is
5295 man:lttng-create(1), which you use to
5296 <<creating-destroying-tracing-sessions,create a tracing session>>. While
5297 this is most of the time the first operation that you do, sometimes it's
5298 not. Some examples are:
5300 * <<list-instrumentation-points,List the available instrumentation points>>.
5301 * <<saving-loading-tracing-session,Load a tracing session configuration>>.
5303 [[tracing-group]] Each Unix user must have its own running session
5304 daemon to trace user applications. The session daemon that the root user
5305 starts is the only one allowed to control the LTTng kernel tracer. Users
5306 that are part of the _tracing group_ can control the root session
5307 daemon. The default tracing group name is `tracing`; you can set it to
5308 something else with the opt:lttng-sessiond(8):--group option when you
5309 start the root session daemon.
5311 To start a user session daemon:
5313 * Run man:lttng-sessiond(8):
5318 lttng-sessiond --daemonize
5322 To start the root session daemon:
5324 * Run man:lttng-sessiond(8) as the root user:
5329 sudo lttng-sessiond --daemonize
5333 In both cases, remove the opt:lttng-sessiond(8):--daemonize option to
5334 start the session daemon in foreground.
5336 To stop a session daemon, use man:kill(1) on its process ID (standard
5339 Note that some Linux distributions could manage the LTTng session daemon
5340 as a service. In this case, you should use the service manager to
5341 start, restart, and stop session daemons.
5344 [[creating-destroying-tracing-sessions]]
5345 === Create and destroy a tracing session
5347 Almost all the LTTng control operations happen in the scope of
5348 a <<tracing-session,tracing session>>, which is the dialogue between the
5349 <<lttng-sessiond,session daemon>> and you.
5351 To create a tracing session with a generated name:
5353 * Use the man:lttng-create(1) command:
5362 The created tracing session's name is `auto` followed by the
5365 To create a tracing session with a specific name:
5367 * Use the optional argument of the man:lttng-create(1) command:
5372 lttng create my-session
5376 Replace `my-session` with the specific tracing session name.
5378 LTTng appends the creation date to the created tracing session's name.
5380 LTTng writes the traces of a tracing session in
5381 +$LTTNG_HOME/lttng-trace/__name__+ by default, where +__name__+ is the
5382 name of the tracing session. Note that the env:LTTNG_HOME environment
5383 variable defaults to `$HOME` if not set.
5385 To output LTTng traces to a non-default location:
5387 * Use the opt:lttng-create(1):--output option of the man:lttng-create(1) command:
5392 lttng create --output=/tmp/some-directory my-session
5396 You may create as many tracing sessions as you wish.
5398 To list all the existing tracing sessions for your Unix user:
5400 * Use the man:lttng-list(1) command:
5409 When you create a tracing session, it is set as the _current tracing
5410 session_. The following man:lttng(1) commands operate on the current
5411 tracing session when you don't specify one:
5413 [role="list-3-cols"]
5430 To change the current tracing session:
5432 * Use the man:lttng-set-session(1) command:
5437 lttng set-session new-session
5441 Replace `new-session` by the name of the new current tracing session.
5443 When you are done tracing in a given tracing session, you can destroy
5444 it. This operation frees the resources taken by the tracing session
5445 to destroy; it does not destroy the trace data that LTTng wrote for
5446 this tracing session.
5448 To destroy the current tracing session:
5450 * Use the man:lttng-destroy(1) command:
5460 [[list-instrumentation-points]]
5461 === List the available instrumentation points
5463 The <<lttng-sessiond,session daemon>> can query the running instrumented
5464 user applications and the Linux kernel to get a list of available
5465 instrumentation points. For the Linux kernel <<domain,tracing domain>>,
5466 they are tracepoints and system calls. For the user space tracing
5467 domain, they are tracepoints. For the other tracing domains, they are
5470 To list the available instrumentation points:
5472 * Use the man:lttng-list(1) command with the requested tracing domain's
5476 * opt:lttng-list(1):--kernel: Linux kernel tracepoints (your Unix user
5477 must be a root user, or it must be a member of the
5478 <<tracing-group,tracing group>>).
5479 * opt:lttng-list(1):--kernel with opt:lttng-list(1):--syscall: Linux
5480 kernel system calls (your Unix user must be a root user, or it must be
5481 a member of the tracing group).
5482 * opt:lttng-list(1):--userspace: user space tracepoints.
5483 * opt:lttng-list(1):--jul: `java.util.logging` loggers.
5484 * opt:lttng-list(1):--log4j: Apache log4j loggers.
5485 * opt:lttng-list(1):--python: Python loggers.
5488 .List the available user space tracepoints.
5492 lttng list --userspace
5496 .List the available Linux kernel system call tracepoints.
5500 lttng list --kernel --syscall
5505 [[enabling-disabling-events]]
5506 === Create and enable an event rule
5508 Once you <<creating-destroying-tracing-sessions,create a tracing
5509 session>>, you can create <<event,event rules>> with the
5510 man:lttng-enable-event(1) command.
5512 You specify each condition with a command-line option. The available
5513 condition options are shown in the following table.
5515 [role="growable",cols="asciidoc,asciidoc,default"]
5516 .Condition command-line options for the man:lttng-enable-event(1) command.
5518 |Option |Description |Applicable tracing domains
5524 . +--probe=__ADDR__+
5525 . +--function=__ADDR__+
5528 Instead of using the default _tracepoint_ instrumentation type, use:
5530 . A Linux system call.
5531 . A Linux https://lwn.net/Articles/132196/[KProbe] (symbol or address).
5532 . The entry and return points of a Linux function (symbol or address).
5536 |First positional argument.
5539 Tracepoint or system call name. In the case of a Linux KProbe or
5540 function, this is a custom name given to the event rule. With the
5541 JUL, log4j, and Python domains, this is a logger name.
5543 With a tracepoint, logger, or system call name, the last character
5544 can be `*` to match anything that remains.
5551 . +--loglevel=__LEVEL__+
5552 . +--loglevel-only=__LEVEL__+
5555 . Match only tracepoints or log statements with a logging level at
5556 least as severe as +__LEVEL__+.
5557 . Match only tracepoints or log statements with a logging level
5558 equal to +__LEVEL__+.
5560 See man:lttng-enable-event(1) for the list of available logging level
5563 |User space, JUL, log4j, and Python.
5565 |+--exclude=__EXCLUSIONS__+
5568 When you use a `*` character at the end of the tracepoint or logger
5569 name (first positional argument), exclude the specific names in the
5570 comma-delimited list +__EXCLUSIONS__+.
5573 User space, JUL, log4j, and Python.
5575 |+--filter=__EXPR__+
5578 Match only events which satisfy the expression +__EXPR__+.
5580 See man:lttng-enable-event(1) to learn more about the syntax of a
5587 You attach an event rule to a <<channel,channel>> on creation. If you do
5588 not specify the channel with the opt:lttng-enable-event(1):--channel
5589 option, and if the event rule to create is the first in its
5590 <<domain,tracing domain>> for a given tracing session, then LTTng
5591 creates a _default channel_ for you. This default channel is reused in
5592 subsequent invocations of the man:lttng-enable-event(1) command for the
5593 same tracing domain.
5595 An event rule is always enabled at creation time.
5597 The following examples show how you can combine the previous
5598 command-line options to create simple to more complex event rules.
5600 .Create an event rule targetting a Linux kernel tracepoint (default channel).
5604 lttng enable-event --kernel sched_switch
5608 .Create an event rule matching four Linux kernel system calls (default channel).
5612 lttng enable-event --kernel --syscall open,write,read,close
5616 .Create event rules matching tracepoints with filter expressions (default channel).
5620 lttng enable-event --kernel sched_switch --filter='prev_comm == "bash"'
5625 lttng enable-event --kernel --all \
5626 --filter='$ctx.tid == 1988 || $ctx.tid == 1534'
5631 lttng enable-event --jul my_logger \
5632 --filter='$app.retriever:cur_msg_id > 3'
5635 IMPORTANT: Make sure to always quote the filter string when you
5636 use man:lttng(1) from a shell.
5639 .Create an event rule matching any user space tracepoint of a given tracepoint provider with a log level range (default channel).
5643 lttng enable-event --userspace my_app:'*' --loglevel=TRACE_INFO
5646 IMPORTANT: Make sure to always quote the wildcard character when you
5647 use man:lttng(1) from a shell.
5650 .Create an event rule matching multiple Python loggers with a wildcard and with exclusions (default channel).
5654 lttng enable-event --python my-app.'*' \
5655 --exclude='my-app.module,my-app.hello'
5659 .Create an event rule matching any Apache log4j logger with a specific log level (default channel).
5663 lttng enable-event --log4j --all --loglevel-only=LOG4J_WARN
5667 .Create an event rule attached to a specific channel matching a specific user space tracepoint provider and tracepoint.
5671 lttng enable-event --userspace my_app:my_tracepoint --channel=my-channel
5675 The event rules of a given channel form a whitelist: as soon as an
5676 emitted event passes one of them, LTTng can record the event. For
5677 example, an event named `my_app:my_tracepoint` emitted from a user space
5678 tracepoint with a `TRACE_ERROR` log level passes both of the following
5683 lttng enable-event --userspace my_app:my_tracepoint
5684 lttng enable-event --userspace my_app:my_tracepoint \
5685 --loglevel=TRACE_INFO
5688 The second event rule is redundant: the first one includes
5692 [[disable-event-rule]]
5693 === Disable an event rule
5695 To disable an event rule that you <<enabling-disabling-events,created>>
5696 previously, use the man:lttng-disable-event(1) command. This command
5697 disables _all_ the event rules (of a given tracing domain and channel)
5698 which match an instrumentation point. The other conditions are not
5699 supported as of LTTng{nbsp}{revision}.
5701 The LTTng tracer does not record an emitted event which passes
5702 a _disabled_ event rule.
5704 .Disable an event rule matching a Python logger (default channel).
5708 lttng disable-event --python my-logger
5712 .Disable an event rule matching all `java.util.logging` loggers (default channel).
5716 lttng disable-event --jul '*'
5720 .Disable _all_ the event rules of the default channel.
5722 The opt:lttng-disable-event(1):--all-events option is not, like the
5723 opt:lttng-enable-event(1):--all option of man:lttng-enable-event(1), the
5724 equivalent of the event name `*` (wildcard): it disables _all_ the event
5725 rules of a given channel.
5729 lttng disable-event --jul --all-events
5733 NOTE: You cannot delete an event rule once you create it.
5737 === Get the status of a tracing session
5739 To get the status of the current tracing session, that is, its
5740 parameters, its channels, event rules, and their attributes:
5742 * Use the man:lttng-status(1) command:
5752 To get the status of any tracing session:
5754 * Use the man:lttng-list(1) command with the tracing session's name:
5759 lttng list my-session
5763 Replace `my-session` with the desired tracing session's name.
5766 [[basic-tracing-session-control]]
5767 === Start and stop a tracing session
5769 Once you <<creating-destroying-tracing-sessions,create a tracing
5771 <<enabling-disabling-events,create one or more event rules>>,
5772 you can start and stop the tracers for this tracing session.
5774 To start tracing in the current tracing session:
5776 * Use the man:lttng-start(1) command:
5785 LTTng is very flexible: you can launch user applications before
5786 or after the you start the tracers. The tracers only record the events
5787 if they pass enabled event rules and if they occur while the tracers are
5790 To stop tracing in the current tracing session:
5792 * Use the man:lttng-stop(1) command:
5801 If there were <<channel-overwrite-mode-vs-discard-mode,lost event
5802 records>> or lost sub-buffers since the last time you ran
5803 man:lttng-start(1), warnings are printed when you run the
5804 man:lttng-stop(1) command.
5807 [[enabling-disabling-channels]]
5808 === Create a channel
5810 Once you create a tracing session, you can create a <<channel,channel>>
5811 with the man:lttng-enable-channel(1) command.
5813 Note that LTTng automatically creates a default channel when, for a
5814 given <<domain,tracing domain>>, no channels exist and you
5815 <<enabling-disabling-events,create>> the first event rule. This default
5816 channel is named `channel0` and its attributes are set to reasonable
5817 values. Therefore, you only need to create a channel when you need
5818 non-default attributes.
5820 You specify each non-default channel attribute with a command-line
5821 option when you use the man:lttng-enable-channel(1) command. The
5822 available command-line options are:
5824 [role="growable",cols="asciidoc,asciidoc"]
5825 .Command-line options for the man:lttng-enable-channel(1) command.
5827 |Option |Description
5833 <<channel-overwrite-mode-vs-discard-mode,event loss mode>> instead of
5834 the default _discard_ mode.
5836 |`--buffers-pid` (user space tracing domain only)
5839 Use the per-process <<channel-buffering-schemes,buffering scheme>>
5840 instead of the default per-user buffering scheme.
5842 |+--subbuf-size=__SIZE__+
5845 Allocate sub-buffers of +__SIZE__+ bytes (power of two), for each CPU,
5846 either for each Unix user (default), or for each instrumented process.
5848 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
5850 |+--num-subbuf=__COUNT__+
5853 Allocate +__COUNT__+ sub-buffers (power of two), for each CPU, either
5854 for each Unix user (default), or for each instrumented process.
5856 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
5858 |+--tracefile-size=__SIZE__+
5861 Set the maximum size of each trace file that this channel writes within
5862 a stream to +__SIZE__+ bytes instead of no maximum.
5864 See <<tracefile-rotation,Trace file count and size>>.
5866 |+--tracefile-count=__COUNT__+
5869 Limit the number of trace files that this channel creates to
5870 +__COUNT__+ channels instead of no limit.
5872 See <<tracefile-rotation,Trace file count and size>>.
5874 |+--switch-timer=__PERIODUS__+
5877 Set the <<channel-switch-timer,switch timer period>>
5878 to +__PERIODUS__+{nbsp}µs.
5880 |+--read-timer=__PERIODUS__+
5883 Set the <<channel-read-timer,read timer period>>
5884 to +__PERIODUS__+{nbsp}µs.
5886 |+--output=__TYPE__+ (Linux kernel tracing domain only)
5889 Set the channel's output type to +__TYPE__+, either `mmap` or `splice`.
5893 You can only create a channel in the Linux kernel and user space
5894 <<domain,tracing domains>>: other tracing domains have their own channel
5895 created on the fly when <<enabling-disabling-events,creating event
5900 Because of a current LTTng limitation, you must create all channels
5901 _before_ you <<basic-tracing-session-control,start tracing>> in a given
5902 tracing session, that is, before the first time you run
5905 Since LTTng automatically creates a default channel when you use the
5906 man:lttng-enable-event(1) command with a specific tracing domain, you
5907 cannot, for example, create a Linux kernel event rule, start tracing,
5908 and then create a user space event rule, because no user space channel
5909 exists yet and it's too late to create one.
5911 For this reason, make sure to configure your channels properly
5912 before starting the tracers for the first time!
5915 The following examples show how you can combine the previous
5916 command-line options to create simple to more complex channels.
5918 .Create a Linux kernel channel with default attributes.
5922 lttng enable-channel --kernel my-channel
5926 .Create a user space channel with 4 sub-buffers or 1{nbsp}MiB each, per CPU, per instrumented process.
5930 lttng enable-channel --userspace --num-subbuf=4 --subbuf-size=1M \
5931 --buffers-pid my-channel
5935 .Create a Linux kernel channel which rotates 8 trace files of 4{nbsp}MiB each for each stream
5939 lttng enable-channel --kernel --tracefile-count=8 \
5940 --tracefile-size=4194304 my-channel
5944 .Create a user space channel in overwrite (or _flight recorder_) mode.
5948 lttng enable-channel --userspace --overwrite my-channel
5952 You can <<enabling-disabling-events,create>> the same event rule in
5953 two different channels:
5957 lttng enable-event --userspace --channel=my-channel app:tp
5958 lttng enable-event --userspace --channel=other-channel app:tp
5961 If both channels are enabled, when a tracepoint named `app:tp` is
5962 reached, LTTng records two events, one for each channel.
5966 === Disable a channel
5968 To disable a specific channel that you <<enabling-disabling-channels,created>>
5969 previously, use the man:lttng-disable-channel(1) command.
5971 .Disable a specific Linux kernel channel.
5975 lttng disable-channel --kernel my-channel
5979 The state of a channel precedes the individual states of event rules
5980 attached to it: event rules which belong to a disabled channel, even if
5981 they are enabled, are also considered disabled.
5985 === Add context fields to a channel
5987 Event record fields in trace files provide important information about
5988 events that occured previously, but sometimes some external context may
5989 help you solve a problem faster. Examples of context fields are:
5991 * The **process ID**, **thread ID**, **process name**, and
5992 **process priority** of the thread in which the event occurs.
5993 * The **hostname** of the system on which the event occurs.
5994 * The current values of many possible **performance counters** using
5996 ** CPU cycles, stalled cycles, idle cycles, and the other cycle types.
5998 ** Branch instructions, misses, and loads.
6000 * Any context defined at the application level (supported for the
6001 JUL and log4j <<domain,tracing domains>>).
6003 To get the full list of available context fields, see
6004 `lttng add-context --list`. Some context fields are reserved for a
6005 specific <<domain,tracing domain>> (Linux kernel or user space).
6007 You add context fields to <<channel,channels>>. All the events
6008 that a channel with added context fields records contain those fields.
6010 To add context fields to one or all the channels of a given tracing
6013 * Use the man:lttng-add-context(1) command.
6015 .Add context fields to all the channels of the current tracing session.
6017 The following command line adds the virtual process identifier and
6018 the per-thread CPU cycles count fields to all the user space channels
6019 of the current tracing session.
6023 lttng add-context --userspace --type=vpid --type=perf:thread:cpu-cycles
6027 .Add performance counter context fields by raw ID
6029 See man:lttng-add-context(1) for the exact format of the context field
6030 type, which is partly compatible with the format used in
6035 lttng add-context --userspace --type=perf:thread:raw:r0110:test
6036 lttng add-context --kernel --type=perf:cpu:raw:r0013c:x86unhalted
6040 .Add a context field to a specific channel.
6042 The following command line adds the thread identifier context field
6043 to the Linux kernel channel named `my-channel` in the current
6048 lttng add-context --kernel --channel=my-channel --type=tid
6052 .Add an application-specific context field to a specific channel.
6054 The following command line adds the `cur_msg_id` context field of the
6055 `retriever` context retriever for all the instrumented
6056 <<java-application,Java applications>> recording <<event,event records>>
6057 in the channel named `my-channel`:
6061 lttng add-context --kernel --channel=my-channel \
6062 --type='$app:retriever:cur_msg_id'
6065 IMPORTANT: Make sure to always quote the `$` character when you
6066 use man:lttng-add-context(1) from a shell.
6069 NOTE: You cannot remove context fields from a channel once you add it.
6074 === Track process IDs
6076 It's often useful to allow only specific process IDs (PIDs) to emit
6077 events. For example, you may wish to record all the system calls made by
6078 a given process (à la http://linux.die.net/man/1/strace[strace]).
6080 The man:lttng-track(1) and man:lttng-untrack(1) commands serve this
6081 purpose. Both commands operate on a whitelist of process IDs. You _add_
6082 entries to this whitelist with the man:lttng-track(1) command and remove
6083 entries with the man:lttng-untrack(1) command. Any process which has one
6084 of the PIDs in the whitelist is allowed to emit LTTng events which pass
6085 an enabled <<event,event rule>>.
6087 NOTE: The PID tracker tracks the _numeric process IDs_. Should a
6088 process with a given tracked ID exit and another process be given this
6089 ID, then the latter would also be allowed to emit events.
6091 .Track and untrack process IDs.
6093 For the sake of the following example, assume the target system has 16
6097 <<creating-destroying-tracing-sessions,create a tracing session>>,
6098 the whitelist contains all the possible PIDs:
6101 .All PIDs are tracked.
6102 image::track-all.png[]
6104 When the whitelist is full and you use the man:lttng-track(1) command to
6105 specify some PIDs to track, LTTng first clears the whitelist, then it
6106 tracks the specific PIDs. After:
6110 lttng track --pid=3,4,7,10,13
6116 .PIDs 3, 4, 7, 10, and 13 are tracked.
6117 image::track-3-4-7-10-13.png[]
6119 You can add more PIDs to the whitelist afterwards:
6123 lttng track --pid=1,15,16
6129 .PIDs 1, 15, and 16 are added to the whitelist.
6130 image::track-1-3-4-7-10-13-15-16.png[]
6132 The man:lttng-untrack(1) command removes entries from the PID tracker's
6133 whitelist. Given the previous example, the following command:
6137 lttng untrack --pid=3,7,10,13
6140 leads to this whitelist:
6143 .PIDs 3, 7, 10, and 13 are removed from the whitelist.
6144 image::track-1-4-15-16.png[]
6146 LTTng can track all possible PIDs again using the opt:track(1):--all
6151 lttng track --pid --all
6154 The result is, again:
6157 .All PIDs are tracked.
6158 image::track-all.png[]
6161 .Track only specific PIDs
6163 A very typical use case with PID tracking is to start with an empty
6164 whitelist, then <<basic-tracing-session-control,start the tracers>>, and
6165 then add PIDs manually while tracers are active. You can accomplish this
6166 by using the opt:lttng-untrack(1):--all option of the
6167 man:lttng-untrack(1) command to clear the whitelist after you
6168 <<creating-destroying-tracing-sessions,create a tracing session>>:
6172 lttng untrack --pid --all
6178 .No PIDs are tracked.
6179 image::untrack-all.png[]
6181 If you trace with this whitelist configuration, the tracer records no
6182 events for this <<domain,tracing domain>> because no processes are
6183 tracked. You can use the man:lttng-track(1) command as usual to track
6184 specific PIDs, for example:
6188 lttng track --pid=6,11
6194 .PIDs 6 and 11 are tracked.
6195 image::track-6-11.png[]
6200 [[saving-loading-tracing-session]]
6201 === Save and load tracing session configurations
6203 Configuring a <<tracing-session,tracing session>> can be long. Some of
6204 the tasks involved are:
6206 * <<enabling-disabling-channels,Create channels>> with
6207 specific attributes.
6208 * <<adding-context,Add context fields>> to specific channels.
6209 * <<enabling-disabling-events,Create event rules>> with specific log
6210 level and filter conditions.
6212 If you use LTTng to solve real world problems, chances are you have to
6213 record events using the same tracing session setup over and over,
6214 modifying a few variables each time in your instrumented program
6215 or environment. To avoid constant tracing session reconfiguration,
6216 the man:lttng(1) command-line tool can save and load tracing session
6217 configurations to/from XML files.
6219 To save a given tracing session configuration:
6221 * Use the man:lttng-save(1) command:
6226 lttng save my-session
6230 Replace `my-session` with the name of the tracing session to save.
6232 LTTng saves tracing session configurations to
6233 dir:{$LTTNG_HOME/.lttng/sessions} by default. Note that the
6234 env:LTTNG_HOME environment variable defaults to `$HOME` if not set. Use
6235 the opt:lttng-save(1):--output-path option to change this destination
6238 LTTng saves all configuration parameters, for example:
6240 * The tracing session name.
6241 * The trace data output path.
6242 * The channels with their state and all their attributes.
6243 * The context fields you added to channels.
6244 * The event rules with their state, log level and filter conditions.
6246 To load a tracing session:
6248 * Use the man:lttng-load(1) command:
6253 lttng load my-session
6257 Replace `my-session` with the name of the tracing session to load.
6259 When LTTng loads a configuration, it restores your saved tracing session
6260 as if you just configured it manually.
6262 See man:lttng(1) for the complete list of command-line options. You
6263 can also save and load all many sessions at a time, and decide in which
6264 directory to output the XML files.
6267 [[sending-trace-data-over-the-network]]
6268 === Send trace data over the network
6270 LTTng can send the recorded trace data to a remote system over the
6271 network instead of writing it to the local file system.
6273 To send the trace data over the network:
6275 . On the _remote_ system (which can also be the target system),
6276 start an LTTng <<lttng-relayd,relay daemon>> (man:lttng-relayd(8)):
6285 . On the _target_ system, create a tracing session configured to
6286 send trace data over the network:
6291 lttng create my-session --set-url=net://remote-system
6295 Replace `remote-system` by the host name or IP address of the
6296 remote system. See man:lttng-create(1) for the exact URL format.
6298 . On the target system, use the man:lttng(1) command-line tool as usual.
6299 When tracing is active, the target's consumer daemon sends sub-buffers
6300 to the relay daemon running on the remote system intead of flushing
6301 them to the local file system. The relay daemon writes the received
6302 packets to the local file system.
6304 The relay daemon writes trace files to
6305 +$LTTNG_HOME/lttng-traces/__hostname__/__session__+ by default, where
6306 +__hostname__+ is the host name of the target system and +__session__+
6307 is the tracing session name. Note that the env:LTTNG_HOME environment
6308 variable defaults to `$HOME` if not set. Use the
6309 opt:lttng-relayd(8):--output option of man:lttng-relayd(8) to write
6310 trace files to another base directory.
6315 === View events as LTTng emits them (noch:{LTTng} live)
6317 LTTng live is a network protocol implemented by the <<lttng-relayd,relay
6318 daemon>> (man:lttng-relayd(8)) to allow compatible trace viewers to
6319 display events as LTTng emits them on the target system while tracing is
6322 The relay daemon creates a _tee_: it forwards the trace data to both
6323 the local file system and to connected live viewers:
6326 .The relay daemon creates a _tee_, forwarding the trace data to both trace files and a connected live viewer.
6331 . On the _target system_, create a <<tracing-session,tracing session>>
6337 lttng create --live my-session
6341 This spawns a local relay daemon.
6343 . Start the live viewer and configure it to connect to the relay
6344 daemon. For example, with http://diamon.org/babeltrace[Babeltrace]:
6349 babeltrace --input-format=lttng-live net://localhost/host/hostname/my-session
6356 * `hostname` with the host name of the target system.
6357 * `my-session` with the name of the tracing session to view.
6360 . Configure the tracing session as usual with the man:lttng(1)
6361 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6363 You can list the available live tracing sessions with Babeltrace:
6367 babeltrace --input-format=lttng-live net://localhost
6370 You can start the relay daemon on another system. In this case, you need
6371 to specify the relay daemon's URL when you create the tracing session
6372 with the opt:lttng-create(1):--set-url option. You also need to replace
6373 `localhost` in the procedure above with the host name of the system on
6374 which the relay daemon is running.
6376 See man:lttng-create(1) and man:lttng-relayd(8) for the complete list of
6377 command-line options.
6381 [[taking-a-snapshot]]
6382 === Take a snapshot of the current sub-buffers of a tracing session
6384 The normal behavior of LTTng is to append full sub-buffers to growing
6385 trace data files. This is ideal to keep a full history of the events
6386 that occurred on the target system, but it can
6387 represent too much data in some situations. For example, you may wish
6388 to trace your application continuously until some critical situation
6389 happens, in which case you only need the latest few recorded
6390 events to perform the desired analysis, not multi-gigabyte trace files.
6392 With the man:lttng-snapshot(1) command, you can take a snapshot of the
6393 current sub-buffers of a given <<tracing-session,tracing session>>.
6394 LTTng can write the snapshot to the local file system or send it over
6399 . Create a tracing session in _snapshot mode_:
6404 lttng create --snapshot my-session
6408 The <<channel-overwrite-mode-vs-discard-mode,event loss mode>> of
6409 <<channel,channels>> created in this mode is automatically set to
6410 _overwrite_ (flight recorder mode).
6412 . Configure the tracing session as usual with the man:lttng(1)
6413 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6415 . **Optional**: When you need to take a snapshot,
6416 <<basic-tracing-session-control,stop tracing>>.
6418 You can take a snapshot when the tracers are active, but if you stop
6419 them first, you are sure that the data in the sub-buffers does not
6420 change before you actually take the snapshot.
6427 lttng snapshot record --name=my-first-snapshot
6431 LTTng writes the current sub-buffers of all the current tracing
6432 session's channels to trace files on the local file system. Those trace
6433 files have `my-first-snapshot` in their name.
6435 There is no difference between the format of a normal trace file and the
6436 format of a snapshot: viewers of LTTng traces also support LTTng
6439 By default, LTTng writes snapshot files to the path shown by
6440 `lttng snapshot list-output`. You can change this path or decide to send
6441 snapshots over the network using either:
6443 . An output path or URL that you specify when you create the
6445 . An snapshot output path or URL that you add using
6446 `lttng snapshot add-output`
6447 . An output path or URL that you provide directly to the
6448 `lttng snapshot record` command.
6450 Method 3 overrides method 2, which overrides method 1. When you
6451 specify a URL, a relay daemon must listen on a remote system (see
6452 <<sending-trace-data-over-the-network,Send trace data over the network>>).
6457 === Use the machine interface
6459 With any command of the man:lttng(1) command-line tool, you can set the
6460 opt:lttng(1):--mi option to `xml` (before the command name) to get an
6461 XML machine interface output, for example:
6465 lttng --mi=xml enable-event --kernel --syscall open
6468 A schema definition (XSD) is
6469 https://github.com/lttng/lttng-tools/blob/stable-2.9/src/common/mi-lttng-3.0.xsd[available]
6470 to ease the integration with external tools as much as possible.
6474 [[metadata-regenerate]]
6475 === Regenerate the metadata of an LTTng trace
6477 An LTTng trace, which is a http://diamon.org/ctf[CTF] trace, has both
6478 data stream files and a metadata file. This metadata file contains,
6479 amongst other things, information about the offset of the clock sources
6480 used to timestamp <<event,event records>> when tracing.
6482 If, once a <<tracing-session,tracing session>> is
6483 <<basic-tracing-session-control,started>>, a major
6484 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
6485 happens, the trace's clock offset also needs to be updated. You
6486 can use the `metadata` item of the man:lttng-regenerate(1) command
6489 The main use case of this command is to allow a system to boot with
6490 an incorrect wall time and trace it with LTTng before its wall time
6491 is corrected. Once the system is known to be in a state where its
6492 wall time is correct, it can run `lttng regenerate metadata`.
6494 To regenerate the metadata of an LTTng trace:
6496 * Use the `metadata` item of the man:lttng-regenerate(1) command:
6501 lttng regenerate metadata
6507 `lttng regenerate metadata` has the following limitations:
6509 * Tracing session <<creating-destroying-tracing-sessions,created>>
6511 * User space <<channel,channels>>, if any, are using
6512 <<channel-buffering-schemes,per-user buffering>>.
6517 [[regenerate-statedump]]
6518 === Regenerate the state dump of a tracing session
6520 The LTTng kernel and user space tracers generate state dump
6521 <<event,event records>> when the application starts or when you
6522 <<basic-tracing-session-control,start a tracing session>>. An analysis
6523 can use the state dump event records to set an initial state before it
6524 builds the rest of the state from the following event records.
6525 http://tracecompass.org/[Trace Compass] is a notable example of an
6526 application which uses the state dump of an LTTng trace.
6528 When you <<taking-a-snapshot,take a snapshot>>, it's possible that the
6529 state dump event records are not included in the snapshot because they
6530 were recorded to a sub-buffer that has been consumed or overwritten
6533 You can use the `lttng regenerate statedump` command to emit the state
6534 dump event records again.
6536 To regenerate the state dump of the current tracing session, provided
6537 create it in snapshot mode, before you take a snapshot:
6539 . Use the `statedump` item of the man:lttng-regenerate(1) command:
6544 lttng regenerate statedump
6548 . <<basic-tracing-session-control,Stop the tracing session>>:
6557 . <<taking-a-snapshot,Take a snapshot>>:
6562 lttng snapshot record --name=my-snapshot
6566 Depending on the event throughput, you should run steps 1 and 2
6567 as closely as possible.
6569 NOTE: To record the state dump events, you need to
6570 <<enabling-disabling-events,create event rules>> which enable them.
6571 LTTng-UST state dump tracepoints start with `lttng_ust_statedump:`.
6572 LTTng-modules state dump tracepoints start with `lttng_statedump_`.
6576 [[persistent-memory-file-systems]]
6577 === Record trace data on persistent memory file systems
6579 https://en.wikipedia.org/wiki/Non-volatile_random-access_memory[Non-volatile random-access memory]
6580 (NVRAM) is random-access memory that retains its information when power
6581 is turned off (non-volatile). Systems with such memory can store data
6582 structures in RAM and retrieve them after a reboot, without flushing
6583 to typical _storage_.
6585 Linux supports NVRAM file systems thanks to either
6586 http://pramfs.sourceforge.net/[PRAMFS] or
6587 https://www.kernel.org/doc/Documentation/filesystems/dax.txt[DAX]{nbsp}+{nbsp}http://lkml.iu.edu/hypermail/linux/kernel/1504.1/03463.html[pmem]
6588 (requires Linux 4.1+).
6590 This section does not describe how to operate such file systems;
6591 we assume that you have a working persistent memory file system.
6593 When you create a <<tracing-session,tracing session>>, you can specify
6594 the path of the shared memory holding the sub-buffers. If you specify a
6595 location on an NVRAM file system, then you can retrieve the latest
6596 recorded trace data when the system reboots after a crash.
6598 To record trace data on a persistent memory file system and retrieve the
6599 trace data after a system crash:
6601 . Create a tracing session with a sub-buffer shared memory path located
6602 on an NVRAM file system:
6607 lttng create --shm-path=/path/to/shm
6611 . Configure the tracing session as usual with the man:lttng(1)
6612 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6614 . After a system crash, use the man:lttng-crash(1) command-line tool to
6615 view the trace data recorded on the NVRAM file system:
6620 lttng-crash /path/to/shm
6624 The binary layout of the ring buffer files is not exactly the same as
6625 the trace files layout. This is why you need to use man:lttng-crash(1)
6626 instead of your preferred trace viewer directly.
6628 To convert the ring buffer files to LTTng trace files:
6630 * Use the opt:lttng-crash(1):--extract option of man:lttng-crash(1):
6635 lttng-crash --extract=/path/to/trace /path/to/shm
6643 [[lttng-modules-ref]]
6644 === noch:{LTTng-modules}
6648 [[lttng-tracepoint-enum]]
6649 ==== `LTTNG_TRACEPOINT_ENUM()` usage
6651 Use the `LTTNG_TRACEPOINT_ENUM()` macro to define an enumeration:
6655 LTTNG_TRACEPOINT_ENUM(name, TP_ENUM_VALUES(entries))
6660 * `name` with the name of the enumeration (C identifier, unique
6661 amongst all the defined enumerations).
6662 * `entries` with a list of enumeration entries.
6664 The available enumeration entry macros are:
6666 +ctf_enum_value(__name__, __value__)+::
6667 Entry named +__name__+ mapped to the integral value +__value__+.
6669 +ctf_enum_range(__name__, __begin__, __end__)+::
6670 Entry named +__name__+ mapped to the range of integral values between
6671 +__begin__+ (included) and +__end__+ (included).
6673 +ctf_enum_auto(__name__)+::
6674 Entry named +__name__+ mapped to the integral value following the
6675 last mapping's value.
6677 The last value of a `ctf_enum_value()` entry is its +__value__+
6680 The last value of a `ctf_enum_range()` entry is its +__end__+ parameter.
6682 If `ctf_enum_auto()` is the first entry in the list, its integral
6685 Use the `ctf_enum()` <<lttng-modules-tp-fields,field definition macro>>
6686 to use a defined enumeration as a tracepoint field.
6688 .Define an enumeration with `LTTNG_TRACEPOINT_ENUM()`.
6692 LTTNG_TRACEPOINT_ENUM(
6695 ctf_enum_auto("AUTO: EXPECT 0")
6696 ctf_enum_value("VALUE: 23", 23)
6697 ctf_enum_value("VALUE: 27", 27)
6698 ctf_enum_auto("AUTO: EXPECT 28")
6699 ctf_enum_range("RANGE: 101 TO 303", 101, 303)
6700 ctf_enum_auto("AUTO: EXPECT 304")
6708 [[lttng-modules-tp-fields]]
6709 ==== Tracepoint fields macros (for `TP_FIELDS()`)
6711 [[tp-fast-assign]][[tp-struct-entry]]The available macros to define
6712 tracepoint fields, which must be listed within `TP_FIELDS()` in
6713 `LTTNG_TRACEPOINT_EVENT()`, are:
6715 [role="func-desc growable",cols="asciidoc,asciidoc"]
6716 .Available macros to define LTTng-modules tracepoint fields
6718 |Macro |Description and parameters
6721 +ctf_integer(__t__, __n__, __e__)+
6723 +ctf_integer_nowrite(__t__, __n__, __e__)+
6725 +ctf_user_integer(__t__, __n__, __e__)+
6727 +ctf_user_integer_nowrite(__t__, __n__, __e__)+
6729 Standard integer, displayed in base 10.
6732 Integer C type (`int`, `long`, `size_t`, ...).
6738 Argument expression.
6741 +ctf_integer_hex(__t__, __n__, __e__)+
6743 +ctf_user_integer_hex(__t__, __n__, __e__)+
6745 Standard integer, displayed in base 16.
6754 Argument expression.
6756 |+ctf_integer_oct(__t__, __n__, __e__)+
6758 Standard integer, displayed in base 8.
6767 Argument expression.
6770 +ctf_integer_network(__t__, __n__, __e__)+
6772 +ctf_user_integer_network(__t__, __n__, __e__)+
6774 Integer in network byte order (big-endian), displayed in base 10.
6783 Argument expression.
6786 +ctf_integer_network_hex(__t__, __n__, __e__)+
6788 +ctf_user_integer_network_hex(__t__, __n__, __e__)+
6790 Integer in network byte order, displayed in base 16.
6799 Argument expression.
6802 +ctf_enum(__N__, __t__, __n__, __e__)+
6804 +ctf_enum_nowrite(__N__, __t__, __n__, __e__)+
6806 +ctf_user_enum(__N__, __t__, __n__, __e__)+
6808 +ctf_user_enum_nowrite(__N__, __t__, __n__, __e__)+
6813 Name of a <<lttng-tracepoint-enum,previously defined enumeration>>.
6816 Integer C type (`int`, `long`, `size_t`, ...).
6822 Argument expression.
6825 +ctf_string(__n__, __e__)+
6827 +ctf_string_nowrite(__n__, __e__)+
6829 +ctf_user_string(__n__, __e__)+
6831 +ctf_user_string_nowrite(__n__, __e__)+
6833 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
6839 Argument expression.
6842 +ctf_array(__t__, __n__, __e__, __s__)+
6844 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
6846 +ctf_user_array(__t__, __n__, __e__, __s__)+
6848 +ctf_user_array_nowrite(__t__, __n__, __e__, __s__)+
6850 Statically-sized array of integers.
6853 Array element C type.
6859 Argument expression.
6865 +ctf_array_bitfield(__t__, __n__, __e__, __s__)+
6867 +ctf_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
6869 +ctf_user_array_bitfield(__t__, __n__, __e__, __s__)+
6871 +ctf_user_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
6873 Statically-sized array of bits.
6875 The type of +__e__+ must be an integer type. +__s__+ is the number
6876 of elements of such type in +__e__+, not the number of bits.
6879 Array element C type.
6885 Argument expression.
6891 +ctf_array_text(__t__, __n__, __e__, __s__)+
6893 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
6895 +ctf_user_array_text(__t__, __n__, __e__, __s__)+
6897 +ctf_user_array_text_nowrite(__t__, __n__, __e__, __s__)+
6899 Statically-sized array, printed as text.
6901 The string does not need to be null-terminated.
6904 Array element C type (always `char`).
6910 Argument expression.
6916 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
6918 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
6920 +ctf_user_sequence(__t__, __n__, __e__, __T__, __E__)+
6922 +ctf_user_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
6924 Dynamically-sized array of integers.
6926 The type of +__E__+ must be unsigned.
6929 Array element C type.
6935 Argument expression.
6938 Length expression C type.
6944 +ctf_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
6946 +ctf_user_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
6948 Dynamically-sized array of integers, displayed in base 16.
6950 The type of +__E__+ must be unsigned.
6953 Array element C type.
6959 Argument expression.
6962 Length expression C type.
6967 |+ctf_sequence_network(__t__, __n__, __e__, __T__, __E__)+
6969 Dynamically-sized array of integers in network byte order (big-endian),
6970 displayed in base 10.
6972 The type of +__E__+ must be unsigned.
6975 Array element C type.
6981 Argument expression.
6984 Length expression C type.
6990 +ctf_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
6992 +ctf_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
6994 +ctf_user_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
6996 +ctf_user_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
6998 Dynamically-sized array of bits.
7000 The type of +__e__+ must be an integer type. +__s__+ is the number
7001 of elements of such type in +__e__+, not the number of bits.
7003 The type of +__E__+ must be unsigned.
7006 Array element C type.
7012 Argument expression.
7015 Length expression C type.
7021 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7023 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7025 +ctf_user_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7027 +ctf_user_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7029 Dynamically-sized array, displayed as text.
7031 The string does not need to be null-terminated.
7033 The type of +__E__+ must be unsigned.
7035 The behaviour is undefined if +__e__+ is `NULL`.
7038 Sequence element C type (always `char`).
7044 Argument expression.
7047 Length expression C type.
7053 Use the `_user` versions when the argument expression, `e`, is
7054 a user space address. In the cases of `ctf_user_integer*()` and
7055 `ctf_user_float*()`, `&e` must be a user space address, thus `e` must
7058 The `_nowrite` versions omit themselves from the session trace, but are
7059 otherwise identical. This means the `_nowrite` fields won't be written
7060 in the recorded trace. Their primary purpose is to make some
7061 of the event context available to the
7062 <<enabling-disabling-events,event filters>> without having to
7063 commit the data to sub-buffers.
7069 Terms related to LTTng and to tracing in general:
7072 The http://diamon.org/babeltrace[Babeltrace] project, which includes
7073 the cmd:babeltrace command, some libraries, and Python bindings.
7075 <<channel-buffering-schemes,buffering scheme>>::
7076 A layout of sub-buffers applied to a given channel.
7078 <<channel,channel>>::
7079 An entity which is responsible for a set of ring buffers.
7081 <<event,Event rules>> are always attached to a specific channel.
7084 A reference of time for a tracer.
7086 <<lttng-consumerd,consumer daemon>>::
7087 A process which is responsible for consuming the full sub-buffers
7088 and write them to a file system or send them over the network.
7090 <<channel-overwrite-mode-vs-discard-mode,discard mode>>:: The event loss
7091 mode in which the tracer _discards_ new event records when there's no
7092 sub-buffer space left to store them.
7095 The consequence of the execution of an instrumentation
7096 point, like a tracepoint that you manually place in some source code,
7097 or a Linux kernel KProbe.
7099 An event is said to _occur_ at a specific time. Different actions can
7100 be taken upon the occurance of an event, like record the event's payload
7103 <<channel-overwrite-mode-vs-discard-mode,event loss mode>>::
7104 The mechanism by which event records of a given channel are lost
7105 (not recorded) when there is no sub-buffer space left to store them.
7107 [[def-event-name]]event name::
7108 The name of an event, which is also the name of the event record.
7109 This is also called the _instrumentation point name_.
7112 A record, in a trace, of the payload of an event which occured.
7114 <<event,event rule>>::
7115 Set of conditions which must be satisfied for one or more occuring
7116 events to be recorded.
7118 `java.util.logging`::
7120 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[core logging facilities].
7122 <<instrumenting,instrumentation>>::
7123 The use of LTTng probes to make a piece of software traceable.
7125 instrumentation point::
7126 A point in the execution path of a piece of software that, when
7127 reached by this execution, can emit an event.
7129 instrumentation point name::
7130 See _<<def-event-name,event name>>_.
7133 A http://logging.apache.org/log4j/1.2/[logging library] for Java
7134 developed by the Apache Software Foundation.
7137 Level of severity of a log statement or user space
7138 instrumentation point.
7141 The _Linux Trace Toolkit: next generation_ project.
7143 <<lttng-cli,cmd:lttng>>::
7144 A command-line tool provided by the LTTng-tools project which you
7145 can use to send and receive control messages to and from a
7149 The https://github.com/lttng/lttng-analyses[LTTng analyses] project,
7150 which is a set of analyzing programs that are used to obtain a
7151 higher level view of an LTTng trace.
7153 cmd:lttng-consumerd::
7154 The name of the consumer daemon program.
7157 A utility provided by the LTTng-tools project which can convert
7158 ring buffer files (usually
7159 <<persistent-memory-file-systems,saved on a persistent memory file system>>)
7162 LTTng Documentation::
7165 <<lttng-live,LTTng live>>::
7166 A communication protocol between the relay daemon and live viewers
7167 which makes it possible to see events "live", as they are received by
7170 <<lttng-modules,LTTng-modules>>::
7171 The https://github.com/lttng/lttng-modules[LTTng-modules] project,
7172 which contains the Linux kernel modules to make the Linux kernel
7173 instrumentation points available for LTTng tracing.
7176 The name of the relay daemon program.
7178 cmd:lttng-sessiond::
7179 The name of the session daemon program.
7182 The https://github.com/lttng/lttng-tools[LTTng-tools] project, which
7183 contains the various programs and libraries used to
7184 <<controlling-tracing,control tracing>>.
7186 <<lttng-ust,LTTng-UST>>::
7187 The https://github.com/lttng/lttng-ust[LTTng-UST] project, which
7188 contains libraries to instrument user applications.
7190 <<lttng-ust-agents,LTTng-UST Java agent>>::
7191 A Java package provided by the LTTng-UST project to allow the
7192 LTTng instrumentation of `java.util.logging` and Apache log4j 1.2
7195 <<lttng-ust-agents,LTTng-UST Python agent>>::
7196 A Python package provided by the LTTng-UST project to allow the
7197 LTTng instrumentation of Python logging statements.
7199 <<channel-overwrite-mode-vs-discard-mode,overwrite mode>>::
7200 The event loss mode in which new event records overwrite older
7201 event records when there's no sub-buffer space left to store them.
7203 <<channel-buffering-schemes,per-process buffering>>::
7204 A buffering scheme in which each instrumented process has its own
7205 sub-buffers for a given user space channel.
7207 <<channel-buffering-schemes,per-user buffering>>::
7208 A buffering scheme in which all the processes of a Unix user share the
7209 same sub-buffer for a given user space channel.
7211 <<lttng-relayd,relay daemon>>::
7212 A process which is responsible for receiving the trace data sent by
7213 a distant consumer daemon.
7216 A set of sub-buffers.
7218 <<lttng-sessiond,session daemon>>::
7219 A process which receives control commands from you and orchestrates
7220 the tracers and various LTTng daemons.
7222 <<taking-a-snapshot,snapshot>>::
7223 A copy of the current data of all the sub-buffers of a given tracing
7224 session, saved as trace files.
7227 One part of an LTTng ring buffer which contains event records.
7230 The time information attached to an event when it is emitted.
7233 A set of files which are the concatenations of one or more
7234 flushed sub-buffers.
7237 The action of recording the events emitted by an application
7238 or by a system, or to initiate such recording by controlling
7242 The http://tracecompass.org[Trace Compass] project and application.
7245 An instrumentation point using the tracepoint mechanism of the Linux
7246 kernel or of LTTng-UST.
7248 tracepoint definition::
7249 The definition of a single tracepoint.
7252 The name of a tracepoint.
7254 tracepoint provider::
7255 A set of functions providing tracepoints to an instrumented user
7258 Not to be confused with a _tracepoint provider package_: many tracepoint
7259 providers can exist within a tracepoint provider package.
7261 tracepoint provider package::
7262 One or more tracepoint providers compiled as an object file or as
7266 A software which records emitted events.
7268 <<domain,tracing domain>>::
7269 A namespace for event sources.
7272 The Unix group in which a Unix user can be to be allowed to trace the
7275 <<tracing-session,tracing session>>::
7276 A stateful dialogue between you and a <<lttng-sessiond,session
7280 An application running in user space, as opposed to a Linux kernel
7281 module, for example.