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 Skip this section if you’re familiar with software tracing and with the
28 * **<<installing-lttng,Installation>>** describes the steps to
29 install the LTTng packages on common Linux distributions and from
32 Skip this section if you already properly installed LTTng on your target
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 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 Skip this section if you don't 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{nbsp}{revision}?
77 LTTng{nbsp}{revision} bears the name _Ta Meilleure_, a Northeast IPA
78 beer brewed by https://lagabiere.com/[Lagabière]. Translating to ``Your
79 best one'', this beer gives out strong aromas of passion fruit, lemon,
80 and peaches. Tastewise, expect a lot of fruit, a creamy texture, and a
81 smooth lingering hop bitterness.
83 New features and changes in LTTng{nbsp}{revision}:
87 * Clear the contents of one or more <<tracing-session,tracing sessions>>
88 without having to destroy and reconfigure them
89 with the new man:lttng-clear(1) command.
91 This is especially useful to clear a tracing session's tracing data
92 between attempts to reproduce a problem.
94 See <<clear,Clear a tracing session>>.
96 * Before LTTng{nbsp}{revision}, the man:lttng-track(1) and
97 man:lttng-untrack(1) commands used to add and remove process IDs
98 (PIDs) to a whitelist so that LTTng would only trace processes with
101 LTTng{nbsp}{revision} adds Unix user IDs (UIDs) and Unix group IDs
102 (GIDs) to the available <<pid-tracking,process attributes to track>>.
103 You can specify numeric user/group IDs and user/group names to track,
108 $ lttng track --userspace --vuid=http,999 --vgid=mysql,9
111 While you can also track UIDs and GIDs with the
112 opt:lttng-enable-event(1):--filter option of the `enable-event` command,
113 this dedicated process attribute tracking approach reduces tracing
114 overhead and prevents the creation of <<def-sub-buffer,sub-buffers>> for
115 the users and groups which LTTng doesn't track.
117 In the command manual pages, the term ``whitelist'' is renamed to
118 ``inclusion set'' to clarify the concept.
120 * The <<lttng-relayd,relay daemon>> can now maintain many files
121 virtually opened without using as many file descriptors (FD). It does
122 so by closing and reopening FDs as needed.
124 This feature is meant as a workaround for users who can't bump the
125 system limit because of permission restrictions.
127 The new opt:lttng-relayd(8):--fd-pool-size relay daemon option
128 sets the maximum number of simultaneously opened file descriptors
129 (using the soft `RLIMIT_NOFILE` resource limit of the process by
130 default; see man:getrlimit(2)).
132 * By default, the relay daemon writes its traces under a predefined
134 +$LTTNG_HOME/lttng-traces/__host__/__session__/__domain__+, with:
141 <<tracing-session,Tracing session>> name.
144 <<domain,Tracing domain>> name (`ust` or `kernel`).
147 Change this hierarchy to group traces by tracing session name rather
149 (+$LTTNG_HOME/lttng-traces/__session__/__host__/__domain__+) with the
150 new relay daemon's opt:lttng-relayd(8):--group-output-by-session option.
152 This feature is especially useful if you're tracing two or more hosts,
153 having different hostnames, which share the same tracing session name as
154 part of their configuration. In this scenario, you can use a descriptive
155 tracing session name (for example, `connection-hang`) across a fleet of
156 machines streaming to a single relay daemon.
158 * The relay daemon has a new opt:lttng-relayd(8):--working-directory
159 option to override its working directory.
161 Linux kernel tracing::
163 * New instrumentation hooks to trace the entry and exit tracepoints of
164 the network reception code paths of the Linux kernel.
166 Use the resulting event records to identify the bounds of a network
167 reception and link the events that occur in the interim (for example,
168 wake-ups) to a specific network reception instance. You can also
169 analyze the network stack's latency thanks to those event records.
171 * The `irqaction` structure's `thread` field, which specifies the
172 process to wake up when a threaded interrupt request (IRQ) occurs, is
173 now part of the `lttng_statedump_interrupt` event record.
175 Use this information to discover which processes handle the various
176 IRQs. You can also associate the events occurring in the context of
177 those processes with their respective IRQ.
179 * New `lttng_statedump_cpu_topology` tracepoint to record the active
182 Use this information to discover which CPUs are SMT siblings or part of
183 the same socket. As of LTTng{nbsp}{revision}, only the x86 architecture
184 is supported since all architectures describe their topologies
187 The tracepoint's `architecture` field is statically defined and exists
188 for all architecture implementations. Analysis tools can therefore
189 anticipate the event record's layout.
191 Event record example:
195 lttng_statedump_cpu_topology:
201 model_name: Intel(R) Core(TM) i7-7600U CPU @ 2.80GHz
207 * New product UUID metadata environment field, `product_uuid`,
208 which LTTng copies from the
209 https://en.wikipedia.org/wiki/Desktop_Management_Interface[Desktop
210 Management Interface] (DMI).
212 Use this environment field to uniquely identify a machine (virtual or
213 physical) in order to correlate traces from multiple virtual machines.
219 What is LTTng? As its name suggests, the _Linux Trace Toolkit: next
220 generation_ is a modern toolkit for tracing Linux systems and
221 applications. So your first question might be:
228 As the history of software engineering progressed and led to what
229 we now take for granted--complex, numerous and
230 interdependent software applications running in parallel on
231 sophisticated operating systems like Linux--the authors of such
232 components, software developers, began feeling a natural
233 urge to have tools that would ensure the robustness and good performance
234 of their masterpieces.
236 One major achievement in this field is, inarguably, the
237 https://www.gnu.org/software/gdb/[GNU debugger (GDB)],
238 an essential tool for developers to find and fix bugs. But even the best
239 debugger won't help make your software run faster, and nowadays, faster
240 software means either more work done by the same hardware, or cheaper
241 hardware for the same work.
243 A _profiler_ is often the tool of choice to identify performance
244 bottlenecks. Profiling is suitable to identify _where_ performance is
245 lost in a given software. The profiler outputs a profile, a statistical
246 summary of observed events, which you may use to discover which
247 functions took the most time to execute. However, a profiler won't
248 report _why_ some identified functions are the bottleneck. Bottlenecks
249 might only occur when specific conditions are met, conditions that are
250 sometimes impossible to capture by a statistical profiler, or impossible
251 to reproduce with an application altered by the overhead of an
252 event-based profiler. For a thorough investigation of software
253 performance issues, a history of execution is essential, with the
254 recorded values of variables and context fields you choose, and
255 with as little influence as possible on the instrumented software. This
256 is where tracing comes in handy.
258 _Tracing_ is a technique used to understand what goes on in a running
259 software system. The software used for tracing is called a _tracer_,
260 which is conceptually similar to a tape recorder. When recording,
261 specific instrumentation points placed in the software source code
262 generate events that are saved on a giant tape: a _trace_ file. You
263 can trace user applications and the operating system at the same time,
264 opening the possibility of resolving a wide range of problems that would
265 otherwise be extremely challenging.
267 Tracing is often compared to _logging_. However, tracers and loggers are
268 two different tools, serving two different purposes. Tracers are
269 designed to record much lower-level events that occur much more
270 frequently than log messages, often in the range of thousands per
271 second, with very little execution overhead. Logging is more appropriate
272 for a very high-level analysis of less frequent events: user accesses,
273 exceptional conditions (errors and warnings, for example), database
274 transactions, instant messaging communications, and such. Simply put,
275 logging is one of the many use cases that can be satisfied with tracing.
277 The list of recorded events inside a trace file can be read manually
278 like a log file for the maximum level of detail, but it is generally
279 much more interesting to perform application-specific analyses to
280 produce reduced statistics and graphs that are useful to resolve a
281 given problem. Trace viewers and analyzers are specialized tools
284 In the end, this is what LTTng is: a powerful, open source set of
285 tools to trace the Linux kernel and user applications at the same time.
286 LTTng is composed of several components actively maintained and
287 developed by its link:/community/#where[community].
290 [[lttng-alternatives]]
291 === Alternatives to noch:{LTTng}
293 Excluding proprietary solutions, a few competing software tracers
296 https://github.com/dtrace4linux/linux[dtrace4linux]::
297 A port of Sun Microsystems's DTrace to Linux.
299 The cmd:dtrace tool interprets user scripts and is responsible for
300 loading code into the Linux kernel for further execution and collecting
303 https://en.wikipedia.org/wiki/Berkeley_Packet_Filter[eBPF]::
304 A subsystem in the Linux kernel in which a virtual machine can
305 execute programs passed from the user space to the kernel.
307 You can attach such programs to tracepoints and kprobes thanks to a
308 system call, and they can output data to the user space when executed
309 thanks to different mechanisms (pipe, VM register values, and eBPF maps,
312 https://www.kernel.org/doc/Documentation/trace/ftrace.txt[ftrace]::
313 The de facto function tracer of the Linux kernel.
315 Its user interface is a set of special files in sysfs.
317 https://perf.wiki.kernel.org/[perf]::
318 A performance analysis tool for Linux which supports hardware
319 performance counters, tracepoints, as well as other counters and
322 perf's controlling utility is the cmd:perf command line/text UI tool.
324 http://linux.die.net/man/1/strace[strace]::
325 A command-line utility which records system calls made by a
326 user process, as well as signal deliveries and changes of process
329 strace makes use of https://en.wikipedia.org/wiki/Ptrace[ptrace] to
330 fulfill its function.
332 http://www.sysdig.org/[sysdig]::
333 Like SystemTap, uses scripts to analyze Linux kernel events.
335 You write scripts, or _chisels_ in sysdig's jargon, in Lua and sysdig
336 executes them while it traces the system or afterwards. sysdig's
337 interface is the cmd:sysdig command-line tool as well as the text
338 UI-based cmd:csysdig tool.
340 https://sourceware.org/systemtap/[SystemTap]::
341 A Linux kernel and user space tracer which uses custom user scripts
342 to produce plain text traces.
344 SystemTap converts the scripts to the C language, and then compiles them
345 as Linux kernel modules which are loaded to produce trace data.
346 SystemTap's primary user interface is the cmd:stap command-line tool.
348 The main distinctive features of LTTng is that it produces correlated
349 kernel and user space traces, as well as doing so with the lowest
350 overhead amongst other solutions. It produces trace files in the
351 http://diamon.org/ctf[CTF] format, a file format optimized
352 for the production and analyses of multi-gigabyte data.
354 LTTng is the result of more than 10{nbsp}years of active open source
355 development by a community of passionate developers. LTTng is currently
356 available on major desktop and server Linux distributions.
358 The main interface for tracing control is a single command-line tool
359 named cmd:lttng. The latter can create several tracing sessions, enable
360 and disable events on the fly, filter events efficiently with custom
361 user expressions, start and stop tracing, and much more. LTTng can
362 record the traces on the file system or send them over the network, and
363 keep them totally or partially. You can view the traces once tracing
364 becomes inactive or in real-time.
366 <<installing-lttng,Install LTTng now>> and
367 <<getting-started,start tracing>>!
373 **LTTng** is a set of software <<plumbing,components>> which interact to
374 <<instrumenting,instrument>> the Linux kernel and user applications, and
375 to <<controlling-tracing,control tracing>> (start and stop
376 tracing, enable and disable event rules, and the rest). Those
377 components are bundled into the following packages:
380 Libraries and command-line interface to control tracing.
383 Linux kernel modules to instrument and trace the kernel.
386 Libraries and Java/Python packages to instrument and trace user
389 Most distributions mark the LTTng-modules and LTTng-UST packages as
390 optional when installing LTTng-tools (which is always required). In the
391 following sections, we always provide the steps to install all three,
394 * You only need to install LTTng-modules if you intend to trace the
396 * You only need to install LTTng-UST if you intend to trace user
400 .Availability of LTTng{nbsp}{revision} for major Linux distributions as
401 of 5{nbsp}August{nbsp}2020.
403 |Distribution |Available in releases
405 |https://www.ubuntu.com/[Ubuntu]
406 |Ubuntu{nbsp}16.04 _Xenial Xerus_, Ubuntu{nbsp}18.04 _Bionic Beaver_,
407 and Ubuntu{nbsp}20.04 _Focal Fossa_:
408 <<ubuntu-ppa,use the LTTng Stable{nbsp}{revision} PPA>>.
410 |https://www.debian.org/[Debian]
411 |<<debian,Debian "bullseye" (testing)>>.
413 |https://www.archlinux.org/[Arch Linux]
414 |<<arch-linux,_Community_ repository and AUR>>.
416 |https://alpinelinux.org/[Alpine Linux]
417 |<<alpine-linux,Alpine Linux{nbsp}3.12>>.
419 |https://www.redhat.com/[RHEL] and https://www.suse.com/[SLES]
420 |See http://packages.efficios.com/[EfficiOS Enterprise Packages].
425 === Ubuntu: noch:{LTTng} Stable {revision} PPA
427 The https://launchpad.net/~lttng/+archive/ubuntu/stable-{revision}[LTTng
428 Stable{nbsp}{revision} PPA] offers the latest stable
429 LTTng{nbsp}{revision} packages for Ubuntu{nbsp}16.04 _Xenial Xerus_,
430 Ubuntu{nbsp}18.04 _Bionic Beaver_, and Ubuntu{nbsp}20.04 _Focal Fossa_.
432 To install LTTng{nbsp}{revision} from the LTTng Stable{nbsp}{revision}
435 . Add the LTTng Stable{nbsp}{revision} PPA repository and update the
441 # apt-add-repository ppa:lttng/stable-2.12
446 . Install the main LTTng{nbsp}{revision} packages:
451 # apt-get install lttng-tools
452 # apt-get install lttng-modules-dkms
453 # apt-get install liblttng-ust-dev
457 . **If you need to instrument and trace
458 <<java-application,Java applications>>**, install the LTTng-UST
464 # apt-get install liblttng-ust-agent-java
468 . **If you need to instrument and trace
469 <<python-application,Python{nbsp}3 applications>>**, install the
470 LTTng-UST Python agent:
475 # apt-get install python3-lttngust
483 To install LTTng{nbsp}{revision} on Debian "bullseye" (testing):
485 . Install the main LTTng{nbsp}{revision} packages:
490 # apt-get install lttng-modules-dkms
491 # apt-get install liblttng-ust-dev
492 # apt-get install lttng-tools
496 . **If you need to instrument and trace <<java-application,Java
497 applications>>**, install the LTTng-UST Java agent:
502 # apt-get install liblttng-ust-agent-java
506 . **If you need to instrument and trace <<python-application,Python
507 applications>>**, install the LTTng-UST Python agent:
512 # apt-get install python3-lttngust
520 LTTng-UST{nbsp}{revision} is available in Arch Linux's _community_
521 repository, while LTTng-tools{nbsp}{revision} and
522 LTTng-modules{nbsp}{revision} are available in the
523 https://aur.archlinux.org/[AUR].
525 To install LTTng{nbsp}{revision} on Arch Linux, using
526 https://github.com/Jguer/yay[yay] for the AUR packages:
528 . Install the main LTTng{nbsp}{revision} packages:
533 # pacman -Sy lttng-ust
534 $ yay -Sy lttng-tools
535 $ yay -Sy lttng-modules
539 . **If you need to instrument and trace <<python-application,Python
540 applications>>**, install the LTTng-UST Python agent:
545 # pacman -Sy python-lttngust
546 # pacman -Sy python2-lttngust
554 To install LTTng-tools{nbsp}{revision} and LTTng-UST{nbsp}{revision} on
555 Alpine Linux{nbsp}3.12:
557 . Add the LTTng packages:
562 # apk add lttng-tools
563 # apk add lttng-ust-dev
567 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
573 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.12.tar.bz2 &&
574 tar -xf lttng-modules-latest-2.12.tar.bz2 &&
575 cd lttng-modules-2.12.* &&
577 sudo make modules_install &&
583 [[building-from-source]]
584 === Build from source
586 To build and install LTTng{nbsp}{revision} from source:
588 . Using your distribution's package manager, or from source, install
589 the following dependencies of LTTng-tools and LTTng-UST:
592 * https://sourceforge.net/projects/libuuid/[libuuid]
593 * http://directory.fsf.org/wiki/Popt[popt]
594 * http://liburcu.org/[Userspace RCU]
595 * http://www.xmlsoft.org/[libxml2]
596 * **Optional**: https://github.com/numactl/numactl[numactl]
599 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
605 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.12.tar.bz2 &&
606 tar -xf lttng-modules-latest-2.12.tar.bz2 &&
607 cd lttng-modules-2.12.* &&
609 sudo make modules_install &&
614 . Download, build, and install the latest LTTng-UST{nbsp}{revision}:
620 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.12.tar.bz2 &&
621 tar -xf lttng-ust-latest-2.12.tar.bz2 &&
622 cd lttng-ust-2.12.* &&
630 Add `--disable-numa` to `./configure` if you don't have
631 https://github.com/numactl/numactl[numactl].
635 .Java and Python application tracing
637 If you need to instrument and trace <<java-application,Java
638 applications>>, pass the `--enable-java-agent-jul`,
639 `--enable-java-agent-log4j`, or `--enable-java-agent-all` options to the
640 `configure` script, depending on which Java logging framework you use.
642 If you need to instrument and trace <<python-application,Python
643 applications>>, pass the `--enable-python-agent` option to the
644 `configure` script. You can set the `PYTHON` environment variable to the
645 path to the Python interpreter for which to install the LTTng-UST Python
653 By default, LTTng-UST libraries are installed to
654 dir:{/usr/local/lib}, which is the de facto directory in which to
655 keep self-compiled and third-party libraries.
657 When <<building-tracepoint-providers-and-user-application,linking an
658 instrumented user application with `liblttng-ust`>>:
660 * Append `/usr/local/lib` to the env:LD_LIBRARY_PATH environment
662 * Pass the `-L/usr/local/lib` and `-Wl,-rpath,/usr/local/lib` options to
663 man:gcc(1), man:g++(1), or man:clang(1).
667 . Download, build, and install the latest LTTng-tools{nbsp}{revision}:
673 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.12.tar.bz2 &&
674 tar -xf lttng-tools-latest-2.12.tar.bz2 &&
675 cd lttng-tools-2.12.* &&
683 TIP: The https://github.com/eepp/vlttng[vlttng tool] can do all the
684 previous steps automatically for a given version of LTTng and confine
685 the installed files in a specific directory. This can be useful to test
686 LTTng without installing it on your system.
692 This is a short guide to get started quickly with LTTng kernel and user
695 Before you follow this guide, make sure to <<installing-lttng,install>>
698 This tutorial walks you through the steps to:
700 . <<tracing-the-linux-kernel,Trace the Linux kernel>>.
701 . <<tracing-your-own-user-application,Trace a user application>> written
703 . <<viewing-and-analyzing-your-traces,View and analyze the
707 [[tracing-the-linux-kernel]]
708 === Trace the Linux kernel
710 The following command lines start with the `#` prompt because you need
711 root privileges to trace the Linux kernel. You can also trace the kernel
712 as a regular user if your Unix user is a member of the
713 <<tracing-group,tracing group>>.
715 . Create a <<tracing-session,tracing session>> which writes its traces
716 to dir:{/tmp/my-kernel-trace}:
721 # lttng create my-kernel-session --output=/tmp/my-kernel-trace
725 . List the available kernel tracepoints and system calls:
730 # lttng list --kernel
731 # lttng list --kernel --syscall
735 . Create <<event,event rules>> which match the desired instrumentation
736 point names, for example the `sched_switch` and `sched_process_fork`
737 tracepoints, and the man:open(2) and man:close(2) system calls:
742 # lttng enable-event --kernel sched_switch,sched_process_fork
743 # lttng enable-event --kernel --syscall open,close
747 Create an event rule which matches _all_ the Linux kernel
748 tracepoints with the opt:lttng-enable-event(1):--all option
749 (this will generate a lot of data when tracing):
754 # lttng enable-event --kernel --all
758 . <<basic-tracing-session-control,Start tracing>>:
767 . Do some operation on your system for a few seconds. For example,
768 load a website, or list the files of a directory.
769 . <<creating-destroying-tracing-sessions,Destroy>> the current
779 The man:lttng-destroy(1) command doesn't destroy the trace data; it
780 only destroys the state of the tracing session.
782 The man:lttng-destroy(1) command also runs the man:lttng-stop(1) command
783 implicitly (see <<basic-tracing-session-control,Start and stop a tracing
784 session>>). You need to stop tracing to make LTTng flush the remaining
785 trace data and make the trace readable.
787 . For the sake of this example, make the recorded trace accessible to
793 # chown -R $(whoami) /tmp/my-kernel-trace
797 See <<viewing-and-analyzing-your-traces,View and analyze the
798 recorded events>> to view the recorded events.
801 [[tracing-your-own-user-application]]
802 === Trace a user application
804 This section steps you through a simple example to trace a
805 _Hello world_ program written in C.
807 To create the traceable user application:
809 . Create the tracepoint provider header file, which defines the
810 tracepoints and the events they can generate:
816 #undef TRACEPOINT_PROVIDER
817 #define TRACEPOINT_PROVIDER hello_world
819 #undef TRACEPOINT_INCLUDE
820 #define TRACEPOINT_INCLUDE "./hello-tp.h"
822 #if !defined(_HELLO_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
825 #include <lttng/tracepoint.h>
835 ctf_string(my_string_field, my_string_arg)
836 ctf_integer(int, my_integer_field, my_integer_arg)
840 #endif /* _HELLO_TP_H */
842 #include <lttng/tracepoint-event.h>
846 . Create the tracepoint provider package source file:
852 #define TRACEPOINT_CREATE_PROBES
853 #define TRACEPOINT_DEFINE
855 #include "hello-tp.h"
859 . Build the tracepoint provider package:
864 $ gcc -c -I. hello-tp.c
868 . Create the _Hello World_ application source file:
875 #include "hello-tp.h"
877 int main(int argc, char *argv[])
881 puts("Hello, World!\nPress Enter to continue...");
884 * The following getchar() call is only placed here for the purpose
885 * of this demonstration, to pause the application in order for
886 * you to have time to list its tracepoints. It's not needed
892 * A tracepoint() call.
894 * Arguments, as defined in hello-tp.h:
896 * 1. Tracepoint provider name (required)
897 * 2. Tracepoint name (required)
898 * 3. my_integer_arg (first user-defined argument)
899 * 4. my_string_arg (second user-defined argument)
901 * Notice the tracepoint provider and tracepoint names are
902 * NOT strings: they are in fact parts of variables that the
903 * macros in hello-tp.h create.
905 tracepoint(hello_world, my_first_tracepoint, 23, "hi there!");
907 for (x = 0; x < argc; ++x) {
908 tracepoint(hello_world, my_first_tracepoint, x, argv[x]);
911 puts("Quitting now!");
912 tracepoint(hello_world, my_first_tracepoint, x * x, "x^2");
919 . Build the application:
928 . Link the application with the tracepoint provider package,
929 `liblttng-ust`, and `libdl`:
934 $ gcc -o hello hello.o hello-tp.o -llttng-ust -ldl
938 Here's the whole build process:
941 .User space tracing tutorial's build steps.
942 image::ust-flow.png[]
944 To trace the user application:
946 . Run the application with a few arguments:
951 $ ./hello world and beyond
960 Press Enter to continue...
964 . Start an LTTng <<lttng-sessiond,session daemon>>:
969 $ lttng-sessiond --daemonize
973 Note that a session daemon might already be running, for example as
974 a service that the distribution's service manager started.
976 . List the available user space tracepoints:
981 $ lttng list --userspace
985 You see the `hello_world:my_first_tracepoint` tracepoint listed
986 under the `./hello` process.
988 . Create a <<tracing-session,tracing session>>:
993 $ lttng create my-user-space-session
997 . Create an <<event,event rule>> which matches the
998 `hello_world:my_first_tracepoint` event name:
1003 $ lttng enable-event --userspace hello_world:my_first_tracepoint
1007 . <<basic-tracing-session-control,Start tracing>>:
1016 . Go back to the running `hello` application and press Enter. The
1017 program executes all `tracepoint()` instrumentation points and exits.
1018 . <<creating-destroying-tracing-sessions,Destroy>> the current
1028 The man:lttng-destroy(1) command doesn't destroy the trace data; it
1029 only destroys the state of the tracing session.
1031 The man:lttng-destroy(1) command also runs the man:lttng-stop(1) command
1032 implicitly (see <<basic-tracing-session-control,Start and stop a tracing
1033 session>>). You need to stop tracing to make LTTng flush the remaining
1034 trace data and make the trace readable.
1036 By default, LTTng saves the traces in
1037 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
1038 where +__name__+ is the tracing session name. The
1039 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
1041 See <<viewing-and-analyzing-your-traces,View and analyze the
1042 recorded events>> to view the recorded events.
1045 [[viewing-and-analyzing-your-traces]]
1046 === View and analyze the recorded events
1048 Once you have completed the <<tracing-the-linux-kernel,Trace the Linux
1049 kernel>> and <<tracing-your-own-user-application,Trace a user
1050 application>> tutorials, you can inspect the recorded events.
1052 There are many tools you can use to read LTTng traces:
1054 https://babeltrace.org/[Babeltrace{nbsp}2]::
1055 A rich, flexible trace manipulation toolkit which includes
1056 a versatile command-line interface
1057 (https://babeltrace.org/docs/v2.0/man1/babeltrace2.1/[cmd:babeltrace2]),
1058 a https://babeltrace.org/docs/v2.0/libbabeltrace2/[C library],
1059 and https://babeltrace.org/docs/v2.0/python/bt2/[Python{nbsp}3 bindings]
1060 so that you can easily process or convert an LTTng trace with
1063 The Babeltrace{nbsp}2 project ships with a
1064 https://babeltrace.org/docs/v2.0/man7/babeltrace2-plugin-ctf.7/[plugin]
1065 which supports the format of the traces which LTTng produces,
1066 https://diamon.org/ctf/[CTF].
1068 http://tracecompass.org/[Trace Compass]::
1069 A graphical user interface for viewing and analyzing any type of
1070 logs or traces, including LTTng's.
1072 https://github.com/lttng/lttng-analyses[**LTTng analyses**]::
1073 An experimental project which includes many high-level analyses of
1074 LTTng kernel traces, like scheduling statistics, interrupt
1075 frequency distribution, top CPU usage, and more.
1077 NOTE: This section assumes that LTTng saved the traces it recorded
1078 during the previous tutorials to their default location, in the
1079 dir:{$LTTNG_HOME/lttng-traces} directory. The env:LTTNG_HOME
1080 environment variable defaults to `$HOME` if not set.
1083 [[viewing-and-analyzing-your-traces-bt]]
1084 ==== Use the cmd:babeltrace2 command-line tool
1086 The simplest way to list all the recorded events of an LTTng trace is to
1088 https://babeltrace.org/docs/v2.0/man1/babeltrace2.1/[cmd:babeltrace2]
1093 $ babeltrace2 ~/lttng-traces/my-user-space-session*
1096 cmd:babeltrace2 finds all traces recursively within the given path and
1097 prints all their events, sorting them chronologically.
1099 Pipe the output of cmd:babeltrace2 into a tool like man:grep(1) for
1104 $ babeltrace2 /tmp/my-kernel-trace | grep _switch
1107 Pipe the output of cmd:babeltrace2 into a tool like man:wc(1) to count
1108 the recorded events:
1112 $ babeltrace2 /tmp/my-kernel-trace | grep _open | wc --lines
1116 [[viewing-and-analyzing-your-traces-bt-python]]
1117 ==== Use the Babeltrace{nbsp}2 Python bindings
1119 The <<viewing-and-analyzing-your-traces-bt,text output of
1120 cmd:babeltrace2>> is useful to isolate events by simple matching using
1121 man:grep(1) and similar utilities. However, more elaborate filters, such
1122 as keeping only event records with a field value falling within a
1123 specific range, are not trivial to write using a shell. Moreover,
1124 reductions and even the most basic computations involving multiple event
1125 records are virtually impossible to implement.
1127 Fortunately, Babeltrace{nbsp}2 ships with
1128 https://babeltrace.org/docs/v2.0/python/bt2/[Python{nbsp}3 bindings]
1129 which make it easy to read the event records of an LTTng trace
1130 sequentially and compute the desired information.
1132 The following script accepts an LTTng Linux kernel trace path as its
1133 first argument and prints the short names of the top five running
1134 processes on CPU{nbsp}0 during the whole trace:
1144 # Get the trace path from the first command-line argument.
1145 it = bt2.TraceCollectionMessageIterator(sys.argv[1])
1147 # This counter dictionary will hold execution times:
1149 # Task command name -> Total execution time (ns)
1150 exec_times = collections.Counter()
1152 # This holds the last `sched_switch` timestamp.
1156 # We only care about event messages.
1157 if type(msg) is not bt2._EventMessageConst:
1160 # Event message's event.
1163 # Keep only `sched_switch` events.
1164 if event.cls.name != 'sched_switch':
1167 # Keep only events which occurred on CPU 0.
1168 if event.packet.context_field['cpu_id'] != 0:
1171 # Event timestamp (ns).
1172 cur_ts = msg.default_clock_snapshot.ns_from_origin
1178 # Previous task command's (short) name.
1179 prev_comm = str(event.payload_field['prev_comm'])
1181 # Initialize an entry in our dictionary if not yet done.
1182 if prev_comm not in exec_times:
1183 exec_times[prev_comm] = 0
1185 # Compute previous command execution time.
1186 diff = cur_ts - last_ts
1188 # Update execution time of this command.
1189 exec_times[prev_comm] += diff
1191 # Update last timestamp.
1195 for name, ns in exec_times.most_common(5):
1196 print('{:20}{} s'.format(name, ns / 1e9))
1199 if __name__ == '__main__':
1207 $ python3 top5proc.py /tmp/my-kernel-trace/kernel
1213 swapper/0 48.607245889 s
1214 chromium 7.192738188 s
1215 pavucontrol 0.709894415 s
1216 Compositor 0.660867933 s
1217 Xorg.bin 0.616753786 s
1220 Note that `swapper/0` is the ``idle'' process of CPU{nbsp}0 on Linux;
1221 since we weren't using the CPU that much when tracing, its first
1222 position in the list makes sense.
1226 == [[understanding-lttng]]Core concepts
1228 From a user's perspective, the LTTng system is built on a few concepts,
1229 or objects, on which the <<lttng-cli,cmd:lttng command-line tool>>
1230 operates by sending commands to the <<lttng-sessiond,session daemon>>.
1231 Understanding how those objects relate to eachother is key in mastering
1234 The core concepts are:
1236 * <<tracing-session,Tracing session>>
1237 * <<domain,Tracing domain>>
1238 * <<channel,Channel and ring buffer>>
1239 * <<"event","Instrumentation point, event rule, event, and event record">>
1245 A _tracing session_ is a stateful dialogue between you and
1246 a <<lttng-sessiond,session daemon>>. You can
1247 <<creating-destroying-tracing-sessions,create a new tracing
1248 session>> with the `lttng create` command.
1250 Most of what you do when you control LTTng tracers happens within a
1251 tracing session. In particular, a tracing session:
1254 * Has its own set of trace files.
1255 * Has its own state of activity (started or stopped).
1256 * Has its own <<tracing-session-mode,mode>> (local, network streaming,
1258 * Has its own <<channel,channels>> to which are associated their own
1259 <<event,event rules>>.
1260 * Has its own <<pid-tracking,process attribute tracking>> inclusion
1264 .A _tracing session_ contains <<channel,channels>> that are members of <<domain,tracing domains>> and contain <<event,event rules>>.
1265 image::concepts.png[]
1267 Those attributes and objects are completely isolated between different
1270 A tracing session is analogous to a cash machine session:
1271 the operations you do on the banking system through the cash machine do
1272 not alter the data of other users of the same system. In the case of
1273 the cash machine, a session lasts as long as your bank card is inside.
1274 In the case of LTTng, a tracing session lasts from the `lttng create`
1275 command to the `lttng destroy` command.
1278 .Each Unix user has its own set of tracing sessions.
1279 image::many-sessions.png[]
1282 [[tracing-session-mode]]
1283 ==== Tracing session mode
1285 LTTng can send the generated trace data to different locations. The
1286 _tracing session mode_ dictates where to send it. The following modes
1287 are available in LTTng{nbsp}{revision}:
1290 LTTng writes the traces to the file system of the machine it traces
1293 Network streaming mode::
1294 LTTng sends the traces over the network to a
1295 <<lttng-relayd,relay daemon>> running on a remote system.
1298 LTTng doesn't write the traces by default.
1300 Instead, you can request LTTng to <<taking-a-snapshot,take a snapshot>>,
1301 that is, a copy of the tracing session's current sub-buffers, and to
1302 write it to the target's file system or to send it over the network to a
1303 <<lttng-relayd,relay daemon>> running on a remote system.
1306 This mode is similar to the network streaming mode, but a live
1307 trace viewer can connect to the distant relay daemon to
1308 <<lttng-live,view event records as LTTng generates them>>.
1314 A _tracing domain_ is a namespace for event sources. A tracing domain
1315 has its own properties and features.
1317 There are currently five available tracing domains:
1321 * `java.util.logging` (JUL)
1325 You must specify a tracing domain when using some commands to avoid
1326 ambiguity. For example, since all the domains support named tracepoints
1327 as event sources (instrumentation points that you manually insert in the
1328 source code), you need to specify a tracing domain when
1329 <<enabling-disabling-events,creating an event rule>> because all the
1330 tracing domains could have tracepoints with the same names.
1332 You can create <<channel,channels>> in the Linux kernel and user space
1333 tracing domains. The other tracing domains have a single default
1338 === Channel and ring buffer
1340 A _channel_ is an object which is responsible for a set of ring buffers.
1341 Each ring buffer is divided into multiple sub-buffers. When an LTTng
1342 tracer emits an event, it can record it to one or more
1343 sub-buffers. The attributes of a channel determine what to do when
1344 there's no space left for a new event record because all sub-buffers
1345 are full, where to send a full sub-buffer, and other behaviours.
1347 A channel is always associated to a <<domain,tracing domain>>. The
1348 `java.util.logging` (JUL), log4j, and Python tracing domains each have
1349 a default channel which you can't configure.
1351 A channel also owns <<event,event rules>>. When an LTTng tracer emits
1352 an event, it records it to the sub-buffers of all
1353 the enabled channels with a satisfied event rule, as long as those
1354 channels are part of active <<tracing-session,tracing sessions>>.
1357 [[channel-buffering-schemes]]
1358 ==== Per-user vs. per-process buffering schemes
1360 A channel has at least one ring buffer _per CPU_. LTTng always
1361 records an event to the ring buffer associated to the CPU on which it
1364 Two _buffering schemes_ are available when you
1365 <<enabling-disabling-channels,create a channel>> in the
1366 user space <<domain,tracing domain>>:
1368 Per-user buffering::
1369 Allocate one set of ring buffers--one per CPU--shared by all the
1370 instrumented processes of each Unix user.
1374 .Per-user buffering scheme.
1375 image::per-user-buffering.png[]
1378 Per-process buffering::
1379 Allocate one set of ring buffers--one per CPU--for each
1380 instrumented process.
1384 .Per-process buffering scheme.
1385 image::per-process-buffering.png[]
1388 The per-process buffering scheme tends to consume more memory than the
1389 per-user option because systems generally have more instrumented
1390 processes than Unix users running instrumented processes. However, the
1391 per-process buffering scheme ensures that one process having a high
1392 event throughput won't fill all the shared sub-buffers of the same
1395 The Linux kernel tracing domain has only one available buffering scheme
1396 which is to allocate a single set of ring buffers for the whole system.
1397 This scheme is similar to the per-user option, but with a single, global
1398 user ``running'' the kernel.
1401 [[channel-overwrite-mode-vs-discard-mode]]
1402 ==== Overwrite vs. discard event record loss modes
1404 When an event occurs, LTTng records it to a specific sub-buffer (yellow
1405 arc in the following animations) of a specific channel's ring buffer.
1406 When there's no space left in a sub-buffer, the tracer marks it as
1407 consumable (red) and another, empty sub-buffer starts receiving the
1408 following event records. A <<lttng-consumerd,consumer daemon>>
1409 eventually consumes the marked sub-buffer (returns to white).
1412 [role="docsvg-channel-subbuf-anim"]
1417 In an ideal world, sub-buffers are consumed faster than they are filled,
1418 as it is the case in the previous animation. In the real world,
1419 however, all sub-buffers can be full at some point, leaving no space to
1420 record the following events.
1422 By default, LTTng-modules and LTTng-UST are _non-blocking_ tracers: when
1423 no empty sub-buffer is available, it is acceptable to lose event records
1424 when the alternative would be to cause substantial delays in the
1425 instrumented application's execution. LTTng privileges performance over
1426 integrity; it aims at perturbing the target system as little as possible
1427 in order to make tracing of subtle race conditions and rare interrupt
1430 Since LTTng{nbsp}2.10, the LTTng user space tracer, LTTng-UST, supports
1431 a _blocking mode_. See the <<blocking-timeout-example,blocking timeout
1432 example>> to learn how to use the blocking mode.
1434 When it comes to losing event records because no empty sub-buffer is
1435 available, or because the <<opt-blocking-timeout,blocking timeout>> is
1436 reached, the channel's _event record loss mode_ determines what to do.
1437 The available event record loss modes are:
1440 Drop the newest event records until the tracer releases a sub-buffer.
1442 This is the only available mode when you specify a
1443 <<opt-blocking-timeout,blocking timeout>>.
1446 Clear the sub-buffer containing the oldest event records and start
1447 writing the newest event records there.
1449 This mode is sometimes called _flight recorder mode_ because it's
1451 https://en.wikipedia.org/wiki/Flight_recorder[flight recorder]:
1452 always keep a fixed amount of the latest data.
1454 Which mechanism you should choose depends on your context: prioritize
1455 the newest or the oldest event records in the ring buffer?
1457 Beware that, in overwrite mode, the tracer abandons a _whole sub-buffer_
1458 as soon as a there's no space left for a new event record, whereas in
1459 discard mode, the tracer only discards the event record that doesn't
1462 In discard mode, LTTng increments a count of lost event records when an
1463 event record is lost and saves this count to the trace. In overwrite
1464 mode, since LTTng{nbsp}2.8, LTTng increments a count of lost sub-buffers
1465 when a sub-buffer is lost and saves this count to the trace. In this
1466 mode, LTTng doesn't write to the trace the exact number of lost event
1467 records in those lost sub-buffers. Trace analyses can use the trace's
1468 saved discarded event record and sub-buffer counts to decide whether or
1469 not to perform the analyses even if trace data is known to be missing.
1471 There are a few ways to decrease your probability of losing event
1473 <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>> shows
1474 how to fine-tune the sub-buffer count and size of a channel to virtually
1475 stop losing event records, though at the cost of greater memory usage.
1478 [[channel-subbuf-size-vs-subbuf-count]]
1479 ==== Sub-buffer count and size
1481 When you <<enabling-disabling-channels,create a channel>>, you can
1482 set its number of sub-buffers and their size.
1484 Note that there is noticeable CPU overhead introduced when
1485 switching sub-buffers (marking a full one as consumable and switching
1486 to an empty one for the following events to be recorded). Knowing this,
1487 the following list presents a few practical situations along with how
1488 to configure the sub-buffer count and size for them:
1490 * **High event throughput**: In general, prefer bigger sub-buffers to
1491 lower the risk of losing event records.
1493 Having bigger sub-buffers also ensures a lower
1494 <<channel-switch-timer,sub-buffer switching frequency>>.
1496 The number of sub-buffers is only meaningful if you create the channel
1497 in overwrite mode: in this case, if a sub-buffer overwrite happens, the
1498 other sub-buffers are left unaltered.
1500 * **Low event throughput**: In general, prefer smaller sub-buffers
1501 since the risk of losing event records is low.
1503 Because events occur less frequently, the sub-buffer switching frequency
1504 should remain low and thus the tracer's overhead shouldn't be a
1507 * **Low memory system**: If your target system has a low memory
1508 limit, prefer fewer first, then smaller sub-buffers.
1510 Even if the system is limited in memory, you want to keep the
1511 sub-buffers as big as possible to avoid a high sub-buffer switching
1514 Note that LTTng uses http://diamon.org/ctf/[CTF] as its trace format,
1515 which means event data is very compact. For example, the average
1516 LTTng kernel event record weights about 32{nbsp}bytes. Thus, a
1517 sub-buffer size of 1{nbsp}MiB is considered big.
1519 The previous situations highlight the major trade-off between a few big
1520 sub-buffers and more, smaller sub-buffers: sub-buffer switching
1521 frequency vs. how much data is lost in overwrite mode. Assuming a
1522 constant event throughput and using the overwrite mode, the two
1523 following configurations have the same ring buffer total size:
1526 [role="docsvg-channel-subbuf-size-vs-count-anim"]
1531 * **Two sub-buffers of 4{nbsp}MiB each**: Expect a very low sub-buffer
1532 switching frequency, but if a sub-buffer overwrite happens, half of
1533 the event records so far (4{nbsp}MiB) are definitely lost.
1534 * **Eight sub-buffers of 1{nbsp}MiB each**: Expect four times the tracer's
1535 overhead as the previous configuration, but if a sub-buffer
1536 overwrite happens, only the eighth of event records so far are
1539 In discard mode, the sub-buffers count parameter is pointless: use two
1540 sub-buffers and set their size according to the requirements of your
1544 [[channel-switch-timer]]
1545 ==== Switch timer period
1547 The _switch timer period_ is an important configurable attribute of
1548 a channel to ensure periodic sub-buffer flushing.
1550 When the _switch timer_ expires, a sub-buffer switch happens. Set
1551 the switch timer period attribute when you
1552 <<enabling-disabling-channels,create a channel>> to ensure that LTTng
1553 consumes and commits trace data to trace files or to a distant relay
1554 daemon periodically in case of a low event throughput.
1557 [role="docsvg-channel-switch-timer"]
1562 This attribute is also convenient when you use big sub-buffers to cope
1563 with a sporadic high event throughput, even if the throughput is
1567 [[channel-read-timer]]
1568 ==== Read timer period
1570 By default, the LTTng tracers use a notification mechanism to signal a
1571 full sub-buffer so that a consumer daemon can consume it. When such
1572 notifications must be avoided, for example in real-time applications,
1573 use the channel's _read timer_ instead. When the read timer fires, the
1574 <<lttng-consumerd,consumer daemon>> checks for full, consumable
1578 [[tracefile-rotation]]
1579 ==== Trace file count and size
1581 By default, trace files can grow as large as needed. Set the
1582 maximum size of each trace file that a channel writes when you
1583 <<enabling-disabling-channels,create a channel>>. When the size of
1584 a trace file reaches the channel's fixed maximum size, LTTng creates
1585 another file to contain the next event records. LTTng appends a file
1586 count to each trace file name in this case.
1588 If you set the trace file size attribute when you create a channel, the
1589 maximum number of trace files that LTTng creates is _unlimited_ by
1590 default. To limit them, set a maximum number of trace files. When the
1591 number of trace files reaches the channel's fixed maximum count, the
1592 oldest trace file is overwritten. This mechanism is called _trace file
1597 Even if you don't limit the trace file count, you can't assume that
1598 LTTng doesn't manage any trace file.
1600 In other words, there is no safe way to know if LTTng still holds a
1601 given trace file open with the trace file rotation feature.
1603 The only way to obtain an unmanaged, self-contained LTTng trace before
1604 you <<creating-destroying-tracing-sessions,destroy>> the tracing session
1605 is with the <<session-rotation,tracing session rotation>> feature
1606 (available since LTTng{nbsp}2.11).
1611 === Instrumentation point, event rule, event, and event record
1613 An _event rule_ is a set of conditions which must be **all** satisfied
1614 for LTTng to record an occuring event.
1616 You set the conditions when you <<enabling-disabling-events,create
1619 You always attach an event rule to a <<channel,channel>> when you create
1622 When an event passes the conditions of an event rule, LTTng records it
1623 in one of the attached channel's sub-buffers.
1625 The available conditions, as of LTTng{nbsp}{revision}, are:
1627 * The event rule _is enabled_.
1628 * The instrumentation point's type _is{nbsp}T_.
1629 * The instrumentation point's name (sometimes called _event name_)
1630 _matches{nbsp}N_, but _isn't{nbsp}E_.
1631 * The instrumentation point's log level _is as severe as{nbsp}L_, or
1632 _is exactly{nbsp}L_.
1633 * The fields of the event's payload _satisfy_ a filter
1634 expression{nbsp}__F__.
1636 As you can see, all the conditions but the dynamic filter are related to
1637 the event rule's status or to the instrumentation point, not to the
1638 occurring events. This is why, without a filter, checking if an event
1639 passes an event rule isn't a dynamic task: when you create or modify an
1640 event rule, all the tracers of its tracing domain enable or disable the
1641 instrumentation points themselves once. This is possible because the
1642 attributes of an instrumentation point (type, name, and log level) are
1643 defined statically. In other words, without a dynamic filter, the tracer
1644 _doesn't evaluate_ the arguments of an instrumentation point unless it
1645 matches an enabled event rule.
1647 Note that, for LTTng to record an event, the <<channel,channel>> to
1648 which a matching event rule is attached must also be enabled, and the
1649 <<tracing-session,tracing session>> owning this channel must be active
1653 .Logical path from an instrumentation point to an event record.
1654 image::event-rule.png[]
1656 .Event, event record, or event rule?
1658 With so many similar terms, it's easy to get confused.
1660 An **event** is the consequence of the execution of an _instrumentation
1661 point_, like a tracepoint that you manually place in some source code,
1662 or a Linux kernel kprobe. An event is said to _occur_ at a specific
1663 time. Different actions can be taken upon the occurrence of an event,
1664 like record the event's payload to a buffer.
1666 An **event record** is the representation of an event in a sub-buffer. A
1667 tracer is responsible for capturing the payload of an event, current
1668 context variables, the event's ID, and the event's timestamp. LTTng
1669 can append this sub-buffer to a trace file.
1671 An **event rule** is a set of conditions which must _all_ be satisfied
1672 for LTTng to record an occuring event. Events still occur without
1673 satisfying event rules, but LTTng doesn't record them.
1678 == Components of noch:{LTTng}
1680 The second _T_ in _LTTng_ stands for _toolkit_: it would be wrong
1681 to call LTTng a simple _tool_ since it is composed of multiple
1682 interacting components. This section describes those components,
1683 explains their respective roles, and shows how they connect together to
1684 form the LTTng ecosystem.
1686 The following diagram shows how the most important components of LTTng
1687 interact with user applications, the Linux kernel, and you:
1690 .Control and trace data paths between LTTng components.
1691 image::plumbing.png[]
1693 The LTTng project incorporates:
1695 * **LTTng-tools**: Libraries and command-line interface to
1696 control tracing sessions.
1697 ** <<lttng-sessiond,Session daemon>> (man:lttng-sessiond(8)).
1698 ** <<lttng-consumerd,Consumer daemon>> (cmd:lttng-consumerd).
1699 ** <<lttng-relayd,Relay daemon>> (man:lttng-relayd(8)).
1700 ** <<liblttng-ctl-lttng,Tracing control library>> (`liblttng-ctl`).
1701 ** <<lttng-cli,Tracing control command-line tool>> (man:lttng(1)).
1702 * **LTTng-UST**: Libraries and Java/Python packages to trace user
1704 ** <<lttng-ust,User space tracing library>> (`liblttng-ust`) and its
1705 headers to instrument and trace any native user application.
1706 ** <<prebuilt-ust-helpers,Preloadable user space tracing helpers>>:
1707 *** `liblttng-ust-libc-wrapper`
1708 *** `liblttng-ust-pthread-wrapper`
1709 *** `liblttng-ust-cyg-profile`
1710 *** `liblttng-ust-cyg-profile-fast`
1711 *** `liblttng-ust-dl`
1712 ** User space tracepoint provider source files generator command-line
1713 tool (man:lttng-gen-tp(1)).
1714 ** <<lttng-ust-agents,LTTng-UST Java agent>> to instrument and trace
1715 Java applications using `java.util.logging` or
1716 Apache log4j{nbsp}1.2 logging.
1717 ** <<lttng-ust-agents,LTTng-UST Python agent>> to instrument
1718 Python applications using the standard `logging` package.
1719 * **LTTng-modules**: <<lttng-modules,Linux kernel modules>> to trace
1721 ** LTTng kernel tracer module.
1722 ** Tracing ring buffer kernel modules.
1723 ** Probe kernel modules.
1724 ** LTTng logger kernel module.
1728 === Tracing control command-line interface
1731 .The tracing control command-line interface.
1732 image::plumbing-lttng-cli.png[]
1734 The _man:lttng(1) command-line tool_ is the standard user interface to
1735 control LTTng <<tracing-session,tracing sessions>>. The cmd:lttng tool
1736 is part of LTTng-tools.
1738 The cmd:lttng tool is linked with
1739 <<liblttng-ctl-lttng,`liblttng-ctl`>> to communicate with
1740 one or more <<lttng-sessiond,session daemons>> behind the scenes.
1742 The cmd:lttng tool has a Git-like interface:
1746 $ lttng <GENERAL OPTIONS> <COMMAND> <COMMAND OPTIONS>
1749 The <<controlling-tracing,Tracing control>> section explores the
1750 available features of LTTng using the cmd:lttng tool.
1753 [[liblttng-ctl-lttng]]
1754 === Tracing control library
1757 .The tracing control library.
1758 image::plumbing-liblttng-ctl.png[]
1760 The _LTTng control library_, `liblttng-ctl`, is used to communicate
1761 with a <<lttng-sessiond,session daemon>> using a C API that hides the
1762 underlying protocol's details. `liblttng-ctl` is part of LTTng-tools.
1764 The <<lttng-cli,cmd:lttng command-line tool>>
1765 is linked with `liblttng-ctl`.
1767 Use `liblttng-ctl` in C or $$C++$$ source code by including its
1772 #include <lttng/lttng.h>
1775 Some objects are referenced by name (C string), such as tracing
1776 sessions, but most of them require to create a handle first using
1777 `lttng_create_handle()`.
1779 As of LTTng{nbsp}{revision}, the best available developer documentation for
1780 `liblttng-ctl` is its installed header files. Every function and structure is
1781 thoroughly documented.
1785 === User space tracing library
1788 .The user space tracing library.
1789 image::plumbing-liblttng-ust.png[]
1791 The _user space tracing library_, `liblttng-ust` (see man:lttng-ust(3)),
1792 is the LTTng user space tracer. It receives commands from a
1793 <<lttng-sessiond,session daemon>>, for example to
1794 enable and disable specific instrumentation points, and writes event
1795 records to ring buffers shared with a
1796 <<lttng-consumerd,consumer daemon>>.
1797 `liblttng-ust` is part of LTTng-UST.
1799 Public C header files are installed beside `liblttng-ust` to
1800 instrument any <<c-application,C or $$C++$$ application>>.
1802 <<lttng-ust-agents,LTTng-UST agents>>, which are regular Java and Python
1803 packages, use their own library providing tracepoints which is
1804 linked with `liblttng-ust`.
1806 An application or library doesn't have to initialize `liblttng-ust`
1807 manually: its constructor does the necessary tasks to properly register
1808 to a session daemon. The initialization phase also enables the
1809 instrumentation points matching the <<event,event rules>> that you
1813 [[lttng-ust-agents]]
1814 === User space tracing agents
1817 .The user space tracing agents.
1818 image::plumbing-lttng-ust-agents.png[]
1820 The _LTTng-UST Java and Python agents_ are regular Java and Python
1821 packages which add LTTng tracing capabilities to the
1822 native logging frameworks. The LTTng-UST agents are part of LTTng-UST.
1824 In the case of Java, the
1825 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[`java.util.logging`
1826 core logging facilities] and
1827 https://logging.apache.org/log4j/1.2/[Apache log4j{nbsp}1.2] are supported.
1828 Note that Apache Log4{nbsp}2 isn't supported.
1830 In the case of Python, the standard
1831 https://docs.python.org/3/library/logging.html[`logging`] package
1832 is supported. Both Python{nbsp}2 and Python{nbsp}3 modules can import the
1833 LTTng-UST Python agent package.
1835 The applications using the LTTng-UST agents are in the
1836 `java.util.logging` (JUL),
1837 log4j, and Python <<domain,tracing domains>>.
1839 Both agents use the same mechanism to trace the log statements. When an
1840 agent initializes, it creates a log handler that attaches to the root
1841 logger. The agent also registers to a <<lttng-sessiond,session daemon>>.
1842 When the application executes a log statement, the root logger passes it
1843 to the agent's log handler. The agent's log handler calls a native
1844 function in a tracepoint provider package shared library linked with
1845 <<lttng-ust,`liblttng-ust`>>, passing the formatted log message and
1846 other fields, like its logger name and its log level. This native
1847 function contains a user space instrumentation point, hence tracing the
1850 The log level condition of an
1851 <<event,event rule>> is considered when tracing
1852 a Java or a Python application, and it's compatible with the standard
1853 JUL, log4j, and Python log levels.
1857 === LTTng kernel modules
1860 .The LTTng kernel modules.
1861 image::plumbing-lttng-modules.png[]
1863 The _LTTng kernel modules_ are a set of Linux kernel modules
1864 which implement the kernel tracer of the LTTng project. The LTTng
1865 kernel modules are part of LTTng-modules.
1867 The LTTng kernel modules include:
1869 * A set of _probe_ modules.
1871 Each module attaches to a specific subsystem
1872 of the Linux kernel using its tracepoint instrument points. There are
1873 also modules to attach to the entry and return points of the Linux
1874 system call functions.
1876 * _Ring buffer_ modules.
1878 A ring buffer implementation is provided as kernel modules. The LTTng
1879 kernel tracer writes to the ring buffer; a
1880 <<lttng-consumerd,consumer daemon>> reads from the ring buffer.
1882 * The _LTTng kernel tracer_ module.
1883 * The _LTTng logger_ module.
1885 The LTTng logger module implements the special path:{/proc/lttng-logger}
1886 (and path:{/dev/lttng-logger} since LTTng{nbsp}2.11) files so that any
1887 executable can generate LTTng events by opening and writing to those
1890 See <<proc-lttng-logger-abi,LTTng logger>>.
1892 Generally, you don't have to load the LTTng kernel modules manually
1893 (using man:modprobe(8), for example): a root <<lttng-sessiond,session
1894 daemon>> loads the necessary modules when starting. If you have extra
1895 probe modules, you can specify to load them to the session daemon on
1898 The LTTng kernel modules are installed in
1899 +/usr/lib/modules/__release__/extra+ by default, where +__release__+ is
1900 the kernel release (see `uname --kernel-release`).
1907 .The session daemon.
1908 image::plumbing-sessiond.png[]
1910 The _session daemon_, man:lttng-sessiond(8), is a daemon responsible for
1911 managing tracing sessions and for controlling the various components of
1912 LTTng. The session daemon is part of LTTng-tools.
1914 The session daemon sends control requests to and receives control
1917 * The <<lttng-ust,user space tracing library>>.
1919 Any instance of the user space tracing library first registers to
1920 a session daemon. Then, the session daemon can send requests to
1921 this instance, such as:
1924 ** Get the list of tracepoints.
1925 ** Share an <<event,event rule>> so that the user space tracing library
1926 can enable or disable tracepoints. Amongst the possible conditions
1927 of an event rule is a filter expression which `liblttng-ust` evalutes
1928 when an event occurs.
1929 ** Share <<channel,channel>> attributes and ring buffer locations.
1932 The session daemon and the user space tracing library use a Unix
1933 domain socket for their communication.
1935 * The <<lttng-ust-agents,user space tracing agents>>.
1937 Any instance of a user space tracing agent first registers to
1938 a session daemon. Then, the session daemon can send requests to
1939 this instance, such as:
1942 ** Get the list of loggers.
1943 ** Enable or disable a specific logger.
1946 The session daemon and the user space tracing agent use a TCP connection
1947 for their communication.
1949 * The <<lttng-modules,LTTng kernel tracer>>.
1950 * The <<lttng-consumerd,consumer daemon>>.
1952 The session daemon sends requests to the consumer daemon to instruct
1953 it where to send the trace data streams, amongst other information.
1955 * The <<lttng-relayd,relay daemon>>.
1957 The session daemon receives commands from the
1958 <<liblttng-ctl-lttng,tracing control library>>.
1960 The root session daemon loads the appropriate
1961 <<lttng-modules,LTTng kernel modules>> on startup. It also spawns
1962 a <<lttng-consumerd,consumer daemon>> as soon as you create
1963 an <<event,event rule>>.
1965 The session daemon doesn't send and receive trace data: this is the
1966 role of the <<lttng-consumerd,consumer daemon>> and
1967 <<lttng-relayd,relay daemon>>. It does, however, generate the
1968 http://diamon.org/ctf/[CTF] metadata stream.
1970 Each Unix user can have its own session daemon instance. The
1971 tracing sessions which different session daemons manage are completely
1974 The root user's session daemon is the only one which is
1975 allowed to control the LTTng kernel tracer, and its spawned consumer
1976 daemon is the only one which is allowed to consume trace data from the
1977 LTTng kernel tracer. Note, however, that any Unix user which is a member
1978 of the <<tracing-group,tracing group>> is allowed
1979 to create <<channel,channels>> in the
1980 Linux kernel <<domain,tracing domain>>, and thus to trace the Linux
1983 The <<lttng-cli,cmd:lttng command-line tool>> automatically starts a
1984 session daemon when using its `create` command if none is currently
1985 running. You can also start the session daemon manually.
1992 .The consumer daemon.
1993 image::plumbing-consumerd.png[]
1995 The _consumer daemon_, cmd:lttng-consumerd, is a daemon which shares
1996 ring buffers with user applications or with the LTTng kernel modules to
1997 collect trace data and send it to some location (on disk or to a
1998 <<lttng-relayd,relay daemon>> over the network). The consumer daemon
1999 is part of LTTng-tools.
2001 You don't start a consumer daemon manually: a consumer daemon is always
2002 spawned by a <<lttng-sessiond,session daemon>> as soon as you create an
2003 <<event,event rule>>, that is, before you start tracing. When you kill
2004 its owner session daemon, the consumer daemon also exits because it is
2005 the session daemon's child process. Command-line options of
2006 man:lttng-sessiond(8) target the consumer daemon process.
2008 There are up to two running consumer daemons per Unix user, whereas only
2009 one session daemon can run per user. This is because each process can be
2010 either 32-bit or 64-bit: if the target system runs a mixture of 32-bit
2011 and 64-bit processes, it is more efficient to have separate
2012 corresponding 32-bit and 64-bit consumer daemons. The root user is an
2013 exception: it can have up to _three_ running consumer daemons: 32-bit
2014 and 64-bit instances for its user applications, and one more
2015 reserved for collecting kernel trace data.
2023 image::plumbing-relayd.png[]
2025 The _relay daemon_, man:lttng-relayd(8), is a daemon acting as a bridge
2026 between remote session and consumer daemons, local trace files, and a
2027 remote live trace viewer. The relay daemon is part of LTTng-tools.
2029 The main purpose of the relay daemon is to implement a receiver of
2030 <<sending-trace-data-over-the-network,trace data over the network>>.
2031 This is useful when the target system doesn't have much file system
2032 space to record trace files locally.
2034 The relay daemon is also a server to which a
2035 <<lttng-live,live trace viewer>> can
2036 connect. The live trace viewer sends requests to the relay daemon to
2037 receive trace data as the target system emits events. The
2038 communication protocol is named _LTTng live_; it is used over TCP
2041 Note that you can start the relay daemon on the target system directly.
2042 This is the setup of choice when the use case is to view events as
2043 the target system emits them without the need of a remote system.
2047 == [[using-lttng]]Instrumentation
2049 There are many examples of tracing and monitoring in our everyday life:
2051 * You have access to real-time and historical weather reports and
2052 forecasts thanks to weather stations installed around the country.
2053 * You know your heart is safe thanks to an electrocardiogram.
2054 * You make sure not to drive your car too fast and to have enough fuel
2055 to reach your destination thanks to gauges visible on your dashboard.
2057 All the previous examples have something in common: they rely on
2058 **instruments**. Without the electrodes attached to the surface of your
2059 body's skin, cardiac monitoring is futile.
2061 LTTng, as a tracer, is no different from those real life examples. If
2062 you're about to trace a software system or, in other words, record its
2063 history of execution, you better have **instrumentation points** in the
2064 subject you're tracing, that is, the actual software.
2066 Various ways were developed to instrument a piece of software for LTTng
2067 tracing. The most straightforward one is to manually place
2068 instrumentation points, called _tracepoints_, in the software's source
2069 code. It is also possible to add instrumentation points dynamically in
2070 the Linux kernel <<domain,tracing domain>>.
2072 If you're only interested in tracing the Linux kernel, your
2073 instrumentation needs are probably already covered by LTTng's built-in
2074 <<lttng-modules,Linux kernel tracepoints>>. You may also wish to trace a
2075 user application which is already instrumented for LTTng tracing.
2076 In such cases, skip this whole section and read the topics of
2077 the <<controlling-tracing,Tracing control>> section.
2079 Many methods are available to instrument a piece of software for LTTng
2082 * <<c-application,User space instrumentation for C and $$C++$$
2084 * <<prebuilt-ust-helpers,Prebuilt user space tracing helpers>>.
2085 * <<java-application,User space Java agent>>.
2086 * <<python-application,User space Python agent>>.
2087 * <<proc-lttng-logger-abi,LTTng logger>>.
2088 * <<instrumenting-linux-kernel,LTTng kernel tracepoints>>.
2092 === [[cxx-application]]User space instrumentation for C and $$C++$$ applications
2094 The procedure to instrument a C or $$C++$$ user application with
2095 the <<lttng-ust,LTTng user space tracing library>>, `liblttng-ust`, is:
2097 . <<tracepoint-provider,Create the source files of a tracepoint provider
2099 . <<probing-the-application-source-code,Add tracepoints to
2100 the application's source code>>.
2101 . <<building-tracepoint-providers-and-user-application,Build and link
2102 a tracepoint provider package and the user application>>.
2104 If you need quick, man:printf(3)-like instrumentation, skip
2105 those steps and use <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>>
2108 IMPORTANT: You need to <<installing-lttng,install>> LTTng-UST to
2109 instrument a user application with `liblttng-ust`.
2112 [[tracepoint-provider]]
2113 ==== Create the source files of a tracepoint provider package
2115 A _tracepoint provider_ is a set of compiled functions which provide
2116 **tracepoints** to an application, the type of instrumentation point
2117 supported by LTTng-UST. Those functions can emit events with
2118 user-defined fields and serialize those events as event records to one
2119 or more LTTng-UST <<channel,channel>> sub-buffers. The `tracepoint()`
2120 macro, which you <<probing-the-application-source-code,insert in a user
2121 application's source code>>, calls those functions.
2123 A _tracepoint provider package_ is an object file (`.o`) or a shared
2124 library (`.so`) which contains one or more tracepoint providers.
2125 Its source files are:
2127 * One or more <<tpp-header,tracepoint provider header>> (`.h`).
2128 * A <<tpp-source,tracepoint provider package source>> (`.c`).
2130 A tracepoint provider package is dynamically linked with `liblttng-ust`,
2131 the LTTng user space tracer, at run time.
2134 .User application linked with `liblttng-ust` and containing a tracepoint provider.
2135 image::ust-app.png[]
2137 NOTE: If you need quick, man:printf(3)-like instrumentation,
2138 skip creating and using a tracepoint provider and use
2139 <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>> instead.
2143 ===== Create a tracepoint provider header file template
2145 A _tracepoint provider header file_ contains the tracepoint
2146 definitions of a tracepoint provider.
2148 To create a tracepoint provider header file:
2150 . Start from this template:
2154 .Tracepoint provider header file template (`.h` file extension).
2156 #undef TRACEPOINT_PROVIDER
2157 #define TRACEPOINT_PROVIDER provider_name
2159 #undef TRACEPOINT_INCLUDE
2160 #define TRACEPOINT_INCLUDE "./tp.h"
2162 #if !defined(_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
2165 #include <lttng/tracepoint.h>
2168 * Use TRACEPOINT_EVENT(), TRACEPOINT_EVENT_CLASS(),
2169 * TRACEPOINT_EVENT_INSTANCE(), and TRACEPOINT_LOGLEVEL() here.
2174 #include <lttng/tracepoint-event.h>
2180 * `provider_name` with the name of your tracepoint provider.
2181 * `"tp.h"` with the name of your tracepoint provider header file.
2183 . Below the `#include <lttng/tracepoint.h>` line, put your
2184 <<defining-tracepoints,tracepoint definitions>>.
2186 Your tracepoint provider name must be unique amongst all the possible
2187 tracepoint provider names used on the same target system. We
2188 suggest to include the name of your project or company in the name,
2189 for example, `org_lttng_my_project_tpp`.
2191 TIP: [[lttng-gen-tp]]Use the man:lttng-gen-tp(1) tool to create
2192 this boilerplate for you. When using cmd:lttng-gen-tp, all you need to
2193 write are the <<defining-tracepoints,tracepoint definitions>>.
2196 [[defining-tracepoints]]
2197 ===== Create a tracepoint definition
2199 A _tracepoint definition_ defines, for a given tracepoint:
2201 * Its **input arguments**. They are the macro parameters that the
2202 `tracepoint()` macro accepts for this particular tracepoint
2203 in the user application's source code.
2204 * Its **output event fields**. They are the sources of event fields
2205 that form the payload of any event that the execution of the
2206 `tracepoint()` macro emits for this particular tracepoint.
2208 Create a tracepoint definition by using the
2209 `TRACEPOINT_EVENT()` macro below the `#include <lttng/tracepoint.h>`
2211 <<tpp-header,tracepoint provider header file template>>.
2213 The syntax of the `TRACEPOINT_EVENT()` macro is:
2216 .`TRACEPOINT_EVENT()` macro syntax.
2219 /* Tracepoint provider name */
2222 /* Tracepoint name */
2225 /* Input arguments */
2230 /* Output event fields */
2239 * `provider_name` with your tracepoint provider name.
2240 * `tracepoint_name` with your tracepoint name.
2241 * `arguments` with the <<tpp-def-input-args,input arguments>>.
2242 * `fields` with the <<tpp-def-output-fields,output event field>>
2245 This tracepoint emits events named `provider_name:tracepoint_name`.
2248 .Event name's length limitation
2250 The concatenation of the tracepoint provider name and the
2251 tracepoint name must not exceed **254{nbsp}characters**. If it does, the
2252 instrumented application compiles and runs, but LTTng throws multiple
2253 warnings and you could experience serious issues.
2256 [[tpp-def-input-args]]The syntax of the `TP_ARGS()` macro is:
2259 .`TP_ARGS()` macro syntax.
2268 * `type` with the C type of the argument.
2269 * `arg_name` with the argument name.
2271 You can repeat `type` and `arg_name` up to 10{nbsp}times to have more
2274 .`TP_ARGS()` usage with three arguments.
2286 The `TP_ARGS()` and `TP_ARGS(void)` forms are valid to create a
2287 tracepoint definition with no input arguments.
2289 [[tpp-def-output-fields]]The `TP_FIELDS()` macro contains a list of
2290 `ctf_*()` macros. Each `ctf_*()` macro defines one event field. See
2291 man:lttng-ust(3) for a complete description of the available `ctf_*()`
2292 macros. A `ctf_*()` macro specifies the type, size, and byte order of
2295 Each `ctf_*()` macro takes an _argument expression_ parameter. This is a
2296 C expression that the tracer evalutes at the `tracepoint()` macro site
2297 in the application's source code. This expression provides a field's
2298 source of data. The argument expression can include input argument names
2299 listed in the `TP_ARGS()` macro.
2301 Each `ctf_*()` macro also takes a _field name_ parameter. Field names
2302 must be unique within a given tracepoint definition.
2304 Here's a complete tracepoint definition example:
2306 .Tracepoint definition.
2308 The following tracepoint definition defines a tracepoint which takes
2309 three input arguments and has four output event fields.
2313 #include "my-custom-structure.h"
2319 const struct my_custom_structure*, my_custom_structure,
2324 ctf_string(query_field, query)
2325 ctf_float(double, ratio_field, ratio)
2326 ctf_integer(int, recv_size, my_custom_structure->recv_size)
2327 ctf_integer(int, send_size, my_custom_structure->send_size)
2332 Refer to this tracepoint definition with the `tracepoint()` macro in
2333 your application's source code like this:
2337 tracepoint(my_provider, my_tracepoint,
2338 my_structure, some_ratio, the_query);
2342 NOTE: The LTTng tracer only evaluates tracepoint arguments at run time
2343 if they satisfy an enabled <<event,event rule>>.
2346 [[using-tracepoint-classes]]
2347 ===== Use a tracepoint class
2349 A _tracepoint class_ is a class of tracepoints which share the same
2350 output event field definitions. A _tracepoint instance_ is one
2351 instance of such a defined tracepoint class, with its own tracepoint
2354 The <<defining-tracepoints,`TRACEPOINT_EVENT()` macro>> is actually a
2355 shorthand which defines both a tracepoint class and a tracepoint
2356 instance at the same time.
2358 When you build a tracepoint provider package, the C or $$C++$$ compiler
2359 creates one serialization function for each **tracepoint class**. A
2360 serialization function is responsible for serializing the event fields
2361 of a tracepoint to a sub-buffer when tracing.
2363 For various performance reasons, when your situation requires multiple
2364 tracepoint definitions with different names, but with the same event
2365 fields, we recommend that you manually create a tracepoint class
2366 and instantiate as many tracepoint instances as needed. One positive
2367 effect of such a design, amongst other advantages, is that all
2368 tracepoint instances of the same tracepoint class reuse the same
2369 serialization function, thus reducing
2370 https://en.wikipedia.org/wiki/Cache_pollution[cache pollution].
2372 .Use a tracepoint class and tracepoint instances.
2374 Consider the following three tracepoint definitions:
2386 ctf_integer(int, userid, userid)
2387 ctf_integer(size_t, len, len)
2399 ctf_integer(int, userid, userid)
2400 ctf_integer(size_t, len, len)
2412 ctf_integer(int, userid, userid)
2413 ctf_integer(size_t, len, len)
2418 In this case, we create three tracepoint classes, with one implicit
2419 tracepoint instance for each of them: `get_account`, `get_settings`, and
2420 `get_transaction`. However, they all share the same event field names
2421 and types. Hence three identical, yet independent serialization
2422 functions are created when you build the tracepoint provider package.
2424 A better design choice is to define a single tracepoint class and three
2425 tracepoint instances:
2429 /* The tracepoint class */
2430 TRACEPOINT_EVENT_CLASS(
2431 /* Tracepoint provider name */
2434 /* Tracepoint class name */
2437 /* Input arguments */
2443 /* Output event fields */
2445 ctf_integer(int, userid, userid)
2446 ctf_integer(size_t, len, len)
2450 /* The tracepoint instances */
2451 TRACEPOINT_EVENT_INSTANCE(
2452 /* Tracepoint provider name */
2455 /* Tracepoint class name */
2458 /* Tracepoint name */
2461 /* Input arguments */
2467 TRACEPOINT_EVENT_INSTANCE(
2476 TRACEPOINT_EVENT_INSTANCE(
2489 [[assigning-log-levels]]
2490 ===== Assign a log level to a tracepoint definition
2492 Assign a _log level_ to a <<defining-tracepoints,tracepoint definition>>
2493 with the `TRACEPOINT_LOGLEVEL()` macro.
2495 Assigning different levels of severity to tracepoint definitions can
2496 be useful: when you <<enabling-disabling-events,create an event rule>>,
2497 you can target tracepoints having a log level as severe as a specific
2500 The concept of LTTng-UST log levels is similar to the levels found
2501 in typical logging frameworks:
2503 * In a logging framework, the log level is given by the function
2504 or method name you use at the log statement site: `debug()`,
2505 `info()`, `warn()`, `error()`, and so on.
2506 * In LTTng-UST, you statically assign the log level to a tracepoint
2507 definition; any `tracepoint()` macro invocation which refers to
2508 this definition has this log level.
2510 You must use `TRACEPOINT_LOGLEVEL()` _after_ the
2511 <<defining-tracepoints,`TRACEPOINT_EVENT()`>> or
2512 <<using-tracepoint-classes,`TRACEPOINT_INSTANCE()`>> macro for a given
2515 The syntax of the `TRACEPOINT_LOGLEVEL()` macro is:
2518 .`TRACEPOINT_LOGLEVEL()` macro syntax.
2520 TRACEPOINT_LOGLEVEL(provider_name, tracepoint_name, log_level)
2525 * `provider_name` with the tracepoint provider name.
2526 * `tracepoint_name` with the tracepoint name.
2527 * `log_level` with the log level to assign to the tracepoint
2528 definition named `tracepoint_name` in the `provider_name`
2529 tracepoint provider.
2531 See man:lttng-ust(3) for a list of available log level names.
2533 .Assign the `TRACE_DEBUG_UNIT` log level to a tracepoint definition.
2537 /* Tracepoint definition */
2546 ctf_integer(int, userid, userid)
2547 ctf_integer(size_t, len, len)
2551 /* Log level assignment */
2552 TRACEPOINT_LOGLEVEL(my_app, get_transaction, TRACE_DEBUG_UNIT)
2558 ===== Create a tracepoint provider package source file
2560 A _tracepoint provider package source file_ is a C source file which
2561 includes a <<tpp-header,tracepoint provider header file>> to expand its
2562 macros into event serialization and other functions.
2564 Use the following tracepoint provider package source file template:
2567 .Tracepoint provider package source file template.
2569 #define TRACEPOINT_CREATE_PROBES
2574 Replace `tp.h` with the name of your <<tpp-header,tracepoint provider
2575 header file>> name. You may also include more than one tracepoint
2576 provider header file here to create a tracepoint provider package
2577 holding more than one tracepoint providers.
2580 [[probing-the-application-source-code]]
2581 ==== Add tracepoints to an application's source code
2583 Once you <<tpp-header,create a tracepoint provider header file>>,
2584 use the `tracepoint()` macro in your application's source code to insert
2585 the tracepoints that this header <<defining-tracepoints,defines>>.
2587 The `tracepoint()` macro takes at least two parameters: the tracepoint
2588 provider name and the tracepoint name. The corresponding tracepoint
2589 definition defines the other parameters.
2591 .`tracepoint()` usage.
2593 The following <<defining-tracepoints,tracepoint definition>> defines a
2594 tracepoint which takes two input arguments and has two output event
2598 .Tracepoint provider header file.
2600 #include "my-custom-structure.h"
2607 const char*, cmd_name
2610 ctf_string(cmd_name, cmd_name)
2611 ctf_integer(int, number_of_args, argc)
2616 Refer to this tracepoint definition with the `tracepoint()` macro in
2617 your application's source code like this:
2620 .Application's source file.
2624 int main(int argc, char* argv[])
2626 tracepoint(my_provider, my_tracepoint, argc, argv[0]);
2632 Note how the application's source code includes
2633 the tracepoint provider header file containing the tracepoint
2634 definitions to use, path:{tp.h}.
2637 .`tracepoint()` usage with a complex tracepoint definition.
2639 Consider this complex tracepoint definition, where multiple event
2640 fields refer to the same input arguments in their argument expression
2644 .Tracepoint provider header file.
2646 /* For `struct stat` */
2647 #include <sys/types.h>
2648 #include <sys/stat.h>
2660 ctf_integer(int, my_constant_field, 23 + 17)
2661 ctf_integer(int, my_int_arg_field, my_int_arg)
2662 ctf_integer(int, my_int_arg_field2, my_int_arg * my_int_arg)
2663 ctf_integer(int, sum4_field, my_str_arg[0] + my_str_arg[1] +
2664 my_str_arg[2] + my_str_arg[3])
2665 ctf_string(my_str_arg_field, my_str_arg)
2666 ctf_integer_hex(off_t, size_field, st->st_size)
2667 ctf_float(double, size_dbl_field, (double) st->st_size)
2668 ctf_sequence_text(char, half_my_str_arg_field, my_str_arg,
2669 size_t, strlen(my_str_arg) / 2)
2674 Refer to this tracepoint definition with the `tracepoint()` macro in
2675 your application's source code like this:
2678 .Application's source file.
2680 #define TRACEPOINT_DEFINE
2687 stat("/etc/fstab", &s);
2688 tracepoint(my_provider, my_tracepoint, 23, "Hello, World!", &s);
2694 If you look at the event record that LTTng writes when tracing this
2695 program, assuming the file size of path:{/etc/fstab} is 301{nbsp}bytes,
2696 it should look like this:
2698 .Event record fields
2700 |Field's name |Field's value
2701 |`my_constant_field` |40
2702 |`my_int_arg_field` |23
2703 |`my_int_arg_field2` |529
2705 |`my_str_arg_field` |`Hello, World!`
2706 |`size_field` |0x12d
2707 |`size_dbl_field` |301.0
2708 |`half_my_str_arg_field` |`Hello,`
2712 Sometimes, the arguments you pass to `tracepoint()` are expensive to
2713 compute--they use the call stack, for example. To avoid this computation
2714 when the tracepoint is disabled, use the `tracepoint_enabled()` and
2715 `do_tracepoint()` macros.
2717 The syntax of the `tracepoint_enabled()` and `do_tracepoint()` macros
2721 .`tracepoint_enabled()` and `do_tracepoint()` macros syntax.
2723 tracepoint_enabled(provider_name, tracepoint_name)
2724 do_tracepoint(provider_name, tracepoint_name, ...)
2729 * `provider_name` with the tracepoint provider name.
2730 * `tracepoint_name` with the tracepoint name.
2732 `tracepoint_enabled()` returns a non-zero value if the tracepoint named
2733 `tracepoint_name` from the provider named `provider_name` is enabled
2736 `do_tracepoint()` is like `tracepoint()`, except that it doesn't check
2737 if the tracepoint is enabled. Using `tracepoint()` with
2738 `tracepoint_enabled()` is dangerous since `tracepoint()` also contains
2739 the `tracepoint_enabled()` check, thus a race condition is
2740 possible in this situation:
2743 .Possible race condition when using `tracepoint_enabled()` with `tracepoint()`.
2745 if (tracepoint_enabled(my_provider, my_tracepoint)) {
2746 stuff = prepare_stuff();
2749 tracepoint(my_provider, my_tracepoint, stuff);
2752 If the tracepoint is enabled after the condition, then `stuff` isn't
2753 prepared: the emitted event will either contain wrong data, or the whole
2754 application could crash (segmentation fault, for example).
2756 NOTE: Neither `tracepoint_enabled()` nor `do_tracepoint()` have an
2757 `STAP_PROBEV()` call. If you need it, you must emit
2761 [[building-tracepoint-providers-and-user-application]]
2762 ==== Build and link a tracepoint provider package and an application
2764 Once you have one or more <<tpp-header,tracepoint provider header
2765 files>> and a <<tpp-source,tracepoint provider package source file>>,
2766 create the tracepoint provider package by compiling its source
2767 file. From here, multiple build and run scenarios are possible. The
2768 following table shows common application and library configurations
2769 along with the required command lines to achieve them.
2771 In the following diagrams, we use the following file names:
2774 Executable application.
2777 Application's object file.
2780 Tracepoint provider package object file.
2783 Tracepoint provider package archive file.
2786 Tracepoint provider package shared object file.
2789 User library object file.
2792 User library shared object file.
2794 We use the following symbols in the diagrams of table below:
2797 .Symbols used in the build scenario diagrams.
2798 image::ust-sit-symbols.png[]
2800 We assume that path:{.} is part of the env:LD_LIBRARY_PATH environment
2801 variable in the following instructions.
2803 [role="growable ust-scenarios",cols="asciidoc,asciidoc"]
2804 .Common tracepoint provider package scenarios.
2806 |Scenario |Instructions
2809 The instrumented application is statically linked with
2810 the tracepoint provider package object.
2812 image::ust-sit+app-linked-with-tp-o+app-instrumented.png[]
2815 include::../common/ust-sit-step-tp-o.txt[]
2817 To build the instrumented application:
2819 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2824 #define TRACEPOINT_DEFINE
2828 . Compile the application source file:
2837 . Build the application:
2842 $ gcc -o app app.o tpp.o -llttng-ust -ldl
2846 To run the instrumented application:
2848 * Start the application:
2858 The instrumented application is statically linked with the
2859 tracepoint provider package archive file.
2861 image::ust-sit+app-linked-with-tp-a+app-instrumented.png[]
2864 To create the tracepoint provider package archive file:
2866 . Compile the <<tpp-source,tracepoint provider package source file>>:
2875 . Create the tracepoint provider package archive file:
2880 $ ar rcs tpp.a tpp.o
2884 To build the instrumented application:
2886 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2891 #define TRACEPOINT_DEFINE
2895 . Compile the application source file:
2904 . Build the application:
2909 $ gcc -o app app.o tpp.a -llttng-ust -ldl
2913 To run the instrumented application:
2915 * Start the application:
2925 The instrumented application is linked with the tracepoint provider
2926 package shared object.
2928 image::ust-sit+app-linked-with-tp-so+app-instrumented.png[]
2931 include::../common/ust-sit-step-tp-so.txt[]
2933 To build the instrumented application:
2935 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2940 #define TRACEPOINT_DEFINE
2944 . Compile the application source file:
2953 . Build the application:
2958 $ gcc -o app app.o -ldl -L. -ltpp
2962 To run the instrumented application:
2964 * Start the application:
2974 The tracepoint provider package shared object is preloaded before the
2975 instrumented application starts.
2977 image::ust-sit+tp-so-preloaded+app-instrumented.png[]
2980 include::../common/ust-sit-step-tp-so.txt[]
2982 To build the instrumented application:
2984 . In path:{app.c}, before including path:{tpp.h}, add the
2990 #define TRACEPOINT_DEFINE
2991 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
2995 . Compile the application source file:
3004 . Build the application:
3009 $ gcc -o app app.o -ldl
3013 To run the instrumented application with tracing support:
3015 * Preload the tracepoint provider package shared object and
3016 start the application:
3021 $ LD_PRELOAD=./libtpp.so ./app
3025 To run the instrumented application without tracing support:
3027 * Start the application:
3037 The instrumented application dynamically loads the tracepoint provider
3038 package shared object.
3040 image::ust-sit+app-dlopens-tp-so+app-instrumented.png[]
3043 include::../common/ust-sit-step-tp-so.txt[]
3045 To build the instrumented application:
3047 . In path:{app.c}, before including path:{tpp.h}, add the
3053 #define TRACEPOINT_DEFINE
3054 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3058 . Compile the application source file:
3067 . Build the application:
3072 $ gcc -o app app.o -ldl
3076 To run the instrumented application:
3078 * Start the application:
3088 The application is linked with the instrumented user library.
3090 The instrumented user library is statically linked with the tracepoint
3091 provider package object file.
3093 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-o+lib-instrumented.png[]
3096 include::../common/ust-sit-step-tp-o-fpic.txt[]
3098 To build the instrumented user library:
3100 . In path:{emon.c}, before including path:{tpp.h}, add the
3106 #define TRACEPOINT_DEFINE
3110 . Compile the user library source file:
3115 $ gcc -I. -fpic -c emon.c
3119 . Build the user library shared object:
3124 $ gcc -shared -o libemon.so emon.o tpp.o -llttng-ust -ldl
3128 To build the application:
3130 . Compile the application source file:
3139 . Build the application:
3144 $ gcc -o app app.o -L. -lemon
3148 To run the application:
3150 * Start the application:
3160 The application is linked with the instrumented user library.
3162 The instrumented user library is linked with the tracepoint provider
3163 package shared object.
3165 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3168 include::../common/ust-sit-step-tp-so.txt[]
3170 To build the instrumented user library:
3172 . In path:{emon.c}, before including path:{tpp.h}, add the
3178 #define TRACEPOINT_DEFINE
3182 . Compile the user library source file:
3187 $ gcc -I. -fpic -c emon.c
3191 . Build the user library shared object:
3196 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3200 To build the application:
3202 . Compile the application source file:
3211 . Build the application:
3216 $ gcc -o app app.o -L. -lemon
3220 To run the application:
3222 * Start the application:
3232 The tracepoint provider package shared object is preloaded before the
3235 The application is linked with the instrumented user library.
3237 image::ust-sit+tp-so-preloaded+app-linked-with-lib+lib-instrumented.png[]
3240 include::../common/ust-sit-step-tp-so.txt[]
3242 To build the instrumented user library:
3244 . In path:{emon.c}, before including path:{tpp.h}, add the
3250 #define TRACEPOINT_DEFINE
3251 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3255 . Compile the user library source file:
3260 $ gcc -I. -fpic -c emon.c
3264 . Build the user library shared object:
3269 $ gcc -shared -o libemon.so emon.o -ldl
3273 To build the application:
3275 . Compile the application source file:
3284 . Build the application:
3289 $ gcc -o app app.o -L. -lemon
3293 To run the application with tracing support:
3295 * Preload the tracepoint provider package shared object and
3296 start the application:
3301 $ LD_PRELOAD=./libtpp.so ./app
3305 To run the application without tracing support:
3307 * Start the application:
3317 The application is linked with the instrumented user library.
3319 The instrumented user library dynamically loads the tracepoint provider
3320 package shared object.
3322 image::ust-sit+app-linked-with-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3325 include::../common/ust-sit-step-tp-so.txt[]
3327 To build the instrumented user library:
3329 . In path:{emon.c}, before including path:{tpp.h}, add the
3335 #define TRACEPOINT_DEFINE
3336 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3340 . Compile the user library source file:
3345 $ gcc -I. -fpic -c emon.c
3349 . Build the user library shared object:
3354 $ gcc -shared -o libemon.so emon.o -ldl
3358 To build the application:
3360 . Compile the application source file:
3369 . Build the application:
3374 $ gcc -o app app.o -L. -lemon
3378 To run the application:
3380 * Start the application:
3390 The application dynamically loads the instrumented user library.
3392 The instrumented user library is linked with the tracepoint provider
3393 package shared object.
3395 image::ust-sit+app-dlopens-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3398 include::../common/ust-sit-step-tp-so.txt[]
3400 To build the instrumented user library:
3402 . In path:{emon.c}, before including path:{tpp.h}, add the
3408 #define TRACEPOINT_DEFINE
3412 . Compile the user library source file:
3417 $ gcc -I. -fpic -c emon.c
3421 . Build the user library shared object:
3426 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3430 To build the application:
3432 . Compile the application source file:
3441 . Build the application:
3446 $ gcc -o app app.o -ldl -L. -lemon
3450 To run the application:
3452 * Start the application:
3462 The application dynamically loads the instrumented user library.
3464 The instrumented user library dynamically loads the tracepoint provider
3465 package shared object.
3467 image::ust-sit+app-dlopens-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3470 include::../common/ust-sit-step-tp-so.txt[]
3472 To build the instrumented user library:
3474 . In path:{emon.c}, before including path:{tpp.h}, add the
3480 #define TRACEPOINT_DEFINE
3481 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3485 . Compile the user library source file:
3490 $ gcc -I. -fpic -c emon.c
3494 . Build the user library shared object:
3499 $ gcc -shared -o libemon.so emon.o -ldl
3503 To build the application:
3505 . Compile the application source file:
3514 . Build the application:
3519 $ gcc -o app app.o -ldl -L. -lemon
3523 To run the application:
3525 * Start the application:
3535 The tracepoint provider package shared object is preloaded before the
3538 The application dynamically loads the instrumented user library.
3540 image::ust-sit+tp-so-preloaded+app-dlopens-lib+lib-instrumented.png[]
3543 include::../common/ust-sit-step-tp-so.txt[]
3545 To build the instrumented user library:
3547 . In path:{emon.c}, before including path:{tpp.h}, add the
3553 #define TRACEPOINT_DEFINE
3554 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3558 . Compile the user library source file:
3563 $ gcc -I. -fpic -c emon.c
3567 . Build the user library shared object:
3572 $ gcc -shared -o libemon.so emon.o -ldl
3576 To build the application:
3578 . Compile the application source file:
3587 . Build the application:
3592 $ gcc -o app app.o -L. -lemon
3596 To run the application with tracing support:
3598 * Preload the tracepoint provider package shared object and
3599 start the application:
3604 $ LD_PRELOAD=./libtpp.so ./app
3608 To run the application without tracing support:
3610 * Start the application:
3620 The application is statically linked with the tracepoint provider
3621 package object file.
3623 The application is linked with the instrumented user library.
3625 image::ust-sit+app-linked-with-tp-o+app-linked-with-lib+lib-instrumented.png[]
3628 include::../common/ust-sit-step-tp-o.txt[]
3630 To build the instrumented user library:
3632 . In path:{emon.c}, before including path:{tpp.h}, add the
3638 #define TRACEPOINT_DEFINE
3642 . Compile the user library source file:
3647 $ gcc -I. -fpic -c emon.c
3651 . Build the user library shared object:
3656 $ gcc -shared -o libemon.so emon.o
3660 To build the application:
3662 . Compile the application source file:
3671 . Build the application:
3676 $ gcc -o app app.o tpp.o -llttng-ust -ldl -L. -lemon
3680 To run the instrumented application:
3682 * Start the application:
3692 The application is statically linked with the tracepoint provider
3693 package object file.
3695 The application dynamically loads the instrumented user library.
3697 image::ust-sit+app-linked-with-tp-o+app-dlopens-lib+lib-instrumented.png[]
3700 include::../common/ust-sit-step-tp-o.txt[]
3702 To build the application:
3704 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3709 #define TRACEPOINT_DEFINE
3713 . Compile the application source file:
3722 . Build the application:
3727 $ gcc -Wl,--export-dynamic -o app app.o tpp.o \
3732 The `--export-dynamic` option passed to the linker is necessary for the
3733 dynamically loaded library to ``see'' the tracepoint symbols defined in
3736 To build the instrumented user library:
3738 . Compile the user library source file:
3743 $ gcc -I. -fpic -c emon.c
3747 . Build the user library shared object:
3752 $ gcc -shared -o libemon.so emon.o
3756 To run the application:
3758 * Start the application:
3769 [[using-lttng-ust-with-daemons]]
3770 ===== Use noch:{LTTng-UST} with daemons
3772 If your instrumented application calls man:fork(2), man:clone(2),
3773 or BSD's man:rfork(2), without a following man:exec(3)-family
3774 system call, you must preload the path:{liblttng-ust-fork.so} shared
3775 object when you start the application.
3779 $ LD_PRELOAD=liblttng-ust-fork.so ./my-app
3782 If your tracepoint provider package is
3783 a shared library which you also preload, you must put both
3784 shared objects in env:LD_PRELOAD:
3788 $ LD_PRELOAD=liblttng-ust-fork.so:/path/to/tp.so ./my-app
3794 ===== Use noch:{LTTng-UST} with applications which close file descriptors that don't belong to them
3796 If your instrumented application closes one or more file descriptors
3797 which it did not open itself, you must preload the
3798 path:{liblttng-ust-fd.so} shared object when you start the application:
3802 $ LD_PRELOAD=liblttng-ust-fd.so ./my-app
3805 Typical use cases include closing all the file descriptors after
3806 man:fork(2) or man:rfork(2) and buggy applications doing
3810 [[lttng-ust-pkg-config]]
3811 ===== Use noch:{pkg-config}
3813 On some distributions, LTTng-UST ships with a
3814 https://www.freedesktop.org/wiki/Software/pkg-config/[pkg-config]
3815 metadata file. If this is your case, then use cmd:pkg-config to
3816 build an application on the command line:
3820 $ gcc -o my-app my-app.o tp.o $(pkg-config --cflags --libs lttng-ust)
3824 [[instrumenting-32-bit-app-on-64-bit-system]]
3825 ===== [[advanced-instrumenting-techniques]]Build a 32-bit instrumented application for a 64-bit target system
3827 In order to trace a 32-bit application running on a 64-bit system,
3828 LTTng must use a dedicated 32-bit
3829 <<lttng-consumerd,consumer daemon>>.
3831 The following steps show how to build and install a 32-bit consumer
3832 daemon, which is _not_ part of the default 64-bit LTTng build, how to
3833 build and install the 32-bit LTTng-UST libraries, and how to build and
3834 link an instrumented 32-bit application in that context.
3836 To build a 32-bit instrumented application for a 64-bit target system,
3837 assuming you have a fresh target system with no installed Userspace RCU
3840 . Download, build, and install a 32-bit version of Userspace RCU:
3845 $ cd $(mktemp -d) &&
3846 wget http://lttng.org/files/urcu/userspace-rcu-latest-0.9.tar.bz2 &&
3847 tar -xf userspace-rcu-latest-0.9.tar.bz2 &&
3848 cd userspace-rcu-0.9.* &&
3849 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 &&
3851 sudo make install &&
3856 . Using your distribution's package manager, or from source, install
3857 the following 32-bit versions of the following dependencies of
3858 LTTng-tools and LTTng-UST:
3861 * https://sourceforge.net/projects/libuuid/[libuuid]
3862 * http://directory.fsf.org/wiki/Popt[popt]
3863 * http://www.xmlsoft.org/[libxml2]
3866 . Download, build, and install a 32-bit version of the latest
3867 LTTng-UST{nbsp}{revision}:
3872 $ cd $(mktemp -d) &&
3873 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.12.tar.bz2 &&
3874 tar -xf lttng-ust-latest-2.12.tar.bz2 &&
3875 cd lttng-ust-2.12.* &&
3876 ./configure --libdir=/usr/local/lib32 \
3877 CFLAGS=-m32 CXXFLAGS=-m32 \
3878 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
3880 sudo make install &&
3887 Depending on your distribution,
3888 32-bit libraries could be installed at a different location than
3889 `/usr/lib32`. For example, Debian is known to install
3890 some 32-bit libraries in `/usr/lib/i386-linux-gnu`.
3892 In this case, make sure to set `LDFLAGS` to all the
3893 relevant 32-bit library paths, for example:
3897 $ LDFLAGS='-L/usr/lib/i386-linux-gnu -L/usr/lib32'
3901 . Download the latest LTTng-tools{nbsp}{revision}, build, and install
3902 the 32-bit consumer daemon:
3907 $ cd $(mktemp -d) &&
3908 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.12.tar.bz2 &&
3909 tar -xf lttng-tools-latest-2.12.tar.bz2 &&
3910 cd lttng-tools-2.12.* &&
3911 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 CXXFLAGS=-m32 \
3912 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' \
3913 --disable-bin-lttng --disable-bin-lttng-crash \
3914 --disable-bin-lttng-relayd --disable-bin-lttng-sessiond &&
3916 cd src/bin/lttng-consumerd &&
3917 sudo make install &&
3922 . From your distribution or from source,
3923 <<installing-lttng,install>> the 64-bit versions of
3924 LTTng-UST and Userspace RCU.
3925 . Download, build, and install the 64-bit version of the
3926 latest LTTng-tools{nbsp}{revision}:
3931 $ cd $(mktemp -d) &&
3932 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.12.tar.bz2 &&
3933 tar -xf lttng-tools-latest-2.12.tar.bz2 &&
3934 cd lttng-tools-2.12.* &&
3935 ./configure --with-consumerd32-libdir=/usr/local/lib32 \
3936 --with-consumerd32-bin=/usr/local/lib32/lttng/libexec/lttng-consumerd &&
3938 sudo make install &&
3943 . Pass the following options to man:gcc(1), man:g++(1), or man:clang(1)
3944 when linking your 32-bit application:
3947 -m32 -L/usr/lib32 -L/usr/local/lib32 \
3948 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32
3951 For example, let's rebuild the quick start example in
3952 <<tracing-your-own-user-application,Trace a user application>> as an
3953 instrumented 32-bit application:
3958 $ gcc -m32 -c -I. hello-tp.c
3959 $ gcc -m32 -c hello.c
3960 $ gcc -m32 -o hello hello.o hello-tp.o \
3961 -L/usr/lib32 -L/usr/local/lib32 \
3962 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32 \
3967 No special action is required to execute the 32-bit application and
3968 to trace it: use the command-line man:lttng(1) tool as usual.
3975 man:tracef(3) is a small LTTng-UST API designed for quick,
3976 man:printf(3)-like instrumentation without the burden of
3977 <<tracepoint-provider,creating>> and
3978 <<building-tracepoint-providers-and-user-application,building>>
3979 a tracepoint provider package.
3981 To use `tracef()` in your application:
3983 . In the C or C++ source files where you need to use `tracef()`,
3984 include `<lttng/tracef.h>`:
3989 #include <lttng/tracef.h>
3993 . In the application's source code, use `tracef()` like you would use
4001 tracef("my message: %d (%s)", my_integer, my_string);
4007 . Link your application with `liblttng-ust`:
4012 $ gcc -o app app.c -llttng-ust
4016 To trace the events that `tracef()` calls emit:
4018 * <<enabling-disabling-events,Create an event rule>> which matches the
4019 `lttng_ust_tracef:*` event name:
4024 $ lttng enable-event --userspace 'lttng_ust_tracef:*'
4029 .Limitations of `tracef()`
4031 The `tracef()` utility function was developed to make user space tracing
4032 super simple, albeit with notable disadvantages compared to
4033 <<defining-tracepoints,user-defined tracepoints>>:
4035 * All the emitted events have the same tracepoint provider and
4036 tracepoint names, respectively `lttng_ust_tracef` and `event`.
4037 * There is no static type checking.
4038 * The only event record field you actually get, named `msg`, is a string
4039 potentially containing the values you passed to `tracef()`
4040 using your own format string. This also means that you can't filter
4041 events with a custom expression at run time because there are no
4043 * Since `tracef()` uses the C standard library's man:vasprintf(3)
4044 function behind the scenes to format the strings at run time, its
4045 expected performance is lower than with user-defined tracepoints,
4046 which don't require a conversion to a string.
4048 Taking this into consideration, `tracef()` is useful for some quick
4049 prototyping and debugging, but you shouldn't consider it for any
4050 permanent and serious applicative instrumentation.
4056 ==== Use `tracelog()`
4058 The man:tracelog(3) API is very similar to <<tracef,`tracef()`>>, with
4059 the difference that it accepts an additional log level parameter.
4061 The goal of `tracelog()` is to ease the migration from logging to
4064 To use `tracelog()` in your application:
4066 . In the C or C++ source files where you need to use `tracelog()`,
4067 include `<lttng/tracelog.h>`:
4072 #include <lttng/tracelog.h>
4076 . In the application's source code, use `tracelog()` like you would use
4077 man:printf(3), except for the first parameter which is the log
4085 tracelog(TRACE_WARNING, "my message: %d (%s)",
4086 my_integer, my_string);
4092 See man:lttng-ust(3) for a list of available log level names.
4094 . Link your application with `liblttng-ust`:
4099 $ gcc -o app app.c -llttng-ust
4103 To trace the events that `tracelog()` calls emit with a log level
4104 _as severe as_ a specific log level:
4106 * <<enabling-disabling-events,Create an event rule>> which matches the
4107 `lttng_ust_tracelog:*` event name and a minimum level
4113 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4114 --loglevel=TRACE_WARNING
4118 To trace the events that `tracelog()` calls emit with a
4119 _specific log level_:
4121 * Create an event rule which matches the `lttng_ust_tracelog:*`
4122 event name and a specific log level:
4127 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
4128 --loglevel-only=TRACE_INFO
4133 [[prebuilt-ust-helpers]]
4134 === Prebuilt user space tracing helpers
4136 The LTTng-UST package provides a few helpers in the form of preloadable
4137 shared objects which automatically instrument system functions and
4140 The helper shared objects are normally found in dir:{/usr/lib}. If you
4141 built LTTng-UST <<building-from-source,from source>>, they are probably
4142 located in dir:{/usr/local/lib}.
4144 The installed user space tracing helpers in LTTng-UST{nbsp}{revision}
4147 path:{liblttng-ust-libc-wrapper.so}::
4148 path:{liblttng-ust-pthread-wrapper.so}::
4149 <<liblttng-ust-libc-pthread-wrapper,C{nbsp}standard library
4150 memory and POSIX threads function tracing>>.
4152 path:{liblttng-ust-cyg-profile.so}::
4153 path:{liblttng-ust-cyg-profile-fast.so}::
4154 <<liblttng-ust-cyg-profile,Function entry and exit tracing>>.
4156 path:{liblttng-ust-dl.so}::
4157 <<liblttng-ust-dl,Dynamic linker tracing>>.
4159 To use a user space tracing helper with any user application:
4161 * Preload the helper shared object when you start the application:
4166 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so my-app
4170 You can preload more than one helper:
4175 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so:liblttng-ust-dl.so my-app
4181 [[liblttng-ust-libc-pthread-wrapper]]
4182 ==== Instrument C standard library memory and POSIX threads functions
4184 The path:{liblttng-ust-libc-wrapper.so} and
4185 path:{liblttng-ust-pthread-wrapper.so} helpers
4186 add instrumentation to some C standard library and POSIX
4190 .Functions instrumented by preloading path:{liblttng-ust-libc-wrapper.so}.
4192 |TP provider name |TP name |Instrumented function
4194 .6+|`lttng_ust_libc` |`malloc` |man:malloc(3)
4195 |`calloc` |man:calloc(3)
4196 |`realloc` |man:realloc(3)
4197 |`free` |man:free(3)
4198 |`memalign` |man:memalign(3)
4199 |`posix_memalign` |man:posix_memalign(3)
4203 .Functions instrumented by preloading path:{liblttng-ust-pthread-wrapper.so}.
4205 |TP provider name |TP name |Instrumented function
4207 .4+|`lttng_ust_pthread` |`pthread_mutex_lock_req` |man:pthread_mutex_lock(3p) (request time)
4208 |`pthread_mutex_lock_acq` |man:pthread_mutex_lock(3p) (acquire time)
4209 |`pthread_mutex_trylock` |man:pthread_mutex_trylock(3p)
4210 |`pthread_mutex_unlock` |man:pthread_mutex_unlock(3p)
4213 When you preload the shared object, it replaces the functions listed
4214 in the previous tables by wrappers which contain tracepoints and call
4215 the replaced functions.
4218 [[liblttng-ust-cyg-profile]]
4219 ==== Instrument function entry and exit
4221 The path:{liblttng-ust-cyg-profile*.so} helpers can add instrumentation
4222 to the entry and exit points of functions.
4224 man:gcc(1) and man:clang(1) have an option named
4225 https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html[`-finstrument-functions`]
4226 which generates instrumentation calls for entry and exit to functions.
4227 The LTTng-UST function tracing helpers,
4228 path:{liblttng-ust-cyg-profile.so} and
4229 path:{liblttng-ust-cyg-profile-fast.so}, take advantage of this feature
4230 to add tracepoints to the two generated functions (which contain
4231 `cyg_profile` in their names, hence the helper's name).
4233 To use the LTTng-UST function tracing helper, the source files to
4234 instrument must be built using the `-finstrument-functions` compiler
4237 There are two versions of the LTTng-UST function tracing helper:
4239 * **path:{liblttng-ust-cyg-profile-fast.so}** is a lightweight variant
4240 that you should only use when it can be _guaranteed_ that the
4241 complete event stream is recorded without any lost event record.
4242 Any kind of duplicate information is left out.
4244 Assuming no event record is lost, having only the function addresses on
4245 entry is enough to create a call graph, since an event record always
4246 contains the ID of the CPU that generated it.
4248 Use a tool like man:addr2line(1) to convert function addresses back to
4249 source file names and line numbers.
4251 * **path:{liblttng-ust-cyg-profile.so}** is a more robust variant
4252 which also works in use cases where event records might get discarded or
4253 not recorded from application startup.
4254 In these cases, the trace analyzer needs more information to be
4255 able to reconstruct the program flow.
4257 See man:lttng-ust-cyg-profile(3) to learn more about the instrumentation
4258 points of this helper.
4260 All the tracepoints that this helper provides have the
4261 log level `TRACE_DEBUG_FUNCTION` (see man:lttng-ust(3)).
4263 TIP: It's sometimes a good idea to limit the number of source files that
4264 you compile with the `-finstrument-functions` option to prevent LTTng
4265 from writing an excessive amount of trace data at run time. When using
4267 `-finstrument-functions-exclude-function-list` option to avoid
4268 instrument entries and exits of specific function names.
4273 ==== Instrument the dynamic linker
4275 The path:{liblttng-ust-dl.so} helper adds instrumentation to the
4276 man:dlopen(3) and man:dlclose(3) function calls.
4278 See man:lttng-ust-dl(3) to learn more about the instrumentation points
4283 [[java-application]]
4284 === User space Java agent
4286 You can instrument any Java application which uses one of the following
4289 * The https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[**`java.util.logging`**]
4290 (JUL) core logging facilities.
4291 * http://logging.apache.org/log4j/1.2/[**Apache log4j{nbsp}1.2**], since
4292 LTTng{nbsp}2.6. Note that Apache Log4j{nbsp}2 isn't supported.
4295 .LTTng-UST Java agent imported by a Java application.
4296 image::java-app.png[]
4298 Note that the methods described below are new in LTTng{nbsp}2.8.
4299 Previous LTTng versions use another technique.
4301 NOTE: We use http://openjdk.java.net/[OpenJDK]{nbsp}8 for development
4302 and https://ci.lttng.org/[continuous integration], thus this version is
4303 directly supported. However, the LTTng-UST Java agent is also tested
4304 with OpenJDK{nbsp}7.
4309 ==== Use the LTTng-UST Java agent for `java.util.logging`
4311 To use the LTTng-UST Java agent in a Java application which uses
4312 `java.util.logging` (JUL):
4314 . In the Java application's source code, import the LTTng-UST
4315 log handler package for `java.util.logging`:
4320 import org.lttng.ust.agent.jul.LttngLogHandler;
4324 . Create an LTTng-UST JUL log handler:
4329 Handler lttngUstLogHandler = new LttngLogHandler();
4333 . Add this handler to the JUL loggers which should emit LTTng events:
4338 Logger myLogger = Logger.getLogger("some-logger");
4340 myLogger.addHandler(lttngUstLogHandler);
4344 . Use `java.util.logging` log statements and configuration as usual.
4345 The loggers with an attached LTTng-UST log handler can emit
4348 . Before exiting the application, remove the LTTng-UST log handler from
4349 the loggers attached to it and call its `close()` method:
4354 myLogger.removeHandler(lttngUstLogHandler);
4355 lttngUstLogHandler.close();
4359 This isn't strictly necessary, but it is recommended for a clean
4360 disposal of the handler's resources.
4362 . Include the LTTng-UST Java agent's common and JUL-specific JAR files,
4363 path:{lttng-ust-agent-common.jar} and path:{lttng-ust-agent-jul.jar},
4365 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4366 path] when you build the Java application.
4368 The JAR files are typically located in dir:{/usr/share/java}.
4370 IMPORTANT: The LTTng-UST Java agent must be
4371 <<installing-lttng,installed>> for the logging framework your
4374 .Use the LTTng-UST Java agent for `java.util.logging`.
4379 import java.io.IOException;
4380 import java.util.logging.Handler;
4381 import java.util.logging.Logger;
4382 import org.lttng.ust.agent.jul.LttngLogHandler;
4386 private static final int answer = 42;
4388 public static void main(String[] argv) throws Exception
4391 Logger logger = Logger.getLogger("jello");
4393 // Create an LTTng-UST log handler
4394 Handler lttngUstLogHandler = new LttngLogHandler();
4396 // Add the LTTng-UST log handler to our logger
4397 logger.addHandler(lttngUstLogHandler);
4400 logger.info("some info");
4401 logger.warning("some warning");
4403 logger.finer("finer information; the answer is " + answer);
4405 logger.severe("error!");
4407 // Not mandatory, but cleaner
4408 logger.removeHandler(lttngUstLogHandler);
4409 lttngUstLogHandler.close();
4418 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4421 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4422 <<enabling-disabling-events,create an event rule>> matching the
4423 `jello` JUL logger, and <<basic-tracing-session-control,start tracing>>:
4428 $ lttng enable-event --jul jello
4432 Run the compiled class:
4436 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4439 <<basic-tracing-session-control,Stop tracing>> and inspect the
4449 In the resulting trace, an <<event,event record>> generated by a Java
4450 application using `java.util.logging` is named `lttng_jul:event` and
4451 has the following fields:
4454 Log record's message.
4460 Name of the class in which the log statement was executed.
4463 Name of the method in which the log statement was executed.
4466 Logging time (timestamp in milliseconds).
4469 Log level integer value.
4472 ID of the thread in which the log statement was executed.
4474 Use the opt:lttng-enable-event(1):--loglevel or
4475 opt:lttng-enable-event(1):--loglevel-only option of the
4476 man:lttng-enable-event(1) command to target a range of JUL log levels
4477 or a specific JUL log level.
4482 ==== Use the LTTng-UST Java agent for Apache log4j
4484 To use the LTTng-UST Java agent in a Java application which uses
4485 Apache log4j{nbsp}1.2:
4487 . In the Java application's source code, import the LTTng-UST
4488 log appender package for Apache log4j:
4493 import org.lttng.ust.agent.log4j.LttngLogAppender;
4497 . Create an LTTng-UST log4j log appender:
4502 Appender lttngUstLogAppender = new LttngLogAppender();
4506 . Add this appender to the log4j loggers which should emit LTTng events:
4511 Logger myLogger = Logger.getLogger("some-logger");
4513 myLogger.addAppender(lttngUstLogAppender);
4517 . Use Apache log4j log statements and configuration as usual. The
4518 loggers with an attached LTTng-UST log appender can emit LTTng events.
4520 . Before exiting the application, remove the LTTng-UST log appender from
4521 the loggers attached to it and call its `close()` method:
4526 myLogger.removeAppender(lttngUstLogAppender);
4527 lttngUstLogAppender.close();
4531 This isn't strictly necessary, but it is recommended for a clean
4532 disposal of the appender's resources.
4534 . Include the LTTng-UST Java agent's common and log4j-specific JAR
4535 files, path:{lttng-ust-agent-common.jar} and
4536 path:{lttng-ust-agent-log4j.jar}, in the
4537 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4538 path] when you build the Java application.
4540 The JAR files are typically located in dir:{/usr/share/java}.
4542 IMPORTANT: The LTTng-UST Java agent must be
4543 <<installing-lttng,installed>> for the logging framework your
4546 .Use the LTTng-UST Java agent for Apache log4j.
4551 import org.apache.log4j.Appender;
4552 import org.apache.log4j.Logger;
4553 import org.lttng.ust.agent.log4j.LttngLogAppender;
4557 private static final int answer = 42;
4559 public static void main(String[] argv) throws Exception
4562 Logger logger = Logger.getLogger("jello");
4564 // Create an LTTng-UST log appender
4565 Appender lttngUstLogAppender = new LttngLogAppender();
4567 // Add the LTTng-UST log appender to our logger
4568 logger.addAppender(lttngUstLogAppender);
4571 logger.info("some info");
4572 logger.warn("some warning");
4574 logger.debug("debug information; the answer is " + answer);
4576 logger.fatal("error!");
4578 // Not mandatory, but cleaner
4579 logger.removeAppender(lttngUstLogAppender);
4580 lttngUstLogAppender.close();
4586 Build this example (`$LOG4JPATH` is the path to the Apache log4j JAR
4591 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH Test.java
4594 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4595 <<enabling-disabling-events,create an event rule>> matching the
4596 `jello` log4j logger, and <<basic-tracing-session-control,start tracing>>:
4601 $ lttng enable-event --log4j jello
4605 Run the compiled class:
4609 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH:. Test
4612 <<basic-tracing-session-control,Stop tracing>> and inspect the
4622 In the resulting trace, an <<event,event record>> generated by a Java
4623 application using log4j is named `lttng_log4j:event` and
4624 has the following fields:
4627 Log record's message.
4633 Name of the class in which the log statement was executed.
4636 Name of the method in which the log statement was executed.
4639 Name of the file in which the executed log statement is located.
4642 Line number at which the log statement was executed.
4648 Log level integer value.
4651 Name of the Java thread in which the log statement was executed.
4653 Use the opt:lttng-enable-event(1):--loglevel or
4654 opt:lttng-enable-event(1):--loglevel-only option of the
4655 man:lttng-enable-event(1) command to target a range of Apache log4j
4656 log levels or a specific log4j log level.
4660 [[java-application-context]]
4661 ==== Provide application-specific context fields in a Java application
4663 A Java application-specific context field is a piece of state provided
4664 by the application which <<adding-context,you can add>>, using the
4665 man:lttng-add-context(1) command, to each <<event,event record>>
4666 produced by the log statements of this application.
4668 For example, a given object might have a current request ID variable.
4669 You can create a context information retriever for this object and
4670 assign a name to this current request ID. You can then, using the
4671 man:lttng-add-context(1) command, add this context field by name to
4672 the JUL or log4j <<channel,channel>>.
4674 To provide application-specific context fields in a Java application:
4676 . In the Java application's source code, import the LTTng-UST
4677 Java agent context classes and interfaces:
4682 import org.lttng.ust.agent.context.ContextInfoManager;
4683 import org.lttng.ust.agent.context.IContextInfoRetriever;
4687 . Create a context information retriever class, that is, a class which
4688 implements the `IContextInfoRetriever` interface:
4693 class MyContextInfoRetriever implements IContextInfoRetriever
4696 public Object retrieveContextInfo(String key)
4698 if (key.equals("intCtx")) {
4700 } else if (key.equals("strContext")) {
4701 return "context value!";
4710 This `retrieveContextInfo()` method is the only member of the
4711 `IContextInfoRetriever` interface. Its role is to return the current
4712 value of a state by name to create a context field. The names of the
4713 context fields and which state variables they return depends on your
4716 All primitive types and objects are supported as context fields.
4717 When `retrieveContextInfo()` returns an object, the context field
4718 serializer calls its `toString()` method to add a string field to
4719 event records. The method can also return `null`, which means that
4720 no context field is available for the required name.
4722 . Register an instance of your context information retriever class to
4723 the context information manager singleton:
4728 IContextInfoRetriever cir = new MyContextInfoRetriever();
4729 ContextInfoManager cim = ContextInfoManager.getInstance();
4730 cim.registerContextInfoRetriever("retrieverName", cir);
4734 . Before exiting the application, remove your context information
4735 retriever from the context information manager singleton:
4740 ContextInfoManager cim = ContextInfoManager.getInstance();
4741 cim.unregisterContextInfoRetriever("retrieverName");
4745 This isn't strictly necessary, but it is recommended for a clean
4746 disposal of some manager's resources.
4748 . Build your Java application with LTTng-UST Java agent support as
4749 usual, following the procedure for either the <<jul,JUL>> or
4750 <<log4j,Apache log4j>> framework.
4753 .Provide application-specific context fields in a Java application.
4758 import java.util.logging.Handler;
4759 import java.util.logging.Logger;
4760 import org.lttng.ust.agent.jul.LttngLogHandler;
4761 import org.lttng.ust.agent.context.ContextInfoManager;
4762 import org.lttng.ust.agent.context.IContextInfoRetriever;
4766 // Our context information retriever class
4767 private static class MyContextInfoRetriever
4768 implements IContextInfoRetriever
4771 public Object retrieveContextInfo(String key) {
4772 if (key.equals("intCtx")) {
4774 } else if (key.equals("strContext")) {
4775 return "context value!";
4782 private static final int answer = 42;
4784 public static void main(String args[]) throws Exception
4786 // Get the context information manager instance
4787 ContextInfoManager cim = ContextInfoManager.getInstance();
4789 // Create and register our context information retriever
4790 IContextInfoRetriever cir = new MyContextInfoRetriever();
4791 cim.registerContextInfoRetriever("myRetriever", cir);
4794 Logger logger = Logger.getLogger("jello");
4796 // Create an LTTng-UST log handler
4797 Handler lttngUstLogHandler = new LttngLogHandler();
4799 // Add the LTTng-UST log handler to our logger
4800 logger.addHandler(lttngUstLogHandler);
4803 logger.info("some info");
4804 logger.warning("some warning");
4806 logger.finer("finer information; the answer is " + answer);
4808 logger.severe("error!");
4810 // Not mandatory, but cleaner
4811 logger.removeHandler(lttngUstLogHandler);
4812 lttngUstLogHandler.close();
4813 cim.unregisterContextInfoRetriever("myRetriever");
4822 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4825 <<creating-destroying-tracing-sessions,Create a tracing session>>
4826 and <<enabling-disabling-events,create an event rule>> matching the
4832 $ lttng enable-event --jul jello
4835 <<adding-context,Add the application-specific context fields>> to the
4840 $ lttng add-context --jul --type='$app.myRetriever:intCtx'
4841 $ lttng add-context --jul --type='$app.myRetriever:strContext'
4844 <<basic-tracing-session-control,Start tracing>>:
4851 Run the compiled class:
4855 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4858 <<basic-tracing-session-control,Stop tracing>> and inspect the
4870 [[python-application]]
4871 === User space Python agent
4873 You can instrument a Python{nbsp}2 or Python{nbsp}3 application which
4875 https://docs.python.org/3/library/logging.html[`logging`] package.
4877 Each log statement emits an LTTng event once the
4878 application module imports the
4879 <<lttng-ust-agents,LTTng-UST Python agent>> package.
4882 .A Python application importing the LTTng-UST Python agent.
4883 image::python-app.png[]
4885 To use the LTTng-UST Python agent:
4887 . In the Python application's source code, import the LTTng-UST Python
4897 The LTTng-UST Python agent automatically adds its logging handler to the
4898 root logger at import time.
4900 Any log statement that the application executes before this import does
4901 not emit an LTTng event.
4903 IMPORTANT: The LTTng-UST Python agent must be
4904 <<installing-lttng,installed>>.
4906 . Use log statements and logging configuration as usual.
4907 Since the LTTng-UST Python agent adds a handler to the _root_
4908 logger, you can trace any log statement from any logger.
4910 .Use the LTTng-UST Python agent.
4921 logging.basicConfig()
4922 logger = logging.getLogger('my-logger')
4925 logger.debug('debug message')
4926 logger.info('info message')
4927 logger.warn('warn message')
4928 logger.error('error message')
4929 logger.critical('critical message')
4933 if __name__ == '__main__':
4937 NOTE: `logging.basicConfig()`, which adds to the root logger a basic
4938 logging handler which prints to the standard error stream, isn't
4939 strictly required for LTTng-UST tracing to work, but in versions of
4940 Python preceding{nbsp}3.2, you could see a warning message which indicates
4941 that no handler exists for the logger `my-logger`.
4943 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4944 <<enabling-disabling-events,create an event rule>> matching the
4945 `my-logger` Python logger, and <<basic-tracing-session-control,start
4951 $ lttng enable-event --python my-logger
4955 Run the Python script:
4962 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
4972 In the resulting trace, an <<event,event record>> generated by a Python
4973 application is named `lttng_python:event` and has the following fields:
4976 Logging time (string).
4979 Log record's message.
4985 Name of the function in which the log statement was executed.
4988 Line number at which the log statement was executed.
4991 Log level integer value.
4994 ID of the Python thread in which the log statement was executed.
4997 Name of the Python thread in which the log statement was executed.
4999 Use the opt:lttng-enable-event(1):--loglevel or
5000 opt:lttng-enable-event(1):--loglevel-only option of the
5001 man:lttng-enable-event(1) command to target a range of Python log levels
5002 or a specific Python log level.
5004 When an application imports the LTTng-UST Python agent, the agent tries
5005 to register to a <<lttng-sessiond,session daemon>>. Note that you must
5006 <<start-sessiond,start the session daemon>> _before_ you run the Python
5007 application. If a session daemon is found, the agent tries to register
5008 to it during five seconds, after which the application continues
5009 without LTTng tracing support. Override this timeout value with
5010 the env:LTTNG_UST_PYTHON_REGISTER_TIMEOUT environment variable
5013 If the session daemon stops while a Python application with an imported
5014 LTTng-UST Python agent runs, the agent retries to connect and to
5015 register to a session daemon every three seconds. Override this
5016 delay with the env:LTTNG_UST_PYTHON_REGISTER_RETRY_DELAY environment
5021 [[proc-lttng-logger-abi]]
5024 The `lttng-tracer` Linux kernel module, part of
5025 <<lttng-modules,LTTng-modules>>, creates the special LTTng logger files
5026 path:{/proc/lttng-logger} and path:{/dev/lttng-logger} (since
5027 LTTng{nbsp}2.11) when it's loaded. Any application can write text data
5028 to any of those files to emit an LTTng event.
5031 .An application writes to the LTTng logger file to emit an LTTng event.
5032 image::lttng-logger.png[]
5034 The LTTng logger is the quickest method--not the most efficient,
5035 however--to add instrumentation to an application. It is designed
5036 mostly to instrument shell scripts:
5040 $ echo "Some message, some $variable" > /dev/lttng-logger
5043 Any event that the LTTng logger emits is named `lttng_logger` and
5044 belongs to the Linux kernel <<domain,tracing domain>>. However, unlike
5045 other instrumentation points in the kernel tracing domain, **any Unix
5046 user** can <<enabling-disabling-events,create an event rule>> which
5047 matches its event name, not only the root user or users in the
5048 <<tracing-group,tracing group>>.
5050 To use the LTTng logger:
5052 * From any application, write text data to the path:{/dev/lttng-logger}
5055 The `msg` field of `lttng_logger` event records contains the
5058 NOTE: The maximum message length of an LTTng logger event is
5059 1024{nbsp}bytes. Writing more than this makes the LTTng logger emit more
5060 than one event to contain the remaining data.
5062 You shouldn't use the LTTng logger to trace a user application which
5063 can be instrumented in a more efficient way, namely:
5065 * <<c-application,C and $$C++$$ applications>>.
5066 * <<java-application,Java applications>>.
5067 * <<python-application,Python applications>>.
5069 .Use the LTTng logger.
5074 echo 'Hello, World!' > /dev/lttng-logger
5076 df --human-readable --print-type / > /dev/lttng-logger
5079 <<creating-destroying-tracing-sessions,Create a tracing session>>,
5080 <<enabling-disabling-events,create an event rule>> matching the
5081 `lttng_logger` Linux kernel tracepoint, and
5082 <<basic-tracing-session-control,start tracing>>:
5087 $ lttng enable-event --kernel lttng_logger
5091 Run the Bash script:
5098 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
5109 [[instrumenting-linux-kernel]]
5110 === LTTng kernel tracepoints
5112 NOTE: This section shows how to _add_ instrumentation points to the
5113 Linux kernel. The kernel's subsystems are already thoroughly
5114 instrumented at strategic places for LTTng when you
5115 <<installing-lttng,install>> the <<lttng-modules,LTTng-modules>>
5119 There are two methods to instrument the Linux kernel:
5121 . <<linux-add-lttng-layer,Add an LTTng layer>> over an existing ftrace
5122 tracepoint which uses the `TRACE_EVENT()` API.
5124 Choose this if you want to instrumentation a Linux kernel tree with an
5125 instrumentation point compatible with ftrace, perf, and SystemTap.
5127 . Use an <<linux-lttng-tracepoint-event,LTTng-only approach>> to
5128 instrument an out-of-tree kernel module.
5130 Choose this if you don't need ftrace, perf, or SystemTap support.
5134 [[linux-add-lttng-layer]]
5135 ==== [[instrumenting-linux-kernel-itself]][[mainline-trace-event]][[lttng-adaptation-layer]]Add an LTTng layer to an existing ftrace tracepoint
5137 This section shows how to add an LTTng layer to existing ftrace
5138 instrumentation using the `TRACE_EVENT()` API.
5140 This section doesn't document the `TRACE_EVENT()` macro. Read the
5141 following articles to learn more about this API:
5143 * http://lwn.net/Articles/379903/[Using the TRACE_EVENT() macro (Part{nbsp}1)]
5144 * http://lwn.net/Articles/381064/[Using the TRACE_EVENT() macro (Part{nbsp}2)]
5145 * http://lwn.net/Articles/383362/[Using the TRACE_EVENT() macro (Part{nbsp}3)]
5147 The following procedure assumes that your ftrace tracepoints are
5148 correctly defined in their own header and that they are created in
5149 one source file using the `CREATE_TRACE_POINTS` definition.
5151 To add an LTTng layer over an existing ftrace tracepoint:
5153 . Make sure the following kernel configuration options are
5159 * `CONFIG_HIGH_RES_TIMERS`
5160 * `CONFIG_TRACEPOINTS`
5163 . Build the Linux source tree with your custom ftrace tracepoints.
5164 . Boot the resulting Linux image on your target system.
5166 Confirm that the tracepoints exist by looking for their names in the
5167 dir:{/sys/kernel/debug/tracing/events/subsys} directory, where `subsys`
5168 is your subsystem's name.
5170 . Get a copy of the latest LTTng-modules{nbsp}{revision}:
5175 $ cd $(mktemp -d) &&
5176 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.12.tar.bz2 &&
5177 tar -xf lttng-modules-latest-2.12.tar.bz2 &&
5178 cd lttng-modules-2.12.*
5182 . In dir:{instrumentation/events/lttng-module}, relative to the root
5183 of the LTTng-modules source tree, create a header file named
5184 +__subsys__.h+ for your custom subsystem +__subsys__+ and write your
5185 LTTng-modules tracepoint definitions using the LTTng-modules
5188 Start with this template:
5192 .path:{instrumentation/events/lttng-module/my_subsys.h}
5195 #define TRACE_SYSTEM my_subsys
5197 #if !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ)
5198 #define _LTTNG_MY_SUBSYS_H
5200 #include "../../../probes/lttng-tracepoint-event.h"
5201 #include <linux/tracepoint.h>
5203 LTTNG_TRACEPOINT_EVENT(
5205 * Format is identical to TRACE_EVENT()'s version for the three
5206 * following macro parameters:
5209 TP_PROTO(int my_int, const char *my_string),
5210 TP_ARGS(my_int, my_string),
5212 /* LTTng-modules specific macros */
5214 ctf_integer(int, my_int_field, my_int)
5215 ctf_string(my_bar_field, my_bar)
5219 #endif /* !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ) */
5221 #include "../../../probes/define_trace.h"
5225 The entries in the `TP_FIELDS()` section are the list of fields for the
5226 LTTng tracepoint. This is similar to the `TP_STRUCT__entry()` part of
5227 ftrace's `TRACE_EVENT()` macro.
5229 See <<lttng-modules-tp-fields,Tracepoint fields macros>> for a
5230 complete description of the available `ctf_*()` macros.
5232 . Create the LTTng-modules probe's kernel module C source file,
5233 +probes/lttng-probe-__subsys__.c+, where +__subsys__+ is your
5238 .path:{probes/lttng-probe-my-subsys.c}
5240 #include <linux/module.h>
5241 #include "../lttng-tracer.h"
5244 * Build-time verification of mismatch between mainline
5245 * TRACE_EVENT() arguments and the LTTng-modules adaptation
5246 * layer LTTNG_TRACEPOINT_EVENT() arguments.
5248 #include <trace/events/my_subsys.h>
5250 /* Create LTTng tracepoint probes */
5251 #define LTTNG_PACKAGE_BUILD
5252 #define CREATE_TRACE_POINTS
5253 #define TRACE_INCLUDE_PATH ../instrumentation/events/lttng-module
5255 #include "../instrumentation/events/lttng-module/my_subsys.h"
5257 MODULE_LICENSE("GPL and additional rights");
5258 MODULE_AUTHOR("Your name <your-email>");
5259 MODULE_DESCRIPTION("LTTng my_subsys probes");
5260 MODULE_VERSION(__stringify(LTTNG_MODULES_MAJOR_VERSION) "."
5261 __stringify(LTTNG_MODULES_MINOR_VERSION) "."
5262 __stringify(LTTNG_MODULES_PATCHLEVEL_VERSION)
5263 LTTNG_MODULES_EXTRAVERSION);
5267 . Edit path:{probes/KBuild} and add your new kernel module object
5268 next to the existing ones:
5272 .path:{probes/KBuild}
5276 obj-m += lttng-probe-module.o
5277 obj-m += lttng-probe-power.o
5279 obj-m += lttng-probe-my-subsys.o
5285 . Build and install the LTTng kernel modules:
5290 $ make KERNELDIR=/path/to/linux
5291 # make modules_install && depmod -a
5295 Replace `/path/to/linux` with the path to the Linux source tree where
5296 you defined and used tracepoints with ftrace's `TRACE_EVENT()` macro.
5298 Note that you can also use the
5299 <<lttng-tracepoint-event-code,`LTTNG_TRACEPOINT_EVENT_CODE()` macro>>
5300 instead of `LTTNG_TRACEPOINT_EVENT()` to use custom local variables and
5301 C code that need to be executed before the event fields are recorded.
5303 The best way to learn how to use the previous LTTng-modules macros is to
5304 inspect the existing LTTng-modules tracepoint definitions in the
5305 dir:{instrumentation/events/lttng-module} header files. Compare them
5306 with the Linux kernel mainline versions in the
5307 dir:{include/trace/events} directory of the Linux source tree.
5311 [[lttng-tracepoint-event-code]]
5312 ===== Use custom C code to access the data for tracepoint fields
5314 Although we recommended to always use the
5315 <<lttng-adaptation-layer,`LTTNG_TRACEPOINT_EVENT()`>> macro to describe
5316 the arguments and fields of an LTTng-modules tracepoint when possible,
5317 sometimes you need a more complex process to access the data that the
5318 tracer records as event record fields. In other words, you need local
5319 variables and multiple C{nbsp}statements instead of simple
5320 argument-based expressions that you pass to the
5321 <<lttng-modules-tp-fields,`ctf_*()` macros of `TP_FIELDS()`>>.
5323 Use the `LTTNG_TRACEPOINT_EVENT_CODE()` macro instead of
5324 `LTTNG_TRACEPOINT_EVENT()` to declare custom local variables and define
5325 a block of C{nbsp}code to be executed before LTTng records the fields.
5326 The structure of this macro is:
5329 .`LTTNG_TRACEPOINT_EVENT_CODE()` macro syntax.
5331 LTTNG_TRACEPOINT_EVENT_CODE(
5333 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5334 * version for the following three macro parameters:
5337 TP_PROTO(int my_int, const char *my_string),
5338 TP_ARGS(my_int, my_string),
5340 /* Declarations of custom local variables */
5343 unsigned long b = 0;
5344 const char *name = "(undefined)";
5345 struct my_struct *my_struct;
5349 * Custom code which uses both tracepoint arguments
5350 * (in TP_ARGS()) and local variables (in TP_locvar()).
5352 * Local variables are actually members of a structure pointed
5353 * to by the special variable tp_locvar.
5357 tp_locvar->a = my_int + 17;
5358 tp_locvar->my_struct = get_my_struct_at(tp_locvar->a);
5359 tp_locvar->b = my_struct_compute_b(tp_locvar->my_struct);
5360 tp_locvar->name = my_struct_get_name(tp_locvar->my_struct);
5361 put_my_struct(tp_locvar->my_struct);
5370 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5371 * version for this, except that tp_locvar members can be
5372 * used in the argument expression parameters of
5373 * the ctf_*() macros.
5376 ctf_integer(unsigned long, my_struct_b, tp_locvar->b)
5377 ctf_integer(int, my_struct_a, tp_locvar->a)
5378 ctf_string(my_string_field, my_string)
5379 ctf_string(my_struct_name, tp_locvar->name)
5384 IMPORTANT: The C code defined in `TP_code()` must not have any side
5385 effects when executed. In particular, the code must not allocate
5386 memory or get resources without deallocating this memory or putting
5387 those resources afterwards.
5390 [[instrumenting-linux-kernel-tracing]]
5391 ==== Load and unload a custom probe kernel module
5393 You must load a <<lttng-adaptation-layer,created LTTng-modules probe
5394 kernel module>> in the kernel before it can emit LTTng events.
5396 To load the default probe kernel modules and a custom probe kernel
5399 * Use the opt:lttng-sessiond(8):--extra-kmod-probes option to give extra
5400 probe modules to load when starting a root <<lttng-sessiond,session
5404 .Load the `my_subsys`, `usb`, and the default probe modules.
5408 # lttng-sessiond --extra-kmod-probes=my_subsys,usb
5413 You only need to pass the subsystem name, not the whole kernel module
5416 To load _only_ a given custom probe kernel module:
5418 * Use the opt:lttng-sessiond(8):--kmod-probes option to give the probe
5419 modules to load when starting a root session daemon:
5422 .Load only the `my_subsys` and `usb` probe modules.
5426 # lttng-sessiond --kmod-probes=my_subsys,usb
5431 To confirm that a probe module is loaded:
5438 $ lsmod | grep lttng_probe_usb
5442 To unload the loaded probe modules:
5444 * Kill the session daemon with `SIGTERM`:
5449 # pkill lttng-sessiond
5453 You can also use man:modprobe(8)'s `--remove` option if the session
5454 daemon terminates abnormally.
5457 [[controlling-tracing]]
5460 Once an application or a Linux kernel is
5461 <<instrumenting,instrumented>> for LTTng tracing,
5464 This section is divided in topics on how to use the various
5465 <<plumbing,components of LTTng>>, in particular the <<lttng-cli,cmd:lttng
5466 command-line tool>>, to _control_ the LTTng daemons and tracers.
5468 NOTE: In the following subsections, we refer to an man:lttng(1) command
5469 using its man page name. For example, instead of _Run the `create`
5470 command to..._, we use _Run the man:lttng-create(1) command to..._.
5474 === Start a session daemon
5476 In some situations, you need to run a <<lttng-sessiond,session daemon>>
5477 (man:lttng-sessiond(8)) _before_ you can use the man:lttng(1)
5480 You will see the following error when you run a command while no session
5484 Error: No session daemon is available
5487 The only command that automatically runs a session daemon is
5488 man:lttng-create(1), which you use to
5489 <<creating-destroying-tracing-sessions,create a tracing session>>. While
5490 this is most of the time the first operation that you do, sometimes it's
5491 not. Some examples are:
5493 * <<list-instrumentation-points,List the available instrumentation points>>.
5494 * <<saving-loading-tracing-session,Load a tracing session configuration>>.
5496 [[tracing-group]] Each Unix user must have its own running session
5497 daemon to trace user applications. The session daemon that the root user
5498 starts is the only one allowed to control the LTTng kernel tracer. Users
5499 that are part of the _tracing group_ can control the root session
5500 daemon. The default tracing group name is `tracing`; set it to something
5501 else with the opt:lttng-sessiond(8):--group option when you start the
5502 root session daemon.
5504 To start a user session daemon:
5506 * Run man:lttng-sessiond(8):
5511 $ lttng-sessiond --daemonize
5515 To start the root session daemon:
5517 * Run man:lttng-sessiond(8) as the root user:
5522 # lttng-sessiond --daemonize
5526 In both cases, remove the opt:lttng-sessiond(8):--daemonize option to
5527 start the session daemon in foreground.
5529 To stop a session daemon, use man:kill(1) on its process ID (standard
5532 Note that some Linux distributions could manage the LTTng session daemon
5533 as a service. In this case, you should use the service manager to
5534 start, restart, and stop session daemons.
5537 [[creating-destroying-tracing-sessions]]
5538 === Create and destroy a tracing session
5540 Almost all the LTTng control operations happen in the scope of
5541 a <<tracing-session,tracing session>>, which is the dialogue between the
5542 <<lttng-sessiond,session daemon>> and you.
5544 To create a tracing session with a generated name:
5546 * Use the man:lttng-create(1) command:
5555 The created tracing session's name is `auto` followed by the
5558 To create a tracing session with a specific name:
5560 * Use the optional argument of the man:lttng-create(1) command:
5565 $ lttng create my-session
5569 Replace `my-session` with the specific tracing session name.
5571 LTTng appends the creation date to the created tracing session's name.
5573 LTTng writes the traces of a tracing session in
5574 +$LTTNG_HOME/lttng-trace/__name__+ by default, where +__name__+ is the
5575 name of the tracing session. Note that the env:LTTNG_HOME environment
5576 variable defaults to `$HOME` if not set.
5578 To output LTTng traces to a non-default location:
5580 * Use the opt:lttng-create(1):--output option of the man:lttng-create(1) command:
5585 $ lttng create my-session --output=/tmp/some-directory
5589 You may create as many tracing sessions as you wish.
5591 To list all the existing tracing sessions for your Unix user:
5593 * Use the man:lttng-list(1) command:
5602 [[cur-tracing-session]]When you create a tracing session, it is set as
5603 the _current tracing session_. The following man:lttng(1) commands
5604 operate on the current tracing session when you don't specify one:
5606 [role="list-3-cols"]
5607 * man:lttng-add-context(1)
5608 * man:lttng-clear(1)
5609 * man:lttng-destroy(1)
5610 * man:lttng-disable-channel(1)
5611 * man:lttng-disable-event(1)
5612 * man:lttng-disable-rotation(1)
5613 * man:lttng-enable-channel(1)
5614 * man:lttng-enable-event(1)
5615 * man:lttng-enable-rotation(1)
5617 * man:lttng-regenerate(1)
5618 * man:lttng-rotate(1)
5620 * man:lttng-snapshot(1)
5621 * man:lttng-start(1)
5622 * man:lttng-status(1)
5624 * man:lttng-track(1)
5625 * man:lttng-untrack(1)
5628 To change the current tracing session:
5630 * Use the man:lttng-set-session(1) command:
5635 $ lttng set-session new-session
5639 Replace `new-session` by the name of the new current tracing session.
5641 When you're done tracing in a given tracing session, destroy it. This
5642 operation frees the resources taken by the tracing session to destroy;
5643 it doesn't destroy the trace data that LTTng wrote for this tracing
5644 session (see <<clear,Clear a tracing session>> for one way to do this).
5646 To destroy the current tracing session:
5648 * Use the man:lttng-destroy(1) command:
5657 The man:lttng-destroy(1) command also runs the man:lttng-stop(1)
5658 command implicitly (see <<basic-tracing-session-control,Start and stop a
5659 tracing session>>). You need to stop tracing to make LTTng flush the
5660 remaining trace data and make the trace readable.
5663 [[list-instrumentation-points]]
5664 === List the available instrumentation points
5666 The <<lttng-sessiond,session daemon>> can query the running instrumented
5667 user applications and the Linux kernel to get a list of available
5668 instrumentation points. For the Linux kernel <<domain,tracing domain>>,
5669 they are tracepoints and system calls. For the user space tracing
5670 domain, they are tracepoints. For the other tracing domains, they are
5673 To list the available instrumentation points:
5675 * Use the man:lttng-list(1) command with the requested tracing domain's
5679 opt:lttng-list(1):--kernel::
5680 Linux kernel tracepoints (your Unix user must be a root user, or it
5681 must be a member of the <<tracing-group,tracing group>>).
5683 opt:lttng-list(1):--kernel with opt:lttng-list(1):--syscall::
5684 Linux kernel system calls (your Unix user must be a root user, or it
5685 must be a member of the tracing group).
5687 opt:lttng-list(1):--userspace::
5688 User space tracepoints.
5690 opt:lttng-list(1):--jul::
5691 `java.util.logging` loggers.
5693 opt:lttng-list(1):--log4j::
5694 Apache log4j loggers.
5696 opt:lttng-list(1):--python::
5700 .List the available user space tracepoints.
5704 $ lttng list --userspace
5708 .List the available Linux kernel system call tracepoints.
5712 $ lttng list --kernel --syscall
5717 [[enabling-disabling-events]]
5718 === Create and enable an event rule
5720 Once you <<creating-destroying-tracing-sessions,create a tracing
5721 session>>, you can create <<event,event rules>> with the
5722 man:lttng-enable-event(1) command.
5724 You specify each condition with a command-line option. The available
5725 condition arguments are shown in the following table.
5727 [role="growable",cols="asciidoc,asciidoc,default"]
5728 .Condition command-line arguments for the man:lttng-enable-event(1) command.
5730 |Argument |Description |Applicable tracing domains
5736 . +--probe=__ADDR__+
5737 . +--function=__ADDR__+
5738 . +--userspace-probe=__PATH__:__SYMBOL__+
5739 . +--userspace-probe=sdt:__PATH__:__PROVIDER__:__NAME__+
5742 Instead of using the default _tracepoint_ instrumentation type, use:
5744 . A Linux system call (entry and exit).
5745 . A Linux https://lwn.net/Articles/132196/[kprobe] (symbol or address).
5746 . The entry and return points of a Linux function (symbol or address).
5747 . The entry point of a user application or library function (path to
5748 application/library and symbol).
5749 . A https://www.sourceware.org/systemtap/wiki/AddingUserSpaceProbingToApps[SystemTap
5750 Statically Defined Tracing] (USDT) probe (path to application/library,
5751 provider and probe names).
5755 |First positional argument.
5758 Tracepoint or system call name.
5760 With the opt:lttng-enable-event(1):--probe,
5761 opt:lttng-enable-event(1):--function, and
5762 opt:lttng-enable-event(1):--userspace-probe options, this is a custom
5763 name given to the event rule. With the JUL, log4j, and Python domains,
5764 this is a logger name.
5766 With a tracepoint, logger, or system call name, use the special
5767 `*` globbing character to match anything (for example, `sched_*`,
5775 . +--loglevel=__LEVEL__+
5776 . +--loglevel-only=__LEVEL__+
5779 . Match only tracepoints or log statements with a logging level at
5780 least as severe as +__LEVEL__+.
5781 . Match only tracepoints or log statements with a logging level
5782 equal to +__LEVEL__+.
5784 See man:lttng-enable-event(1) for the list of available logging level
5787 |User space, JUL, log4j, and Python.
5789 |+--exclude=__EXCLUSIONS__+
5792 When you use a `*` character at the end of the tracepoint or logger
5793 name (first positional argument), exclude the specific names in the
5794 comma-delimited list +__EXCLUSIONS__+.
5797 User space, JUL, log4j, and Python.
5799 |+--filter=__EXPR__+
5802 Match only events which satisfy the expression +__EXPR__+.
5804 See man:lttng-enable-event(1) to learn more about the syntax of a
5811 You attach an event rule to a <<channel,channel>> on creation. If you do
5812 not specify the channel with the opt:lttng-enable-event(1):--channel
5813 option, and if the event rule to create is the first in its
5814 <<domain,tracing domain>> for a given tracing session, then LTTng
5815 creates a _default channel_ for you. This default channel is reused in
5816 subsequent invocations of the man:lttng-enable-event(1) command for the
5817 same tracing domain.
5819 An event rule is always enabled at creation time.
5821 The following examples show how to combine the previous
5822 command-line options to create simple to more complex event rules.
5824 .Create an event rule targetting a Linux kernel tracepoint (default channel).
5828 $ lttng enable-event --kernel sched_switch
5832 .Create an event rule matching four Linux kernel system calls (default channel).
5836 $ lttng enable-event --kernel --syscall open,write,read,close
5840 .Create event rules matching tracepoints with filter expressions (default channel).
5844 $ lttng enable-event --kernel sched_switch --filter='prev_comm == "bash"'
5849 $ lttng enable-event --kernel --all \
5850 --filter='$ctx.tid == 1988 || $ctx.tid == 1534'
5855 $ lttng enable-event --jul my_logger \
5856 --filter='$app.retriever:cur_msg_id > 3'
5859 IMPORTANT: Make sure to always quote the filter string when you
5860 use man:lttng(1) from a shell.
5862 See also <<pid-tracking,Track process attributes>> which offers another,
5863 more efficient filtering mechanism for process ID, user ID, and group
5867 .Create an event rule matching any user space tracepoint of a given tracepoint provider with a log level range (default channel).
5871 $ lttng enable-event --userspace my_app:'*' --loglevel=TRACE_INFO
5874 IMPORTANT: Make sure to always quote the wildcard character when you
5875 use man:lttng(1) from a shell.
5878 .Create an event rule matching multiple Python loggers with a wildcard and with exclusions (default channel).
5882 $ lttng enable-event --python my-app.'*' \
5883 --exclude='my-app.module,my-app.hello'
5887 .Create an event rule matching any Apache log4j logger with a specific log level (default channel).
5891 $ lttng enable-event --log4j --all --loglevel-only=LOG4J_WARN
5895 .Create an event rule attached to a specific channel matching a specific user space tracepoint provider and tracepoint.
5899 $ lttng enable-event --userspace my_app:my_tracepoint --channel=my-channel
5903 .Create an event rule matching the `malloc` function entry in path:{/usr/lib/libc.so.6}:
5907 $ lttng enable-event --kernel --userspace-probe=/usr/lib/libc.so.6:malloc \
5912 .Create an event rule matching the `server`/`accept_request` https://www.sourceware.org/systemtap/wiki/AddingUserSpaceProbingToApps[USDT probe] in path:{/usr/bin/serv}:
5916 $ lttng enable-event --kernel --userspace-probe=sdt:serv:server:accept_request \
5917 server_accept_request
5921 The event rules of a given channel form a whitelist: as soon as an
5922 emitted event passes one of them, LTTng can record the event. For
5923 example, an event named `my_app:my_tracepoint` emitted from a user space
5924 tracepoint with a `TRACE_ERROR` log level passes both of the following
5929 $ lttng enable-event --userspace my_app:my_tracepoint
5930 $ lttng enable-event --userspace my_app:my_tracepoint \
5931 --loglevel=TRACE_INFO
5934 The second event rule is redundant: the first one includes
5938 [[disable-event-rule]]
5939 === Disable an event rule
5941 To disable an event rule that you <<enabling-disabling-events,created>>
5942 previously, use the man:lttng-disable-event(1) command. This command
5943 disables _all_ the event rules (of a given tracing domain and channel)
5944 which match an instrumentation point. The other conditions aren't
5945 supported as of LTTng{nbsp}{revision}.
5947 The LTTng tracer doesn't record an emitted event which passes
5948 a _disabled_ event rule.
5950 .Disable an event rule matching a Python logger (default channel).
5954 $ lttng disable-event --python my-logger
5958 .Disable an event rule matching all `java.util.logging` loggers (default channel).
5962 $ lttng disable-event --jul '*'
5966 .Disable _all_ the event rules of the default channel.
5968 The opt:lttng-disable-event(1):--all-events option isn't, like the
5969 opt:lttng-enable-event(1):--all option of man:lttng-enable-event(1), the
5970 equivalent of the event name `*` (wildcard): it disables _all_ the event
5971 rules of a given channel.
5975 $ lttng disable-event --jul --all-events
5979 NOTE: You can't delete an event rule once you create it.
5983 === Get the status of a tracing session
5985 To get the status of the <<cur-tracing-session,current tracing
5986 session>>, that is, its parameters, its channels, event rules, and their
5989 * Use the man:lttng-status(1) command:
5998 To get the status of any tracing session:
6000 * Use the man:lttng-list(1) command with the tracing session's name:
6005 $ lttng list my-session
6009 Replace `my-session` with the desired tracing session's name.
6012 [[basic-tracing-session-control]]
6013 === Start and stop a tracing session
6015 Once you <<creating-destroying-tracing-sessions,create a tracing
6017 <<enabling-disabling-events,create one or more event rules>>,
6018 you can start and stop the tracers for this tracing session.
6020 To start tracing in the <<cur-tracing-session,current tracing session>>:
6022 * Use the man:lttng-start(1) command:
6031 LTTng is very flexible: you can launch user applications before
6032 or after the you start the tracers. The tracers only record the events
6033 if they pass enabled event rules and if they occur while the tracers are
6036 To stop tracing in the current tracing session:
6038 * Use the man:lttng-stop(1) command:
6047 If there were <<channel-overwrite-mode-vs-discard-mode,lost event
6048 records>> or lost sub-buffers since the last time you ran
6049 man:lttng-start(1), warnings are printed when you run the
6050 man:lttng-stop(1) command.
6052 IMPORTANT: You need to stop tracing to make LTTng flush the remaining
6053 trace data and make the trace readable. Note that the
6054 man:lttng-destroy(1) command (see
6055 <<creating-destroying-tracing-sessions,Create and destroy a tracing
6056 session>>) also runs the man:lttng-stop(1) command implicitly.
6060 === Clear a tracing session
6062 You might need to remove all the current tracing data of one or more
6063 <<tracing-session,tracing sessions>> between multiple attempts to
6064 reproduce a problem without interrupting the LTTng tracing activity.
6066 To clear the tracing data of the
6067 <<cur-tracing-session,current tracing session>>:
6069 * Use the man:lttng-clear(1) command:
6078 To clear the tracing data of all the tracing sessions:
6080 * Use the `lttng clear` command with the opt:lttng-clear(1):--all
6091 [[enabling-disabling-channels]]
6092 === Create a channel
6094 Once you create a tracing session, you can create a <<channel,channel>>
6095 with the man:lttng-enable-channel(1) command.
6097 Note that LTTng automatically creates a default channel when, for a
6098 given <<domain,tracing domain>>, no channels exist and you
6099 <<enabling-disabling-events,create>> the first event rule. This default
6100 channel is named `channel0` and its attributes are set to reasonable
6101 values. Therefore, you only need to create a channel when you need
6102 non-default attributes.
6104 You specify each non-default channel attribute with a command-line
6105 option when you use the man:lttng-enable-channel(1) command. The
6106 available command-line options are:
6108 [role="growable",cols="asciidoc,asciidoc"]
6109 .Command-line options for the man:lttng-enable-channel(1) command.
6111 |Option |Description
6117 <<channel-overwrite-mode-vs-discard-mode,event record loss mode>> instead
6118 of the default _discard_ mode.
6120 |`--buffers-pid` (user space tracing domain only)
6123 Use the per-process <<channel-buffering-schemes,buffering scheme>>
6124 instead of the default per-user buffering scheme.
6126 |+--subbuf-size=__SIZE__+
6129 Allocate sub-buffers of +__SIZE__+ bytes (power of two), for each CPU,
6130 either for each Unix user (default), or for each instrumented process.
6132 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6134 |+--num-subbuf=__COUNT__+
6137 Allocate +__COUNT__+ sub-buffers (power of two), for each CPU, either
6138 for each Unix user (default), or for each instrumented process.
6140 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
6142 |+--tracefile-size=__SIZE__+
6145 Set the maximum size of each trace file that this channel writes within
6146 a stream to +__SIZE__+ bytes instead of no maximum.
6148 See <<tracefile-rotation,Trace file count and size>>.
6150 |+--tracefile-count=__COUNT__+
6153 Limit the number of trace files that this channel creates to
6154 +__COUNT__+ channels instead of no limit.
6156 See <<tracefile-rotation,Trace file count and size>>.
6158 |+--switch-timer=__PERIODUS__+
6161 Set the <<channel-switch-timer,switch timer period>>
6162 to +__PERIODUS__+{nbsp}µs.
6164 |+--read-timer=__PERIODUS__+
6167 Set the <<channel-read-timer,read timer period>>
6168 to +__PERIODUS__+{nbsp}µs.
6170 |[[opt-blocking-timeout]]+--blocking-timeout=__TIMEOUTUS__+
6173 Set the timeout of user space applications which load LTTng-UST
6174 in blocking mode to +__TIMEOUTUS__+:
6177 Never block (non-blocking mode).
6180 Block forever until space is available in a sub-buffer to record
6183 __n__, a positive value::
6184 Wait for at most __n__ µs when trying to write into a sub-buffer.
6186 Note that, for this option to have any effect on an instrumented
6187 user space application, you need to run the application with a set
6188 env:LTTNG_UST_ALLOW_BLOCKING environment variable.
6190 |+--output=__TYPE__+ (Linux kernel tracing domain only)
6193 Set the channel's output type to +__TYPE__+, either `mmap` or `splice`.
6197 You can only create a channel in the Linux kernel and user space
6198 <<domain,tracing domains>>: other tracing domains have their own channel
6199 created on the fly when <<enabling-disabling-events,creating event
6204 Because of a current LTTng limitation, you must create all channels
6205 _before_ you <<basic-tracing-session-control,start tracing>> in a given
6206 tracing session, that is, before the first time you run
6209 Since LTTng automatically creates a default channel when you use the
6210 man:lttng-enable-event(1) command with a specific tracing domain, you
6211 can't, for example, create a Linux kernel event rule, start tracing,
6212 and then create a user space event rule, because no user space channel
6213 exists yet and it's too late to create one.
6215 For this reason, make sure to configure your channels properly
6216 before starting the tracers for the first time!
6219 The following examples show how to combine the previous
6220 command-line options to create simple to more complex channels.
6222 .Create a Linux kernel channel with default attributes.
6226 $ lttng enable-channel --kernel my-channel
6230 .Create a user space channel with four sub-buffers or 1{nbsp}MiB each, per CPU, per instrumented process.
6234 $ lttng enable-channel --userspace --num-subbuf=4 --subbuf-size=1M \
6235 --buffers-pid my-channel
6239 .[[blocking-timeout-example]]Create a default user space channel with an infinite blocking timeout.
6241 <<creating-destroying-tracing-sessions,Create a tracing-session>>,
6242 create the channel, <<enabling-disabling-events,create an event rule>>,
6243 and <<basic-tracing-session-control,start tracing>>:
6248 $ lttng enable-channel --userspace --blocking-timeout=inf blocking-channel
6249 $ lttng enable-event --userspace --channel=blocking-channel --all
6253 Run an application instrumented with LTTng-UST and allow it to block:
6257 $ LTTNG_UST_ALLOW_BLOCKING=1 my-app
6261 .Create a Linux kernel channel which rotates eight trace files of 4{nbsp}MiB each for each stream
6265 $ lttng enable-channel --kernel --tracefile-count=8 \
6266 --tracefile-size=4194304 my-channel
6270 .Create a user space channel in overwrite (or _flight recorder_) mode.
6274 $ lttng enable-channel --userspace --overwrite my-channel
6278 <<enabling-disabling-events,Create>> the same event rule in
6279 two different channels:
6283 $ lttng enable-event --userspace --channel=my-channel app:tp
6284 $ lttng enable-event --userspace --channel=other-channel app:tp
6287 If both channels are enabled, when a tracepoint named `app:tp` is
6288 reached, LTTng records two events, one for each channel.
6292 === Disable a channel
6294 To disable a specific channel that you <<enabling-disabling-channels,created>>
6295 previously, use the man:lttng-disable-channel(1) command.
6297 .Disable a specific Linux kernel channel.
6301 $ lttng disable-channel --kernel my-channel
6305 The state of a channel precedes the individual states of event rules
6306 attached to it: event rules which belong to a disabled channel, even if
6307 they are enabled, are also considered disabled.
6311 === Add context fields to a channel
6313 Event record fields in trace files provide important information about
6314 events that occured previously, but sometimes some external context may
6315 help you solve a problem faster.
6317 Examples of context fields are:
6319 * The **process ID**, **thread ID**, **process name**, and
6320 **process priority** of the thread in which the event occurs.
6321 * The **hostname** of the system on which the event occurs.
6322 * The Linux kernel and user call stacks (since
6324 * The current values of many possible **performance counters** using
6326 ** CPU cycles, stalled cycles, idle cycles, and the other cycle types.
6328 ** Branch instructions, misses, and loads.
6330 * Any context defined at the application level (supported for the
6331 JUL and log4j <<domain,tracing domains>>).
6333 To get the full list of available context fields, see
6334 `lttng add-context --list`. Some context fields are reserved for a
6335 specific <<domain,tracing domain>> (Linux kernel or user space).
6337 You add context fields to <<channel,channels>>. All the events
6338 that a channel with added context fields records contain those fields.
6340 To add context fields to one or all the channels of a given tracing
6343 * Use the man:lttng-add-context(1) command.
6345 .Add context fields to all the channels of the current tracing session.
6347 The following command line adds the virtual process identifier and
6348 the per-thread CPU cycles count fields to all the user space channels
6350 <<cur-tracing-session,current tracing session>>.
6354 $ lttng add-context --userspace --type=vpid --type=perf:thread:cpu-cycles
6358 .Add performance counter context fields by raw ID
6360 See man:lttng-add-context(1) for the exact format of the context field
6361 type, which is partly compatible with the format used in
6366 $ lttng add-context --userspace --type=perf:thread:raw:r0110:test
6367 $ lttng add-context --kernel --type=perf:cpu:raw:r0013c:x86unhalted
6371 .Add context fields to a specific channel.
6373 The following command line adds the thread identifier and user call
6374 stack context fields to the Linux kernel channel named `my-channel` in
6375 the current tracing session.
6379 $ lttng add-context --kernel --channel=my-channel \
6380 --type=tid --type=callstack-user
6384 .Add an application-specific context field to a specific channel.
6386 The following command line adds the `cur_msg_id` context field of the
6387 `retriever` context retriever for all the instrumented
6388 <<java-application,Java applications>> recording <<event,event records>>
6389 in the channel named `my-channel`:
6393 $ lttng add-context --kernel --channel=my-channel \
6394 --type='$app:retriever:cur_msg_id'
6397 IMPORTANT: Make sure to always quote the `$` character when you
6398 use man:lttng-add-context(1) from a shell.
6401 NOTE: You can't remove context fields from a channel once you add it.
6406 === Track process attributes
6408 It's often useful to only allow processes with specific attributes to
6409 emit events. For example, you may wish to record all the system calls
6410 which a given process makes (à la
6411 http://linux.die.net/man/1/strace[strace]).
6413 The man:lttng-track(1) and man:lttng-untrack(1) commands serve this
6414 purpose. Both commands operate on _inclusion sets_ of process attribute
6415 values. The available process attribute types are:
6417 Linux kernel <<domain,tracing domain>> only::
6421 * Virtual process ID (VPID).
6423 This is the PID as seen by the application.
6425 * Unix user ID (UID) (since LTTng{nbsp}2.12).
6427 * Virtual Unix user ID (VUID) (since LTTng{nbsp}2.12).
6429 This is the UID as seen by the application.
6431 * Unix group ID (GID) (since LTTng{nbsp}2.12).
6433 * Virtual Unix group ID (VGID) (since LTTng{nbsp}2.12).
6435 This is the GID as seen by the application.
6438 User space tracing domain::
6441 * VUID (since LTTng{nbsp}2.12).
6442 * VGID (since LTTng{nbsp}2.12).
6444 Each tracing domain has one inclusion set per process attribute type:
6445 the Linux kernel tracing domain has six while the user space tracing
6448 For a given event which passes an enabled <<event,event rule>> to be
6449 recorded, _all_ the attributes of its executing process must be part of
6450 the inclusion sets of the event rule's tracing domain.
6452 Add entries to an inclusion set with the man:lttng-track(1) command and
6453 remove entries with the man:lttng-untrack(1) command. A process
6454 attribute is _tracked_ when it's part of an inclusion set and
6455 _untracked_ otherwise.
6459 The process attribute values are _numeric_.
6461 Should a process with a given tracked process ID, for example, exit, and
6462 then a new process be given this ID, then the latter would also be
6463 allowed to emit events.
6465 With the `lttng track` command, you can add Unix user and group _names_
6466 to the user and group inclusion sets: the <<lttng-sessiond,session
6467 daemon>> finds the corresponding UID, VUID, GID, or VGID once on
6468 _addition_ to the inclusion set. This means that if you rename the user
6469 or group after you run `lttng track`, its user/group ID remains tracked.
6472 .Track and untrack virtual process IDs.
6474 For the sake of the following example, assume the target system has
6475 16{nbsp}possible VPIDs.
6478 <<creating-destroying-tracing-sessions,create a tracing session>>,
6479 the user space VPID inclusion set contains _all_ the possible VPIDs:
6482 .All VPIDs are tracked.
6483 image::track-all.png[]
6485 When the inclusion set is full and you use the man:lttng-track(1)
6486 command to specify some VPIDs to track, LTTng first clears the inclusion
6487 set, and then it adds the specific VPIDs to track. After:
6491 $ lttng track --userspace --vpid=3,4,7,10,13
6494 the VPID inclusion set is:
6497 .VPIDs 3, 4, 7, 10, and 13 are tracked.
6498 image::track-3-4-7-10-13.png[]
6500 Add more VPIDs to the inclusion set afterwards:
6504 $ lttng track --userspace --vpid=1,15,16
6510 .VPIDs 1, 15, and 16 are added to the inclusion set.
6511 image::track-1-3-4-7-10-13-15-16.png[]
6513 The man:lttng-untrack(1) command removes entries from process attribute
6514 inclusion sets. Given the previous example, the following command:
6518 $ lttng untrack --userspace --vpid=3,7,10,13
6521 leads to this VPID inclusion set:
6524 .VPIDs 3, 7, 10, and 13 are removed from the inclusion set.
6525 image::track-1-4-15-16.png[]
6527 LTTng can track all the possible VPIDs again using the
6528 opt:lttng-track(1):--all option:
6532 $ lttng track --userspace --vpid --all
6535 The result is, again:
6538 .All VPIDs are tracked.
6539 image::track-all.png[]
6542 .Track only specific process attributes.
6544 A typical use case with process attribute tracking is to start with an
6545 empty inclusion set, then <<basic-tracing-session-control,start the
6546 tracers>>, and then add entries manually while the tracers are active.
6548 Use the opt:lttng-untrack(1):--all option of the
6549 man:lttng-untrack(1) command to clear the inclusion set after you
6550 <<creating-destroying-tracing-sessions,create a tracing session>>, for
6551 example (with UIDs):
6555 $ lttng untrack --kernel --uid --all
6561 .No UIDs are tracked.
6562 image::untrack-all.png[]
6564 If you trace with this inclusion set configuration, the LTTng kernel
6565 tracer records no events within the <<cur-tracing-session,current
6566 tracing session>> because it doesn't track any UID. Use the
6567 man:lttng-track(1) command as usual to track specific UIDs when you need
6572 $ lttng track --kernel --uid=http,11
6578 .UIDs 6 (`http`) and 11 are tracked.
6579 image::track-6-11.png[]
6584 [[saving-loading-tracing-session]]
6585 === Save and load tracing session configurations
6587 Configuring a <<tracing-session,tracing session>> can be long. Some of
6588 the tasks involved are:
6590 * <<enabling-disabling-channels,Create channels>> with
6591 specific attributes.
6592 * <<adding-context,Add context fields>> to specific channels.
6593 * <<enabling-disabling-events,Create event rules>> with specific log
6594 level and filter conditions.
6596 If you use LTTng to solve real world problems, chances are you have to
6597 record events using the same tracing session setup over and over,
6598 modifying a few variables each time in your instrumented program
6599 or environment. To avoid constant tracing session reconfiguration,
6600 the man:lttng(1) command-line tool can save and load tracing session
6601 configurations to/from XML files.
6603 To save a given tracing session configuration:
6605 * Use the man:lttng-save(1) command:
6610 $ lttng save my-session
6614 Replace `my-session` with the name of the tracing session to save.
6616 LTTng saves tracing session configurations to
6617 dir:{$LTTNG_HOME/.lttng/sessions} by default. Note that the
6618 env:LTTNG_HOME environment variable defaults to `$HOME` if not set. Use
6619 the opt:lttng-save(1):--output-path option to change this destination
6622 LTTng saves all configuration parameters, for example:
6624 * The tracing session name.
6625 * The trace data output path.
6626 * The channels with their state and all their attributes.
6627 * The context fields you added to channels.
6628 * The event rules with their state, log level and filter conditions.
6630 To load a tracing session:
6632 * Use the man:lttng-load(1) command:
6637 $ lttng load my-session
6641 Replace `my-session` with the name of the tracing session to load.
6643 When LTTng loads a configuration, it restores your saved tracing session
6644 as if you just configured it manually.
6646 See man:lttng-load(1) for the complete list of command-line options. You
6647 can also save and load many sessions at a time, and decide in which
6648 directory to output the XML files.
6651 [[sending-trace-data-over-the-network]]
6652 === Send trace data over the network
6654 LTTng can send the recorded trace data to a remote system over the
6655 network instead of writing it to the local file system.
6657 To send the trace data over the network:
6659 . On the _remote_ system (which can also be the target system),
6660 start an LTTng <<lttng-relayd,relay daemon>> (man:lttng-relayd(8)):
6669 . On the _target_ system, create a tracing session configured to
6670 send trace data over the network:
6675 $ lttng create my-session --set-url=net://remote-system
6679 Replace `remote-system` by the host name or IP address of the
6680 remote system. See man:lttng-create(1) for the exact URL format.
6682 . On the target system, use the man:lttng(1) command-line tool as usual.
6683 When tracing is active, the target's consumer daemon sends sub-buffers
6684 to the relay daemon running on the remote system instead of flushing
6685 them to the local file system. The relay daemon writes the received
6686 packets to the local file system.
6688 The relay daemon writes trace files to
6689 +$LTTNG_HOME/lttng-traces/__hostname__/__session__+ by default, where
6690 +__hostname__+ is the host name of the target system and +__session__+
6691 is the tracing session name. Note that the env:LTTNG_HOME environment
6692 variable defaults to `$HOME` if not set. Use the
6693 opt:lttng-relayd(8):--output option of man:lttng-relayd(8) to write
6694 trace files to another base directory.
6699 === View events as LTTng emits them (noch:{LTTng} live)
6701 LTTng live is a network protocol implemented by the <<lttng-relayd,relay
6702 daemon>> (man:lttng-relayd(8)) to allow compatible trace viewers to
6703 display events as LTTng emits them on the target system while tracing is
6706 The relay daemon creates a _tee_: it forwards the trace data to both
6707 the local file system and to connected live viewers:
6710 .The relay daemon creates a _tee_, forwarding the trace data to both trace files and a connected live viewer.
6715 . On the _target system_, create a <<tracing-session,tracing session>>
6721 $ lttng create my-session --live
6725 This spawns a local relay daemon.
6727 . Start the live viewer and configure it to connect to the relay
6728 daemon. For example, with
6729 https://babeltrace.org/docs/v2.0/man1/babeltrace2.1/[cmd:babeltrace2]:
6734 $ babeltrace2 net://localhost/host/hostname/my-session
6741 * `hostname` with the host name of the target system.
6742 * `my-session` with the name of the tracing session to view.
6745 . Configure the tracing session as usual with the man:lttng(1)
6746 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6748 List the available live tracing sessions with Babeltrace{nbsp}2:
6752 $ babeltrace2 net://localhost
6755 You can start the relay daemon on another system. In this case, you need
6756 to specify the relay daemon's URL when you create the tracing session
6757 with the opt:lttng-create(1):--set-url option. You also need to replace
6758 `localhost` in the procedure above with the host name of the system on
6759 which the relay daemon is running.
6761 See man:lttng-create(1) and man:lttng-relayd(8) for the complete list of
6762 command-line options.
6766 [[taking-a-snapshot]]
6767 === Take a snapshot of the current sub-buffers of a tracing session
6769 The normal behavior of LTTng is to append full sub-buffers to growing
6770 trace data files. This is ideal to keep a full history of the events
6771 that occurred on the target system, but it can
6772 represent too much data in some situations. For example, you may wish
6773 to trace your application continuously until some critical situation
6774 happens, in which case you only need the latest few recorded
6775 events to perform the desired analysis, not multi-gigabyte trace files.
6777 With the man:lttng-snapshot(1) command, you can take a snapshot of the
6778 current sub-buffers of a given <<tracing-session,tracing session>>.
6779 LTTng can write the snapshot to the local file system or send it over
6783 .A snapshot is a copy of the current sub-buffers, which aren't cleared after the operation.
6784 image::snapshot.png[]
6786 If you wish to create unmanaged, self-contained, non-overlapping
6787 trace chunk archives instead of a simple copy of the current
6788 sub-buffers, see the <<session-rotation,tracing session rotation>>
6789 feature (available since LTTng{nbsp}2.11).
6793 . Create a tracing session in _snapshot mode_:
6798 $ lttng create my-session --snapshot
6802 The <<channel-overwrite-mode-vs-discard-mode,event record loss mode>> of
6803 <<channel,channels>> created in this mode is automatically set to
6804 _overwrite_ (flight recorder mode).
6806 . Configure the tracing session as usual with the man:lttng(1)
6807 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6809 . **Optional**: When you need to take a snapshot,
6810 <<basic-tracing-session-control,stop tracing>>.
6812 You can take a snapshot when the tracers are active, but if you stop
6813 them first, you're sure that the data in the sub-buffers doesn't
6814 change before you actually take the snapshot.
6821 $ lttng snapshot record --name=my-first-snapshot
6825 LTTng writes the current sub-buffers of all the
6826 <<cur-tracing-session,current tracing session>>'s channels to
6827 trace files on the local file system. Those trace files have
6828 `my-first-snapshot` in their name.
6830 There is no difference between the format of a normal trace file and the
6831 format of a snapshot: viewers of LTTng traces also support LTTng
6834 By default, LTTng writes snapshot files to the path shown by
6835 `lttng snapshot list-output`. You can change this path or decide to send
6836 snapshots over the network using either:
6838 . An output path or URL that you specify when you
6839 <<creating-destroying-tracing-sessions,create the tracing session>>.
6840 . A snapshot output path or URL that you add using
6841 `lttng snapshot add-output`.
6842 . An output path or URL that you provide directly to the
6843 `lttng snapshot record` command.
6845 Method{nbsp}3 overrides method{nbsp}2, which overrides method 1. When
6846 you specify a URL, a relay daemon must listen on a remote system (see
6847 <<sending-trace-data-over-the-network,Send trace data over the
6852 [[session-rotation]]
6853 === Archive the current trace chunk (rotate a tracing session)
6855 The <<taking-a-snapshot,snapshot user guide>> shows how to dump
6856 a tracing session's current sub-buffers to the file system or send them
6857 over the network. When you take a snapshot, LTTng doesn't clear the
6858 tracing session's ring buffers: if you take another snapshot immediately
6859 after, both snapshots could contain overlapping trace data.
6861 Inspired by https://en.wikipedia.org/wiki/Log_rotation[log rotation],
6862 _tracing session rotation_ is a feature which appends the content of the
6863 ring buffers to what's already on the file system or sent over the
6864 network since the tracing session's creation or since the last
6865 rotation, and then clears those ring buffers to avoid trace data
6868 What LTTng is about to write when performing a tracing session rotation
6869 is called the _current trace chunk_. When this current trace chunk is
6870 written to the file system or sent over the network, it becomes a _trace
6871 chunk archive_. Therefore, a tracing session rotation _archives_ the
6872 current trace chunk.
6875 .A tracing session rotation operation _archives_ the current trace chunk.
6876 image::rotation.png[]
6878 A trace chunk archive is a self-contained LTTng trace which LTTng
6879 doesn't manage anymore: you can read it, modify it, move it, or remove
6882 There are two methods to perform a tracing session rotation: immediately
6883 or with a rotation schedule.
6885 To perform an immediate tracing session rotation:
6887 . <<creating-destroying-tracing-sessions,Create a tracing session>>
6888 in _normal mode_ or _network streaming mode_
6889 (only those two creation modes support tracing session rotation):
6894 $ lttng create my-session
6898 . <<enabling-disabling-events,Create one or more event rules>>
6899 and <<basic-tracing-session-control,start tracing>>:
6904 $ lttng enable-event --kernel sched_'*'
6909 . When needed, immediately rotate the
6910 <<cur-tracing-session,current tracing session>>:
6919 The cmd:lttng-rotate command prints the path to the created trace
6920 chunk archive. See man:lttng-rotate(1) to learn about the format
6921 of trace chunk archive directory names.
6923 Perform other immediate rotations while the tracing session is
6924 active. It is guaranteed that all the trace chunk archives don't
6925 contain overlapping trace data. You can also perform an immediate
6926 rotation once you have <<basic-tracing-session-control,stopped>> the
6929 . When you're done tracing,
6930 <<creating-destroying-tracing-sessions,destroy the current tracing
6940 The tracing session destruction operation creates one last trace
6941 chunk archive from the current trace chunk.
6943 A tracing session rotation schedule is a planned rotation which LTTng
6944 performs automatically based on one of the following conditions:
6946 * A timer with a configured period times out.
6948 * The total size of the flushed part of the current trace chunk
6949 becomes greater than or equal to a configured value.
6951 To schedule a tracing session rotation, set a _rotation schedule_:
6953 . <<creating-destroying-tracing-sessions,Create a tracing session>>
6954 in _normal mode_ or _network streaming mode_
6955 (only those two creation modes support tracing session rotation):
6960 $ lttng create my-session
6964 . <<enabling-disabling-events,Create one or more event rules>>:
6969 $ lttng enable-event --kernel sched_'*'
6973 . Set a tracing session rotation schedule:
6978 $ lttng enable-rotation --timer=10s
6982 In this example, we set a rotation schedule so that LTTng performs a
6983 tracing session rotation every ten seconds.
6985 See man:lttng-enable-rotation(1) to learn more about other ways to set a
6988 . <<basic-tracing-session-control,Start tracing>>:
6997 LTTng performs tracing session rotations automatically while the tracing
6998 session is active thanks to the rotation schedule.
7000 . When you're done tracing,
7001 <<creating-destroying-tracing-sessions,destroy the current tracing
7011 The tracing session destruction operation creates one last trace chunk
7012 archive from the current trace chunk.
7014 Use man:lttng-disable-rotation(1) to unset a tracing session
7017 NOTE: man:lttng-rotate(1) and man:lttng-enable-rotation(1) list
7018 limitations regarding those two commands.
7023 === Use the machine interface
7025 With any command of the man:lttng(1) command-line tool, set the
7026 opt:lttng(1):--mi option to `xml` (before the command name) to get an
7027 XML machine interface output, for example:
7031 $ lttng --mi=xml enable-event --kernel --syscall open
7034 A schema definition (XSD) is
7035 https://github.com/lttng/lttng-tools/blob/stable-{revision}/src/common/src/common/mi-lttng-4.0.xsd[available]
7036 to ease the integration with external tools as much as possible.
7040 [[metadata-regenerate]]
7041 === Regenerate the metadata of an LTTng trace
7043 An LTTng trace, which is a http://diamon.org/ctf[CTF] trace, has both
7044 data stream files and a metadata file. This metadata file contains,
7045 amongst other things, information about the offset of the clock sources
7046 used to timestamp <<event,event records>> when tracing.
7048 If, once a <<tracing-session,tracing session>> is
7049 <<basic-tracing-session-control,started>>, a major
7050 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
7051 happens, the trace's clock offset also needs to be updated. Use
7052 the `metadata` item of the man:lttng-regenerate(1) command to do so.
7054 The main use case of this command is to allow a system to boot with
7055 an incorrect wall time and trace it with LTTng before its wall time
7056 is corrected. Once the system is known to be in a state where its
7057 wall time is correct, it can run `lttng regenerate metadata`.
7059 To regenerate the metadata of an LTTng trace:
7061 * Use the `metadata` item of the man:lttng-regenerate(1) command:
7066 $ lttng regenerate metadata
7072 `lttng regenerate metadata` has the following limitations:
7074 * Tracing session <<creating-destroying-tracing-sessions,created>>
7076 * User space <<channel,channels>>, if any, are using
7077 <<channel-buffering-schemes,per-user buffering>>.
7082 [[regenerate-statedump]]
7083 === Regenerate the state dump of a tracing session
7085 The LTTng kernel and user space tracers generate state dump
7086 <<event,event records>> when the application starts or when you
7087 <<basic-tracing-session-control,start a tracing session>>. An analysis
7088 can use the state dump event records to set an initial state before it
7089 builds the rest of the state from the following event records.
7090 http://tracecompass.org/[Trace Compass] is a notable example of an
7091 application which uses the state dump of an LTTng trace.
7093 When you <<taking-a-snapshot,take a snapshot>>, it's possible that the
7094 state dump event records aren't included in the snapshot because they
7095 were recorded to a sub-buffer that has been consumed or overwritten
7098 Use the `lttng regenerate statedump` command to emit the state
7099 dump event records again.
7101 To regenerate the state dump of the current tracing session, provided
7102 create it in snapshot mode, before you take a snapshot:
7104 . Use the `statedump` item of the man:lttng-regenerate(1) command:
7109 $ lttng regenerate statedump
7113 . <<basic-tracing-session-control,Stop the tracing session>>:
7122 . <<taking-a-snapshot,Take a snapshot>>:
7127 $ lttng snapshot record --name=my-snapshot
7131 Depending on the event throughput, you should run steps 1 and 2
7132 as closely as possible.
7134 NOTE: To record the state dump events, you need to
7135 <<enabling-disabling-events,create event rules>> which enable them.
7136 LTTng-UST state dump tracepoints start with `lttng_ust_statedump:`.
7137 LTTng-modules state dump tracepoints start with `lttng_statedump_`.
7141 [[persistent-memory-file-systems]]
7142 === Record trace data on persistent memory file systems
7144 https://en.wikipedia.org/wiki/Non-volatile_random-access_memory[Non-volatile random-access memory]
7145 (NVRAM) is random-access memory that retains its information when power
7146 is turned off (non-volatile). Systems with such memory can store data
7147 structures in RAM and retrieve them after a reboot, without flushing
7148 to typical _storage_.
7150 Linux supports NVRAM file systems thanks to either
7151 http://pramfs.sourceforge.net/[PRAMFS] or
7152 https://www.kernel.org/doc/Documentation/filesystems/dax.txt[DAX]{nbsp}+{nbsp}http://lkml.iu.edu/hypermail/linux/kernel/1504.1/03463.html[pmem]
7153 (requires Linux{nbsp}4.1+).
7155 This section doesn't describe how to operate such file systems;
7156 we assume that you have a working persistent memory file system.
7158 When you create a <<tracing-session,tracing session>>, you can specify
7159 the path of the shared memory holding the sub-buffers. If you specify a
7160 location on an NVRAM file system, then you can retrieve the latest
7161 recorded trace data when the system reboots after a crash.
7163 To record trace data on a persistent memory file system and retrieve the
7164 trace data after a system crash:
7166 . Create a tracing session with a sub-buffer shared memory path located
7167 on an NVRAM file system:
7172 $ lttng create my-session --shm-path=/path/to/shm
7176 . Configure the tracing session as usual with the man:lttng(1)
7177 command-line tool, and <<basic-tracing-session-control,start tracing>>.
7179 . After a system crash, use the man:lttng-crash(1) command-line tool to
7180 view the trace data recorded on the NVRAM file system:
7185 $ lttng-crash /path/to/shm
7189 The binary layout of the ring buffer files isn't exactly the same as
7190 the trace files layout. This is why you need to use man:lttng-crash(1)
7191 instead of your preferred trace viewer directly.
7193 To convert the ring buffer files to LTTng trace files:
7195 * Use the opt:lttng-crash(1):--extract option of man:lttng-crash(1):
7200 $ lttng-crash --extract=/path/to/trace /path/to/shm
7206 [[notif-trigger-api]]
7207 === Get notified when a channel's buffer usage is too high or too low
7209 With LTTng's $$C/C++$$ notification and trigger API, your user
7210 application can get notified when the buffer usage of one or more
7211 <<channel,channels>> becomes too low or too high. Use this API
7212 and enable or disable <<event,event rules>> during tracing to avoid
7213 <<channel-overwrite-mode-vs-discard-mode,discarded event records>>.
7215 .Have a user application get notified when an LTTng channel's buffer usage is too high.
7217 In this example, we create and build an application which gets notified
7218 when the buffer usage of a specific LTTng channel is higher than
7219 75{nbsp}%. We only print that it is the case in the example, but we
7220 could as well use the API of <<liblttng-ctl-lttng,`liblttng-ctl`>> to
7221 disable event rules when this happens.
7223 . Create the application's C source file:
7231 #include <lttng/domain.h>
7232 #include <lttng/action/action.h>
7233 #include <lttng/action/notify.h>
7234 #include <lttng/condition/condition.h>
7235 #include <lttng/condition/buffer-usage.h>
7236 #include <lttng/condition/evaluation.h>
7237 #include <lttng/notification/channel.h>
7238 #include <lttng/notification/notification.h>
7239 #include <lttng/trigger/trigger.h>
7240 #include <lttng/endpoint.h>
7242 int main(int argc, char *argv[])
7244 int exit_status = 0;
7245 struct lttng_notification_channel *notification_channel;
7246 struct lttng_condition *condition;
7247 struct lttng_action *action;
7248 struct lttng_trigger *trigger;
7249 const char *tracing_session_name;
7250 const char *channel_name;
7253 tracing_session_name = argv[1];
7254 channel_name = argv[2];
7257 * Create a notification channel. A notification channel
7258 * connects the user application to the LTTng session daemon.
7259 * This notification channel can be used to listen to various
7260 * types of notifications.
7262 notification_channel = lttng_notification_channel_create(
7263 lttng_session_daemon_notification_endpoint);
7266 * Create a "high buffer usage" condition. In this case, the
7267 * condition is reached when the buffer usage is greater than or
7268 * equal to 75 %. We create the condition for a specific tracing
7269 * session name, channel name, and for the user space tracing
7272 * The "low buffer usage" condition type also exists.
7274 condition = lttng_condition_buffer_usage_high_create();
7275 lttng_condition_buffer_usage_set_threshold_ratio(condition, .75);
7276 lttng_condition_buffer_usage_set_session_name(
7277 condition, tracing_session_name);
7278 lttng_condition_buffer_usage_set_channel_name(condition,
7280 lttng_condition_buffer_usage_set_domain_type(condition,
7284 * Create an action (get a notification) to take when the
7285 * condition created above is reached.
7287 action = lttng_action_notify_create();
7290 * Create a trigger. A trigger associates a condition to an
7291 * action: the action is executed when the condition is reached.
7293 trigger = lttng_trigger_create(condition, action);
7295 /* Register the trigger to LTTng. */
7296 lttng_register_trigger(trigger);
7299 * Now that we have registered a trigger, a notification will be
7300 * emitted everytime its condition is met. To receive this
7301 * notification, we must subscribe to notifications that match
7302 * the same condition.
7304 lttng_notification_channel_subscribe(notification_channel,
7308 * Notification loop. Put this in a dedicated thread to avoid
7309 * blocking the main thread.
7312 struct lttng_notification *notification;
7313 enum lttng_notification_channel_status status;
7314 const struct lttng_evaluation *notification_evaluation;
7315 const struct lttng_condition *notification_condition;
7316 double buffer_usage;
7318 /* Receive the next notification. */
7319 status = lttng_notification_channel_get_next_notification(
7320 notification_channel, ¬ification);
7323 case LTTNG_NOTIFICATION_CHANNEL_STATUS_OK:
7325 case LTTNG_NOTIFICATION_CHANNEL_STATUS_NOTIFICATIONS_DROPPED:
7327 * The session daemon can drop notifications if a monitoring
7328 * application isn't consuming the notifications fast
7332 case LTTNG_NOTIFICATION_CHANNEL_STATUS_CLOSED:
7334 * The notification channel has been closed by the
7335 * session daemon. This is typically caused by a session
7336 * daemon shutting down.
7340 /* Unhandled conditions or errors. */
7346 * A notification provides, amongst other things:
7348 * * The condition that caused this notification to be
7350 * * The condition evaluation, which provides more
7351 * specific information on the evaluation of the
7354 * The condition evaluation provides the buffer usage
7355 * value at the moment the condition was reached.
7357 notification_condition = lttng_notification_get_condition(
7359 notification_evaluation = lttng_notification_get_evaluation(
7362 /* We're subscribed to only one condition. */
7363 assert(lttng_condition_get_type(notification_condition) ==
7364 LTTNG_CONDITION_TYPE_BUFFER_USAGE_HIGH);
7367 * Get the exact sampled buffer usage from the
7368 * condition evaluation.
7370 lttng_evaluation_buffer_usage_get_usage_ratio(
7371 notification_evaluation, &buffer_usage);
7374 * At this point, instead of printing a message, we
7375 * could do something to reduce the channel's buffer
7376 * usage, like disable specific events.
7378 printf("Buffer usage is %f %% in tracing session \"%s\", "
7379 "user space channel \"%s\".\n", buffer_usage * 100,
7380 tracing_session_name, channel_name);
7381 lttng_notification_destroy(notification);
7385 lttng_action_destroy(action);
7386 lttng_condition_destroy(condition);
7387 lttng_trigger_destroy(trigger);
7388 lttng_notification_channel_destroy(notification_channel);
7394 . Build the `notif-app` application, linking it to `liblttng-ctl`:
7399 $ gcc -o notif-app notif-app.c -llttng-ctl
7403 . <<creating-destroying-tracing-sessions,Create a tracing session>>,
7404 <<enabling-disabling-events,create an event rule>> matching all the
7405 user space tracepoints, and
7406 <<basic-tracing-session-control,start tracing>>:
7411 $ lttng create my-session
7412 $ lttng enable-event --userspace --all
7417 If you create the channel manually with the man:lttng-enable-channel(1)
7418 command, control how frequently are the current values of the
7419 channel's properties sampled to evaluate user conditions with the
7420 opt:lttng-enable-channel(1):--monitor-timer option.
7422 . Run the `notif-app` application. This program accepts the
7423 <<tracing-session,tracing session>> name and the user space channel
7424 name as its two first arguments. The channel which LTTng automatically
7425 creates with the man:lttng-enable-event(1) command above is named
7431 $ ./notif-app my-session channel0
7435 . In another terminal, run an application with a very high event
7436 throughput so that the 75{nbsp}% buffer usage condition is reached.
7438 In the first terminal, the application should print lines like this:
7441 Buffer usage is 81.45197 % in tracing session "my-session", user space
7445 If you don't see anything, try modifying the condition in
7446 path:{notif-app.c} to a lower value (0.1, for example), rebuilding it
7447 (step{nbsp}2) and running it again (step{nbsp}4).
7454 [[lttng-modules-ref]]
7455 === noch:{LTTng-modules}
7459 [[lttng-tracepoint-enum]]
7460 ==== `LTTNG_TRACEPOINT_ENUM()` usage
7462 Use the `LTTNG_TRACEPOINT_ENUM()` macro to define an enumeration:
7466 LTTNG_TRACEPOINT_ENUM(name, TP_ENUM_VALUES(entries))
7471 * `name` with the name of the enumeration (C identifier, unique
7472 amongst all the defined enumerations).
7473 * `entries` with a list of enumeration entries.
7475 The available enumeration entry macros are:
7477 +ctf_enum_value(__name__, __value__)+::
7478 Entry named +__name__+ mapped to the integral value +__value__+.
7480 +ctf_enum_range(__name__, __begin__, __end__)+::
7481 Entry named +__name__+ mapped to the range of integral values between
7482 +__begin__+ (included) and +__end__+ (included).
7484 +ctf_enum_auto(__name__)+::
7485 Entry named +__name__+ mapped to the integral value following the
7486 last mapping's value.
7488 The last value of a `ctf_enum_value()` entry is its +__value__+
7491 The last value of a `ctf_enum_range()` entry is its +__end__+ parameter.
7493 If `ctf_enum_auto()` is the first entry in the list, its integral
7496 Use the `ctf_enum()` <<lttng-modules-tp-fields,field definition macro>>
7497 to use a defined enumeration as a tracepoint field.
7499 .Define an enumeration with `LTTNG_TRACEPOINT_ENUM()`.
7503 LTTNG_TRACEPOINT_ENUM(
7506 ctf_enum_auto("AUTO: EXPECT 0")
7507 ctf_enum_value("VALUE: 23", 23)
7508 ctf_enum_value("VALUE: 27", 27)
7509 ctf_enum_auto("AUTO: EXPECT 28")
7510 ctf_enum_range("RANGE: 101 TO 303", 101, 303)
7511 ctf_enum_auto("AUTO: EXPECT 304")
7519 [[lttng-modules-tp-fields]]
7520 ==== Tracepoint fields macros (for `TP_FIELDS()`)
7522 [[tp-fast-assign]][[tp-struct-entry]]The available macros to define
7523 tracepoint fields, which must be listed within `TP_FIELDS()` in
7524 `LTTNG_TRACEPOINT_EVENT()`, are:
7526 [role="func-desc growable",cols="asciidoc,asciidoc"]
7527 .Available macros to define LTTng-modules tracepoint fields
7529 |Macro |Description and parameters
7532 +ctf_integer(__t__, __n__, __e__)+
7534 +ctf_integer_nowrite(__t__, __n__, __e__)+
7536 +ctf_user_integer(__t__, __n__, __e__)+
7538 +ctf_user_integer_nowrite(__t__, __n__, __e__)+
7540 Standard integer, displayed in base{nbsp}10.
7543 Integer C type (`int`, `long`, `size_t`, ...).
7549 Argument expression.
7552 +ctf_integer_hex(__t__, __n__, __e__)+
7554 +ctf_user_integer_hex(__t__, __n__, __e__)+
7556 Standard integer, displayed in base{nbsp}16.
7565 Argument expression.
7567 |+ctf_integer_oct(__t__, __n__, __e__)+
7569 Standard integer, displayed in base{nbsp}8.
7578 Argument expression.
7581 +ctf_integer_network(__t__, __n__, __e__)+
7583 +ctf_user_integer_network(__t__, __n__, __e__)+
7585 Integer in network byte order (big-endian), displayed in base{nbsp}10.
7594 Argument expression.
7597 +ctf_integer_network_hex(__t__, __n__, __e__)+
7599 +ctf_user_integer_network_hex(__t__, __n__, __e__)+
7601 Integer in network byte order, displayed in base{nbsp}16.
7610 Argument expression.
7613 +ctf_enum(__N__, __t__, __n__, __e__)+
7615 +ctf_enum_nowrite(__N__, __t__, __n__, __e__)+
7617 +ctf_user_enum(__N__, __t__, __n__, __e__)+
7619 +ctf_user_enum_nowrite(__N__, __t__, __n__, __e__)+
7624 Name of a <<lttng-tracepoint-enum,previously defined enumeration>>.
7627 Integer C type (`int`, `long`, `size_t`, ...).
7633 Argument expression.
7636 +ctf_string(__n__, __e__)+
7638 +ctf_string_nowrite(__n__, __e__)+
7640 +ctf_user_string(__n__, __e__)+
7642 +ctf_user_string_nowrite(__n__, __e__)+
7644 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
7650 Argument expression.
7653 +ctf_array(__t__, __n__, __e__, __s__)+
7655 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
7657 +ctf_user_array(__t__, __n__, __e__, __s__)+
7659 +ctf_user_array_nowrite(__t__, __n__, __e__, __s__)+
7661 Statically-sized array of integers.
7664 Array element C type.
7670 Argument expression.
7676 +ctf_array_bitfield(__t__, __n__, __e__, __s__)+
7678 +ctf_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7680 +ctf_user_array_bitfield(__t__, __n__, __e__, __s__)+
7682 +ctf_user_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7684 Statically-sized array of bits.
7686 The type of +__e__+ must be an integer type. +__s__+ is the number
7687 of elements of such type in +__e__+, not the number of bits.
7690 Array element C type.
7696 Argument expression.
7702 +ctf_array_text(__t__, __n__, __e__, __s__)+
7704 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
7706 +ctf_user_array_text(__t__, __n__, __e__, __s__)+
7708 +ctf_user_array_text_nowrite(__t__, __n__, __e__, __s__)+
7710 Statically-sized array, printed as text.
7712 The string doesn't need to be null-terminated.
7715 Array element C type (always `char`).
7721 Argument expression.
7727 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
7729 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7731 +ctf_user_sequence(__t__, __n__, __e__, __T__, __E__)+
7733 +ctf_user_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7735 Dynamically-sized array of integers.
7737 The type of +__E__+ must be unsigned.
7740 Array element C type.
7746 Argument expression.
7749 Length expression C type.
7755 +ctf_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7757 +ctf_user_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7759 Dynamically-sized array of integers, displayed in base{nbsp}16.
7761 The type of +__E__+ must be unsigned.
7764 Array element C type.
7770 Argument expression.
7773 Length expression C type.
7778 |+ctf_sequence_network(__t__, __n__, __e__, __T__, __E__)+
7780 Dynamically-sized array of integers in network byte order (big-endian),
7781 displayed in base{nbsp}10.
7783 The type of +__E__+ must be unsigned.
7786 Array element C type.
7792 Argument expression.
7795 Length expression C type.
7801 +ctf_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7803 +ctf_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7805 +ctf_user_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7807 +ctf_user_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7809 Dynamically-sized array of bits.
7811 The type of +__e__+ must be an integer type. +__s__+ is the number
7812 of elements of such type in +__e__+, not the number of bits.
7814 The type of +__E__+ must be unsigned.
7817 Array element C type.
7823 Argument expression.
7826 Length expression C type.
7832 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7834 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7836 +ctf_user_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7838 +ctf_user_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7840 Dynamically-sized array, displayed as text.
7842 The string doesn't need to be null-terminated.
7844 The type of +__E__+ must be unsigned.
7846 The behaviour is undefined if +__e__+ is `NULL`.
7849 Sequence element C type (always `char`).
7855 Argument expression.
7858 Length expression C type.
7864 Use the `_user` versions when the argument expression, `e`, is
7865 a user space address. In the cases of `ctf_user_integer*()` and
7866 `ctf_user_float*()`, `&e` must be a user space address, thus `e` must
7869 The `_nowrite` versions omit themselves from the session trace, but are
7870 otherwise identical. This means the `_nowrite` fields won't be written
7871 in the recorded trace. Their primary purpose is to make some
7872 of the event context available to the
7873 <<enabling-disabling-events,event filters>> without having to
7874 commit the data to sub-buffers.
7880 Terms related to LTTng and to tracing in general:
7883 The http://diamon.org/babeltrace[Babeltrace] project, which includes:
7886 https://babeltrace.org/docs/v2.0/man1/babeltrace2.1/[cmd:babeltrace2]
7887 command-line interface.
7888 * The libbabeltrace2 library which offers a
7889 https://babeltrace.org/docs/v2.0/libbabeltrace2/[C API].
7890 * https://babeltrace.org/docs/v2.0/python/bt2/[Python{nbsp}3 bindings].
7893 [[def-buffering-scheme]]<<channel-buffering-schemes,buffering scheme>>::
7894 A layout of <<def-sub-buffer,sub-buffers>> applied to a given channel.
7896 [[def-channel]]<<channel,channel>>::
7897 An entity which is responsible for a set of
7898 <<def-ring-buffer,ring buffers>>.
7900 <<def-event-rule,Event rules>> are always attached to a specific
7904 A source of time for a <<def-tracer,tracer>>.
7906 [[def-consumer-daemon]]<<lttng-consumerd,consumer daemon>>::
7907 A process which is responsible for consuming the full
7908 <<def-sub-buffer,sub-buffers>> and write them to a file system or
7909 send them over the network.
7911 [[def-current-trace-chunk]]current trace chunk::
7912 A <<def-trace-chunk,trace chunk>> which includes the current content
7913 of all the <<def-tracing-session-rotation,tracing session>>'s
7914 <<def-sub-buffer,sub-buffers>> and the stream files produced since the
7915 latest event amongst:
7917 * The creation of the <<def-tracing-session,tracing session>>.
7918 * The last tracing session rotation, if any.
7920 <<channel-overwrite-mode-vs-discard-mode,discard mode>>::
7921 The <<def-event-record-loss-mode,event record loss mode>> in which
7922 the <<def-tracer,tracer>> _discards_ new event records when there's no
7923 <<def-sub-buffer,sub-buffer>> space left to store them.
7925 [[def-event]]event::
7926 The consequence of the execution of an
7927 <<def-instrumentation-point,instrumentation point>>, like a
7928 <<def-tracepoint,tracepoint>> that you manually place in some source
7929 code, or a Linux kernel kprobe.
7931 An event is said to _occur_ at a specific time. <<def-lttng,LTTng>> can
7932 take various actions upon the occurrence of an event, like record the
7933 event's payload to a <<def-sub-buffer,sub-buffer>>.
7935 [[def-event-name]]event name::
7936 The name of an <<def-event,event>>, which is also the name of the
7937 <<def-event-record,event record>>.
7939 This is also called the _instrumentation point name_.
7941 [[def-event-record]]event record::
7942 A record, in a <<def-trace,trace>>, of the payload of an
7943 <<def-event,event>> which occured.
7945 [[def-event-record-loss-mode]]<<channel-overwrite-mode-vs-discard-mode,event record loss mode>>::
7946 The mechanism by which event records of a given
7947 <<def-channel,channel>> are lost (not recorded) when there is no
7948 <<def-sub-buffer,sub-buffer>> space left to store them.
7950 [[def-event-rule]]<<event,event rule>>::
7951 Set of conditions which must be satisfied for one or more occuring
7952 <<def-event,events>> to be recorded.
7954 [[def-incl-set]]inclusion set::
7955 In the <<pid-tracking,process attribute tracking>> context: a
7956 set of <<def-proc-attr,process attributes>> of a given type.
7958 <<instrumenting,instrumentation>>::
7959 The use of <<def-lttng,LTTng>> probes to make a piece of software
7962 [[def-instrumentation-point]]instrumentation point::
7963 A point in the execution path of a piece of software that, when
7964 reached by this execution, can emit an <<def-event,event>>.
7966 instrumentation point name::
7967 See _<<def-event-name,event name>>_.
7969 `java.util.logging`::
7971 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[core logging facilities].
7974 A http://logging.apache.org/log4j/1.2/[logging library] for Java
7975 developed by the Apache Software Foundation.
7978 Level of severity of a log statement or user space
7979 <<def-instrumentation-point,instrumentation point>>.
7981 [[def-lttng]]LTTng::
7982 The _Linux Trace Toolkit: next generation_ project.
7984 <<lttng-cli,cmd:lttng>>::
7985 A command-line tool provided by the <<def-lttng-tools,LTTng-tools>>
7986 project which you can use to send and receive control messages to and
7987 from a <<def-session-daemon,session daemon>>.
7990 The https://github.com/lttng/lttng-analyses[LTTng analyses] project,
7991 which is a set of analyzing programs that you can use to obtain a
7992 higher level view of an <<def-lttng,LTTng>> <<def-trace,trace>>.
7994 cmd:lttng-consumerd::
7995 The name of the <<def-consumer-daemon,consumer daemon>> program.
7998 A utility provided by the <<def-lttng-tools,LTTng-tools>> project
7999 which can convert <<def-ring-buffer,ring buffer>> files (usually
8000 <<persistent-memory-file-systems,saved on a persistent memory file
8001 system>>) to <<def-trace,trace>> files.
8003 See man:lttng-crash(1).
8005 LTTng Documentation::
8008 <<lttng-live,LTTng live>>::
8009 A communication protocol between the <<lttng-relayd,relay daemon>> and
8010 live viewers which makes it possible to see <<def-event-record,event
8011 records>> ``live'', as they are received by the
8012 <<def-relay-daemon,relay daemon>>.
8014 <<lttng-modules,LTTng-modules>>::
8015 The https://github.com/lttng/lttng-modules[LTTng-modules] project,
8016 which contains the Linux kernel modules to make the Linux kernel
8017 <<def-instrumentation-point,instrumentation points>> available for
8018 <<def-lttng,LTTng>> tracing.
8021 The name of the <<def-relay-daemon,relay daemon>> program.
8023 cmd:lttng-sessiond::
8024 The name of the <<def-session-daemon,session daemon>> program.
8026 [[def-lttng-tools]]LTTng-tools::
8027 The https://github.com/lttng/lttng-tools[LTTng-tools] project, which
8028 contains the various programs and libraries used to
8029 <<controlling-tracing,control tracing>>.
8031 [[def-lttng-ust]]<<lttng-ust,LTTng-UST>>::
8032 The https://github.com/lttng/lttng-ust[LTTng-UST] project, which
8033 contains libraries to instrument
8034 <<def-user-application,user applications>>.
8036 <<lttng-ust-agents,LTTng-UST Java agent>>::
8037 A Java package provided by the <<def-lttng-ust,LTTng-UST>> project to
8038 allow the LTTng instrumentation of `java.util.logging` and Apache
8039 log4j{nbsp}1.2 logging statements.
8041 <<lttng-ust-agents,LTTng-UST Python agent>>::
8042 A Python package provided by the <<def-lttng-ust,LTTng-UST>> project
8043 to allow the <<def-lttng,LTTng>> instrumentation of Python logging
8046 <<channel-overwrite-mode-vs-discard-mode,overwrite mode>>::
8047 The <<def-event-record-loss-mode,event record loss mode>> in which new
8048 <<def-event-record,event records>> _overwrite_ older event records
8049 when there's no <<def-sub-buffer,sub-buffer>> space left to store
8052 <<channel-buffering-schemes,per-process buffering>>::
8053 A <<def-buffering-scheme,buffering scheme>> in which each instrumented
8054 process has its own <<def-sub-buffer,sub-buffers>> for a given user
8055 space <<def-channel,channel>>.
8057 <<channel-buffering-schemes,per-user buffering>>::
8058 A <<def-buffering-scheme,buffering scheme>> in which all the processes
8059 of a Unix user share the same <<def-sub-buffer,sub-buffers>> for a
8060 given user space <<def-channel,channel>>.
8062 [[def-proc-attr]]process attribute::
8063 In the <<pid-tracking,process attribute tracking>> context:
8066 * A virtual process ID.
8068 * A virtual Unix user ID.
8070 * A virtual Unix group ID.
8072 [[def-relay-daemon]]<<lttng-relayd,relay daemon>>::
8073 A process which is responsible for receiving the <<def-trace,trace>>
8074 data which a distant <<def-consumer-daemon,consumer daemon>> sends.
8076 [[def-ring-buffer]]ring buffer::
8077 A set of <<def-sub-buffer,sub-buffers>>.
8080 See _<<def-tracing-session-rotation,tracing session rotation>>_.
8082 [[def-session-daemon]]<<lttng-sessiond,session daemon>>::
8083 A process which receives control commands from you and orchestrates
8084 the <<def-tracer,tracers>> and various <<def-lttng,LTTng>> daemons.
8086 <<taking-a-snapshot,snapshot>>::
8087 A copy of the current data of all the <<def-sub-buffer,sub-buffers>>
8088 of a given <<def-tracing-session,tracing session>>, saved as
8089 <<def-trace,trace>> files.
8091 [[def-sub-buffer]]sub-buffer::
8092 One part of an <<def-lttng,LTTng>> <<def-ring-buffer,ring buffer>>
8093 which contains <<def-event-record,event records>>.
8096 The time information attached to an <<def-event,event>> when it is
8099 [[def-trace]]trace (_noun_)::
8102 * One http://diamon.org/ctf/[CTF] metadata stream file.
8103 * One or more CTF data stream files which are the concatenations of one
8104 or more flushed <<def-sub-buffer,sub-buffers>>.
8106 [[def-trace-verb]]trace (_verb_)::
8107 The action of recording the <<def-event,events>> emitted by an
8108 application or by a system, or to initiate such recording by
8109 controlling a <<def-tracer,tracer>>.
8111 [[def-trace-chunk]]trace chunk::
8112 A self-contained <<def-trace,trace>> which is part of a
8113 <<def-tracing-session,tracing session>>. Each
8114 <<def-tracing-session-rotation, tracing session rotation>> produces a
8115 <<def-trace-chunk-archive,trace chunk archive>>.
8117 [[def-trace-chunk-archive]]trace chunk archive::
8118 The result of a <<def-tracing-session-rotation, tracing session rotation>>.
8120 <<def-lttng,LTTng>> doesn't manage any trace chunk archive, even if its
8121 containing <<def-tracing-session,tracing session>> is still active: you
8122 are free to read it, modify it, move it, or remove it.
8125 The http://tracecompass.org[Trace Compass] project and application.
8127 [[def-tracepoint]]tracepoint::
8128 An instrumentation point using the tracepoint mechanism of the Linux
8129 kernel or of <<def-lttng-ust,LTTng-UST>>.
8131 tracepoint definition::
8132 The definition of a single <<def-tracepoint,tracepoint>>.
8135 The name of a <<def-tracepoint,tracepoint>>.
8137 [[def-tracepoint-provider]]tracepoint provider::
8138 A set of functions providing <<def-tracepoint,tracepoints>> to an
8139 instrumented <<def-user-application,user application>>.
8141 Not to be confused with a <<def-tracepoint-provider-package,tracepoint
8142 provider package>>: many tracepoint providers can exist within a
8143 tracepoint provider package.
8145 [[def-tracepoint-provider-package]]tracepoint provider package::
8146 One or more <<def-tracepoint-provider,tracepoint providers>> compiled
8147 as an https://en.wikipedia.org/wiki/Object_file[object file] or as a
8148 link:https://en.wikipedia.org/wiki/Library_(computing)#Shared_libraries[shared
8151 [[def-tracer]]tracer::
8152 A software which records emitted <<def-event,events>>.
8154 <<domain,tracing domain>>::
8155 A namespace for <<def-event,event>> sources.
8157 <<tracing-group,tracing group>>::
8158 The Unix group in which a Unix user can be to be allowed to
8159 <<def-trace-verb,trace>> the Linux kernel.
8161 [[def-tracing-session]]<<tracing-session,tracing session>>::
8162 A stateful dialogue between you and a <<lttng-sessiond,session daemon>>.
8164 [[def-tracing-session-rotation]]<<session-rotation,tracing session rotation>>::
8165 The action of archiving the
8166 <<def-current-trace-chunk,current trace chunk>> of a
8167 <<def-tracing-session,tracing session>>.
8169 tracked <<def-proc-attr,process attribute>>::
8170 A process attribute which is part of an <<def-incl-set,inclusion
8173 untracked process attribute::
8174 A process attribute which isn't part of an <<def-incl-set,inclusion
8177 [[def-user-application]]user application::
8178 An application running in user space, as opposed to a Linux kernel
8179 module, for example.