1 The LTTng Documentation
2 =======================
3 Philippe Proulx <pproulx@efficios.com>
7 include::../common/copyright.txt[]
10 include::../common/welcome.txt[]
13 include::../common/audience.txt[]
17 === What's in this documentation?
19 The LTTng Documentation is divided into the following sections:
21 * **<<nuts-and-bolts,Nuts and bolts>>** explains the
22 rudiments of software tracing and the rationale behind the
25 You can skip this section if you’re familiar with software tracing and
26 with the LTTng project.
28 * **<<installing-lttng,Installation>>** describes the steps to
29 install the LTTng packages on common Linux distributions and from
32 You can skip this section if you already properly installed LTTng on
35 * **<<getting-started,Quick start>>** is a concise guide to
36 getting started quickly with LTTng kernel and user space tracing.
38 We recommend this section if you're new to LTTng or to software tracing
41 You can skip this section if you're not new to LTTng.
43 * **<<core-concepts,Core concepts>>** explains the concepts at
46 It's a good idea to become familiar with the core concepts
47 before attempting to use the toolkit.
49 * **<<plumbing,Components of LTTng>>** describes the various components
50 of the LTTng machinery, like the daemons, the libraries, and the
51 command-line interface.
52 * **<<instrumenting,Instrumentation>>** shows different ways to
53 instrument user applications and the Linux kernel.
55 Instrumenting source code is essential to provide a meaningful
58 You can skip this section if you do not have a programming background.
60 * **<<controlling-tracing,Tracing control>>** is divided into topics
61 which demonstrate how to use the vast array of features that
62 LTTng{nbsp}{revision} offers.
63 * **<<reference,Reference>>** contains reference tables.
64 * **<<glossary,Glossary>>** is a specialized dictionary of terms related
65 to LTTng or to the field of software tracing.
68 include::../common/convention.txt[]
71 include::../common/acknowledgements.txt[]
75 == What's new in LTTng{nbsp}{revision}?
77 LTTng{nbsp}{revision} bears the name _Lafontaine_. This modern
78 https://en.wikipedia.org/wiki/saison[saison] from the
79 https://oshlag.com/[Oshlag] microbrewery is a refreshing--zesty--rice
80 beer with hints of fruit and spices. Some even say it makes for a great
81 https://en.wikipedia.org/wiki/Somaek[Somaek] when mixed with
82 Chamisul Soju, not that we've tried!
84 New features and changes in LTTng{nbsp}{revision}:
86 * Just like you can typically perform
87 https://en.wikipedia.org/wiki/Log_rotation[log rotation], you can
88 now <<session-rotation,_rotate_ a tracing session>>, that
89 is, according to man:lttng-rotate(1), archive the current trace
90 chunk (all the tracing session's trace data since the last rotation
91 or since its inception) so that LTTng does not manage it anymore.
93 Once LTTng archives a trace chunk, you are free to read it, modify it,
94 move it, or remove it.
96 You can rotate a tracing session immediately or set a rotation schedule
97 to automate rotations.
99 * When you <<enabling-disabling-events,create an event rule>>, the
100 filter expression syntax now supports the following new operators:
104 ** `<<` (bitwise left shift)
105 ** `>>` (bitwise right shift)
111 The syntax also supports array indexing with the usual square brackets:
114 regs[3][1] & 0xff7 == 0x240
117 There are peculiarities for both the new operators and the array
118 indexing brackets, like a custom precedence table and implicit casting.
119 See man:lttng-enable-event(1) to get all the details about the filter
122 * You can now dynamically instrument any application's or library's
123 function entry by symbol name thanks to the new
124 opt:lttng-enable-event(1):--userspace-probe option of
125 the `lttng enable-event` command:
129 $ lttng enable-event --kernel \
130 --userspace-probe=/usr/lib/libc.so.6:malloc libc_malloc
133 The option also supports tracing existing
134 https://www.sourceware.org/systemtap/wiki/AddingUserSpaceProbingToApps[SystemTap
135 Statically Defined Tracing] (USDT) probe (DTrace-style marker). For
136 example, given the following probe:
140 DTRACE_PROBE2("server", "accept-request", request_id, ip_addr);
143 You can trace this probe with:
146 $ lttng enable-event --kernel \
147 --userspace-probe=sdt:/path/to/server:server:accept-request \
148 server_accept_request
151 This feature makes use of Linux's
152 https://www.kernel.org/doc/Documentation/trace/uprobetracer.txt[uprobe]
153 mechanism, therefore you must use the `--userspace-probe`
154 instrumentation option with the opt:lttng-enable-event(1):--kernel
157 NOTE: As of LTTng{nbsp}{revision}, LTTng does not record function
158 parameters with the opt:lttng-enable-event(1):--userspace-probe option.
160 * Two new <<adding-context,context>> fields are available for Linux
161 kernel <<channel,channels>>:
164 ** `callstack-kernel`
168 Thanks to those, you can record the Linux kernel and user call stacks
169 when a kernel event occurs. For example:
173 $ lttng enable-event --kernel --syscall read
174 $ lttng add-context --kernel --type=callstack-kernel --type=callstack-user
177 When an man:open(2) system call occurs, LTTng attaches the kernel and
178 user call stacks to the recorded event.
180 NOTE: LTTng cannot always sample the user space call stack reliably.
181 For instance, LTTng cannot sample the call stack of user applications
182 and libraries compiled with the
183 https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html[`-fomit-frame-pointer`]
184 option. In such a case, the tracing is not affected, but the sampled
185 user space call stack may only contain the user call stack's topmost
188 * User applications and libraries instrumented with
189 <<lttng-ust,LTTng-UST>> can now safely unload (man:dlclose(3)) a
191 <<building-tracepoint-providers-and-user-application,tracepoint
194 * The <<lttng-relayd,relay daemon>> is more efficient and presents fewer
195 connectivity issues, especially when a large number of targets send
196 trace data to a given relay daemon.
198 * LTTng-UST uses https://github.com/numactl/numactl[libnuma]
199 when available to allocate <<def-sub-buffer,sub-buffers>>, making them
201 https://en.wikipedia.org/wiki/Non-uniform_memory_access[NUMA] node.
203 This change makes the tracer more efficient on NUMA systems.
209 What is LTTng? As its name suggests, the _Linux Trace Toolkit: next
210 generation_ is a modern toolkit for tracing Linux systems and
211 applications. So your first question might be:
218 As the history of software engineering progressed and led to what
219 we now take for granted--complex, numerous and
220 interdependent software applications running in parallel on
221 sophisticated operating systems like Linux--the authors of such
222 components, software developers, began feeling a natural
223 urge to have tools that would ensure the robustness and good performance
224 of their masterpieces.
226 One major achievement in this field is, inarguably, the
227 https://www.gnu.org/software/gdb/[GNU debugger (GDB)],
228 an essential tool for developers to find and fix bugs. But even the best
229 debugger won't help make your software run faster, and nowadays, faster
230 software means either more work done by the same hardware, or cheaper
231 hardware for the same work.
233 A _profiler_ is often the tool of choice to identify performance
234 bottlenecks. Profiling is suitable to identify _where_ performance is
235 lost in a given software. The profiler outputs a profile, a statistical
236 summary of observed events, which you may use to discover which
237 functions took the most time to execute. However, a profiler won't
238 report _why_ some identified functions are the bottleneck. Bottlenecks
239 might only occur when specific conditions are met, conditions that are
240 sometimes impossible to capture by a statistical profiler, or impossible
241 to reproduce with an application altered by the overhead of an
242 event-based profiler. For a thorough investigation of software
243 performance issues, a history of execution is essential, with the
244 recorded values of variables and context fields you choose, and
245 with as little influence as possible on the instrumented software. This
246 is where tracing comes in handy.
248 _Tracing_ is a technique used to understand what goes on in a running
249 software system. The software used for tracing is called a _tracer_,
250 which is conceptually similar to a tape recorder. When recording,
251 specific instrumentation points placed in the software source code
252 generate events that are saved on a giant tape: a _trace_ file. You
253 can trace user applications and the operating system at the same time,
254 opening the possibility of resolving a wide range of problems that would
255 otherwise be extremely challenging.
257 Tracing is often compared to _logging_. However, tracers and loggers are
258 two different tools, serving two different purposes. Tracers are
259 designed to record much lower-level events that occur much more
260 frequently than log messages, often in the range of thousands per
261 second, with very little execution overhead. Logging is more appropriate
262 for a very high-level analysis of less frequent events: user accesses,
263 exceptional conditions (errors and warnings, for example), database
264 transactions, instant messaging communications, and such. Simply put,
265 logging is one of the many use cases that can be satisfied with tracing.
267 The list of recorded events inside a trace file can be read manually
268 like a log file for the maximum level of detail, but it is generally
269 much more interesting to perform application-specific analyses to
270 produce reduced statistics and graphs that are useful to resolve a
271 given problem. Trace viewers and analyzers are specialized tools
274 In the end, this is what LTTng is: a powerful, open source set of
275 tools to trace the Linux kernel and user applications at the same time.
276 LTTng is composed of several components actively maintained and
277 developed by its link:/community/#where[community].
280 [[lttng-alternatives]]
281 === Alternatives to noch:{LTTng}
283 Excluding proprietary solutions, a few competing software tracers
286 * https://github.com/dtrace4linux/linux[dtrace4linux] is a port of
287 Sun Microsystems's DTrace to Linux. The cmd:dtrace tool interprets
288 user scripts and is responsible for loading code into the
289 Linux kernel for further execution and collecting the outputted data.
290 * https://en.wikipedia.org/wiki/Berkeley_Packet_Filter[eBPF] is a
291 subsystem in the Linux kernel in which a virtual machine can execute
292 programs passed from the user space to the kernel. You can attach
293 such programs to tracepoints and kprobes thanks to a system call, and
294 they can output data to the user space when executed thanks to
295 different mechanisms (pipe, VM register values, and eBPF maps, to name
297 * https://www.kernel.org/doc/Documentation/trace/ftrace.txt[ftrace]
298 is the de facto function tracer of the Linux kernel. Its user
299 interface is a set of special files in sysfs.
300 * https://perf.wiki.kernel.org/[perf] is
301 a performance analyzing tool for Linux which supports hardware
302 performance counters, tracepoints, as well as other counters and
303 types of probes. perf's controlling utility is the cmd:perf command
305 * http://linux.die.net/man/1/strace[strace]
306 is a command-line utility which records system calls made by a
307 user process, as well as signal deliveries and changes of process
308 state. strace makes use of https://en.wikipedia.org/wiki/Ptrace[ptrace]
309 to fulfill its function.
310 * http://www.sysdig.org/[sysdig], like SystemTap, uses scripts to
311 analyze Linux kernel events. You write scripts, or _chisels_ in
312 sysdig's jargon, in Lua and sysdig executes them while it traces the
313 system or afterwards. sysdig's interface is the cmd:sysdig
314 command-line tool as well as the curses-based cmd:csysdig tool.
315 * https://sourceware.org/systemtap/[SystemTap] is a Linux kernel and
316 user space tracer which uses custom user scripts to produce plain text
317 traces. SystemTap converts the scripts to the C language, and then
318 compiles them as Linux kernel modules which are loaded to produce
319 trace data. SystemTap's primary user interface is the cmd:stap
322 The main distinctive features of LTTng is that it produces correlated
323 kernel and user space traces, as well as doing so with the lowest
324 overhead amongst other solutions. It produces trace files in the
325 http://diamon.org/ctf[CTF] format, a file format optimized
326 for the production and analyses of multi-gigabyte data.
328 LTTng is the result of more than 10{nbsp}years of active open source
329 development by a community of passionate developers.
330 LTTng{nbsp}{revision} is currently available on major desktop and server
333 The main interface for tracing control is a single command-line tool
334 named cmd:lttng. The latter can create several tracing sessions, enable
335 and disable events on the fly, filter events efficiently with custom
336 user expressions, start and stop tracing, and much more. LTTng can
337 record the traces on the file system or send them over the network, and
338 keep them totally or partially. You can view the traces once tracing
339 becomes inactive or in real-time.
341 <<installing-lttng,Install LTTng now>> and
342 <<getting-started,start tracing>>!
348 **LTTng** is a set of software <<plumbing,components>> which interact to
349 <<instrumenting,instrument>> the Linux kernel and user applications, and
350 to <<controlling-tracing,control tracing>> (start and stop
351 tracing, enable and disable event rules, and the rest). Those
352 components are bundled into the following packages:
355 Libraries and command-line interface to control tracing.
358 Linux kernel modules to instrument and trace the kernel.
361 Libraries and Java/Python packages to instrument and trace user
364 Most distributions mark the LTTng-modules and LTTng-UST packages as
365 optional when installing LTTng-tools (which is always required). In the
366 following sections, we always provide the steps to install all three,
369 * You only need to install LTTng-modules if you intend to trace the
371 * You only need to install LTTng-UST if you intend to trace user
376 As of 18 October 2019, LTTng{nbsp}{revision} is not available
377 as distribution packages.
379 You can <<building-from-source,build LTTng{nbsp}{revision} from source>>
380 to install and use it.
384 [[building-from-source]]
385 === Build from source
387 To build and install LTTng{nbsp}{revision} from source:
389 . Using your distribution's package manager, or from source, install
390 the following dependencies of LTTng-tools and LTTng-UST:
393 * https://sourceforge.net/projects/libuuid/[libuuid]
394 * http://directory.fsf.org/wiki/Popt[popt]
395 * http://liburcu.org/[Userspace RCU]
396 * http://www.xmlsoft.org/[libxml2]
397 * **Optional**: https://github.com/numactl/numactl[numactl]
400 . Download, build, and install the latest LTTng-modules{nbsp}{revision}:
406 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.11.tar.bz2 &&
407 tar -xf lttng-modules-latest-2.11.tar.bz2 &&
408 cd lttng-modules-2.11.* &&
410 sudo make modules_install &&
415 . Download, build, and install the latest LTTng-UST{nbsp}{revision}:
421 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.11.tar.bz2 &&
422 tar -xf lttng-ust-latest-2.11.tar.bz2 &&
423 cd lttng-ust-2.11.* &&
431 Add `--disable-numa` to `./configure` if you don't have
432 https://github.com/numactl/numactl[numactl].
436 .Java and Python application tracing
438 If you need to instrument and trace <<java-application,Java
439 applications>>, pass the `--enable-java-agent-jul`,
440 `--enable-java-agent-log4j`, or `--enable-java-agent-all` options to the
441 `configure` script, depending on which Java logging framework you use.
443 If you need to instrument and trace <<python-application,Python
444 applications>>, pass the `--enable-python-agent` option to the
445 `configure` script. You can set the `PYTHON` environment variable to the
446 path to the Python interpreter for which to install the LTTng-UST Python
454 By default, LTTng-UST libraries are installed to
455 dir:{/usr/local/lib}, which is the de facto directory in which to
456 keep self-compiled and third-party libraries.
458 When <<building-tracepoint-providers-and-user-application,linking an
459 instrumented user application with `liblttng-ust`>>:
461 * Append `/usr/local/lib` to the env:LD_LIBRARY_PATH environment
463 * Pass the `-L/usr/local/lib` and `-Wl,-rpath,/usr/local/lib` options to
464 man:gcc(1), man:g++(1), or man:clang(1).
468 . Download, build, and install the latest LTTng-tools{nbsp}{revision}:
474 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.11.tar.bz2 &&
475 tar -xf lttng-tools-latest-2.11.tar.bz2 &&
476 cd lttng-tools-2.11.* &&
484 TIP: The https://github.com/eepp/vlttng[vlttng tool] can do all the
485 previous steps automatically for a given version of LTTng and confine
486 the installed files in a specific directory. This can be useful to test
487 LTTng without installing it on your system.
493 This is a short guide to get started quickly with LTTng kernel and user
496 Before you follow this guide, make sure to <<installing-lttng,install>>
499 This tutorial walks you through the steps to:
501 . <<tracing-the-linux-kernel,Trace the Linux kernel>>.
502 . <<tracing-your-own-user-application,Trace a user application>> written
504 . <<viewing-and-analyzing-your-traces,View and analyze the
508 [[tracing-the-linux-kernel]]
509 === Trace the Linux kernel
511 The following command lines start with the `#` prompt because you need
512 root privileges to trace the Linux kernel. You can also trace the kernel
513 as a regular user if your Unix user is a member of the
514 <<tracing-group,tracing group>>.
516 . Create a <<tracing-session,tracing session>> which writes its traces
517 to dir:{/tmp/my-kernel-trace}:
522 # lttng create my-kernel-session --output=/tmp/my-kernel-trace
526 . List the available kernel tracepoints and system calls:
531 # lttng list --kernel
532 # lttng list --kernel --syscall
536 . Create <<event,event rules>> which match the desired instrumentation
537 point names, for example the `sched_switch` and `sched_process_fork`
538 tracepoints, and the man:open(2) and man:close(2) system calls:
543 # lttng enable-event --kernel sched_switch,sched_process_fork
544 # lttng enable-event --kernel --syscall open,close
548 You can also create an event rule which matches _all_ the Linux kernel
549 tracepoints (this will generate a lot of data when tracing):
554 # lttng enable-event --kernel --all
558 . <<basic-tracing-session-control,Start tracing>>:
567 . Do some operation on your system for a few seconds. For example,
568 load a website, or list the files of a directory.
569 . <<creating-destroying-tracing-sessions,Destroy>> the current
579 The man:lttng-destroy(1) command does not destroy the trace data; it
580 only destroys the state of the tracing session.
582 The man:lttng-destroy(1) command also runs the man:lttng-stop(1) command
583 implicitly (see <<basic-tracing-session-control,Start and stop a tracing
584 session>>). You need to stop tracing to make LTTng flush the remaining
585 trace data and make the trace readable.
587 . For the sake of this example, make the recorded trace accessible to
593 # chown -R $(whoami) /tmp/my-kernel-trace
597 See <<viewing-and-analyzing-your-traces,View and analyze the
598 recorded events>> to view the recorded events.
601 [[tracing-your-own-user-application]]
602 === Trace a user application
604 This section steps you through a simple example to trace a
605 _Hello world_ program written in C.
607 To create the traceable user application:
609 . Create the tracepoint provider header file, which defines the
610 tracepoints and the events they can generate:
616 #undef TRACEPOINT_PROVIDER
617 #define TRACEPOINT_PROVIDER hello_world
619 #undef TRACEPOINT_INCLUDE
620 #define TRACEPOINT_INCLUDE "./hello-tp.h"
622 #if !defined(_HELLO_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
625 #include <lttng/tracepoint.h>
635 ctf_string(my_string_field, my_string_arg)
636 ctf_integer(int, my_integer_field, my_integer_arg)
640 #endif /* _HELLO_TP_H */
642 #include <lttng/tracepoint-event.h>
646 . Create the tracepoint provider package source file:
652 #define TRACEPOINT_CREATE_PROBES
653 #define TRACEPOINT_DEFINE
655 #include "hello-tp.h"
659 . Build the tracepoint provider package:
664 $ gcc -c -I. hello-tp.c
668 . Create the _Hello World_ application source file:
675 #include "hello-tp.h"
677 int main(int argc, char *argv[])
681 puts("Hello, World!\nPress Enter to continue...");
684 * The following getchar() call is only placed here for the purpose
685 * of this demonstration, to pause the application in order for
686 * you to have time to list its tracepoints. It is not
692 * A tracepoint() call.
694 * Arguments, as defined in hello-tp.h:
696 * 1. Tracepoint provider name (required)
697 * 2. Tracepoint name (required)
698 * 3. my_integer_arg (first user-defined argument)
699 * 4. my_string_arg (second user-defined argument)
701 * Notice the tracepoint provider and tracepoint names are
702 * NOT strings: they are in fact parts of variables that the
703 * macros in hello-tp.h create.
705 tracepoint(hello_world, my_first_tracepoint, 23, "hi there!");
707 for (x = 0; x < argc; ++x) {
708 tracepoint(hello_world, my_first_tracepoint, x, argv[x]);
711 puts("Quitting now!");
712 tracepoint(hello_world, my_first_tracepoint, x * x, "x^2");
719 . Build the application:
728 . Link the application with the tracepoint provider package,
729 `liblttng-ust`, and `libdl`:
734 $ gcc -o hello hello.o hello-tp.o -llttng-ust -ldl
738 Here's the whole build process:
741 .User space tracing tutorial's build steps.
742 image::ust-flow.png[]
744 To trace the user application:
746 . Run the application with a few arguments:
751 $ ./hello world and beyond
760 Press Enter to continue...
764 . Start an LTTng <<lttng-sessiond,session daemon>>:
769 $ lttng-sessiond --daemonize
773 Note that a session daemon might already be running, for example as
774 a service that the distribution's service manager started.
776 . List the available user space tracepoints:
781 $ lttng list --userspace
785 You see the `hello_world:my_first_tracepoint` tracepoint listed
786 under the `./hello` process.
788 . Create a <<tracing-session,tracing session>>:
793 $ lttng create my-user-space-session
797 . Create an <<event,event rule>> which matches the
798 `hello_world:my_first_tracepoint` event name:
803 $ lttng enable-event --userspace hello_world:my_first_tracepoint
807 . <<basic-tracing-session-control,Start tracing>>:
816 . Go back to the running `hello` application and press Enter. The
817 program executes all `tracepoint()` instrumentation points and exits.
818 . <<creating-destroying-tracing-sessions,Destroy>> the current
828 The man:lttng-destroy(1) command does not destroy the trace data; it
829 only destroys the state of the tracing session.
831 The man:lttng-destroy(1) command also runs the man:lttng-stop(1) command
832 implicitly (see <<basic-tracing-session-control,Start and stop a tracing
833 session>>). You need to stop tracing to make LTTng flush the remaining
834 trace data and make the trace readable.
836 By default, LTTng saves the traces in
837 +$LTTNG_HOME/lttng-traces/__name__-__date__-__time__+,
838 where +__name__+ is the tracing session name. The
839 env:LTTNG_HOME environment variable defaults to `$HOME` if not set.
841 See <<viewing-and-analyzing-your-traces,View and analyze the
842 recorded events>> to view the recorded events.
845 [[viewing-and-analyzing-your-traces]]
846 === View and analyze the recorded events
848 Once you have completed the <<tracing-the-linux-kernel,Trace the Linux
849 kernel>> and <<tracing-your-own-user-application,Trace a user
850 application>> tutorials, you can inspect the recorded events.
852 There are many tools you can use to read LTTng traces:
854 * **cmd:babeltrace** is a command-line utility which converts trace
855 formats; it supports the format that LTTng produces, CTF, as well as a
856 basic text output which can be ++grep++ed. The cmd:babeltrace command
857 is part of the http://diamon.org/babeltrace[Babeltrace] project.
858 * Babeltrace also includes
859 **https://www.python.org/[Python{nbsp}3] bindings** so
860 that you can easily open and read an LTTng trace with your own script,
861 benefiting from the power of Python.
862 * http://tracecompass.org/[**Trace Compass**]
863 is a graphical user interface for viewing and analyzing any type of
864 logs or traces, including LTTng's.
865 * https://github.com/lttng/lttng-analyses[**LTTng analyses**] is a
866 project which includes many high-level analyses of LTTng kernel
867 traces, like scheduling statistics, interrupt frequency distribution,
868 top CPU usage, and more.
870 NOTE: This section assumes that LTTng saved the traces it recorded
871 during the previous tutorials to their default location, in the
872 dir:{$LTTNG_HOME/lttng-traces} directory. The env:LTTNG_HOME
873 environment variable defaults to `$HOME` if not set.
876 [[viewing-and-analyzing-your-traces-bt]]
877 ==== Use the cmd:babeltrace command-line tool
879 The simplest way to list all the recorded events of a trace is to pass
880 its path to cmd:babeltrace with no options:
884 $ babeltrace ~/lttng-traces/my-user-space-session*
887 cmd:babeltrace finds all traces recursively within the given path and
888 prints all their events, merging them in chronological order.
890 You can pipe the output of cmd:babeltrace into a tool like man:grep(1) for
895 $ babeltrace /tmp/my-kernel-trace | grep _switch
898 You can pipe the output of cmd:babeltrace into a tool like man:wc(1) to
899 count the recorded events:
903 $ babeltrace /tmp/my-kernel-trace | grep _open | wc --lines
907 [[viewing-and-analyzing-your-traces-bt-python]]
908 ==== Use the Babeltrace{nbsp}1 Python bindings
910 The <<viewing-and-analyzing-your-traces-bt,text output of cmd:babeltrace>>
911 is useful to isolate events by simple matching using man:grep(1) and
912 similar utilities. However, more elaborate filters, such as keeping only
913 event records with a field value falling within a specific range, are
914 not trivial to write using a shell. Moreover, reductions and even the
915 most basic computations involving multiple event records are virtually
916 impossible to implement.
918 Fortunately, Babeltrace{nbsp}1 ships with Python{nbsp}3 bindings which
919 makes it easy to read the event records of an LTTng trace sequentially
920 and compute the desired information.
922 The following script accepts an LTTng Linux kernel trace path as its
923 first argument and prints the short names of the top five running
924 processes on CPU{nbsp}0 during the whole trace:
929 from collections import Counter
935 if len(sys.argv) != 2:
936 msg = 'Usage: python3 {} TRACEPATH'.format(sys.argv[0])
937 print(msg, file=sys.stderr)
940 # A trace collection contains one or more traces
941 col = babeltrace.TraceCollection()
943 # Add the trace provided by the user (LTTng traces always have
945 if col.add_trace(sys.argv[1], 'ctf') is None:
946 raise RuntimeError('Cannot add trace')
948 # This counter dict contains execution times:
950 # task command name -> total execution time (ns)
951 exec_times = Counter()
953 # This contains the last `sched_switch` timestamp
957 for event in col.events:
958 # Keep only `sched_switch` events
959 if event.name != 'sched_switch':
962 # Keep only events which happened on CPU 0
963 if event['cpu_id'] != 0:
967 cur_ts = event.timestamp
973 # Previous task command (short) name
974 prev_comm = event['prev_comm']
976 # Initialize entry in our dict if not yet done
977 if prev_comm not in exec_times:
978 exec_times[prev_comm] = 0
980 # Compute previous command execution time
981 diff = cur_ts - last_ts
983 # Update execution time of this command
984 exec_times[prev_comm] += diff
986 # Update last timestamp
990 for name, ns in exec_times.most_common(5):
992 print('{:20}{} s'.format(name, s))
997 if __name__ == '__main__':
998 sys.exit(0 if top5proc() else 1)
1005 $ python3 top5proc.py /tmp/my-kernel-trace/kernel
1011 swapper/0 48.607245889 s
1012 chromium 7.192738188 s
1013 pavucontrol 0.709894415 s
1014 Compositor 0.660867933 s
1015 Xorg.bin 0.616753786 s
1018 Note that `swapper/0` is the "idle" process of CPU{nbsp}0 on Linux;
1019 since we weren't using the CPU that much when tracing, its first
1020 position in the list makes sense.
1024 == [[understanding-lttng]]Core concepts
1026 From a user's perspective, the LTTng system is built on a few concepts,
1027 or objects, on which the <<lttng-cli,cmd:lttng command-line tool>>
1028 operates by sending commands to the <<lttng-sessiond,session daemon>>.
1029 Understanding how those objects relate to eachother is key in mastering
1032 The core concepts are:
1034 * <<tracing-session,Tracing session>>
1035 * <<domain,Tracing domain>>
1036 * <<channel,Channel and ring buffer>>
1037 * <<"event","Instrumentation point, event rule, event, and event record">>
1043 A _tracing session_ is a stateful dialogue between you and
1044 a <<lttng-sessiond,session daemon>>. You can
1045 <<creating-destroying-tracing-sessions,create a new tracing
1046 session>> with the `lttng create` command.
1048 Anything that you do when you control LTTng tracers happens within a
1049 tracing session. In particular, a tracing session:
1052 * Has its own set of trace files.
1053 * Has its own state of activity (started or stopped).
1054 * Has its own <<tracing-session-mode,mode>> (local, network streaming,
1056 * Has its own <<channel,channels>> to which are associated their own
1057 <<event,event rules>>.
1060 .A _tracing session_ contains <<channel,channels>> that are members of <<domain,tracing domains>> and contain <<event,event rules>>.
1061 image::concepts.png[]
1063 Those attributes and objects are completely isolated between different
1066 A tracing session is analogous to a cash machine session:
1067 the operations you do on the banking system through the cash machine do
1068 not alter the data of other users of the same system. In the case of
1069 the cash machine, a session lasts as long as your bank card is inside.
1070 In the case of LTTng, a tracing session lasts from the `lttng create`
1071 command to the `lttng destroy` command.
1074 .Each Unix user has its own set of tracing sessions.
1075 image::many-sessions.png[]
1078 [[tracing-session-mode]]
1079 ==== Tracing session mode
1081 LTTng can send the generated trace data to different locations. The
1082 _tracing session mode_ dictates where to send it. The following modes
1083 are available in LTTng{nbsp}{revision}:
1086 LTTng writes the traces to the file system of the machine it traces
1089 Network streaming mode::
1090 LTTng sends the traces over the network to a
1091 <<lttng-relayd,relay daemon>> running on a remote system.
1094 LTTng does not write the traces by default. Instead, you can request
1095 LTTng to <<taking-a-snapshot,take a snapshot>>, that is, a copy of the
1096 tracing session's current sub-buffers, and to write it to the
1097 target's file system or to send it over the network to a
1098 <<lttng-relayd,relay daemon>> running on a remote system.
1101 This mode is similar to the network streaming mode, but a live
1102 trace viewer can connect to the distant relay daemon to
1103 <<lttng-live,view event records as LTTng generates them>>.
1109 A _tracing domain_ is a namespace for event sources. A tracing domain
1110 has its own properties and features.
1112 There are currently five available tracing domains:
1116 * `java.util.logging` (JUL)
1120 You must specify a tracing domain when using some commands to avoid
1121 ambiguity. For example, since all the domains support named tracepoints
1122 as event sources (instrumentation points that you manually insert in the
1123 source code), you need to specify a tracing domain when
1124 <<enabling-disabling-events,creating an event rule>> because all the
1125 tracing domains could have tracepoints with the same names.
1127 Some features are reserved to specific tracing domains. Dynamic function
1128 entry and return instrumentation points, for example, are currently only
1129 supported in the Linux kernel tracing domain, but support for other
1130 tracing domains could be added in the future.
1132 You can create <<channel,channels>> in the Linux kernel and user space
1133 tracing domains. The other tracing domains have a single default
1138 === Channel and ring buffer
1140 A _channel_ is an object which is responsible for a set of ring buffers.
1141 Each ring buffer is divided into multiple sub-buffers. When an LTTng
1142 tracer emits an event, it can record it to one or more
1143 sub-buffers. The attributes of a channel determine what to do when
1144 there's no space left for a new event record because all sub-buffers
1145 are full, where to send a full sub-buffer, and other behaviours.
1147 A channel is always associated to a <<domain,tracing domain>>. The
1148 `java.util.logging` (JUL), log4j, and Python tracing domains each have
1149 a default channel which you cannot configure.
1151 A channel also owns <<event,event rules>>. When an LTTng tracer emits
1152 an event, it records it to the sub-buffers of all
1153 the enabled channels with a satisfied event rule, as long as those
1154 channels are part of active <<tracing-session,tracing sessions>>.
1157 [[channel-buffering-schemes]]
1158 ==== Per-user vs. per-process buffering schemes
1160 A channel has at least one ring buffer _per CPU_. LTTng always
1161 records an event to the ring buffer associated to the CPU on which it
1164 Two _buffering schemes_ are available when you
1165 <<enabling-disabling-channels,create a channel>> in the
1166 user space <<domain,tracing domain>>:
1168 Per-user buffering::
1169 Allocate one set of ring buffers--one per CPU--shared by all the
1170 instrumented processes of each Unix user.
1174 .Per-user buffering scheme.
1175 image::per-user-buffering.png[]
1178 Per-process buffering::
1179 Allocate one set of ring buffers--one per CPU--for each
1180 instrumented process.
1184 .Per-process buffering scheme.
1185 image::per-process-buffering.png[]
1188 The per-process buffering scheme tends to consume more memory than the
1189 per-user option because systems generally have more instrumented
1190 processes than Unix users running instrumented processes. However, the
1191 per-process buffering scheme ensures that one process having a high
1192 event throughput won't fill all the shared sub-buffers of the same
1195 The Linux kernel tracing domain has only one available buffering scheme
1196 which is to allocate a single set of ring buffers for the whole system.
1197 This scheme is similar to the per-user option, but with a single, global
1198 user "running" the kernel.
1201 [[channel-overwrite-mode-vs-discard-mode]]
1202 ==== Overwrite vs. discard event record loss modes
1204 When an event occurs, LTTng records it to a specific sub-buffer (yellow
1205 arc in the following animations) of a specific channel's ring buffer.
1206 When there's no space left in a sub-buffer, the tracer marks it as
1207 consumable (red) and another, empty sub-buffer starts receiving the
1208 following event records. A <<lttng-consumerd,consumer daemon>>
1209 eventually consumes the marked sub-buffer (returns to white).
1212 [role="docsvg-channel-subbuf-anim"]
1217 In an ideal world, sub-buffers are consumed faster than they are filled,
1218 as it is the case in the previous animation. In the real world,
1219 however, all sub-buffers can be full at some point, leaving no space to
1220 record the following events.
1222 By default, LTTng-modules and LTTng-UST are _non-blocking_ tracers: when
1223 no empty sub-buffer is available, it is acceptable to lose event records
1224 when the alternative would be to cause substantial delays in the
1225 instrumented application's execution. LTTng privileges performance over
1226 integrity; it aims at perturbing the target system as little as possible
1227 in order to make tracing of subtle race conditions and rare interrupt
1230 Since LTTng{nbsp}2.10, the LTTng user space tracer, LTTng-UST, supports
1231 a _blocking mode_. See the <<blocking-timeout-example,blocking timeout
1232 example>> to learn how to use the blocking mode.
1234 When it comes to losing event records because no empty sub-buffer is
1235 available, or because the <<opt-blocking-timeout,blocking timeout>> is
1236 reached, the channel's _event record loss mode_ determines what to do.
1237 The available event record loss modes are:
1240 Drop the newest event records until a the tracer releases a
1243 This is the only available mode when you specify a
1244 <<opt-blocking-timeout,blocking timeout>>.
1247 Clear the sub-buffer containing the oldest event records and start
1248 writing the newest event records there.
1250 This mode is sometimes called _flight recorder mode_ because it's
1252 https://en.wikipedia.org/wiki/Flight_recorder[flight recorder]:
1253 always keep a fixed amount of the latest data.
1255 Which mechanism you should choose depends on your context: prioritize
1256 the newest or the oldest event records in the ring buffer?
1258 Beware that, in overwrite mode, the tracer abandons a _whole sub-buffer_
1259 as soon as a there's no space left for a new event record, whereas in
1260 discard mode, the tracer only discards the event record that doesn't
1263 In discard mode, LTTng increments a count of lost event records when an
1264 event record is lost and saves this count to the trace. In overwrite
1265 mode, since LTTng{nbsp}2.8, LTTng increments a count of lost sub-buffers
1266 when a sub-buffer is lost and saves this count to the trace. In this
1267 mode, LTTng does not write to the trace the exact number of lost event
1268 records in those lost sub-buffers. Trace analyses can use the trace's
1269 saved discarded event record and sub-buffer counts to decide whether or
1270 not to perform the analyses even if trace data is known to be missing.
1272 There are a few ways to decrease your probability of losing event
1274 <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>> shows
1275 how you can fine-tune the sub-buffer count and size of a channel to
1276 virtually stop losing event records, though at the cost of greater
1280 [[channel-subbuf-size-vs-subbuf-count]]
1281 ==== Sub-buffer count and size
1283 When you <<enabling-disabling-channels,create a channel>>, you can
1284 set its number of sub-buffers and their size.
1286 Note that there is noticeable CPU overhead introduced when
1287 switching sub-buffers (marking a full one as consumable and switching
1288 to an empty one for the following events to be recorded). Knowing this,
1289 the following list presents a few practical situations along with how
1290 to configure the sub-buffer count and size for them:
1292 * **High event throughput**: In general, prefer bigger sub-buffers to
1293 lower the risk of losing event records.
1295 Having bigger sub-buffers also ensures a lower
1296 <<channel-switch-timer,sub-buffer switching frequency>>.
1298 The number of sub-buffers is only meaningful if you create the channel
1299 in overwrite mode: in this case, if a sub-buffer overwrite happens, the
1300 other sub-buffers are left unaltered.
1302 * **Low event throughput**: In general, prefer smaller sub-buffers
1303 since the risk of losing event records is low.
1305 Because events occur less frequently, the sub-buffer switching frequency
1306 should remain low and thus the tracer's overhead should not be a
1309 * **Low memory system**: If your target system has a low memory
1310 limit, prefer fewer first, then smaller sub-buffers.
1312 Even if the system is limited in memory, you want to keep the
1313 sub-buffers as big as possible to avoid a high sub-buffer switching
1316 Note that LTTng uses http://diamon.org/ctf/[CTF] as its trace format,
1317 which means event data is very compact. For example, the average
1318 LTTng kernel event record weights about 32{nbsp}bytes. Thus, a
1319 sub-buffer size of 1{nbsp}MiB is considered big.
1321 The previous situations highlight the major trade-off between a few big
1322 sub-buffers and more, smaller sub-buffers: sub-buffer switching
1323 frequency vs. how much data is lost in overwrite mode. Assuming a
1324 constant event throughput and using the overwrite mode, the two
1325 following configurations have the same ring buffer total size:
1328 [role="docsvg-channel-subbuf-size-vs-count-anim"]
1333 * **Two sub-buffers of 4{nbsp}MiB each**: Expect a very low sub-buffer
1334 switching frequency, but if a sub-buffer overwrite happens, half of
1335 the event records so far (4{nbsp}MiB) are definitely lost.
1336 * **Eight sub-buffers of 1{nbsp}MiB each**: Expect four times the tracer's
1337 overhead as the previous configuration, but if a sub-buffer
1338 overwrite happens, only the eighth of event records so far are
1341 In discard mode, the sub-buffers count parameter is pointless: use two
1342 sub-buffers and set their size according to the requirements of your
1346 [[channel-switch-timer]]
1347 ==== Switch timer period
1349 The _switch timer period_ is an important configurable attribute of
1350 a channel to ensure periodic sub-buffer flushing.
1352 When the _switch timer_ expires, a sub-buffer switch happens. You can
1353 set the switch timer period attribute when you
1354 <<enabling-disabling-channels,create a channel>> to ensure that LTTng
1355 consumes and commits trace data to trace files or to a distant relay
1356 daemon periodically in case of a low event throughput.
1359 [role="docsvg-channel-switch-timer"]
1364 This attribute is also convenient when you use big sub-buffers to cope
1365 with a sporadic high event throughput, even if the throughput is
1369 [[channel-read-timer]]
1370 ==== Read timer period
1372 By default, the LTTng tracers use a notification mechanism to signal a
1373 full sub-buffer so that a consumer daemon can consume it. When such
1374 notifications must be avoided, for example in real-time applications,
1375 you can use the channel's _read timer_ instead. When the read timer
1376 fires, the <<lttng-consumerd,consumer daemon>> checks for full,
1377 consumable sub-buffers.
1380 [[tracefile-rotation]]
1381 ==== Trace file count and size
1383 By default, trace files can grow as large as needed. You can set the
1384 maximum size of each trace file that a channel writes when you
1385 <<enabling-disabling-channels,create a channel>>. When the size of
1386 a trace file reaches the channel's fixed maximum size, LTTng creates
1387 another file to contain the next event records. LTTng appends a file
1388 count to each trace file name in this case.
1390 If you set the trace file size attribute when you create a channel, the
1391 maximum number of trace files that LTTng creates is _unlimited_ by
1392 default. To limit them, you can also set a maximum number of trace
1393 files. When the number of trace files reaches the channel's fixed
1394 maximum count, the oldest trace file is overwritten. This mechanism is
1395 called _trace file rotation_.
1399 Even if you don't limit the trace file count, you cannot assume that
1400 LTTng doesn't manage any trace file.
1402 In other words, there is no safe way to know if LTTng still holds a
1403 given trace file open with the trace file rotation feature.
1405 The only way to obtain an unmanaged, self-contained LTTng trace before
1406 you <<creating-destroying-tracing-sessions,destroy>> the tracing session
1407 is with the <<session-rotation,tracing session rotation>> feature
1408 (available since LTTng{nbsp}2.11).
1413 === Instrumentation point, event rule, event, and event record
1415 An _event rule_ is a set of conditions which must be **all** satisfied
1416 for LTTng to record an occuring event.
1418 You set the conditions when you <<enabling-disabling-events,create
1421 You always attach an event rule to <<channel,channel>> when you create
1424 When an event passes the conditions of an event rule, LTTng records it
1425 in one of the attached channel's sub-buffers.
1427 The available conditions, as of LTTng{nbsp}{revision}, are:
1429 * The event rule _is enabled_.
1430 * The instrumentation point's type _is{nbsp}T_.
1431 * The instrumentation point's name (sometimes called _event name_)
1432 _matches{nbsp}N_, but _is not{nbsp}E_.
1433 * The instrumentation point's log level _is as severe as{nbsp}L_, or
1434 _is exactly{nbsp}L_.
1435 * The fields of the event's payload _satisfy_ a filter
1436 expression{nbsp}__F__.
1438 As you can see, all the conditions but the dynamic filter are related to
1439 the event rule's status or to the instrumentation point, not to the
1440 occurring events. This is why, without a filter, checking if an event
1441 passes an event rule is not a dynamic task: when you create or modify an
1442 event rule, all the tracers of its tracing domain enable or disable the
1443 instrumentation points themselves once. This is possible because the
1444 attributes of an instrumentation point (type, name, and log level) are
1445 defined statically. In other words, without a dynamic filter, the tracer
1446 _does not evaluate_ the arguments of an instrumentation point unless it
1447 matches an enabled event rule.
1449 Note that, for LTTng to record an event, the <<channel,channel>> to
1450 which a matching event rule is attached must also be enabled, and the
1451 <<tracing-session,tracing session>> owning this channel must be active
1455 .Logical path from an instrumentation point to an event record.
1456 image::event-rule.png[]
1458 .Event, event record, or event rule?
1460 With so many similar terms, it's easy to get confused.
1462 An **event** is the consequence of the execution of an _instrumentation
1463 point_, like a tracepoint that you manually place in some source code,
1464 or a Linux kernel kprobe. An event is said to _occur_ at a specific
1465 time. Different actions can be taken upon the occurrence of an event,
1466 like record the event's payload to a buffer.
1468 An **event record** is the representation of an event in a sub-buffer. A
1469 tracer is responsible for capturing the payload of an event, current
1470 context variables, the event's ID, and the event's timestamp. LTTng
1471 can append this sub-buffer to a trace file.
1473 An **event rule** is a set of conditions which must _all_ be satisfied
1474 for LTTng to record an occuring event. Events still occur without
1475 satisfying event rules, but LTTng does not record them.
1480 == Components of noch:{LTTng}
1482 The second _T_ in _LTTng_ stands for _toolkit_: it would be wrong
1483 to call LTTng a simple _tool_ since it is composed of multiple
1484 interacting components. This section describes those components,
1485 explains their respective roles, and shows how they connect together to
1486 form the LTTng ecosystem.
1488 The following diagram shows how the most important components of LTTng
1489 interact with user applications, the Linux kernel, and you:
1492 .Control and trace data paths between LTTng components.
1493 image::plumbing.png[]
1495 The LTTng project incorporates:
1497 * **LTTng-tools**: Libraries and command-line interface to
1498 control tracing sessions.
1499 ** <<lttng-sessiond,Session daemon>> (man:lttng-sessiond(8)).
1500 ** <<lttng-consumerd,Consumer daemon>> (cmd:lttng-consumerd).
1501 ** <<lttng-relayd,Relay daemon>> (man:lttng-relayd(8)).
1502 ** <<liblttng-ctl-lttng,Tracing control library>> (`liblttng-ctl`).
1503 ** <<lttng-cli,Tracing control command-line tool>> (man:lttng(1)).
1504 * **LTTng-UST**: Libraries and Java/Python packages to trace user
1506 ** <<lttng-ust,User space tracing library>> (`liblttng-ust`) and its
1507 headers to instrument and trace any native user application.
1508 ** <<prebuilt-ust-helpers,Preloadable user space tracing helpers>>:
1509 *** `liblttng-ust-libc-wrapper`
1510 *** `liblttng-ust-pthread-wrapper`
1511 *** `liblttng-ust-cyg-profile`
1512 *** `liblttng-ust-cyg-profile-fast`
1513 *** `liblttng-ust-dl`
1514 ** User space tracepoint provider source files generator command-line
1515 tool (man:lttng-gen-tp(1)).
1516 ** <<lttng-ust-agents,LTTng-UST Java agent>> to instrument and trace
1517 Java applications using `java.util.logging` or
1518 Apache log4j{nbsp}1.2 logging.
1519 ** <<lttng-ust-agents,LTTng-UST Python agent>> to instrument
1520 Python applications using the standard `logging` package.
1521 * **LTTng-modules**: <<lttng-modules,Linux kernel modules>> to trace
1523 ** LTTng kernel tracer module.
1524 ** Tracing ring buffer kernel modules.
1525 ** Probe kernel modules.
1526 ** LTTng logger kernel module.
1530 === Tracing control command-line interface
1533 .The tracing control command-line interface.
1534 image::plumbing-lttng-cli.png[]
1536 The _man:lttng(1) command-line tool_ is the standard user interface to
1537 control LTTng <<tracing-session,tracing sessions>>. The cmd:lttng tool
1538 is part of LTTng-tools.
1540 The cmd:lttng tool is linked with
1541 <<liblttng-ctl-lttng,`liblttng-ctl`>> to communicate with
1542 one or more <<lttng-sessiond,session daemons>> behind the scenes.
1544 The cmd:lttng tool has a Git-like interface:
1548 $ lttng <GENERAL OPTIONS> <COMMAND> <COMMAND OPTIONS>
1551 The <<controlling-tracing,Tracing control>> section explores the
1552 available features of LTTng using the cmd:lttng tool.
1555 [[liblttng-ctl-lttng]]
1556 === Tracing control library
1559 .The tracing control library.
1560 image::plumbing-liblttng-ctl.png[]
1562 The _LTTng control library_, `liblttng-ctl`, is used to communicate
1563 with a <<lttng-sessiond,session daemon>> using a C API that hides the
1564 underlying protocol's details. `liblttng-ctl` is part of LTTng-tools.
1566 The <<lttng-cli,cmd:lttng command-line tool>>
1567 is linked with `liblttng-ctl`.
1569 You can use `liblttng-ctl` in C or $$C++$$ source code by including its
1574 #include <lttng/lttng.h>
1577 Some objects are referenced by name (C string), such as tracing
1578 sessions, but most of them require to create a handle first using
1579 `lttng_create_handle()`.
1581 The best available developer documentation for `liblttng-ctl` is, as of
1582 LTTng{nbsp}{revision}, its installed header files. Every function and
1583 structure is thoroughly documented.
1587 === User space tracing library
1590 .The user space tracing library.
1591 image::plumbing-liblttng-ust.png[]
1593 The _user space tracing library_, `liblttng-ust` (see man:lttng-ust(3)),
1594 is the LTTng user space tracer. It receives commands from a
1595 <<lttng-sessiond,session daemon>>, for example to
1596 enable and disable specific instrumentation points, and writes event
1597 records to ring buffers shared with a
1598 <<lttng-consumerd,consumer daemon>>.
1599 `liblttng-ust` is part of LTTng-UST.
1601 Public C header files are installed beside `liblttng-ust` to
1602 instrument any <<c-application,C or $$C++$$ application>>.
1604 <<lttng-ust-agents,LTTng-UST agents>>, which are regular Java and Python
1605 packages, use their own library providing tracepoints which is
1606 linked with `liblttng-ust`.
1608 An application or library does not have to initialize `liblttng-ust`
1609 manually: its constructor does the necessary tasks to properly register
1610 to a session daemon. The initialization phase also enables the
1611 instrumentation points matching the <<event,event rules>> that you
1615 [[lttng-ust-agents]]
1616 === User space tracing agents
1619 .The user space tracing agents.
1620 image::plumbing-lttng-ust-agents.png[]
1622 The _LTTng-UST Java and Python agents_ are regular Java and Python
1623 packages which add LTTng tracing capabilities to the
1624 native logging frameworks. The LTTng-UST agents are part of LTTng-UST.
1626 In the case of Java, the
1627 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[`java.util.logging`
1628 core logging facilities] and
1629 https://logging.apache.org/log4j/1.2/[Apache log4j{nbsp}1.2] are supported.
1630 Note that Apache Log4{nbsp}2 is not supported.
1632 In the case of Python, the standard
1633 https://docs.python.org/3/library/logging.html[`logging`] package
1634 is supported. Both Python{nbsp}2 and Python{nbsp}3 modules can import the
1635 LTTng-UST Python agent package.
1637 The applications using the LTTng-UST agents are in the
1638 `java.util.logging` (JUL),
1639 log4j, and Python <<domain,tracing domains>>.
1641 Both agents use the same mechanism to trace the log statements. When an
1642 agent initializes, it creates a log handler that attaches to the root
1643 logger. The agent also registers to a <<lttng-sessiond,session daemon>>.
1644 When the application executes a log statement, the root logger passes it
1645 to the agent's log handler. The agent's log handler calls a native
1646 function in a tracepoint provider package shared library linked with
1647 <<lttng-ust,`liblttng-ust`>>, passing the formatted log message and
1648 other fields, like its logger name and its log level. This native
1649 function contains a user space instrumentation point, hence tracing the
1652 The log level condition of an
1653 <<event,event rule>> is considered when tracing
1654 a Java or a Python application, and it's compatible with the standard
1655 JUL, log4j, and Python log levels.
1659 === LTTng kernel modules
1662 .The LTTng kernel modules.
1663 image::plumbing-lttng-modules.png[]
1665 The _LTTng kernel modules_ are a set of Linux kernel modules
1666 which implement the kernel tracer of the LTTng project. The LTTng
1667 kernel modules are part of LTTng-modules.
1669 The LTTng kernel modules include:
1671 * A set of _probe_ modules.
1673 Each module attaches to a specific subsystem
1674 of the Linux kernel using its tracepoint instrument points. There are
1675 also modules to attach to the entry and return points of the Linux
1676 system call functions.
1678 * _Ring buffer_ modules.
1680 A ring buffer implementation is provided as kernel modules. The LTTng
1681 kernel tracer writes to the ring buffer; a
1682 <<lttng-consumerd,consumer daemon>> reads from the ring buffer.
1684 * The _LTTng kernel tracer_ module.
1685 * The _LTTng logger_ module.
1687 The LTTng logger module implements the special path:{/proc/lttng-logger}
1688 file so that any executable can generate LTTng events by opening and
1689 writing to this file.
1691 See <<proc-lttng-logger-abi,LTTng logger>>.
1693 Generally, you do not have to load the LTTng kernel modules manually
1694 (using man:modprobe(8), for example): a root <<lttng-sessiond,session
1695 daemon>> loads the necessary modules when starting. If you have extra
1696 probe modules, you can specify to load them to the session daemon on
1699 The LTTng kernel modules are installed in
1700 +/usr/lib/modules/__release__/extra+ by default, where +__release__+ is
1701 the kernel release (see `uname --kernel-release`).
1708 .The session daemon.
1709 image::plumbing-sessiond.png[]
1711 The _session daemon_, man:lttng-sessiond(8), is a daemon responsible for
1712 managing tracing sessions and for controlling the various components of
1713 LTTng. The session daemon is part of LTTng-tools.
1715 The session daemon sends control requests to and receives control
1718 * The <<lttng-ust,user space tracing library>>.
1720 Any instance of the user space tracing library first registers to
1721 a session daemon. Then, the session daemon can send requests to
1722 this instance, such as:
1725 ** Get the list of tracepoints.
1726 ** Share an <<event,event rule>> so that the user space tracing library
1727 can enable or disable tracepoints. Amongst the possible conditions
1728 of an event rule is a filter expression which `liblttng-ust` evalutes
1729 when an event occurs.
1730 ** Share <<channel,channel>> attributes and ring buffer locations.
1733 The session daemon and the user space tracing library use a Unix
1734 domain socket for their communication.
1736 * The <<lttng-ust-agents,user space tracing agents>>.
1738 Any instance of a user space tracing agent first registers to
1739 a session daemon. Then, the session daemon can send requests to
1740 this instance, such as:
1743 ** Get the list of loggers.
1744 ** Enable or disable a specific logger.
1747 The session daemon and the user space tracing agent use a TCP connection
1748 for their communication.
1750 * The <<lttng-modules,LTTng kernel tracer>>.
1751 * The <<lttng-consumerd,consumer daemon>>.
1753 The session daemon sends requests to the consumer daemon to instruct
1754 it where to send the trace data streams, amongst other information.
1756 * The <<lttng-relayd,relay daemon>>.
1758 The session daemon receives commands from the
1759 <<liblttng-ctl-lttng,tracing control library>>.
1761 The root session daemon loads the appropriate
1762 <<lttng-modules,LTTng kernel modules>> on startup. It also spawns
1763 a <<lttng-consumerd,consumer daemon>> as soon as you create
1764 an <<event,event rule>>.
1766 The session daemon does not send and receive trace data: this is the
1767 role of the <<lttng-consumerd,consumer daemon>> and
1768 <<lttng-relayd,relay daemon>>. It does, however, generate the
1769 http://diamon.org/ctf/[CTF] metadata stream.
1771 Each Unix user can have its own session daemon instance. The
1772 tracing sessions which different session daemons manage are completely
1775 The root user's session daemon is the only one which is
1776 allowed to control the LTTng kernel tracer, and its spawned consumer
1777 daemon is the only one which is allowed to consume trace data from the
1778 LTTng kernel tracer. Note, however, that any Unix user which is a member
1779 of the <<tracing-group,tracing group>> is allowed
1780 to create <<channel,channels>> in the
1781 Linux kernel <<domain,tracing domain>>, and thus to trace the Linux
1784 The <<lttng-cli,cmd:lttng command-line tool>> automatically starts a
1785 session daemon when using its `create` command if none is currently
1786 running. You can also start the session daemon manually.
1793 .The consumer daemon.
1794 image::plumbing-consumerd.png[]
1796 The _consumer daemon_, cmd:lttng-consumerd, is a daemon which shares
1797 ring buffers with user applications or with the LTTng kernel modules to
1798 collect trace data and send it to some location (on disk or to a
1799 <<lttng-relayd,relay daemon>> over the network). The consumer daemon
1800 is part of LTTng-tools.
1802 You do not start a consumer daemon manually: a consumer daemon is always
1803 spawned by a <<lttng-sessiond,session daemon>> as soon as you create an
1804 <<event,event rule>>, that is, before you start tracing. When you kill
1805 its owner session daemon, the consumer daemon also exits because it is
1806 the session daemon's child process. Command-line options of
1807 man:lttng-sessiond(8) target the consumer daemon process.
1809 There are up to two running consumer daemons per Unix user, whereas only
1810 one session daemon can run per user. This is because each process can be
1811 either 32-bit or 64-bit: if the target system runs a mixture of 32-bit
1812 and 64-bit processes, it is more efficient to have separate
1813 corresponding 32-bit and 64-bit consumer daemons. The root user is an
1814 exception: it can have up to _three_ running consumer daemons: 32-bit
1815 and 64-bit instances for its user applications, and one more
1816 reserved for collecting kernel trace data.
1824 image::plumbing-relayd.png[]
1826 The _relay daemon_, man:lttng-relayd(8), is a daemon acting as a bridge
1827 between remote session and consumer daemons, local trace files, and a
1828 remote live trace viewer. The relay daemon is part of LTTng-tools.
1830 The main purpose of the relay daemon is to implement a receiver of
1831 <<sending-trace-data-over-the-network,trace data over the network>>.
1832 This is useful when the target system does not have much file system
1833 space to record trace files locally.
1835 The relay daemon is also a server to which a
1836 <<lttng-live,live trace viewer>> can
1837 connect. The live trace viewer sends requests to the relay daemon to
1838 receive trace data as the target system emits events. The
1839 communication protocol is named _LTTng live_; it is used over TCP
1842 Note that you can start the relay daemon on the target system directly.
1843 This is the setup of choice when the use case is to view events as
1844 the target system emits them without the need of a remote system.
1848 == [[using-lttng]]Instrumentation
1850 There are many examples of tracing and monitoring in our everyday life:
1852 * You have access to real-time and historical weather reports and
1853 forecasts thanks to weather stations installed around the country.
1854 * You know your heart is safe thanks to an electrocardiogram.
1855 * You make sure not to drive your car too fast and to have enough fuel
1856 to reach your destination thanks to gauges visible on your dashboard.
1858 All the previous examples have something in common: they rely on
1859 **instruments**. Without the electrodes attached to the surface of your
1860 body's skin, cardiac monitoring is futile.
1862 LTTng, as a tracer, is no different from those real life examples. If
1863 you're about to trace a software system or, in other words, record its
1864 history of execution, you better have **instrumentation points** in the
1865 subject you're tracing, that is, the actual software.
1867 Various ways were developed to instrument a piece of software for LTTng
1868 tracing. The most straightforward one is to manually place
1869 instrumentation points, called _tracepoints_, in the software's source
1870 code. It is also possible to add instrumentation points dynamically in
1871 the Linux kernel <<domain,tracing domain>>.
1873 If you're only interested in tracing the Linux kernel, your
1874 instrumentation needs are probably already covered by LTTng's built-in
1875 <<lttng-modules,Linux kernel tracepoints>>. You may also wish to trace a
1876 user application which is already instrumented for LTTng tracing.
1877 In such cases, you can skip this whole section and read the topics of
1878 the <<controlling-tracing,Tracing control>> section.
1880 Many methods are available to instrument a piece of software for LTTng
1883 * <<c-application,User space instrumentation for C and $$C++$$
1885 * <<prebuilt-ust-helpers,Prebuilt user space tracing helpers>>.
1886 * <<java-application,User space Java agent>>.
1887 * <<python-application,User space Python agent>>.
1888 * <<proc-lttng-logger-abi,LTTng logger>>.
1889 * <<instrumenting-linux-kernel,LTTng kernel tracepoints>>.
1893 === [[cxx-application]]User space instrumentation for C and $$C++$$ applications
1895 The procedure to instrument a C or $$C++$$ user application with
1896 the <<lttng-ust,LTTng user space tracing library>>, `liblttng-ust`, is:
1898 . <<tracepoint-provider,Create the source files of a tracepoint provider
1900 . <<probing-the-application-source-code,Add tracepoints to
1901 the application's source code>>.
1902 . <<building-tracepoint-providers-and-user-application,Build and link
1903 a tracepoint provider package and the user application>>.
1905 If you need quick, man:printf(3)-like instrumentation, you can skip
1906 those steps and use <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>>
1909 IMPORTANT: You need to <<installing-lttng,install>> LTTng-UST to
1910 instrument a user application with `liblttng-ust`.
1913 [[tracepoint-provider]]
1914 ==== Create the source files of a tracepoint provider package
1916 A _tracepoint provider_ is a set of compiled functions which provide
1917 **tracepoints** to an application, the type of instrumentation point
1918 supported by LTTng-UST. Those functions can emit events with
1919 user-defined fields and serialize those events as event records to one
1920 or more LTTng-UST <<channel,channel>> sub-buffers. The `tracepoint()`
1921 macro, which you <<probing-the-application-source-code,insert in a user
1922 application's source code>>, calls those functions.
1924 A _tracepoint provider package_ is an object file (`.o`) or a shared
1925 library (`.so`) which contains one or more tracepoint providers.
1926 Its source files are:
1928 * One or more <<tpp-header,tracepoint provider header>> (`.h`).
1929 * A <<tpp-source,tracepoint provider package source>> (`.c`).
1931 A tracepoint provider package is dynamically linked with `liblttng-ust`,
1932 the LTTng user space tracer, at run time.
1935 .User application linked with `liblttng-ust` and containing a tracepoint provider.
1936 image::ust-app.png[]
1938 NOTE: If you need quick, man:printf(3)-like instrumentation, you can
1939 skip creating and using a tracepoint provider and use
1940 <<tracef,`tracef()`>> or <<tracelog,`tracelog()`>> instead.
1944 ===== Create a tracepoint provider header file template
1946 A _tracepoint provider header file_ contains the tracepoint
1947 definitions of a tracepoint provider.
1949 To create a tracepoint provider header file:
1951 . Start from this template:
1955 .Tracepoint provider header file template (`.h` file extension).
1957 #undef TRACEPOINT_PROVIDER
1958 #define TRACEPOINT_PROVIDER provider_name
1960 #undef TRACEPOINT_INCLUDE
1961 #define TRACEPOINT_INCLUDE "./tp.h"
1963 #if !defined(_TP_H) || defined(TRACEPOINT_HEADER_MULTI_READ)
1966 #include <lttng/tracepoint.h>
1969 * Use TRACEPOINT_EVENT(), TRACEPOINT_EVENT_CLASS(),
1970 * TRACEPOINT_EVENT_INSTANCE(), and TRACEPOINT_LOGLEVEL() here.
1975 #include <lttng/tracepoint-event.h>
1981 * `provider_name` with the name of your tracepoint provider.
1982 * `"tp.h"` with the name of your tracepoint provider header file.
1984 . Below the `#include <lttng/tracepoint.h>` line, put your
1985 <<defining-tracepoints,tracepoint definitions>>.
1987 Your tracepoint provider name must be unique amongst all the possible
1988 tracepoint provider names used on the same target system. We
1989 suggest to include the name of your project or company in the name,
1990 for example, `org_lttng_my_project_tpp`.
1992 TIP: [[lttng-gen-tp]]You can use the man:lttng-gen-tp(1) tool to create
1993 this boilerplate for you. When using cmd:lttng-gen-tp, all you need to
1994 write are the <<defining-tracepoints,tracepoint definitions>>.
1997 [[defining-tracepoints]]
1998 ===== Create a tracepoint definition
2000 A _tracepoint definition_ defines, for a given tracepoint:
2002 * Its **input arguments**. They are the macro parameters that the
2003 `tracepoint()` macro accepts for this particular tracepoint
2004 in the user application's source code.
2005 * Its **output event fields**. They are the sources of event fields
2006 that form the payload of any event that the execution of the
2007 `tracepoint()` macro emits for this particular tracepoint.
2009 You can create a tracepoint definition by using the
2010 `TRACEPOINT_EVENT()` macro below the `#include <lttng/tracepoint.h>`
2012 <<tpp-header,tracepoint provider header file template>>.
2014 The syntax of the `TRACEPOINT_EVENT()` macro is:
2017 .`TRACEPOINT_EVENT()` macro syntax.
2020 /* Tracepoint provider name */
2023 /* Tracepoint name */
2026 /* Input arguments */
2031 /* Output event fields */
2040 * `provider_name` with your tracepoint provider name.
2041 * `tracepoint_name` with your tracepoint name.
2042 * `arguments` with the <<tpp-def-input-args,input arguments>>.
2043 * `fields` with the <<tpp-def-output-fields,output event field>>
2046 This tracepoint emits events named `provider_name:tracepoint_name`.
2049 .Event name's length limitation
2051 The concatenation of the tracepoint provider name and the
2052 tracepoint name must not exceed **254{nbsp}characters**. If it does, the
2053 instrumented application compiles and runs, but LTTng throws multiple
2054 warnings and you could experience serious issues.
2057 [[tpp-def-input-args]]The syntax of the `TP_ARGS()` macro is:
2060 .`TP_ARGS()` macro syntax.
2069 * `type` with the C type of the argument.
2070 * `arg_name` with the argument name.
2072 You can repeat `type` and `arg_name` up to 10{nbsp}times to have
2073 more than one argument.
2075 .`TP_ARGS()` usage with three arguments.
2087 The `TP_ARGS()` and `TP_ARGS(void)` forms are valid to create a
2088 tracepoint definition with no input arguments.
2090 [[tpp-def-output-fields]]The `TP_FIELDS()` macro contains a list of
2091 `ctf_*()` macros. Each `ctf_*()` macro defines one event field. See
2092 man:lttng-ust(3) for a complete description of the available `ctf_*()`
2093 macros. A `ctf_*()` macro specifies the type, size, and byte order of
2096 Each `ctf_*()` macro takes an _argument expression_ parameter. This is a
2097 C expression that the tracer evalutes at the `tracepoint()` macro site
2098 in the application's source code. This expression provides a field's
2099 source of data. The argument expression can include input argument names
2100 listed in the `TP_ARGS()` macro.
2102 Each `ctf_*()` macro also takes a _field name_ parameter. Field names
2103 must be unique within a given tracepoint definition.
2105 Here's a complete tracepoint definition example:
2107 .Tracepoint definition.
2109 The following tracepoint definition defines a tracepoint which takes
2110 three input arguments and has four output event fields.
2114 #include "my-custom-structure.h"
2120 const struct my_custom_structure*, my_custom_structure,
2125 ctf_string(query_field, query)
2126 ctf_float(double, ratio_field, ratio)
2127 ctf_integer(int, recv_size, my_custom_structure->recv_size)
2128 ctf_integer(int, send_size, my_custom_structure->send_size)
2133 You can refer to this tracepoint definition with the `tracepoint()`
2134 macro in your application's source code like this:
2138 tracepoint(my_provider, my_tracepoint,
2139 my_structure, some_ratio, the_query);
2143 NOTE: The LTTng tracer only evaluates tracepoint arguments at run time
2144 if they satisfy an enabled <<event,event rule>>.
2147 [[using-tracepoint-classes]]
2148 ===== Use a tracepoint class
2150 A _tracepoint class_ is a class of tracepoints which share the same
2151 output event field definitions. A _tracepoint instance_ is one
2152 instance of such a defined tracepoint class, with its own tracepoint
2155 The <<defining-tracepoints,`TRACEPOINT_EVENT()` macro>> is actually a
2156 shorthand which defines both a tracepoint class and a tracepoint
2157 instance at the same time.
2159 When you build a tracepoint provider package, the C or $$C++$$ compiler
2160 creates one serialization function for each **tracepoint class**. A
2161 serialization function is responsible for serializing the event fields
2162 of a tracepoint to a sub-buffer when tracing.
2164 For various performance reasons, when your situation requires multiple
2165 tracepoint definitions with different names, but with the same event
2166 fields, we recommend that you manually create a tracepoint class
2167 and instantiate as many tracepoint instances as needed. One positive
2168 effect of such a design, amongst other advantages, is that all
2169 tracepoint instances of the same tracepoint class reuse the same
2170 serialization function, thus reducing
2171 https://en.wikipedia.org/wiki/Cache_pollution[cache pollution].
2173 .Use a tracepoint class and tracepoint instances.
2175 Consider the following three tracepoint definitions:
2187 ctf_integer(int, userid, userid)
2188 ctf_integer(size_t, len, len)
2200 ctf_integer(int, userid, userid)
2201 ctf_integer(size_t, len, len)
2213 ctf_integer(int, userid, userid)
2214 ctf_integer(size_t, len, len)
2219 In this case, we create three tracepoint classes, with one implicit
2220 tracepoint instance for each of them: `get_account`, `get_settings`, and
2221 `get_transaction`. However, they all share the same event field names
2222 and types. Hence three identical, yet independent serialization
2223 functions are created when you build the tracepoint provider package.
2225 A better design choice is to define a single tracepoint class and three
2226 tracepoint instances:
2230 /* The tracepoint class */
2231 TRACEPOINT_EVENT_CLASS(
2232 /* Tracepoint provider name */
2235 /* Tracepoint class name */
2238 /* Input arguments */
2244 /* Output event fields */
2246 ctf_integer(int, userid, userid)
2247 ctf_integer(size_t, len, len)
2251 /* The tracepoint instances */
2252 TRACEPOINT_EVENT_INSTANCE(
2253 /* Tracepoint provider name */
2256 /* Tracepoint class name */
2259 /* Tracepoint name */
2262 /* Input arguments */
2268 TRACEPOINT_EVENT_INSTANCE(
2277 TRACEPOINT_EVENT_INSTANCE(
2290 [[assigning-log-levels]]
2291 ===== Assign a log level to a tracepoint definition
2293 You can assign an optional _log level_ to a
2294 <<defining-tracepoints,tracepoint definition>>.
2296 Assigning different levels of severity to tracepoint definitions can
2297 be useful: when you <<enabling-disabling-events,create an event rule>>,
2298 you can target tracepoints having a log level as severe as a specific
2301 The concept of LTTng-UST log levels is similar to the levels found
2302 in typical logging frameworks:
2304 * In a logging framework, the log level is given by the function
2305 or method name you use at the log statement site: `debug()`,
2306 `info()`, `warn()`, `error()`, and so on.
2307 * In LTTng-UST, you statically assign the log level to a tracepoint
2308 definition; any `tracepoint()` macro invocation which refers to
2309 this definition has this log level.
2311 You can assign a log level to a tracepoint definition with the
2312 `TRACEPOINT_LOGLEVEL()` macro. You must use this macro _after_ the
2313 <<defining-tracepoints,`TRACEPOINT_EVENT()`>> or
2314 <<using-tracepoint-classes,`TRACEPOINT_INSTANCE()`>> macro for a given
2317 The syntax of the `TRACEPOINT_LOGLEVEL()` macro is:
2320 .`TRACEPOINT_LOGLEVEL()` macro syntax.
2322 TRACEPOINT_LOGLEVEL(provider_name, tracepoint_name, log_level)
2327 * `provider_name` with the tracepoint provider name.
2328 * `tracepoint_name` with the tracepoint name.
2329 * `log_level` with the log level to assign to the tracepoint
2330 definition named `tracepoint_name` in the `provider_name`
2331 tracepoint provider.
2333 See man:lttng-ust(3) for a list of available log level names.
2335 .Assign the `TRACE_DEBUG_UNIT` log level to a tracepoint definition.
2339 /* Tracepoint definition */
2348 ctf_integer(int, userid, userid)
2349 ctf_integer(size_t, len, len)
2353 /* Log level assignment */
2354 TRACEPOINT_LOGLEVEL(my_app, get_transaction, TRACE_DEBUG_UNIT)
2360 ===== Create a tracepoint provider package source file
2362 A _tracepoint provider package source file_ is a C source file which
2363 includes a <<tpp-header,tracepoint provider header file>> to expand its
2364 macros into event serialization and other functions.
2366 You can always use the following tracepoint provider package source
2370 .Tracepoint provider package source file template.
2372 #define TRACEPOINT_CREATE_PROBES
2377 Replace `tp.h` with the name of your <<tpp-header,tracepoint provider
2378 header file>> name. You may also include more than one tracepoint
2379 provider header file here to create a tracepoint provider package
2380 holding more than one tracepoint providers.
2383 [[probing-the-application-source-code]]
2384 ==== Add tracepoints to an application's source code
2386 Once you <<tpp-header,create a tracepoint provider header file>>, you
2387 can use the `tracepoint()` macro in your application's
2388 source code to insert the tracepoints that this header
2389 <<defining-tracepoints,defines>>.
2391 The `tracepoint()` macro takes at least two parameters: the tracepoint
2392 provider name and the tracepoint name. The corresponding tracepoint
2393 definition defines the other parameters.
2395 .`tracepoint()` usage.
2397 The following <<defining-tracepoints,tracepoint definition>> defines a
2398 tracepoint which takes two input arguments and has two output event
2402 .Tracepoint provider header file.
2404 #include "my-custom-structure.h"
2411 const char*, cmd_name
2414 ctf_string(cmd_name, cmd_name)
2415 ctf_integer(int, number_of_args, argc)
2420 You can refer to this tracepoint definition with the `tracepoint()`
2421 macro in your application's source code like this:
2424 .Application's source file.
2428 int main(int argc, char* argv[])
2430 tracepoint(my_provider, my_tracepoint, argc, argv[0]);
2436 Note how the application's source code includes
2437 the tracepoint provider header file containing the tracepoint
2438 definitions to use, path:{tp.h}.
2441 .`tracepoint()` usage with a complex tracepoint definition.
2443 Consider this complex tracepoint definition, where multiple event
2444 fields refer to the same input arguments in their argument expression
2448 .Tracepoint provider header file.
2450 /* For `struct stat` */
2451 #include <sys/types.h>
2452 #include <sys/stat.h>
2464 ctf_integer(int, my_constant_field, 23 + 17)
2465 ctf_integer(int, my_int_arg_field, my_int_arg)
2466 ctf_integer(int, my_int_arg_field2, my_int_arg * my_int_arg)
2467 ctf_integer(int, sum4_field, my_str_arg[0] + my_str_arg[1] +
2468 my_str_arg[2] + my_str_arg[3])
2469 ctf_string(my_str_arg_field, my_str_arg)
2470 ctf_integer_hex(off_t, size_field, st->st_size)
2471 ctf_float(double, size_dbl_field, (double) st->st_size)
2472 ctf_sequence_text(char, half_my_str_arg_field, my_str_arg,
2473 size_t, strlen(my_str_arg) / 2)
2478 You can refer to this tracepoint definition with the `tracepoint()`
2479 macro in your application's source code like this:
2482 .Application's source file.
2484 #define TRACEPOINT_DEFINE
2491 stat("/etc/fstab", &s);
2492 tracepoint(my_provider, my_tracepoint, 23, "Hello, World!", &s);
2498 If you look at the event record that LTTng writes when tracing this
2499 program, assuming the file size of path:{/etc/fstab} is 301{nbsp}bytes,
2500 it should look like this:
2502 .Event record fields
2504 |Field's name |Field's value
2505 |`my_constant_field` |40
2506 |`my_int_arg_field` |23
2507 |`my_int_arg_field2` |529
2509 |`my_str_arg_field` |`Hello, World!`
2510 |`size_field` |0x12d
2511 |`size_dbl_field` |301.0
2512 |`half_my_str_arg_field` |`Hello,`
2516 Sometimes, the arguments you pass to `tracepoint()` are expensive to
2517 compute--they use the call stack, for example. To avoid this
2518 computation when the tracepoint is disabled, you can use the
2519 `tracepoint_enabled()` and `do_tracepoint()` macros.
2521 The syntax of the `tracepoint_enabled()` and `do_tracepoint()` macros
2525 .`tracepoint_enabled()` and `do_tracepoint()` macros syntax.
2527 tracepoint_enabled(provider_name, tracepoint_name)
2528 do_tracepoint(provider_name, tracepoint_name, ...)
2533 * `provider_name` with the tracepoint provider name.
2534 * `tracepoint_name` with the tracepoint name.
2536 `tracepoint_enabled()` returns a non-zero value if the tracepoint named
2537 `tracepoint_name` from the provider named `provider_name` is enabled
2540 `do_tracepoint()` is like `tracepoint()`, except that it doesn't check
2541 if the tracepoint is enabled. Using `tracepoint()` with
2542 `tracepoint_enabled()` is dangerous since `tracepoint()` also contains
2543 the `tracepoint_enabled()` check, thus a race condition is
2544 possible in this situation:
2547 .Possible race condition when using `tracepoint_enabled()` with `tracepoint()`.
2549 if (tracepoint_enabled(my_provider, my_tracepoint)) {
2550 stuff = prepare_stuff();
2553 tracepoint(my_provider, my_tracepoint, stuff);
2556 If the tracepoint is enabled after the condition, then `stuff` is not
2557 prepared: the emitted event will either contain wrong data, or the whole
2558 application could crash (segmentation fault, for example).
2560 NOTE: Neither `tracepoint_enabled()` nor `do_tracepoint()` have an
2561 `STAP_PROBEV()` call. If you need it, you must emit
2565 [[building-tracepoint-providers-and-user-application]]
2566 ==== Build and link a tracepoint provider package and an application
2568 Once you have one or more <<tpp-header,tracepoint provider header
2569 files>> and a <<tpp-source,tracepoint provider package source file>>,
2570 you can create the tracepoint provider package by compiling its source
2571 file. From here, multiple build and run scenarios are possible. The
2572 following table shows common application and library configurations
2573 along with the required command lines to achieve them.
2575 In the following diagrams, we use the following file names:
2578 Executable application.
2581 Application's object file.
2584 Tracepoint provider package object file.
2587 Tracepoint provider package archive file.
2590 Tracepoint provider package shared object file.
2593 User library object file.
2596 User library shared object file.
2598 We use the following symbols in the diagrams of table below:
2601 .Symbols used in the build scenario diagrams.
2602 image::ust-sit-symbols.png[]
2604 We assume that path:{.} is part of the env:LD_LIBRARY_PATH environment
2605 variable in the following instructions.
2607 [role="growable ust-scenarios",cols="asciidoc,asciidoc"]
2608 .Common tracepoint provider package scenarios.
2610 |Scenario |Instructions
2613 The instrumented application is statically linked with
2614 the tracepoint provider package object.
2616 image::ust-sit+app-linked-with-tp-o+app-instrumented.png[]
2619 include::../common/ust-sit-step-tp-o.txt[]
2621 To build the instrumented application:
2623 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2628 #define TRACEPOINT_DEFINE
2632 . Compile the application source file:
2641 . Build the application:
2646 $ gcc -o app app.o tpp.o -llttng-ust -ldl
2650 To run the instrumented application:
2652 * Start the application:
2662 The instrumented application is statically linked with the
2663 tracepoint provider package archive file.
2665 image::ust-sit+app-linked-with-tp-a+app-instrumented.png[]
2668 To create the tracepoint provider package archive file:
2670 . Compile the <<tpp-source,tracepoint provider package source file>>:
2679 . Create the tracepoint provider package archive file:
2684 $ ar rcs tpp.a tpp.o
2688 To build the instrumented application:
2690 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2695 #define TRACEPOINT_DEFINE
2699 . Compile the application source file:
2708 . Build the application:
2713 $ gcc -o app app.o tpp.a -llttng-ust -ldl
2717 To run the instrumented application:
2719 * Start the application:
2729 The instrumented application is linked with the tracepoint provider
2730 package shared object.
2732 image::ust-sit+app-linked-with-tp-so+app-instrumented.png[]
2735 include::../common/ust-sit-step-tp-so.txt[]
2737 To build the instrumented application:
2739 . In path:{app.c}, before including path:{tpp.h}, add the following line:
2744 #define TRACEPOINT_DEFINE
2748 . Compile the application source file:
2757 . Build the application:
2762 $ gcc -o app app.o -ldl -L. -ltpp
2766 To run the instrumented application:
2768 * Start the application:
2778 The tracepoint provider package shared object is preloaded before the
2779 instrumented application starts.
2781 image::ust-sit+tp-so-preloaded+app-instrumented.png[]
2784 include::../common/ust-sit-step-tp-so.txt[]
2786 To build the instrumented application:
2788 . In path:{app.c}, before including path:{tpp.h}, add the
2794 #define TRACEPOINT_DEFINE
2795 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
2799 . Compile the application source file:
2808 . Build the application:
2813 $ gcc -o app app.o -ldl
2817 To run the instrumented application with tracing support:
2819 * Preload the tracepoint provider package shared object and
2820 start the application:
2825 $ LD_PRELOAD=./libtpp.so ./app
2829 To run the instrumented application without tracing support:
2831 * Start the application:
2841 The instrumented application dynamically loads the tracepoint provider
2842 package shared object.
2844 image::ust-sit+app-dlopens-tp-so+app-instrumented.png[]
2847 include::../common/ust-sit-step-tp-so.txt[]
2849 To build the instrumented application:
2851 . In path:{app.c}, before including path:{tpp.h}, add the
2857 #define TRACEPOINT_DEFINE
2858 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
2862 . Compile the application source file:
2871 . Build the application:
2876 $ gcc -o app app.o -ldl
2880 To run the instrumented application:
2882 * Start the application:
2892 The application is linked with the instrumented user library.
2894 The instrumented user library is statically linked with the tracepoint
2895 provider package object file.
2897 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-o+lib-instrumented.png[]
2900 include::../common/ust-sit-step-tp-o-fpic.txt[]
2902 To build the instrumented user library:
2904 . In path:{emon.c}, before including path:{tpp.h}, add the
2910 #define TRACEPOINT_DEFINE
2914 . Compile the user library source file:
2919 $ gcc -I. -fpic -c emon.c
2923 . Build the user library shared object:
2928 $ gcc -shared -o libemon.so emon.o tpp.o -llttng-ust -ldl
2932 To build the application:
2934 . Compile the application source file:
2943 . Build the application:
2948 $ gcc -o app app.o -L. -lemon
2952 To run the application:
2954 * Start the application:
2964 The application is linked with the instrumented user library.
2966 The instrumented user library is linked with the tracepoint provider
2967 package shared object.
2969 image::ust-sit+app-linked-with-lib+lib-linked-with-tp-so+lib-instrumented.png[]
2972 include::../common/ust-sit-step-tp-so.txt[]
2974 To build the instrumented user library:
2976 . In path:{emon.c}, before including path:{tpp.h}, add the
2982 #define TRACEPOINT_DEFINE
2986 . Compile the user library source file:
2991 $ gcc -I. -fpic -c emon.c
2995 . Build the user library shared object:
3000 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3004 To build the application:
3006 . Compile the application source file:
3015 . Build the application:
3020 $ gcc -o app app.o -L. -lemon
3024 To run the application:
3026 * Start the application:
3036 The tracepoint provider package shared object is preloaded before the
3039 The application is linked with the instrumented user library.
3041 image::ust-sit+tp-so-preloaded+app-linked-with-lib+lib-instrumented.png[]
3044 include::../common/ust-sit-step-tp-so.txt[]
3046 To build the instrumented user library:
3048 . In path:{emon.c}, before including path:{tpp.h}, add the
3054 #define TRACEPOINT_DEFINE
3055 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3059 . Compile the user library source file:
3064 $ gcc -I. -fpic -c emon.c
3068 . Build the user library shared object:
3073 $ gcc -shared -o libemon.so emon.o -ldl
3077 To build the application:
3079 . Compile the application source file:
3088 . Build the application:
3093 $ gcc -o app app.o -L. -lemon
3097 To run the application with tracing support:
3099 * Preload the tracepoint provider package shared object and
3100 start the application:
3105 $ LD_PRELOAD=./libtpp.so ./app
3109 To run the application without tracing support:
3111 * Start the application:
3121 The application is linked with the instrumented user library.
3123 The instrumented user library dynamically loads the tracepoint provider
3124 package shared object.
3126 image::ust-sit+app-linked-with-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3129 include::../common/ust-sit-step-tp-so.txt[]
3131 To build the instrumented user library:
3133 . In path:{emon.c}, before including path:{tpp.h}, add the
3139 #define TRACEPOINT_DEFINE
3140 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3144 . Compile the user library source file:
3149 $ gcc -I. -fpic -c emon.c
3153 . Build the user library shared object:
3158 $ gcc -shared -o libemon.so emon.o -ldl
3162 To build the application:
3164 . Compile the application source file:
3173 . Build the application:
3178 $ gcc -o app app.o -L. -lemon
3182 To run the application:
3184 * Start the application:
3194 The application dynamically loads the instrumented user library.
3196 The instrumented user library is linked with the tracepoint provider
3197 package shared object.
3199 image::ust-sit+app-dlopens-lib+lib-linked-with-tp-so+lib-instrumented.png[]
3202 include::../common/ust-sit-step-tp-so.txt[]
3204 To build the instrumented user library:
3206 . In path:{emon.c}, before including path:{tpp.h}, add the
3212 #define TRACEPOINT_DEFINE
3216 . Compile the user library source file:
3221 $ gcc -I. -fpic -c emon.c
3225 . Build the user library shared object:
3230 $ gcc -shared -o libemon.so emon.o -ldl -L. -ltpp
3234 To build the application:
3236 . Compile the application source file:
3245 . Build the application:
3250 $ gcc -o app app.o -ldl -L. -lemon
3254 To run the application:
3256 * Start the application:
3266 The application dynamically loads the instrumented user library.
3268 The instrumented user library dynamically loads the tracepoint provider
3269 package shared object.
3271 image::ust-sit+app-dlopens-lib+lib-dlopens-tp-so+lib-instrumented.png[]
3274 include::../common/ust-sit-step-tp-so.txt[]
3276 To build the instrumented user library:
3278 . In path:{emon.c}, before including path:{tpp.h}, add the
3284 #define TRACEPOINT_DEFINE
3285 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3289 . Compile the user library source file:
3294 $ gcc -I. -fpic -c emon.c
3298 . Build the user library shared object:
3303 $ gcc -shared -o libemon.so emon.o -ldl
3307 To build the application:
3309 . Compile the application source file:
3318 . Build the application:
3323 $ gcc -o app app.o -ldl -L. -lemon
3327 To run the application:
3329 * Start the application:
3339 The tracepoint provider package shared object is preloaded before the
3342 The application dynamically loads the instrumented user library.
3344 image::ust-sit+tp-so-preloaded+app-dlopens-lib+lib-instrumented.png[]
3347 include::../common/ust-sit-step-tp-so.txt[]
3349 To build the instrumented user library:
3351 . In path:{emon.c}, before including path:{tpp.h}, add the
3357 #define TRACEPOINT_DEFINE
3358 #define TRACEPOINT_PROBE_DYNAMIC_LINKAGE
3362 . Compile the user library source file:
3367 $ gcc -I. -fpic -c emon.c
3371 . Build the user library shared object:
3376 $ gcc -shared -o libemon.so emon.o -ldl
3380 To build the application:
3382 . Compile the application source file:
3391 . Build the application:
3396 $ gcc -o app app.o -L. -lemon
3400 To run the application with tracing support:
3402 * Preload the tracepoint provider package shared object and
3403 start the application:
3408 $ LD_PRELOAD=./libtpp.so ./app
3412 To run the application without tracing support:
3414 * Start the application:
3424 The application is statically linked with the tracepoint provider
3425 package object file.
3427 The application is linked with the instrumented user library.
3429 image::ust-sit+app-linked-with-tp-o+app-linked-with-lib+lib-instrumented.png[]
3432 include::../common/ust-sit-step-tp-o.txt[]
3434 To build the instrumented user library:
3436 . In path:{emon.c}, before including path:{tpp.h}, add the
3442 #define TRACEPOINT_DEFINE
3446 . Compile the user library source file:
3451 $ gcc -I. -fpic -c emon.c
3455 . Build the user library shared object:
3460 $ gcc -shared -o libemon.so emon.o
3464 To build the application:
3466 . Compile the application source file:
3475 . Build the application:
3480 $ gcc -o app app.o tpp.o -llttng-ust -ldl -L. -lemon
3484 To run the instrumented application:
3486 * Start the application:
3496 The application is statically linked with the tracepoint provider
3497 package object file.
3499 The application dynamically loads the instrumented user library.
3501 image::ust-sit+app-linked-with-tp-o+app-dlopens-lib+lib-instrumented.png[]
3504 include::../common/ust-sit-step-tp-o.txt[]
3506 To build the application:
3508 . In path:{app.c}, before including path:{tpp.h}, add the following line:
3513 #define TRACEPOINT_DEFINE
3517 . Compile the application source file:
3526 . Build the application:
3531 $ gcc -Wl,--export-dynamic -o app app.o tpp.o \
3536 The `--export-dynamic` option passed to the linker is necessary for the
3537 dynamically loaded library to ``see'' the tracepoint symbols defined in
3540 To build the instrumented user library:
3542 . Compile the user library source file:
3547 $ gcc -I. -fpic -c emon.c
3551 . Build the user library shared object:
3556 $ gcc -shared -o libemon.so emon.o
3560 To run the application:
3562 * Start the application:
3573 [[using-lttng-ust-with-daemons]]
3574 ===== Use noch:{LTTng-UST} with daemons
3576 If your instrumented application calls man:fork(2), man:clone(2),
3577 or BSD's man:rfork(2), without a following man:exec(3)-family
3578 system call, you must preload the path:{liblttng-ust-fork.so} shared
3579 object when you start the application.
3583 $ LD_PRELOAD=liblttng-ust-fork.so ./my-app
3586 If your tracepoint provider package is
3587 a shared library which you also preload, you must put both
3588 shared objects in env:LD_PRELOAD:
3592 $ LD_PRELOAD=liblttng-ust-fork.so:/path/to/tp.so ./my-app
3598 ===== Use noch:{LTTng-UST} with applications which close file descriptors that don't belong to them
3600 If your instrumented application closes one or more file descriptors
3601 which it did not open itself, you must preload the
3602 path:{liblttng-ust-fd.so} shared object when you start the application:
3606 $ LD_PRELOAD=liblttng-ust-fd.so ./my-app
3609 Typical use cases include closing all the file descriptors after
3610 man:fork(2) or man:rfork(2) and buggy applications doing
3614 [[lttng-ust-pkg-config]]
3615 ===== Use noch:{pkg-config}
3617 On some distributions, LTTng-UST ships with a
3618 https://www.freedesktop.org/wiki/Software/pkg-config/[pkg-config]
3619 metadata file. If this is your case, then you can use cmd:pkg-config to
3620 build an application on the command line:
3624 $ gcc -o my-app my-app.o tp.o $(pkg-config --cflags --libs lttng-ust)
3628 [[instrumenting-32-bit-app-on-64-bit-system]]
3629 ===== [[advanced-instrumenting-techniques]]Build a 32-bit instrumented application for a 64-bit target system
3631 In order to trace a 32-bit application running on a 64-bit system,
3632 LTTng must use a dedicated 32-bit
3633 <<lttng-consumerd,consumer daemon>>.
3635 The following steps show how to build and install a 32-bit consumer
3636 daemon, which is _not_ part of the default 64-bit LTTng build, how to
3637 build and install the 32-bit LTTng-UST libraries, and how to build and
3638 link an instrumented 32-bit application in that context.
3640 To build a 32-bit instrumented application for a 64-bit target system,
3641 assuming you have a fresh target system with no installed Userspace RCU
3644 . Download, build, and install a 32-bit version of Userspace RCU:
3649 $ cd $(mktemp -d) &&
3650 wget http://lttng.org/files/urcu/userspace-rcu-latest-0.9.tar.bz2 &&
3651 tar -xf userspace-rcu-latest-0.9.tar.bz2 &&
3652 cd userspace-rcu-0.9.* &&
3653 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 &&
3655 sudo make install &&
3660 . Using your distribution's package manager, or from source, install
3661 the following 32-bit versions of the following dependencies of
3662 LTTng-tools and LTTng-UST:
3665 * https://sourceforge.net/projects/libuuid/[libuuid]
3666 * http://directory.fsf.org/wiki/Popt[popt]
3667 * http://www.xmlsoft.org/[libxml2]
3670 . Download, build, and install a 32-bit version of the latest
3671 LTTng-UST{nbsp}{revision}:
3676 $ cd $(mktemp -d) &&
3677 wget http://lttng.org/files/lttng-ust/lttng-ust-latest-2.11.tar.bz2 &&
3678 tar -xf lttng-ust-latest-2.11.tar.bz2 &&
3679 cd lttng-ust-2.11.* &&
3680 ./configure --libdir=/usr/local/lib32 \
3681 CFLAGS=-m32 CXXFLAGS=-m32 \
3682 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' &&
3684 sudo make install &&
3691 Depending on your distribution,
3692 32-bit libraries could be installed at a different location than
3693 `/usr/lib32`. For example, Debian is known to install
3694 some 32-bit libraries in `/usr/lib/i386-linux-gnu`.
3696 In this case, make sure to set `LDFLAGS` to all the
3697 relevant 32-bit library paths, for example:
3701 $ LDFLAGS='-L/usr/lib/i386-linux-gnu -L/usr/lib32'
3705 . Download the latest LTTng-tools{nbsp}{revision}, build, and install
3706 the 32-bit consumer daemon:
3711 $ cd $(mktemp -d) &&
3712 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.11.tar.bz2 &&
3713 tar -xf lttng-tools-latest-2.11.tar.bz2 &&
3714 cd lttng-tools-2.11.* &&
3715 ./configure --libdir=/usr/local/lib32 CFLAGS=-m32 CXXFLAGS=-m32 \
3716 LDFLAGS='-L/usr/local/lib32 -L/usr/lib32' \
3717 --disable-bin-lttng --disable-bin-lttng-crash \
3718 --disable-bin-lttng-relayd --disable-bin-lttng-sessiond &&
3720 cd src/bin/lttng-consumerd &&
3721 sudo make install &&
3726 . From your distribution or from source,
3727 <<installing-lttng,install>> the 64-bit versions of
3728 LTTng-UST and Userspace RCU.
3729 . Download, build, and install the 64-bit version of the
3730 latest LTTng-tools{nbsp}{revision}:
3735 $ cd $(mktemp -d) &&
3736 wget http://lttng.org/files/lttng-tools/lttng-tools-latest-2.11.tar.bz2 &&
3737 tar -xf lttng-tools-latest-2.11.tar.bz2 &&
3738 cd lttng-tools-2.11.* &&
3739 ./configure --with-consumerd32-libdir=/usr/local/lib32 \
3740 --with-consumerd32-bin=/usr/local/lib32/lttng/libexec/lttng-consumerd &&
3742 sudo make install &&
3747 . Pass the following options to man:gcc(1), man:g++(1), or man:clang(1)
3748 when linking your 32-bit application:
3751 -m32 -L/usr/lib32 -L/usr/local/lib32 \
3752 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32
3755 For example, let's rebuild the quick start example in
3756 <<tracing-your-own-user-application,Trace a user application>> as an
3757 instrumented 32-bit application:
3762 $ gcc -m32 -c -I. hello-tp.c
3763 $ gcc -m32 -c hello.c
3764 $ gcc -m32 -o hello hello.o hello-tp.o \
3765 -L/usr/lib32 -L/usr/local/lib32 \
3766 -Wl,-rpath,/usr/lib32,-rpath,/usr/local/lib32 \
3771 No special action is required to execute the 32-bit application and
3772 to trace it: use the command-line man:lttng(1) tool as usual.
3779 man:tracef(3) is a small LTTng-UST API designed for quick,
3780 man:printf(3)-like instrumentation without the burden of
3781 <<tracepoint-provider,creating>> and
3782 <<building-tracepoint-providers-and-user-application,building>>
3783 a tracepoint provider package.
3785 To use `tracef()` in your application:
3787 . In the C or C++ source files where you need to use `tracef()`,
3788 include `<lttng/tracef.h>`:
3793 #include <lttng/tracef.h>
3797 . In the application's source code, use `tracef()` like you would use
3805 tracef("my message: %d (%s)", my_integer, my_string);
3811 . Link your application with `liblttng-ust`:
3816 $ gcc -o app app.c -llttng-ust
3820 To trace the events that `tracef()` calls emit:
3822 * <<enabling-disabling-events,Create an event rule>> which matches the
3823 `lttng_ust_tracef:*` event name:
3828 $ lttng enable-event --userspace 'lttng_ust_tracef:*'
3833 .Limitations of `tracef()`
3835 The `tracef()` utility function was developed to make user space tracing
3836 super simple, albeit with notable disadvantages compared to
3837 <<defining-tracepoints,user-defined tracepoints>>:
3839 * All the emitted events have the same tracepoint provider and
3840 tracepoint names, respectively `lttng_ust_tracef` and `event`.
3841 * There is no static type checking.
3842 * The only event record field you actually get, named `msg`, is a string
3843 potentially containing the values you passed to `tracef()`
3844 using your own format string. This also means that you cannot filter
3845 events with a custom expression at run time because there are no
3847 * Since `tracef()` uses the C standard library's man:vasprintf(3)
3848 function behind the scenes to format the strings at run time, its
3849 expected performance is lower than with user-defined tracepoints,
3850 which do not require a conversion to a string.
3852 Taking this into consideration, `tracef()` is useful for some quick
3853 prototyping and debugging, but you should not consider it for any
3854 permanent and serious applicative instrumentation.
3860 ==== Use `tracelog()`
3862 The man:tracelog(3) API is very similar to <<tracef,`tracef()`>>, with
3863 the difference that it accepts an additional log level parameter.
3865 The goal of `tracelog()` is to ease the migration from logging to
3868 To use `tracelog()` in your application:
3870 . In the C or C++ source files where you need to use `tracelog()`,
3871 include `<lttng/tracelog.h>`:
3876 #include <lttng/tracelog.h>
3880 . In the application's source code, use `tracelog()` like you would use
3881 man:printf(3), except for the first parameter which is the log
3889 tracelog(TRACE_WARNING, "my message: %d (%s)",
3890 my_integer, my_string);
3896 See man:lttng-ust(3) for a list of available log level names.
3898 . Link your application with `liblttng-ust`:
3903 $ gcc -o app app.c -llttng-ust
3907 To trace the events that `tracelog()` calls emit with a log level
3908 _as severe as_ a specific log level:
3910 * <<enabling-disabling-events,Create an event rule>> which matches the
3911 `lttng_ust_tracelog:*` event name and a minimum level
3917 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
3918 --loglevel=TRACE_WARNING
3922 To trace the events that `tracelog()` calls emit with a
3923 _specific log level_:
3925 * Create an event rule which matches the `lttng_ust_tracelog:*`
3926 event name and a specific log level:
3931 $ lttng enable-event --userspace 'lttng_ust_tracelog:*'
3932 --loglevel-only=TRACE_INFO
3937 [[prebuilt-ust-helpers]]
3938 === Prebuilt user space tracing helpers
3940 The LTTng-UST package provides a few helpers in the form or preloadable
3941 shared objects which automatically instrument system functions and
3944 The helper shared objects are normally found in dir:{/usr/lib}. If you
3945 built LTTng-UST <<building-from-source,from source>>, they are probably
3946 located in dir:{/usr/local/lib}.
3948 The installed user space tracing helpers in LTTng-UST{nbsp}{revision}
3951 path:{liblttng-ust-libc-wrapper.so}::
3952 path:{liblttng-ust-pthread-wrapper.so}::
3953 <<liblttng-ust-libc-pthread-wrapper,C{nbsp}standard library
3954 memory and POSIX threads function tracing>>.
3956 path:{liblttng-ust-cyg-profile.so}::
3957 path:{liblttng-ust-cyg-profile-fast.so}::
3958 <<liblttng-ust-cyg-profile,Function entry and exit tracing>>.
3960 path:{liblttng-ust-dl.so}::
3961 <<liblttng-ust-dl,Dynamic linker tracing>>.
3963 To use a user space tracing helper with any user application:
3965 * Preload the helper shared object when you start the application:
3970 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so my-app
3974 You can preload more than one helper:
3979 $ LD_PRELOAD=liblttng-ust-libc-wrapper.so:liblttng-ust-dl.so my-app
3985 [[liblttng-ust-libc-pthread-wrapper]]
3986 ==== Instrument C standard library memory and POSIX threads functions
3988 The path:{liblttng-ust-libc-wrapper.so} and
3989 path:{liblttng-ust-pthread-wrapper.so} helpers
3990 add instrumentation to some C standard library and POSIX
3994 .Functions instrumented by preloading path:{liblttng-ust-libc-wrapper.so}.
3996 |TP provider name |TP name |Instrumented function
3998 .6+|`lttng_ust_libc` |`malloc` |man:malloc(3)
3999 |`calloc` |man:calloc(3)
4000 |`realloc` |man:realloc(3)
4001 |`free` |man:free(3)
4002 |`memalign` |man:memalign(3)
4003 |`posix_memalign` |man:posix_memalign(3)
4007 .Functions instrumented by preloading path:{liblttng-ust-pthread-wrapper.so}.
4009 |TP provider name |TP name |Instrumented function
4011 .4+|`lttng_ust_pthread` |`pthread_mutex_lock_req` |man:pthread_mutex_lock(3p) (request time)
4012 |`pthread_mutex_lock_acq` |man:pthread_mutex_lock(3p) (acquire time)
4013 |`pthread_mutex_trylock` |man:pthread_mutex_trylock(3p)
4014 |`pthread_mutex_unlock` |man:pthread_mutex_unlock(3p)
4017 When you preload the shared object, it replaces the functions listed
4018 in the previous tables by wrappers which contain tracepoints and call
4019 the replaced functions.
4022 [[liblttng-ust-cyg-profile]]
4023 ==== Instrument function entry and exit
4025 The path:{liblttng-ust-cyg-profile*.so} helpers can add instrumentation
4026 to the entry and exit points of functions.
4028 man:gcc(1) and man:clang(1) have an option named
4029 https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html[`-finstrument-functions`]
4030 which generates instrumentation calls for entry and exit to functions.
4031 The LTTng-UST function tracing helpers,
4032 path:{liblttng-ust-cyg-profile.so} and
4033 path:{liblttng-ust-cyg-profile-fast.so}, take advantage of this feature
4034 to add tracepoints to the two generated functions (which contain
4035 `cyg_profile` in their names, hence the helper's name).
4037 To use the LTTng-UST function tracing helper, the source files to
4038 instrument must be built using the `-finstrument-functions` compiler
4041 There are two versions of the LTTng-UST function tracing helper:
4043 * **path:{liblttng-ust-cyg-profile-fast.so}** is a lightweight variant
4044 that you should only use when it can be _guaranteed_ that the
4045 complete event stream is recorded without any lost event record.
4046 Any kind of duplicate information is left out.
4048 Assuming no event record is lost, having only the function addresses on
4049 entry is enough to create a call graph, since an event record always
4050 contains the ID of the CPU that generated it.
4052 You can use a tool like man:addr2line(1) to convert function addresses
4053 back to source file names and line numbers.
4055 * **path:{liblttng-ust-cyg-profile.so}** is a more robust variant
4056 which also works in use cases where event records might get discarded or
4057 not recorded from application startup.
4058 In these cases, the trace analyzer needs more information to be
4059 able to reconstruct the program flow.
4061 See man:lttng-ust-cyg-profile(3) to learn more about the instrumentation
4062 points of this helper.
4064 All the tracepoints that this helper provides have the
4065 log level `TRACE_DEBUG_FUNCTION` (see man:lttng-ust(3)).
4067 TIP: It's sometimes a good idea to limit the number of source files that
4068 you compile with the `-finstrument-functions` option to prevent LTTng
4069 from writing an excessive amount of trace data at run time. When using
4070 man:gcc(1), you can use the
4071 `-finstrument-functions-exclude-function-list` option to avoid
4072 instrument entries and exits of specific function names.
4077 ==== Instrument the dynamic linker
4079 The path:{liblttng-ust-dl.so} helper adds instrumentation to the
4080 man:dlopen(3) and man:dlclose(3) function calls.
4082 See man:lttng-ust-dl(3) to learn more about the instrumentation points
4087 [[java-application]]
4088 === User space Java agent
4090 You can instrument any Java application which uses one of the following
4093 * The https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[**`java.util.logging`**]
4094 (JUL) core logging facilities.
4095 * http://logging.apache.org/log4j/1.2/[**Apache log4j{nbsp}1.2**], since
4096 LTTng{nbsp}2.6. Note that Apache Log4j{nbsp}2 is not supported.
4099 .LTTng-UST Java agent imported by a Java application.
4100 image::java-app.png[]
4102 Note that the methods described below are new in LTTng{nbsp}{revision}.
4103 Previous LTTng versions use another technique.
4105 NOTE: We use http://openjdk.java.net/[OpenJDK]{nbsp}8 for development
4106 and https://ci.lttng.org/[continuous integration], thus this version is
4107 directly supported. However, the LTTng-UST Java agent is also tested
4108 with OpenJDK{nbsp}7.
4113 ==== Use the LTTng-UST Java agent for `java.util.logging`
4115 To use the LTTng-UST Java agent in a Java application which uses
4116 `java.util.logging` (JUL):
4118 . In the Java application's source code, import the LTTng-UST
4119 log handler package for `java.util.logging`:
4124 import org.lttng.ust.agent.jul.LttngLogHandler;
4128 . Create an LTTng-UST JUL log handler:
4133 Handler lttngUstLogHandler = new LttngLogHandler();
4137 . Add this handler to the JUL loggers which should emit LTTng events:
4142 Logger myLogger = Logger.getLogger("some-logger");
4144 myLogger.addHandler(lttngUstLogHandler);
4148 . Use `java.util.logging` log statements and configuration as usual.
4149 The loggers with an attached LTTng-UST log handler can emit
4152 . Before exiting the application, remove the LTTng-UST log handler from
4153 the loggers attached to it and call its `close()` method:
4158 myLogger.removeHandler(lttngUstLogHandler);
4159 lttngUstLogHandler.close();
4163 This is not strictly necessary, but it is recommended for a clean
4164 disposal of the handler's resources.
4166 . Include the LTTng-UST Java agent's common and JUL-specific JAR files,
4167 path:{lttng-ust-agent-common.jar} and path:{lttng-ust-agent-jul.jar},
4169 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4170 path] when you build the Java application.
4172 The JAR files are typically located in dir:{/usr/share/java}.
4174 IMPORTANT: The LTTng-UST Java agent must be
4175 <<installing-lttng,installed>> for the logging framework your
4178 .Use the LTTng-UST Java agent for `java.util.logging`.
4183 import java.io.IOException;
4184 import java.util.logging.Handler;
4185 import java.util.logging.Logger;
4186 import org.lttng.ust.agent.jul.LttngLogHandler;
4190 private static final int answer = 42;
4192 public static void main(String[] argv) throws Exception
4195 Logger logger = Logger.getLogger("jello");
4197 // Create an LTTng-UST log handler
4198 Handler lttngUstLogHandler = new LttngLogHandler();
4200 // Add the LTTng-UST log handler to our logger
4201 logger.addHandler(lttngUstLogHandler);
4204 logger.info("some info");
4205 logger.warning("some warning");
4207 logger.finer("finer information; the answer is " + answer);
4209 logger.severe("error!");
4211 // Not mandatory, but cleaner
4212 logger.removeHandler(lttngUstLogHandler);
4213 lttngUstLogHandler.close();
4222 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4225 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4226 <<enabling-disabling-events,create an event rule>> matching the
4227 `jello` JUL logger, and <<basic-tracing-session-control,start tracing>>:
4232 $ lttng enable-event --jul jello
4236 Run the compiled class:
4240 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4243 <<basic-tracing-session-control,Stop tracing>> and inspect the
4253 In the resulting trace, an <<event,event record>> generated by a Java
4254 application using `java.util.logging` is named `lttng_jul:event` and
4255 has the following fields:
4258 Log record's message.
4264 Name of the class in which the log statement was executed.
4267 Name of the method in which the log statement was executed.
4270 Logging time (timestamp in milliseconds).
4273 Log level integer value.
4276 ID of the thread in which the log statement was executed.
4278 You can use the opt:lttng-enable-event(1):--loglevel or
4279 opt:lttng-enable-event(1):--loglevel-only option of the
4280 man:lttng-enable-event(1) command to target a range of JUL log levels
4281 or a specific JUL log level.
4286 ==== Use the LTTng-UST Java agent for Apache log4j
4288 To use the LTTng-UST Java agent in a Java application which uses
4289 Apache log4j{nbsp}1.2:
4291 . In the Java application's source code, import the LTTng-UST
4292 log appender package for Apache log4j:
4297 import org.lttng.ust.agent.log4j.LttngLogAppender;
4301 . Create an LTTng-UST log4j log appender:
4306 Appender lttngUstLogAppender = new LttngLogAppender();
4310 . Add this appender to the log4j loggers which should emit LTTng events:
4315 Logger myLogger = Logger.getLogger("some-logger");
4317 myLogger.addAppender(lttngUstLogAppender);
4321 . Use Apache log4j log statements and configuration as usual. The
4322 loggers with an attached LTTng-UST log appender can emit LTTng events.
4324 . Before exiting the application, remove the LTTng-UST log appender from
4325 the loggers attached to it and call its `close()` method:
4330 myLogger.removeAppender(lttngUstLogAppender);
4331 lttngUstLogAppender.close();
4335 This is not strictly necessary, but it is recommended for a clean
4336 disposal of the appender's resources.
4338 . Include the LTTng-UST Java agent's common and log4j-specific JAR
4339 files, path:{lttng-ust-agent-common.jar} and
4340 path:{lttng-ust-agent-log4j.jar}, in the
4341 https://docs.oracle.com/javase/tutorial/essential/environment/paths.html[class
4342 path] when you build the Java application.
4344 The JAR files are typically located in dir:{/usr/share/java}.
4346 IMPORTANT: The LTTng-UST Java agent must be
4347 <<installing-lttng,installed>> for the logging framework your
4350 .Use the LTTng-UST Java agent for Apache log4j.
4355 import org.apache.log4j.Appender;
4356 import org.apache.log4j.Logger;
4357 import org.lttng.ust.agent.log4j.LttngLogAppender;
4361 private static final int answer = 42;
4363 public static void main(String[] argv) throws Exception
4366 Logger logger = Logger.getLogger("jello");
4368 // Create an LTTng-UST log appender
4369 Appender lttngUstLogAppender = new LttngLogAppender();
4371 // Add the LTTng-UST log appender to our logger
4372 logger.addAppender(lttngUstLogAppender);
4375 logger.info("some info");
4376 logger.warn("some warning");
4378 logger.debug("debug information; the answer is " + answer);
4380 logger.fatal("error!");
4382 // Not mandatory, but cleaner
4383 logger.removeAppender(lttngUstLogAppender);
4384 lttngUstLogAppender.close();
4390 Build this example (`$LOG4JPATH` is the path to the Apache log4j JAR
4395 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH Test.java
4398 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4399 <<enabling-disabling-events,create an event rule>> matching the
4400 `jello` log4j logger, and <<basic-tracing-session-control,start tracing>>:
4405 $ lttng enable-event --log4j jello
4409 Run the compiled class:
4413 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-log4j.jar:$LOG4JPATH:. Test
4416 <<basic-tracing-session-control,Stop tracing>> and inspect the
4426 In the resulting trace, an <<event,event record>> generated by a Java
4427 application using log4j is named `lttng_log4j:event` and
4428 has the following fields:
4431 Log record's message.
4437 Name of the class in which the log statement was executed.
4440 Name of the method in which the log statement was executed.
4443 Name of the file in which the executed log statement is located.
4446 Line number at which the log statement was executed.
4452 Log level integer value.
4455 Name of the Java thread in which the log statement was executed.
4457 You can use the opt:lttng-enable-event(1):--loglevel or
4458 opt:lttng-enable-event(1):--loglevel-only option of the
4459 man:lttng-enable-event(1) command to target a range of Apache log4j log levels
4460 or a specific log4j log level.
4464 [[java-application-context]]
4465 ==== Provide application-specific context fields in a Java application
4467 A Java application-specific context field is a piece of state provided
4468 by the application which <<adding-context,you can add>>, using the
4469 man:lttng-add-context(1) command, to each <<event,event record>>
4470 produced by the log statements of this application.
4472 For example, a given object might have a current request ID variable.
4473 You can create a context information retriever for this object and
4474 assign a name to this current request ID. You can then, using the
4475 man:lttng-add-context(1) command, add this context field by name to
4476 the JUL or log4j <<channel,channel>>.
4478 To provide application-specific context fields in a Java application:
4480 . In the Java application's source code, import the LTTng-UST
4481 Java agent context classes and interfaces:
4486 import org.lttng.ust.agent.context.ContextInfoManager;
4487 import org.lttng.ust.agent.context.IContextInfoRetriever;
4491 . Create a context information retriever class, that is, a class which
4492 implements the `IContextInfoRetriever` interface:
4497 class MyContextInfoRetriever implements IContextInfoRetriever
4500 public Object retrieveContextInfo(String key)
4502 if (key.equals("intCtx")) {
4504 } else if (key.equals("strContext")) {
4505 return "context value!";
4514 This `retrieveContextInfo()` method is the only member of the
4515 `IContextInfoRetriever` interface. Its role is to return the current
4516 value of a state by name to create a context field. The names of the
4517 context fields and which state variables they return depends on your
4520 All primitive types and objects are supported as context fields.
4521 When `retrieveContextInfo()` returns an object, the context field
4522 serializer calls its `toString()` method to add a string field to
4523 event records. The method can also return `null`, which means that
4524 no context field is available for the required name.
4526 . Register an instance of your context information retriever class to
4527 the context information manager singleton:
4532 IContextInfoRetriever cir = new MyContextInfoRetriever();
4533 ContextInfoManager cim = ContextInfoManager.getInstance();
4534 cim.registerContextInfoRetriever("retrieverName", cir);
4538 . Before exiting the application, remove your context information
4539 retriever from the context information manager singleton:
4544 ContextInfoManager cim = ContextInfoManager.getInstance();
4545 cim.unregisterContextInfoRetriever("retrieverName");
4549 This is not strictly necessary, but it is recommended for a clean
4550 disposal of some manager's resources.
4552 . Build your Java application with LTTng-UST Java agent support as
4553 usual, following the procedure for either the <<jul,JUL>> or
4554 <<log4j,Apache log4j>> framework.
4557 .Provide application-specific context fields in a Java application.
4562 import java.util.logging.Handler;
4563 import java.util.logging.Logger;
4564 import org.lttng.ust.agent.jul.LttngLogHandler;
4565 import org.lttng.ust.agent.context.ContextInfoManager;
4566 import org.lttng.ust.agent.context.IContextInfoRetriever;
4570 // Our context information retriever class
4571 private static class MyContextInfoRetriever
4572 implements IContextInfoRetriever
4575 public Object retrieveContextInfo(String key) {
4576 if (key.equals("intCtx")) {
4578 } else if (key.equals("strContext")) {
4579 return "context value!";
4586 private static final int answer = 42;
4588 public static void main(String args[]) throws Exception
4590 // Get the context information manager instance
4591 ContextInfoManager cim = ContextInfoManager.getInstance();
4593 // Create and register our context information retriever
4594 IContextInfoRetriever cir = new MyContextInfoRetriever();
4595 cim.registerContextInfoRetriever("myRetriever", cir);
4598 Logger logger = Logger.getLogger("jello");
4600 // Create an LTTng-UST log handler
4601 Handler lttngUstLogHandler = new LttngLogHandler();
4603 // Add the LTTng-UST log handler to our logger
4604 logger.addHandler(lttngUstLogHandler);
4607 logger.info("some info");
4608 logger.warning("some warning");
4610 logger.finer("finer information; the answer is " + answer);
4612 logger.severe("error!");
4614 // Not mandatory, but cleaner
4615 logger.removeHandler(lttngUstLogHandler);
4616 lttngUstLogHandler.close();
4617 cim.unregisterContextInfoRetriever("myRetriever");
4626 $ javac -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar Test.java
4629 <<creating-destroying-tracing-sessions,Create a tracing session>>
4630 and <<enabling-disabling-events,create an event rule>> matching the
4636 $ lttng enable-event --jul jello
4639 <<adding-context,Add the application-specific context fields>> to the
4644 $ lttng add-context --jul --type='$app.myRetriever:intCtx'
4645 $ lttng add-context --jul --type='$app.myRetriever:strContext'
4648 <<basic-tracing-session-control,Start tracing>>:
4655 Run the compiled class:
4659 $ java -cp /usr/share/java/jarpath/lttng-ust-agent-common.jar:/usr/share/java/jarpath/lttng-ust-agent-jul.jar:. Test
4662 <<basic-tracing-session-control,Stop tracing>> and inspect the
4674 [[python-application]]
4675 === User space Python agent
4677 You can instrument a Python{nbsp}2 or Python{nbsp}3 application which
4679 https://docs.python.org/3/library/logging.html[`logging`] package.
4681 Each log statement emits an LTTng event once the
4682 application module imports the
4683 <<lttng-ust-agents,LTTng-UST Python agent>> package.
4686 .A Python application importing the LTTng-UST Python agent.
4687 image::python-app.png[]
4689 To use the LTTng-UST Python agent:
4691 . In the Python application's source code, import the LTTng-UST Python
4701 The LTTng-UST Python agent automatically adds its logging handler to the
4702 root logger at import time.
4704 Any log statement that the application executes before this import does
4705 not emit an LTTng event.
4707 IMPORTANT: The LTTng-UST Python agent must be
4708 <<installing-lttng,installed>>.
4710 . Use log statements and logging configuration as usual.
4711 Since the LTTng-UST Python agent adds a handler to the _root_
4712 logger, you can trace any log statement from any logger.
4714 .Use the LTTng-UST Python agent.
4725 logging.basicConfig()
4726 logger = logging.getLogger('my-logger')
4729 logger.debug('debug message')
4730 logger.info('info message')
4731 logger.warn('warn message')
4732 logger.error('error message')
4733 logger.critical('critical message')
4737 if __name__ == '__main__':
4741 NOTE: `logging.basicConfig()`, which adds to the root logger a basic
4742 logging handler which prints to the standard error stream, is not
4743 strictly required for LTTng-UST tracing to work, but in versions of
4744 Python preceding{nbsp}3.2, you could see a warning message which indicates
4745 that no handler exists for the logger `my-logger`.
4747 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4748 <<enabling-disabling-events,create an event rule>> matching the
4749 `my-logger` Python logger, and <<basic-tracing-session-control,start
4755 $ lttng enable-event --python my-logger
4759 Run the Python script:
4766 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
4776 In the resulting trace, an <<event,event record>> generated by a Python
4777 application is named `lttng_python:event` and has the following fields:
4780 Logging time (string).
4783 Log record's message.
4789 Name of the function in which the log statement was executed.
4792 Line number at which the log statement was executed.
4795 Log level integer value.
4798 ID of the Python thread in which the log statement was executed.
4801 Name of the Python thread in which the log statement was executed.
4803 You can use the opt:lttng-enable-event(1):--loglevel or
4804 opt:lttng-enable-event(1):--loglevel-only option of the
4805 man:lttng-enable-event(1) command to target a range of Python log levels
4806 or a specific Python log level.
4808 When an application imports the LTTng-UST Python agent, the agent tries
4809 to register to a <<lttng-sessiond,session daemon>>. Note that you must
4810 <<start-sessiond,start the session daemon>> _before_ you run the Python
4811 application. If a session daemon is found, the agent tries to register
4812 to it during five seconds, after which the application continues
4813 without LTTng tracing support. You can override this timeout value with
4814 the env:LTTNG_UST_PYTHON_REGISTER_TIMEOUT environment variable
4817 If the session daemon stops while a Python application with an imported
4818 LTTng-UST Python agent runs, the agent retries to connect and to
4819 register to a session daemon every three seconds. You can override this
4820 delay with the env:LTTNG_UST_PYTHON_REGISTER_RETRY_DELAY environment
4825 [[proc-lttng-logger-abi]]
4828 The `lttng-tracer` Linux kernel module, part of
4829 <<lttng-modules,LTTng-modules>>, creates the special LTTng logger file
4830 path:{/proc/lttng-logger} when it's loaded. Any application can write
4831 text data to this file to emit an LTTng event.
4834 .An application writes to the LTTng logger file to emit an LTTng event.
4835 image::lttng-logger.png[]
4837 The LTTng logger is the quickest method--not the most efficient,
4838 however--to add instrumentation to an application. It is designed
4839 mostly to instrument shell scripts:
4843 $ echo "Some message, some $variable" > /proc/lttng-logger
4846 Any event that the LTTng logger emits is named `lttng_logger` and
4847 belongs to the Linux kernel <<domain,tracing domain>>. However, unlike
4848 other instrumentation points in the kernel tracing domain, **any Unix
4849 user** can <<enabling-disabling-events,create an event rule>> which
4850 matches its event name, not only the root user or users in the
4851 <<tracing-group,tracing group>>.
4853 To use the LTTng logger:
4855 * From any application, write text data to the path:{/proc/lttng-logger}
4858 The `msg` field of `lttng_logger` event records contains the
4861 NOTE: The maximum message length of an LTTng logger event is
4862 1024{nbsp}bytes. Writing more than this makes the LTTng logger emit more
4863 than one event to contain the remaining data.
4865 You should not use the LTTng logger to trace a user application which
4866 can be instrumented in a more efficient way, namely:
4868 * <<c-application,C and $$C++$$ applications>>.
4869 * <<java-application,Java applications>>.
4870 * <<python-application,Python applications>>.
4872 .Use the LTTng logger.
4877 echo 'Hello, World!' > /proc/lttng-logger
4879 df --human-readable --print-type / > /proc/lttng-logger
4882 <<creating-destroying-tracing-sessions,Create a tracing session>>,
4883 <<enabling-disabling-events,create an event rule>> matching the
4884 `lttng_logger` Linux kernel tracepoint, and
4885 <<basic-tracing-session-control,start tracing>>:
4890 $ lttng enable-event --kernel lttng_logger
4894 Run the Bash script:
4901 <<basic-tracing-session-control,Stop tracing>> and inspect the recorded
4912 [[instrumenting-linux-kernel]]
4913 === LTTng kernel tracepoints
4915 NOTE: This section shows how to _add_ instrumentation points to the
4916 Linux kernel. The kernel's subsystems are already thoroughly
4917 instrumented at strategic places for LTTng when you
4918 <<installing-lttng,install>> the <<lttng-modules,LTTng-modules>>
4922 There are two methods to instrument the Linux kernel:
4924 . <<linux-add-lttng-layer,Add an LTTng layer>> over an existing ftrace
4925 tracepoint which uses the `TRACE_EVENT()` API.
4927 Choose this if you want to instrumentation a Linux kernel tree with an
4928 instrumentation point compatible with ftrace, perf, and SystemTap.
4930 . Use an <<linux-lttng-tracepoint-event,LTTng-only approach>> to
4931 instrument an out-of-tree kernel module.
4933 Choose this if you don't need ftrace, perf, or SystemTap support.
4937 [[linux-add-lttng-layer]]
4938 ==== [[instrumenting-linux-kernel-itself]][[mainline-trace-event]][[lttng-adaptation-layer]]Add an LTTng layer to an existing ftrace tracepoint
4940 This section shows how to add an LTTng layer to existing ftrace
4941 instrumentation using the `TRACE_EVENT()` API.
4943 This section does not document the `TRACE_EVENT()` macro. You can
4944 read the following articles to learn more about this API:
4946 * http://lwn.net/Articles/379903/[Using the TRACE_EVENT() macro (Part{nbsp}1)]
4947 * http://lwn.net/Articles/381064/[Using the TRACE_EVENT() macro (Part{nbsp}2)]
4948 * http://lwn.net/Articles/383362/[Using the TRACE_EVENT() macro (Part{nbsp}3)]
4950 The following procedure assumes that your ftrace tracepoints are
4951 correctly defined in their own header and that they are created in
4952 one source file using the `CREATE_TRACE_POINTS` definition.
4954 To add an LTTng layer over an existing ftrace tracepoint:
4956 . Make sure the following kernel configuration options are
4962 * `CONFIG_HIGH_RES_TIMERS`
4963 * `CONFIG_TRACEPOINTS`
4966 . Build the Linux source tree with your custom ftrace tracepoints.
4967 . Boot the resulting Linux image on your target system.
4969 Confirm that the tracepoints exist by looking for their names in the
4970 dir:{/sys/kernel/debug/tracing/events/subsys} directory, where `subsys`
4971 is your subsystem's name.
4973 . Get a copy of the latest LTTng-modules{nbsp}{revision}:
4978 $ cd $(mktemp -d) &&
4979 wget http://lttng.org/files/lttng-modules/lttng-modules-latest-2.11.tar.bz2 &&
4980 tar -xf lttng-modules-latest-2.11.tar.bz2 &&
4981 cd lttng-modules-2.11.*
4985 . In dir:{instrumentation/events/lttng-module}, relative to the root
4986 of the LTTng-modules source tree, create a header file named
4987 +__subsys__.h+ for your custom subsystem +__subsys__+ and write your
4988 LTTng-modules tracepoint definitions using the LTTng-modules
4991 Start with this template:
4995 .path:{instrumentation/events/lttng-module/my_subsys.h}
4998 #define TRACE_SYSTEM my_subsys
5000 #if !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ)
5001 #define _LTTNG_MY_SUBSYS_H
5003 #include "../../../probes/lttng-tracepoint-event.h"
5004 #include <linux/tracepoint.h>
5006 LTTNG_TRACEPOINT_EVENT(
5008 * Format is identical to TRACE_EVENT()'s version for the three
5009 * following macro parameters:
5012 TP_PROTO(int my_int, const char *my_string),
5013 TP_ARGS(my_int, my_string),
5015 /* LTTng-modules specific macros */
5017 ctf_integer(int, my_int_field, my_int)
5018 ctf_string(my_bar_field, my_bar)
5022 #endif /* !defined(_LTTNG_MY_SUBSYS_H) || defined(TRACE_HEADER_MULTI_READ) */
5024 #include "../../../probes/define_trace.h"
5028 The entries in the `TP_FIELDS()` section are the list of fields for the
5029 LTTng tracepoint. This is similar to the `TP_STRUCT__entry()` part of
5030 ftrace's `TRACE_EVENT()` macro.
5032 See <<lttng-modules-tp-fields,Tracepoint fields macros>> for a
5033 complete description of the available `ctf_*()` macros.
5035 . Create the LTTng-modules probe's kernel module C source file,
5036 +probes/lttng-probe-__subsys__.c+, where +__subsys__+ is your
5041 .path:{probes/lttng-probe-my-subsys.c}
5043 #include <linux/module.h>
5044 #include "../lttng-tracer.h"
5047 * Build-time verification of mismatch between mainline
5048 * TRACE_EVENT() arguments and the LTTng-modules adaptation
5049 * layer LTTNG_TRACEPOINT_EVENT() arguments.
5051 #include <trace/events/my_subsys.h>
5053 /* Create LTTng tracepoint probes */
5054 #define LTTNG_PACKAGE_BUILD
5055 #define CREATE_TRACE_POINTS
5056 #define TRACE_INCLUDE_PATH ../instrumentation/events/lttng-module
5058 #include "../instrumentation/events/lttng-module/my_subsys.h"
5060 MODULE_LICENSE("GPL and additional rights");
5061 MODULE_AUTHOR("Your name <your-email>");
5062 MODULE_DESCRIPTION("LTTng my_subsys probes");
5063 MODULE_VERSION(__stringify(LTTNG_MODULES_MAJOR_VERSION) "."
5064 __stringify(LTTNG_MODULES_MINOR_VERSION) "."
5065 __stringify(LTTNG_MODULES_PATCHLEVEL_VERSION)
5066 LTTNG_MODULES_EXTRAVERSION);
5070 . Edit path:{probes/KBuild} and add your new kernel module object
5071 next to the existing ones:
5075 .path:{probes/KBuild}
5079 obj-m += lttng-probe-module.o
5080 obj-m += lttng-probe-power.o
5082 obj-m += lttng-probe-my-subsys.o
5088 . Build and install the LTTng kernel modules:
5093 $ make KERNELDIR=/path/to/linux
5094 # make modules_install && depmod -a
5098 Replace `/path/to/linux` with the path to the Linux source tree where
5099 you defined and used tracepoints with ftrace's `TRACE_EVENT()` macro.
5101 Note that you can also use the
5102 <<lttng-tracepoint-event-code,`LTTNG_TRACEPOINT_EVENT_CODE()` macro>>
5103 instead of `LTTNG_TRACEPOINT_EVENT()` to use custom local variables and
5104 C code that need to be executed before the event fields are recorded.
5106 The best way to learn how to use the previous LTTng-modules macros is to
5107 inspect the existing LTTng-modules tracepoint definitions in the
5108 dir:{instrumentation/events/lttng-module} header files. Compare them
5109 with the Linux kernel mainline versions in the
5110 dir:{include/trace/events} directory of the Linux source tree.
5114 [[lttng-tracepoint-event-code]]
5115 ===== Use custom C code to access the data for tracepoint fields
5117 Although we recommended to always use the
5118 <<lttng-adaptation-layer,`LTTNG_TRACEPOINT_EVENT()`>> macro to describe
5119 the arguments and fields of an LTTng-modules tracepoint when possible,
5120 sometimes you need a more complex process to access the data that the
5121 tracer records as event record fields. In other words, you need local
5122 variables and multiple C{nbsp}statements instead of simple
5123 argument-based expressions that you pass to the
5124 <<lttng-modules-tp-fields,`ctf_*()` macros of `TP_FIELDS()`>>.
5126 You can use the `LTTNG_TRACEPOINT_EVENT_CODE()` macro instead of
5127 `LTTNG_TRACEPOINT_EVENT()` to declare custom local variables and define
5128 a block of C{nbsp}code to be executed before LTTng records the fields.
5129 The structure of this macro is:
5132 .`LTTNG_TRACEPOINT_EVENT_CODE()` macro syntax.
5134 LTTNG_TRACEPOINT_EVENT_CODE(
5136 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5137 * version for the following three macro parameters:
5140 TP_PROTO(int my_int, const char *my_string),
5141 TP_ARGS(my_int, my_string),
5143 /* Declarations of custom local variables */
5146 unsigned long b = 0;
5147 const char *name = "(undefined)";
5148 struct my_struct *my_struct;
5152 * Custom code which uses both tracepoint arguments
5153 * (in TP_ARGS()) and local variables (in TP_locvar()).
5155 * Local variables are actually members of a structure pointed
5156 * to by the special variable tp_locvar.
5160 tp_locvar->a = my_int + 17;
5161 tp_locvar->my_struct = get_my_struct_at(tp_locvar->a);
5162 tp_locvar->b = my_struct_compute_b(tp_locvar->my_struct);
5163 tp_locvar->name = my_struct_get_name(tp_locvar->my_struct);
5164 put_my_struct(tp_locvar->my_struct);
5173 * Format identical to the LTTNG_TRACEPOINT_EVENT()
5174 * version for this, except that tp_locvar members can be
5175 * used in the argument expression parameters of
5176 * the ctf_*() macros.
5179 ctf_integer(unsigned long, my_struct_b, tp_locvar->b)
5180 ctf_integer(int, my_struct_a, tp_locvar->a)
5181 ctf_string(my_string_field, my_string)
5182 ctf_string(my_struct_name, tp_locvar->name)
5187 IMPORTANT: The C code defined in `TP_code()` must not have any side
5188 effects when executed. In particular, the code must not allocate
5189 memory or get resources without deallocating this memory or putting
5190 those resources afterwards.
5193 [[instrumenting-linux-kernel-tracing]]
5194 ==== Load and unload a custom probe kernel module
5196 You must load a <<lttng-adaptation-layer,created LTTng-modules probe
5197 kernel module>> in the kernel before it can emit LTTng events.
5199 To load the default probe kernel modules and a custom probe kernel
5202 * Use the opt:lttng-sessiond(8):--extra-kmod-probes option to give extra
5203 probe modules to load when starting a root <<lttng-sessiond,session
5207 .Load the `my_subsys`, `usb`, and the default probe modules.
5211 # lttng-sessiond --extra-kmod-probes=my_subsys,usb
5216 You only need to pass the subsystem name, not the whole kernel module
5219 To load _only_ a given custom probe kernel module:
5221 * Use the opt:lttng-sessiond(8):--kmod-probes option to give the probe
5222 modules to load when starting a root session daemon:
5225 .Load only the `my_subsys` and `usb` probe modules.
5229 # lttng-sessiond --kmod-probes=my_subsys,usb
5234 To confirm that a probe module is loaded:
5241 $ lsmod | grep lttng_probe_usb
5245 To unload the loaded probe modules:
5247 * Kill the session daemon with `SIGTERM`:
5252 # pkill lttng-sessiond
5256 You can also use man:modprobe(8)'s `--remove` option if the session
5257 daemon terminates abnormally.
5260 [[controlling-tracing]]
5263 Once an application or a Linux kernel is
5264 <<instrumenting,instrumented>> for LTTng tracing,
5267 This section is divided in topics on how to use the various
5268 <<plumbing,components of LTTng>>, in particular the <<lttng-cli,cmd:lttng
5269 command-line tool>>, to _control_ the LTTng daemons and tracers.
5271 NOTE: In the following subsections, we refer to an man:lttng(1) command
5272 using its man page name. For example, instead of _Run the `create`
5273 command to..._, we use _Run the man:lttng-create(1) command to..._.
5277 === Start a session daemon
5279 In some situations, you need to run a <<lttng-sessiond,session daemon>>
5280 (man:lttng-sessiond(8)) _before_ you can use the man:lttng(1)
5283 You will see the following error when you run a command while no session
5287 Error: No session daemon is available
5290 The only command that automatically runs a session daemon is
5291 man:lttng-create(1), which you use to
5292 <<creating-destroying-tracing-sessions,create a tracing session>>. While
5293 this is most of the time the first operation that you do, sometimes it's
5294 not. Some examples are:
5296 * <<list-instrumentation-points,List the available instrumentation points>>.
5297 * <<saving-loading-tracing-session,Load a tracing session configuration>>.
5299 [[tracing-group]] Each Unix user must have its own running session
5300 daemon to trace user applications. The session daemon that the root user
5301 starts is the only one allowed to control the LTTng kernel tracer. Users
5302 that are part of the _tracing group_ can control the root session
5303 daemon. The default tracing group name is `tracing`; you can set it to
5304 something else with the opt:lttng-sessiond(8):--group option when you
5305 start the root session daemon.
5307 To start a user session daemon:
5309 * Run man:lttng-sessiond(8):
5314 $ lttng-sessiond --daemonize
5318 To start the root session daemon:
5320 * Run man:lttng-sessiond(8) as the root user:
5325 # lttng-sessiond --daemonize
5329 In both cases, remove the opt:lttng-sessiond(8):--daemonize option to
5330 start the session daemon in foreground.
5332 To stop a session daemon, use man:kill(1) on its process ID (standard
5335 Note that some Linux distributions could manage the LTTng session daemon
5336 as a service. In this case, you should use the service manager to
5337 start, restart, and stop session daemons.
5340 [[creating-destroying-tracing-sessions]]
5341 === Create and destroy a tracing session
5343 Almost all the LTTng control operations happen in the scope of
5344 a <<tracing-session,tracing session>>, which is the dialogue between the
5345 <<lttng-sessiond,session daemon>> and you.
5347 To create a tracing session with a generated name:
5349 * Use the man:lttng-create(1) command:
5358 The created tracing session's name is `auto` followed by the
5361 To create a tracing session with a specific name:
5363 * Use the optional argument of the man:lttng-create(1) command:
5368 $ lttng create my-session
5372 Replace `my-session` with the specific tracing session name.
5374 LTTng appends the creation date to the created tracing session's name.
5376 LTTng writes the traces of a tracing session in
5377 +$LTTNG_HOME/lttng-trace/__name__+ by default, where +__name__+ is the
5378 name of the tracing session. Note that the env:LTTNG_HOME environment
5379 variable defaults to `$HOME` if not set.
5381 To output LTTng traces to a non-default location:
5383 * Use the opt:lttng-create(1):--output option of the man:lttng-create(1) command:
5388 $ lttng create my-session --output=/tmp/some-directory
5392 You may create as many tracing sessions as you wish.
5394 To list all the existing tracing sessions for your Unix user:
5396 * Use the man:lttng-list(1) command:
5405 When you create a tracing session, it is set as the _current tracing
5406 session_. The following man:lttng(1) commands operate on the current
5407 tracing session when you don't specify one:
5409 [role="list-3-cols"]
5410 * man:lttng-add-context(1)
5411 * man:lttng-destroy(1)
5412 * man:lttng-disable-channel(1)
5413 * man:lttng-disable-event(1)
5414 * man:lttng-disable-rotation(1)
5415 * man:lttng-enable-channel(1)
5416 * man:lttng-enable-event(1)
5417 * man:lttng-enable-rotation(1)
5419 * man:lttng-regenerate(1)
5420 * man:lttng-rotate(1)
5422 * man:lttng-snapshot(1)
5423 * man:lttng-start(1)
5424 * man:lttng-status(1)
5426 * man:lttng-track(1)
5427 * man:lttng-untrack(1)
5430 To change the current tracing session:
5432 * Use the man:lttng-set-session(1) command:
5437 $ lttng set-session new-session
5441 Replace `new-session` by the name of the new current tracing session.
5443 When you are done tracing in a given tracing session, you can destroy
5444 it. This operation frees the resources taken by the tracing session
5445 to destroy; it does not destroy the trace data that LTTng wrote for
5446 this tracing session.
5448 To destroy the current tracing session:
5450 * Use the man:lttng-destroy(1) command:
5459 The man:lttng-destroy(1) command also runs the man:lttng-stop(1)
5460 command implicitly (see <<basic-tracing-session-control,Start and stop a
5461 tracing session>>). You need to stop tracing to make LTTng flush the
5462 remaining trace data and make the trace readable.
5465 [[list-instrumentation-points]]
5466 === List the available instrumentation points
5468 The <<lttng-sessiond,session daemon>> can query the running instrumented
5469 user applications and the Linux kernel to get a list of available
5470 instrumentation points. For the Linux kernel <<domain,tracing domain>>,
5471 they are tracepoints and system calls. For the user space tracing
5472 domain, they are tracepoints. For the other tracing domains, they are
5475 To list the available instrumentation points:
5477 * Use the man:lttng-list(1) command with the requested tracing domain's
5481 * opt:lttng-list(1):--kernel: Linux kernel tracepoints (your Unix user
5482 must be a root user, or it must be a member of the
5483 <<tracing-group,tracing group>>).
5484 * opt:lttng-list(1):--kernel with opt:lttng-list(1):--syscall: Linux
5485 kernel system calls (your Unix user must be a root user, or it must be
5486 a member of the tracing group).
5487 * opt:lttng-list(1):--userspace: user space tracepoints.
5488 * opt:lttng-list(1):--jul: `java.util.logging` loggers.
5489 * opt:lttng-list(1):--log4j: Apache log4j loggers.
5490 * opt:lttng-list(1):--python: Python loggers.
5493 .List the available user space tracepoints.
5497 $ lttng list --userspace
5501 .List the available Linux kernel system call tracepoints.
5505 $ lttng list --kernel --syscall
5510 [[enabling-disabling-events]]
5511 === Create and enable an event rule
5513 Once you <<creating-destroying-tracing-sessions,create a tracing
5514 session>>, you can create <<event,event rules>> with the
5515 man:lttng-enable-event(1) command.
5517 You specify each condition with a command-line option. The available
5518 condition arguments are shown in the following table.
5520 [role="growable",cols="asciidoc,asciidoc,default"]
5521 .Condition command-line arguments for the man:lttng-enable-event(1) command.
5523 |Argument |Description |Applicable tracing domains
5529 . +--probe=__ADDR__+
5530 . +--function=__ADDR__+
5531 . +--userspace-probe=__PATH__:__SYMBOL__+
5532 . +--userspace-probe=sdt:__PATH__:__PROVIDER__:__NAME__+
5535 Instead of using the default _tracepoint_ instrumentation type, use:
5537 . A Linux system call (entry and exit).
5538 . A Linux https://lwn.net/Articles/132196/[kprobe] (symbol or address).
5539 . The entry and return points of a Linux function (symbol or address).
5540 . The entry point of a user application or library function (path to
5541 application/library and symbol).
5542 . A https://www.sourceware.org/systemtap/wiki/AddingUserSpaceProbingToApps[SystemTap
5543 Statically Defined Tracing] (USDT) probe (path to application/library,
5544 provider and probe names).
5548 |First positional argument.
5551 Tracepoint or system call name.
5553 With the opt:lttng-enable-event(1):--probe,
5554 opt:lttng-enable-event(1):--function, and
5555 opt:lttng-enable-event(1):--userspace-probe options, this is a custom
5556 name given to the event rule. With the JUL, log4j, and Python domains,
5557 this is a logger name.
5559 With a tracepoint, logger, or system call name, you can use the special
5560 `*` globbing character to match anything (for example, `sched_*`,
5568 . +--loglevel=__LEVEL__+
5569 . +--loglevel-only=__LEVEL__+
5572 . Match only tracepoints or log statements with a logging level at
5573 least as severe as +__LEVEL__+.
5574 . Match only tracepoints or log statements with a logging level
5575 equal to +__LEVEL__+.
5577 See man:lttng-enable-event(1) for the list of available logging level
5580 |User space, JUL, log4j, and Python.
5582 |+--exclude=__EXCLUSIONS__+
5585 When you use a `*` character at the end of the tracepoint or logger
5586 name (first positional argument), exclude the specific names in the
5587 comma-delimited list +__EXCLUSIONS__+.
5590 User space, JUL, log4j, and Python.
5592 |+--filter=__EXPR__+
5595 Match only events which satisfy the expression +__EXPR__+.
5597 See man:lttng-enable-event(1) to learn more about the syntax of a
5604 You attach an event rule to a <<channel,channel>> on creation. If you do
5605 not specify the channel with the opt:lttng-enable-event(1):--channel
5606 option, and if the event rule to create is the first in its
5607 <<domain,tracing domain>> for a given tracing session, then LTTng
5608 creates a _default channel_ for you. This default channel is reused in
5609 subsequent invocations of the man:lttng-enable-event(1) command for the
5610 same tracing domain.
5612 An event rule is always enabled at creation time.
5614 The following examples show how you can combine the previous
5615 command-line options to create simple to more complex event rules.
5617 .Create an event rule targetting a Linux kernel tracepoint (default channel).
5621 $ lttng enable-event --kernel sched_switch
5625 .Create an event rule matching four Linux kernel system calls (default channel).
5629 $ lttng enable-event --kernel --syscall open,write,read,close
5633 .Create event rules matching tracepoints with filter expressions (default channel).
5637 $ lttng enable-event --kernel sched_switch --filter='prev_comm == "bash"'
5642 $ lttng enable-event --kernel --all \
5643 --filter='$ctx.tid == 1988 || $ctx.tid == 1534'
5648 $ lttng enable-event --jul my_logger \
5649 --filter='$app.retriever:cur_msg_id > 3'
5652 IMPORTANT: Make sure to always quote the filter string when you
5653 use man:lttng(1) from a shell.
5656 .Create an event rule matching any user space tracepoint of a given tracepoint provider with a log level range (default channel).
5660 $ lttng enable-event --userspace my_app:'*' --loglevel=TRACE_INFO
5663 IMPORTANT: Make sure to always quote the wildcard character when you
5664 use man:lttng(1) from a shell.
5667 .Create an event rule matching multiple Python loggers with a wildcard and with exclusions (default channel).
5671 $ lttng enable-event --python my-app.'*' \
5672 --exclude='my-app.module,my-app.hello'
5676 .Create an event rule matching any Apache log4j logger with a specific log level (default channel).
5680 $ lttng enable-event --log4j --all --loglevel-only=LOG4J_WARN
5684 .Create an event rule attached to a specific channel matching a specific user space tracepoint provider and tracepoint.
5688 $ lttng enable-event --userspace my_app:my_tracepoint --channel=my-channel
5692 .Create an event rule matching the `malloc` function entry in path:{/usr/lib/libc.so.6}:
5696 $ lttng enable-event --kernel --userspace-probe=/usr/lib/libc.so.6:malloc \
5701 .Create an event rule matching the `server`/`accept_request` https://www.sourceware.org/systemtap/wiki/AddingUserSpaceProbingToApps[USDT probe] in path:{/usr/bin/serv}:
5705 $ lttng enable-event --kernel --userspace-probe=sdt:serv:server:accept_request \
5706 server_accept_request
5710 The event rules of a given channel form a whitelist: as soon as an
5711 emitted event passes one of them, LTTng can record the event. For
5712 example, an event named `my_app:my_tracepoint` emitted from a user space
5713 tracepoint with a `TRACE_ERROR` log level passes both of the following
5718 $ lttng enable-event --userspace my_app:my_tracepoint
5719 $ lttng enable-event --userspace my_app:my_tracepoint \
5720 --loglevel=TRACE_INFO
5723 The second event rule is redundant: the first one includes
5727 [[disable-event-rule]]
5728 === Disable an event rule
5730 To disable an event rule that you <<enabling-disabling-events,created>>
5731 previously, use the man:lttng-disable-event(1) command. This command
5732 disables _all_ the event rules (of a given tracing domain and channel)
5733 which match an instrumentation point. The other conditions are not
5734 supported as of LTTng{nbsp}{revision}.
5736 The LTTng tracer does not record an emitted event which passes
5737 a _disabled_ event rule.
5739 .Disable an event rule matching a Python logger (default channel).
5743 $ lttng disable-event --python my-logger
5747 .Disable an event rule matching all `java.util.logging` loggers (default channel).
5751 $ lttng disable-event --jul '*'
5755 .Disable _all_ the event rules of the default channel.
5757 The opt:lttng-disable-event(1):--all-events option is not, like the
5758 opt:lttng-enable-event(1):--all option of man:lttng-enable-event(1), the
5759 equivalent of the event name `*` (wildcard): it disables _all_ the event
5760 rules of a given channel.
5764 $ lttng disable-event --jul --all-events
5768 NOTE: You cannot delete an event rule once you create it.
5772 === Get the status of a tracing session
5774 To get the status of the current tracing session, that is, its
5775 parameters, its channels, event rules, and their attributes:
5777 * Use the man:lttng-status(1) command:
5787 To get the status of any tracing session:
5789 * Use the man:lttng-list(1) command with the tracing session's name:
5794 $ lttng list my-session
5798 Replace `my-session` with the desired tracing session's name.
5801 [[basic-tracing-session-control]]
5802 === Start and stop a tracing session
5804 Once you <<creating-destroying-tracing-sessions,create a tracing
5806 <<enabling-disabling-events,create one or more event rules>>,
5807 you can start and stop the tracers for this tracing session.
5809 To start tracing in the current tracing session:
5811 * Use the man:lttng-start(1) command:
5820 LTTng is very flexible: you can launch user applications before
5821 or after the you start the tracers. The tracers only record the events
5822 if they pass enabled event rules and if they occur while the tracers are
5825 To stop tracing in the current tracing session:
5827 * Use the man:lttng-stop(1) command:
5836 If there were <<channel-overwrite-mode-vs-discard-mode,lost event
5837 records>> or lost sub-buffers since the last time you ran
5838 man:lttng-start(1), warnings are printed when you run the
5839 man:lttng-stop(1) command.
5841 IMPORTANT: You need to stop tracing to make LTTng flush the remaining
5842 trace data and make the trace readable. Note that the
5843 man:lttng-destroy(1) command (see
5844 <<creating-destroying-tracing-sessions,Create and destroy a tracing
5845 session>>) also runs the man:lttng-stop(1) command implicitly.
5848 [[enabling-disabling-channels]]
5849 === Create a channel
5851 Once you create a tracing session, you can create a <<channel,channel>>
5852 with the man:lttng-enable-channel(1) command.
5854 Note that LTTng automatically creates a default channel when, for a
5855 given <<domain,tracing domain>>, no channels exist and you
5856 <<enabling-disabling-events,create>> the first event rule. This default
5857 channel is named `channel0` and its attributes are set to reasonable
5858 values. Therefore, you only need to create a channel when you need
5859 non-default attributes.
5861 You specify each non-default channel attribute with a command-line
5862 option when you use the man:lttng-enable-channel(1) command. The
5863 available command-line options are:
5865 [role="growable",cols="asciidoc,asciidoc"]
5866 .Command-line options for the man:lttng-enable-channel(1) command.
5868 |Option |Description
5874 <<channel-overwrite-mode-vs-discard-mode,event record loss mode>> instead
5875 of the default _discard_ mode.
5877 |`--buffers-pid` (user space tracing domain only)
5880 Use the per-process <<channel-buffering-schemes,buffering scheme>>
5881 instead of the default per-user buffering scheme.
5883 |+--subbuf-size=__SIZE__+
5886 Allocate sub-buffers of +__SIZE__+ bytes (power of two), for each CPU,
5887 either for each Unix user (default), or for each instrumented process.
5889 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
5891 |+--num-subbuf=__COUNT__+
5894 Allocate +__COUNT__+ sub-buffers (power of two), for each CPU, either
5895 for each Unix user (default), or for each instrumented process.
5897 See <<channel-subbuf-size-vs-subbuf-count,Sub-buffer count and size>>.
5899 |+--tracefile-size=__SIZE__+
5902 Set the maximum size of each trace file that this channel writes within
5903 a stream to +__SIZE__+ bytes instead of no maximum.
5905 See <<tracefile-rotation,Trace file count and size>>.
5907 |+--tracefile-count=__COUNT__+
5910 Limit the number of trace files that this channel creates to
5911 +__COUNT__+ channels instead of no limit.
5913 See <<tracefile-rotation,Trace file count and size>>.
5915 |+--switch-timer=__PERIODUS__+
5918 Set the <<channel-switch-timer,switch timer period>>
5919 to +__PERIODUS__+{nbsp}µs.
5921 |+--read-timer=__PERIODUS__+
5924 Set the <<channel-read-timer,read timer period>>
5925 to +__PERIODUS__+{nbsp}µs.
5927 |[[opt-blocking-timeout]]+--blocking-timeout=__TIMEOUTUS__+
5930 Set the timeout of user space applications which load LTTng-UST
5931 in blocking mode to +__TIMEOUTUS__+:
5934 Never block (non-blocking mode).
5937 Block forever until space is available in a sub-buffer to record
5940 __n__, a positive value::
5941 Wait for at most __n__ µs when trying to write into a sub-buffer.
5943 Note that, for this option to have any effect on an instrumented
5944 user space application, you need to run the application with a set
5945 env:LTTNG_UST_ALLOW_BLOCKING environment variable.
5947 |+--output=__TYPE__+ (Linux kernel tracing domain only)
5950 Set the channel's output type to +__TYPE__+, either `mmap` or `splice`.
5954 You can only create a channel in the Linux kernel and user space
5955 <<domain,tracing domains>>: other tracing domains have their own channel
5956 created on the fly when <<enabling-disabling-events,creating event
5961 Because of a current LTTng limitation, you must create all channels
5962 _before_ you <<basic-tracing-session-control,start tracing>> in a given
5963 tracing session, that is, before the first time you run
5966 Since LTTng automatically creates a default channel when you use the
5967 man:lttng-enable-event(1) command with a specific tracing domain, you
5968 cannot, for example, create a Linux kernel event rule, start tracing,
5969 and then create a user space event rule, because no user space channel
5970 exists yet and it's too late to create one.
5972 For this reason, make sure to configure your channels properly
5973 before starting the tracers for the first time!
5976 The following examples show how you can combine the previous
5977 command-line options to create simple to more complex channels.
5979 .Create a Linux kernel channel with default attributes.
5983 $ lttng enable-channel --kernel my-channel
5987 .Create a user space channel with four sub-buffers or 1{nbsp}MiB each, per CPU, per instrumented process.
5991 $ lttng enable-channel --userspace --num-subbuf=4 --subbuf-size=1M \
5992 --buffers-pid my-channel
5996 .[[blocking-timeout-example]]Create a default user space channel with an infinite blocking timeout.
5998 <<creating-destroying-tracing-sessions,Create a tracing-session>>,
5999 create the channel, <<enabling-disabling-events,create an event rule>>,
6000 and <<basic-tracing-session-control,start tracing>>:
6005 $ lttng enable-channel --userspace --blocking-timeout=inf blocking-channel
6006 $ lttng enable-event --userspace --channel=blocking-channel --all
6010 Run an application instrumented with LTTng-UST and allow it to block:
6014 $ LTTNG_UST_ALLOW_BLOCKING=1 my-app
6018 .Create a Linux kernel channel which rotates eight trace files of 4{nbsp}MiB each for each stream
6022 $ lttng enable-channel --kernel --tracefile-count=8 \
6023 --tracefile-size=4194304 my-channel
6027 .Create a user space channel in overwrite (or _flight recorder_) mode.
6031 $ lttng enable-channel --userspace --overwrite my-channel
6035 You can <<enabling-disabling-events,create>> the same event rule in
6036 two different channels:
6040 $ lttng enable-event --userspace --channel=my-channel app:tp
6041 $ lttng enable-event --userspace --channel=other-channel app:tp
6044 If both channels are enabled, when a tracepoint named `app:tp` is
6045 reached, LTTng records two events, one for each channel.
6049 === Disable a channel
6051 To disable a specific channel that you <<enabling-disabling-channels,created>>
6052 previously, use the man:lttng-disable-channel(1) command.
6054 .Disable a specific Linux kernel channel.
6058 $ lttng disable-channel --kernel my-channel
6062 The state of a channel precedes the individual states of event rules
6063 attached to it: event rules which belong to a disabled channel, even if
6064 they are enabled, are also considered disabled.
6068 === Add context fields to a channel
6070 Event record fields in trace files provide important information about
6071 events that occured previously, but sometimes some external context may
6072 help you solve a problem faster. Examples of context fields are:
6074 * The **process ID**, **thread ID**, **process name**, and
6075 **process priority** of the thread in which the event occurs.
6076 * The **hostname** of the system on which the event occurs.
6077 * The Linux kernel and user call stacks (since
6078 LTTng{nbsp}{revision}).
6079 * The current values of many possible **performance counters** using
6081 ** CPU cycles, stalled cycles, idle cycles, and the other cycle types.
6083 ** Branch instructions, misses, and loads.
6085 * Any context defined at the application level (supported for the
6086 JUL and log4j <<domain,tracing domains>>).
6088 To get the full list of available context fields, see
6089 `lttng add-context --list`. Some context fields are reserved for a
6090 specific <<domain,tracing domain>> (Linux kernel or user space).
6092 You add context fields to <<channel,channels>>. All the events
6093 that a channel with added context fields records contain those fields.
6095 To add context fields to one or all the channels of a given tracing
6098 * Use the man:lttng-add-context(1) command.
6100 .Add context fields to all the channels of the current tracing session.
6102 The following command line adds the virtual process identifier and
6103 the per-thread CPU cycles count fields to all the user space channels
6104 of the current tracing session.
6108 $ lttng add-context --userspace --type=vpid --type=perf:thread:cpu-cycles
6112 .Add performance counter context fields by raw ID
6114 See man:lttng-add-context(1) for the exact format of the context field
6115 type, which is partly compatible with the format used in
6120 $ lttng add-context --userspace --type=perf:thread:raw:r0110:test
6121 $ lttng add-context --kernel --type=perf:cpu:raw:r0013c:x86unhalted
6125 .Add context fields to a specific channel.
6127 The following command line adds the thread identifier and user call
6128 stack context fields to the Linux kernel channel named `my-channel` in
6129 the current tracing session.
6133 $ lttng add-context --kernel --channel=my-channel \
6134 --type=tid --type=callstack-user
6138 .Add an application-specific context field to a specific channel.
6140 The following command line adds the `cur_msg_id` context field of the
6141 `retriever` context retriever for all the instrumented
6142 <<java-application,Java applications>> recording <<event,event records>>
6143 in the channel named `my-channel`:
6147 $ lttng add-context --kernel --channel=my-channel \
6148 --type='$app:retriever:cur_msg_id'
6151 IMPORTANT: Make sure to always quote the `$` character when you
6152 use man:lttng-add-context(1) from a shell.
6155 NOTE: You cannot remove context fields from a channel once you add it.
6160 === Track process IDs
6162 It's often useful to allow only specific process IDs (PIDs) to emit
6163 events. For example, you may wish to record all the system calls made by
6164 a given process (Ă la http://linux.die.net/man/1/strace[strace]).
6166 The man:lttng-track(1) and man:lttng-untrack(1) commands serve this
6167 purpose. Both commands operate on a whitelist of process IDs. You _add_
6168 entries to this whitelist with the man:lttng-track(1) command and remove
6169 entries with the man:lttng-untrack(1) command. Any process which has one
6170 of the PIDs in the whitelist is allowed to emit LTTng events which pass
6171 an enabled <<event,event rule>>.
6173 NOTE: The PID tracker tracks the _numeric process IDs_. Should a
6174 process with a given tracked ID exit and another process be given this
6175 ID, then the latter would also be allowed to emit events.
6177 .Track and untrack process IDs.
6179 For the sake of the following example, assume the target system has
6180 16{nbsp}possible PIDs.
6183 <<creating-destroying-tracing-sessions,create a tracing session>>,
6184 the whitelist contains all the possible PIDs:
6187 .All PIDs are tracked.
6188 image::track-all.png[]
6190 When the whitelist is full and you use the man:lttng-track(1) command to
6191 specify some PIDs to track, LTTng first clears the whitelist, then it
6192 tracks the specific PIDs. After:
6196 $ lttng track --pid=3,4,7,10,13
6202 .PIDs 3, 4, 7, 10, and 13 are tracked.
6203 image::track-3-4-7-10-13.png[]
6205 You can add more PIDs to the whitelist afterwards:
6209 $ lttng track --pid=1,15,16
6215 .PIDs 1, 15, and 16 are added to the whitelist.
6216 image::track-1-3-4-7-10-13-15-16.png[]
6218 The man:lttng-untrack(1) command removes entries from the PID tracker's
6219 whitelist. Given the previous example, the following command:
6223 $ lttng untrack --pid=3,7,10,13
6226 leads to this whitelist:
6229 .PIDs 3, 7, 10, and 13 are removed from the whitelist.
6230 image::track-1-4-15-16.png[]
6232 LTTng can track all possible PIDs again using the
6233 opt:lttng-track(1):--all option:
6237 $ lttng track --pid --all
6240 The result is, again:
6243 .All PIDs are tracked.
6244 image::track-all.png[]
6247 .Track only specific PIDs
6249 A very typical use case with PID tracking is to start with an empty
6250 whitelist, then <<basic-tracing-session-control,start the tracers>>, and
6251 then add PIDs manually while tracers are active. You can accomplish this
6252 by using the opt:lttng-untrack(1):--all option of the
6253 man:lttng-untrack(1) command to clear the whitelist after you
6254 <<creating-destroying-tracing-sessions,create a tracing session>>:
6258 $ lttng untrack --pid --all
6264 .No PIDs are tracked.
6265 image::untrack-all.png[]
6267 If you trace with this whitelist configuration, the tracer records no
6268 events for this <<domain,tracing domain>> because no processes are
6269 tracked. You can use the man:lttng-track(1) command as usual to track
6270 specific PIDs, for example:
6274 $ lttng track --pid=6,11
6280 .PIDs 6 and 11 are tracked.
6281 image::track-6-11.png[]
6286 [[saving-loading-tracing-session]]
6287 === Save and load tracing session configurations
6289 Configuring a <<tracing-session,tracing session>> can be long. Some of
6290 the tasks involved are:
6292 * <<enabling-disabling-channels,Create channels>> with
6293 specific attributes.
6294 * <<adding-context,Add context fields>> to specific channels.
6295 * <<enabling-disabling-events,Create event rules>> with specific log
6296 level and filter conditions.
6298 If you use LTTng to solve real world problems, chances are you have to
6299 record events using the same tracing session setup over and over,
6300 modifying a few variables each time in your instrumented program
6301 or environment. To avoid constant tracing session reconfiguration,
6302 the man:lttng(1) command-line tool can save and load tracing session
6303 configurations to/from XML files.
6305 To save a given tracing session configuration:
6307 * Use the man:lttng-save(1) command:
6312 $ lttng save my-session
6316 Replace `my-session` with the name of the tracing session to save.
6318 LTTng saves tracing session configurations to
6319 dir:{$LTTNG_HOME/.lttng/sessions} by default. Note that the
6320 env:LTTNG_HOME environment variable defaults to `$HOME` if not set. Use
6321 the opt:lttng-save(1):--output-path option to change this destination
6324 LTTng saves all configuration parameters, for example:
6326 * The tracing session name.
6327 * The trace data output path.
6328 * The channels with their state and all their attributes.
6329 * The context fields you added to channels.
6330 * The event rules with their state, log level and filter conditions.
6332 To load a tracing session:
6334 * Use the man:lttng-load(1) command:
6339 $ lttng load my-session
6343 Replace `my-session` with the name of the tracing session to load.
6345 When LTTng loads a configuration, it restores your saved tracing session
6346 as if you just configured it manually.
6348 See man:lttng(1) for the complete list of command-line options. You
6349 can also save and load all many sessions at a time, and decide in which
6350 directory to output the XML files.
6353 [[sending-trace-data-over-the-network]]
6354 === Send trace data over the network
6356 LTTng can send the recorded trace data to a remote system over the
6357 network instead of writing it to the local file system.
6359 To send the trace data over the network:
6361 . On the _remote_ system (which can also be the target system),
6362 start an LTTng <<lttng-relayd,relay daemon>> (man:lttng-relayd(8)):
6371 . On the _target_ system, create a tracing session configured to
6372 send trace data over the network:
6377 $ lttng create my-session --set-url=net://remote-system
6381 Replace `remote-system` by the host name or IP address of the
6382 remote system. See man:lttng-create(1) for the exact URL format.
6384 . On the target system, use the man:lttng(1) command-line tool as usual.
6385 When tracing is active, the target's consumer daemon sends sub-buffers
6386 to the relay daemon running on the remote system instead of flushing
6387 them to the local file system. The relay daemon writes the received
6388 packets to the local file system.
6390 The relay daemon writes trace files to
6391 +$LTTNG_HOME/lttng-traces/__hostname__/__session__+ by default, where
6392 +__hostname__+ is the host name of the target system and +__session__+
6393 is the tracing session name. Note that the env:LTTNG_HOME environment
6394 variable defaults to `$HOME` if not set. Use the
6395 opt:lttng-relayd(8):--output option of man:lttng-relayd(8) to write
6396 trace files to another base directory.
6401 === View events as LTTng emits them (noch:{LTTng} live)
6403 LTTng live is a network protocol implemented by the <<lttng-relayd,relay
6404 daemon>> (man:lttng-relayd(8)) to allow compatible trace viewers to
6405 display events as LTTng emits them on the target system while tracing is
6408 The relay daemon creates a _tee_: it forwards the trace data to both
6409 the local file system and to connected live viewers:
6412 .The relay daemon creates a _tee_, forwarding the trace data to both trace files and a connected live viewer.
6417 . On the _target system_, create a <<tracing-session,tracing session>>
6423 $ lttng create my-session --live
6427 This spawns a local relay daemon.
6429 . Start the live viewer and configure it to connect to the relay
6430 daemon. For example, with http://diamon.org/babeltrace[Babeltrace]:
6435 $ babeltrace --input-format=lttng-live \
6436 net://localhost/host/hostname/my-session
6443 * `hostname` with the host name of the target system.
6444 * `my-session` with the name of the tracing session to view.
6447 . Configure the tracing session as usual with the man:lttng(1)
6448 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6450 You can list the available live tracing sessions with Babeltrace:
6454 $ babeltrace --input-format=lttng-live net://localhost
6457 You can start the relay daemon on another system. In this case, you need
6458 to specify the relay daemon's URL when you create the tracing session
6459 with the opt:lttng-create(1):--set-url option. You also need to replace
6460 `localhost` in the procedure above with the host name of the system on
6461 which the relay daemon is running.
6463 See man:lttng-create(1) and man:lttng-relayd(8) for the complete list of
6464 command-line options.
6468 [[taking-a-snapshot]]
6469 === Take a snapshot of the current sub-buffers of a tracing session
6471 The normal behavior of LTTng is to append full sub-buffers to growing
6472 trace data files. This is ideal to keep a full history of the events
6473 that occurred on the target system, but it can
6474 represent too much data in some situations. For example, you may wish
6475 to trace your application continuously until some critical situation
6476 happens, in which case you only need the latest few recorded
6477 events to perform the desired analysis, not multi-gigabyte trace files.
6479 With the man:lttng-snapshot(1) command, you can take a snapshot of the
6480 current sub-buffers of a given <<tracing-session,tracing session>>.
6481 LTTng can write the snapshot to the local file system or send it over
6485 .A snapshot is a copy of the current sub-buffers, which are not cleared after the operation.
6486 image::snapshot.png[]
6488 If you wish to create unmanaged, self-contained, non-overlapping
6489 trace chunk archives instead of a simple copy of the current
6490 sub-buffers, see the <<session-rotation,tracing session rotation>>
6491 feature (available since LTTng{nbsp}2.11).
6495 . Create a tracing session in _snapshot mode_:
6500 $ lttng create my-session --snapshot
6504 The <<channel-overwrite-mode-vs-discard-mode,event record loss mode>> of
6505 <<channel,channels>> created in this mode is automatically set to
6506 _overwrite_ (flight recorder mode).
6508 . Configure the tracing session as usual with the man:lttng(1)
6509 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6511 . **Optional**: When you need to take a snapshot,
6512 <<basic-tracing-session-control,stop tracing>>.
6514 You can take a snapshot when the tracers are active, but if you stop
6515 them first, you are sure that the data in the sub-buffers does not
6516 change before you actually take the snapshot.
6523 $ lttng snapshot record --name=my-first-snapshot
6527 LTTng writes the current sub-buffers of all the current tracing
6528 session's channels to trace files on the local file system. Those trace
6529 files have `my-first-snapshot` in their name.
6531 There is no difference between the format of a normal trace file and the
6532 format of a snapshot: viewers of LTTng traces also support LTTng
6535 By default, LTTng writes snapshot files to the path shown by
6536 `lttng snapshot list-output`. You can change this path or decide to send
6537 snapshots over the network using either:
6539 . An output path or URL that you specify when you
6540 <<creating-destroying-tracing-sessions,create the tracing session>>.
6541 . A snapshot output path or URL that you add using
6542 `lttng snapshot add-output`.
6543 . An output path or URL that you provide directly to the
6544 `lttng snapshot record` command.
6546 Method{nbsp}3 overrides method{nbsp}2, which overrides method 1. When
6547 you specify a URL, a relay daemon must listen on a remote system (see
6548 <<sending-trace-data-over-the-network,Send trace data over the
6553 [[session-rotation]]
6554 === Archive the current trace chunk (rotate a tracing session)
6556 The <<taking-a-snapshot,snapshot user guide>> shows how you can dump
6557 a tracing session's current sub-buffers to the file system or send them
6558 over the network. When you take a snapshot, LTTng does not clear the
6559 tracing session's ring buffers: if you take another snapshot immediately
6560 after, both snapshots could contain overlapping trace data.
6562 Inspired by https://en.wikipedia.org/wiki/Log_rotation[log rotation],
6563 _tracing session rotation_ is a feature which appends the content of the
6564 ring buffers to what's already on the file system or sent over the
6565 network since the tracing session's creation or since the last
6566 rotation, and then clears those ring buffers to avoid trace data
6569 What LTTng is about to write when performing a tracing session rotation
6570 is called the _current trace chunk_. When this current trace chunk is
6571 written to the file system or sent over the network, it becomes a _trace
6572 chunk archive_. Therefore, a tracing session rotation _archives_ the
6573 current trace chunk.
6576 .A tracing session rotation operation _archives_ the current trace chunk.
6577 image::rotation.png[]
6579 A trace chunk archive is a self-contained LTTng trace which LTTng
6580 doesn't manage anymore: you can read it, modify it, move it, or remove
6583 There are two methods to perform a tracing session rotation: immediately
6584 or with a rotation schedule.
6586 To perform an immediate tracing session rotation:
6588 . <<creating-destroying-tracing-sessions,Create a tracing session>>
6589 in _normal mode_ or _network streaming mode_
6590 (only those two creation modes support tracing session rotation):
6595 $ lttng create my-session
6599 . <<enabling-disabling-events,Create one or more event rules>>
6600 and <<basic-tracing-session-control,start tracing>>:
6605 $ lttng enable-event --kernel sched_'*'
6610 . When needed, immediately rotate the current tracing session:
6619 The cmd:lttng-rotate command prints the path to the created trace
6620 chunk archive. See man:lttng-rotate(1) to learn about the format
6621 of trace chunk archive directory names.
6623 You can perform other immediate rotations while the tracing session is
6624 active. It is guaranteed that all the trace chunk archives do not
6625 contain overlapping trace data. You can also perform an immediate
6626 rotation once you have <<basic-tracing-session-control,stopped>> the
6629 . When you are done tracing,
6630 <<creating-destroying-tracing-sessions,destroy the current tracing
6640 The tracing session destruction operation creates one last trace
6641 chunk archive from the current trace chunk.
6643 A tracing session rotation schedule is a planned rotation which LTTng
6644 performs automatically based on one of the following conditions:
6646 * A timer with a configured period times out.
6648 * The total size of the flushed part of the current trace chunk
6649 becomes greater than or equal to a configured value.
6651 To schedule a tracing session rotation, set a _rotation schedule_:
6653 . <<creating-destroying-tracing-sessions,Create a tracing session>>
6654 in _normal mode_ or _network streaming mode_
6655 (only those two creation modes support tracing session rotation):
6660 $ lttng create my-session
6664 . <<enabling-disabling-events,Create one or more event rules>>:
6669 $ lttng enable-event --kernel sched_'*'
6673 . Set a tracing session rotation schedule:
6678 $ lttng enable-rotation --timer=10s
6682 In this example, we set a rotation schedule so that LTTng performs a
6683 tracing session rotation every ten seconds.
6685 See man:lttng-enable-rotation(1) to learn more about other ways to set a
6688 . <<basic-tracing-session-control,Start tracing>>:
6697 LTTng performs tracing session rotations automatically while the tracing
6698 session is active thanks to the rotation schedule.
6700 . When you are done tracing,
6701 <<creating-destroying-tracing-sessions,destroy the current tracing
6711 The tracing session destruction operation creates one last trace chunk
6712 archive from the current trace chunk.
6714 You can use man:lttng-disable-rotation(1) to unset a tracing session
6717 NOTE: man:lttng-rotate(1) and man:lttng-enable-rotation(1) list
6718 limitations regarding those two commands.
6723 === Use the machine interface
6725 With any command of the man:lttng(1) command-line tool, you can set the
6726 opt:lttng(1):--mi option to `xml` (before the command name) to get an
6727 XML machine interface output, for example:
6731 $ lttng --mi=xml enable-event --kernel --syscall open
6734 A schema definition (XSD) is
6735 https://github.com/lttng/lttng-tools/blob/stable-2.11/src/common/mi-lttng-3.0.xsd[available]
6736 to ease the integration with external tools as much as possible.
6740 [[metadata-regenerate]]
6741 === Regenerate the metadata of an LTTng trace
6743 An LTTng trace, which is a http://diamon.org/ctf[CTF] trace, has both
6744 data stream files and a metadata file. This metadata file contains,
6745 amongst other things, information about the offset of the clock sources
6746 used to timestamp <<event,event records>> when tracing.
6748 If, once a <<tracing-session,tracing session>> is
6749 <<basic-tracing-session-control,started>>, a major
6750 https://en.wikipedia.org/wiki/Network_Time_Protocol[NTP] correction
6751 happens, the trace's clock offset also needs to be updated. You
6752 can use the `metadata` item of the man:lttng-regenerate(1) command
6755 The main use case of this command is to allow a system to boot with
6756 an incorrect wall time and trace it with LTTng before its wall time
6757 is corrected. Once the system is known to be in a state where its
6758 wall time is correct, it can run `lttng regenerate metadata`.
6760 To regenerate the metadata of an LTTng trace:
6762 * Use the `metadata` item of the man:lttng-regenerate(1) command:
6767 $ lttng regenerate metadata
6773 `lttng regenerate metadata` has the following limitations:
6775 * Tracing session <<creating-destroying-tracing-sessions,created>>
6777 * User space <<channel,channels>>, if any, are using
6778 <<channel-buffering-schemes,per-user buffering>>.
6783 [[regenerate-statedump]]
6784 === Regenerate the state dump of a tracing session
6786 The LTTng kernel and user space tracers generate state dump
6787 <<event,event records>> when the application starts or when you
6788 <<basic-tracing-session-control,start a tracing session>>. An analysis
6789 can use the state dump event records to set an initial state before it
6790 builds the rest of the state from the following event records.
6791 http://tracecompass.org/[Trace Compass] is a notable example of an
6792 application which uses the state dump of an LTTng trace.
6794 When you <<taking-a-snapshot,take a snapshot>>, it's possible that the
6795 state dump event records are not included in the snapshot because they
6796 were recorded to a sub-buffer that has been consumed or overwritten
6799 You can use the `lttng regenerate statedump` command to emit the state
6800 dump event records again.
6802 To regenerate the state dump of the current tracing session, provided
6803 create it in snapshot mode, before you take a snapshot:
6805 . Use the `statedump` item of the man:lttng-regenerate(1) command:
6810 $ lttng regenerate statedump
6814 . <<basic-tracing-session-control,Stop the tracing session>>:
6823 . <<taking-a-snapshot,Take a snapshot>>:
6828 $ lttng snapshot record --name=my-snapshot
6832 Depending on the event throughput, you should run steps 1 and 2
6833 as closely as possible.
6835 NOTE: To record the state dump events, you need to
6836 <<enabling-disabling-events,create event rules>> which enable them.
6837 LTTng-UST state dump tracepoints start with `lttng_ust_statedump:`.
6838 LTTng-modules state dump tracepoints start with `lttng_statedump_`.
6842 [[persistent-memory-file-systems]]
6843 === Record trace data on persistent memory file systems
6845 https://en.wikipedia.org/wiki/Non-volatile_random-access_memory[Non-volatile random-access memory]
6846 (NVRAM) is random-access memory that retains its information when power
6847 is turned off (non-volatile). Systems with such memory can store data
6848 structures in RAM and retrieve them after a reboot, without flushing
6849 to typical _storage_.
6851 Linux supports NVRAM file systems thanks to either
6852 http://pramfs.sourceforge.net/[PRAMFS] or
6853 https://www.kernel.org/doc/Documentation/filesystems/dax.txt[DAX]{nbsp}+{nbsp}http://lkml.iu.edu/hypermail/linux/kernel/1504.1/03463.html[pmem]
6854 (requires Linux{nbsp}4.1+).
6856 This section does not describe how to operate such file systems;
6857 we assume that you have a working persistent memory file system.
6859 When you create a <<tracing-session,tracing session>>, you can specify
6860 the path of the shared memory holding the sub-buffers. If you specify a
6861 location on an NVRAM file system, then you can retrieve the latest
6862 recorded trace data when the system reboots after a crash.
6864 To record trace data on a persistent memory file system and retrieve the
6865 trace data after a system crash:
6867 . Create a tracing session with a sub-buffer shared memory path located
6868 on an NVRAM file system:
6873 $ lttng create my-session --shm-path=/path/to/shm
6877 . Configure the tracing session as usual with the man:lttng(1)
6878 command-line tool, and <<basic-tracing-session-control,start tracing>>.
6880 . After a system crash, use the man:lttng-crash(1) command-line tool to
6881 view the trace data recorded on the NVRAM file system:
6886 $ lttng-crash /path/to/shm
6890 The binary layout of the ring buffer files is not exactly the same as
6891 the trace files layout. This is why you need to use man:lttng-crash(1)
6892 instead of your preferred trace viewer directly.
6894 To convert the ring buffer files to LTTng trace files:
6896 * Use the opt:lttng-crash(1):--extract option of man:lttng-crash(1):
6901 $ lttng-crash --extract=/path/to/trace /path/to/shm
6907 [[notif-trigger-api]]
6908 === Get notified when a channel's buffer usage is too high or too low
6910 With LTTng's $$C/C++$$ notification and trigger API, your user
6911 application can get notified when the buffer usage of one or more
6912 <<channel,channels>> becomes too low or too high. You can use this API
6913 and enable or disable <<event,event rules>> during tracing to avoid
6914 <<channel-overwrite-mode-vs-discard-mode,discarded event records>>.
6916 .Have a user application get notified when an LTTng channel's buffer usage is too high.
6918 In this example, we create and build an application which gets notified
6919 when the buffer usage of a specific LTTng channel is higher than
6920 75{nbsp}%. We only print that it is the case in the example, but we
6921 could as well use the API of <<liblttng-ctl-lttng,`liblttng-ctl`>> to
6922 disable event rules when this happens.
6924 . Create the application's C source file:
6932 #include <lttng/domain.h>
6933 #include <lttng/action/action.h>
6934 #include <lttng/action/notify.h>
6935 #include <lttng/condition/condition.h>
6936 #include <lttng/condition/buffer-usage.h>
6937 #include <lttng/condition/evaluation.h>
6938 #include <lttng/notification/channel.h>
6939 #include <lttng/notification/notification.h>
6940 #include <lttng/trigger/trigger.h>
6941 #include <lttng/endpoint.h>
6943 int main(int argc, char *argv[])
6945 int exit_status = 0;
6946 struct lttng_notification_channel *notification_channel;
6947 struct lttng_condition *condition;
6948 struct lttng_action *action;
6949 struct lttng_trigger *trigger;
6950 const char *tracing_session_name;
6951 const char *channel_name;
6954 tracing_session_name = argv[1];
6955 channel_name = argv[2];
6958 * Create a notification channel. A notification channel
6959 * connects the user application to the LTTng session daemon.
6960 * This notification channel can be used to listen to various
6961 * types of notifications.
6963 notification_channel = lttng_notification_channel_create(
6964 lttng_session_daemon_notification_endpoint);
6967 * Create a "high buffer usage" condition. In this case, the
6968 * condition is reached when the buffer usage is greater than or
6969 * equal to 75 %. We create the condition for a specific tracing
6970 * session name, channel name, and for the user space tracing
6973 * The "low buffer usage" condition type also exists.
6975 condition = lttng_condition_buffer_usage_high_create();
6976 lttng_condition_buffer_usage_set_threshold_ratio(condition, .75);
6977 lttng_condition_buffer_usage_set_session_name(
6978 condition, tracing_session_name);
6979 lttng_condition_buffer_usage_set_channel_name(condition,
6981 lttng_condition_buffer_usage_set_domain_type(condition,
6985 * Create an action (get a notification) to take when the
6986 * condition created above is reached.
6988 action = lttng_action_notify_create();
6991 * Create a trigger. A trigger associates a condition to an
6992 * action: the action is executed when the condition is reached.
6994 trigger = lttng_trigger_create(condition, action);
6996 /* Register the trigger to LTTng. */
6997 lttng_register_trigger(trigger);
7000 * Now that we have registered a trigger, a notification will be
7001 * emitted everytime its condition is met. To receive this
7002 * notification, we must subscribe to notifications that match
7003 * the same condition.
7005 lttng_notification_channel_subscribe(notification_channel,
7009 * Notification loop. You can put this in a dedicated thread to
7010 * avoid blocking the main thread.
7013 struct lttng_notification *notification;
7014 enum lttng_notification_channel_status status;
7015 const struct lttng_evaluation *notification_evaluation;
7016 const struct lttng_condition *notification_condition;
7017 double buffer_usage;
7019 /* Receive the next notification. */
7020 status = lttng_notification_channel_get_next_notification(
7021 notification_channel, ¬ification);
7024 case LTTNG_NOTIFICATION_CHANNEL_STATUS_OK:
7026 case LTTNG_NOTIFICATION_CHANNEL_STATUS_NOTIFICATIONS_DROPPED:
7028 * The session daemon can drop notifications if
7029 * a monitoring application is not consuming the
7030 * notifications fast enough.
7033 case LTTNG_NOTIFICATION_CHANNEL_STATUS_CLOSED:
7035 * The notification channel has been closed by the
7036 * session daemon. This is typically caused by a session
7037 * daemon shutting down.
7041 /* Unhandled conditions or errors. */
7047 * A notification provides, amongst other things:
7049 * * The condition that caused this notification to be
7051 * * The condition evaluation, which provides more
7052 * specific information on the evaluation of the
7055 * The condition evaluation provides the buffer usage
7056 * value at the moment the condition was reached.
7058 notification_condition = lttng_notification_get_condition(
7060 notification_evaluation = lttng_notification_get_evaluation(
7063 /* We're subscribed to only one condition. */
7064 assert(lttng_condition_get_type(notification_condition) ==
7065 LTTNG_CONDITION_TYPE_BUFFER_USAGE_HIGH);
7068 * Get the exact sampled buffer usage from the
7069 * condition evaluation.
7071 lttng_evaluation_buffer_usage_get_usage_ratio(
7072 notification_evaluation, &buffer_usage);
7075 * At this point, instead of printing a message, we
7076 * could do something to reduce the channel's buffer
7077 * usage, like disable specific events.
7079 printf("Buffer usage is %f %% in tracing session \"%s\", "
7080 "user space channel \"%s\".\n", buffer_usage * 100,
7081 tracing_session_name, channel_name);
7082 lttng_notification_destroy(notification);
7086 lttng_action_destroy(action);
7087 lttng_condition_destroy(condition);
7088 lttng_trigger_destroy(trigger);
7089 lttng_notification_channel_destroy(notification_channel);
7095 . Build the `notif-app` application, linking it to `liblttng-ctl`:
7100 $ gcc -o notif-app notif-app.c -llttng-ctl
7104 . <<creating-destroying-tracing-sessions,Create a tracing session>>,
7105 <<enabling-disabling-events,create an event rule>> matching all the
7106 user space tracepoints, and
7107 <<basic-tracing-session-control,start tracing>>:
7112 $ lttng create my-session
7113 $ lttng enable-event --userspace --all
7118 If you create the channel manually with the man:lttng-enable-channel(1)
7119 command, you can control how frequently are the current values of the
7120 channel's properties sampled to evaluate user conditions with the
7121 opt:lttng-enable-channel(1):--monitor-timer option.
7123 . Run the `notif-app` application. This program accepts the
7124 <<tracing-session,tracing session>> name and the user space channel
7125 name as its two first arguments. The channel which LTTng automatically
7126 creates with the man:lttng-enable-event(1) command above is named
7132 $ ./notif-app my-session channel0
7136 . In another terminal, run an application with a very high event
7137 throughput so that the 75{nbsp}% buffer usage condition is reached.
7139 In the first terminal, the application should print lines like this:
7142 Buffer usage is 81.45197 % in tracing session "my-session", user space
7146 If you don't see anything, try modifying the condition in
7147 path:{notif-app.c} to a lower value (0.1, for example), rebuilding it
7148 (step{nbsp}2) and running it again (step{nbsp}4).
7155 [[lttng-modules-ref]]
7156 === noch:{LTTng-modules}
7160 [[lttng-tracepoint-enum]]
7161 ==== `LTTNG_TRACEPOINT_ENUM()` usage
7163 Use the `LTTNG_TRACEPOINT_ENUM()` macro to define an enumeration:
7167 LTTNG_TRACEPOINT_ENUM(name, TP_ENUM_VALUES(entries))
7172 * `name` with the name of the enumeration (C identifier, unique
7173 amongst all the defined enumerations).
7174 * `entries` with a list of enumeration entries.
7176 The available enumeration entry macros are:
7178 +ctf_enum_value(__name__, __value__)+::
7179 Entry named +__name__+ mapped to the integral value +__value__+.
7181 +ctf_enum_range(__name__, __begin__, __end__)+::
7182 Entry named +__name__+ mapped to the range of integral values between
7183 +__begin__+ (included) and +__end__+ (included).
7185 +ctf_enum_auto(__name__)+::
7186 Entry named +__name__+ mapped to the integral value following the
7187 last mapping's value.
7189 The last value of a `ctf_enum_value()` entry is its +__value__+
7192 The last value of a `ctf_enum_range()` entry is its +__end__+ parameter.
7194 If `ctf_enum_auto()` is the first entry in the list, its integral
7197 Use the `ctf_enum()` <<lttng-modules-tp-fields,field definition macro>>
7198 to use a defined enumeration as a tracepoint field.
7200 .Define an enumeration with `LTTNG_TRACEPOINT_ENUM()`.
7204 LTTNG_TRACEPOINT_ENUM(
7207 ctf_enum_auto("AUTO: EXPECT 0")
7208 ctf_enum_value("VALUE: 23", 23)
7209 ctf_enum_value("VALUE: 27", 27)
7210 ctf_enum_auto("AUTO: EXPECT 28")
7211 ctf_enum_range("RANGE: 101 TO 303", 101, 303)
7212 ctf_enum_auto("AUTO: EXPECT 304")
7220 [[lttng-modules-tp-fields]]
7221 ==== Tracepoint fields macros (for `TP_FIELDS()`)
7223 [[tp-fast-assign]][[tp-struct-entry]]The available macros to define
7224 tracepoint fields, which must be listed within `TP_FIELDS()` in
7225 `LTTNG_TRACEPOINT_EVENT()`, are:
7227 [role="func-desc growable",cols="asciidoc,asciidoc"]
7228 .Available macros to define LTTng-modules tracepoint fields
7230 |Macro |Description and parameters
7233 +ctf_integer(__t__, __n__, __e__)+
7235 +ctf_integer_nowrite(__t__, __n__, __e__)+
7237 +ctf_user_integer(__t__, __n__, __e__)+
7239 +ctf_user_integer_nowrite(__t__, __n__, __e__)+
7241 Standard integer, displayed in base{nbsp}10.
7244 Integer C type (`int`, `long`, `size_t`, ...).
7250 Argument expression.
7253 +ctf_integer_hex(__t__, __n__, __e__)+
7255 +ctf_user_integer_hex(__t__, __n__, __e__)+
7257 Standard integer, displayed in base{nbsp}16.
7266 Argument expression.
7268 |+ctf_integer_oct(__t__, __n__, __e__)+
7270 Standard integer, displayed in base{nbsp}8.
7279 Argument expression.
7282 +ctf_integer_network(__t__, __n__, __e__)+
7284 +ctf_user_integer_network(__t__, __n__, __e__)+
7286 Integer in network byte order (big-endian), displayed in base{nbsp}10.
7295 Argument expression.
7298 +ctf_integer_network_hex(__t__, __n__, __e__)+
7300 +ctf_user_integer_network_hex(__t__, __n__, __e__)+
7302 Integer in network byte order, displayed in base{nbsp}16.
7311 Argument expression.
7314 +ctf_enum(__N__, __t__, __n__, __e__)+
7316 +ctf_enum_nowrite(__N__, __t__, __n__, __e__)+
7318 +ctf_user_enum(__N__, __t__, __n__, __e__)+
7320 +ctf_user_enum_nowrite(__N__, __t__, __n__, __e__)+
7325 Name of a <<lttng-tracepoint-enum,previously defined enumeration>>.
7328 Integer C type (`int`, `long`, `size_t`, ...).
7334 Argument expression.
7337 +ctf_string(__n__, __e__)+
7339 +ctf_string_nowrite(__n__, __e__)+
7341 +ctf_user_string(__n__, __e__)+
7343 +ctf_user_string_nowrite(__n__, __e__)+
7345 Null-terminated string; undefined behavior if +__e__+ is `NULL`.
7351 Argument expression.
7354 +ctf_array(__t__, __n__, __e__, __s__)+
7356 +ctf_array_nowrite(__t__, __n__, __e__, __s__)+
7358 +ctf_user_array(__t__, __n__, __e__, __s__)+
7360 +ctf_user_array_nowrite(__t__, __n__, __e__, __s__)+
7362 Statically-sized array of integers.
7365 Array element C type.
7371 Argument expression.
7377 +ctf_array_bitfield(__t__, __n__, __e__, __s__)+
7379 +ctf_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7381 +ctf_user_array_bitfield(__t__, __n__, __e__, __s__)+
7383 +ctf_user_array_bitfield_nowrite(__t__, __n__, __e__, __s__)+
7385 Statically-sized array of bits.
7387 The type of +__e__+ must be an integer type. +__s__+ is the number
7388 of elements of such type in +__e__+, not the number of bits.
7391 Array element C type.
7397 Argument expression.
7403 +ctf_array_text(__t__, __n__, __e__, __s__)+
7405 +ctf_array_text_nowrite(__t__, __n__, __e__, __s__)+
7407 +ctf_user_array_text(__t__, __n__, __e__, __s__)+
7409 +ctf_user_array_text_nowrite(__t__, __n__, __e__, __s__)+
7411 Statically-sized array, printed as text.
7413 The string does not need to be null-terminated.
7416 Array element C type (always `char`).
7422 Argument expression.
7428 +ctf_sequence(__t__, __n__, __e__, __T__, __E__)+
7430 +ctf_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7432 +ctf_user_sequence(__t__, __n__, __e__, __T__, __E__)+
7434 +ctf_user_sequence_nowrite(__t__, __n__, __e__, __T__, __E__)+
7436 Dynamically-sized array of integers.
7438 The type of +__E__+ must be unsigned.
7441 Array element C type.
7447 Argument expression.
7450 Length expression C type.
7456 +ctf_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7458 +ctf_user_sequence_hex(__t__, __n__, __e__, __T__, __E__)+
7460 Dynamically-sized array of integers, displayed in base{nbsp}16.
7462 The type of +__E__+ must be unsigned.
7465 Array element C type.
7471 Argument expression.
7474 Length expression C type.
7479 |+ctf_sequence_network(__t__, __n__, __e__, __T__, __E__)+
7481 Dynamically-sized array of integers in network byte order (big-endian),
7482 displayed in base{nbsp}10.
7484 The type of +__E__+ must be unsigned.
7487 Array element C type.
7493 Argument expression.
7496 Length expression C type.
7502 +ctf_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7504 +ctf_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7506 +ctf_user_sequence_bitfield(__t__, __n__, __e__, __T__, __E__)+
7508 +ctf_user_sequence_bitfield_nowrite(__t__, __n__, __e__, __T__, __E__)+
7510 Dynamically-sized array of bits.
7512 The type of +__e__+ must be an integer type. +__s__+ is the number
7513 of elements of such type in +__e__+, not the number of bits.
7515 The type of +__E__+ must be unsigned.
7518 Array element C type.
7524 Argument expression.
7527 Length expression C type.
7533 +ctf_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7535 +ctf_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7537 +ctf_user_sequence_text(__t__, __n__, __e__, __T__, __E__)+
7539 +ctf_user_sequence_text_nowrite(__t__, __n__, __e__, __T__, __E__)+
7541 Dynamically-sized array, displayed as text.
7543 The string does not need to be null-terminated.
7545 The type of +__E__+ must be unsigned.
7547 The behaviour is undefined if +__e__+ is `NULL`.
7550 Sequence element C type (always `char`).
7556 Argument expression.
7559 Length expression C type.
7565 Use the `_user` versions when the argument expression, `e`, is
7566 a user space address. In the cases of `ctf_user_integer*()` and
7567 `ctf_user_float*()`, `&e` must be a user space address, thus `e` must
7570 The `_nowrite` versions omit themselves from the session trace, but are
7571 otherwise identical. This means the `_nowrite` fields won't be written
7572 in the recorded trace. Their primary purpose is to make some
7573 of the event context available to the
7574 <<enabling-disabling-events,event filters>> without having to
7575 commit the data to sub-buffers.
7581 Terms related to LTTng and to tracing in general:
7584 The http://diamon.org/babeltrace[Babeltrace] project, which includes:
7586 * The cmd:babeltrace (Babeltrace{nbsp}1) or cmd:babeltrace2
7587 (Babeltrace{nbsp}2) command.
7588 * Libraries with a C{nbsp}API.
7589 * Python{nbsp}3 bindings.
7590 * Plugins (Babeltrace{nbsp}2).
7592 [[def-buffering-scheme]]<<channel-buffering-schemes,buffering scheme>>::
7593 A layout of <<def-sub-buffer,sub-buffers>> applied to a given channel.
7595 [[def-channel]]<<channel,channel>>::
7596 An entity which is responsible for a set of
7597 <<def-ring-buffer,ring buffers>>.
7599 <<def-event-rule,Event rules>> are always attached to a specific
7603 A source of time for a <<def-tracer,tracer>>.
7605 [[def-consumer-daemon]]<<lttng-consumerd,consumer daemon>>::
7606 A process which is responsible for consuming the full
7607 <<def-sub-buffer,sub-buffers>> and write them to a file system or
7608 send them over the network.
7610 [[def-current-trace-chunk]]current trace chunk::
7611 A <<def-trace-chunk,trace chunk>> which includes the current content
7612 of all the <<def-tracing-session-rotation,tracing session>>'s
7613 <<def-sub-buffer,sub-buffers>> and the stream files produced since the
7614 latest event amongst:
7616 * The creation of the <<def-tracing-session,tracing session>>.
7617 * The last tracing session rotation, if any.
7619 <<channel-overwrite-mode-vs-discard-mode,discard mode>>::
7620 The <<def-event-record-loss-mode,event record loss mode>> in which
7621 the <<def-tracer,tracer>> _discards_ new event records when there's no
7622 <<def-sub-buffer,sub-buffer>> space left to store them.
7624 [[def-event]]event::
7625 The consequence of the execution of an
7626 <<def-instrumentation-point,instrumentation point>>, like a
7627 <<def-tracepoint,tracepoint>> that you manually place in some source
7628 code, or a Linux kernel kprobe.
7630 An event is said to _occur_ at a specific time. <<def-lttng,LTTng>> can
7631 take various actions upon the occurrence of an event, like record the
7632 event's payload to a <<def-sub-buffer,sub-buffer>>.
7634 [[def-event-name]]event name::
7635 The name of an <<def-event,event>>, which is also the name of the
7636 <<def-event-record,event record>>.
7638 This is also called the _instrumentation point name_.
7640 [[def-event-record]]event record::
7641 A record, in a <<def-trace,trace>>, of the payload of an
7642 <<def-event,event>> which occured.
7644 [[def-event-record-loss-mode]]<<channel-overwrite-mode-vs-discard-mode,event record loss mode>>::
7645 The mechanism by which event records of a given
7646 <<def-channel,channel>> are lost (not recorded) when there is no
7647 <<def-sub-buffer,sub-buffer>> space left to store them.
7649 [[def-event-rule]]<<event,event rule>>::
7650 Set of conditions which must be satisfied for one or more occuring
7651 <<def-event,events>> to be recorded.
7653 `java.util.logging`::
7655 https://docs.oracle.com/javase/7/docs/api/java/util/logging/package-summary.html[core logging facilities].
7657 <<instrumenting,instrumentation>>::
7658 The use of <<def-lttng,LTTng>> probes to make a piece of software
7661 [[def-instrumentation-point]]instrumentation point::
7662 A point in the execution path of a piece of software that, when
7663 reached by this execution, can emit an <<def-event,event>>.
7665 instrumentation point name::
7666 See _<<def-event-name,event name>>_.
7669 A http://logging.apache.org/log4j/1.2/[logging library] for Java
7670 developed by the Apache Software Foundation.
7673 Level of severity of a log statement or user space
7674 <<def-instrumentation-point,instrumentation point>>.
7676 [[def-lttng]]LTTng::
7677 The _Linux Trace Toolkit: next generation_ project.
7679 <<lttng-cli,cmd:lttng>>::
7680 A command-line tool provided by the <<def-lttng-tools,LTTng-tools>>
7681 project which you can use to send and receive control messages to and
7682 from a <<def-session-daemon,session daemon>>.
7685 The https://github.com/lttng/lttng-analyses[LTTng analyses] project,
7686 which is a set of analyzing programs that you can use to obtain a
7687 higher level view of an <<def-lttng,LTTng>> <<def-trace,trace>>.
7689 cmd:lttng-consumerd::
7690 The name of the <<def-consumer-daemon,consumer daemon>> program.
7693 A utility provided by the <<def-lttng-tools,LTTng-tools>> project
7694 which can convert <<def-ring-buffer,ring buffer>> files (usually
7695 <<persistent-memory-file-systems,saved on a persistent memory file
7696 system>>) to <<def-trace,trace>> files.
7698 See man:lttng-crash(1).
7700 LTTng Documentation::
7703 <<lttng-live,LTTng live>>::
7704 A communication protocol between the <<lttng-relayd,relay daemon>> and
7705 live viewers which makes it possible to see <<def-event-record,event
7706 records>> "live", as they are received by the
7707 <<def-relay-daemon,relay daemon>>.
7709 <<lttng-modules,LTTng-modules>>::
7710 The https://github.com/lttng/lttng-modules[LTTng-modules] project,
7711 which contains the Linux kernel modules to make the Linux kernel
7712 <<def-instrumentation-point,instrumentation points>> available for
7713 <<def-lttng,LTTng>> tracing.
7716 The name of the <<def-relay-daemon,relay daemon>> program.
7718 cmd:lttng-sessiond::
7719 The name of the <<def-session-daemon,session daemon>> program.
7721 [[def-lttng-tools]]LTTng-tools::
7722 The https://github.com/lttng/lttng-tools[LTTng-tools] project, which
7723 contains the various programs and libraries used to
7724 <<controlling-tracing,control tracing>>.
7726 [[def-lttng-ust]]<<lttng-ust,LTTng-UST>>::
7727 The https://github.com/lttng/lttng-ust[LTTng-UST] project, which
7728 contains libraries to instrument
7729 <<def-user-application,user applications>>.
7731 <<lttng-ust-agents,LTTng-UST Java agent>>::
7732 A Java package provided by the <<def-lttng-ust,LTTng-UST>> project to
7733 allow the LTTng instrumentation of `java.util.logging` and Apache
7734 log4j{nbsp}1.2 logging statements.
7736 <<lttng-ust-agents,LTTng-UST Python agent>>::
7737 A Python package provided by the <<def-lttng-ust,LTTng-UST>> project
7738 to allow the <<def-lttng,LTTng>> instrumentation of Python logging
7741 <<channel-overwrite-mode-vs-discard-mode,overwrite mode>>::
7742 The <<def-event-record-loss-mode,event record loss mode>> in which new
7743 <<def-event-record,event records>> _overwrite_ older event records
7744 when there's no <<def-sub-buffer,sub-buffer>> space left to store
7747 <<channel-buffering-schemes,per-process buffering>>::
7748 A <<def-buffering-scheme,buffering scheme>> in which each instrumented
7749 process has its own <<def-sub-buffer,sub-buffers>> for a given user
7750 space <<def-channel,channel>>.
7752 <<channel-buffering-schemes,per-user buffering>>::
7753 A <<def-buffering-scheme,buffering scheme>> in which all the processes
7754 of a Unix user share the same <<def-sub-buffer,sub-buffers>> for a
7755 given user space <<def-channel,channel>>.
7757 [[def-relay-daemon]]<<lttng-relayd,relay daemon>>::
7758 A process which is responsible for receiving the <<def-trace,trace>>
7759 data which a distant <<def-consumer-daemon,consumer daemon>> sends.
7761 [[def-ring-buffer]]ring buffer::
7762 A set of <<def-sub-buffer,sub-buffers>>.
7765 See _<<def-tracing-session-rotation,tracing session rotation>>_.
7767 [[def-session-daemon]]<<lttng-sessiond,session daemon>>::
7768 A process which receives control commands from you and orchestrates
7769 the <<def-tracer,tracers>> and various <<def-lttng,LTTng>> daemons.
7771 <<taking-a-snapshot,snapshot>>::
7772 A copy of the current data of all the <<def-sub-buffer,sub-buffers>>
7773 of a given <<def-tracing-session,tracing session>>, saved as
7774 <<def-trace,trace>> files.
7776 [[def-sub-buffer]]sub-buffer::
7777 One part of an <<def-lttng,LTTng>> <<def-ring-buffer,ring buffer>>
7778 which contains <<def-event-record,event records>>.
7781 The time information attached to an <<def-event,event>> when it is
7784 [[def-trace]]trace (_noun_)::
7787 * One http://diamon.org/ctf/[CTF] metadata stream file.
7788 * One or more CTF data stream files which are the concatenations of one
7789 or more flushed <<def-sub-buffer,sub-buffers>>.
7791 [[def-trace-verb]]trace (_verb_)::
7792 The action of recording the <<def-event,events>> emitted by an
7793 application or by a system, or to initiate such recording by
7794 controlling a <<def-tracer,tracer>>.
7796 [[def-trace-chunk]]trace chunk::
7797 A self-contained <<def-trace,trace>> which is part of a
7798 <<def-tracing-session,tracing session>>. Each
7799 <<def-tracing-session-rotation, tracing session rotation>> produces a
7800 <<def-trace-chunk-archive,trace chunk archive>>.
7802 [[def-trace-chunk-archive]]trace chunk archive::
7803 The result of a <<def-tracing-session-rotation, tracing session rotation>>.
7805 <<def-lttng,LTTng>> does not manage any trace chunk archive, even if its
7806 containing <<def-tracing-session,tracing session>> is still active: you
7807 are free to read it, modify it, move it, or remove it.
7810 The http://tracecompass.org[Trace Compass] project and application.
7812 [[def-tracepoint]]tracepoint::
7813 An instrumentation point using the tracepoint mechanism of the Linux
7814 kernel or of <<def-lttng-ust,LTTng-UST>>.
7816 tracepoint definition::
7817 The definition of a single <<def-tracepoint,tracepoint>>.
7820 The name of a <<def-tracepoint,tracepoint>>.
7822 [[def-tracepoint-provider]]tracepoint provider::
7823 A set of functions providing <<def-tracepoint,tracepoints>> to an
7824 instrumented <<def-user-application,user application>>.
7826 Not to be confused with a <<def-tracepoint-provider-package,tracepoint
7827 provider package>>: many tracepoint providers can exist within a
7828 tracepoint provider package.
7830 [[def-tracepoint-provider-package]]tracepoint provider package::
7831 One or more <<def-tracepoint-provider,tracepoint providers>> compiled
7832 as an https://en.wikipedia.org/wiki/Object_file[object file] or as a
7833 link:https://en.wikipedia.org/wiki/Library_(computing)#Shared_libraries[shared
7836 [[def-tracer]]tracer::
7837 A software which records emitted <<def-event,events>>.
7839 <<domain,tracing domain>>::
7840 A namespace for <<def-event,event>> sources.
7842 <<tracing-group,tracing group>>::
7843 The Unix group in which a Unix user can be to be allowed to
7844 <<def-trace-verb,trace>> the Linux kernel.
7846 [[def-tracing-session]]<<tracing-session,tracing session>>::
7847 A stateful dialogue between you and a <<lttng-sessiond,session daemon>>.
7849 [[def-tracing-session-rotation]]<<session-rotation,tracing session rotation>>::
7850 The action of archiving the
7851 <<def-current-trace-chunk,current trace chunk>> of a
7852 <<def-tracing-session,tracing session>>.
7854 [[def-user-application]]user application::
7855 An application running in user space, as opposed to a Linux kernel
7856 module, for example.