1 .TH "LTTNG-UST" "3" "February 16, 2012" "" ""
4 lttng-ust \(em Linux Trace Toolkit Next Generation User-Space Tracer 2.x
10 Link liblttng-ust.so with applications, following this manpage.
15 LTTng-UST, the Linux Trace Toolkit Next Generation Userspace Tracer, is a
16 port of the low-overhead tracing capabilities of the LTTng kernel tracer
17 to user-space. The library "liblttng-ust" enables tracing of
18 applications and libraries.
20 .SH "USAGE WITH TRACEF"
22 The simplest way to add instrumentation to your code is by far the
23 tracef() API. To do it, in a nutshell:
25 1) #include <lttng/tracef.h>
27 2) /* in your code, use like a printf */
28 tracef("my message, this integer %d", 1234);
30 3) Link your program against liblttng-ust.so.
32 4) Enable UST events when tracing with the following sequence of commands
36 lttng enable-event -u -a
38 [... run your program ...]
44 If you want to have more flexibility and control on the event names,
45 payload typing, etc, you can continue reading on and use the tracepoints
46 below. "tracef()" is there for quick and dirty ad hoc instrumentation,
47 whereas tracepoint.h is meant for thorough instrumentation of a code
48 base to be integrated with an upstream project.
51 .SH "USAGE WITH TRACELOG"
53 If you want to migrate existing logging (info, errors, ...)
54 to LTTng UST, you can use the tracelog() interface.
55 To do it, in a nutshell:
57 1) #include <lttng/tracelog.h>
59 2) /* in your code, use like a printf, with extra loglevel info. */
60 tracelog(TRACE_INFO, "Message with integer %d", 1234);
62 3) Link your program against liblttng-ust.so.
64 4) Enable UST events when tracing with the following sequence of commands
68 lttng enable-event -u "lttng_ust_tracelog:*"
70 [... run your program ...]
76 You can replace the enable-event line above with a selection of
79 lttng enable-event -u -a --loglevel TRACE_INFO
81 Which will gather all events from TRACE_INFO and more important
86 .SH "USAGE WITH TRACEPOINT"
88 The simple way to generate the lttng-ust tracepoint probes is to use the
89 lttng-gen-tp(1) tool. See the lttng-gen-tp(1) manpage for explanation.
93 Here is the way to do it manually, without the lttng-gen-tp(1) helper
94 script, through an example:
97 .SH "CREATION OF TRACEPOINT PROVIDER"
101 To create a tracepoint provider, within a build tree similar to
102 examples/easy-ust installed with lttng-ust documentation, see
103 sample_component_provider.h for the general layout. You will need to
104 define TRACEPOINT_CREATE_PROBES before including your tracepoint
105 provider probe in one source file of your application. See tp.c from
106 easy-ust for an example of a tracepoint probe source file. This manpage
107 will focus on the various types that can be recorded into a trace
112 * provider name, not a variable but a string starting with a
113 * letter and containing either letters, numbers or underscores.
114 * Needs to be the same as TRACEPOINT_PROVIDER. Needs to
115 * follow the namespacing guide-lines in lttng/tracepoint.h:
117 * Must be included before include tracepoint provider
119 * ex.: project_component_event
121 * Optional company name goes here
122 * ex.: com_efficios_project_component_event
124 * In this example, "sample" is the project, and "component" is the
130 * tracepoint name, characters permitted follow the same
131 * constraints as the provider name. The name of this example
132 * event is "sample_event".
137 * TP_ARGS macro contains the arguments passed for the tracepoint
138 * it is in the following format
139 * TP_ARGS(type1, name1, type2, name2, ... type10,
141 * where there can be from zero to ten elements.
142 * typeN is the datatype, such as int, struct or double **.
143 * name is the variable name (in "int myInt" the name would be
145 * TP_ARGS() is valid to mean no arguments
146 * TP_ARGS(void) is valid too
148 TP_ARGS(int, anint, int, netint, long *, values,
149 char *, text, size_t, textlen,
150 double, doublearg, float, floatarg),
153 * TP_FIELDS describes how to write the fields of the trace event.
154 * You can put expressions in the "argument expression" area,
155 * typically using the input arguments from TP_ARGS.
159 * ctf_integer: standard integer field.
160 * args: (type, field name, argument expression)
162 ctf_integer(int, intfield, anint)
163 ctf_integer(long, longfield, anint)
166 * ctf_integer_hex: integer field printed as hexadecimal.
167 * args: (type, field name, argument expression)
169 ctf_integer_hex(int, intfield2, anint)
172 * ctf_integer_network: integer field in network byte
173 * order. (_hex: printed as hexadecimal too)
174 * args: (type, field name, argument expression)
176 ctf_integer_network(int, netintfield, netint)
177 ctf_integer_network_hex(int, netintfieldhex, netint)
180 * ctf_array: a statically-sized array.
181 * args: (type, field name, argument expression, value)
183 ctf_array(long, arrfield1, values, 3)
186 * ctf_array_text: a statically-sized array, printed as
187 * a string. No need to be terminated by a null
189 * Behavior is undefined if "text" argument is NULL.
191 ctf_array_text(char, arrfield2, text, 10)
194 * ctf_sequence: a dynamically-sized array.
195 * args: (type, field name, argument expression,
196 * type of length expression, length expression)
197 * The "type of length expression" needs to be an
198 * unsigned type. As a reminder, "unsigned char" should
199 * be preferred to "char", since the signedness of
200 * "char" is implementation-defined.
201 * Behavior is undefined if "text" argument is NULL.
203 ctf_sequence(char, seqfield1, text,
207 * ctf_sequence_text: a dynamically-sized array, printed
208 * as string. No need to be null-terminated.
209 * Behavior is undefined if "text" argument is NULL.
211 ctf_sequence_text(char, seqfield2, text,
215 * ctf_string: null-terminated string.
216 * args: (field name, argument expression)
217 * Behavior is undefined if "text" argument is NULL.
219 ctf_string(stringfield, text)
222 * ctf_float: floating-point number.
223 * args: (type, field name, argument expression)
225 ctf_float(float, floatfield, floatarg)
226 ctf_float(double, doublefield, doublearg)
230 There can be an arbitrary number of tracepoint providers within an
231 application, but they must each have their own provider name. Duplicate
232 provider names are not allowed.
236 .SH "ASSIGNING LOGLEVEL TO EVENTS"
240 Optionally, a loglevel can be assigned to a TRACEPOINT_EVENT using the
243 TRACEPOINT_LOGLEVEL(< [com_company_]project[_component] >,
244 < event >, < loglevel_name >)
246 The first field is the provider name, the second field is the name of
247 the tracepoint, and the third field is the loglevel name. A
248 TRACEPOINT_EVENT should be declared prior to the the TRACEPOINT_LOGLEVEL
249 for a given tracepoint name. The TRACEPOINT_PROVIDER must be already
250 declared before declaring a TRACEPOINT_LOGLEVEL.
252 The loglevels go from 0 to 14. Higher numbers imply the most verbosity
253 (higher event throughput expected.
255 Loglevels 0 through 6, and loglevel 14, match syslog(3) loglevels
256 semantic. Loglevels 7 through 13 offer more fine-grained selection of
263 action must be taken immediately
275 normal, but significant, condition
278 informational message
281 debug information with system-level scope (set of programs)
283 TRACE_DEBUG_PROGRAM 8
284 debug information with program-level scope (set of processes)
286 TRACE_DEBUG_PROCESS 9
287 debug information with process-level scope (set of modules)
289 TRACE_DEBUG_MODULE 10
290 debug information with module (executable/library) scope (set of
294 debug information with compilation unit scope (set of functions)
296 TRACE_DEBUG_FUNCTION 12
297 debug information with function-level scope
300 debug information with line-level scope (TRACEPOINT_EVENT default)
305 See lttng(1) for information on how to use LTTng-UST loglevels.
309 .SH "ADDING TRACEPOINTS TO YOUR CODE"
313 Include the provider header in each C files you plan to instrument,
314 following the building/linking directives in the next section.
316 For instance, add within a function:
318 tracepoint(ust_tests_hello, tptest, i, netint, values,
319 text, strlen(text), dbl, flt);
321 As a call to the tracepoint. It will only be activated when requested by
322 lttng(1) through lttng-sessiond(8).
324 Even though LTTng-UST supports tracepoint() call site duplicates having
325 the same provider and event name, it is recommended to use a
326 provider event name pair only once within the source code to help
327 map events back to their call sites when analyzing the trace.
329 Sometimes arguments to the probe are expensive to compute (e.g.
330 take call stack). To avoid the computation when the tracepoint is
331 disabled one can use more 'low level' tracepoint_enabled() and
332 do_tracepoint() macros as following:
334 if (tracepoint_enabled(ust_tests_hello, tptest)) {
335 /* prepare arguments */
336 do_tracepoint(ust_tests_hello, tptest, i, netint, values,
337 text, strlen(text), dbl, flt);
340 Here do_tracepoint() doesn't contain check if the tracepoint is enabled.
341 Using tracepoint() in such scenario is dangerous since it also contains
342 enabled check and thus race condition is possible in the following code
343 if the tracepoint has been enabled after check in tracepoint_enabled()
344 but before tracepoint():
346 if (tracepoint_enabled(provider, name)) { /* tracepoint is disabled */
349 /* tracepoint is enabled by 'lttng' tool */
350 tracepoint(provider, name, args); /* args wasn't prepared properly */
352 Note also that neither tracepoint_enabled() nor do_tracepoint() have
353 STAP_PROBEV() call so if you need it you should emit this call yourself.
357 .SH "BUILDING/LINKING THE TRACEPOINT PROVIDER"
360 There are 2 ways to compile the Tracepoint Provider with the
361 application: either statically or dynamically. Please follow
364 1) Compile the Tracepoint Provider with the application, either
365 directly or through a static library (.a):
366 - Into exactly one object of your application, define
367 "TRACEPOINT_DEFINE" and include the tracepoint provider.
368 - Use "\-I." for the compilation unit containing the tracepoint
369 provider include (e.g., tp.c).
370 - Link the application with "\-llttng-ust" and "\-ldl".
371 - Include the tracepoint provider header into all C files using
374 - doc/examples/easy-ust/ sample.c sample_component_provider.h tp.c
376 - doc/examples/hello-static-lib/ hello.c tp.c ust_test_hello.h Makefile
378 2) Compile the Tracepoint Provider separately from the application,
379 using dynamic linking:
380 - Into exactly one object of your application: define
381 "TRACEPOINT_DEFINE" _and_ also define
382 "TRACEPOINT_PROBE_DYNAMIC_LINKAGE", then include the tracepoint
384 - Include the tracepoint provider header into all instrumented C
385 files that use the provider.
386 - Compile the tracepoint provider with "\-I.".
387 - Link the tracepoint provider with "\-llttng-ust".
388 - Link application with "\-ldl".
389 - Set a LD_PRELOAD environment to preload the tracepoint provider
390 shared object before starting the application when tracing is
391 needed. Another way is to dlopen the tracepoint probe when needed
394 - doc/examples/demo demo.c tp*.c ust_tests_demo*.h demo-trace Makefile
396 - Note about dlclose() usage: it is not safe to use dlclose on a
397 provider shared object that is being actively used for tracing due
398 to a lack of reference counting from lttng-ust to the used shared
400 - Enable instrumentation and control tracing with the "lttng" command
401 from lttng-tools. See lttng-tools doc/quickstart.txt.
402 - Note for C++ support: although an application instrumented with
403 tracepoints can be compiled with g++, tracepoint probes should be
404 compiled with gcc (only tested with gcc so far).
408 .SH "USING LTTNG UST WITH DAEMONS"
411 Some extra care is needed when using liblttng-ust with daemon
412 applications that call fork(), clone(), or BSD rfork() without a
413 following exec() family system call. The library "liblttng-ust-fork.so"
414 needs to be preloaded for the application (launch with e.g.
415 LD_PRELOAD=liblttng-ust-fork.so appname).
422 Context information can be prepended by the tracer before each, or some,
423 events. The following context information is supported by LTTng-UST:
428 Virtual thread ID: thread ID as seen from the point of view of the
434 Virtual process ID: process ID as seen from the point of view of the
440 Instruction pointer: Enables recording of the exact location where a tracepoint
441 was emitted. Can be used to reverse-lookup the source location that caused the
447 Thread name, as set by exec() or prctl(). It is recommended that
448 programs set their thread name with prctl() before hitting the first
449 tracepoint for that thread.
454 Pthread identifier. Can be used on architectures where pthread_t maps
455 nicely to an unsigned long type.
458 .SH "BASE ADDRESS STATEDUMP"
461 If an application that uses liblttng-ust.so becomes part of a session,
462 information about its currently loaded shared objects will be traced to the
463 session at session-enable time. To record this information, the following event
466 .IP "ust_baddr_statedump:soinfo"
467 This event is used to trace a currently loaded shared object. The base address
468 (where the dynamic linker has placed the shared object) is recorded in the
469 "baddr" field. The path to the shared object gets recorded in the
470 "sopath" field (as string). The file size of the loaded object (in
471 bytes) is recorded to the "size" field and its time of last modification
472 (in seconds since Epoch) is recorded in the "mtime" field.
474 If the event above is enabled, a series of "ust_baddr_statedump:soinfo"
475 events is recorded at session-enable time. It represents the state of
476 currently loaded shared objects for the traced process. If this
477 information gets combined with the lttng-ust-dl(3) instrumentation, all
478 aspects of dynamic loading that are relevant for symbol and
479 line number lookup are traced by LTTng.
481 .SH "ENVIRONMENT VARIABLES"
484 .IP "LTTNG_UST_DEBUG"
485 Activate liblttng-ust debug and error output.
487 .IP "LTTNG_UST_REGISTER_TIMEOUT"
488 The environment variable "LTTNG_UST_REGISTER_TIMEOUT" can be used to
489 specify how long the applications should wait for sessiond
490 "registration done" command before proceeding to execute the main
491 program. The default is 3000ms (3 seconds). The timeout value is
492 specified in milliseconds. The value 0 means "don't wait". The value
493 \-1 means "wait forever". Setting this environment variable to 0 is
494 recommended for applications with time constraints on the process
497 .IP "LTTNG_UST_WITHOUT_BADDR_STATEDUMP"
498 Prevent liblttng-ust to perform a base-address statedump on session-enable.
504 lttng-gen-tp(1), lttng(1), babeltrace(1), lttng-ust-cyg-profile(3),
505 lttng-ust-dl(3), lttng-sessiond(8)
511 Older lttng-ust libraries reject more recent, and incompatible, probe
512 providers. Newer lttng-ust libraries accept older probe providers, even
513 though some newer features might not be available with those providers.
519 LTTng-UST 2.0 and 2.1 lttng-ust libraries do not check for probe
520 provider version compatibility. This can lead to out-of-bound accesses
521 when using a more recent probe provider with an older lttng-ust library.
522 These error only trigger when tracing is active. This issue has been
523 fixed in LTTng-UST 2.2.
525 If you encounter any issues or usability problem, please report it on
526 our mailing list <lttng-dev@lists.lttng.org> to help improve this
530 liblttng-ust is distributed under the GNU Lesser General Public License
531 version 2.1. The headers are distributed under the MIT license.
533 See http://lttng.org for more information on the LTTng project.
535 Mailing list for support and development: <lttng-dev@lists.lttng.org>.
537 You can find us on IRC server irc.oftc.net (OFTC) in #lttng.
541 Thanks to Ericsson for funding this work, providing real-life use-cases,
544 Special thanks to Michel Dagenais and the DORSAL laboratory at
545 Polytechnique de Montreal for the LTTng journey.
550 liblttng-ust was originally written by Mathieu Desnoyers, with additional
551 contributions from various other people. It is currently maintained by
552 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>.