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(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 INFO
81 Which will gather all events from INFO and more important loglevels.
85 .SH "USAGE WITH TRACEPOINT"
87 The simple way to generate the lttng-ust tracepoint probes is to use the
88 lttng-gen-tp(1) tool. See the lttng-gen-tp(1) manpage for explanation.
92 Here is the way to do it manually, without the lttng-gen-tp(1) helper
93 script, through an example:
96 .SH "CREATION OF TRACEPOINT PROVIDER"
100 To create a tracepoint provider, within a build tree similar to
101 examples/easy-ust installed with lttng-ust documentation, see
102 sample_component_provider.h for the general layout. You will need to
103 define TRACEPOINT_CREATE_PROBES before including your tracepoint
104 provider probe in one source file of your application. See tp.c from
105 easy-ust for an example of a tracepoint probe source file. This manpage
106 will focus on the various types that can be recorded into a trace
111 * provider name, not a variable but a string starting with a
112 * letter and containing either letters, numbers or underscores.
113 * Needs to be the same as TRACEPOINT_PROVIDER. Needs to
114 * follow the namespacing guide-lines in lttng/tracepoint.h:
116 * Must be included before include tracepoint provider
118 * ex.: project_component_event
120 * Optional company name goes here
121 * ex.: com_efficios_project_component_event
123 * In this example, "sample" is the project, and "component" is the
129 * tracepoint name, characters permitted follow the same
130 * constraints as the provider name. The name of this example
131 * event is "sample_event".
136 * TP_ARGS macro contains the arguments passed for the tracepoint
137 * it is in the following format
138 * TP_ARGS(type1, name1, type2, name2, ... type10,
140 * where there can be from zero to ten elements.
141 * typeN is the datatype, such as int, struct or double **.
142 * name is the variable name (in "int myInt" the name would be
144 * TP_ARGS() is valid to mean no arguments
145 * TP_ARGS(void) is valid too
147 TP_ARGS(int, anint, int, netint, long *, values,
148 char *, text, size_t, textlen,
149 double, doublearg, float, floatarg),
152 * TP_FIELDS describes how to write the fields of the trace event.
153 * You can put expressions in the "argument expression" area,
154 * typically using the input arguments from TP_ARGS.
158 * ctf_integer: standard integer field.
159 * args: (type, field name, argument expression)
161 ctf_integer(int, intfield, anint)
162 ctf_integer(long, longfield, anint)
165 * ctf_integer_hex: integer field printed as hexadecimal.
166 * args: (type, field name, argument expression)
168 ctf_integer_hex(int, intfield2, anint)
171 * ctf_integer_network: integer field in network byte
172 * order. (_hex: printed as hexadecimal too)
173 * args: (type, field name, argument expression)
175 ctf_integer_network(int, netintfield, netint)
176 ctf_integer_network_hex(int, netintfieldhex, netint)
179 * ctf_array: a statically-sized array.
180 * args: (type, field name, argument expression, value)
182 ctf_array(long, arrfield1, values, 3)
185 * ctf_array_text: a statically-sized array, printed as
186 * a string. No need to be terminated by a null
188 * Behavior is undefined if "text" argument is NULL.
190 ctf_array_text(char, arrfield2, text, 10)
193 * ctf_sequence: a dynamically-sized array.
194 * args: (type, field name, argument expression,
195 * type of length expression, length expression)
196 * The "type of length expression" needs to be an
197 * unsigned type. As a reminder, "unsigned char" should
198 * be preferred to "char", since the signedness of
199 * "char" is implementation-defined.
200 * Behavior is undefined if "text" argument is NULL.
202 ctf_sequence(char, seqfield1, text,
206 * ctf_sequence_text: a dynamically-sized array, printed
207 * as string. No need to be null-terminated.
208 * Behavior is undefined if "text" argument is NULL.
210 ctf_sequence_text(char, seqfield2, text,
214 * ctf_string: null-terminated string.
215 * args: (field name, argument expression)
216 * Behavior is undefined if "text" argument is NULL.
218 ctf_string(stringfield, text)
221 * ctf_float: floating-point number.
222 * args: (type, field name, argument expression)
224 ctf_float(float, floatfield, floatarg)
225 ctf_float(double, doublefield, doublearg)
229 There can be an arbitrary number of tracepoint providers within an
230 application, but they must each have their own provider name. Duplicate
231 provider names are not allowed.
235 .SH "ASSIGNING LOGLEVEL TO EVENTS"
239 Optionally, a loglevel can be assigned to a TRACEPOINT_EVENT using the
242 TRACEPOINT_LOGLEVEL(< [com_company_]project[_component] >,
243 < event >, < loglevel_name >)
245 The first field is the provider name, the second field is the name of
246 the tracepoint, and the third field is the loglevel name. A
247 TRACEPOINT_EVENT should be declared prior to the the TRACEPOINT_LOGLEVEL
248 for a given tracepoint name. The TRACEPOINT_PROVIDER must be already
249 declared before declaring a TRACEPOINT_LOGLEVEL.
251 The loglevels go from 0 to 14. Higher numbers imply the most verbosity
252 (higher event throughput expected.
254 Loglevels 0 through 6, and loglevel 14, match syslog(3) loglevels
255 semantic. Loglevels 7 through 13 offer more fine-grained selection of
262 action must be taken immediately
274 normal, but significant, condition
277 informational message
280 debug information with system-level scope (set of programs)
282 TRACE_DEBUG_PROGRAM 8
283 debug information with program-level scope (set of processes)
285 TRACE_DEBUG_PROCESS 9
286 debug information with process-level scope (set of modules)
288 TRACE_DEBUG_MODULE 10
289 debug information with module (executable/library) scope (set of
293 debug information with compilation unit scope (set of functions)
295 TRACE_DEBUG_FUNCTION 12
296 debug information with function-level scope
299 debug information with line-level scope (TRACEPOINT_EVENT default)
304 See lttng(1) for information on how to use LTTng-UST loglevels.
308 .SH "ADDING TRACEPOINTS TO YOUR CODE"
312 Include the provider header in each C files you plan to instrument,
313 following the building/linking directives in the next section.
315 For instance, add within a function:
317 tracepoint(ust_tests_hello, tptest, i, netint, values,
318 text, strlen(text), dbl, flt);
320 As a call to the tracepoint. It will only be activated when requested by
321 lttng(1) through lttng-sessiond(8).
323 Even though LTTng-UST supports tracepoint() call site duplicates having
324 the same provider and event name, it is recommended to use a
325 provider event name pair only once within the source code to help
326 map events back to their call sites when analyzing the trace.
328 Sometimes arguments to the probe are expensive to compute (e.g.
329 take call stack). To avoid the computation when the tracepoint is
330 disabled one can use more 'low level' tracepoint_enabled() and
331 do_tracepoint() macros as following:
333 if (tracepoint_enabled(ust_tests_hello, tptest)) {
334 /* prepare arguments */
335 do_tracepoint(ust_tests_hello, tptest, i, netint, values,
336 text, strlen(text), dbl, flt);
339 Here do_tracepoint() doesn't contain check if the tracepoint is enabled.
340 Using tracepoint() in such scenario is dangerous since it also contains
341 enabled check and thus race condition is possible in the following code
342 if the tracepoint has been enabled after check in tracepoint_enabled()
343 but before tracepoint():
345 if (tracepoint_enabled(provider, name)) { /* tracepoint is disabled */
348 /* tracepoint is enabled by 'lttng' tool */
349 tracepoint(provider, name, args); /* args wasn't prepared properly */
351 Note also that neither tracepoint_enabled() nor do_tracepoint() have
352 STAP_PROBEV() call so if you need it you should emit this call yourself.
356 .SH "BUILDING/LINKING THE TRACEPOINT PROVIDER"
359 There are 2 ways to compile the Tracepoint Provider with the
360 application: either statically or dynamically. Please follow
363 1) Compile the Tracepoint Provider with the application, either
364 directly or through a static library (.a):
365 - Into exactly one object of your application, define
366 "TRACEPOINT_DEFINE" and include the tracepoint provider.
367 - Use "\-I." for the compilation unit containing the tracepoint
368 provider include (e.g., tp.c).
369 - Link the application with "\-llttng-ust" and "\-ldl".
370 - Include the tracepoint provider header into all C files using
373 - doc/examples/easy-ust/ sample.c sample_component_provider.h tp.c
375 - doc/examples/hello-static-lib/ hello.c tp.c ust_test_hello.h Makefile
377 2) Compile the Tracepoint Provider separately from the application,
378 using dynamic linking:
379 - Into exactly one object of your application: define
380 "TRACEPOINT_DEFINE" _and_ also define
381 "TRACEPOINT_PROBE_DYNAMIC_LINKAGE", then include the tracepoint
383 - Include the tracepoint provider header into all instrumented C
384 files that use the provider.
385 - Compile the tracepoint provider with "\-I.".
386 - Link the tracepoint provider with "\-llttng-ust".
387 - Link application with "\-ldl".
388 - Set a LD_PRELOAD environment to preload the tracepoint provider
389 shared object before starting the application when tracing is
390 needed. Another way is to dlopen the tracepoint probe when needed
393 - doc/examples/demo demo.c tp*.c ust_tests_demo*.h demo-trace Makefile
395 - Note about dlclose() usage: it is not safe to use dlclose on a
396 provider shared object that is being actively used for tracing due
397 to a lack of reference counting from lttng-ust to the used shared
399 - Enable instrumentation and control tracing with the "lttng" command
400 from lttng-tools. See lttng-tools doc/quickstart.txt.
401 - Note for C++ support: although an application instrumented with
402 tracepoints can be compiled with g++, tracepoint probes should be
403 compiled with gcc (only tested with gcc so far).
407 .SH "USING LTTNG UST WITH DAEMONS"
410 Some extra care is needed when using liblttng-ust with daemon
411 applications that call fork(), clone(), or BSD rfork() without a
412 following exec() family system call. The library "liblttng-ust-fork.so"
413 needs to be preloaded for the application (launch with e.g.
414 LD_PRELOAD=liblttng-ust-fork.so appname).
421 Context information can be prepended by the tracer before each, or some,
422 events. The following context information is supported by LTTng-UST:
427 Virtual thread ID: thread ID as seen from the point of view of the
433 Virtual process ID: process ID as seen from the point of view of the
439 Instruction pointer: Enables recording of the exact location where a tracepoint
440 was emitted. Can be used to reverse-lookup the source location that caused the
446 Thread name, as set by exec() or prctl(). It is recommended that
447 programs set their thread name with prctl() before hitting the first
448 tracepoint for that thread.
453 Pthread identifier. Can be used on architectures where pthread_t maps
454 nicely to an unsigned long type.
457 .SH "BASE ADDRESS STATEDUMP"
460 If an application that uses liblttng-ust.so becomes part of a session,
461 information about its currently loaded shared objects will be traced to the
462 session at session-enable time. To record this information, the following event
465 .IP "ust_baddr_statedump:soinfo"
466 This event is used to trace a currently loaded shared object. The base address
467 (where the dynamic linker has placed the shared object) is recorded in the
468 "baddr" field. The path to the shared object gets recorded in the
469 "sopath" field (as string). The file size of the loaded object (in
470 bytes) is recorded to the "size" field and its time of last modification
471 (in seconds since Epoch) is recorded in the "mtime" field.
473 If the event above is enabled, a series of "ust_baddr_statedump:soinfo"
474 events is recorded at session-enable time. It represents the state of
475 currently loaded shared objects for the traced process. If this
476 information gets combined with the lttng-ust-dl(3) instrumentation, all
477 aspects of dynamic loading that are relevant for symbol and
478 line number lookup are traced by LTTng.
480 .SH "ENVIRONMENT VARIABLES"
483 .IP "LTTNG_UST_DEBUG"
484 Activate liblttng-ust debug and error output.
486 .IP "LTTNG_UST_REGISTER_TIMEOUT"
487 The environment variable "LTTNG_UST_REGISTER_TIMEOUT" can be used to
488 specify how long the applications should wait for sessiond
489 "registration done" command before proceeding to execute the main
490 program. The default is 3000ms (3 seconds). The timeout value is
491 specified in milliseconds. The value 0 means "don't wait". The value
492 \-1 means "wait forever". Setting this environment variable to 0 is
493 recommended for applications with time constraints on the process
496 .IP "LTTNG_UST_WITHOUT_BADDR_STATEDUMP"
497 Prevent liblttng-ust to perform a base-address statedump on session-enable.
503 lttng-gen-tp(1), lttng(1), babeltrace(1), lttng-ust-cyg-profile(3),
504 lttng-ust-dl(3), lttng-sessiond(8)
510 Older lttng-ust libraries reject more recent, and incompatible, probe
511 providers. Newer lttng-ust libraries accept older probe providers, even
512 though some newer features might not be available with those providers.
518 LTTng-UST 2.0 and 2.1 lttng-ust libraries do not check for probe
519 provider version compatibility. This can lead to out-of-bound accesses
520 when using a more recent probe provider with an older lttng-ust library.
521 These error only trigger when tracing is active. This issue has been
522 fixed in LTTng-UST 2.2.
524 If you encounter any issues or usability problem, please report it on
525 our mailing list <lttng-dev@lists.lttng.org> to help improve this
529 liblttng-ust is distributed under the GNU Lesser General Public License
530 version 2.1. The headers are distributed under the MIT license.
532 See http://lttng.org for more information on the LTTng project.
534 Mailing list for support and development: <lttng-dev@lists.lttng.org>.
536 You can find us on IRC server irc.oftc.net (OFTC) in #lttng.
540 Thanks to Ericsson for funding this work, providing real-life use-cases,
543 Special thanks to Michel Dagenais and the DORSAL laboratory at
544 Polytechnique de Montreal for the LTTng journey.
549 liblttng-ust was originally written by Mathieu Desnoyers, with additional
550 contributions from various other people. It is currently maintained by
551 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>.