Update version to 0.16
[ust.git] / doc / info / ust.texi
1 \input texinfo @c -*-texinfo-*-
2 @c %**start of header
3 @setfilename ust.info
4 @settitle LTTng Userspace Tracer (UST) Manual
5 @c %**end of header
6
7 @copying
8 This manual is for program, version version.
9
10 Copyright @copyright{} copyright-owner.
11
12 @quotation
13 Permission is granted to ...
14 @end quotation
15 @end copying
16
17 @titlepage
18 @title LTTng Userspace Tracer (UST) Manual
19 @c @subtitle subtitle-if-any
20 @c @subtitle second-subtitle
21 @c @author author
22
23 @c The following two commands
24 @c start the copyright page.
25 @c @page
26 @c @vskip 0pt plus 1filll
27 @c @insertcopying
28
29 @c Published by ...
30 @end titlepage
31
32 @c So the toc is printed at the start.
33 @contents
34
35 @ifnottex
36 @node Top
37 @top LTTng Userspace Tracer
38
39 This manual is for UST 0.14.
40 @end ifnottex
41
42 @menu
43 * Overview::
44 * Installation::
45 * Quick start::
46 * Instrumenting an application::
47 * Recording a trace::
48 * Viewing traces::
49 * Performance::
50 * Resource Usage::
51 * List of environment variables detected by libust::
52 * GDB integration::
53 @c * Copying:: Your rights and freedoms.
54 @end menu
55
56 @node Overview
57 @chapter Overview
58
59 @menu
60 * What is UST?::
61 * License::
62 * Supported platforms::
63 @end menu
64
65 @node What is UST?
66 @section What is UST?
67
68 The LTTng Userspace Tracer (UST) is a library accompanied by a set of tools to
69 trace userspace code.
70
71 Code may be instrumented with either markers or tracepoints. A highly efficient
72 lockless tracer records these events to a trace buffers. These buffers are reaped
73 by a deamon which writes trace data to disk.
74
75 High performance is achieved by the use of lockless buffering algorithms, RCU and
76 per-cpu buffers. In addition, special care is taken to minize cache impact.
77
78 @node License
79 @section License
80 The LTTng Userspace Tracer is intended to be linkable to open source software
81 as well as to proprietary applications. This was accomplished by licensing
82 the code that needs to be linked to the traced program as @acronym{LGPL}.
83
84 Components licensed as LGPL v2.1:
85 @itemize @bullet
86 @item libust
87 @item libinterfork
88 @item libustcomm
89 @end itemize
90
91 Components licensed as GPL v2:
92 @itemize @bullet
93 @item ustctl
94 @item libustctl
95 @item ust-consumerd
96 @end itemize
97
98 @node Supported platforms
99 @section Supported platforms
100
101 UST can currently trace applications running on Linux, on the x86-32, x86-64
102 and PowerPC 32 architectures.
103
104 @node Installation
105 @chapter Installation
106
107 The LTTng userspace tracer is a library and a set of userspace tools.
108
109 The following packages are required:
110
111 @itemize @bullet
112 @item
113 ust
114
115 This contains the tracing library, the ust-consumerd daemon, trace control tools
116 and other helper tools.
117
118 Repository: @url{http://lttng.org/ust}
119
120 @item
121 liburcu
122
123 This is the userspace read-copy update library by Mathieu Desnoyers.
124
125 Available in Debian as package liburcu-dev.
126
127 Home page: @url{http://lttng.org/urcu}
128
129 @item
130 LTTV
131
132 LTTV is a graphical (and text) viewer for LTTng traces.
133
134 Home page: @url{http://lttng.org}
135
136 @end itemize
137
138 Liburcu should be installed first. UST may then be compiled and installed. LTTV
139 has no dependency on the other packages; it may therefore be installed on a
140 system which does not have UST installed.
141
142 Refer to the README in each of these packages for installation instructions.
143
144 @c @menu
145 @c @end menu
146
147 @node Quick start
148 @chapter Quick start
149
150 First, instrument a program with a marker.
151
152 @example
153 @verbatim
154
155 #include <ust/marker.h>
156
157 int main(int argc, char **argv)
158 {
159 int v;
160 char *st;
161
162 /* ... set values of v and st ... */
163
164 /* a marker: */
165 ust_marker(myevent, "firstarg %d secondarg %s", v, st);
166
167 /* a marker without arguments: */
168 ust_marker(myotherevent, MARK_NOARGS);
169
170 return 0;
171 }
172
173 @end verbatim
174 @end example
175
176 Then compile it in the regular way, linking it with libust. For example:
177
178 @example
179 gcc -o foo -lust foo.c
180 @end example
181
182 Run the program with @command{usttrace}. The @command{usttrace} output says where the trace
183 was written.
184
185 @example
186 usttrace ./foo
187 @end example
188
189 Finally, open the trace in LTTV.
190
191 @example
192 lttv-gui -t /path/to/trace
193 @end example
194
195 The trace can also be dumped as text in the console:
196
197 @example
198 lttv -m textDump -t /path/to/trace
199 @end example
200
201 @node Instrumenting an application
202 @chapter Instrumenting an application
203
204 In order to record a trace of events occurring in a application, the
205 application must be instrumented. Instrumentation points resemble function
206 calls. When the program reaches an instrumentation point, an event is
207 generated.
208
209 There are no limitations on the type of code that may be instrumented.
210 Multi-threaded programs may be instrumented without problem. Signal handlers
211 may be instrumented as well.
212
213 There are two APIs to instrument programs: markers and tracepoints. Markers are
214 quick to add and are usually used for temporary instrumentation. Tracepoints
215 provide a way to instrument code more cleanly and are suited for permanent
216 instrumentation.
217
218 In addition to executable programs, shared libraries may also be instrumented
219 with the methods described in this chapter.
220
221 @menu
222 * Markers::
223 * Tracepoints::
224 @end menu
225
226 @node Markers
227 @section Markers
228
229 Adding a marker is simply a matter of inserting one line in the program.
230
231 @example
232 @verbatim
233 #include <ust/marker.h>
234
235 int main(int argc, char **argv)
236 {
237 int v;
238 char *st;
239
240 /* ... set values of v and st ... */
241
242 /* a marker: */
243 ust_marker(myevent, "firstarg %d secondarg %s", v, st);
244
245 /* another marker without arguments: */
246 ust_marker(myotherevent, MARK_NOARGS);
247
248 return 0;
249 }
250 @end verbatim
251 @end example
252
253 The invocation of the ust_marker() macro requires at least 2 arguments. The
254 first, "myevent", is the name of the event. The second is a format string
255 that announces the names and the types of the event arguments. Its
256 format resembles that of a printf() format string; it is described
257 thoroughly in Appendix x.
258
259 A given Marker may appear more than once in the same program. Other Markers may
260 have the same name and a different format string, although this might induce
261 some confusion at analysis time.
262
263 @node Tracepoints
264 @section Tracepoints
265
266 The Tracepoint API is superseded by TRACEPOINT_EVENT() (which uses the
267 Tracepoints internally). At this stage, TRACEPOINT_EVENT() is work in
268 progress. While we complete this API, please use the ust_marker() API
269 provided by ust/marker.h.
270
271 @node Recording a trace
272 @chapter Recording a trace
273
274 @menu
275 * Using @command{usttrace}::
276 * Setting up the recording manually::
277 * Using early tracing::
278 * Crash recovery::
279 * Tracing across @code{fork()} and @code{clone()}::
280 * Tracing programs and libraries that were not linked to libust::
281 @end menu
282
283 @node Using @command{usttrace}
284 @section Using @command{usttrace}
285
286 The simplest way to record a trace is to use the @command{usttrace} script. An
287 example is given in the quickstart above.
288
289 The @command{usttrace} script automatically:
290 @itemize @bullet
291 @item creates a daemon
292 @item enables all markers
293 @item runs the command specified on the command line
294 @item after the command ends, prints the location where the trace was saved
295 @end itemize
296
297 Each subdirectory of the save location contains the trace of one process that
298 was generated by the command. The name of a subdirectory consists in the the PID
299 of the process, followed by the timestamp of its creation.
300
301 The save location also contains logs of the tracing.
302
303 When using @command{usttrace}, the early tracing is always active, which means
304 that the tracing is guaranteed to be started by the time the process enters its
305 @code{main()} function.
306
307 Several @command{usttrace}'s may be run simultaneously without risk of
308 conflict. This facilitates the use of the tracer by idependent users on a
309 system. Each instance of @command{usttrace} starts its own daemon which
310 collects the events of the processes it creates.
311
312 @node Setting up the recording manually
313 @section Setting up the recording manually
314
315 Instead of using @command{usttrace}, a trace may be recorded on an already
316 running process.
317
318 First the daemon must be started.
319
320 @example
321 @verbatim
322 # Make sure the directory for the communication sockets exists.
323 $ mkdir /tmp/ustsocks
324
325 # Make sure the directory where ust-consumerd will write the trace exists.
326 $ mkdir /tmp/trace
327
328 # Start the daemon
329 $ ust-consumerd
330
331 # We assume the program we want to trace is already running and that
332 # it has pid 1234.
333
334 # List the available markers
335 $ ustctl list-markers 1234
336 # A column indicates 0 for an inactive marker and 1 for an active marker.
337
338 # Enable a marker
339 $ ustctl enable-marker 1234 auto ust/mymark
340
341 # Create a trace
342 $ ustctl create-trace 1234 auto
343
344 # Start tracing
345 $ ustctl start-trace 1234 auto
346
347 # Do things...
348
349 # Stop tracing
350 $ ustctl stop-trace 1234 auto
351
352 # Destroy the trace
353 $ ustctl destroy-trace 1234 auto
354 @end verbatim
355 @end example
356
357 For more information about the manual mode, see the ustctl(1) man page.
358
359 @node Using early tracing
360 @section Using early tracing
361
362 Early tracing consists in starting the tracing as early as possible in the
363 program, so no events are lost between program start and the point where the
364 command to start the tracing is given. When using early tracing, it is
365 guaranteed that by the time the traced program enters its @code{main()}
366 function, the tracing will be started.
367
368 When using @command{usttrace}, the early tracing is always active.
369
370 When using the manual mode (@command{ustctl}), early tracing is enabled using
371 environment variables. Setting @env{UST_TRACE} to @code{1}, enables early
372 tracing, while setting @env{UST_AUTOPROBE} to @code{1} enables all markers
373 automatically.
374
375
376 @node Crash recovery
377 @section Crash recovery
378
379 When a process being traced crashes, the daemon is able to recover all the
380 events in its buffers that were successfully commited. This is possible because
381 the buffers are in a shared memory segment which remains available to the
382 daemon even after the termination of the traced process.
383
384 @node Tracing across @code{fork()} and @code{clone()}
385 @section Tracing across @code{fork()} and @code{clone()}
386
387 Tracing across @code{clone()} when the @code{CLONE_VM} flag is specified is
388 supported without any particular action.
389
390 When @code{clone()} is called without @code{CLONE_VM} or @code{fork()} is
391 called, a new address space is created and the tracer must be notified to
392 create new buffers for it.
393
394 This can be done automatically, by @env{LD_PRELOAD}'ing @file{libinterfork.so}.
395 This library intercepts calls to @code{fork()} and informs the tracer it is
396 being called. When using @command{usttrace}, this is accomplied by specifying
397 the @option{-f} command line argument.
398
399 Alternatively, the program can call @code{ust_before_fork()} before calling
400 @code{fork()} or @code{clone()} with @code{CLONE_VM}. After the call,
401 @code{ust_after_fork_parent()} must be called in the parent process and
402 @code{ust_after_fork_child()} must be called in the child process.
403
404
405 @node Tracing programs and libraries that were not linked to libust
406 @section Tracing programs and libraries that were not linked to libust
407
408 Some programs need to be traced even though they were not linked to libust
409 either because they were not instrumented or because it was not practical.
410
411 An executable that is not instrumented can still yield interesting traces when
412 at least one of its dynamic libraries is instrumented. It is also possible to
413 trace certain function calls by intercepting them with a specially crafted
414 library that is linked with @env{LD_PRELOAD} at program start.
415
416 In any case, a program that was not linked to libust at compile time must be
417 linked to it at run time with @env{LD_PRELOAD}. This can be accomplished with
418 @command{usttrace}'s @option{-l} option. It can also be done by setting the
419 @env{LD_PRELOAD} environment variable on the command line. For example:
420
421 @example
422 @verbatim
423 # Run ls with usttrace, LD_PRELOAD'ing libust
424 # (assuming one of the libraries used by ls is instrumented).
425 $ usttrace -l ls
426
427 # Run ls, manually adding the LD_PRELOAD.
428 $ LD_PRELOAD=/usr/local/lib/libust.so.0 ls
429 @end verbatim
430 @end example
431
432
433 @node Performance
434 @chapter Performance
435
436 Todo.
437
438 @node Viewing traces
439 @chapter Viewing traces
440
441 Traces may be viewed with LTTV. An example of command for launching LTTV is
442 given in the quickstart.
443
444 @menu
445 * Viewing multiple traces::
446 * Combined kernel-userspace tracing::
447 @end menu
448
449 @node Viewing multiple traces
450 @section Viewing multiple traces
451
452 When tracing multi-process applications or several applications simultaneously,
453 more than one trace will be obtained. LTTV can open and display all these
454 traces simultaneously.
455
456 @node Combined kernel-userspace tracing
457 @section Combined kernel-userspace tracing
458
459 In addition to multiple userspace traces, LTTV can open a kernel trace recorded
460 with the LTTng kernel tracer. This provides events that enable the rendering of
461 the Control Flow View and the Resource View.
462
463 When doing so, it is necessary to use the same time source for the kernel
464 tracer as well as the userspace tracer. Currently, the recommended method is to
465 use the timestamp counter for both. The TSC can however only be used on architectures
466 where it is synchronized across cores.
467
468 @node Resource Usage
469 @chapter Resource Usage
470
471 The purpose of this section is to give an overview of the resource usage of libust. For
472 a developer, knowing this can be important: because libust is linked with applications, it
473 needs to share some resources with it. Some applications may make some assumptions that are in
474 conflict with libust's usage of resources.
475
476 In practice however, libust is designed to be transparent and is compatible
477 with the vast majority of applications. This means no changes are required in
478 the application (or library) being linked to libust.
479
480 Libust is initialized by a constructor, which by definition runs before the
481 @code{main()} function of the application starts. This constructor creates a
482 thread called the @emph{listener thread}. The listener thread initializes a
483 named socket and waits for connections for ust-consumerd or ustctl.
484
485 Libust-specific code may:
486 @itemize @bullet
487 @item use @code{malloc()} and @code{free()}
488 @item map shared memory segment in the process adress space
489 @item intercept some library calls, specifically @code{fork()} and @code{clone()}
490 @item do interprocess communication with the daemon or ustctl
491 @item create and open named sockets
492
493 @end itemize
494
495 It will not:
496 @itemize @bullet
497 @item handle any signal (all signals are blocked in the listener thread)
498 @item change any process-wide setting that could confuse the application
499 @end itemize
500
501 @node List of environment variables detected by libust
502 @appendix List of environment variables detected by libust
503
504 The behavior of tracing can be influenced by setting special environment
505 variables in the environment of the traced application. This section
506 describes these variables.
507
508 @itemize @bullet
509
510 @item
511 @env{UST_TRACE}
512
513 If set to 1, start tracing as soon as the program starts. Tracing is
514 guaranteed to be started by the time the @code{main()} function starts.
515
516 @item
517 @env{UST_AUTOPROBE}
518
519 If set to @code{1}, enable all markers by the time the @code{main()} function starts.
520
521 @item
522 @env{UST_AUTOCOLLECT}
523
524 If set to @code{0}, disable notification of daemon on trace start. Useful for
525 performance tests.
526
527 @item
528 @env{UST_OVERWRITE}
529
530 If set to @code{1}, enable overwriting of buffers on overrun.
531
532 @item
533 @env{UST_SUBBUF_NUM}
534
535 If set, defines the default number of subbuffers per buffer.
536
537 @item
538 @env{UST_SUBBUF_SIZE}
539
540 If set, defines the default size of subbuffers, in bytes.
541
542 @end itemize
543
544 @node GDB integration
545 @appendix GDB integration
546
547 GDB, the GNU Debugger, can use UST markers as GDB tracepoints (note GDB has its
548 own concept of tracepoint). This feature is called GDB Static Tracepoints. When
549 a GDB tracepoint is hit, GDB collects the marker arguments, as well as the
550 state of the registers. Support for GDB is currently work in progress.
551
552 @bye
This page took 0.04048 seconds and 5 git commands to generate.