1<?xml version="1.0"?> <!-- -*- sgml -*- -->
2<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN"
3  "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd"
4[ <!ENTITY % vg-entities SYSTEM "vg-entities.xml"> %vg-entities; ]>
5
6
7<book id="FAQ" xreflabel="Valgrind FAQ">
8
9<bookinfo>
10  <title>Valgrind FAQ</title>
11  <releaseinfo>&rel-type; &rel-version; &rel-date;</releaseinfo>
12  <copyright>
13    <year>&vg-lifespan;</year>
14    <holder><ulink url="&vg-devs-url;">Valgrind Developers</ulink></holder>
15  </copyright>
16  <legalnotice>
17    <para>Email: <ulink url="mailto:&vg-vemail;">&vg-vemail;</ulink></para>
18  </legalnotice>
19</bookinfo>
20
21
22<article id="faq">
23<title>Valgrind Frequently Asked Questions</title>
24
25
26<!-- FAQ starts here -->
27<qandaset>
28
29
30<!-- Background -->
31<qandadiv id="faq.background" xreflabel="Background">
32<title>Background</title>
33
34<qandaentry id="faq.pronounce">
35 <question id="q-pronounce">
36  <para>How do you pronounce "Valgrind"?</para>
37 </question>
38 <answer id="a-pronounce">
39  <para>The "Val" as in the word "value".  The "grind" is pronounced
40  with a short 'i' -- ie. "grinned" (rhymes with "tinned") rather than
41  "grined" (rhymes with "find").</para> <para>Don't feel bad: almost
42  everyone gets it wrong at first.</para>
43 </answer>
44</qandaentry>
45
46<qandaentry id="faq.whence">
47 <question id="q-whence">
48  <para>Where does the name "Valgrind" come from?</para>
49 </question>
50 <answer id="a-whence">
51
52  <para>From Nordic mythology.  Originally (before release) the project
53  was named Heimdall, after the watchman of the Nordic gods.  He could
54  "see a hundred miles by day or night, hear the grass growing, see the
55  wool growing on a sheep's back", etc.  This would have been a great
56  name, but it was already taken by a security package "Heimdal".</para>
57
58  <para>Keeping with the Nordic theme, Valgrind was chosen.  Valgrind is
59  the name of the main entrance to Valhalla (the Hall of the Chosen
60  Slain in Asgard).  Over this entrance there resides a wolf and over it
61  there is the head of a boar and on it perches a huge eagle, whose eyes
62  can see to the far regions of the nine worlds.  Only those judged
63  worthy by the guardians are allowed to pass through Valgrind.  All
64  others are refused entrance.</para>
65
66  <para>It's not short for "value grinder", although that's not a bad
67  guess.</para>
68  </answer>
69 </qandaentry>
70
71</qandadiv>
72
73
74
75<!-- Compiling, Installing and Configuring -->
76<qandadiv id="faq.installing" xreflabel="Compiling, installing and configuring">
77<title>Compiling, installing and configuring</title>
78
79<qandaentry id="faq.make_dies">
80 <question id="q-make_dies">
81  <para>When building Valgrind, 'make' dies partway with
82  an assertion failure, something like this:</para>
83<screen>
84% make: expand.c:489: allocated_variable_append:
85        Assertion 'current_variable_set_list->next != 0' failed.
86</screen>
87 </question>
88 <answer id="a-make_dies">
89  <para>It's probably a bug in 'make'.  Some, but not all, instances of
90  version 3.79.1 have this bug, see
91  <ulink url="http://www.mail-archive.com/bug-make@gnu.org/msg01658.html">this</ulink>.
92  Try upgrading to a more recent version of 'make'.  Alternatively, we have
93  heard that unsetting the CFLAGS environment variable avoids the
94  problem.</para>
95 </answer>
96</qandaentry>
97
98<qandaentry id="faq.glibc_devel">
99 <question>
100  <para>When building Valgrind, 'make' fails with this:</para>
101<screen>
102/usr/bin/ld: cannot find -lc
103collect2: ld returned 1 exit status
104</screen>
105 </question>
106 <answer>
107  <para>You need to install the glibc-static-devel package.</para>
108 </answer>
109</qandaentry>
110
111</qandadiv>
112
113
114<!-- Valgrind aborts unexpectedly -->
115<qandadiv id="faq.abort" xreflabel="Valgrind aborts unexpectedly">
116<title>Valgrind aborts unexpectedly</title>
117
118<qandaentry id="faq.exit_errors">
119  <question id="q-exit_errors">
120    <para>Programs run OK on Valgrind, but at exit produce a bunch of
121    errors involving <literal>__libc_freeres</literal> and then die
122    with a segmentation fault.</para>
123  </question>
124  <answer id="a-exit_errors">
125    <para>When the program exits, Valgrind runs the procedure
126    <function>__libc_freeres</function> in glibc.  This is a hook for
127    memory debuggers, so they can ask glibc to free up any memory it has
128    used.  Doing that is needed to ensure that Valgrind doesn't
129    incorrectly report space leaks in glibc.</para>
130
131    <para>The problem is that running <literal>__libc_freeres</literal> in
132    older glibc versions causes this crash.</para>
133
134    <para>Workaround for 1.1.X and later versions of Valgrind: use the
135    <option>--run-libc-freeres=no</option> option.  You may then get space
136    leak reports for glibc allocations (please don't report these to
137    the glibc people, since they are not real leaks), but at least the
138    program runs.</para>
139  </answer>
140</qandaentry>
141
142<qandaentry id="faq.bugdeath">
143  <question id="q-bugdeath">
144    <para>My (buggy) program dies like this:</para>
145<screen>valgrind: m_mallocfree.c:248 (get_bszB_as_is): Assertion 'bszB_lo == bszB_hi' failed.</screen>
146    <para>or like this:</para>
147<screen>valgrind: m_mallocfree.c:442 (mk_inuse_bszB): Assertion 'bszB != 0' failed.</screen>
148    <para>or otherwise aborts or crashes in m_mallocfree.c.</para>
149
150  </question>
151  <answer id="a-bugdeath">
152    <para>If Memcheck (the memory checker) shows any invalid reads,
153    invalid writes or invalid frees in your program, the above may
154    happen.  Reason is that your program may trash Valgrind's low-level
155    memory manager, which then dies with the above assertion, or
156    something similar.  The cure is to fix your program so that it
157    doesn't do any illegal memory accesses.  The above failure will
158    hopefully go away after that.</para>
159  </answer>
160</qandaentry>
161
162<qandaentry id="faq.msgdeath">
163  <question id="q-msgdeath">
164    <para>My program dies, printing a message like this along the
165    way:</para>
166<screen>vex x86->IR: unhandled instruction bytes: 0x66 0xF 0x2E 0x5</screen>
167  </question>
168  <answer id="a-msgdeath">
169    <para>One possibility is that your program has a bug and erroneously
170    jumps to a non-code address, in which case you'll get a SIGILL signal.
171    Memcheck may issue a warning just before this happens, but it might not
172    if the jump happens to land in addressable memory.</para>
173
174    <para>Another possibility is that Valgrind does not handle the
175    instruction.  If you are using an older Valgrind, a newer version might
176    handle the instruction.  However, all instruction sets have some
177    obscure, rarely used instructions.  Also, on amd64 there are an almost
178    limitless number of combinations of redundant instruction prefixes, many
179    of them undocumented but accepted by CPUs.  So Valgrind will still have
180    decoding failures from time to time.  If this happens, please file a bug
181    report.</para>
182  </answer>
183</qandaentry>
184
185<qandaentry id="faq.java">
186  <question id="q-java">
187    <para>I tried running a Java program (or another program that uses a
188    just-in-time compiler) under Valgrind but something went wrong.
189    Does Valgrind handle such programs?</para>
190  </question>
191  <answer id="a-java">
192    <para>Valgrind can handle dynamically generated code, so long as
193    none of the generated code is later overwritten by other generated
194    code.  If this happens, though, things will go wrong as Valgrind
195    will continue running its translations of the old code (this is true
196    on x86 and amd64, on PowerPC there are explicit cache flush
197    instructions which Valgrind detects and honours).
198    You should try running with
199    <option>--smc-check=all</option> in this case.  Valgrind will run
200    much more slowly, but should detect the use of the out-of-date
201    code.</para>
202
203    <para>Alternatively, if you have the source code to the JIT compiler
204    you can insert calls to the
205    <computeroutput>VALGRIND_DISCARD_TRANSLATIONS</computeroutput>
206    client request to mark out-of-date code, saving you from using
207    <option>--smc-check=all</option>.</para>
208
209    <para>Apart from this, in theory Valgrind can run any Java program
210    just fine, even those that use JNI and are partially implemented in
211    other languages like C and C++.  In practice, Java implementations
212    tend to do nasty things that most programs do not, and Valgrind
213    sometimes falls over these corner cases.</para>
214
215    <para>If your Java programs do not run under Valgrind, even with
216    <option>--smc-check=all</option>, please file a bug report and
217    hopefully we'll be able to fix the problem.</para>
218  </answer>
219</qandaentry>
220
221</qandadiv>
222
223
224<!-- Valgrind behaves unexpectedly -->
225<qandadiv id="faq.unexpected" xreflabel="Valgrind behaves unexpectedly">
226<title>Valgrind behaves unexpectedly</title>
227
228<qandaentry id="faq.reports">
229  <question id="q-reports">
230    <para>My program uses the C++ STL and string classes.  Valgrind
231    reports 'still reachable' memory leaks involving these classes at
232    the exit of the program, but there should be none.</para>
233  </question>
234  <answer id="a-reports">
235    <para>First of all: relax, it's probably not a bug, but a feature.
236    Many implementations of the C++ standard libraries use their own
237    memory pool allocators.  Memory for quite a number of destructed
238    objects is not immediately freed and given back to the OS, but kept
239    in the pool(s) for later re-use.  The fact that the pools are not
240    freed at the exit of the program cause Valgrind to report this
241    memory as still reachable.  The behaviour not to free pools at the
242    exit could be called a bug of the library though.</para>
243
244    <para>Using GCC, you can force the STL to use malloc and to free
245    memory as soon as possible by globally disabling memory caching.
246    Beware!  Doing so will probably slow down your program, sometimes
247    drastically.</para>
248    <itemizedlist>
249      <listitem>
250        <para>With GCC 2.91, 2.95, 3.0 and 3.1, compile all source using
251        the STL with <literal>-D__USE_MALLOC</literal>. Beware!  This was
252        removed from GCC starting with version 3.3.</para>
253      </listitem>
254      <listitem>
255        <para>With GCC 3.2.2 and later, you should export the
256        environment variable <literal>GLIBCPP_FORCE_NEW</literal> before
257        running your program.</para>
258      </listitem>
259      <listitem>
260        <para>With GCC 3.4 and later, that variable has changed name to
261        <literal>GLIBCXX_FORCE_NEW</literal>.</para>
262      </listitem>
263    </itemizedlist>
264
265    <para>There are other ways to disable memory pooling: using the
266    <literal>malloc_alloc</literal> template with your objects (not
267    portable, but should work for GCC) or even writing your own memory
268    allocators. But all this goes beyond the scope of this FAQ.  Start
269    by reading
270    <ulink
271    url="http://gcc.gnu.org/onlinedocs/libstdc++/faq/index.html#4_4_leak">
272         http://gcc.gnu.org/onlinedocs/libstdc++/faq/index.html#4_4_leak</ulink>
273    if you absolutely want to do that. But beware:
274    allocators belong to the more messy parts of the STL and
275    people went to great lengths to make the STL portable across
276    platforms. Chances are good that your solution will work on your
277    platform, but not on others.</para>
278 </answer>
279</qandaentry>
280
281
282<qandaentry id="faq.unhelpful">
283  <question id="q-unhelpful">
284    <para>The stack traces given by Memcheck (or another tool) aren't
285    helpful.  How can I improve them?</para>
286  </question>
287  <answer id="a-unhelpful">
288    <para>If they're not long enough, use <option>--num-callers</option>
289    to make them longer.</para>
290
291    <para>If they're not detailed enough, make sure you are compiling
292    with <option>-g</option> to add debug information.  And don't strip
293    symbol tables (programs should be unstripped unless you run 'strip'
294    on them; some libraries ship stripped).</para>
295
296    <para>Also, for leak reports involving shared objects, if the shared
297    object is unloaded before the program terminates, Valgrind will
298    discard the debug information and the error message will be full of
299    <literal>???</literal> entries.  The workaround here is to avoid
300    calling <function>dlclose</function> on these shared objects.</para>
301
302    <para>Also, <option>-fomit-frame-pointer</option> and
303    <option>-fstack-check</option> can make stack traces worse.</para>
304
305  <para>Some example sub-traces:</para>
306
307   <itemizedlist>
308     <listitem>
309       <para>With debug information and unstripped (best):</para>
310<programlisting>
311Invalid write of size 1
312   at 0x80483BF: really (malloc1.c:20)
313   by 0x8048370: main (malloc1.c:9)
314</programlisting>
315     </listitem>
316
317     <listitem>
318       <para>With no debug information, unstripped:</para>
319<programlisting>
320Invalid write of size 1
321   at 0x80483BF: really (in /auto/homes/njn25/grind/head5/a.out)
322   by 0x8048370: main (in /auto/homes/njn25/grind/head5/a.out)
323</programlisting>
324     </listitem>
325
326     <listitem>
327       <para>With no debug information, stripped:</para>
328<programlisting>
329Invalid write of size 1
330   at 0x80483BF: (within /auto/homes/njn25/grind/head5/a.out)
331   by 0x8048370: (within /auto/homes/njn25/grind/head5/a.out)
332   by 0x42015703: __libc_start_main (in /lib/tls/libc-2.3.2.so)
333   by 0x80482CC: (within /auto/homes/njn25/grind/head5/a.out)
334</programlisting>
335     </listitem>
336
337     <listitem>
338       <para>With debug information and -fomit-frame-pointer:</para>
339<programlisting>
340Invalid write of size 1
341   at 0x80483C4: really (malloc1.c:20)
342   by 0x42015703: __libc_start_main (in /lib/tls/libc-2.3.2.so)
343   by 0x80482CC: ??? (start.S:81)
344</programlisting>
345     </listitem>
346
347     <listitem>
348      <para>A leak error message involving an unloaded shared object:</para>
349<programlisting>
35084 bytes in 1 blocks are possibly lost in loss record 488 of 713
351   at 0x1B9036DA: operator new(unsigned) (vg_replace_malloc.c:132)
352   by 0x1DB63EEB: ???
353   by 0x1DB4B800: ???
354   by 0x1D65E007: ???
355   by 0x8049EE6: main (main.cpp:24)
356</programlisting>
357     </listitem>
358   </itemizedlist>
359
360 </answer>
361</qandaentry>
362
363<qandaentry id="faq.aliases">
364  <question id="q-aliases">
365    <para>The stack traces given by Memcheck (or another tool) seem to
366    have the wrong function name in them.  What's happening?</para>
367  </question>
368  <answer id="a-aliases">
369    <para>Occasionally Valgrind stack traces get the wrong function
370    names.  This is caused by glibc using aliases to effectively give
371    one function two names.  Most of the time Valgrind chooses a
372    suitable name, but very occasionally it gets it wrong.  Examples we know
373    of are printing <function>bcmp</function> instead of
374    <function>memcmp</function>, <function>index</function> instead of
375    <function>strchr</function>, and <function>rindex</function> instead of
376    <function>strrchr</function>.</para>
377  </answer>
378</qandaentry>
379
380
381<qandaentry id="faq.crashes">
382  <question id="q-crashes">
383    <para>My program crashes normally, but doesn't under Valgrind, or vice
384    versa.  What's happening?</para>
385  </question>
386  <answer id="a-crashes">
387    <para>When a program runs under Valgrind, its environment is slightly
388    different to when it runs natively.  For example, the memory layout is
389    different, and the way that threads are scheduled is different.</para>
390
391    <para>Most of the time this doesn't make any difference, but it can,
392    particularly if your program is buggy.  For example, if your program
393    crashes because it erroneously accesses memory that is unaddressable,
394    it's possible that this memory will not be unaddressable when run under
395    Valgrind.  Alternatively, if your program has data races, these may not
396    manifest under Valgrind.</para>
397
398    <para>There isn't anything you can do to change this, it's just the
399    nature of the way Valgrind works that it cannot exactly replicate a
400    native execution environment.  In the case where your program crashes
401    due to a memory error when run natively but not when run under Valgrind,
402    in most cases Memcheck should identify the bad memory operation.</para>.
403  </answer>
404</qandaentry>
405
406
407
408<qandaentry id="faq.hiddenbug">
409  <question id="q-hiddenbug">
410    <para> Memcheck doesn't report any errors and I know my program has
411    errors.</para>
412  </question>
413  <answer id="a-hiddenbug">
414    <para>There are two possible causes of this.</para>
415
416    <para>First, by default, Valgrind only traces the top-level process.
417    So if your program spawns children, they won't be traced by Valgrind
418    by default.  Also, if your program is started by a shell script,
419    Perl script, or something similar, Valgrind will trace the shell, or
420    the Perl interpreter, or equivalent.</para>
421
422    <para>To trace child processes, use the
423    <option>--trace-children=yes</option> option.</para>
424
425    <para>If you are tracing large trees of processes, it can be less
426    disruptive to have the output sent over the network.  Give Valgrind
427    the option <option>--log-socket=127.0.0.1:12345</option> (if you want
428    logging output sent to port <literal>12345</literal> on
429    <literal>localhost</literal>).  You can use the valgrind-listener
430    program to listen on that port:</para>
431<programlisting>
432valgrind-listener 12345
433</programlisting>
434
435    <para>Obviously you have to start the listener process first.  See
436    the manual for more details.</para>
437
438    <para>Second, if your program is statically linked, most Valgrind
439    tools will only work well if they are able to replace certain
440    functions, such as <function>malloc</function>, with their own
441    versions.  By default, statically linked <function>malloc
442    functions</function> are not replaced. A key indicator of this is
443    if Memcheck says:
444<programlisting>
445All heap blocks were freed -- no leaks are possible
446</programlisting>
447    when you know your program calls <function>malloc</function>.  The
448    workaround is to use the option
449    <option>--soname-synonyms=somalloc=NONE</option>
450    or to avoid statically linking your program.</para>
451
452    <para>There will also be no replacement if you use an alternative
453    <function>malloc library</function> such as tcmalloc, jemalloc,
454    ...  In such a case, the
455    option <option>--soname-synonyms=somalloc=zzzz</option> (where
456    zzzz is the soname of the alternative malloc library) will allow
457    Valgrind to replace the functions.</para>
458  </answer>
459</qandaentry>
460
461
462<qandaentry id="faq.overruns">
463  <question id="q-overruns">
464    <para>Why doesn't Memcheck find the array overruns in this
465    program?</para>
466<programlisting>
467int static[5];
468
469int main(void)
470{
471  int stack[5];
472
473  static[5] = 0;
474  stack [5] = 0;
475
476  return 0;
477}
478</programlisting>
479  </question>
480  <answer id="a-overruns">
481    <para>Unfortunately, Memcheck doesn't do bounds checking on global
482    or stack arrays.  We'd like to, but it's just not possible to do in
483    a reasonable way that fits with how Memcheck works.  Sorry.</para>
484
485    <para>However, the experimental tool SGcheck can detect errors like
486    this.  Run Valgrind with the <option>--tool=exp-sgcheck</option> option
487    to try it, but be aware that it is not as robust as Memcheck.</para>
488  </answer>
489</qandaentry>
490
491</qandadiv>
492
493
494
495<!-- Miscellaneous -->
496<qandadiv id="faq.misc" xreflabel="Miscellaneous">
497<title>Miscellaneous</title>
498
499<qandaentry id="faq.writesupp">
500  <question id="q-writesupp">
501    <para>I tried writing a suppression but it didn't work.  Can you
502    write my suppression for me?</para>
503  </question>
504  <answer id="a-writesupp">
505    <para>Yes!  Use the <option>--gen-suppressions=yes</option> feature
506    to spit out suppressions automatically for you.  You can then edit
507    them if you like, eg.  combining similar automatically generated
508    suppressions using wildcards like <literal>'*'</literal>.</para>
509
510    <para>If you really want to write suppressions by hand, read the
511    manual carefully.  Note particularly that C++ function names must be
512    mangled (that is, not demangled).</para>
513  </answer>
514</qandaentry>
515
516
517<qandaentry id="faq.deflost">
518  <question id="q-deflost">
519    <para>With Memcheck's memory leak detector, what's the
520    difference between "definitely lost", "indirectly lost", "possibly
521    lost", "still reachable", and "suppressed"?</para>
522  </question>
523  <answer id="a-deflost">
524    <para>The details are in the Memcheck section of the user manual.</para>
525
526    <para>In short:</para>
527    <itemizedlist>
528      <listitem>
529        <para>"definitely lost" means your program is leaking memory --
530        fix those leaks!</para>
531      </listitem>
532      <listitem>
533        <para>"indirectly lost" means your program is leaking memory in
534        a pointer-based structure.  (E.g. if the root node of a binary tree
535        is "definitely lost", all the children will be "indirectly lost".)
536        If you fix the "definitely lost" leaks, the "indirectly lost" leaks
537        should go away.
538        </para>
539      </listitem>
540      <listitem>
541        <para>"possibly lost" means your program is leaking
542        memory, unless you're doing unusual things with pointers that could
543        cause them to point into the middle of an allocated block;  see the
544        user manual for some possible causes.  Use
545        <option>--show-possibly-lost=no</option> if you don't want to see
546        these reports.</para>
547      </listitem>
548      <listitem>
549        <para>"still reachable" means your program is probably ok -- it
550        didn't free some memory it could have.  This is quite common and
551        often reasonable.  Don't use
552        <option>--show-reachable=yes</option> if you don't want to see
553        these reports.</para>
554      </listitem>
555      <listitem>
556        <para>"suppressed" means that a leak error has been suppressed.
557        There are some suppressions in the default suppression files.
558        You can ignore suppressed errors.</para>
559      </listitem>
560    </itemizedlist>
561  </answer>
562</qandaentry>
563
564<qandaentry id="faq.undeferrors">
565  <question id="q-undeferrors">
566    <para>Memcheck's uninitialised value errors are hard to track down,
567    because they are often reported some time after they are caused.  Could
568    Memcheck record a trail of operations to better link the cause to the
569    effect?  Or maybe just eagerly report any copies of uninitialised
570    memory values?</para>
571  </question>
572  <answer id="a-undeferrors">
573    <para>Prior to version 3.4.0, the answer was "we don't know how to do it
574    without huge performance penalties".  As of 3.4.0, try using the
575    <option>--track-origins=yes</option> option.  It will run slower than
576    usual, but will give you extra information about the origin of
577    uninitialised values.</para>
578
579    <para>Or if you want to do it the old fashioned way, you can use the
580    client request
581    <computeroutput>VALGRIND_CHECK_VALUE_IS_DEFINED</computeroutput> to help
582    track these errors down -- work backwards from the point where the
583    uninitialised error occurs, checking suspect values until you find the
584    cause.  This requires editing, compiling and re-running your program
585    multiple times, which is a pain, but still easier than debugging the
586    problem without Memcheck's help.</para>
587
588    <para>As for eager reporting of copies of uninitialised memory values,
589    this has been suggested multiple times.  Unfortunately, almost all
590    programs legitimately copy uninitialised memory values around (because
591    compilers pad structs to preserve alignment) and eager checking leads to
592    hundreds of false positives.  Therefore Memcheck does not support eager
593    checking at this time.</para>
594  </answer>
595</qandaentry>
596
597
598<qandaentry id="faq.attach">
599  <question id="q-attach">
600    <para>Is it possible to attach Valgrind to a program that is already
601    running?</para>
602  </question>
603  <answer id="a-attach">
604    <para>No.  The environment that Valgrind provides for running programs
605    is significantly different to that for normal programs, e.g. due to
606    different layout of memory.  Therefore Valgrind has to have full control
607    from the very start.</para>
608
609    <para>It is possible to achieve something like this by running your
610    program without any instrumentation (which involves a slow-down of about
611    5x, less than that of most tools), and then adding instrumentation once
612    you get to a point of interest.  Support for this must be provided by
613    the tool, however, and Callgrind is the only tool that currently has
614    such support.  See the instructions on the
615    <computeroutput>callgrind_control</computeroutput> program for details.
616    </para>
617  </answer>
618</qandaentry>
619
620
621</qandadiv>
622
623
624
625<!-- Further Assistance -->
626<qandadiv id="faq.help" xreflabel="How To Get Further Assistance">
627<title>How To Get Further Assistance</title>
628
629<!-- WARNING: this file should not xref other parts of the docs, because it
630is built standalone as FAQ.txt.  That's why we link to, for example, the
631online copy of the manual. -->
632
633<qandaentry id="e-help">
634  <!-- <question><para/></question> -->
635  <answer id="a-help">
636  <para>Read the appropriate section(s) of the
637  <ulink url="&vg-docs-url;">Valgrind Documentation</ulink>.</para>
638
639  <para><ulink url="http://search.gmane.org">Search</ulink> the
640  <ulink url="http://news.gmane.org/gmane.comp.debugging.valgrind">valgrind-users</ulink> mailing list archives, using the group name
641  <computeroutput>gmane.comp.debugging.valgrind</computeroutput>.</para>
642
643  <para>If you think an answer in this FAQ is incomplete or inaccurate, please
644  e-mail <ulink url="mailto:&vg-vemail;">&vg-vemail;</ulink>.</para>
645
646  <para>If you have tried all of these things and are still
647  stuck, you can try mailing the
648  <ulink url="&vg-lists-url;">valgrind-users mailing list</ulink>.
649  Note that an email has a better change of being answered usefully if it is
650  clearly written.  Also remember that, despite the fact that most of the
651  community are very helpful and responsive to emailed questions, you are
652  probably requesting help from unpaid volunteers, so you have no guarantee
653  of receiving an answer.</para>
654</answer>
655
656</qandaentry>
657</qandadiv>
658
659
660<!-- FAQ ends here -->
661</qandaset>
662
663
664
665<!-- template
666<qandadiv id="faq.installing" xreflabel="Installing">
667<title>Installing</title>
668
669<qandaentry id="faq.problem">
670 <question id="q-problem">
671  <para></para>
672 </question>
673 <answer id="a-problem">
674  <para></para>
675 </answer>
676</qandaentry>
677
678</qandadiv>
679-->
680
681</article>
682
683</book>
684