llvm-project/lldb/www/varformats.html

1323 lines
75 KiB
HTML
Executable File

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta http-equiv="Content-Type" content="text/html;
charset=ISO-8859-1">
<link href="style.css" rel="stylesheet" type="text/css">
<title>LLDB Data Formatters</title>
</head>
<body>
<div class="www_title"> The <strong>LLDB</strong> Debugger </div>
<div id="container">
<div id="content">
<!--#include virtual="sidebar.incl"-->
<div id="middle">
<div class="post">
<h1 class="postheader">Variable display</h1>
<div class="postcontent">
<p>LLDB has a data formatters subsystem that allows users to define custom display options for their variables.</p>
<p>Usually, when you type <code>frame variable</code> or
run some <code>expression</code> LLDB will
automatically choose the way to display your results on
a per-type basis, as in the following example:</p>
<p> <code> <b>(lldb)</b> frame variable<br>
(uint8_t) x = 'a'<br>
(intptr_t) y = 124752287<br>
</code> </p>
<p>However, in certain cases, you may want to associate a
different style to the display for certain datatypes.
To do so, you need to give hints to the debugger as to
how variables should be displayed.<br>
The LLDB <b>type</b> command allows you to do just that.<br>
</p>
<p>Using it you can change your visualization to look like this: </p>
<p> <code> <b>(lldb)</b> frame variable<br>
(uint8_t) x = chr='a' dec=65 hex=0x41<br>
(intptr_t) y = 0x76f919f<br>
</code> </p>
<p>There are several features related to data visualization: <span
style="font-style: italic;">formats</span>, <span
style="font-style: italic;">summaries</span>, <span
style="font-style: italic;">filters</span>, <span
style="font-style: italic;">synthetic children</span>.</p>
<p>To reflect this, the <b>type</b> command has five
subcommands:<br>
</p>
<p><code>type format</code></p>
<p><code>type summary</code></p>
<p><code>type filter</code></p>
<p><code>type synthetic</code></p>
<p><code>type category</code></p>
<p>These commands are meant to bind printing options to
types. When variables are printed, LLDB will first check
if custom printing options have been associated to a
variable's type and, if so, use them instead of picking
the default choices.<br>
</p>
<p>Each of the commands (except <code>type category</code>) has four subcommands available:<br>
</p>
<p><code>add</code>: associates a new printing option to one
or more types</p>
<p><code>delete</code>: deletes an existing association</p>
<p><code>list</code>: provides a listing of all
associations</p>
<p><code>clear</code>: deletes all associations</p>
</div>
</div>
<div class="post">
<h1 class="postheader">type format</h1>
<div class="postcontent">
<p>Type formats enable you to quickly override the default
format for displaying primitive types (the usual basic
C/C++/ObjC types: <code><font color="blue">int</font></code>, <code><font color="blue">float</font></code>, <code><font color="blue">char</font></code>, ...).</p>
<p>If for some reason you want all <code>int</code>
variables in your program to print out as hex, you can add
a format to the <code>int</code> type.<br></p>
<p>This is done by typing
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type format add --format hex int
</td>
<table>
at the LLDB command line.</p>
<p>The <code>--format</code> (which you can shorten to <code>-f</code>) option accepts a <a
href="#formatstable">format name</a>. Then, you provide one or more
types to which you want the new format applied.</p>
<p>A frequent scenario is that your program has a <code>typedef</code>
for a numeric type that you know represents something
that must be printed in a certain way. Again, you can
add a format just to that typedef by using <code>type
format add</code> with the name alias.</p>
<p>But things can quickly get hierarchical. Let's say you
have a situation like the following:</p>
<p><code><font color="blue">typedef int</font> A;<br>
<font color="blue">typedef</font> A B;<br>
<font color="blue">typedef</font> B C;<br>
<font color="blue">typedef</font> C D;<br>
</code></p>
<p>and you want to show all <code>A</code>'s as hex, all
<code>C'</code>s as byte arrays and leave the defaults
untouched for other types (albeit its contrived look, the example is far
from unrealistic in large software systems).</p>
<p>If you simply type <br>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type format add -f hex A<br>
<b>(lldb)</b> type format add -f uint8_t[] C
</td>
<table>
<br>
values of type <code>B</code> will be shown as hex
and values of type <code>D</code> as byte arrays, as in:</p>
<p> <code>
<b>(lldb)</b> frame variable -T<br/>
(A) a = 0x00000001<br/>
(B) b = 0x00000002<br/>
(C) c = {0x03 0x00 0x00 0x00}<br/>
(D) d = {0x04 0x00 0x00 0x00}<br/>
</code> </p>
<p>This is because by default LLDB <i>cascades</i>
formats through typedef chains. In order to avoid that
you can use the option <code>-C no</code> to prevent
cascading, thus making the two commands required to
achieve your goal:<br>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type format add -C no -f hex A<br>
<b>(lldb)</b> type format add -C no -f uint8_t[] C
</td>
<table>
<p>which provides the desired output:</p>
<p> <code>
<b>(lldb)</b> frame variable -T<br/>
(A) a = 0x00000001<br/>
(B) b = 2<br/>
(C) c = {0x03 0x00 0x00 0x00}<br/>
(D) d = 4<br/>
</code> </p>
<p>Two additional options that you will want to look at
are <code>--skip-pointers</code> (<code>-p</code>) and <code>--skip-references</code> (<code>-r</code>). These two
options prevent LLDB from applying a format for type <code>T</code>
to values of type <code>T*</code> and <code>T&amp;</code>
respectively.</p>
<p> <code> <b>(lldb)</b> type format add -f float32[]
int<br>
<b>(lldb)</b> frame variable pointer *pointer -T<br>
(int *) pointer = {1.46991e-39 1.4013e-45}<br>
(int) *pointer = {1.53302e-42}<br>
<b>(lldb)</b> type format add -f float32[] int -p<br>
<b>(lldb)</b> frame variable pointer *pointer -T<br>
(int *) pointer = 0x0000000100100180<br>
(int) *pointer = {1.53302e-42}<br>
</code> </p>
<p>While they can be applied to pointers and references, formats will make no attempt
to dereference the pointer and extract the value before applying the format, which means you
are effectively formatting the address stored in the pointer rather than the pointee value.
For this reason, you may want to use the <code>-p</code> option when defining formats.</p>
<p>If you need to delete a custom format simply type <code>type
format delete</code> followed by the name of the type
to which the format applies.Even if you
defined the same format for multiple types on the same command,
<code>type format delete</code> will only remove the format for
the type name passed as argument.<br>
</p>
<p>
To delete ALL formats, use
<code>type format clear</code>. To see all the formats
defined, use <code>type format list</code>.</p>
<p>If all you need to do, however, is display one variable
in a custom format, while leaving the others of the same
type untouched, you can simply type:<br>
<br>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> frame variable counter -f hex
</td>
<table>
<p>This has the effect of displaying the value of <code>counter</code>
as an hexadecimal number, and will keep showing it this
way until you either pick a different format or till you
let your program run again.</p>
<p>Finally, this is a list of formatting options available
out of
which you can pick:</p><a name="formatstable"></a>
<table border="1">
<tbody>
<tr valign="top">
<td width="23%"><b>Format name</b></td>
<td><b>Abbreviation</b></td>
<td><b>Description</b></td>
</tr>
<tr valign="top">
<td><b>default</b></td>
<td><br>
</td>
<td>the default LLDB algorithm is used to pick a
format</td>
</tr>
<tr valign="top">
<td><b>boolean</b></td>
<td>B</td>
<td>show this as a true/false boolean, using the
customary rule that 0 is false and everything else
is true</td>
</tr>
<tr valign="top">
<td><b>binary</b></td>
<td>b</td>
<td>show this as a sequence of bits</td>
</tr>
<tr valign="top">
<td><b>bytes</b></td>
<td>y</td>
<td>show the bytes one after the other<br>
e.g. <code>(int) s.x = 07 00 00 00</code></td>
</tr>
<tr valign="top">
<td><b>bytes with ASCII</b></td>
<td>Y</td>
<td>show the bytes, but try to display them as ASCII
characters as well<br>
e.g. <code>(int *) c.sp.x = 50 f8 bf 5f ff 7f 00
00 P.._....</code></td>
</tr>
<tr valign="top">
<td><b>character</b></td>
<td>c</td>
<td>show the bytes as ASCII characters<br>
e.g. <code>(int *) c.sp.x =
P\xf8\xbf_\xff\x7f\0\0</code></td>
</tr>
<tr valign="top">
<td><b>printable character</b></td>
<td>C</td>
<td>show the bytes as printable ASCII
characters<br>
e.g. <code>(int *) c.sp.x = P.._....</code></td>
</tr>
<tr valign="top">
<td><b>complex float</b></td>
<td>F</td>
<td>interpret this value as the real and imaginary
part of a complex floating-point number<br>
e.g. <code>(int *) c.sp.x = 2.76658e+19 +
4.59163e-41i</code></td>
</tr>
<tr valign="top">
<td><b>c-string</b></td>
<td>s</td>
<td>show this as a 0-terminated C string</td>
</tr>
<tr valign="top">
<td><b>decimal</b></td>
<td>i</td>
<td>show this as a signed integer number (this does
not perform a cast, it simply shows the bytes as
an integer with sign)</td>
</tr>
<tr valign="top">
<td><b>enumeration</b></td>
<td>E</td>
<td>show this as an enumeration, printing the
value's name if available or the integer value
otherwise<br>
e.g. <code>(enum enumType) val_type = eValue2</code></td>
</tr>
<tr valign="top">
<td><b>hex</b></td>
<td>x</td>
<td>show this as in hexadecimal notation (this does
not perform a cast, it simply shows the bytes as
hex)</td>
</tr>
<tr valign="top">
<td><b>float</b></td>
<td>f</td>
<td>show this as a floating-point number (this does
not perform a cast, it simply interprets the bytes
as an IEEE754 floating-point value)</td>
</tr>
<tr valign="top">
<td><b>octal</b></td>
<td>o</td>
<td>show this in octal notation</td>
</tr>
<tr valign="top">
<td><b>OSType</b></td>
<td>O</td>
<td>show this as a MacOS OSType<br>
e.g. <code>(float) x = '\n\x1f\xd7\n'</code></td>
</tr>
<tr valign="top">
<td><b>unicode16</b></td>
<td>U</td>
<td>show this as UTF-16 characters<br>
e.g. <code>(float) x = 0xd70a 0x411f</code></td>
</tr>
<tr valign="top">
<td><b>unicode32</b></td>
<td><br>
</td>
<td>show this as UTF-32 characters<br>
e.g. <code>(float) x = 0x411fd70a</code></td>
</tr>
<tr valign="top">
<td><b>unsigned decimal</b></td>
<td>u</td>
<td>show this as an unsigned integer number (this
does not perform a cast, it simply shows the bytes
as unsigned integer)</td>
</tr>
<tr valign="top">
<td><b>pointer</b></td>
<td>p</td>
<td>show this as a native pointer (unless this is
really a pointer, the resulting address will
probably be invalid)</td>
</tr>
<tr valign="top">
<td><b>char[]</b></td>
<td><br>
</td>
<td>show this as an array of characters<br>
e.g. <code>(char) *c.sp.z = {X}</code></td>
</tr>
<tr valign="top">
<td><b>int8_t[], uint8_t[]<br>
int16_t[], uint16_t[]<br>
int32_t[], uint32_t[]<br>
int64_t[], uint64_t[]<br>
uint128_t[]</b></td>
<td><br>
</td>
<td>show this as an array of the corresponding
integer type<br>
e.g.<br>
<code>(int) x = {1 0 0 0}</code> (with uint8_t[])<br>
<code>(int) y = {0x00000001}</code> (with uint32_t[])</td>
</tr>
<tr valign="top">
<td><b>float32[], float64[]</b></td>
<td><br>
</td>
<td>show this as an array of the corresponding
floating-point type<br>
e.g. <code>(int *) pointer = {1.46991e-39
1.4013e-45}</code></td>
</tr>
<tr valign="top">
<td><b>complex integer</b></td>
<td>I</td>
<td>interpret this value as the real and imaginary
part of a complex integer number<br>
e.g. <code>(int *) pointer = 1048960 + 1i</code></td>
</tr>
<tr valign="top">
<td><b>character array</b></td>
<td>a</td>
<td>show this as a character array<br>
e.g. <code>(int *) pointer =
\x80\x01\x10\0\x01\0\0\0</code></td>
</tr>
</tbody>
</table>
</div>
</div>
<div class="post">
<h1 class="postheader">type summary</h1>
<div class="postcontent">
<p>Type formats work by showing a different kind of display for
the value of a variable. However, they only work for basic types.
When you want to display a class or struct in a custom format, you
cannot do that using formats.</p>
<p>A different feature, type summaries, works by extracting
information from classes, structures, ... (<i>aggregate types</i>)
and arranging it in a user-defined format, as in the following example:</p>
<p> <i>before adding a summary...</i><br>
<code> <b>(lldb)</b> frame variable -T one<br>
(i_am_cool) one = {<br>
&nbsp;&nbsp;&nbsp;&nbsp;(int) x = 3<br>
&nbsp;&nbsp;&nbsp;&nbsp;(float) y = 3.14159<br>
&nbsp;&nbsp;&nbsp;&nbsp;(char) z = 'E'<br>
}<br>
</code> <br>
<i>after adding a summary...</i><br>
<code> <b>(lldb)</b> frame variable one<br>
(i_am_cool) one = int = 3, float = 3.14159, char = 69<br>
</code> </p>
<p>There are two ways to use type summaries: the first one is to bind a <i>
summary string</i> to the type; the second is to write a Python script that returns
the string to be used as summary. Both options are enabled by the <code>type summary add</code>
command.</p>
<p>The command to obtain the output shown in the example is:</p>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --summary-string "int = ${var.x}, float = ${var.y}, char = ${var.z%u}" i_am_cool
</td>
<table>
<p>Initially, we will focus on summary strings, and then describe the Python binding
mechanism.</p>
</div>
</div>
<div class="post">
<h1 class="postheader">Summary Strings</h1>
<div class="postcontent">
<p>Summary strings are written using a simple control language, exemplified by the snippet above.
A summary string contains a sequence of tokens that are processed by LLDB to generate the summary.</p>
<p>Summary strings can contain plain text, control characters and
special variables that have access to information about
the current object and the overall program state.</p>
<p>Plain text is any sequence of characters that doesn't contain a <code><b>'{'</b></code>,
<code><b>'}'</b></code>, <code><b>'$'</b></code>, or <code><b>'\'</b></code>
character, which are the syntax control characters.</p>
<p>The special variables are found in between a <code><b>"${"</b></code>
prefix, and end with a <code><b>"}"</b></code> suffix. Variables can be a simple name
or they can refer to complex objects that have subitems themselves.
In other words, a variable looks like <code>"<b>${object}</b>"</code> or
<code>"<b>${object.child.otherchild}</b>"</code>. A variable can also be prefixed or
suffixed with other symbols meant to change the way its value is handled. An example is
<code>"<b>${*var.int_pointer[0-3]}</b>".</code></p>
<p>Basically, the syntax is the same one described <a
href="formats.html">Frame and Thread Formatting</a>
plus additional symbols specific for summary strings. The main of them is <code>${var</code>,
which is used refer to the variable that a summary is being created for.</p>
<p>The simplest thing you can do is grab a member variable
of a class or structure by typing its <i>expression
path</i>. In the previous example, the expression path
for the field <code>float y</code> is simply <code>.y</code>.
Thus, to ask the summary string to display <code>y</code>
you would type <code>${var.y}</code>.</p>
<p>If you have code like the following: <br>
<code> <font color="blue">struct</font> A {<br>
&nbsp;&nbsp;&nbsp;&nbsp;<font color="blue">int</font> x;<br>
&nbsp;&nbsp;&nbsp;&nbsp;<font color="blue">int</font> y;<br>
};<br>
<font color="blue">struct</font> B {<br>
&nbsp;&nbsp;&nbsp;&nbsp;A x;<br>
&nbsp;&nbsp;&nbsp;&nbsp;A y;<br>
&nbsp;&nbsp;&nbsp;&nbsp;<font color="blue">int</font> *z;<br>
};<br>
</code> the expression path for the <code>y</code>
member of the <code>x</code> member of an object of
type <code>B</code> would be <code>.x.y</code> and you
would type <code>${var.x.y}</code> to display it in a
summary string for type <code>B</code>. </p>
<p>By default, a summary defined for type <code>T</code>, also works for types
<code>T*</code> and <code>T&amp;</code> (you can disable this behavior if desired).
For this reason, expression paths do not differentiate between <code>.</code>
and <code>-&gt;</code>, and the above expression path <code>.x.y</code>
would be just as good if you were displaying a <code>B*</code>,
or even if the actual definition of <code>B</code>
were: <code><br>
<font color="blue">struct</font> B {<br>
&nbsp;&nbsp;&nbsp;&nbsp;A *x;<br>
&nbsp;&nbsp;&nbsp;&nbsp;A y;<br>
&nbsp;&nbsp;&nbsp;&nbsp;<font color="blue">int</font> *z;<br>
};<br>
</code> </p>
<p>This is unlike the behavior of <code>frame variable</code>
which, on the contrary, will enforce the distinction. As
hinted above, the rationale for this choice is that
waiving this distinction enables you to write a summary
string once for type <code>T</code> and use it for both
<code>T</code> and <code>T*</code> instances. As a
summary string is mostly about extracting nested
members' information, a pointer to an object is just as
good as the object itself for the purpose.</p>
<p>If you need to access the value of the integer pointed to by <code>B::z</code>, you
cannot simply say <code>${var.z}</code> because that symbol refers to the pointer <code>z</code>.
In order to dereference it and get the pointed value, you should say <code>${*var.z}</code>. The <code>${*var</code>
tells LLDB to get the object that the expression paths leads to, and then dereference it. In this example is it
equivalent to <code>*(bObject.z)</code> in C/C++ syntax. Because <code>.</code> and <code>-></code> operators can both be
used, there is no need to have dereferences in the middle of an expression path (e.g. you do not need to type
<code>${*(var.x).x})</code> to read <code>A::x</code> as contained in <code>*(B::x)</code>. To achieve that effect
you can simply write <code>${var.x->x}</code>, or even <code>${var.x.x}</code>. The <code>*</code> operator only binds
to the result of the whole expression path, rather than piecewise, and there is no way to use parentheses to change
that behavior.</p>
<p>Of course, a summary string can contain more than one <code>${var</code> specifier,
and can use <code>${var</code> and <code>${*var</code> specifiers together.</p>
</div>
</div>
<div class="post">
<h1 class="postheader">Formatting summary elements</h1>
<div class="postcontent">
<p>An expression path can include formatting codes.
Much like the type formats discussed previously, you can also customize
the way variables are displayed in summary strings, regardless of the format they have
applied to their types. To do that, you can use <code>%<i>format</i></code> inside an expression path,
as in <code>${var.x->x%u}</code>, which would display the value of <code>x</code> as an unsigned integer.
<p>You can also use some other special format markers, not available
for formats themselves, but which carry a special meaning when used in this
context:</p>
<table border="1">
<tbody>
<tr valign="top">
<td width="23%"><b>Symbol</b></td>
<td><b>Description</b></td>
</tr>
<tr valign="top">
<td><b>%S</b></td>
<td>Use this object's summary (the default for aggregate types)</td>
</tr>
<tr valign="top">
<td><b>%V</b></td>
<td>Use this object's value (the default for non-aggregate types)</td>
</tr>
<tr valign="top">
<td><b>%@</b></td>
<td>Use a language-runtime specific description (for C++ this does nothing,
for Objective-C it calls the NSPrintForDebugger API)</td>
</tr>
<tr valign="top">
<td><b>%L</b></td>
<td>Use this object's location (memory address, register name, ...)</td>
</tr>
<tr valign="top">
<td><b>%#</b></td>
<td>Use the count of the children of this object</td>
</tr>
<tr valign="top">
<td><b>%T</b></td>
<td>Use this object's datatype name</td>
</tr>
</tbody>
</table>
<p>Starting with SVN r220821, you can also specify ${var.script:<i>pythonFuncName</i>}.
<br/>It is expected that the function name you use specifies a function whose signature is the same
as a Python summary function. The return string from the function will be placed verbatim in the output.
<br/><br/>
You cannot use element access, or formatting symbols, in combination with this syntax. For example the following:
<table class="stats" width="620" cellspacing="0">
<td class="content">
${var.element[0].script:myFunctionName%@}
</td>
<table>
is not valid and will cause the summary to fail to evaluate.
</p>
</div>
</div>
<div class="post">
<h1 class="postheader">Element inlining</h1>
<div class="postcontent">
<p>Option <code>--inline-children</code> (<code>-c</code>) to <code>type summary add</code>
tells LLDB not to look for a summary string, but instead
to just print a listing of all the object's children on
one line.</p>
<p> As an example, given a type <code>pair</code>:
<code> <br>
<b>(lldb)</b> frame variable --show-types a_pair<br>
(pair) a_pair = {<br>
&nbsp;&nbsp;&nbsp;&nbsp;(int) first = 1;<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) second = 2;<br/>
}<br>
</code><br>
If one types the following commands:
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --inline-children pair<br>
</td>
<table>
the output becomes: <br><code>
<b>(lldb)</b> frame variable a_pair<br>
(pair) a_pair = (first=1, second=2)<br>
</code> </p>
Of course, one can obtain the same effect by typing
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add pair --summary-string "(first=${var.first}, second=${var.second})"<br>
</td>
<table>
While the final result is the same, using <code>--inline-children</code> can often save time. If one does not need to
see the names of the variables, but just their values, the option <code>--omit-names</code> (<code>-O</code>, uppercase letter o), can be combined with <code>--inline-children</code> to obtain:
<br><code>
<b>(lldb)</b> frame variable a_pair<br>
(pair) a_pair = (1, 2)<br>
</code> </p>
which is of course the same as
typing
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add pair --summary-string "(${var.first}, ${var.second})"<br>
</td>
<table>
</div>
</div>
<div class="post">
<h1 class="postheader">Bitfields and array syntax</h1>
<div class="postcontent">
<p>Sometimes, a basic type's value actually represents
several different values packed together in a bitfield.<br/>
With the classical view, there is no way to look at
them. Hexadecimal display can help, but if the bits
actually span nibble boundaries, the help is limited.<br/>
Binary view would show it all without ambiguity, but is
often too detailed and hard to read for real-life
scenarios.
<p>
To cope with the issue, LLDB supports native
bitfield formatting in summary strings. If your
expression paths leads to a so-called <i>scalar type</i>
(the usual int, float, char, double, short, long, long
long, double, long double and unsigned variants), you
can ask LLDB to only grab some bits out of the value and
display them in any format you like. If you only need one bit
you can use the <code>[</code><i>n</i><code>]</code>, just like
indexing an array. To extract multiple bits, you can use
a slice-like syntax: <code>[</code><i>n</i>-<i>m</i><code>]</code>, e.g. <br><p>
<code> <b>(lldb)</b> frame variable float_point<br>
(float) float_point = -3.14159<br> </code>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --summary-string "Sign: ${var[31]%B}
Exponent: ${var[30-23]%x} Mantissa: ${var[0-22]%u}"
float
</td>
</table><br></code>
<code>
<b>(lldb)</b> frame variable float_point<br>
(float) float_point = -3.14159 Sign: true Exponent:
0x00000080 Mantissa: 4788184<br>
</code> In this example, LLDB shows the internal
representation of a <code>float</code> variable by
extracting bitfields out of a float object.</p>
<p> When typing a range, the extremes <i>n</i> and <i>m</i> are always
included, and the order of the indices is irrelevant. </p>
<p>LLDB also allows to use a similar syntax to display
array members inside a summary string. For instance, you
may want to display all arrays of a given type using a
more compact notation than the default, and then just
delve into individual array members that prove
interesting to your debugging task. You can tell
LLDB to format arrays in special ways, possibly
independent of the way the array members' datatype is formatted. <br>
e.g. <br>
<code> <b>(lldb)</b> frame variable sarray<br>
(Simple [3]) sarray = {<br>
&nbsp;&nbsp;&nbsp;&nbsp;[0] = {<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;x = 1<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;y = 2<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;z = '\x03'<br>
&nbsp;&nbsp;&nbsp;&nbsp;}<br>
&nbsp;&nbsp;&nbsp;&nbsp;[1] = {<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;x = 4<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;y = 5<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;z = '\x06'<br>
&nbsp;&nbsp;&nbsp;&nbsp;}<br>
&nbsp;&nbsp;&nbsp;&nbsp;[2] = {<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;x = 7<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;y = 8<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;z = '\t'<br>
&nbsp;&nbsp;&nbsp;&nbsp;}<br>
}<br></code>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --summary-string "${var[].x}" "Simple
[3]"
</td>
<table><br>
<code>
<b>(lldb)</b> frame variable sarray<br>
(Simple [3]) sarray = [1,4,7]<br></code></p>
<p>The <code>[]</code> symbol amounts to: <i>if <code>var</code>
is an array and I know its size, apply this summary
string to every element of the array</i>. Here, we are
asking LLDB to display <code>.x</code> for every
element of the array, and in fact this is what happens.
If you find some of those integers anomalous, you can
then inspect that one item in greater detail, without
the array format getting in the way: <br>
<code> <b>(lldb)</b> frame variable sarray[1]<br>
(Simple) sarray[1] = {<br>
&nbsp;&nbsp;&nbsp;&nbsp;x = 4<br>
&nbsp;&nbsp;&nbsp;&nbsp;y = 5<br>
&nbsp;&nbsp;&nbsp;&nbsp;z = '\x06'<br>
}<br>
</code> </p>
<p>You can also ask LLDB to only print a subset of the
array range by using the same syntax used to extract bit
for bitfields:
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --summary-string "${var[1-2].x}" "Simple
[3]"
</td>
<table><br>
<code>
<b>(lldb)</b> frame variable sarray<br>
(Simple [3]) sarray = [4,7]<br></code></p>
<p>If you are dealing with a pointer that you know is an array, you can use this
syntax to display the elements contained in the pointed array instead of just
the pointer value. However, because pointers have no notion of their size, the
empty brackets <code>[]</code> operator does not work, and you must explicitly provide
higher and lower bounds.</p>
<p>In general, LLDB needs the square brackets operator <code>[]</code> in
order to handle arrays and pointers correctly, and for pointers it also
needs a range. However, a few special cases are defined to make your life easier:
<ul>
<li>you can print a 0-terminated string (<i>C-string</i>) using the %s format,
omitting square brackets, as in:
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --summary-string "${var%s}" "char *"
</td>
<table>
<p>
This syntax works for <code>char*</code> as well as for <code>char[]</code>
because LLDB can rely on the final <code>\0</code> terminator to know when the string
has ended.</p>
LLDB has default summary strings for <code>char*</code> and <code>char[]</code> that use
this special case. On debugger startup, the following are defined automatically:
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --summary-string "${var%s}" "char *"<br/>
<b>(lldb)</b> type summary add --summary-string "${var%s}" -x "char \[[0-9]+]"<br/>
</td>
<table>
</li>
</ul>
<ul>
<li>any of the array formats (<code>int8_t[]</code>,
<code>float32{}</code>, ...), and the <code>y</code>, <code>Y</code>
and <code>a</code> formats
work to print an array of a non-aggregate
type, even if square brackets are omitted.
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --summary-string "${var%int32_t[]}" "int [10]"
</td>
<table>
</ul>
This feature, however, is not enabled for pointers because there is no
way for LLDB to detect the end of the pointed data.
<br>
This also does not work for other formats (e.g. <code>boolean</code>), and you must
specify the square brackets operator to get the expected output.
</p>
</div>
</div>
<div class="post">
<h1 class="postheader">Python scripting</h1>
<div class="postcontent">
<p>Most of the times, summary strings prove good enough for the job of summarizing
the contents of a variable. However, as soon as you need to do more than picking
some values and rearranging them for display, summary strings stop being an
effective tool. This is because summary strings lack the power to actually perform
any kind of computation on the value of variables.</p>
<p>To solve this issue, you can bind some Python scripting code as a summary for
your datatype, and that script has the ability to both extract children variables
as the summary strings do and to perform active computation on the extracted
values. As a small example, let's say we have a Rectangle class:</p>
<code>
<font color="blue">class</font> Rectangle<br/>
{<br/>
<font color="blue">private</font>:<br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color="blue">int</font> height;<br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color="blue">int</font> width;<br/>
<font color="blue">public</font>:<br/>
&nbsp;&nbsp;&nbsp;&nbsp;Rectangle() : height(3), width(5) {}<br/>
&nbsp;&nbsp;&nbsp;&nbsp;Rectangle(<font color="blue">int</font> H) : height(H), width(H*2-1) {}<br/>
&nbsp;&nbsp;&nbsp;&nbsp;Rectangle(<font color="blue">int</font> H, <font color="blue">int</font> W) : height(H), width(W) {}<br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color="blue">int</font> GetHeight() { return height; }<br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color="blue">int</font> GetWidth() { return width; }<br/>
};<br/>
</code>
<p>Summary strings are effective to reduce the screen real estate used by
the default viewing mode, but are not effective if we want to display the
area and perimeter of <code>Rectangle</code> objects</p>
<p>To obtain this, we can simply attach a small Python script to the <code>Rectangle</code>
class, as shown in this example:</p>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add -P Rectangle<br/>
Enter your Python command(s). Type 'DONE' to end.<br/>
def function (valobj,internal_dict):<br/>
&nbsp;&nbsp;&nbsp;&nbsp;height_val = valobj.GetChildMemberWithName('height')<br/>
&nbsp;&nbsp;&nbsp;&nbsp;width_val = valobj.GetChildMemberWithName('width')<br/>
&nbsp;&nbsp;&nbsp;&nbsp;height = height_val.GetValueAsUnsigned(0)<br/>
&nbsp;&nbsp;&nbsp;&nbsp;width = width_val.GetValueAsUnsigned(0)<br/>
&nbsp;&nbsp;&nbsp;&nbsp;area = height*width<br/>
&nbsp;&nbsp;&nbsp;&nbsp;perimeter = 2*(height + width)<br/>
&nbsp;&nbsp;&nbsp;&nbsp;return 'Area: ' + str(area) + ', Perimeter: ' + str(perimeter)<br/>
&nbsp;&nbsp;&nbsp;&nbsp;DONE<br/>
<b>(lldb)</b> frame variable<br/>
(Rectangle) r1 = Area: 20, Perimeter: 18<br/>
(Rectangle) r2 = Area: 72, Perimeter: 36<br/>
(Rectangle) r3 = Area: 16, Perimeter: 16<br/>
</td>
</table>
<p>In order to write effective summary scripts, you need to know the LLDB public
API, which is the way Python code can access the LLDB object model. For further
details on the API you should look at <a href="scripting.html">this page</a>, or at
the LLDB <a href="docs.html">API reference documentation</a>.</p>
<p>As a brief introduction, your script is encapsulated into a function that is
passed two parameters: <code>valobj</code> and <code>internal_dict</code>.</p>
<p><code>internal_dict</code> is an internal support parameter used by LLDB and you should
not touch it.<br/><code>valobj</code> is the object encapsulating the actual
variable being displayed, and its type is <a href="http://llvm.org/svn/llvm-project/lldb/trunk/include/lldb/API/SBValue.h">SBValue</a>.
Out of the many possible operations on an SBValue, the basic one is retrieve the children objects
it contains (essentially, the fields of the object wrapped by it), by calling
<code>GetChildMemberWithName()</code>, passing it the child's name as a string.<br/>
If the variable has a value, you can ask for it, and return it as a string using <code>GetValue()</code>,
or as a signed/unsigned number using <code>GetValueAsSigned()</code>, <code>GetValueAsUnsigned()</code>.
It is also possible to retrieve an <a href="http://llvm.org/svn/llvm-project/lldb/trunk/include/lldb/API/SBData.h"><code>SBData</code></a> object by calling <code>GetData()</code> and then read
the object's contents out of the <code>SBData</code>.
<p>If you need to delve into several levels of hierarchy, as you can do with summary
strings, you can use the method <code>GetValueForExpressionPath()</code>, passing it
an expression path just like those you could use for summary strings (one of the differences
is that dereferencing a pointer does not occur by prefixing the path with a <code>*</code>,
but by calling the <code>Dereference()</code> method on the returned SBValue).
If you need to access array slices, you cannot do that (yet) via this method call, and you must
use <code>GetChildAtIndex()</code> querying it for the array items one by one.
Also, handling custom formats is something you have to deal with on your own.
<p>Other than interactively typing a Python script there are two other ways for you
to input a Python script as a summary:
<ul>
<li> using the --python-script option to <code>type summary add </code> and typing the script
code as an option argument; as in: </ul>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --python-script "height =
valobj.GetChildMemberWithName('height').GetValueAsUnsigned(0);width =
valobj.GetChildMemberWithName('width').GetValueAsUnsigned(0);
return 'Area: %d' % (height*width)" Rectangle<br/>
</td>
</table>
<ul>
<li> using the <code>--python-function</code> (<code>-F</code>) option to <code>type summary add </code> and giving the name of a
Python function with the correct prototype. Most probably, you will define (or have
already defined) the function in the interactive interpreter, or somehow
loaded it from a file, using the <code>command script import</code> command. LLDB will emit a warning if it is unable to find the function you passed, but will still register the binding.
</ul>
</p>
<p>Starting in SVN r222593, Python summary formatters can optionally define a third argument: <code>options</code><br/>
This is an object of type lldb.SBTypeSummaryOptions that can be passed into the formatter, allowing for a few customizations of the result.
The decision to adopt or not this third argument - and the meaning of options thereof - is within the individual formatters' writer.<br/>
</div>
</div>
<div class="post">
<h1 class="postheader">Regular expression typenames</h1>
<div class="postcontent">
<p>As you noticed, in order to associate the custom
summary string to the array types, one must give the
array size as part of the typename. This can long become
tiresome when using arrays of different sizes, <code>Simple
[3]</code>, <code>Simple [9]</code>, <code>Simple
[12]</code>, ...</p>
<p>If you use the <code>-x</code> option, type names are
treated as regular expressions instead of type names.
This would let you rephrase the above example
for arrays of type <code>Simple [3]</code> as: <br>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --summary-string "${var[].x}"
-x "Simple \[[0-9]+\]"
</td>
<table>
<code>
<b>(lldb)</b> frame variable<br>
(Simple [3]) sarray = [1,4,7]<br>
(Simple [2]) sother = [3,6]<br>
</code> The above scenario works for <code>Simple [3]</code>
as well as for any other array of <code>Simple</code>
objects. </p>
<p>While this feature is mostly useful for arrays, you
could also use regular expressions to catch other type
sets grouped by name. However, as regular expression
matching is slower than normal name matching, LLDB will
first try to match by name in any way it can, and only
when this fails, will it resort to regular expression
matching. </p>
<p>One of the ways LLDB uses this feature internally, is to match
the names of STL container classes, regardless of the template
arguments provided. The details for this are found at <a href="http://llvm.org/svn/llvm-project/lldb/trunk/source/DataFormatters/FormatManager.cpp">FormatManager.cpp</a></p>
<p>The regular expression language used by LLDB is the <a href="http://en.wikipedia.org/wiki/Regular_expression#POSIX_Extended_Regular_Expressions">POSIX extended language</a>, as defined by the <a href="http://pubs.opengroup.org/onlinepubs/7908799/xsh/regex.h.html">Single UNIX Specification</a>, of which Mac OS X is a
compliant implementation.
</div>
</div>
<div class="post">
<h1 class="postheader">Named summaries</h1>
<div class="postcontent">
<p>For a given type, there may be different meaningful summary
representations. However, currently, only one summary can be associated
to a type at each moment. If you need to temporarily override the association
for a variable, without changing the summary string for to its type,
you can use named summaries.</p>
<p>Named summaries work by attaching a name to a summary when creating
it. Then, when there is a need to attach the summary to a variable, the
<code>frame variable</code> command, supports a <code>--summary</code> option
that tells LLDB to use the named summary given instead of the default one.</p>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --summary-string "x=${var.integer}" --name NamedSummary
</td>
<table>
<code> <b>(lldb)</b> frame variable one<br>
(i_am_cool) one = int = 3, float = 3.14159, char = 69<br>
<b>(lldb)</b> frame variable one --summary NamedSummary<br>
(i_am_cool) one = x=3<br>
</code> </p>
<p>When defining a named summmary, binding it to one or more types becomes optional.
Even if you bind the named summary to a type, and later change the summary string
for that type, the named summary will not be changed by that. You can delete
named summaries by using the <code>type summary delete</code> command, as if the
summary name was the datatype that the summary is applied to</p>
<p>A summary attached to a variable using the </code>--summary</code> option,
has the same semantics that a custom format attached using the <code>-f</code>
option has: it stays attached till you attach a new one, or till you let
your program run again.</p>
</div>
</div>
<div class="post">
<h1 class="postheader">Synthetic children</h1>
<div class="postcontent">
<p>Summaries work well when one is able to navigate through an expression path.
In order for LLDB to do so, appropriate debugging information must be available.</p>
<p>Some types are <i>opaque</i>, i.e. no knowledge of their internals is provided.
When that's the case, expression paths do not work correctly.</p>
<p>In other cases, the internals are available to use in expression paths, but they
do not provide a user-friendly representation of the object's value.</p>
<p>For instance, consider an STL vector, as implemented by the <a href="http://gcc.gnu.org/onlinedocs/libstdc++/">GNU C++ Library</a>:</p>
<code>
<b>(lldb)</b> frame variable numbers -T<br/>
(std::vector&lt;int&gt;) numbers = {<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(std::_Vector_base&lt;int, std::allocator&lt;int&gt; &gt;) std::_Vector_base&lt;int, std::allocator&lt;int&gt; &gt; = {<br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(std::_Vector_base&lt;int, std::allocator&tl;int&gt; &gt;::_Vector_impl) _M_impl = {<br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(int *) _M_start = 0x00000001001008a0<br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(int *) _M_finish = 0x00000001001008a8<br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(int *) _M_end_of_storage = 0x00000001001008a8<br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;}<br/>
&nbsp;&nbsp;&nbsp;&nbsp;}<br/>
}<br/>
</code>
<p>Here, you can see how the type is implemented, and you can write a summary for that implementation
but that is not going to help you infer what items are actually stored in the vector.</p>
<p>What you would like to see is probably something like:</p>
<code>
<b>(lldb)</b> frame variable numbers -T<br/>
(std::vector&lt;int&gt;) numbers = {<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) [0] = 1<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) [1] = 12<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) [2] = 123<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) [3] = 1234<br/>
}<br/>
</code>
<p>Synthetic children are a way to get that result.</p>
<p>The feature is based upon the idea of providing a new set of children for a variable that replaces the ones
available by default through the debug information. In the example, we can use synthetic children to provide
the vector items as children for the std::vector object.</p>
<p>In order to create synthetic children, you need to provide a Python class that adheres to a given <i>interface</i>
(the word is italicized because <a href="http://en.wikipedia.org/wiki/Duck_typing">Python has no explicit notion of interface</a>, by that word we mean a given set of methods
must be implemented by the Python class):</p>
<code>
<font color=blue>class</font> SyntheticChildrenProvider:<br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color=blue>def</font> __init__(self, valobj, internal_dict):<br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<i>this call should initialize the Python object using valobj as the variable to provide synthetic children for</i> <br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color=blue>def</font> num_children(self): <br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<i>this call should return the number of children that you want your object to have</i> <br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color=blue>def</font> get_child_index(self,name): <br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<i>this call should return the index of the synthetic child whose name is given as argument</i> <br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color=blue>def</font> get_child_at_index(self,index): <br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<i>this call should return a new LLDB SBValue object representing the child at the index given as argument</i> <br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color=blue>def</font> update(self): <br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<i>this call should be used to update the internal state of this Python object whenever the state of the variables in LLDB changes.</i><sup>[1]</sup><br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color=blue>def</font> has_children(self): <br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<i>this call should return True if this object might have children, and False if this object can be guaranteed not to have children.</i><sup>[2]</sup><br/>
&nbsp;&nbsp;&nbsp;&nbsp;<font color=blue>def</font> get_value(self): <br/>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<i>this call can return an SBValue to be presented as the value of the synthetic value under consideration.</i><sup>[3]</sup><br/>
</code>
<sup>[1]</sup> This method is optional. Also, it may optionally choose to return a value (starting with SVN rev153061/LLDB-134). If it returns a value, and that value is <font color=blue><code>True</code></font>, LLDB will be allowed to cache the children and the children count it previously obtained, and will not return to the provider class to ask. If nothing, <font color=blue><code>None</code></font>, or anything other than <font color=blue><code>True</code></font> is returned, LLDB will discard the cached information and ask. Regardless, whenever necessary LLDB will call <code>update</code>.
<br/>
<sup>[2]</sup> This method is optional (starting with SVN rev166495/LLDB-175). While implementing it in terms of <code>num_children</code> is acceptable, implementors are encouraged to look for optimized coding alternatives whenever reasonable.
<br/>
<sup>[3]</sup> This method is optional (starting with SVN revision 219330). The SBValue you return here will most likely be a numeric type (int, float, ...) as its value bytes will be used as-if they were the value of the root SBValue proper. As a shortcut for this, you can inherit from lldb.SBSyntheticValueProvider, and just define get_value as other methods are defaulted in the superclass as returning default no-children responses.
<p>For examples of how synthetic children are created, you are encouraged to look at <a href="http://llvm.org/svn/llvm-project/lldb/trunk/examples/synthetic/">examples/synthetic</a> in the LLDB trunk. Please, be aware that the code in those files (except bitfield/)
is legacy code and is not maintained.
You may especially want to begin looking at <a href="http://llvm.org/svn/llvm-project/lldb/trunk/examples/synthetic/bitfield">this example</a> to get
a feel for this feature, as it is a very easy and well commented example.</p>
The design pattern consistently used in synthetic providers shipping with LLDB
is to use the <code>__init__</code> to store the SBValue instance as a part of <code>self</code>. The <code>update</code> function is then used
to perform the actual initialization.
<p>Once a synthetic children provider is written, one must load it into LLDB before it can be used.
Currently, one can use the LLDB <code>script</code> command to type Python code interactively,
or use the <code>command script import <i>fileName </i></code> command to load Python code from a Python module
(ordinary rules apply to importing modules this way). A third option is to type the code for
the provider class interactively while adding it.</p>
<p>For example, let's pretend we have a class <code>Foo</code> for which a synthetic children provider class
<code>Foo_Provider</code> is available, in a Python module contained in file <code>~/Foo_Tools.py</code>. The following interaction
sets <code>Foo_Provider</code> as a synthetic children provider in LLDB:</p>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> command script import ~/Foo_Tools.py<br/>
<b>(lldb)</b> type synthetic add Foo --python-class Foo_Tools.Foo_Provider
</td>
<table>
<code> <b>(lldb)</b> frame variable a_foo<br/>
(Foo) a_foo = {<br/>
&nbsp;&nbsp;&nbsp;&nbsp;x = 1<br/>
&nbsp;&nbsp;&nbsp;&nbsp;y = "Hello world"<br/>
} <br/>
</code> </p>
<p>LLDB has synthetic children providers for a core subset of STL classes, both in the version provided by <a href="http://gcc.gnu.org/libstdc++/">libstdcpp</a> and by <a href="http://libcxx.llvm.org/">libcxx</a>, as well as for several Foundation classes.</p>
<p>Synthetic children extend summary strings by enabling a new special variable: <code>${svar</code>.<br/>
This symbol tells LLDB to refer expression paths to the
synthetic children instead of the real ones. For instance,</p>
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add --expand -x "std::vector&lt;" --summary-string "${svar%#} items"
</td>
</table>
<code> <b>(lldb)</b> frame variable numbers<br/>
(std::vector&lt;int&gt;) numbers = 4 items {<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) [0] = 1<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) [1] = 12<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) [2] = 123<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) [3] = 1234<br/>
}<br/>
</code> </p>
<p>In some cases, if LLDB is unable to use the real object to get a child specified in an expression path, it will automatically refer to the
synthetic children. While in summaries it is best to always use <code>${svar</code> to make your intentions clearer, interactive debugging
can benefit from this behavior, as in:
<code> <b>(lldb)</b> frame variable numbers[0] numbers[1]<br/>
(int) numbers[0] = 1<br/>
(int) numbers[1] = 12<br/>
</code> </p>
Unlike many other visualization features, however, the access to synthetic children only works when using <code>frame variable</code>, and is
not supported in <code>expression</code>:<br/>
<code> <b>(lldb)</b> expression numbers[0]<br/>
Error [IRForTarget]: Call to a function '_ZNSt33vector&lt;int, std::allocator&lt;int&gt; &gt;ixEm' that is not present in the target<br/>
error: Couldn't convert the expression to DWARF<br/>
</code> </p>
The reason for this is that classes might have an overloaded <code><font color="blue">operator</font> []</code>, or other special provisions
and the <code>expression</code> command chooses to ignore synthetic children in the interest of equivalency with code you asked to have compiled from source.
</div>
</div>
<div class="post">
<h1 class="postheader">Filters</h1>
<div class="postcontent">
<p>Filters are a solution to the display of complex classes.
At times, classes have many member variables but not all of these are actually
necessary for the user to see.</p>
<p>A filter will solve this issue by only letting the user see those member
variables he cares about. Of course, the equivalent of a filter can be implemented easily
using synthetic children, but a filter lets you get the job done without having to write
Python code.</p>
<p>For instance, if your class <code>Foobar</code> has member variables named <code>A</code> thru <code>Z</code>, but you only need to see
the ones named <code>B</code>, <code>H</code> and <code>Q</code>, you can define a filter:
<table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type filter add Foobar --child B --child H --child Q
</td>
</table>
<code> <b>(lldb)</b> frame variable a_foobar<br/>
(Foobar) a_foobar = {<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(int) B = 1<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(char) H = 'H'<br/>
&nbsp;&nbsp;&nbsp;&nbsp;(std::string) Q = "Hello world"<br/>
}<br/>
</code> </p>
</div>
</div>
<div class="post">
<h1 class="postheader">Objective-C dynamic type discovery</h1>
<div class="postcontent">
<p>When doing Objective-C development, you may notice that some of your variables
come out as of type <code>id</code> (for instance, items extracted from <code>NSArray</code>).
By default, LLDB will not show you the real type of the object. it can actually dynamically discover the type of an Objective-C
variable, much like the runtime itself does when invoking a selector. In order
to be shown the result of that discovery that, however, a special option to <code>frame variable</code> or <code>expression</code> is
required: <br/><code>--dynamic-type</code>.</p>
<p><code>--dynamic-type</code> can have one of three values:
<ul>
<li><code>no-dynamic-values</code>: the default, prevents dynamic type discovery</li>
<li><code>no-run-target</code>: enables dynamic type discovery as long as running
code on the target is not required</li>
<li><code>run-target</code>: enables code execution on the target in order to perform
dynamic type discovery</li>
</ul>
</p>
<p>
If you specify a value of either <code>no-run-target</code> or <code>run-target</code>,
LLDB will detect the dynamic type of your variables and show the appropriate formatters
for them. As an example:
</p>
<p><table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> expr @&quot;Hello&quot;
</td>
</table>
<code>(NSString *) $0 = 0x00000001048000b0 @&quot;Hello&quot;<br/>
</code>
<p><table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> expr -d no-run @&quot;Hello&quot;
</td>
</table>
<code>(__NSCFString *) $1 = 0x00000001048000b0 @&quot;Hello&quot;<br/>
</code>
<p>
Because LLDB uses a detection algorithm that does not need to invoke any functions
on the target process, <code>no-run-target</code> is enough for this to work.</p>
As a side note, the summary for NSString shown in the example is built right into LLDB.
It was initially implemented through Python (the code is still available for reference at <a href="http://llvm.org/svn/llvm-project/lldb/trunk/examples/summaries/cocoa/CFString.py">CFString.py</a>).
However, this is out of sync with the current implementation of the NSString formatter (which is a C++ function compiled into the LLDB core).
</p>
</div>
</div>
<div class="post">
<h1 class="postheader">Categories</h1>
<div class="postcontent">
<p>Categories are a way to group related formatters. For instance, LLDB itself groups
the formatters for the libstdc++ types in a category named <code>gnu-libstdc++</code>.
Basically, categories act like containers in which to store formatters for a same library
or OS release.</p>
<p>By default, several categories are created in LLDB:
<ul>
<li><code>default</code>: this is the category where every formatter ends up, unless another category is specified
<li><code>objc</code>: formatters for basic and common Objective-C types that do not specifically depend on Mac OS X
<li><code>gnu-libstdc++</code>: formatters for std::string, std::vector, std::list and std::map as implemented by libstdcpp
<li><code>libcxx</code>: formatters for std::string, std::vector, std::list and std::map as implemented by <a href="http://libcxx.llvm.org/">libcxx</a>
<li><code>system</code>: truly basic types for which a formatter is required
<li><a href="https://developer.apple.com/library/mac/#documentation/Cocoa/Reference/Foundation/ObjC_classic/_index.html#//apple_ref/doc/uid/20001091"><code>AppKit</code></a>: Cocoa classes
<li><a href="https://developer.apple.com/corefoundation/"><code>CoreFoundation</code></a>: CF classes
<li><a href="https://developer.apple.com/library/mac/#documentation/CoreGraphics/Reference/CoreGraphicsConstantsRef/Reference/reference.html"><code>CoreGraphics</code></a>: CG classes
<li><a href="http://developer.apple.com/library/mac/#documentation/Carbon/reference/CoreServicesReferenceCollection/_index.html"><code>CoreServices</code></a>: CS classes
<li><code>VectorTypes</code>: compact display for several vector types
</ul>
If you want to use a custom category for your formatters, all the <code>type ... add</code>
provide a <code>--category</code> (<code>-w</code>) option, that names the category to add the formatter to.
To delete the formatter, you then have to specify the correct category.</p>
<p>Categories can be in one of two states: enabled and disabled. A category is initially disabled,
and can be enabled using the <code>type category enable</code> command. To disable an enabled category,
the command to use is <code>type category disable</code>.
<p>The order in which categories are enabled or disabled
is significant, in that LLDB uses that order when looking for formatters. Therefore, when you enable a category, it becomes
the second one to be searched (after <code>default</code>, which always stays on top of the list). The default categories are enabled in such a way that the search order is:
<ul>
<li>default</li>
<li>objc</li>
<li>CoreFoundation</li>
<li>AppKit</li>
<li>CoreServices</li>
<li>CoreGraphics</li>
<li>gnu-libstdc++</li>
<li>libcxx</li>
<li>VectorTypes</li>
<li>system</li>
</ul>
<p>As said, <code>gnu-libstdc++</code> and <code>libcxx</code> contain formatters for C++ STL
data types. <code>system</code> contains formatters for <code>char*</code> and <code>char[]</code>, which reflect the behavior
of older versions of LLDB which had built-in formatters for these types. Because now these are formatters, you can even
replace them with your own if so you wish.</p>
<p>There is no special command to create a category. When you place a formatter in a category, if that category does not
exist, it is automatically created. For instance,</p>
<p><table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type summary add Foobar --summary-string "a foobar" --category newcategory
</td>
</table>
automatically creates a (disabled) category named newcategory.</p>
<p>Another way to create a new (empty) category, is to enable it, as in:</p>
<p><table class="stats" width="620" cellspacing="0">
<td class="content">
<b>(lldb)</b> type category enable newcategory
</td>
</table>
<p>However, in this case LLDB warns you that enabling an empty category has no effect. If you add formatters to the
category after enabling it, they will be honored. But an empty category <i>per se</i> does not change the way any
type is displayed. The reason the debugger warns you is that enabling an empty category might be a typo, and you
effectively wanted to enable a similarly-named but not-empty category.</p>
</div>
</div>
<div class="post">
<h1 class="postheader">Finding formatters 101</h1>
<div class="postcontent">
<p>Searching for a formatter
(including formats, starting in SVN rev <a href="http://llvm.org/viewvc/llvm-project?view=revision&amp;revision=192217">r192217</a>)
given a variable goes through
a rather intricate set of rules. Namely, what happens is that LLDB
starts looking in each enabled category, according to the order in which
they were enabled (latest enabled first). In each category, LLDB does
the following:</p>
<ul>
<li>If there is a formatter for the type of the variable,
use it</li>
<li>If this object is a pointer, and there is a formatter
for the pointee type that does not skip pointers, use
it</li>
<li>If this object is a reference, and there is a
formatter for the referred type that does not skip
references, use it</li>
<li>If this object is an Objective-C class and dynamic types are enabled,
look for a formatter for the dynamic type of the object. If dynamic types are disabled,
or the search failed, look for a formatter for the declared type of the object</li>
<li>If this object's type is a typedef, go through
typedef hierarchy (LLDB might not be able to do this if
the compiler has not emitted enough information. If the
required information to traverse typedef hierarchies is
missing, type cascading will not work. The
<a href="http://clang.llvm.org/">clang compiler</a>,
part of the LLVM project, emits the correct debugging
information for LLDB to cascade). If at any level of the hierarchy
there is a valid formatter that can cascade, use it.</li>
<li>If everything has failed, repeat the above search,
looking for regular expressions instead of exact
matches</li>
</ul>
<p>If any of those attempts returned a valid formatter to be used,
that one is used, and the search is terminated (without going to look
in other categories). If nothing was found in the current category, the next
enabled category is scanned according to the same algorithm. If there are no
more enabled categories, the search has failed.</p>
<p><font color=red>Warning</font>: previous versions of LLDB defined cascading to mean
not only going through typedef chains, but also through inheritance chains.
This feature has been removed since it significantly degrades performance.
You need to set up your formatters for every type in inheritance chains to which
you want the formatter to apply.</p>
</div>
</div>
</div>
</div>
</div>
</body>
</html>