Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / clang / 39a8de10c18365bde7062d8959b7ed525449c561 / . / docs / InternalsManual.html
blob: 9250fd1dd777f0e50f2caf2a4dc9b833b6c611ed [file] [log] [blame]
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]1<html>
2<head>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]3<title>"Clang" CFE Internals Manual</title>
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]4<link type="text/css" rel="stylesheet" href="../menu.css" />
5<link type="text/css" rel="stylesheet" href="../content.css" />
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]6<style type="text/css">
7td {
8 vertical-align: top;
9}
10</style>
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]11</head>
12<body>
13
14<!--#include virtual="../menu.html.incl"-->
15
16<div id="content">
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]17
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]18<h1>"Clang" CFE Internals Manual</h1>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]19
20<ul>
21<li><a href="#intro">Introduction</a></li>
22<li><a href="#libsystem">LLVM System and Support Libraries</a></li>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]23<li><a href="#libbasic">The Clang 'Basic' Library</a>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]24 <ul>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]25 <li><a href="#Diagnostics">The Diagnostics Subsystem</a></li>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]26 <li><a href="#SourceLocation">The SourceLocation and SourceManager
27 classes</a></li>
28 </ul>
29</li>
30<li><a href="#liblex">The Lexer and Preprocessor Library</a>
31 <ul>
32 <li><a href="#Token">The Token class</a></li>
33 <li><a href="#Lexer">The Lexer class</a></li>
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]34 <li><a href="#AnnotationToken">Annotation Tokens</a></li>
Chris Lattner79281252008-03-09 02:27:26 +0000[diff] [blame]35 <li><a href="#TokenLexer">The TokenLexer class</a></li>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]36 <li><a href="#MultipleIncludeOpt">The MultipleIncludeOpt class</a></li>
37 </ul>
38</li>
39<li><a href="#libparse">The Parser Library</a>
40 <ul>
41 </ul>
42</li>
43<li><a href="#libast">The AST Library</a>
44 <ul>
45 <li><a href="#Type">The Type class and its subclasses</a></li>
46 <li><a href="#QualType">The QualType class</a></li>
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]47 <li><a href="#DeclarationName">Declaration names</a></li>
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]48 <li><a href="#DeclContext">Declaration contexts</a>
49 <ul>
50 <li><a href="#Redeclarations">Redeclarations and Overloads</a></li>
51 <li><a href="#LexicalAndSemanticContexts">Lexical and Semantic
52 Contexts</a></li>
53 <li><a href="#TransparentContexts">Transparent Declaration Contexts</a></li>
54 <li><a href="#MultiDeclContext">Multiply-Defined Declaration Contexts</a></li>
55 </ul>
56 </li>
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]57 <li><a href="#CFG">The CFG class</a></li>
Chris Lattner7bad1992008-11-16 21:48:07 +0000[diff] [blame]58 <li><a href="#Constants">Constant Folding in the Clang AST</a></li>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]59 </ul>
60</li>
61</ul>
62
63
64<!-- ======================================================================= -->
65<h2 id="intro">Introduction</h2>
66<!-- ======================================================================= -->
67
68<p>This document describes some of the more important APIs and internal design
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]69decisions made in the Clang C front-end. The purpose of this document is to
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]70both capture some of this high level information and also describe some of the
71design decisions behind it. This is meant for people interested in hacking on
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]72Clang, not for end-users. The description below is categorized by
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]73libraries, and does not describe any of the clients of the libraries.</p>
74
75<!-- ======================================================================= -->
76<h2 id="libsystem">LLVM System and Support Libraries</h2>
77<!-- ======================================================================= -->
78
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]79<p>The LLVM libsystem library provides the basic Clang system abstraction layer,
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]80which is used for file system access. The LLVM libsupport library provides many
81underlying libraries and <a
82href="http://llvm.org/docs/ProgrammersManual.html">data-structures</a>,
83 including command line option
84processing and various containers.</p>
85
86<!-- ======================================================================= -->
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]87<h2 id="libbasic">The Clang 'Basic' Library</h2>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]88<!-- ======================================================================= -->
89
90<p>This library certainly needs a better name. The 'basic' library contains a
91number of low-level utilities for tracking and manipulating source buffers,
92locations within the source buffers, diagnostics, tokens, target abstraction,
93and information about the subset of the language being compiled for.</p>
94
95<p>Part of this infrastructure is specific to C (such as the TargetInfo class),
96other parts could be reused for other non-C-based languages (SourceLocation,
97SourceManager, Diagnostics, FileManager). When and if there is future demand
98we can figure out if it makes sense to introduce a new library, move the general
99classes somewhere else, or introduce some other solution.</p>
100
101<p>We describe the roles of these classes in order of their dependencies.</p>
102
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]103
104<!-- ======================================================================= -->
105<h3 id="Diagnostics">The Diagnostics Subsystem</h3>
106<!-- ======================================================================= -->
107
108<p>The Clang Diagnostics subsystem is an important part of how the compiler
109communicates with the human. Diagnostics are the warnings and errors produced
110when the code is incorrect or dubious. In Clang, each diagnostic produced has
111(at the minimum) a unique ID, a <a href="#SourceLocation">SourceLocation</a> to
112"put the caret", an English translation associated with it, and a severity (e.g.
113<tt>WARNING</tt> or <tt>ERROR</tt>). They can also optionally include a number
114of arguments to the dianostic (which fill in "%0"'s in the string) as well as a
115number of source ranges that related to the diagnostic.</p>
116
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]117<p>In this section, we'll be giving examples produced by the Clang command line
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]118driver, but diagnostics can be <a href="#DiagnosticClient">rendered in many
119different ways</a> depending on how the DiagnosticClient interface is
120implemented. A representative example of a diagonstic is:</p>
121
122<pre>
123t.c:38:15: error: invalid operands to binary expression ('int *' and '_Complex float')
124 <font color="darkgreen">P = (P-42) + Gamma*4;</font>
125 <font color="blue">~~~~~~ ^ ~~~~~~~</font>
126</pre>
127
128<p>In this example, you can see the English translation, the severity (error),
129you can see the source location (the caret ("^") and file/line/column info),
130the source ranges "~~~~", arguments to the diagnostic ("int*" and "_Complex
131float"). You'll have to believe me that there is a unique ID backing the
132diagnostic :).</p>
133
134<p>Getting all of this to happen has several steps and involves many moving
135pieces, this section describes them and talks about best practices when adding
136a new diagnostic.</p>
137
138<!-- ============================ -->
139<h4>The DiagnosticKinds.def file</h4>
140<!-- ============================ -->
141
142<p>Diagnostics are created by adding an entry to the <tt><a
143href="http://llvm.org/svn/llvm-project/cfe/trunk/include/clang/Basic/DiagnosticKinds.def"
144>DiagnosticKinds.def</a></tt> file. This file encodes the unique ID of the
145diagnostic (as an enum, the first argument), the severity of the diagnostic
146(second argument) and the English translation + format string.</p>
147
148<p>There is little sanity with the naming of the unique ID's right now. Some
149start with err_, warn_, ext_ to encode the severity into the name. Since the
150enum is referenced in the C++ code that produces the diagnostic, it is somewhat
151useful for it to be reasonably short.</p>
152
153<p>The severity of the diagnostic comes from the set {<tt>NOTE</tt>,
154<tt>WARNING</tt>, <tt>EXTENSION</tt>, <tt>EXTWARN</tt>, <tt>ERROR</tt>}. The
155<tt>ERROR</tt> severity is used for diagnostics indicating the program is never
156acceptable under any circumstances. When an error is emitted, the AST for the
157input code may not be fully built. The <tt>EXTENSION</tt> and <tt>EXTWARN</tt>
158severities are used for extensions to the language that Clang accepts. This
159means that Clang fully understands and can represent them in the AST, but we
160produce diagnostics to tell the user their code is non-portable. The difference
161is that the former are ignored by default, and the later warn by default. The
162<tt>WARNING</tt> severity is used for constructs that are valid in the currently
163selected source language but that are dubious in some way. The <tt>NOTE</tt>
Daniel Dunbar426b8632009-02-17 15:49:03 +0000[diff] [blame]164level is used to staple more information onto previous diagnostics.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]165
166<p>These <em>severities</em> are mapped into a smaller set (the
167Diagnostic::Level enum, {<tt>Ignored</tt>, <tt>Note</tt>, <tt>Warning</tt>,
Chris Lattner0aad2972009-02-05 22:49:08 +0000[diff] [blame]168<tt>Error</tt>, <tt>Fatal</tt> }) of output <em>levels</em> by the diagnostics
Chris Lattnera180fdd2009-02-17 07:07:29 +0000[diff] [blame]169subsystem based on various configuration options. Clang internally supports a
170fully fine grained mapping mechanism that allows you to map almost any
171diagnostic to the output level that you want. The only diagnostics that cannot
172be mapped are <tt>NOTE</tt>s, which always follow the severity of the previously
173emitted diagnostic and <tt>ERROR</tt>s, which can only be mapped to
174<tt>Fatal</tt> (it is not possible to turn an error into a warning,
175for example).</p>
176
177<p>Diagnostic mappings are used in many ways. For example, if the user
178specifies <tt>-pedantic</tt>, <tt>EXTENSION</tt> maps to <tt>Warning</tt>, if
179they specify <tt>-pedantic-errors</tt>, it turns into <tt>Error</tt>. This is
180used to implement options like <tt>-Wunused_macros</tt>, <tt>-Wundef</tt> etc.
181</p>
182
183<p>
184Mapping to <tt>Fatal</tt> should only be used for diagnostics that are
185considered so severe that error recovery won't be able to recover sensibly from
186them (thus spewing a ton of bogus errors). One example of this class of error
187are failure to #include a file.
188</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]189
190<!-- ================= -->
191<h4>The Format String</h4>
192<!-- ================= -->
193
194<p>The format string for the diagnostic is very simple, but it has some power.
195It takes the form of a string in English with markers that indicate where and
196how arguments to the diagnostic are inserted and formatted. For example, here
197are some simple format strings:</p>
198
199<pre>
200 "binary integer literals are an extension"
201 "format string contains '\\0' within the string body"
202 "more '<b>%%</b>' conversions than data arguments"
Chris Lattner545b3682008-11-23 20:27:13 +0000[diff] [blame]203 "invalid operands to binary expression (<b>%0</b> and <b>%1</b>)"
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]204 "overloaded '<b>%0</b>' must be a <b>%select{unary|binary|unary or binary}2</b> operator"
205 " (has <b>%1</b> parameter<b>%s1</b>)"
206</pre>
207
208<p>These examples show some important points of format strings. You can use any
209 plain ASCII character in the diagnostic string except "%" without a problem,
210 but these are C strings, so you have to use and be aware of all the C escape
211 sequences (as in the second example). If you want to produce a "%" in the
212 output, use the "%%" escape sequence, like the third diagnostic. Finally,
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]213 Clang uses the "%...[digit]" sequences to specify where and how arguments to
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]214 the diagnostic are formatted.</p>
215
216<p>Arguments to the diagnostic are numbered according to how they are specified
217 by the C++ code that <a href="#producingdiag">produces them</a>, and are
218 referenced by <tt>%0</tt> .. <tt>%9</tt>. If you have more than 10 arguments
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]219 to your diagnostic, you are doing something wrong :). Unlike printf, there
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]220 is no requirement that arguments to the diagnostic end up in the output in
221 the same order as they are specified, you could have a format string with
222 <tt>"%1 %0"</tt> that swaps them, for example. The text in between the
223 percent and digit are formatting instructions. If there are no instructions,
224 the argument is just turned into a string and substituted in.</p>
225
226<p>Here are some "best practices" for writing the English format string:</p>
227
228<ul>
229<li>Keep the string short. It should ideally fit in the 80 column limit of the
230 <tt>DiagnosticKinds.def</tt> file. This avoids the diagnostic wrapping when
231 printed, and forces you to think about the important point you are conveying
232 with the diagnostic.</li>
233<li>Take advantage of location information. The user will be able to see the
234 line and location of the caret, so you don't need to tell them that the
235 problem is with the 4th argument to the function: just point to it.</li>
236<li>Do not capitalize the diagnostic string, and do not end it with a
237 period.</li>
238<li>If you need to quote something in the diagnostic string, use single
239 quotes.</li>
240</ul>
241
242<p>Diagnostics should never take random English strings as arguments: you
243shouldn't use <tt>"you have a problem with %0"</tt> and pass in things like
244<tt>"your argument"</tt> or <tt>"your return value"</tt> as arguments. Doing
245this prevents <a href="translation">translating</a> the Clang diagnostics to
246other languages (because they'll get random English words in their otherwise
247localized diagnostic). The exceptions to this are C/C++ language keywords
248(e.g. auto, const, mutable, etc) and C/C++ operators (<tt>/=</tt>). Note
249that things like "pointer" and "reference" are not keywords. On the other
250hand, you <em>can</em> include anything that comes from the user's source code,
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]251including variable names, types, labels, etc. The 'select' format can be
252used to achieve this sort of thing in a localizable way, see below.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]253
254<!-- ==================================== -->
255<h4>Formatting a Diagnostic Argument</a></h4>
256<!-- ==================================== -->
257
258<p>Arguments to diagnostics are fully typed internally, and come from a couple
259different classes: integers, types, names, and random strings. Depending on
260the class of the argument, it can be optionally formatted in different ways.
261This gives the DiagnosticClient information about what the argument means
262without requiring it to use a specific presentation (consider this MVC for
263Clang :).</p>
264
265<p>Here are the different diagnostic argument formats currently supported by
266Clang:</p>
267
268<table>
269<tr><td colspan="2"><b>"s" format</b></td></tr>
270<tr><td>Example:</td><td><tt>"requires %1 parameter%s1"</tt></td></tr>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]271<tr><td>Class:</td><td>Integers</td></tr>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]272<tr><td>Description:</td><td>This is a simple formatter for integers that is
273 useful when producing English diagnostics. When the integer is 1, it prints
274 as nothing. When the integer is not 1, it prints as "s". This allows some
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]275 simple grammatical forms to be to be handled correctly, and eliminates the
276 need to use gross things like <tt>"requires %1 parameter(s)"</tt>.</td></tr>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]277
278<tr><td colspan="2"><b>"select" format</b></td></tr>
279<tr><td>Example:</td><td><tt>"must be a %select{unary|binary|unary or binary}2
280 operator"</tt></td></tr>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]281<tr><td>Class:</td><td>Integers</td></tr>
Chris Lattnercc543342008-11-22 23:50:47 +0000[diff] [blame]282<tr><td>Description:</td><td>This format specifier is used to merge multiple
283 related diagnostics together into one common one, without requiring the
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]284 difference to be specified as an English string argument. Instead of
Chris Lattnercc543342008-11-22 23:50:47 +0000[diff] [blame]285 specifying the string, the diagnostic gets an integer argument and the
286 format string selects the numbered option. In this case, the "%2" value
287 must be an integer in the range [0..2]. If it is 0, it prints 'unary', if
288 it is 1 it prints 'binary' if it is 2, it prints 'unary or binary'. This
289 allows other language translations to substitute reasonable words (or entire
290 phrases) based on the semantics of the diagnostic instead of having to do
291 things textually.</td></tr>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]292
293<tr><td colspan="2"><b>"plural" format</b></td></tr>
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]294<tr><td>Example:</td><td><tt>"you have %1 %plural{1:mouse|:mice}1 connected to
295 your computer"</tt></td></tr>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]296<tr><td>Class:</td><td>Integers</td></tr>
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]297<tr><td>Description:</td><td><p>This is a formatter for complex plural forms.
298 It is designed to handle even the requirements of languages with very
299 complex plural forms, as many Baltic languages have. The argument consists
300 of a series of expression/form pairs, separated by ':', where the first form
301 whose expression evaluates to true is the result of the modifier.</p>
302 <p>An expression can be empty, in which case it is always true. See the
303 example at the top. Otherwise, it is a series of one or more numeric
304 conditions, separated by ','. If any condition matches, the expression
305 matches. Each numeric condition can take one of three forms.</p>
306 <ul>
307 <li>number: A simple decimal number matches if the argument is the same
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]308 as the number. Example: <tt>"%plural{1:mouse|:mice}4"</tt></li>
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]309 <li>range: A range in square brackets matches if the argument is within
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]310 the range. Then range is inclusive on both ends. Example:
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]311 <tt>"%plural{0:none|1:one|[2,5]:some|:many}2"</tt></li>
312 <li>modulo: A modulo operator is followed by a number, and
313 equals sign and either a number or a range. The tests are the
314 same as for plain
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]315 numbers and ranges, but the argument is taken modulo the number first.
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]316 Example: <tt>"%plural{%100=0:even hundred|%100=[1,50]:lower half|:everything
317 else}1"</tt></li>
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]318 </ul>
319 <p>The parser is very unforgiving. A syntax error, even whitespace, will
320 abort, as will a failure to match the argument against any
321 expression.</p></td></tr>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]322
Chris Lattner077bf5e2008-11-24 03:33:13 +0000[diff] [blame]323<tr><td colspan="2"><b>"objcclass" format</b></td></tr>
324<tr><td>Example:</td><td><tt>"method %objcclass0 not found"</tt></td></tr>
325<tr><td>Class:</td><td>DeclarationName</td></tr>
326<tr><td>Description:</td><td><p>This is a simple formatter that indicates the
327 DeclarationName corresponds to an Objective-C class method selector. As
328 such, it prints the selector with a leading '+'.</p></td></tr>
329
330<tr><td colspan="2"><b>"objcinstance" format</b></td></tr>
331<tr><td>Example:</td><td><tt>"method %objcinstance0 not found"</tt></td></tr>
332<tr><td>Class:</td><td>DeclarationName</td></tr>
333<tr><td>Description:</td><td><p>This is a simple formatter that indicates the
334 DeclarationName corresponds to an Objective-C instance method selector. As
335 such, it prints the selector with a leading '-'.</p></td></tr>
336
Douglas Gregor47b9a1c2009-02-04 17:27:36 +0000[diff] [blame]337<tr><td colspan="2"><b>"q" format</b></td></tr>
338<tr><td>Example:</td><td><tt>"candidate found by name lookup is %q0"</tt></td></tr>
339<tr><td>Class:</td><td>NamedDecl*</td></tr>
340<tr><td>Description</td><td><p>This formatter indicates that the fully-qualified name of the declaration should be printed, e.g., "std::vector" rather than "vector".</p></td></tr>
341
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]342</table>
343
Chris Lattnercc543342008-11-22 23:50:47 +0000[diff] [blame]344<p>It is really easy to add format specifiers to the Clang diagnostics system,
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]345but they should be discussed before they are added. If you are creating a lot
346of repetitive diagnostics and/or have an idea for a useful formatter, please
347bring it up on the cfe-dev mailing list.</p>
Chris Lattnercc543342008-11-22 23:50:47 +0000[diff] [blame]348
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]349<!-- ===================================================== -->
350<h4><a name="#producingdiag">Producing the Diagnostic</a></h4>
351<!-- ===================================================== -->
352
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]353<p>Now that you've created the diagnostic in the DiagnosticKinds.def file, you
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]354need to write the code that detects the condition in question and emits the
355new diagnostic. Various components of Clang (e.g. the preprocessor, Sema,
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]356etc) provide a helper function named "Diag". It creates a diagnostic and
357accepts the arguments, ranges, and other information that goes along with
358it.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]359
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]360<p>For example, the binary expression error comes from code like this:</p>
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]361
362<pre>
363 if (various things that are bad)
364 Diag(Loc, diag::err_typecheck_invalid_operands)
365 &lt;&lt; lex-&gt;getType() &lt;&lt; rex-&gt;getType()
366 &lt;&lt; lex-&gt;getSourceRange() &lt;&lt; rex-&gt;getSourceRange();
367</pre>
368
369<p>This shows that use of the Diag method: they take a location (a <a
370href="#SourceLocation">SourceLocation</a> object) and a diagnostic enum value
371(which matches the name from DiagnosticKinds.def). If the diagnostic takes
372arguments, they are specified with the &lt;&lt; operator: the first argument
373becomes %0, the second becomes %1, etc. The diagnostic interface allows you to
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]374specify arguments of many different types, including <tt>int</tt> and
375<tt>unsigned</tt> for integer arguments, <tt>const char*</tt> and
376<tt>std::string</tt> for string arguments, <tt>DeclarationName</tt> and
377<tt>const IdentifierInfo*</tt> for names, <tt>QualType</tt> for types, etc.
378SourceRanges are also specified with the &lt;&lt; operator, but do not have a
379specific ordering requirement.</p>
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]380
381<p>As you can see, adding and producing a diagnostic is pretty straightforward.
382The hard part is deciding exactly what you need to say to help the user, picking
383a suitable wording, and providing the information needed to format it correctly.
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]384The good news is that the call site that issues a diagnostic should be
385completely independent of how the diagnostic is formatted and in what language
386it is rendered.
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]387</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]388
389<!-- ============================================================= -->
390<h4><a name="DiagnosticClient">The DiagnosticClient Interface</a></h4>
391<!-- ============================================================= -->
392
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]393<p>Once code generates a diagnostic with all of the arguments and the rest of
394the relevant information, Clang needs to know what to do with it. As previously
395mentioned, the diagnostic machinery goes through some filtering to map a
396severity onto a diagnostic level, then (assuming the diagnostic is not mapped to
397"<tt>Ignore</tt>") it invokes an object that implements the DiagnosticClient
398interface with the information.</p>
399
400<p>It is possible to implement this interface in many different ways. For
401example, the normal Clang DiagnosticClient (named 'TextDiagnosticPrinter') turns
402the arguments into strings (according to the various formatting rules), prints
403out the file/line/column information and the string, then prints out the line of
404code, the source ranges, and the caret. However, this behavior isn't required.
405</p>
406
407<p>Another implementation of the DiagnosticClient interface is the
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]408'TextDiagnosticBuffer' class, which is used when Clang is in -verify mode.
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]409Instead of formatting and printing out the diagnostics, this implementation just
410captures and remembers the diagnostics as they fly by. Then -verify compares
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]411the list of produced diagnostics to the list of expected ones. If they disagree,
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]412it prints out its own output.
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]413</p>
414
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]415<p>There are many other possible implementations of this interface, and this is
416why we prefer diagnostics to pass down rich structured information in arguments.
417For example, an HTML output might want declaration names be linkified to where
418they come from in the source. Another example is that a GUI might let you click
419on typedefs to expand them. This application would want to pass significantly
420more information about types through to the GUI than a simple flat string. The
421interface allows this to happen.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]422
423<!-- ====================================================== -->
424<h4><a name="translation">Adding Translations to Clang</a></h4>
425<!-- ====================================================== -->
426
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]427<p>Not possible yet! Diagnostic strings should be written in UTF-8, the client
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]428can translate to the relevant code page if needed. Each translation completely
429replaces the format string for the diagnostic.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]430
431
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]432<!-- ======================================================================= -->
433<h3 id="SourceLocation">The SourceLocation and SourceManager classes</h3>
434<!-- ======================================================================= -->
435
436<p>Strangely enough, the SourceLocation class represents a location within the
437source code of the program. Important design points include:</p>
438
439<ol>
440<li>sizeof(SourceLocation) must be extremely small, as these are embedded into
441 many AST nodes and are passed around often. Currently it is 32 bits.</li>
442<li>SourceLocation must be a simple value object that can be efficiently
443 copied.</li>
444<li>We should be able to represent a source location for any byte of any input
445 file. This includes in the middle of tokens, in whitespace, in trigraphs,
446 etc.</li>
447<li>A SourceLocation must encode the current #include stack that was active when
448 the location was processed. For example, if the location corresponds to a
449 token, it should contain the set of #includes active when the token was
450 lexed. This allows us to print the #include stack for a diagnostic.</li>
451<li>SourceLocation must be able to describe macro expansions, capturing both
452 the ultimate instantiation point and the source of the original character
453 data.</li>
454</ol>
455
456<p>In practice, the SourceLocation works together with the SourceManager class
Chris Lattner18376dd2009-01-16 07:00:50 +0000[diff] [blame]457to encode two pieces of information about a location: it's spelling location
Chris Lattner88054de2009-01-16 07:15:35 +0000[diff] [blame]458and it's instantiation location. For most tokens, these will be the same. However,
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]459for a macro expansion (or tokens that came from a _Pragma directive) these will
460describe the location of the characters corresponding to the token and the
461location where the token was used (i.e. the macro instantiation point or the
462location of the _Pragma itself).</p>
463
Chris Lattner3fcbb892008-11-23 08:32:53 +0000[diff] [blame]464<p>For efficiency, we only track one level of macro instantiations: if a token was
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]465produced by multiple instantiations, we only track the source and ultimate
466destination. Though we could track the intermediate instantiation points, this
467would require extra bookkeeping and no known client would benefit substantially
468from this.</p>
469
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]470<p>The Clang front-end inherently depends on the location of a token being
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]471tracked correctly. If it is ever incorrect, the front-end may get confused and
472die. The reason for this is that the notion of the 'spelling' of a Token in
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]473Clang depends on being able to find the original input characters for the token.
Chris Lattner18376dd2009-01-16 07:00:50 +0000[diff] [blame]474This concept maps directly to the "spelling location" for the token.</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]475
476<!-- ======================================================================= -->
477<h2 id="liblex">The Lexer and Preprocessor Library</h2>
478<!-- ======================================================================= -->
479
480<p>The Lexer library contains several tightly-connected classes that are involved
481with the nasty process of lexing and preprocessing C source code. The main
482interface to this library for outside clients is the large <a
483href="#Preprocessor">Preprocessor</a> class.
484It contains the various pieces of state that are required to coherently read
485tokens out of a translation unit.</p>
486
487<p>The core interface to the Preprocessor object (once it is set up) is the
488Preprocessor::Lex method, which returns the next <a href="#Token">Token</a> from
489the preprocessor stream. There are two types of token providers that the
490preprocessor is capable of reading from: a buffer lexer (provided by the <a
491href="#Lexer">Lexer</a> class) and a buffered token stream (provided by the <a
Chris Lattner79281252008-03-09 02:27:26 +0000[diff] [blame]492href="#TokenLexer">TokenLexer</a> class).
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]493
494
495<!-- ======================================================================= -->
496<h3 id="Token">The Token class</h3>
497<!-- ======================================================================= -->
498
499<p>The Token class is used to represent a single lexed token. Tokens are
500intended to be used by the lexer/preprocess and parser libraries, but are not
501intended to live beyond them (for example, they should not live in the ASTs).<p>
502
503<p>Tokens most often live on the stack (or some other location that is efficient
504to access) as the parser is running, but occasionally do get buffered up. For
505example, macro definitions are stored as a series of tokens, and the C++
Chris Lattner3fcbb892008-11-23 08:32:53 +0000[diff] [blame]506front-end periodically needs to buffer tokens up for tentative parsing and
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]507various pieces of look-ahead. As such, the size of a Token matter. On a 32-bit
508system, sizeof(Token) is currently 16 bytes.</p>
509
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]510<p>Tokens occur in two forms: "<a href="#AnnotationToken">Annotation
511Tokens</a>" and normal tokens. Normal tokens are those returned by the lexer,
512annotation tokens represent semantic information and are produced by the parser,
513replacing normal tokens in the token stream. Normal tokens contain the
514following information:</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]515
516<ul>
517<li><b>A SourceLocation</b> - This indicates the location of the start of the
518token.</li>
519
520<li><b>A length</b> - This stores the length of the token as stored in the
521SourceBuffer. For tokens that include them, this length includes trigraphs and
522escaped newlines which are ignored by later phases of the compiler. By pointing
523into the original source buffer, it is always possible to get the original
524spelling of a token completely accurately.</li>
525
526<li><b>IdentifierInfo</b> - If a token takes the form of an identifier, and if
527identifier lookup was enabled when the token was lexed (e.g. the lexer was not
528reading in 'raw' mode) this contains a pointer to the unique hash value for the
529identifier. Because the lookup happens before keyword identification, this
530field is set even for language keywords like 'for'.</li>
531
532<li><b>TokenKind</b> - This indicates the kind of token as classified by the
533lexer. This includes things like <tt>tok::starequal</tt> (for the "*="
534operator), <tt>tok::ampamp</tt> for the "&amp;&amp;" token, and keyword values
535(e.g. <tt>tok::kw_for</tt>) for identifiers that correspond to keywords. Note
536that some tokens can be spelled multiple ways. For example, C++ supports
537"operator keywords", where things like "and" are treated exactly like the
538"&amp;&amp;" operator. In these cases, the kind value is set to
539<tt>tok::ampamp</tt>, which is good for the parser, which doesn't have to
540consider both forms. For something that cares about which form is used (e.g.
541the preprocessor 'stringize' operator) the spelling indicates the original
542form.</li>
543
544<li><b>Flags</b> - There are currently four flags tracked by the
545lexer/preprocessor system on a per-token basis:
546
547 <ol>
548 <li><b>StartOfLine</b> - This was the first token that occurred on its input
549 source line.</li>
550 <li><b>LeadingSpace</b> - There was a space character either immediately
551 before the token or transitively before the token as it was expanded
552 through a macro. The definition of this flag is very closely defined by
553 the stringizing requirements of the preprocessor.</li>
554 <li><b>DisableExpand</b> - This flag is used internally to the preprocessor to
555 represent identifier tokens which have macro expansion disabled. This
556 prevents them from being considered as candidates for macro expansion ever
557 in the future.</li>
558 <li><b>NeedsCleaning</b> - This flag is set if the original spelling for the
559 token includes a trigraph or escaped newline. Since this is uncommon,
560 many pieces of code can fast-path on tokens that did not need cleaning.
561 </p>
562 </ol>
563</li>
564</ul>
565
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]566<p>One interesting (and somewhat unusual) aspect of normal tokens is that they
567don't contain any semantic information about the lexed value. For example, if
568the token was a pp-number token, we do not represent the value of the number
569that was lexed (this is left for later pieces of code to decide). Additionally,
570the lexer library has no notion of typedef names vs variable names: both are
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]571returned as identifiers, and the parser is left to decide whether a specific
572identifier is a typedef or a variable (tracking this requires scope information
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]573among other things). The parser can do this translation by replacing tokens
574returned by the preprocessor with "Annotation Tokens".</p>
575
576<!-- ======================================================================= -->
577<h3 id="AnnotationToken">Annotation Tokens</h3>
578<!-- ======================================================================= -->
579
580<p>Annotation Tokens are tokens that are synthesized by the parser and injected
581into the preprocessor's token stream (replacing existing tokens) to record
582semantic information found by the parser. For example, if "foo" is found to be
583a typedef, the "foo" <tt>tok::identifier</tt> token is replaced with an
584<tt>tok::annot_typename</tt>. This is useful for a couple of reasons: 1) this
585makes it easy to handle qualified type names (e.g. "foo::bar::baz&lt;42&gt;::t")
586in C++ as a single "token" in the parser. 2) if the parser backtracks, the
587reparse does not need to redo semantic analysis to determine whether a token
588sequence is a variable, type, template, etc.</p>
589
590<p>Annotation Tokens are created by the parser and reinjected into the parser's
591token stream (when backtracking is enabled). Because they can only exist in
592tokens that the preprocessor-proper is done with, it doesn't need to keep around
593flags like "start of line" that the preprocessor uses to do its job.
594Additionally, an annotation token may "cover" a sequence of preprocessor tokens
595(e.g. <tt>a::b::c</tt> is five preprocessor tokens). As such, the valid fields
596of an annotation token are different than the fields for a normal token (but
597they are multiplexed into the normal Token fields):</p>
598
599<ul>
600<li><b>SourceLocation "Location"</b> - The SourceLocation for the annotation
601token indicates the first token replaced by the annotation token. In the example
602above, it would be the location of the "a" identifier.</li>
603
604<li><b>SourceLocation "AnnotationEndLoc"</b> - This holds the location of the
605last token replaced with the annotation token. In the example above, it would
606be the location of the "c" identifier.</li>
607
608<li><b>void* "AnnotationValue"</b> - This contains an opaque object that the
609parser gets from Sema through an Actions module, it is passed around and Sema
610intepretes it, based on the type of annotation token.</li>
611
612<li><b>TokenKind "Kind"</b> - This indicates the kind of Annotation token this
613is. See below for the different valid kinds.</li>
614</ul>
615
616<p>Annotation tokens currently come in three kinds:</p>
617
618<ol>
619<li><b>tok::annot_typename</b>: This annotation token represents a
620resolved typename token that is potentially qualified. The AnnotationValue
Steve Naroffb43a50f2009-01-28 19:39:02 +0000[diff] [blame]621field contains a pointer returned by Action::getTypeName(). In the case of the
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]622Sema actions module, this is a <tt>Decl*</tt> for the type.</li>
623
624<li><b>tok::annot_cxxscope</b>: This annotation token represents a C++ scope
625specifier, such as "A::B::". This corresponds to the grammar productions "::"
626and ":: [opt] nested-name-specifier". The AnnotationValue pointer is returned
627by the Action::ActOnCXXGlobalScopeSpecifier and
628Action::ActOnCXXNestedNameSpecifier callbacks. In the case of Sema, this is a
629<tt>DeclContext*</tt>.</li>
630
Douglas Gregor39a8de12009-02-25 19:37:18 +0000[diff] [blame^]631<li><b>tok::annot_template_id</b>: This annotation token represents a
632C++ template-id such as "foo&lt;int, 4&gt;", where "foo" is the name
633of a template. The AnnotationValue pointer is a pointer to a malloc'd
634TemplateIdAnnotation object. Depending on the context, a parsed template-id that names a type might become a typename annotation token (if all we care about is the named type, e.g., because it occurs in a type specifier) or might remain a template-id token (if we want to retain more source location information or produce a new type, e.g., in a declaration of a class template specialization). template-id annotation tokens that refer to a type can be "upgraded" to typename annotation tokens by the parser.</li>
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]635
636</ol>
637
Cedric Venetda76b282009-01-06 16:22:54 +0000[diff] [blame]638<p>As mentioned above, annotation tokens are not returned by the preprocessor,
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]639they are formed on demand by the parser. This means that the parser has to be
640aware of cases where an annotation could occur and form it where appropriate.
641This is somewhat similar to how the parser handles Translation Phase 6 of C99:
642String Concatenation (see C99 5.1.1.2). In the case of string concatenation,
643the preprocessor just returns distinct tok::string_literal and
644tok::wide_string_literal tokens and the parser eats a sequence of them wherever
645the grammar indicates that a string literal can occur.</p>
646
647<p>In order to do this, whenever the parser expects a tok::identifier or
648tok::coloncolon, it should call the TryAnnotateTypeOrScopeToken or
649TryAnnotateCXXScopeToken methods to form the annotation token. These methods
650will maximally form the specified annotation tokens and replace the current
651token with them, if applicable. If the current tokens is not valid for an
652annotation token, it will remain an identifier or :: token.</p>
653
654
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]655
656<!-- ======================================================================= -->
657<h3 id="Lexer">The Lexer class</h3>
658<!-- ======================================================================= -->
659
660<p>The Lexer class provides the mechanics of lexing tokens out of a source
661buffer and deciding what they mean. The Lexer is complicated by the fact that
662it operates on raw buffers that have not had spelling eliminated (this is a
663necessity to get decent performance), but this is countered with careful coding
664as well as standard performance techniques (for example, the comment handling
665code is vectorized on X86 and PowerPC hosts).</p>
666
667<p>The lexer has a couple of interesting modal features:</p>
668
669<ul>
670<li>The lexer can operate in 'raw' mode. This mode has several features that
671 make it possible to quickly lex the file (e.g. it stops identifier lookup,
672 doesn't specially handle preprocessor tokens, handles EOF differently, etc).
673 This mode is used for lexing within an "<tt>#if 0</tt>" block, for
674 example.</li>
675<li>The lexer can capture and return comments as tokens. This is required to
676 support the -C preprocessor mode, which passes comments through, and is
677 used by the diagnostic checker to identifier expect-error annotations.</li>
678<li>The lexer can be in ParsingFilename mode, which happens when preprocessing
Chris Lattner84386242007-09-16 19:25:23 +0000[diff] [blame]679 after reading a #include directive. This mode changes the parsing of '&lt;'
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]680 to return an "angled string" instead of a bunch of tokens for each thing
681 within the filename.</li>
682<li>When parsing a preprocessor directive (after "<tt>#</tt>") the
683 ParsingPreprocessorDirective mode is entered. This changes the parser to
684 return EOM at a newline.</li>
685<li>The Lexer uses a LangOptions object to know whether trigraphs are enabled,
686 whether C++ or ObjC keywords are recognized, etc.</li>
687</ul>
688
689<p>In addition to these modes, the lexer keeps track of a couple of other
690 features that are local to a lexed buffer, which change as the buffer is
691 lexed:</p>
692
693<ul>
694<li>The Lexer uses BufferPtr to keep track of the current character being
695 lexed.</li>
696<li>The Lexer uses IsAtStartOfLine to keep track of whether the next lexed token
697 will start with its "start of line" bit set.</li>
698<li>The Lexer keeps track of the current #if directives that are active (which
699 can be nested).</li>
700<li>The Lexer keeps track of an <a href="#MultipleIncludeOpt">
701 MultipleIncludeOpt</a> object, which is used to
702 detect whether the buffer uses the standard "<tt>#ifndef XX</tt> /
703 <tt>#define XX</tt>" idiom to prevent multiple inclusion. If a buffer does,
704 subsequent includes can be ignored if the XX macro is defined.</li>
705</ul>
706
707<!-- ======================================================================= -->
Chris Lattner79281252008-03-09 02:27:26 +0000[diff] [blame]708<h3 id="TokenLexer">The TokenLexer class</h3>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]709<!-- ======================================================================= -->
710
Chris Lattner79281252008-03-09 02:27:26 +0000[diff] [blame]711<p>The TokenLexer class is a token provider that returns tokens from a list
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]712of tokens that came from somewhere else. It typically used for two things: 1)
713returning tokens from a macro definition as it is being expanded 2) returning
714tokens from an arbitrary buffer of tokens. The later use is used by _Pragma and
715will most likely be used to handle unbounded look-ahead for the C++ parser.</p>
716
717<!-- ======================================================================= -->
718<h3 id="MultipleIncludeOpt">The MultipleIncludeOpt class</h3>
719<!-- ======================================================================= -->
720
721<p>The MultipleIncludeOpt class implements a really simple little state machine
722that is used to detect the standard "<tt>#ifndef XX</tt> / <tt>#define XX</tt>"
723idiom that people typically use to prevent multiple inclusion of headers. If a
724buffer uses this idiom and is subsequently #include'd, the preprocessor can
725simply check to see whether the guarding condition is defined or not. If so,
726the preprocessor can completely ignore the include of the header.</p>
727
728
729
730<!-- ======================================================================= -->
731<h2 id="libparse">The Parser Library</h2>
732<!-- ======================================================================= -->
733
734<!-- ======================================================================= -->
735<h2 id="libast">The AST Library</h2>
736<!-- ======================================================================= -->
737
738<!-- ======================================================================= -->
739<h3 id="Type">The Type class and its subclasses</h3>
740<!-- ======================================================================= -->
741
742<p>The Type class (and its subclasses) are an important part of the AST. Types
743are accessed through the ASTContext class, which implicitly creates and uniques
744them as they are needed. Types have a couple of non-obvious features: 1) they
745do not capture type qualifiers like const or volatile (See
746<a href="#QualType">QualType</a>), and 2) they implicitly capture typedef
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]747information. Once created, types are immutable (unlike decls).</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]748
749<p>Typedefs in C make semantic analysis a bit more complex than it would
750be without them. The issue is that we want to capture typedef information
751and represent it in the AST perfectly, but the semantics of operations need to
752"see through" typedefs. For example, consider this code:</p>
753
754<code>
755void func() {<br>
Bill Wendling30d17752007-10-06 01:56:01 +0000[diff] [blame]756&nbsp;&nbsp;typedef int foo;<br>
757&nbsp;&nbsp;foo X, *Y;<br>
758&nbsp;&nbsp;typedef foo* bar;<br>
759&nbsp;&nbsp;bar Z;<br>
760&nbsp;&nbsp;*X; <i>// error</i><br>
761&nbsp;&nbsp;**Y; <i>// error</i><br>
762&nbsp;&nbsp;**Z; <i>// error</i><br>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]763}<br>
764</code>
765
766<p>The code above is illegal, and thus we expect there to be diagnostics emitted
767on the annotated lines. In this example, we expect to get:</p>
768
769<pre>
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]770<b>test.c:6:1: error: indirection requires pointer operand ('foo' invalid)</b>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]771*X; // error
772<font color="blue">^~</font>
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]773<b>test.c:7:1: error: indirection requires pointer operand ('foo' invalid)</b>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]774**Y; // error
775<font color="blue">^~~</font>
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]776<b>test.c:8:1: error: indirection requires pointer operand ('foo' invalid)</b>
777**Z; // error
778<font color="blue">^~~</font>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]779</pre>
780
781<p>While this example is somewhat silly, it illustrates the point: we want to
782retain typedef information where possible, so that we can emit errors about
783"<tt>std::string</tt>" instead of "<tt>std::basic_string&lt;char, std:...</tt>".
784Doing this requires properly keeping typedef information (for example, the type
785of "X" is "foo", not "int"), and requires properly propagating it through the
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]786various operators (for example, the type of *Y is "foo", not "int"). In order
787to retain this information, the type of these expressions is an instance of the
788TypedefType class, which indicates that the type of these expressions is a
789typedef for foo.
790</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]791
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]792<p>Representing types like this is great for diagnostics, because the
793user-specified type is always immediately available. There are two problems
794with this: first, various semantic checks need to make judgements about the
Chris Lattner33fc68a2007-07-31 18:54:50 +0000[diff] [blame]795<em>actual structure</em> of a type, ignoring typdefs. Second, we need an
796efficient way to query whether two types are structurally identical to each
797other, ignoring typedefs. The solution to both of these problems is the idea of
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]798canonical types.</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]799
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]800<!-- =============== -->
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]801<h4>Canonical Types</h4>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]802<!-- =============== -->
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]803
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]804<p>Every instance of the Type class contains a canonical type pointer. For
805simple types with no typedefs involved (e.g. "<tt>int</tt>", "<tt>int*</tt>",
806"<tt>int**</tt>"), the type just points to itself. For types that have a
807typedef somewhere in their structure (e.g. "<tt>foo</tt>", "<tt>foo*</tt>",
808"<tt>foo**</tt>", "<tt>bar</tt>"), the canonical type pointer points to their
809structurally equivalent type without any typedefs (e.g. "<tt>int</tt>",
810"<tt>int*</tt>", "<tt>int**</tt>", and "<tt>int*</tt>" respectively).</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]811
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]812<p>This design provides a constant time operation (dereferencing the canonical
813type pointer) that gives us access to the structure of types. For example,
814we can trivially tell that "bar" and "foo*" are the same type by dereferencing
815their canonical type pointers and doing a pointer comparison (they both point
816to the single "<tt>int*</tt>" type).</p>
817
818<p>Canonical types and typedef types bring up some complexities that must be
819carefully managed. Specifically, the "isa/cast/dyncast" operators generally
820shouldn't be used in code that is inspecting the AST. For example, when type
821checking the indirection operator (unary '*' on a pointer), the type checker
822must verify that the operand has a pointer type. It would not be correct to
823check that with "<tt>isa&lt;PointerType&gt;(SubExpr-&gt;getType())</tt>",
824because this predicate would fail if the subexpression had a typedef type.</p>
825
826<p>The solution to this problem are a set of helper methods on Type, used to
827check their properties. In this case, it would be correct to use
828"<tt>SubExpr-&gt;getType()-&gt;isPointerType()</tt>" to do the check. This
829predicate will return true if the <em>canonical type is a pointer</em>, which is
830true any time the type is structurally a pointer type. The only hard part here
831is remembering not to use the <tt>isa/cast/dyncast</tt> operations.</p>
832
833<p>The second problem we face is how to get access to the pointer type once we
834know it exists. To continue the example, the result type of the indirection
835operator is the pointee type of the subexpression. In order to determine the
836type, we need to get the instance of PointerType that best captures the typedef
837information in the program. If the type of the expression is literally a
838PointerType, we can return that, otherwise we have to dig through the
839typedefs to find the pointer type. For example, if the subexpression had type
840"<tt>foo*</tt>", we could return that type as the result. If the subexpression
841had type "<tt>bar</tt>", we want to return "<tt>foo*</tt>" (note that we do
842<em>not</em> want "<tt>int*</tt>"). In order to provide all of this, Type has
Chris Lattner11406c12007-07-31 16:50:51 +0000[diff] [blame]843a getAsPointerType() method that checks whether the type is structurally a
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]844PointerType and, if so, returns the best one. If not, it returns a null
845pointer.</p>
846
847<p>This structure is somewhat mystical, but after meditating on it, it will
848make sense to you :).</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]849
850<!-- ======================================================================= -->
851<h3 id="QualType">The QualType class</h3>
852<!-- ======================================================================= -->
853
854<p>The QualType class is designed as a trivial value class that is small,
855passed by-value and is efficient to query. The idea of QualType is that it
856stores the type qualifiers (const, volatile, restrict) separately from the types
857themselves: QualType is conceptually a pair of "Type*" and bits for the type
858qualifiers.</p>
859
860<p>By storing the type qualifiers as bits in the conceptual pair, it is
861extremely efficient to get the set of qualifiers on a QualType (just return the
862field of the pair), add a type qualifier (which is a trivial constant-time
863operation that sets a bit), and remove one or more type qualifiers (just return
864a QualType with the bitfield set to empty).</p>
865
866<p>Further, because the bits are stored outside of the type itself, we do not
867need to create duplicates of types with different sets of qualifiers (i.e. there
868is only a single heap allocated "int" type: "const int" and "volatile const int"
869both point to the same heap allocated "int" type). This reduces the heap size
870used to represent bits and also means we do not have to consider qualifiers when
871uniquing types (<a href="#Type">Type</a> does not even contain qualifiers).</p>
872
873<p>In practice, on hosts where it is safe, the 3 type qualifiers are stored in
874the low bit of the pointer to the Type object. This means that QualType is
875exactly the same size as a pointer, and this works fine on any system where
876malloc'd objects are at least 8 byte aligned.</p>
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]877
878<!-- ======================================================================= -->
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]879<h3 id="DeclarationName">Declaration names</h3>
880<!-- ======================================================================= -->
881
882<p>The <tt>DeclarationName</tt> class represents the name of a
883 declaration in Clang. Declarations in the C family of languages can
Chris Lattner3fcbb892008-11-23 08:32:53 +0000[diff] [blame]884 take several different forms. Most declarations are named by
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]885 simple identifiers, e.g., "<code>f</code>" and "<code>x</code>" in
886 the function declaration <code>f(int x)</code>. In C++, declaration
887 names can also name class constructors ("<code>Class</code>"
888 in <code>struct Class { Class(); }</code>), class destructors
889 ("<code>~Class</code>"), overloaded operator names ("operator+"),
890 and conversion functions ("<code>operator void const *</code>"). In
891 Objective-C, declaration names can refer to the names of Objective-C
892 methods, which involve the method name and the parameters,
Chris Lattner3fcbb892008-11-23 08:32:53 +0000[diff] [blame]893 collectively called a <i>selector</i>, e.g.,
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]894 "<code>setWidth:height:</code>". Since all of these kinds of
Chris Lattner3fcbb892008-11-23 08:32:53 +0000[diff] [blame]895 entities - variables, functions, Objective-C methods, C++
896 constructors, destructors, and operators - are represented as
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]897 subclasses of Clang's common <code>NamedDecl</code>
898 class, <code>DeclarationName</code> is designed to efficiently
899 represent any kind of name.</p>
900
901<p>Given
902 a <code>DeclarationName</code> <code>N</code>, <code>N.getNameKind()</code>
Douglas Gregor2def4832008-11-17 20:34:05 +0000[diff] [blame]903 will produce a value that describes what kind of name <code>N</code>
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]904 stores. There are 8 options (all of the names are inside
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]905 the <code>DeclarationName</code> class)</p>
906<dl>
907 <dt>Identifier</dt>
908 <dd>The name is a simple
909 identifier. Use <code>N.getAsIdentifierInfo()</code> to retrieve the
910 corresponding <code>IdentifierInfo*</code> pointing to the actual
911 identifier. Note that C++ overloaded operators (e.g.,
912 "<code>operator+</code>") are represented as special kinds of
913 identifiers. Use <code>IdentifierInfo</code>'s <code>getOverloadedOperatorID</code>
914 function to determine whether an identifier is an overloaded
915 operator name.</dd>
916
917 <dt>ObjCZeroArgSelector, ObjCOneArgSelector,
918 ObjCMultiArgSelector</dt>
919 <dd>The name is an Objective-C selector, which can be retrieved as a
920 <code>Selector</code> instance
921 via <code>N.getObjCSelector()</code>. The three possible name
922 kinds for Objective-C reflect an optimization within
923 the <code>DeclarationName</code> class: both zero- and
924 one-argument selectors are stored as a
925 masked <code>IdentifierInfo</code> pointer, and therefore require
926 very little space, since zero- and one-argument selectors are far
927 more common than multi-argument selectors (which use a different
928 structure).</dd>
929
930 <dt>CXXConstructorName</dt>
931 <dd>The name is a C++ constructor
932 name. Use <code>N.getCXXNameType()</code> to retrieve
933 the <a href="#QualType">type</a> that this constructor is meant to
934 construct. The type is always the canonical type, since all
935 constructors for a given type have the same name.</dd>
936
937 <dt>CXXDestructorName</dt>
938 <dd>The name is a C++ destructor
939 name. Use <code>N.getCXXNameType()</code> to retrieve
940 the <a href="#QualType">type</a> whose destructor is being
941 named. This type is always a canonical type.</dd>
942
943 <dt>CXXConversionFunctionName</dt>
944 <dd>The name is a C++ conversion function. Conversion functions are
945 named according to the type they convert to, e.g., "<code>operator void
946 const *</code>". Use <code>N.getCXXNameType()</code> to retrieve
947 the type that this conversion function converts to. This type is
948 always a canonical type.</dd>
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]949
950 <dt>CXXOperatorName</dt>
951 <dd>The name is a C++ overloaded operator name. Overloaded operators
952 are named according to their spelling, e.g.,
953 "<code>operator+</code>" or "<code>operator new
954 []</code>". Use <code>N.getCXXOverloadedOperator()</code> to
955 retrieve the overloaded operator (a value of
956 type <code>OverloadedOperatorKind</code>).</dd>
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]957</dl>
958
959<p><code>DeclarationName</code>s are cheap to create, copy, and
960 compare. They require only a single pointer's worth of storage in
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]961 the common cases (identifiers, zero-
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]962 and one-argument Objective-C selectors) and use dense, uniqued
963 storage for the other kinds of
964 names. Two <code>DeclarationName</code>s can be compared for
965 equality (<code>==</code>, <code>!=</code>) using a simple bitwise
966 comparison, can be ordered
967 with <code>&lt;</code>, <code>&gt;</code>, <code>&lt;=</code>,
968 and <code>&gt;=</code> (which provide a lexicographical ordering for
969 normal identifiers but an unspecified ordering for other kinds of
970 names), and can be placed into LLVM <code>DenseMap</code>s
971 and <code>DenseSet</code>s.</p>
972
973<p><code>DeclarationName</code> instances can be created in different
974 ways depending on what kind of name the instance will store. Normal
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]975 identifiers (<code>IdentifierInfo</code> pointers) and Objective-C selectors
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]976 (<code>Selector</code>) can be implicitly converted
977 to <code>DeclarationName</code>s. Names for C++ constructors,
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]978 destructors, conversion functions, and overloaded operators can be retrieved from
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]979 the <code>DeclarationNameTable</code>, an instance of which is
980 available as <code>ASTContext::DeclarationNames</code>. The member
981 functions <code>getCXXConstructorName</code>, <code>getCXXDestructorName</code>,
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]982 <code>getCXXConversionFunctionName</code>, and <code>getCXXOperatorName</code>, respectively,
983 return <code>DeclarationName</code> instances for the four kinds of
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]984 C++ special function names.</p>
985
986<!-- ======================================================================= -->
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]987<h3 id="DeclContext">Declaration contexts</h3>
988<!-- ======================================================================= -->
989<p>Every declaration in a program exists within some <i>declaration
990 context</i>, such as a translation unit, namespace, class, or
991 function. Declaration contexts in Clang are represented by
992 the <code>DeclContext</code> class, from which the various
993 declaration-context AST nodes
994 (<code>TranslationUnitDecl</code>, <code>NamespaceDecl</code>, <code>RecordDecl</code>, <code>FunctionDecl</code>,
995 etc.) will derive. The <code>DeclContext</code> class provides
996 several facilities common to each declaration context:</p>
997<dl>
998 <dt>Source-centric vs. Semantics-centric View of Declarations</dt>
999 <dd><code>DeclContext</code> provides two views of the declarations
1000 stored within a declaration context. The source-centric view
1001 accurately represents the program source code as written, including
1002 multiple declarations of entities where present (see the
1003 section <a href="#Redeclarations">Redeclarations and
1004 Overloads</a>), while the semantics-centric view represents the
1005 program semantics. The two views are kept synchronized by semantic
1006 analysis while the ASTs are being constructed.</dd>
1007
1008 <dt>Storage of declarations within that context</dt>
1009 <dd>Every declaration context can contain some number of
1010 declarations. For example, a C++ class (represented
1011 by <code>RecordDecl</code>) contains various member functions,
1012 fields, nested types, and so on. All of these declarations will be
1013 stored within the <code>DeclContext</code>, and one can iterate
1014 over the declarations via
1015 [<code>DeclContext::decls_begin()</code>,
1016 <code>DeclContext::decls_end()</code>). This mechanism provides
1017 the source-centric view of declarations in the context.</dd>
1018
1019 <dt>Lookup of declarations within that context</dt>
1020 <dd>The <code>DeclContext</code> structure provides efficient name
1021 lookup for names within that declaration context. For example,
1022 if <code>N</code> is a namespace we can look for the
1023 name <code>N::f</code>
1024 using <code>DeclContext::lookup</code>. The lookup itself is
1025 based on a lazily-constructed array (for declaration contexts
1026 with a small number of declarations) or hash table (for
1027 declaration contexts with more declarations). The lookup
1028 operation provides the semantics-centric view of the declarations
1029 in the context.</dd>
1030
1031 <dt>Ownership of declarations</dt>
1032 <dd>The <code>DeclContext</code> owns all of the declarations that
1033 were declared within its declaration context, and is responsible
1034 for the management of their memory as well as their
1035 (de-)serialization.</dd>
1036</dl>
1037
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1038<p>All declarations are stored within a declaration context, and one
1039 can query
1040 information about the context in which each declaration lives. One
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1041 can retrieve the <code>DeclContext</code> that contains a
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1042 particular <code>Decl</code>
1043 using <code>Decl::getDeclContext</code>. However, see the
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1044 section <a href="#LexicalAndSemanticContexts">Lexical and Semantic
1045 Contexts</a> for more information about how to interpret this
1046 context information.</p>
1047
1048<h4 id="Redeclarations">Redeclarations and Overloads</h4>
1049<p>Within a translation unit, it is common for an entity to be
1050declared several times. For example, we might declare a function "f"
1051 and then later re-declare it as part of an inlined definition:</p>
1052
1053<pre>
1054void f(int x, int y, int z = 1);
1055
1056inline void f(int x, int y, int z) { /* ... */ }
1057</pre>
1058
1059<p>The representation of "f" differs in the source-centric and
1060 semantics-centric views of a declaration context. In the
1061 source-centric view, all redeclarations will be present, in the
1062 order they occurred in the source code, making
1063 this view suitable for clients that wish to see the structure of
1064 the source code. In the semantics-centric view, only the most recent "f"
1065 will be found by the lookup, since it effectively replaces the first
1066 declaration of "f".</p>
1067
1068<p>In the semantics-centric view, overloading of functions is
1069 represented explicitly. For example, given two declarations of a
1070 function "g" that are overloaded, e.g.,</p>
1071<pre>
1072void g();
1073void g(int);
1074</pre>
1075<p>the <code>DeclContext::lookup</code> operation will return
1076 an <code>OverloadedFunctionDecl</code> that contains both
1077 declarations of "g". Clients that perform semantic analysis on a
1078 program that is not concerned with the actual source code will
1079 primarily use this semantics-centric view.</p>
1080
1081<h4 id="LexicalAndSemanticContexts">Lexical and Semantic Contexts</h4>
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1082<p>Each declaration has two potentially different
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1083 declaration contexts: a <i>lexical</i> context, which corresponds to
1084 the source-centric view of the declaration context, and
1085 a <i>semantic</i> context, which corresponds to the
1086 semantics-centric view. The lexical context is accessible
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1087 via <code>Decl::getLexicalDeclContext</code> while the
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1088 semantic context is accessible
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1089 via <code>Decl::getDeclContext</code>, both of which return
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1090 <code>DeclContext</code> pointers. For most declarations, the two
1091 contexts are identical. For example:</p>
1092
1093<pre>
1094class X {
1095public:
1096 void f(int x);
1097};
1098</pre>
1099
1100<p>Here, the semantic and lexical contexts of <code>X::f</code> are
1101 the <code>DeclContext</code> associated with the
1102 class <code>X</code> (itself stored as a <code>RecordDecl</code> AST
1103 node). However, we can now define <code>X::f</code> out-of-line:</p>
1104
1105<pre>
1106void X::f(int x = 17) { /* ... */ }
1107</pre>
1108
1109<p>This definition of has different lexical and semantic
1110 contexts. The lexical context corresponds to the declaration
1111 context in which the actual declaration occurred in the source
1112 code, e.g., the translation unit containing <code>X</code>. Thus,
1113 this declaration of <code>X::f</code> can be found by traversing
1114 the declarations provided by
1115 [<code>decls_begin()</code>, <code>decls_end()</code>) in the
1116 translation unit.</p>
1117
1118<p>The semantic context of <code>X::f</code> corresponds to the
1119 class <code>X</code>, since this member function is (semantically) a
1120 member of <code>X</code>. Lookup of the name <code>f</code> into
1121 the <code>DeclContext</code> associated with <code>X</code> will
1122 then return the definition of <code>X::f</code> (including
1123 information about the default argument).</p>
1124
1125<h4 id="TransparentContexts">Transparent Declaration Contexts</h4>
1126<p>In C and C++, there are several contexts in which names that are
1127 logically declared inside another declaration will actually "leak"
1128 out into the enclosing scope from the perspective of name
1129 lookup. The most obvious instance of this behavior is in
1130 enumeration types, e.g.,</p>
1131<pre>
1132enum Color {
1133 Red,
1134 Green,
1135 Blue
1136};
1137</pre>
1138
1139<p>Here, <code>Color</code> is an enumeration, which is a declaration
1140 context that contains the
1141 enumerators <code>Red</code>, <code>Green</code>,
1142 and <code>Blue</code>. Thus, traversing the list of declarations
1143 contained in the enumeration <code>Color</code> will
1144 yield <code>Red</code>, <code>Green</code>,
1145 and <code>Blue</code>. However, outside of the scope
1146 of <code>Color</code> one can name the enumerator <code>Red</code>
1147 without qualifying the name, e.g.,</p>
1148
1149<pre>
1150Color c = Red;
1151</pre>
1152
1153<p>There are other entities in C++ that provide similar behavior. For
1154 example, linkage specifications that use curly braces:</p>
1155
1156<pre>
1157extern "C" {
1158 void f(int);
1159 void g(int);
1160}
1161// f and g are visible here
1162</pre>
1163
1164<p>For source-level accuracy, we treat the linkage specification and
1165 enumeration type as a
1166 declaration context in which its enclosed declarations ("Red",
1167 "Green", and "Blue"; "f" and "g")
1168 are declared. However, these declarations are visible outside of the
1169 scope of the declaration context.</p>
1170
1171<p>These language features (and several others, described below) have
1172 roughly the same set of
1173 requirements: declarations are declared within a particular lexical
1174 context, but the declarations are also found via name lookup in
1175 scopes enclosing the declaration itself. This feature is implemented
1176 via <i>transparent</i> declaration contexts
1177 (see <code>DeclContext::isTransparentContext()</code>), whose
1178 declarations are visible in the nearest enclosing non-transparent
1179 declaration context. This means that the lexical context of the
1180 declaration (e.g., an enumerator) will be the
1181 transparent <code>DeclContext</code> itself, as will the semantic
1182 context, but the declaration will be visible in every outer context
1183 up to and including the first non-transparent declaration context (since
1184 transparent declaration contexts can be nested).</p>
1185
1186<p>The transparent <code>DeclContexts</code> are:</p>
1187<ul>
1188 <li>Enumerations (but not C++0x "scoped enumerations"):
1189 <pre>
1190enum Color {
1191 Red,
1192 Green,
1193 Blue
1194};
1195// Red, Green, and Blue are in scope
1196 </pre></li>
1197 <li>C++ linkage specifications:
1198 <pre>
1199extern "C" {
1200 void f(int);
1201 void g(int);
1202}
1203// f and g are in scope
1204 </pre></li>
1205 <li>Anonymous unions and structs:
1206 <pre>
1207struct LookupTable {
1208 bool IsVector;
1209 union {
1210 std::vector&lt;Item&gt; *Vector;
1211 std::set&lt;Item&gt; *Set;
1212 };
1213};
1214
1215LookupTable LT;
1216LT.Vector = 0; // Okay: finds Vector inside the unnamed union
1217 </pre>
1218 </li>
1219 <li>C++0x inline namespaces:
1220<pre>
1221namespace mylib {
1222 inline namespace debug {
1223 class X;
1224 }
1225}
1226mylib::X *xp; // okay: mylib::X refers to mylib::debug::X
1227</pre>
1228</li>
1229</ul>
1230
1231
1232<h4 id="MultiDeclContext">Multiply-Defined Declaration Contexts</h4>
1233<p>C++ namespaces have the interesting--and, so far, unique--property that
1234the namespace can be defined multiple times, and the declarations
1235provided by each namespace definition are effectively merged (from
1236the semantic point of view). For example, the following two code
1237snippets are semantically indistinguishable:</p>
1238<pre>
1239// Snippet #1:
1240namespace N {
1241 void f();
1242}
1243namespace N {
1244 void f(int);
1245}
1246
1247// Snippet #2:
1248namespace N {
1249 void f();
1250 void f(int);
1251}
1252</pre>
1253
1254<p>In Clang's representation, the source-centric view of declaration
1255 contexts will actually have two separate <code>NamespaceDecl</code>
1256 nodes in Snippet #1, each of which is a declaration context that
1257 contains a single declaration of "f". However, the semantics-centric
1258 view provided by name lookup into the namespace <code>N</code> for
1259 "f" will return an <code>OverloadedFunctionDecl</code> that contains
1260 both declarations of "f".</p>
1261
1262<p><code>DeclContext</code> manages multiply-defined declaration
1263 contexts internally. The
1264 function <code>DeclContext::getPrimaryContext</code> retrieves the
1265 "primary" context for a given <code>DeclContext</code> instance,
1266 which is the <code>DeclContext</code> responsible for maintaining
1267 the lookup table used for the semantics-centric view. Given the
1268 primary context, one can follow the chain
1269 of <code>DeclContext</code> nodes that define additional
1270 declarations via <code>DeclContext::getNextContext</code>. Note that
1271 these functions are used internally within the lookup and insertion
1272 methods of the <code>DeclContext</code>, so the vast majority of
1273 clients can ignore them.</p>
1274
1275<!-- ======================================================================= -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1276<h3 id="CFG">The <tt>CFG</tt> class</h3>
1277<!-- ======================================================================= -->
1278
1279<p>The <tt>CFG</tt> class is designed to represent a source-level
1280control-flow graph for a single statement (<tt>Stmt*</tt>). Typically
1281instances of <tt>CFG</tt> are constructed for function bodies (usually
1282an instance of <tt>CompoundStmt</tt>), but can also be instantiated to
1283represent the control-flow of any class that subclasses <tt>Stmt</tt>,
1284which includes simple expressions. Control-flow graphs are especially
1285useful for performing
1286<a href="http://en.wikipedia.org/wiki/Data_flow_analysis#Sensitivities">flow-
1287or path-sensitive</a> program analyses on a given function.</p>
1288
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1289<!-- ============ -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1290<h4>Basic Blocks</h4>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1291<!-- ============ -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1292
1293<p>Concretely, an instance of <tt>CFG</tt> is a collection of basic
1294blocks. Each basic block is an instance of <tt>CFGBlock</tt>, which
1295simply contains an ordered sequence of <tt>Stmt*</tt> (each referring
1296to statements in the AST). The ordering of statements within a block
1297indicates unconditional flow of control from one statement to the
1298next. <a href="#ConditionalControlFlow">Conditional control-flow</a>
1299is represented using edges between basic blocks. The statements
1300within a given <tt>CFGBlock</tt> can be traversed using
1301the <tt>CFGBlock::*iterator</tt> interface.</p>
1302
1303<p>
Ted Kremenek18e17e72007-10-18 22:50:52 +0000[diff] [blame]1304A <tt>CFG</tt> object owns the instances of <tt>CFGBlock</tt> within
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1305the control-flow graph it represents. Each <tt>CFGBlock</tt> within a
1306CFG is also uniquely numbered (accessible
1307via <tt>CFGBlock::getBlockID()</tt>). Currently the number is
1308based on the ordering the blocks were created, but no assumptions
1309should be made on how <tt>CFGBlock</tt>s are numbered other than their
1310numbers are unique and that they are numbered from 0..N-1 (where N is
1311the number of basic blocks in the CFG).</p>
1312
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1313<!-- ===================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1314<h4>Entry and Exit Blocks</h4>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1315<!-- ===================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1316
1317Each instance of <tt>CFG</tt> contains two special blocks:
1318an <i>entry</i> block (accessible via <tt>CFG::getEntry()</tt>), which
1319has no incoming edges, and an <i>exit</i> block (accessible
1320via <tt>CFG::getExit()</tt>), which has no outgoing edges. Neither
1321block contains any statements, and they serve the role of providing a
1322clear entrance and exit for a body of code such as a function body.
1323The presence of these empty blocks greatly simplifies the
1324implementation of many analyses built on top of CFGs.
1325
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1326<!-- ===================================================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1327<h4 id ="ConditionalControlFlow">Conditional Control-Flow</h4>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1328<!-- ===================================================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1329
1330<p>Conditional control-flow (such as those induced by if-statements
1331and loops) is represented as edges between <tt>CFGBlock</tt>s.
1332Because different C language constructs can induce control-flow,
1333each <tt>CFGBlock</tt> also records an extra <tt>Stmt*</tt> that
1334represents the <i>terminator</i> of the block. A terminator is simply
1335the statement that caused the control-flow, and is used to identify
1336the nature of the conditional control-flow between blocks. For
1337example, in the case of an if-statement, the terminator refers to
1338the <tt>IfStmt</tt> object in the AST that represented the given
1339branch.</p>
1340
1341<p>To illustrate, consider the following code example:</p>
1342
1343<code>
1344int foo(int x) {<br>
1345&nbsp;&nbsp;x = x + 1;<br>
1346<br>
1347&nbsp;&nbsp;if (x > 2) x++;<br>
1348&nbsp;&nbsp;else {<br>
1349&nbsp;&nbsp;&nbsp;&nbsp;x += 2;<br>
1350&nbsp;&nbsp;&nbsp;&nbsp;x *= 2;<br>
1351&nbsp;&nbsp;}<br>
1352<br>
1353&nbsp;&nbsp;return x;<br>
1354}
1355</code>
1356
1357<p>After invoking the parser+semantic analyzer on this code fragment,
1358the AST of the body of <tt>foo</tt> is referenced by a
1359single <tt>Stmt*</tt>. We can then construct an instance
1360of <tt>CFG</tt> representing the control-flow graph of this function
1361body by single call to a static class method:</p>
1362
1363<code>
1364&nbsp;&nbsp;Stmt* FooBody = ...<br>
1365&nbsp;&nbsp;CFG* FooCFG = <b>CFG::buildCFG</b>(FooBody);
1366</code>
1367
1368<p>It is the responsibility of the caller of <tt>CFG::buildCFG</tt>
1369to <tt>delete</tt> the returned <tt>CFG*</tt> when the CFG is no
1370longer needed.</p>
1371
1372<p>Along with providing an interface to iterate over
1373its <tt>CFGBlock</tt>s, the <tt>CFG</tt> class also provides methods
1374that are useful for debugging and visualizing CFGs. For example, the
1375method
1376<tt>CFG::dump()</tt> dumps a pretty-printed version of the CFG to
1377standard error. This is especially useful when one is using a
1378debugger such as gdb. For example, here is the output
1379of <tt>FooCFG->dump()</tt>:</p>
1380
1381<code>
1382&nbsp;[ B5 (ENTRY) ]<br>
1383&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (0):<br>
1384&nbsp;&nbsp;&nbsp;&nbsp;Successors (1): B4<br>
1385<br>
1386&nbsp;[ B4 ]<br>
1387&nbsp;&nbsp;&nbsp;&nbsp;1: x = x + 1<br>
1388&nbsp;&nbsp;&nbsp;&nbsp;2: (x > 2)<br>
1389&nbsp;&nbsp;&nbsp;&nbsp;<b>T: if [B4.2]</b><br>
1390&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (1): B5<br>
1391&nbsp;&nbsp;&nbsp;&nbsp;Successors (2): B3 B2<br>
1392<br>
1393&nbsp;[ B3 ]<br>
1394&nbsp;&nbsp;&nbsp;&nbsp;1: x++<br>
1395&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (1): B4<br>
1396&nbsp;&nbsp;&nbsp;&nbsp;Successors (1): B1<br>
1397<br>
1398&nbsp;[ B2 ]<br>
1399&nbsp;&nbsp;&nbsp;&nbsp;1: x += 2<br>
1400&nbsp;&nbsp;&nbsp;&nbsp;2: x *= 2<br>
1401&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (1): B4<br>
1402&nbsp;&nbsp;&nbsp;&nbsp;Successors (1): B1<br>
1403<br>
1404&nbsp;[ B1 ]<br>
1405&nbsp;&nbsp;&nbsp;&nbsp;1: return x;<br>
1406&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (2): B2 B3<br>
1407&nbsp;&nbsp;&nbsp;&nbsp;Successors (1): B0<br>
1408<br>
1409&nbsp;[ B0 (EXIT) ]<br>
1410&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (1): B1<br>
1411&nbsp;&nbsp;&nbsp;&nbsp;Successors (0):
1412</code>
1413
1414<p>For each block, the pretty-printed output displays for each block
1415the number of <i>predecessor</i> blocks (blocks that have outgoing
1416control-flow to the given block) and <i>successor</i> blocks (blocks
1417that have control-flow that have incoming control-flow from the given
1418block). We can also clearly see the special entry and exit blocks at
1419the beginning and end of the pretty-printed output. For the entry
1420block (block B5), the number of predecessor blocks is 0, while for the
1421exit block (block B0) the number of successor blocks is 0.</p>
1422
1423<p>The most interesting block here is B4, whose outgoing control-flow
1424represents the branching caused by the sole if-statement
1425in <tt>foo</tt>. Of particular interest is the second statement in
1426the block, <b><tt>(x > 2)</tt></b>, and the terminator, printed
1427as <b><tt>if [B4.2]</tt></b>. The second statement represents the
1428evaluation of the condition of the if-statement, which occurs before
1429the actual branching of control-flow. Within the <tt>CFGBlock</tt>
1430for B4, the <tt>Stmt*</tt> for the second statement refers to the
1431actual expression in the AST for <b><tt>(x > 2)</tt></b>. Thus
1432pointers to subclasses of <tt>Expr</tt> can appear in the list of
1433statements in a block, and not just subclasses of <tt>Stmt</tt> that
1434refer to proper C statements.</p>
1435
1436<p>The terminator of block B4 is a pointer to the <tt>IfStmt</tt>
1437object in the AST. The pretty-printer outputs <b><tt>if
1438[B4.2]</tt></b> because the condition expression of the if-statement
1439has an actual place in the basic block, and thus the terminator is
1440essentially
1441<i>referring</i> to the expression that is the second statement of
1442block B4 (i.e., B4.2). In this manner, conditions for control-flow
1443(which also includes conditions for loops and switch statements) are
1444hoisted into the actual basic block.</p>
1445
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1446<!-- ===================== -->
1447<!-- <h4>Implicit Control-Flow</h4> -->
1448<!-- ===================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1449
1450<!--
1451<p>A key design principle of the <tt>CFG</tt> class was to not require
1452any transformations to the AST in order to represent control-flow.
1453Thus the <tt>CFG</tt> does not perform any "lowering" of the
1454statements in an AST: loops are not transformed into guarded gotos,
1455short-circuit operations are not converted to a set of if-statements,
1456and so on.</p>
1457-->
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]1458
Chris Lattner7bad1992008-11-16 21:48:07 +0000[diff] [blame]1459
1460<!-- ======================================================================= -->
1461<h3 id="Constants">Constant Folding in the Clang AST</h3>
1462<!-- ======================================================================= -->
1463
1464<p>There are several places where constants and constant folding matter a lot to
1465the Clang front-end. First, in general, we prefer the AST to retain the source
1466code as close to how the user wrote it as possible. This means that if they
1467wrote "5+4", we want to keep the addition and two constants in the AST, we don't
1468want to fold to "9". This means that constant folding in various ways turns
1469into a tree walk that needs to handle the various cases.</p>
1470
1471<p>However, there are places in both C and C++ that require constants to be
1472folded. For example, the C standard defines what an "integer constant
1473expression" (i-c-e) is with very precise and specific requirements. The
1474language then requires i-c-e's in a lot of places (for example, the size of a
1475bitfield, the value for a case statement, etc). For these, we have to be able
1476to constant fold the constants, to do semantic checks (e.g. verify bitfield size
1477is non-negative and that case statements aren't duplicated). We aim for Clang
1478to be very pedantic about this, diagnosing cases when the code does not use an
1479i-c-e where one is required, but accepting the code unless running with
1480<tt>-pedantic-errors</tt>.</p>
1481
1482<p>Things get a little bit more tricky when it comes to compatibility with
1483real-world source code. Specifically, GCC has historically accepted a huge
1484superset of expressions as i-c-e's, and a lot of real world code depends on this
1485unfortuate accident of history (including, e.g., the glibc system headers). GCC
1486accepts anything its "fold" optimizer is capable of reducing to an integer
1487constant, which means that the definition of what it accepts changes as its
1488optimizer does. One example is that GCC accepts things like "case X-X:" even
1489when X is a variable, because it can fold this to 0.</p>
1490
1491<p>Another issue are how constants interact with the extensions we support, such
1492as __builtin_constant_p, __builtin_inf, __extension__ and many others. C99
1493obviously does not specify the semantics of any of these extensions, and the
1494definition of i-c-e does not include them. However, these extensions are often
1495used in real code, and we have to have a way to reason about them.</p>
1496
1497<p>Finally, this is not just a problem for semantic analysis. The code
1498generator and other clients have to be able to fold constants (e.g. to
1499initialize global variables) and has to handle a superset of what C99 allows.
1500Further, these clients can benefit from extended information. For example, we
1501know that "foo()||1" always evaluates to true, but we can't replace the
1502expression with true because it has side effects.</p>
1503
1504<!-- ======================= -->
1505<h4>Implementation Approach</h4>
1506<!-- ======================= -->
1507
1508<p>After trying several different approaches, we've finally converged on a
1509design (Note, at the time of this writing, not all of this has been implemented,
1510consider this a design goal!). Our basic approach is to define a single
1511recursive method evaluation method (<tt>Expr::Evaluate</tt>), which is
1512implemented in <tt>AST/ExprConstant.cpp</tt>. Given an expression with 'scalar'
1513type (integer, fp, complex, or pointer) this method returns the following
1514information:</p>
1515
1516<ul>
1517<li>Whether the expression is an integer constant expression, a general
1518 constant that was folded but has no side effects, a general constant that
1519 was folded but that does have side effects, or an uncomputable/unfoldable
1520 value.
1521</li>
1522<li>If the expression was computable in any way, this method returns the APValue
1523 for the result of the expression.</li>
1524<li>If the expression is not evaluatable at all, this method returns
1525 information on one of the problems with the expression. This includes a
1526 SourceLocation for where the problem is, and a diagnostic ID that explains
1527 the problem. The diagnostic should be have ERROR type.</li>
1528<li>If the expression is not an integer constant expression, this method returns
1529 information on one of the problems with the expression. This includes a
1530 SourceLocation for where the problem is, and a diagnostic ID that explains
1531 the problem. The diagnostic should be have EXTENSION type.</li>
1532</ul>
1533
1534<p>This information gives various clients the flexibility that they want, and we
1535will eventually have some helper methods for various extensions. For example,
1536Sema should have a <tt>Sema::VerifyIntegerConstantExpression</tt> method, which
1537calls Evaluate on the expression. If the expression is not foldable, the error
1538is emitted, and it would return true. If the expression is not an i-c-e, the
1539EXTENSION diagnostic is emitted. Finally it would return false to indicate that
1540the AST is ok.</p>
1541
1542<p>Other clients can use the information in other ways, for example, codegen can
1543just use expressions that are foldable in any way.</p>
1544
1545<!-- ========== -->
1546<h4>Extensions</h4>
1547<!-- ========== -->
1548
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]1549<p>This section describes how some of the various extensions Clang supports
Chris Lattner7bad1992008-11-16 21:48:07 +0000[diff] [blame]1550interacts with constant evaluation:</p>
1551
1552<ul>
1553<li><b><tt>__extension__</tt></b>: The expression form of this extension causes
1554 any evaluatable subexpression to be accepted as an integer constant
1555 expression.</li>
1556<li><b><tt>__builtin_constant_p</tt></b>: This returns true (as a integer
Chris Lattner28daa532008-12-12 06:55:44 +0000[diff] [blame]1557 constant expression) if the operand is any evaluatable constant. As a
1558 special case, if <tt>__builtin_constant_p</tt> is the (potentially
1559 parenthesized) condition of a conditional operator expression ("?:"), only
Chris Lattner42b83dd2008-12-12 18:00:51 +0000[diff] [blame]1560 the true side of the conditional operator is considered, and it is evaluated
1561 with full constant folding.</li>
Chris Lattner7bad1992008-11-16 21:48:07 +0000[diff] [blame]1562<li><b><tt>__builtin_choose_expr</tt></b>: The condition is required to be an
1563 integer constant expression, but we accept any constant as an "extension of
1564 an extension". This only evaluates one operand depending on which way the
1565 condition evaluates.</li>
1566<li><b><tt>__builtin_classify_type</tt></b>: This always returns an integer
1567 constant expression.</li>
1568<li><b><tt>__builtin_inf,nan,..</tt></b>: These are treated just like a
1569 floating-point literal.</li>
1570<li><b><tt>__builtin_abs,copysign,..</tt></b>: These are constant folded as
1571 general constant expressions.</li>
1572</ul>
1573
1574
1575
1576
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]1577</div>
1578</body>
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]1579</html>