blob: 7ee8f010a32baea1cf2f3856ec19b9e4b2afa327 [file] [log] [blame]
Sean Hunt7e98b472011-06-23 01:21:01 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
Chris Lattner5ce933f2009-02-09 08:46:11 +00004<html>
5<head>
Sean Hunt7e98b472011-06-23 01:21:01 +00006 <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
7 <title>Clang LanguageExtensions</title>
Sean Hunt64f857b2011-06-23 01:22:53 +00008 <link type="text/css" rel="stylesheet" href="../menu.css">
9 <link type="text/css" rel="stylesheet" href="../content.css">
Sean Hunt7e98b472011-06-23 01:21:01 +000010 <style type="text/css">
11 td {
12 vertical-align: top;
13 }
14 </style>
Chris Lattner5ce933f2009-02-09 08:46:11 +000015</head>
16<body>
17
18<!--#include virtual="../menu.html.incl"-->
19
20<div id="content">
21
22<h1>Clang Language Extensions</h1>
23
24<ul>
25<li><a href="#intro">Introduction</a></li>
Chris Lattner148772a2009-06-13 07:13:28 +000026<li><a href="#feature_check">Feature Checking Macros</a></li>
John Thompson92bd8c72009-11-02 22:28:12 +000027<li><a href="#has_include">Include File Checking Macros</a></li>
Chris Lattner81edc9f2009-04-13 02:45:46 +000028<li><a href="#builtinmacros">Builtin Macros</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +000029<li><a href="#vectors">Vectors and Extended Vectors</a></li>
John McCall48209082010-11-08 19:48:17 +000030<li><a href="#deprecated">Messages on <tt>deprecated</tt> and <tt>unavailable</tt> attributes</a></li>
31<li><a href="#attributes-on-enumerators">Attributes on enumerators</a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +000032<li><a href="#checking_language_features">Checks for Standard Language Features</a>
Ted Kremenek22c34102009-12-03 02:05:57 +000033 <ul>
34 <li><a href="#cxx_exceptions">C++ exceptions</a></li>
35 <li><a href="#cxx_rtti">C++ RTTI</a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +000036 </ul></li>
37<li><a href="#checking_upcoming_features">Checks for Upcoming Standard Language Features</a>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +000038 <ul>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000039 <li><a href="#cxx0x">C++0x</a>
40 <ul>
41 <li><a href="#cxx_decltype">C++0x <tt>decltype()</tt></a></li>
42 <li><a href="#cxx_access_control_sfinae">C++0x SFINAE includes access control</a></li>
43 <li><a href="#cxx_alias_templates">C++0x alias templates</a></li>
44 <li><a href="#cxx_attributes">C++0x attributes</a></li>
45 <li><a href="#cxx_default_function_template_args">C++0x default template arguments in function templates</a></li>
Sean Huntd9624992011-06-23 06:11:37 +000046 <li><a href="#cxx_delegating_constructor">C++0x delegating constructors</a></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000047 <li><a href="#cxx_deleted_functions">C++0x deleted functions</a></li>
48 <li><a href="#cxx_lambdas">C++0x lambdas</a></li>
49 <li><a href="#cxx_nullptr">C++0x nullptr</a></li>
50 <li><a href="#cxx_override_control">C++0x override control</a></li>
51 <li><a href="#cxx_range_for">C++0x range-based for loop</a></li>
52 <li><a href="#cxx_rvalue_references">C++0x rvalue references</a></li>
53 <li><a href="#cxx_reference_qualified_functions">C++0x reference-qualified functions</a></li>
54 <li><a href="#cxx_static_assert">C++0x <tt>static_assert()</tt></a></li>
55 <li><a href="#cxx_auto_type">C++0x type inference</a></li>
56 <li><a href="#cxx_variadic_templates">C++0x variadic templates</a></li>
57 <li><a href="#cxx_inline_namespaces">C++0x inline namespaces</a></li>
58 <li><a href="#cxx_strong_enums">C++0x strongly-typed enumerations</a></li>
59 <li><a href="#cxx_trailing_return">C++0x trailing return type</a></li>
60 <li><a href="#cxx_noexcept">C++0x noexcept specification</a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +000061 </ul></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000062 <li><a href="#c1x">C1X</a>
63 <ul>
64 <li><a href="#c_generic_selections">C1X generic selections</a></li>
65 <li><a href="#c_static_assert">C1X <tt>_Static_assert()</tt></a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +000066 </ul></li>
67 </ul> </li>
Douglas Gregorafdf1372011-02-03 21:57:35 +000068<li><a href="#checking_type_traits">Checks for Type Traits</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +000069<li><a href="#blocks">Blocks</a></li>
Douglas Gregor926df6c2011-06-11 01:09:30 +000070<li><a href="#objc_features">Objective-C Features</a>
71 <ul>
72 <li><a href="#objc_instancetype">Related result types</a></li>
John McCallf85e1932011-06-15 23:02:42 +000073 <li><a href="#objc_arc">Automatic reference counting</a></li>
Douglas Gregor926df6c2011-06-11 01:09:30 +000074 </ul>
75</li>
Douglas Gregorcb54d432009-02-13 00:57:04 +000076<li><a href="#overloading-in-c">Function Overloading in C</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +000077<li><a href="#builtins">Builtin Functions</a>
78 <ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +000079 <li><a href="#__builtin_shufflevector">__builtin_shufflevector</a></li>
Chris Lattner21190d52009-09-21 03:09:59 +000080 <li><a href="#__builtin_unreachable">__builtin_unreachable</a></li>
Chris Lattner23aa9c82011-04-09 03:57:26 +000081 <li><a href="#__sync_swap">__sync_swap</a></li>
Douglas Gregorafdf1372011-02-03 21:57:35 +000082 </ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +000083</li>
Chris Lattner1177f912009-04-09 19:58:15 +000084<li><a href="#targetspecific">Target-Specific Extensions</a>
85 <ul>
86 <li><a href="#x86-specific">X86/X86-64 Language Extensions</a></li>
87 </ul>
88</li>
John McCall87494012011-03-18 03:51:49 +000089<li><a href="#analyzerspecific">Static Analysis-Specific Extensions</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +000090</ul>
91
Chris Lattner5ce933f2009-02-09 08:46:11 +000092<!-- ======================================================================= -->
93<h2 id="intro">Introduction</h2>
94<!-- ======================================================================= -->
95
96<p>This document describes the language extensions provided by Clang. In
Chris Lattner148772a2009-06-13 07:13:28 +000097addition to the language extensions listed here, Clang aims to support a broad
Chris Lattner5ce933f2009-02-09 08:46:11 +000098range of GCC extensions. Please see the <a
99href="http://gcc.gnu.org/onlinedocs/gcc/C-Extensions.html">GCC manual</a> for
100more information on these extensions.</p>
101
102<!-- ======================================================================= -->
Chris Lattner148772a2009-06-13 07:13:28 +0000103<h2 id="feature_check">Feature Checking Macros</h2>
104<!-- ======================================================================= -->
105
106<p>Language extensions can be very useful, but only if you know you can depend
Chris Lattnerc70e1932011-03-21 16:25:11 +0000107on them. In order to allow fine-grain features checks, we support three builtin
Chris Lattner148772a2009-06-13 07:13:28 +0000108function-like macros. This allows you to directly test for a feature in your
109code without having to resort to something like autoconf or fragile "compiler
110version checks".</p>
111
112<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000113<h3><a name="__has_builtin">__has_builtin</a></h3>
Chris Lattner148772a2009-06-13 07:13:28 +0000114<!-- ======================================================================= -->
115
116<p>This function-like macro takes a single identifier argument that is the name
117of a builtin function. It evaluates to 1 if the builtin is supported or 0 if
118not. It can be used like this:</p>
119
120<blockquote>
121<pre>
122#ifndef __has_builtin // Optional of course.
123 #define __has_builtin(x) 0 // Compatibility with non-clang compilers.
124#endif
125
126...
127#if __has_builtin(__builtin_trap)
128 __builtin_trap();
129#else
130 abort();
131#endif
132...
133</pre>
134</blockquote>
135
136
137<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000138<h3><a name="__has_feature_extension"> __has_feature and __has_extension</a></h3>
Chris Lattner148772a2009-06-13 07:13:28 +0000139<!-- ======================================================================= -->
140
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000141<p>These function-like macros take a single identifier argument that is the
142name of a feature. <code>__has_feature</code> evaluates to 1 if the feature
143is both supported by Clang and standardized in the current language standard
144or 0 if not (but see <a href="#has_feature_back_compat">below</a>), while
145<code>__has_extension</code> evaluates to 1 if the feature is supported by
146Clang in the current language (either as a language extension or a standard
147language feature) or 0 if not. They can be used like this:</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000148
149<blockquote>
150<pre>
151#ifndef __has_feature // Optional of course.
152 #define __has_feature(x) 0 // Compatibility with non-clang compilers.
153#endif
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000154#ifndef __has_extension
155 #define __has_extension __has_feature // Compatibility with pre-3.0 compilers.
156#endif
Chris Lattner148772a2009-06-13 07:13:28 +0000157
158...
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000159#if __has_feature(cxx_rvalue_references)
160// This code will only be compiled with the -std=c++0x and -std=gnu++0x
161// options, because rvalue references are only standardized in C++0x.
Chris Lattner148772a2009-06-13 07:13:28 +0000162#endif
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000163
164#if __has_extension(cxx_rvalue_references)
165// This code will be compiled with the -std=c++0x, -std=gnu++0x, -std=c++98
166// and -std=gnu++98 options, because rvalue references are supported as a
167// language extension in C++98.
168#endif
Chris Lattner148772a2009-06-13 07:13:28 +0000169</pre>
170</blockquote>
171
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000172<p id="has_feature_back_compat">For backwards compatibility reasons,
173<code>__has_feature</code> can also be used to test for support for
174non-standardized features, i.e. features not prefixed <code>c_</code>,
175<code>cxx_</code> or <code>objc_</code>.</p>
176
177<p>If the <code>-pedantic-errors</code> option is given,
178<code>__has_extension</code> is equivalent to <code>__has_feature</code>.</p>
179
Chris Lattner148772a2009-06-13 07:13:28 +0000180<p>The feature tag is described along with the language feature below.</p>
181
John Thompson92bd8c72009-11-02 22:28:12 +0000182<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000183<h3><a name="__has_attribute">__has_attribute</a></h3>
Anders Carlssoncae50952010-10-20 02:31:43 +0000184<!-- ======================================================================= -->
185
186<p>This function-like macro takes a single identifier argument that is the name
187of an attribute. It evaluates to 1 if the attribute is supported or 0 if not. It
188can be used like this:</p>
189
190<blockquote>
191<pre>
192#ifndef __has_attribute // Optional of course.
193 #define __has_attribute(x) 0 // Compatibility with non-clang compilers.
194#endif
195
196...
Anders Carlsson961003d2011-01-24 03:54:51 +0000197#if __has_attribute(always_inline)
198#define ALWAYS_INLINE __attribute__((always_inline))
Anders Carlssoncae50952010-10-20 02:31:43 +0000199#else
Anders Carlsson961003d2011-01-24 03:54:51 +0000200#define ALWAYS_INLINE
Anders Carlssoncae50952010-10-20 02:31:43 +0000201#endif
202...
203</pre>
204</blockquote>
205
206<!-- ======================================================================= -->
John Thompson92bd8c72009-11-02 22:28:12 +0000207<h2 id="has_include">Include File Checking Macros</h2>
208<!-- ======================================================================= -->
209
210<p>Not all developments systems have the same include files.
211The <a href="#__has_include">__has_include</a> and
212<a href="#__has_include_next">__has_include_next</a> macros allow you to
213check for the existence of an include file before doing
214a possibly failing #include directive.</p>
215
216<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000217<h3><a name="__has_include">__has_include</a></h3>
John Thompson92bd8c72009-11-02 22:28:12 +0000218<!-- ======================================================================= -->
219
220<p>This function-like macro takes a single file name string argument that
221is the name of an include file. It evaluates to 1 if the file can
222be found using the include paths, or 0 otherwise:</p>
223
224<blockquote>
225<pre>
226// Note the two possible file name string formats.
Sean Hunt7e98b472011-06-23 01:21:01 +0000227#if __has_include("myinclude.h") &amp;&amp; __has_include(&lt;stdint.h&gt;)
John Thompson92bd8c72009-11-02 22:28:12 +0000228# include "myinclude.h"
229#endif
230
231// To avoid problem with non-clang compilers not having this macro.
Sean Hunt7e98b472011-06-23 01:21:01 +0000232#if defined(__has_include) &amp;&amp; __has_include("myinclude.h")
John Thompson92bd8c72009-11-02 22:28:12 +0000233# include "myinclude.h"
234#endif
235</pre>
236</blockquote>
237
238<p>To test for this feature, use #if defined(__has_include).</p>
239
240<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000241<h3><a name="__has_include_next">__has_include_next</a></h3>
John Thompson92bd8c72009-11-02 22:28:12 +0000242<!-- ======================================================================= -->
243
244<p>This function-like macro takes a single file name string argument that
245is the name of an include file. It is like __has_include except that it
246looks for the second instance of the given file found in the include
247paths. It evaluates to 1 if the second instance of the file can
248be found using the include paths, or 0 otherwise:</p>
249
250<blockquote>
251<pre>
252// Note the two possible file name string formats.
Sean Hunt7e98b472011-06-23 01:21:01 +0000253#if __has_include_next("myinclude.h") &amp;&amp; __has_include_next(&lt;stdint.h&gt;)
John Thompson92bd8c72009-11-02 22:28:12 +0000254# include_next "myinclude.h"
255#endif
256
257// To avoid problem with non-clang compilers not having this macro.
Sean Hunt7e98b472011-06-23 01:21:01 +0000258#if defined(__has_include_next) &amp;&amp; __has_include_next("myinclude.h")
John Thompson92bd8c72009-11-02 22:28:12 +0000259# include_next "myinclude.h"
260#endif
261</pre>
262</blockquote>
263
264<p>Note that __has_include_next, like the GNU extension
265#include_next directive, is intended for use in headers only,
266and will issue a warning if used in the top-level compilation
267file. A warning will also be issued if an absolute path
268is used in the file argument.</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000269
270<!-- ======================================================================= -->
Chris Lattner81edc9f2009-04-13 02:45:46 +0000271<h2 id="builtinmacros">Builtin Macros</h2>
272<!-- ======================================================================= -->
273
Douglas Gregor4290fbd2010-04-30 02:51:06 +0000274<dl>
275 <dt><code>__BASE_FILE__</code></dt>
276 <dd>Defined to a string that contains the name of the main input
277 file passed to Clang.</dd>
278
279 <dt><code>__COUNTER__</code></dt>
280 <dd>Defined to an integer value that starts at zero and is
281 incremented each time the <code>__COUNTER__</code> macro is
282 expanded.</dd>
283
284 <dt><code>__INCLUDE_LEVEL__</code></dt>
285 <dd>Defined to an integral value that is the include depth of the
286 file currently being translated. For the main file, this value is
287 zero.</dd>
288
289 <dt><code>__TIMESTAMP__</code></dt>
290 <dd>Defined to the date and time of the last modification of the
291 current source file.</dd>
292
293 <dt><code>__clang__</code></dt>
294 <dd>Defined when compiling with Clang</dd>
295
296 <dt><code>__clang_major__</code></dt>
297 <dd>Defined to the major version number of Clang (e.g., the 2 in
298 2.0.1).</dd>
299
300 <dt><code>__clang_minor__</code></dt>
301 <dd>Defined to the minor version number of Clang (e.g., the 0 in
302 2.0.1).</dd>
303
304 <dt><code>__clang_patchlevel__</code></dt>
305 <dd>Defined to the patch level of Clang (e.g., the 1 in 2.0.1).</dd>
306
307 <dt><code>__clang_version__</code></dt>
308 <dd>Defined to a string that captures the Clang version, including
309 the Subversion tag or revision number, e.g., "1.5 (trunk
310 102332)".</dd>
311</dl>
Chris Lattner81edc9f2009-04-13 02:45:46 +0000312
313<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000314<h2 id="vectors">Vectors and Extended Vectors</h2>
315<!-- ======================================================================= -->
316
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000317<p>Supports the GCC vector extensions, plus some stuff like V[1].</p>
318
319<p>Also supports <tt>ext_vector</tt>, which additionally support for V.xyzw
320syntax and other tidbits as seen in OpenCL. An example is:</p>
321
322<blockquote>
323<pre>
324typedef float float4 <b>__attribute__((ext_vector_type(4)))</b>;
325typedef float float2 <b>__attribute__((ext_vector_type(2)))</b>;
326
327float4 foo(float2 a, float2 b) {
328 float4 c;
329 c.xz = a;
330 c.yw = b;
331 return c;
332}
John McCall48209082010-11-08 19:48:17 +0000333</pre>
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000334</blockquote>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000335
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000336<p>Query for this feature with __has_extension(attribute_ext_vector_type).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000337
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000338<p>See also <a href="#__builtin_shufflevector">__builtin_shufflevector</a>.</p>
339
Chris Lattner5ce933f2009-02-09 08:46:11 +0000340<!-- ======================================================================= -->
John McCall48209082010-11-08 19:48:17 +0000341<h2 id="deprecated">Messages on <tt>deprecated</tt> and <tt>unavailable</tt> Attributes</h2>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000342<!-- ======================================================================= -->
343
John McCall48209082010-11-08 19:48:17 +0000344<p>An optional string message can be added to the <tt>deprecated</tt>
345and <tt>unavailable</tt> attributes. For example:</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000346
John McCall48209082010-11-08 19:48:17 +0000347<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000348<pre>void explode(void) __attribute__((deprecated("extremely unsafe, use 'combust' instead!!!")));</pre>
John McCall48209082010-11-08 19:48:17 +0000349</blockquote>
350
351<p>If the deprecated or unavailable declaration is used, the message
352will be incorporated into the appropriate diagnostic:</p>
353
354<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000355<pre>harmless.c:4:3: warning: 'explode' is deprecated: extremely unsafe, use 'combust' instead!!! [-Wdeprecated-declarations]
John McCall48209082010-11-08 19:48:17 +0000356 explode();
357 ^</pre>
358</blockquote>
359
360<p>Query for this feature
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000361with <tt>__has_extension(attribute_deprecated_with_message)</tt>
362and <tt>__has_extension(attribute_unavailable_with_message)</tt>.</p>
John McCall48209082010-11-08 19:48:17 +0000363
364<!-- ======================================================================= -->
365<h2 id="attributes-on-enumerators">Attributes on Enumerators</h2>
366<!-- ======================================================================= -->
367
368<p>Clang allows attributes to be written on individual enumerators.
369This allows enumerators to be deprecated, made unavailable, etc. The
370attribute must appear after the enumerator name and before any
371initializer, like so:</p>
372
373<blockquote>
374<pre>enum OperationMode {
375 OM_Invalid,
376 OM_Normal,
377 OM_Terrified __attribute__((deprecated)),
378 OM_AbortOnError __attribute__((deprecated)) = 4
379};</pre>
380</blockquote>
381
382<p>Attributes on the <tt>enum</tt> declaration do not apply to
383individual enumerators.</p>
384
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000385<p>Query for this feature with <tt>__has_extension(enumerator_attributes)</tt>.</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000386
387<!-- ======================================================================= -->
Ted Kremenek87774fd2009-12-03 02:04:01 +0000388<h2 id="checking_language_features">Checks for Standard Language Features</h2>
389<!-- ======================================================================= -->
390
391<p>The <tt>__has_feature</tt> macro can be used to query if certain standard language features are
392enabled. Those features are listed here.</p>
393
Ted Kremenek22c34102009-12-03 02:05:57 +0000394<h3 id="cxx_exceptions">C++ exceptions</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000395
Ted Kremenek22c34102009-12-03 02:05:57 +0000396<p>Use <tt>__has_feature(cxx_exceptions)</tt> to determine if C++ exceptions have been enabled. For
Sean Hunt647ba1b2011-06-23 00:42:53 +0000397example, compiling code with <tt>-fexceptions</tt> enables C++ exceptions.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000398
Ted Kremenek22c34102009-12-03 02:05:57 +0000399<h3 id="cxx_rtti">C++ RTTI</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000400
Ted Kremenek0eb95602009-12-03 02:06:43 +0000401<p>Use <tt>__has_feature(cxx_rtti)</tt> to determine if C++ RTTI has been enabled. For example,
Ted Kremenek22c34102009-12-03 02:05:57 +0000402compiling code with <tt>-fno-rtti</tt> disables the use of RTTI.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000403
404<!-- ======================================================================= -->
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000405<h2 id="checking_upcoming_features">Checks for Upcoming Standard Language Features</h2>
406<!-- ======================================================================= -->
407
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000408<p>The <tt>__has_feature</tt> or <tt>__has_extension</tt> macros can be used
409to query if certain upcoming standard language features are enabled. Those
410features are listed here. Features that are not yet implemented will be
411noted.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000412
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000413<h3 id="cxx0x">C++0x</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000414
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000415<p>The features listed below are slated for inclusion in the upcoming
416C++0x standard. As a result, all these features are enabled
417with the <tt>-std=c++0x</tt> option when compiling C++ code.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000418
Sean Hunt7e98b472011-06-23 01:21:01 +0000419<h4 id="cxx_decltype">C++0x <tt>decltype()</tt></h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000420
421<p>Use <tt>__has_feature(cxx_decltype)</tt> or
422<tt>__has_extension(cxx_decltype)</tt> to determine if support for the
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000423<tt>decltype()</tt> specifier is enabled.</p>
424
Sean Hunt7e98b472011-06-23 01:21:01 +0000425<h4 id="cxx_access_control_sfinae">C++0x SFINAE includes access control</h4>
Douglas Gregor7822ee32011-05-11 23:45:11 +0000426
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000427<p>Use <tt>__has_feature(cxx_access_control_sfinae)</tt> or <tt>__has_extension(cxx_access_control_sfinae)</tt> to determine whether access-control errors (e.g., calling a private constructor) are considered to be template argument deduction errors (aka SFINAE errors), per <a href="http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#1170">C++ DR1170</a>.</p>
Douglas Gregor7822ee32011-05-11 23:45:11 +0000428
Sean Hunt7e98b472011-06-23 01:21:01 +0000429<h4 id="cxx_alias_templates">C++0x alias templates</h4>
Richard Smith3e4c6c42011-05-05 21:57:07 +0000430
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000431<p>Use <tt>__has_feature(cxx_alias_templates)</tt> or
432<tt>__has_extension(cxx_alias_templates)</tt> to determine if support for
Richard Smith3e4c6c42011-05-05 21:57:07 +0000433C++0x's alias declarations and alias templates is enabled.</p>
434
Sean Hunt7e98b472011-06-23 01:21:01 +0000435<h4 id="cxx_attributes">C++0x attributes</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000436
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000437<p>Use <tt>__has_feature(cxx_attributes)</tt> or
438<tt>__has_extension(cxx_attributes)</tt> to determine if support for attribute
439parsing with C++0x's square bracket notation is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000440
Sean Hunt7e98b472011-06-23 01:21:01 +0000441<h4 id="cxx_default_function_template_args">C++0x default template arguments in function templates</h4>
Douglas Gregor07508002011-02-05 20:35:30 +0000442
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000443<p>Use <tt>__has_feature(cxx_default_function_template_args)</tt> or
444<tt>__has_extension(cxx_default_function_template_args)</tt> to determine
445if support for default template arguments in function templates is enabled.</p>
Douglas Gregor07508002011-02-05 20:35:30 +0000446
Sean Huntd9624992011-06-23 06:11:37 +0000447<h4 id="cxx_delegating_constructors">C++0x delegating constructors</h4>
448
449<p>Use <tt>__has_feature(cxx_delegating_constructors)</tt> to determine if
450support for delegating constructors is enabled.</p>
451
Sean Hunt7e98b472011-06-23 01:21:01 +0000452<h4 id="cxx_deleted_functions">C++0x <tt>delete</tt>d functions</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000453
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000454<p>Use <tt>__has_feature(cxx_deleted_functions)</tt> or
455<tt>__has_extension(cxx_deleted_functions)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000456deleted function definitions (with <tt>= delete</tt>) is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000457
Sean Hunt7e98b472011-06-23 01:21:01 +0000458<h4 id="cxx_lambdas">C++0x lambdas</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000459
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000460<p>Use <tt>__has_feature(cxx_lambdas)</tt> or
461<tt>__has_extension(cxx_lambdas)</tt> to determine if support for lambdas
462is enabled. clang does not currently implement this feature.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000463
Sean Hunt7e98b472011-06-23 01:21:01 +0000464<h4 id="cxx_nullptr">C++0x <tt>nullptr</tt></h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000465
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000466<p>Use <tt>__has_feature(cxx_nullptr)</tt> or
467<tt>__has_extension(cxx_nullptr)</tt> to determine if support for
Douglas Gregor84ee2ee2011-05-21 23:15:46 +0000468<tt>nullptr</tt> is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000469
Sean Hunt7e98b472011-06-23 01:21:01 +0000470<h4 id="cxx_override_control">C++0x <tt>override control</tt></h4>
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000471
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000472<p>Use <tt>__has_feature(cxx_override_control)</tt> or
473<tt>__has_extension(cxx_override_control)</tt> to determine if support for
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000474the override control keywords is enabled.</p>
475
Sean Hunt7e98b472011-06-23 01:21:01 +0000476<h4 id="cxx_reference_qualified_functions">C++0x reference-qualified functions</h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000477<p>Use <tt>__has_feature(cxx_reference_qualified_functions)</tt> or
478<tt>__has_extension(cxx_reference_qualified_functions)</tt> to determine
479if support for reference-qualified functions (e.g., member functions with
480<code>&amp;</code> or <code>&amp;&amp;</code> applied to <code>*this</code>)
481is enabled.</p>
Douglas Gregor56209ff2011-01-26 21:25:54 +0000482
Sean Hunt7e98b472011-06-23 01:21:01 +0000483<h4 id="cxx_range_for">C++0x range-based <tt>for</tt> loop</h4>
Richard Smitha391a462011-04-15 15:14:40 +0000484
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000485<p>Use <tt>__has_feature(cxx_range_for)</tt> or
486<tt>__has_extension(cxx_range_for)</tt> to determine if support for the
487range-based for loop is enabled. </p>
Richard Smitha391a462011-04-15 15:14:40 +0000488
Sean Hunt7e98b472011-06-23 01:21:01 +0000489<h4 id="cxx_rvalue_references">C++0x rvalue references</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000490
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000491<p>Use <tt>__has_feature(cxx_rvalue_references)</tt> or
492<tt>__has_extension(cxx_rvalue_references)</tt> to determine if support for
Douglas Gregor56209ff2011-01-26 21:25:54 +0000493rvalue references is enabled. </p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000494
Sean Hunt7e98b472011-06-23 01:21:01 +0000495<h4 id="cxx_static_assert">C++0x <tt>static_assert()</tt></h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000496
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000497<p>Use <tt>__has_feature(cxx_static_assert)</tt> or
498<tt>__has_extension(cxx_static_assert)</tt> to determine if support for
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000499compile-time assertions using <tt>static_assert</tt> is enabled.</p>
500
Sean Hunt7e98b472011-06-23 01:21:01 +0000501<h4 id="cxx_auto_type">C++0x type inference</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000502
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000503<p>Use <tt>__has_feature(cxx_auto_type)</tt> or
504<tt>__has_extension(cxx_auto_type)</tt> to determine C++0x type inference is
505supported using the <tt>auto</tt> specifier. If this is disabled, <tt>auto</tt>
506will instead be a storage class specifier, as in C or C++98.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000507
Sean Hunt7e98b472011-06-23 01:21:01 +0000508<h4 id="cxx_variadic_templates">C++0x variadic templates</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000509
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000510<p>Use <tt>__has_feature(cxx_variadic_templates)</tt> or
511<tt>__has_extension(cxx_variadic_templates)</tt> to determine if support
Douglas Gregor83d77812011-01-19 23:15:20 +0000512for variadic templates is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000513
Sean Hunt7e98b472011-06-23 01:21:01 +0000514<h4 id="cxx_inline_namespaces">C++0x inline namespaces</h4>
Sebastian Redlf6c09772010-08-31 23:28:47 +0000515
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000516<p>Use <tt>__has_feature(cxx_inline_namespaces)</tt> or
517<tt>__has_extension(cxx_inline_namespaces)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000518inline namespaces is enabled.</p>
519
Sean Hunt7e98b472011-06-23 01:21:01 +0000520<h4 id="cxx_trailing_return">C++0x trailing return type</h4>
Douglas Gregordab60ad2010-10-01 18:44:50 +0000521
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000522<p>Use <tt>__has_feature(cxx_trailing_return)</tt> or
523<tt>__has_extension(cxx_trailing_return)</tt> to determine if support for the
524alternate function declaration syntax with trailing return type is enabled.</p>
Douglas Gregordab60ad2010-10-01 18:44:50 +0000525
Sean Hunt7e98b472011-06-23 01:21:01 +0000526<h4 id="cxx_noexcept">C++0x noexcept</h4>
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000527
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000528<p>Use <tt>__has_feature(cxx_noexcept)</tt> or
529<tt>__has_extension(cxx_noexcept)</tt> to determine if support for noexcept
530exception specifications is enabled.</p>
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000531
Sean Hunt7e98b472011-06-23 01:21:01 +0000532<h4 id="cxx_strong_enums">C++0x strongly typed enumerations</h4>
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000533
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000534<p>Use <tt>__has_feature(cxx_strong_enums)</tt> or
535<tt>__has_extension(cxx_strong_enums)</tt> to determine if support for
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000536strongly typed, scoped enumerations is enabled.</p>
537
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000538<h3 id="c1x">C1X</h3>
539
540<p>The features listed below are slated for inclusion in the upcoming
541C1X standard. As a result, all these features are enabled
542with the <tt>-std=c1x</tt> option when compiling C code.</p>
543
Sean Hunt7e98b472011-06-23 01:21:01 +0000544<h4 id="c_generic_selections">C1X generic selections</h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000545
546<p>Use <tt>__has_feature(c_generic_selections)</tt> or
547<tt>__has_extension(c_generic_selections)</tt> to determine if support for
548generic selections is enabled.</p>
549
550<p>As an extension, the C1X generic selection expression is available in all
551languages supported by Clang. The syntax is the same as that given in the
552C1X draft standard.</p>
553
554<p>In C, type compatibility is decided according to the rules given in the
555appropriate standard, but in C++, which lacks the type compatibility rules
556used in C, types are considered compatible only if they are equivalent.</p>
557
Sean Hunt7e98b472011-06-23 01:21:01 +0000558<h4 id="c_static_assert">C1X <tt>_Static_assert()</tt></h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000559
560<p>Use <tt>__has_feature(c_static_assert)</tt> or
561<tt>__has_extension(c_static_assert)</tt> to determine if support for
562compile-time assertions using <tt>_Static_assert</tt> is enabled.</p>
563
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000564<!-- ======================================================================= -->
Douglas Gregorafdf1372011-02-03 21:57:35 +0000565<h2 id="checking_type_traits">Checks for Type Traits</h2>
566<!-- ======================================================================= -->
567
Sean Hunt7e98b472011-06-23 01:21:01 +0000568<p>Clang supports the <a href="http://gcc.gnu.org/onlinedocs/gcc/Type-Traits.html">GNU C++ type traits</a> and a subset of the <a href="http://msdn.microsoft.com/en-us/library/ms177194(v=VS.100).aspx">Microsoft Visual C++ Type traits</a>. For each supported type trait <code>__X</code>, <code>__has_extension(X)</code> indicates the presence of the type trait. For example:
Douglas Gregorafdf1372011-02-03 21:57:35 +0000569<blockquote>
570<pre>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000571#if __has_extension(is_convertible_to)
Douglas Gregorafdf1372011-02-03 21:57:35 +0000572template&lt;typename From, typename To&gt;
573struct is_convertible_to {
574 static const bool value = __is_convertible_to(From, To);
575};
576#else
577// Emulate type trait
578#endif
579</pre>
580</blockquote>
581
582<p>The following type traits are supported by Clang:</p>
583<ul>
584 <li><code>__has_nothrow_assign</code> (GNU, Microsoft)</li>
585 <li><code>__has_nothrow_copy</code> (GNU, Microsoft)</li>
586 <li><code>__has_nothrow_constructor</code> (GNU, Microsoft)</li>
587 <li><code>__has_trivial_assign</code> (GNU, Microsoft)</li>
588 <li><code>__has_trivial_copy</code> (GNU, Microsoft)</li>
589 <li><code>__has_trivial_constructor</code> (GNU, Microsoft)</li>
590 <li><code>__has_trivial_destructor</code> (GNU, Microsoft)</li>
591 <li><code>__has_virtual_destructor</code> (GNU, Microsoft)</li>
592 <li><code>__is_abstract</code> (GNU, Microsoft)</li>
593 <li><code>__is_base_of</code> (GNU, Microsoft)</li>
594 <li><code>__is_class</code> (GNU, Microsoft)</li>
595 <li><code>__is_convertible_to</code> (Microsoft)</li>
596 <li><code>__is_empty</code> (GNU, Microsoft)</li>
597 <li><code>__is_enum</code> (GNU, Microsoft)</li>
598 <li><code>__is_pod</code> (GNU, Microsoft)</li>
599 <li><code>__is_polymorphic</code> (GNU, Microsoft)</li>
600 <li><code>__is_union</code> (GNU, Microsoft)</li>
601 <li><code>__is_literal(type)</code>: Determines whether the given type is a literal type</li>
602</ul>
603
604<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000605<h2 id="blocks">Blocks</h2>
606<!-- ======================================================================= -->
607
Chris Lattnera7dbdf52009-03-09 07:03:22 +0000608<p>The syntax and high level language feature description is in <a
609href="BlockLanguageSpec.txt">BlockLanguageSpec.txt</a>. Implementation and ABI
610details for the clang implementation are in <a
Chris Lattner5d7650b2010-03-16 21:43:03 +0000611href="Block-ABI-Apple.txt">Block-ABI-Apple.txt</a>.</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000612
Chris Lattner148772a2009-06-13 07:13:28 +0000613
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000614<p>Query for this feature with __has_extension(blocks).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000615
Chris Lattner5ce933f2009-02-09 08:46:11 +0000616<!-- ======================================================================= -->
Douglas Gregor926df6c2011-06-11 01:09:30 +0000617<h2 id="objc_features">Objective-C Features</h2>
618<!-- ======================================================================= -->
619
620<h3 id="objc_instancetype">Related result types</h3>
621
622<p>According to Cocoa conventions, Objective-C methods with certain names ("init", "alloc", etc.) always return objects that are an instance of the receiving class's type. Such methods are said to have a "related result type", meaning that a message send to one of these methods will have the same static type as an instance of the receiver class. For example, given the following classes:</p>
623
624<blockquote>
625<pre>
626@interface NSObject
627+ (id)alloc;
628- (id)init;
629@end
630
631@interface NSArray : NSObject
632@end
633</pre>
634</blockquote>
635
636<p>and this common initialization pattern</p>
637
638<blockquote>
639<pre>
640NSArray *array = [[NSArray alloc] init];
641</pre>
642</blockquote>
643
644<p>the type of the expression <code>[NSArray alloc]</code> is
645<code>NSArray*</code> because <code>alloc</code> implicitly has a
646related result type. Similarly, the type of the expression
647<code>[[NSArray alloc] init]</code> is <code>NSArray*</code>, since
648<code>init</code> has a related result type and its receiver is known
649to have the type <code>NSArray *</code>. If neither <code>alloc</code> nor <code>init</code> had a related result type, the expressions would have had type <code>id</code>, as declared in the method signature.</p>
650
651<p>To determine whether a method has a related result type, the first
652word in the camel-case selector (e.g., "init" in "initWithObjects") is
653considered, and the method will a related result type if its return
Sean Hunt7e98b472011-06-23 01:21:01 +0000654type is compatible with the type of its class and if</p>
Douglas Gregor926df6c2011-06-11 01:09:30 +0000655
656<ul>
657
658 <li>the first word is "alloc" or "new", and the method is a class
659 method, or</li>
660
661 <li>the first word is "autorelease", "init", "retain", or "self",
662 and the method is an instance method.</li>
663
Sean Hunt7e98b472011-06-23 01:21:01 +0000664</ul>
Douglas Gregor926df6c2011-06-11 01:09:30 +0000665
666<p>If a method with a related result type is overridden by a subclass
667method, the subclass method must also return a type that is compatible
668with the subclass type. For example:</p>
669
670<blockquote>
671<pre>
672@interface NSString : NSObject
673- (NSUnrelated *)init; // incorrect usage: NSUnrelated is not NSString or a superclass of NSString
674@end
675</pre>
676</blockquote>
677
678<p>Related result types only affect the type of a message send or
679property access via the given method. In all other respects, a method
680with a related result type is treated the same way as method without a
681related result type.</p>
682
683<!-- ======================================================================= -->
John McCallf85e1932011-06-15 23:02:42 +0000684<h2 id="objc_arc">Automatic reference counting </h2>
685<!-- ======================================================================= -->
686
687<p>Clang provides support for <a href="AutomaticReferenceCounting.html">automated reference counting</a> in Objective-C, which eliminates the need for manual retain/release/autorelease message sends. There are two feature macros associated with automatic reference counting: <code>__has_feature(objc_arc)</code> indicates the availability of automated reference counting in general, while <code>__has_feature(objc_arc_weak)</code> indicates that automated reference counting also includes support for <code>__weak</code> pointers to Objective-C objects.</p>
688
689<!-- ======================================================================= -->
Douglas Gregorcb54d432009-02-13 00:57:04 +0000690<h2 id="overloading-in-c">Function Overloading in C</h2>
691<!-- ======================================================================= -->
692
Chris Lattnerf161d412009-02-13 21:51:45 +0000693<p>Clang provides support for C++ function overloading in C. Function
694overloading in C is introduced using the <tt>overloadable</tt> attribute. For
695example, one might provide several overloaded versions of a <tt>tgsin</tt>
696function that invokes the appropriate standard function computing the sine of a
697value with <tt>float</tt>, <tt>double</tt>, or <tt>long double</tt>
698precision:</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000699
700<blockquote>
701<pre>
702#include &lt;math.h&gt;
703float <b>__attribute__((overloadable))</b> tgsin(float x) { return sinf(x); }
704double <b>__attribute__((overloadable))</b> tgsin(double x) { return sin(x); }
705long double <b>__attribute__((overloadable))</b> tgsin(long double x) { return sinl(x); }
706</pre>
707</blockquote>
708
709<p>Given these declarations, one can call <tt>tgsin</tt> with a
710<tt>float</tt> value to receive a <tt>float</tt> result, with a
711<tt>double</tt> to receive a <tt>double</tt> result, etc. Function
712overloading in C follows the rules of C++ function overloading to pick
713the best overload given the call arguments, with a few C-specific
714semantics:</p>
715<ul>
716 <li>Conversion from <tt>float</tt> or <tt>double</tt> to <tt>long
717 double</tt> is ranked as a floating-point promotion (per C99) rather
718 than as a floating-point conversion (as in C++).</li>
719
720 <li>A conversion from a pointer of type <tt>T*</tt> to a pointer of type
721 <tt>U*</tt> is considered a pointer conversion (with conversion
722 rank) if <tt>T</tt> and <tt>U</tt> are compatible types.</li>
723
724 <li>A conversion from type <tt>T</tt> to a value of type <tt>U</tt>
725 is permitted if <tt>T</tt> and <tt>U</tt> are compatible types. This
726 conversion is given "conversion" rank.</li>
727</ul>
728
729<p>The declaration of <tt>overloadable</tt> functions is restricted to
730function declarations and definitions. Most importantly, if any
731function with a given name is given the <tt>overloadable</tt>
732attribute, then all function declarations and definitions with that
733name (and in that scope) must have the <tt>overloadable</tt>
Chris Lattnerf161d412009-02-13 21:51:45 +0000734attribute. This rule even applies to redeclarations of functions whose original
735declaration had the <tt>overloadable</tt> attribute, e.g.,</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000736
737<blockquote>
738<pre>
739int f(int) __attribute__((overloadable));
740float f(float); <i>// error: declaration of "f" must have the "overloadable" attribute</i>
741
742int g(int) __attribute__((overloadable));
743int g(int) { } <i>// error: redeclaration of "g" must also have the "overloadable" attribute</i>
744</pre>
745</blockquote>
746
Douglas Gregor965acbb2009-02-18 07:07:28 +0000747<p>Functions marked <tt>overloadable</tt> must have
748prototypes. Therefore, the following code is ill-formed:</p>
749
750<blockquote>
751<pre>
752int h() __attribute__((overloadable)); <i>// error: h does not have a prototype</i>
753</pre>
754</blockquote>
755
756<p>However, <tt>overloadable</tt> functions are allowed to use a
757ellipsis even if there are no named parameters (as is permitted in C++). This feature is particularly useful when combined with the <tt>unavailable</tt> attribute:</p>
758
759<blockquote>
760<pre>
Chris Lattner02246802009-02-18 22:27:46 +0000761void honeypot(...) __attribute__((overloadable, unavailable)); <i>// calling me is an error</i>
Douglas Gregor965acbb2009-02-18 07:07:28 +0000762</pre>
763</blockquote>
764
Douglas Gregorcb54d432009-02-13 00:57:04 +0000765<p>Functions declared with the <tt>overloadable</tt> attribute have
766their names mangled according to the same rules as C++ function
767names. For example, the three <tt>tgsin</tt> functions in our
768motivating example get the mangled names <tt>_Z5tgsinf</tt>,
Chris Lattner71b48d62010-11-28 18:19:13 +0000769<tt>_Z5tgsind</tt>, and <tt>_Z5tgsine</tt>, respectively. There are two
Douglas Gregorcb54d432009-02-13 00:57:04 +0000770caveats to this use of name mangling:</p>
771
772<ul>
773
774 <li>Future versions of Clang may change the name mangling of
775 functions overloaded in C, so you should not depend on an specific
776 mangling. To be completely safe, we strongly urge the use of
777 <tt>static inline</tt> with <tt>overloadable</tt> functions.</li>
778
779 <li>The <tt>overloadable</tt> attribute has almost no meaning when
780 used in C++, because names will already be mangled and functions are
781 already overloadable. However, when an <tt>overloadable</tt>
782 function occurs within an <tt>extern "C"</tt> linkage specification,
783 it's name <i>will</i> be mangled in the same way as it would in
784 C.</li>
785</ul>
786
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000787<p>Query for this feature with __has_extension(attribute_overloadable).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000788
789
Douglas Gregorcb54d432009-02-13 00:57:04 +0000790<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000791<h2 id="builtins">Builtin Functions</h2>
792<!-- ======================================================================= -->
793
794<p>Clang supports a number of builtin library functions with the same syntax as
795GCC, including things like <tt>__builtin_nan</tt>,
796<tt>__builtin_constant_p</tt>, <tt>__builtin_choose_expr</tt>,
797<tt>__builtin_types_compatible_p</tt>, <tt>__sync_fetch_and_add</tt>, etc. In
798addition to the GCC builtins, Clang supports a number of builtins that GCC does
799not, which are listed here.</p>
800
801<p>Please note that Clang does not and will not support all of the GCC builtins
802for vector operations. Instead of using builtins, you should use the functions
803defined in target-specific header files like <tt>&lt;xmmintrin.h&gt;</tt>, which
804define portable wrappers for these. Many of the Clang versions of these
805functions are implemented directly in terms of <a href="#vectors">extended
806vector support</a> instead of builtins, in order to reduce the number of
807builtins that we need to implement.</p>
808
Chris Lattner5ce933f2009-02-09 08:46:11 +0000809<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000810<h3><a name="__builtin_shufflevector">__builtin_shufflevector</a></h3>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000811<!-- ======================================================================= -->
812
Chris Lattneraad826b2009-09-16 18:56:12 +0000813<p><tt>__builtin_shufflevector</tt> is used to express generic vector
Chris Lattner6f72da52009-02-13 20:00:20 +0000814permutation/shuffle/swizzle operations. This builtin is also very important for
815the implementation of various target-specific header files like
816<tt>&lt;xmmintrin.h&gt;</tt>.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000817</p>
818
819<p><b>Syntax:</b></p>
820
821<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000822__builtin_shufflevector(vec1, vec2, index1, index2, ...)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000823</pre>
824
825<p><b>Examples:</b></p>
826
827<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000828 // Identity operation - return 4-element vector V1.
829 __builtin_shufflevector(V1, V1, 0, 1, 2, 3)
830
831 // "Splat" element 0 of V1 into a 4-element result.
832 __builtin_shufflevector(V1, V1, 0, 0, 0, 0)
833
834 // Reverse 4-element vector V1.
835 __builtin_shufflevector(V1, V1, 3, 2, 1, 0)
836
837 // Concatenate every other element of 4-element vectors V1 and V2.
838 __builtin_shufflevector(V1, V2, 0, 2, 4, 6)
839
840 // Concatenate every other element of 8-element vectors V1 and V2.
841 __builtin_shufflevector(V1, V2, 0, 2, 4, 6, 8, 10, 12, 14)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000842</pre>
843
844<p><b>Description:</b></p>
845
Chris Lattner6f72da52009-02-13 20:00:20 +0000846<p>The first two arguments to __builtin_shufflevector are vectors that have the
847same element type. The remaining arguments are a list of integers that specify
848the elements indices of the first two vectors that should be extracted and
849returned in a new vector. These element indices are numbered sequentially
850starting with the first vector, continuing into the second vector. Thus, if
851vec1 is a 4-element vector, index 5 would refer to the second element of vec2.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000852</p>
853
Chris Lattner6f72da52009-02-13 20:00:20 +0000854<p>The result of __builtin_shufflevector is a vector
855with the same element type as vec1/vec2 but that has an element count equal to
856the number of indices specified.
857</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000858
Chris Lattner21190d52009-09-21 03:09:59 +0000859<p>Query for this feature with __has_builtin(__builtin_shufflevector).</p>
860
861<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000862<h3><a name="__builtin_unreachable">__builtin_unreachable</a></h3>
Chris Lattner21190d52009-09-21 03:09:59 +0000863<!-- ======================================================================= -->
864
865<p><tt>__builtin_unreachable</tt> is used to indicate that a specific point in
866the program cannot be reached, even if the compiler might otherwise think it
867can. This is useful to improve optimization and eliminates certain warnings.
868For example, without the <tt>__builtin_unreachable</tt> in the example below,
869the compiler assumes that the inline asm can fall through and prints a "function
870declared 'noreturn' should not return" warning.
871</p>
872
873<p><b>Syntax:</b></p>
874
875<pre>
876__builtin_unreachable()
877</pre>
878
879<p><b>Example of Use:</b></p>
880
881<pre>
882void myabort(void) __attribute__((noreturn));
883void myabort(void) {
884 asm("int3");
885 __builtin_unreachable();
886}
887</pre>
888
889<p><b>Description:</b></p>
890
891<p>The __builtin_unreachable() builtin has completely undefined behavior. Since
892it has undefined behavior, it is a statement that it is never reached and the
893optimizer can take advantage of this to produce better code. This builtin takes
894no arguments and produces a void result.
895</p>
896
897<p>Query for this feature with __has_builtin(__builtin_unreachable).</p>
898
Chris Lattner23aa9c82011-04-09 03:57:26 +0000899<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000900<h3><a name="__sync_swap">__sync_swap</a></h3>
Chris Lattner23aa9c82011-04-09 03:57:26 +0000901<!-- ======================================================================= -->
902
903<p><tt>__sync_swap</tt> is used to atomically swap integers or pointers in
904memory.
905</p>
906
907<p><b>Syntax:</b></p>
908
909<pre>
910<i>type</i> __sync_swap(<i>type</i> *ptr, <i>type</i> value, ...)
911</pre>
912
913<p><b>Example of Use:</b></p>
914
915<pre>
Sean Hunt7e98b472011-06-23 01:21:01 +0000916int old_value = __sync_swap(&amp;value, new_value);
Chris Lattner23aa9c82011-04-09 03:57:26 +0000917</pre>
918
919<p><b>Description:</b></p>
920
921<p>The __sync_swap() builtin extends the existing __sync_*() family of atomic
922intrinsics to allow code to atomically swap the current value with the new
923value. More importantly, it helps developers write more efficient and correct
924code by avoiding expensive loops around __sync_bool_compare_and_swap() or
925relying on the platform specific implementation details of
926__sync_lock_test_and_set(). The __sync_swap() builtin is a full barrier.
927</p>
928
Chris Lattner21190d52009-09-21 03:09:59 +0000929
Chris Lattner1177f912009-04-09 19:58:15 +0000930<!-- ======================================================================= -->
931<h2 id="targetspecific">Target-Specific Extensions</h2>
932<!-- ======================================================================= -->
933
934<p>Clang supports some language features conditionally on some targets.</p>
935
936<!-- ======================================================================= -->
937<h3 id="x86-specific">X86/X86-64 Language Extensions</h3>
938<!-- ======================================================================= -->
939
940<p>The X86 backend has these language extensions:</p>
941
942<!-- ======================================================================= -->
943<h4 id="x86-gs-segment">Memory references off the GS segment</h4>
944<!-- ======================================================================= -->
945
946<p>Annotating a pointer with address space #256 causes it to be code generated
Chris Lattnera021e7c2009-05-05 18:54:47 +0000947relative to the X86 GS segment register, and address space #257 causes it to be
948relative to the X86 FS segment. Note that this is a very very low-level
949feature that should only be used if you know what you're doing (for example in
950an OS kernel).</p>
Chris Lattner1177f912009-04-09 19:58:15 +0000951
952<p>Here is an example:</p>
953
954<pre>
955#define GS_RELATIVE __attribute__((address_space(256)))
956int foo(int GS_RELATIVE *P) {
957 return *P;
958}
959</pre>
960
961<p>Which compiles to (on X86-32):</p>
962
963<pre>
964_foo:
965 movl 4(%esp), %eax
966 movl %gs:(%eax), %eax
967 ret
968</pre>
969
Ted Kremeneked869312009-04-10 05:03:33 +0000970<!-- ======================================================================= -->
971<h2 id="analyzerspecific">Static Analysis-Specific Extensions</h2>
972<!-- ======================================================================= -->
973
974<p>Clang supports additional attributes that are useful for documenting program
975invariants and rules for static analysis tools. The extensions documented here
976are used by the <a
977href="http://clang.llvm.org/StaticAnalysis.html">path-sensitive static analyzer
978engine</a> that is part of Clang's Analysis library.</p>
979
John McCall87494012011-03-18 03:51:49 +0000980<h3 id="attr_analyzer_noreturn">The <tt>analyzer_noreturn</tt> attribute</h3>
Ted Kremeneked869312009-04-10 05:03:33 +0000981
982<p>Clang's static analysis engine understands the standard <tt>noreturn</tt>
Ted Kremenek4df21142009-04-10 05:04:22 +0000983attribute. This attribute, which is typically affixed to a function prototype,
984indicates that a call to a given function never returns. Function prototypes for
985common functions like <tt>exit</tt> are typically annotated with this attribute,
986as well as a variety of common assertion handlers. Users can educate the static
987analyzer about their own custom assertion handles (thus cutting down on false
988positives due to false paths) by marking their own &quot;panic&quot; functions
989with this attribute.</p>
Ted Kremeneked869312009-04-10 05:03:33 +0000990
991<p>While useful, <tt>noreturn</tt> is not applicable in all cases. Sometimes
Nick Lewycky625b5862009-06-14 04:08:08 +0000992there are special functions that for all intents and purposes should be
993considered panic functions (i.e., they are only called when an internal program
994error occurs) but may actually return so that the program can fail gracefully.
995The <tt>analyzer_noreturn</tt> attribute allows one to annotate such functions
996as being interpreted as &quot;no return&quot; functions by the analyzer (thus
Chris Lattner28935892009-04-10 05:54:56 +0000997pruning bogus paths) but will not affect compilation (as in the case of
Ted Kremeneked869312009-04-10 05:03:33 +0000998<tt>noreturn</tt>).</p>
999
1000<p><b>Usage</b>: The <tt>analyzer_noreturn</tt> attribute can be placed in the
Chris Lattner28935892009-04-10 05:54:56 +00001001same places where the <tt>noreturn</tt> attribute can be placed. It is commonly
Ted Kremeneked869312009-04-10 05:03:33 +00001002placed at the end of function prototypes:</p>
1003
1004<pre>
1005 void foo() <b>__attribute__((analyzer_noreturn))</b>;
Chris Lattner148772a2009-06-13 07:13:28 +00001006</pre>
1007
John McCall87494012011-03-18 03:51:49 +00001008<p>Query for this feature with
1009<tt>__has_attribute(analyzer_noreturn)</tt>.</p>
Chris Lattner148772a2009-06-13 07:13:28 +00001010
John McCall87494012011-03-18 03:51:49 +00001011<h3 id="attr_method_family">The <tt>objc_method_family</tt> attribute</h3>
1012
1013<p>Many methods in Objective-C have conventional meanings determined
1014by their selectors. For the purposes of static analysis, it is
1015sometimes useful to be able to mark a method as having a particular
1016conventional meaning despite not having the right selector, or as not
1017having the conventional meaning that its selector would suggest.
1018For these use cases, we provide an attribute to specifically describe
1019the <q>method family</q> that a method belongs to.</p>
1020
1021<p><b>Usage</b>: <tt>__attribute__((objc_method_family(X)))</tt>,
1022where <tt>X</tt> is one of <tt>none</tt>, <tt>alloc</tt>, <tt>copy</tt>,
1023<tt>init</tt>, <tt>mutableCopy</tt>, or <tt>new</tt>. This attribute
1024can only be placed at the end of a method declaration:</p>
1025
1026<pre>
1027 - (NSString*) initMyStringValue <b>__attribute__((objc_method_family(none)))</b>;
1028</pre>
1029
1030<p>Users who do not wish to change the conventional meaning of a
1031method, and who merely want to document its non-standard retain and
1032release semantics, should use the
1033<a href="#attr_retain_release">retaining behavior attributes</a>
1034described below.</p>
1035
1036<p>Query for this feature with
1037<tt>__has_attribute(objc_method_family)</tt>.</p>
1038
1039<h3 id="attr_retain_release">Objective-C retaining behavior attributes</h3>
John McCall630b7ae2011-01-25 04:26:21 +00001040
1041<p>In Objective-C, functions and methods are generally assumed to take
1042and return objects with +0 retain counts, with some exceptions for
1043special methods like <tt>+alloc</tt> and <tt>init</tt>. However,
1044there are exceptions, and so Clang provides attributes to allow these
1045exceptions to be documented, which helps the analyzer find leaks (and
John McCall87494012011-03-18 03:51:49 +00001046ignore non-leaks). Some exceptions may be better described using
1047the <a href="#attr_method_family"><tt>objc_method_family</tt></a>
1048attribute instead.</p>
John McCall630b7ae2011-01-25 04:26:21 +00001049
1050<p><b>Usage</b>: The <tt>ns_returns_retained</tt>, <tt>ns_returns_not_retained</tt>,
1051<tt>ns_returns_autoreleased</tt>, <tt>cf_returns_retained</tt>,
1052and <tt>cf_returns_not_retained</tt> attributes can be placed on
1053methods and functions that return Objective-C or CoreFoundation
1054objects. They are commonly placed at the end of a function prototype
1055or method declaration:</p>
1056
1057<pre>
1058 id foo() <b>__attribute__((ns_returns_retained))</b>;
1059
1060 - (NSString*) bar: (int) x <b>__attribute__((ns_returns_retained))</b>;
1061</pre>
1062
1063<p>The <tt>*_returns_retained</tt> attributes specify that the
1064returned object has a +1 retain count.
1065The <tt>*_returns_not_retained</tt> attributes specify that the return
1066object has a +0 retain count, even if the normal convention for its
1067selector would be +1. <tt>ns_returns_autoreleased</tt> specifies that the
1068returned object is +0, but is guaranteed to live at least as long as the
1069next flush of an autorelease pool.</p>
1070
1071<p><b>Usage</b>: The <tt>ns_consumed</tt> and <tt>cf_consumed</tt>
1072attributes can be placed on an parameter declaration; they specify
1073that the argument is expected to have a +1 retain count, which will be
1074balanced in some way by the function or method.
1075The <tt>ns_consumes_self</tt> attribute can only be placed on an
1076Objective-C method; it specifies that the method expects
1077its <tt>self</tt> parameter to have a +1 retain count, which it will
1078balance in some way.</p>
1079
1080<pre>
1081 void <b>foo(__attribute__((ns_consumed))</b> NSString *string);
1082
1083 - (void) bar <b>__attribute__((ns_consumes_self))</b>;
1084 - (void) baz: (id) <b>__attribute__((ns_consumed))</b> x;
1085</pre>
Ted Kremeneked869312009-04-10 05:03:33 +00001086
John McCall87494012011-03-18 03:51:49 +00001087<p>Query for these features with <tt>__has_attribute(ns_consumed)</tt>,
1088<tt>__has_attribute(ns_returns_retained)</tt>, etc.</p>
1089
Chris Lattner5ce933f2009-02-09 08:46:11 +00001090</div>
1091</body>
1092</html>