Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / clang / 84ee2eeec9f63e4938bafd060105fa4f0c9cd89b / . / docs / LanguageExtensions.html
blob: bc8ecb1bc937fef24ee9c44ece71fe481628b23e [file] [log] [blame]
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]1<html>
2<head>
3<title>Clang Language Extensions</title>
4<link type="text/css" rel="stylesheet" href="../menu.css" />
5<link type="text/css" rel="stylesheet" href="../content.css" />
6<style type="text/css">
7td {
8 vertical-align: top;
9}
10</style>
11</head>
12<body>
13
14<!--#include virtual="../menu.html.incl"-->
15
16<div id="content">
17
18<h1>Clang Language Extensions</h1>
19
20<ul>
21<li><a href="#intro">Introduction</a></li>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]22<li><a href="#feature_check">Feature Checking Macros</a></li>
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]23<li><a href="#has_include">Include File Checking Macros</a></li>
Chris Lattner81edc9f2009-04-13 02:45:46 +0000[diff] [blame]24<li><a href="#builtinmacros">Builtin Macros</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]25<li><a href="#vectors">Vectors and Extended Vectors</a></li>
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]26<li><a href="#deprecated">Messages on <tt>deprecated</tt> and <tt>unavailable</tt> attributes</a></li>
27<li><a href="#attributes-on-enumerators">Attributes on enumerators</a></li>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]28<li><a href="#checking_language_features">Checks for Standard Language Features</a></li>
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]29 <ul>
30 <li><a href="#cxx_exceptions">C++ exceptions</a></li>
31 <li><a href="#cxx_rtti">C++ RTTI</a></li>
32 </ul>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]33<li><a href="#checking_upcoming_features">Checks for Upcoming Standard Language Features</a></li>
34 <ul>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]35 <li><a href="#cxx0x">C++0x</a>
36 <ul>
37 <li><a href="#cxx_decltype">C++0x <tt>decltype()</tt></a></li>
38 <li><a href="#cxx_access_control_sfinae">C++0x SFINAE includes access control</a></li>
39 <li><a href="#cxx_alias_templates">C++0x alias templates</a></li>
40 <li><a href="#cxx_attributes">C++0x attributes</a></li>
41 <li><a href="#cxx_default_function_template_args">C++0x default template arguments in function templates</a></li>
42 <li><a href="#cxx_deleted_functions">C++0x deleted functions</a></li>
43 <li><a href="#cxx_lambdas">C++0x lambdas</a></li>
44 <li><a href="#cxx_nullptr">C++0x nullptr</a></li>
45 <li><a href="#cxx_override_control">C++0x override control</a></li>
46 <li><a href="#cxx_range_for">C++0x range-based for loop</a></li>
47 <li><a href="#cxx_rvalue_references">C++0x rvalue references</a></li>
48 <li><a href="#cxx_reference_qualified_functions">C++0x reference-qualified functions</a></li>
49 <li><a href="#cxx_static_assert">C++0x <tt>static_assert()</tt></a></li>
50 <li><a href="#cxx_auto_type">C++0x type inference</a></li>
51 <li><a href="#cxx_variadic_templates">C++0x variadic templates</a></li>
52 <li><a href="#cxx_inline_namespaces">C++0x inline namespaces</a></li>
53 <li><a href="#cxx_strong_enums">C++0x strongly-typed enumerations</a></li>
54 <li><a href="#cxx_trailing_return">C++0x trailing return type</a></li>
55 <li><a href="#cxx_noexcept">C++0x noexcept specification</a></li>
56 </ul>
57 <li><a href="#c1x">C1X</a>
58 <ul>
59 <li><a href="#c_generic_selections">C1X generic selections</a></li>
60 <li><a href="#c_static_assert">C1X <tt>_Static_assert()</tt></a></li>
61 </ul>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]62 </ul>
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]63<li><a href="#checking_type_traits">Checks for Type Traits</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]64<li><a href="#blocks">Blocks</a></li>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]65<li><a href="#overloading-in-c">Function Overloading in C</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]66<li><a href="#builtins">Builtin Functions</a>
67 <ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]68 <li><a href="#__builtin_shufflevector">__builtin_shufflevector</a></li>
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]69 <li><a href="#__builtin_unreachable">__builtin_unreachable</a></li>
Chris Lattner23aa9c82011-04-09 03:57:26 +0000[diff] [blame]70 <li><a href="#__sync_swap">__sync_swap</a></li>
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]71 </ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]72</li>
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]73<li><a href="#targetspecific">Target-Specific Extensions</a>
74 <ul>
75 <li><a href="#x86-specific">X86/X86-64 Language Extensions</a></li>
76 </ul>
77</li>
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]78<li><a href="#analyzerspecific">Static Analysis-Specific Extensions</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]79</ul>
80
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]81<!-- ======================================================================= -->
82<h2 id="intro">Introduction</h2>
83<!-- ======================================================================= -->
84
85<p>This document describes the language extensions provided by Clang. In
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]86addition to the language extensions listed here, Clang aims to support a broad
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]87range of GCC extensions. Please see the <a
88href="http://gcc.gnu.org/onlinedocs/gcc/C-Extensions.html">GCC manual</a> for
89more information on these extensions.</p>
90
91<!-- ======================================================================= -->
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]92<h2 id="feature_check">Feature Checking Macros</h2>
93<!-- ======================================================================= -->
94
95<p>Language extensions can be very useful, but only if you know you can depend
Chris Lattnerc70e1932011-03-21 16:25:11 +0000[diff] [blame]96on them. In order to allow fine-grain features checks, we support three builtin
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]97function-like macros. This allows you to directly test for a feature in your
98code without having to resort to something like autoconf or fragile "compiler
99version checks".</p>
100
101<!-- ======================================================================= -->
102<h3 id="__has_builtin">__has_builtin</h3>
103<!-- ======================================================================= -->
104
105<p>This function-like macro takes a single identifier argument that is the name
106of a builtin function. It evaluates to 1 if the builtin is supported or 0 if
107not. It can be used like this:</p>
108
109<blockquote>
110<pre>
111#ifndef __has_builtin // Optional of course.
112 #define __has_builtin(x) 0 // Compatibility with non-clang compilers.
113#endif
114
115...
116#if __has_builtin(__builtin_trap)
117 __builtin_trap();
118#else
119 abort();
120#endif
121...
122</pre>
123</blockquote>
124
125
126<!-- ======================================================================= -->
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]127<h3 id="__has_feature_extension">__has_feature and __has_extension</h3>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]128<!-- ======================================================================= -->
129
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]130<p>These function-like macros take a single identifier argument that is the
131name of a feature. <code>__has_feature</code> evaluates to 1 if the feature
132is both supported by Clang and standardized in the current language standard
133or 0 if not (but see <a href="#has_feature_back_compat">below</a>), while
134<code>__has_extension</code> evaluates to 1 if the feature is supported by
135Clang in the current language (either as a language extension or a standard
136language feature) or 0 if not. They can be used like this:</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]137
138<blockquote>
139<pre>
140#ifndef __has_feature // Optional of course.
141 #define __has_feature(x) 0 // Compatibility with non-clang compilers.
142#endif
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]143#ifndef __has_extension
144 #define __has_extension __has_feature // Compatibility with pre-3.0 compilers.
145#endif
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]146
147...
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]148#if __has_feature(cxx_rvalue_references)
149// This code will only be compiled with the -std=c++0x and -std=gnu++0x
150// options, because rvalue references are only standardized in C++0x.
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]151#endif
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]152
153#if __has_extension(cxx_rvalue_references)
154// This code will be compiled with the -std=c++0x, -std=gnu++0x, -std=c++98
155// and -std=gnu++98 options, because rvalue references are supported as a
156// language extension in C++98.
157#endif
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]158</pre>
159</blockquote>
160
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]161<p id="has_feature_back_compat">For backwards compatibility reasons,
162<code>__has_feature</code> can also be used to test for support for
163non-standardized features, i.e. features not prefixed <code>c_</code>,
164<code>cxx_</code> or <code>objc_</code>.</p>
165
166<p>If the <code>-pedantic-errors</code> option is given,
167<code>__has_extension</code> is equivalent to <code>__has_feature</code>.</p>
168
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]169<p>The feature tag is described along with the language feature below.</p>
170
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]171<!-- ======================================================================= -->
Anders Carlssoncae50952010-10-20 02:31:43 +0000[diff] [blame]172<h3 id="__has_attribute">__has_attribute</h3>
173<!-- ======================================================================= -->
174
175<p>This function-like macro takes a single identifier argument that is the name
176of an attribute. It evaluates to 1 if the attribute is supported or 0 if not. It
177can be used like this:</p>
178
179<blockquote>
180<pre>
181#ifndef __has_attribute // Optional of course.
182 #define __has_attribute(x) 0 // Compatibility with non-clang compilers.
183#endif
184
185...
Anders Carlsson961003d2011-01-24 03:54:51 +0000[diff] [blame]186#if __has_attribute(always_inline)
187#define ALWAYS_INLINE __attribute__((always_inline))
Anders Carlssoncae50952010-10-20 02:31:43 +0000[diff] [blame]188#else
Anders Carlsson961003d2011-01-24 03:54:51 +0000[diff] [blame]189#define ALWAYS_INLINE
Anders Carlssoncae50952010-10-20 02:31:43 +0000[diff] [blame]190#endif
191...
192</pre>
193</blockquote>
194
195<!-- ======================================================================= -->
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]196<h2 id="has_include">Include File Checking Macros</h2>
197<!-- ======================================================================= -->
198
199<p>Not all developments systems have the same include files.
200The <a href="#__has_include">__has_include</a> and
201<a href="#__has_include_next">__has_include_next</a> macros allow you to
202check for the existence of an include file before doing
203a possibly failing #include directive.</p>
204
205<!-- ======================================================================= -->
206<h3 id="__has_include">__has_include</h3>
207<!-- ======================================================================= -->
208
209<p>This function-like macro takes a single file name string argument that
210is the name of an include file. It evaluates to 1 if the file can
211be found using the include paths, or 0 otherwise:</p>
212
213<blockquote>
214<pre>
215// Note the two possible file name string formats.
216#if __has_include("myinclude.h") && __has_include(&lt;stdint.h&gt;)
217# include "myinclude.h"
218#endif
219
220// To avoid problem with non-clang compilers not having this macro.
221#if defined(__has_include) && __has_include("myinclude.h")
222# include "myinclude.h"
223#endif
224</pre>
225</blockquote>
226
227<p>To test for this feature, use #if defined(__has_include).</p>
228
229<!-- ======================================================================= -->
230<h3 id="__has_include_next">__has_include_next</h3>
231<!-- ======================================================================= -->
232
233<p>This function-like macro takes a single file name string argument that
234is the name of an include file. It is like __has_include except that it
235looks for the second instance of the given file found in the include
236paths. It evaluates to 1 if the second instance of the file can
237be found using the include paths, or 0 otherwise:</p>
238
239<blockquote>
240<pre>
241// Note the two possible file name string formats.
242#if __has_include_next("myinclude.h") && __has_include_next(&lt;stdint.h&gt;)
243# include_next "myinclude.h"
244#endif
245
246// To avoid problem with non-clang compilers not having this macro.
247#if defined(__has_include_next) && __has_include_next("myinclude.h")
248# include_next "myinclude.h"
249#endif
250</pre>
251</blockquote>
252
253<p>Note that __has_include_next, like the GNU extension
254#include_next directive, is intended for use in headers only,
255and will issue a warning if used in the top-level compilation
256file. A warning will also be issued if an absolute path
257is used in the file argument.</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]258
259<!-- ======================================================================= -->
Chris Lattner81edc9f2009-04-13 02:45:46 +0000[diff] [blame]260<h2 id="builtinmacros">Builtin Macros</h2>
261<!-- ======================================================================= -->
262
Douglas Gregor4290fbd2010-04-30 02:51:06 +0000[diff] [blame]263<dl>
264 <dt><code>__BASE_FILE__</code></dt>
265 <dd>Defined to a string that contains the name of the main input
266 file passed to Clang.</dd>
267
268 <dt><code>__COUNTER__</code></dt>
269 <dd>Defined to an integer value that starts at zero and is
270 incremented each time the <code>__COUNTER__</code> macro is
271 expanded.</dd>
272
273 <dt><code>__INCLUDE_LEVEL__</code></dt>
274 <dd>Defined to an integral value that is the include depth of the
275 file currently being translated. For the main file, this value is
276 zero.</dd>
277
278 <dt><code>__TIMESTAMP__</code></dt>
279 <dd>Defined to the date and time of the last modification of the
280 current source file.</dd>
281
282 <dt><code>__clang__</code></dt>
283 <dd>Defined when compiling with Clang</dd>
284
285 <dt><code>__clang_major__</code></dt>
286 <dd>Defined to the major version number of Clang (e.g., the 2 in
287 2.0.1).</dd>
288
289 <dt><code>__clang_minor__</code></dt>
290 <dd>Defined to the minor version number of Clang (e.g., the 0 in
291 2.0.1).</dd>
292
293 <dt><code>__clang_patchlevel__</code></dt>
294 <dd>Defined to the patch level of Clang (e.g., the 1 in 2.0.1).</dd>
295
296 <dt><code>__clang_version__</code></dt>
297 <dd>Defined to a string that captures the Clang version, including
298 the Subversion tag or revision number, e.g., "1.5 (trunk
299 102332)".</dd>
300</dl>
Chris Lattner81edc9f2009-04-13 02:45:46 +0000[diff] [blame]301
302<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]303<h2 id="vectors">Vectors and Extended Vectors</h2>
304<!-- ======================================================================= -->
305
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]306<p>Supports the GCC vector extensions, plus some stuff like V[1].</p>
307
308<p>Also supports <tt>ext_vector</tt>, which additionally support for V.xyzw
309syntax and other tidbits as seen in OpenCL. An example is:</p>
310
311<blockquote>
312<pre>
313typedef float float4 <b>__attribute__((ext_vector_type(4)))</b>;
314typedef float float2 <b>__attribute__((ext_vector_type(2)))</b>;
315
316float4 foo(float2 a, float2 b) {
317 float4 c;
318 c.xz = a;
319 c.yw = b;
320 return c;
321}
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]322</pre>
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]323</blockquote>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]324
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]325<p>Query for this feature with __has_extension(attribute_ext_vector_type).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]326
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]327<p>See also <a href="#__builtin_shufflevector">__builtin_shufflevector</a>.</p>
328
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]329<!-- ======================================================================= -->
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]330<h2 id="deprecated">Messages on <tt>deprecated</tt> and <tt>unavailable</tt> Attributes</h2>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000[diff] [blame]331<!-- ======================================================================= -->
332
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]333<p>An optional string message can be added to the <tt>deprecated</tt>
334and <tt>unavailable</tt> attributes. For example:</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000[diff] [blame]335
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]336<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000[diff] [blame]337<pre>void explode(void) __attribute__((deprecated("extremely unsafe, use 'combust' instead!!!")));</pre>
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]338</blockquote>
339
340<p>If the deprecated or unavailable declaration is used, the message
341will be incorporated into the appropriate diagnostic:</p>
342
343<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000[diff] [blame]344<pre>harmless.c:4:3: warning: 'explode' is deprecated: extremely unsafe, use 'combust' instead!!! [-Wdeprecated-declarations]
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]345 explode();
346 ^</pre>
347</blockquote>
348
349<p>Query for this feature
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]350with <tt>__has_extension(attribute_deprecated_with_message)</tt>
351and <tt>__has_extension(attribute_unavailable_with_message)</tt>.</p>
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]352
353<!-- ======================================================================= -->
354<h2 id="attributes-on-enumerators">Attributes on Enumerators</h2>
355<!-- ======================================================================= -->
356
357<p>Clang allows attributes to be written on individual enumerators.
358This allows enumerators to be deprecated, made unavailable, etc. The
359attribute must appear after the enumerator name and before any
360initializer, like so:</p>
361
362<blockquote>
363<pre>enum OperationMode {
364 OM_Invalid,
365 OM_Normal,
366 OM_Terrified __attribute__((deprecated)),
367 OM_AbortOnError __attribute__((deprecated)) = 4
368};</pre>
369</blockquote>
370
371<p>Attributes on the <tt>enum</tt> declaration do not apply to
372individual enumerators.</p>
373
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]374<p>Query for this feature with <tt>__has_extension(enumerator_attributes)</tt>.</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000[diff] [blame]375
376<!-- ======================================================================= -->
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]377<h2 id="checking_language_features">Checks for Standard Language Features</h2>
378<!-- ======================================================================= -->
379
380<p>The <tt>__has_feature</tt> macro can be used to query if certain standard language features are
381enabled. Those features are listed here.</p>
382
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]383<h3 id="cxx_exceptions">C++ exceptions</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]384
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]385<p>Use <tt>__has_feature(cxx_exceptions)</tt> to determine if C++ exceptions have been enabled. For
386example, compiling code with <tt>-fexceptions</tt> enables C++ exceptions.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]387
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]388<h3 id="cxx_rtti">C++ RTTI</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]389
Ted Kremenek0eb95602009-12-03 02:06:43 +0000[diff] [blame]390<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 +0000[diff] [blame]391compiling code with <tt>-fno-rtti</tt> disables the use of RTTI.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]392
393<!-- ======================================================================= -->
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]394<h2 id="checking_upcoming_features">Checks for Upcoming Standard Language Features</h2>
395<!-- ======================================================================= -->
396
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]397<p>The <tt>__has_feature</tt> or <tt>__has_extension</tt> macros can be used
398to query if certain upcoming standard language features are enabled. Those
399features are listed here. Features that are not yet implemented will be
400noted.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]401
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]402<h3 id="cxx0x">C++0x</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]403
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]404<p>The features listed below are slated for inclusion in the upcoming
405C++0x standard. As a result, all these features are enabled
406with the <tt>-std=c++0x</tt> option when compiling C++ code.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]407
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]408<h4 id="cxx_decltype">C++0x <tt>decltype()</tt></h3>
409
410<p>Use <tt>__has_feature(cxx_decltype)</tt> or
411<tt>__has_extension(cxx_decltype)</tt> to determine if support for the
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]412<tt>decltype()</tt> specifier is enabled.</p>
413
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]414<h4 id="cxx_access_control_sfinae">C++0x SFINAE includes access control</h3>
Douglas Gregor7822ee32011-05-11 23:45:11 +0000[diff] [blame]415
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]416<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 +0000[diff] [blame]417
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]418<h4 id="cxx_alias_templates">C++0x alias templates</h3>
Richard Smith3e4c6c42011-05-05 21:57:07 +0000[diff] [blame]419
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]420<p>Use <tt>__has_feature(cxx_alias_templates)</tt> or
421<tt>__has_extension(cxx_alias_templates)</tt> to determine if support for
Richard Smith3e4c6c42011-05-05 21:57:07 +0000[diff] [blame]422C++0x's alias declarations and alias templates is enabled.</p>
423
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]424<h4 id="cxx_attributes">C++0x attributes</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]425
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]426<p>Use <tt>__has_feature(cxx_attributes)</tt> or
427<tt>__has_extension(cxx_attributes)</tt> to determine if support for attribute
428parsing with C++0x's square bracket notation is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]429
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]430<h4 id="cxx_default_function_template_args">C++0x default template arguments in function templates</h3>
Douglas Gregor07508002011-02-05 20:35:30 +0000[diff] [blame]431
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]432<p>Use <tt>__has_feature(cxx_default_function_template_args)</tt> or
433<tt>__has_extension(cxx_default_function_template_args)</tt> to determine
434if support for default template arguments in function templates is enabled.</p>
Douglas Gregor07508002011-02-05 20:35:30 +0000[diff] [blame]435
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]436<h4 id="cxx_deleted_functions">C++0x deleted functions</tt></h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]437
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]438<p>Use <tt>__has_feature(cxx_deleted_functions)</tt> or
439<tt>__has_extension(cxx_deleted_functions)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]440deleted function definitions (with <tt>= delete</tt>) is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]441
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]442<h4 id="cxx_lambdas">C++0x lambdas</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]443
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]444<p>Use <tt>__has_feature(cxx_lambdas)</tt> or
445<tt>__has_extension(cxx_lambdas)</tt> to determine if support for lambdas
446is enabled. clang does not currently implement this feature.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]447
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]448<h4 id="cxx_nullptr">C++0x <tt>nullptr</tt></h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]449
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]450<p>Use <tt>__has_feature(cxx_nullptr)</tt> or
451<tt>__has_extension(cxx_nullptr)</tt> to determine if support for
Douglas Gregor84ee2ee2011-05-21 23:15:46 +0000[diff] [blame^]452<tt>nullptr</tt> is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]453
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]454<h4 id="cxx_override_control">C++0x <tt>override control</tt></h3>
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000[diff] [blame]455
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]456<p>Use <tt>__has_feature(cxx_override_control)</tt> or
457<tt>__has_extension(cxx_override_control)</tt> to determine if support for
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000[diff] [blame]458the override control keywords is enabled.</p>
459
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]460<h4 id="cxx_reference_qualified_functions">C++0x reference-qualified functions</h3>
461<p>Use <tt>__has_feature(cxx_reference_qualified_functions)</tt> or
462<tt>__has_extension(cxx_reference_qualified_functions)</tt> to determine
463if support for reference-qualified functions (e.g., member functions with
464<code>&amp;</code> or <code>&amp;&amp;</code> applied to <code>*this</code>)
465is enabled.</p>
Douglas Gregor56209ff2011-01-26 21:25:54 +0000[diff] [blame]466
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]467<h4 id="cxx_range_for">C++0x range-based for loop</tt></h3>
Richard Smitha391a462011-04-15 15:14:40 +0000[diff] [blame]468
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]469<p>Use <tt>__has_feature(cxx_range_for)</tt> or
470<tt>__has_extension(cxx_range_for)</tt> to determine if support for the
471range-based for loop is enabled. </p>
Richard Smitha391a462011-04-15 15:14:40 +0000[diff] [blame]472
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]473<h4 id="cxx_rvalue_references">C++0x rvalue references</tt></h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]474
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]475<p>Use <tt>__has_feature(cxx_rvalue_references)</tt> or
476<tt>__has_extension(cxx_rvalue_references)</tt> to determine if support for
Douglas Gregor56209ff2011-01-26 21:25:54 +0000[diff] [blame]477rvalue references is enabled. </p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]478
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]479<h4 id="cxx_static_assert">C++0x <tt>static_assert()</tt></h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]480
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]481<p>Use <tt>__has_feature(cxx_static_assert)</tt> or
482<tt>__has_extension(cxx_static_assert)</tt> to determine if support for
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]483compile-time assertions using <tt>static_assert</tt> is enabled.</p>
484
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]485<h4 id="cxx_auto_type">C++0x type inference</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]486
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]487<p>Use <tt>__has_feature(cxx_auto_type)</tt> or
488<tt>__has_extension(cxx_auto_type)</tt> to determine C++0x type inference is
489supported using the <tt>auto</tt> specifier. If this is disabled, <tt>auto</tt>
490will instead be a storage class specifier, as in C or C++98.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]491
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]492<h4 id="cxx_variadic_templates">C++0x variadic templates</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]493
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]494<p>Use <tt>__has_feature(cxx_variadic_templates)</tt> or
495<tt>__has_extension(cxx_variadic_templates)</tt> to determine if support
Douglas Gregor83d77812011-01-19 23:15:20 +0000[diff] [blame]496for variadic templates is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]497
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]498<h4 id="cxx_inline_namespaces">C++0x inline namespaces</h3>
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]499
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]500<p>Use <tt>__has_feature(cxx_inline_namespaces)</tt> or
501<tt>__has_extension(cxx_inline_namespaces)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]502inline namespaces is enabled.</p>
503
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]504<h4 id="cxx_trailing_return">C++0x trailing return type</h3>
Douglas Gregordab60ad2010-10-01 18:44:50 +0000[diff] [blame]505
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]506<p>Use <tt>__has_feature(cxx_trailing_return)</tt> or
507<tt>__has_extension(cxx_trailing_return)</tt> to determine if support for the
508alternate function declaration syntax with trailing return type is enabled.</p>
Douglas Gregordab60ad2010-10-01 18:44:50 +0000[diff] [blame]509
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]510<h4 id="cxx_noexcept">C++0x noexcept</h3>
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000[diff] [blame]511
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]512<p>Use <tt>__has_feature(cxx_noexcept)</tt> or
513<tt>__has_extension(cxx_noexcept)</tt> to determine if support for noexcept
514exception specifications is enabled.</p>
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000[diff] [blame]515
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]516<h4 id="cxx_strong_enums">C++0x strongly typed enumerations</h3>
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000[diff] [blame]517
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]518<p>Use <tt>__has_feature(cxx_strong_enums)</tt> or
519<tt>__has_extension(cxx_strong_enums)</tt> to determine if support for
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000[diff] [blame]520strongly typed, scoped enumerations is enabled.</p>
521
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]522<h3 id="c1x">C1X</h3>
523
524<p>The features listed below are slated for inclusion in the upcoming
525C1X standard. As a result, all these features are enabled
526with the <tt>-std=c1x</tt> option when compiling C code.</p>
527
528<h4 id="c_generic_selections">C1X generic selections</h2>
529
530<p>Use <tt>__has_feature(c_generic_selections)</tt> or
531<tt>__has_extension(c_generic_selections)</tt> to determine if support for
532generic selections is enabled.</p>
533
534<p>As an extension, the C1X generic selection expression is available in all
535languages supported by Clang. The syntax is the same as that given in the
536C1X draft standard.</p>
537
538<p>In C, type compatibility is decided according to the rules given in the
539appropriate standard, but in C++, which lacks the type compatibility rules
540used in C, types are considered compatible only if they are equivalent.</p>
541
542<h4 id="c_static_assert">C1X <tt>_Static_assert()</tt></h3>
543
544<p>Use <tt>__has_feature(c_static_assert)</tt> or
545<tt>__has_extension(c_static_assert)</tt> to determine if support for
546compile-time assertions using <tt>_Static_assert</tt> is enabled.</p>
547
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]548<!-- ======================================================================= -->
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]549<h2 id="checking_type_traits">Checks for Type Traits</h2>
550<!-- ======================================================================= -->
551
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]552<p>Clang supports the <a hef="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 +0000[diff] [blame]553<blockquote>
554<pre>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]555#if __has_extension(is_convertible_to)
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]556template&lt;typename From, typename To&gt;
557struct is_convertible_to {
558 static const bool value = __is_convertible_to(From, To);
559};
560#else
561// Emulate type trait
562#endif
563</pre>
564</blockquote>
565
566<p>The following type traits are supported by Clang:</p>
567<ul>
568 <li><code>__has_nothrow_assign</code> (GNU, Microsoft)</li>
569 <li><code>__has_nothrow_copy</code> (GNU, Microsoft)</li>
570 <li><code>__has_nothrow_constructor</code> (GNU, Microsoft)</li>
571 <li><code>__has_trivial_assign</code> (GNU, Microsoft)</li>
572 <li><code>__has_trivial_copy</code> (GNU, Microsoft)</li>
573 <li><code>__has_trivial_constructor</code> (GNU, Microsoft)</li>
574 <li><code>__has_trivial_destructor</code> (GNU, Microsoft)</li>
575 <li><code>__has_virtual_destructor</code> (GNU, Microsoft)</li>
576 <li><code>__is_abstract</code> (GNU, Microsoft)</li>
577 <li><code>__is_base_of</code> (GNU, Microsoft)</li>
578 <li><code>__is_class</code> (GNU, Microsoft)</li>
579 <li><code>__is_convertible_to</code> (Microsoft)</li>
580 <li><code>__is_empty</code> (GNU, Microsoft)</li>
581 <li><code>__is_enum</code> (GNU, Microsoft)</li>
582 <li><code>__is_pod</code> (GNU, Microsoft)</li>
583 <li><code>__is_polymorphic</code> (GNU, Microsoft)</li>
584 <li><code>__is_union</code> (GNU, Microsoft)</li>
585 <li><code>__is_literal(type)</code>: Determines whether the given type is a literal type</li>
586</ul>
587
588<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]589<h2 id="blocks">Blocks</h2>
590<!-- ======================================================================= -->
591
Chris Lattnera7dbdf52009-03-09 07:03:22 +0000[diff] [blame]592<p>The syntax and high level language feature description is in <a
593href="BlockLanguageSpec.txt">BlockLanguageSpec.txt</a>. Implementation and ABI
594details for the clang implementation are in <a
Chris Lattner5d7650b2010-03-16 21:43:03 +0000[diff] [blame]595href="Block-ABI-Apple.txt">Block-ABI-Apple.txt</a>.</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]596
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]597
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]598<p>Query for this feature with __has_extension(blocks).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]599
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]600<!-- ======================================================================= -->
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]601<h2 id="overloading-in-c">Function Overloading in C</h2>
602<!-- ======================================================================= -->
603
Chris Lattnerf161d412009-02-13 21:51:45 +0000[diff] [blame]604<p>Clang provides support for C++ function overloading in C. Function
605overloading in C is introduced using the <tt>overloadable</tt> attribute. For
606example, one might provide several overloaded versions of a <tt>tgsin</tt>
607function that invokes the appropriate standard function computing the sine of a
608value with <tt>float</tt>, <tt>double</tt>, or <tt>long double</tt>
609precision:</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]610
611<blockquote>
612<pre>
613#include &lt;math.h&gt;
614float <b>__attribute__((overloadable))</b> tgsin(float x) { return sinf(x); }
615double <b>__attribute__((overloadable))</b> tgsin(double x) { return sin(x); }
616long double <b>__attribute__((overloadable))</b> tgsin(long double x) { return sinl(x); }
617</pre>
618</blockquote>
619
620<p>Given these declarations, one can call <tt>tgsin</tt> with a
621<tt>float</tt> value to receive a <tt>float</tt> result, with a
622<tt>double</tt> to receive a <tt>double</tt> result, etc. Function
623overloading in C follows the rules of C++ function overloading to pick
624the best overload given the call arguments, with a few C-specific
625semantics:</p>
626<ul>
627 <li>Conversion from <tt>float</tt> or <tt>double</tt> to <tt>long
628 double</tt> is ranked as a floating-point promotion (per C99) rather
629 than as a floating-point conversion (as in C++).</li>
630
631 <li>A conversion from a pointer of type <tt>T*</tt> to a pointer of type
632 <tt>U*</tt> is considered a pointer conversion (with conversion
633 rank) if <tt>T</tt> and <tt>U</tt> are compatible types.</li>
634
635 <li>A conversion from type <tt>T</tt> to a value of type <tt>U</tt>
636 is permitted if <tt>T</tt> and <tt>U</tt> are compatible types. This
637 conversion is given "conversion" rank.</li>
638</ul>
639
640<p>The declaration of <tt>overloadable</tt> functions is restricted to
641function declarations and definitions. Most importantly, if any
642function with a given name is given the <tt>overloadable</tt>
643attribute, then all function declarations and definitions with that
644name (and in that scope) must have the <tt>overloadable</tt>
Chris Lattnerf161d412009-02-13 21:51:45 +0000[diff] [blame]645attribute. This rule even applies to redeclarations of functions whose original
646declaration had the <tt>overloadable</tt> attribute, e.g.,</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]647
648<blockquote>
649<pre>
650int f(int) __attribute__((overloadable));
651float f(float); <i>// error: declaration of "f" must have the "overloadable" attribute</i>
652
653int g(int) __attribute__((overloadable));
654int g(int) { } <i>// error: redeclaration of "g" must also have the "overloadable" attribute</i>
655</pre>
656</blockquote>
657
Douglas Gregor965acbb2009-02-18 07:07:28 +0000[diff] [blame]658<p>Functions marked <tt>overloadable</tt> must have
659prototypes. Therefore, the following code is ill-formed:</p>
660
661<blockquote>
662<pre>
663int h() __attribute__((overloadable)); <i>// error: h does not have a prototype</i>
664</pre>
665</blockquote>
666
667<p>However, <tt>overloadable</tt> functions are allowed to use a
668ellipsis 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>
669
670<blockquote>
671<pre>
Chris Lattner02246802009-02-18 22:27:46 +0000[diff] [blame]672void honeypot(...) __attribute__((overloadable, unavailable)); <i>// calling me is an error</i>
Douglas Gregor965acbb2009-02-18 07:07:28 +0000[diff] [blame]673</pre>
674</blockquote>
675
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]676<p>Functions declared with the <tt>overloadable</tt> attribute have
677their names mangled according to the same rules as C++ function
678names. For example, the three <tt>tgsin</tt> functions in our
679motivating example get the mangled names <tt>_Z5tgsinf</tt>,
Chris Lattner71b48d62010-11-28 18:19:13 +0000[diff] [blame]680<tt>_Z5tgsind</tt>, and <tt>_Z5tgsine</tt>, respectively. There are two
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]681caveats to this use of name mangling:</p>
682
683<ul>
684
685 <li>Future versions of Clang may change the name mangling of
686 functions overloaded in C, so you should not depend on an specific
687 mangling. To be completely safe, we strongly urge the use of
688 <tt>static inline</tt> with <tt>overloadable</tt> functions.</li>
689
690 <li>The <tt>overloadable</tt> attribute has almost no meaning when
691 used in C++, because names will already be mangled and functions are
692 already overloadable. However, when an <tt>overloadable</tt>
693 function occurs within an <tt>extern "C"</tt> linkage specification,
694 it's name <i>will</i> be mangled in the same way as it would in
695 C.</li>
696</ul>
697
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]698<p>Query for this feature with __has_extension(attribute_overloadable).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]699
700
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]701<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]702<h2 id="builtins">Builtin Functions</h2>
703<!-- ======================================================================= -->
704
705<p>Clang supports a number of builtin library functions with the same syntax as
706GCC, including things like <tt>__builtin_nan</tt>,
707<tt>__builtin_constant_p</tt>, <tt>__builtin_choose_expr</tt>,
708<tt>__builtin_types_compatible_p</tt>, <tt>__sync_fetch_and_add</tt>, etc. In
709addition to the GCC builtins, Clang supports a number of builtins that GCC does
710not, which are listed here.</p>
711
712<p>Please note that Clang does not and will not support all of the GCC builtins
713for vector operations. Instead of using builtins, you should use the functions
714defined in target-specific header files like <tt>&lt;xmmintrin.h&gt;</tt>, which
715define portable wrappers for these. Many of the Clang versions of these
716functions are implemented directly in terms of <a href="#vectors">extended
717vector support</a> instead of builtins, in order to reduce the number of
718builtins that we need to implement.</p>
719
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]720<!-- ======================================================================= -->
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]721<h3 id="__builtin_shufflevector">__builtin_shufflevector</h3>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]722<!-- ======================================================================= -->
723
Chris Lattneraad826b2009-09-16 18:56:12 +0000[diff] [blame]724<p><tt>__builtin_shufflevector</tt> is used to express generic vector
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]725permutation/shuffle/swizzle operations. This builtin is also very important for
726the implementation of various target-specific header files like
727<tt>&lt;xmmintrin.h&gt;</tt>.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]728</p>
729
730<p><b>Syntax:</b></p>
731
732<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]733__builtin_shufflevector(vec1, vec2, index1, index2, ...)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]734</pre>
735
736<p><b>Examples:</b></p>
737
738<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]739 // Identity operation - return 4-element vector V1.
740 __builtin_shufflevector(V1, V1, 0, 1, 2, 3)
741
742 // "Splat" element 0 of V1 into a 4-element result.
743 __builtin_shufflevector(V1, V1, 0, 0, 0, 0)
744
745 // Reverse 4-element vector V1.
746 __builtin_shufflevector(V1, V1, 3, 2, 1, 0)
747
748 // Concatenate every other element of 4-element vectors V1 and V2.
749 __builtin_shufflevector(V1, V2, 0, 2, 4, 6)
750
751 // Concatenate every other element of 8-element vectors V1 and V2.
752 __builtin_shufflevector(V1, V2, 0, 2, 4, 6, 8, 10, 12, 14)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]753</pre>
754
755<p><b>Description:</b></p>
756
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]757<p>The first two arguments to __builtin_shufflevector are vectors that have the
758same element type. The remaining arguments are a list of integers that specify
759the elements indices of the first two vectors that should be extracted and
760returned in a new vector. These element indices are numbered sequentially
761starting with the first vector, continuing into the second vector. Thus, if
762vec1 is a 4-element vector, index 5 would refer to the second element of vec2.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]763</p>
764
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]765<p>The result of __builtin_shufflevector is a vector
766with the same element type as vec1/vec2 but that has an element count equal to
767the number of indices specified.
768</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]769
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]770<p>Query for this feature with __has_builtin(__builtin_shufflevector).</p>
771
772<!-- ======================================================================= -->
773<h3 id="__builtin_unreachable">__builtin_unreachable</h3>
774<!-- ======================================================================= -->
775
776<p><tt>__builtin_unreachable</tt> is used to indicate that a specific point in
777the program cannot be reached, even if the compiler might otherwise think it
778can. This is useful to improve optimization and eliminates certain warnings.
779For example, without the <tt>__builtin_unreachable</tt> in the example below,
780the compiler assumes that the inline asm can fall through and prints a "function
781declared 'noreturn' should not return" warning.
782</p>
783
784<p><b>Syntax:</b></p>
785
786<pre>
787__builtin_unreachable()
788</pre>
789
790<p><b>Example of Use:</b></p>
791
792<pre>
793void myabort(void) __attribute__((noreturn));
794void myabort(void) {
795 asm("int3");
796 __builtin_unreachable();
797}
798</pre>
799
800<p><b>Description:</b></p>
801
802<p>The __builtin_unreachable() builtin has completely undefined behavior. Since
803it has undefined behavior, it is a statement that it is never reached and the
804optimizer can take advantage of this to produce better code. This builtin takes
805no arguments and produces a void result.
806</p>
807
808<p>Query for this feature with __has_builtin(__builtin_unreachable).</p>
809
Chris Lattner23aa9c82011-04-09 03:57:26 +0000[diff] [blame]810<!-- ======================================================================= -->
811<h3 id="__sync_swap">__sync_swap</h3>
812<!-- ======================================================================= -->
813
814<p><tt>__sync_swap</tt> is used to atomically swap integers or pointers in
815memory.
816</p>
817
818<p><b>Syntax:</b></p>
819
820<pre>
821<i>type</i> __sync_swap(<i>type</i> *ptr, <i>type</i> value, ...)
822</pre>
823
824<p><b>Example of Use:</b></p>
825
826<pre>
827int old_value = __sync_swap(&value, new_value);
828</pre>
829
830<p><b>Description:</b></p>
831
832<p>The __sync_swap() builtin extends the existing __sync_*() family of atomic
833intrinsics to allow code to atomically swap the current value with the new
834value. More importantly, it helps developers write more efficient and correct
835code by avoiding expensive loops around __sync_bool_compare_and_swap() or
836relying on the platform specific implementation details of
837__sync_lock_test_and_set(). The __sync_swap() builtin is a full barrier.
838</p>
839
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]840
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]841<!-- ======================================================================= -->
842<h2 id="targetspecific">Target-Specific Extensions</h2>
843<!-- ======================================================================= -->
844
845<p>Clang supports some language features conditionally on some targets.</p>
846
847<!-- ======================================================================= -->
848<h3 id="x86-specific">X86/X86-64 Language Extensions</h3>
849<!-- ======================================================================= -->
850
851<p>The X86 backend has these language extensions:</p>
852
853<!-- ======================================================================= -->
854<h4 id="x86-gs-segment">Memory references off the GS segment</h4>
855<!-- ======================================================================= -->
856
857<p>Annotating a pointer with address space #256 causes it to be code generated
Chris Lattnera021e7c2009-05-05 18:54:47 +0000[diff] [blame]858relative to the X86 GS segment register, and address space #257 causes it to be
859relative to the X86 FS segment. Note that this is a very very low-level
860feature that should only be used if you know what you're doing (for example in
861an OS kernel).</p>
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]862
863<p>Here is an example:</p>
864
865<pre>
866#define GS_RELATIVE __attribute__((address_space(256)))
867int foo(int GS_RELATIVE *P) {
868 return *P;
869}
870</pre>
871
872<p>Which compiles to (on X86-32):</p>
873
874<pre>
875_foo:
876 movl 4(%esp), %eax
877 movl %gs:(%eax), %eax
878 ret
879</pre>
880
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]881<!-- ======================================================================= -->
882<h2 id="analyzerspecific">Static Analysis-Specific Extensions</h2>
883<!-- ======================================================================= -->
884
885<p>Clang supports additional attributes that are useful for documenting program
886invariants and rules for static analysis tools. The extensions documented here
887are used by the <a
888href="http://clang.llvm.org/StaticAnalysis.html">path-sensitive static analyzer
889engine</a> that is part of Clang's Analysis library.</p>
890
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]891<h3 id="attr_analyzer_noreturn">The <tt>analyzer_noreturn</tt> attribute</h3>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]892
893<p>Clang's static analysis engine understands the standard <tt>noreturn</tt>
Ted Kremenek4df21142009-04-10 05:04:22 +0000[diff] [blame]894attribute. This attribute, which is typically affixed to a function prototype,
895indicates that a call to a given function never returns. Function prototypes for
896common functions like <tt>exit</tt> are typically annotated with this attribute,
897as well as a variety of common assertion handlers. Users can educate the static
898analyzer about their own custom assertion handles (thus cutting down on false
899positives due to false paths) by marking their own &quot;panic&quot; functions
900with this attribute.</p>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]901
902<p>While useful, <tt>noreturn</tt> is not applicable in all cases. Sometimes
Nick Lewycky625b5862009-06-14 04:08:08 +0000[diff] [blame]903there are special functions that for all intents and purposes should be
904considered panic functions (i.e., they are only called when an internal program
905error occurs) but may actually return so that the program can fail gracefully.
906The <tt>analyzer_noreturn</tt> attribute allows one to annotate such functions
907as being interpreted as &quot;no return&quot; functions by the analyzer (thus
Chris Lattner28935892009-04-10 05:54:56 +0000[diff] [blame]908pruning bogus paths) but will not affect compilation (as in the case of
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]909<tt>noreturn</tt>).</p>
910
911<p><b>Usage</b>: The <tt>analyzer_noreturn</tt> attribute can be placed in the
Chris Lattner28935892009-04-10 05:54:56 +0000[diff] [blame]912same places where the <tt>noreturn</tt> attribute can be placed. It is commonly
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]913placed at the end of function prototypes:</p>
914
915<pre>
916 void foo() <b>__attribute__((analyzer_noreturn))</b>;
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]917</pre>
918
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]919<p>Query for this feature with
920<tt>__has_attribute(analyzer_noreturn)</tt>.</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]921
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]922<h3 id="attr_method_family">The <tt>objc_method_family</tt> attribute</h3>
923
924<p>Many methods in Objective-C have conventional meanings determined
925by their selectors. For the purposes of static analysis, it is
926sometimes useful to be able to mark a method as having a particular
927conventional meaning despite not having the right selector, or as not
928having the conventional meaning that its selector would suggest.
929For these use cases, we provide an attribute to specifically describe
930the <q>method family</q> that a method belongs to.</p>
931
932<p><b>Usage</b>: <tt>__attribute__((objc_method_family(X)))</tt>,
933where <tt>X</tt> is one of <tt>none</tt>, <tt>alloc</tt>, <tt>copy</tt>,
934<tt>init</tt>, <tt>mutableCopy</tt>, or <tt>new</tt>. This attribute
935can only be placed at the end of a method declaration:</p>
936
937<pre>
938 - (NSString*) initMyStringValue <b>__attribute__((objc_method_family(none)))</b>;
939</pre>
940
941<p>Users who do not wish to change the conventional meaning of a
942method, and who merely want to document its non-standard retain and
943release semantics, should use the
944<a href="#attr_retain_release">retaining behavior attributes</a>
945described below.</p>
946
947<p>Query for this feature with
948<tt>__has_attribute(objc_method_family)</tt>.</p>
949
950<h3 id="attr_retain_release">Objective-C retaining behavior attributes</h3>
John McCall630b7ae2011-01-25 04:26:21 +0000[diff] [blame]951
952<p>In Objective-C, functions and methods are generally assumed to take
953and return objects with +0 retain counts, with some exceptions for
954special methods like <tt>+alloc</tt> and <tt>init</tt>. However,
955there are exceptions, and so Clang provides attributes to allow these
956exceptions to be documented, which helps the analyzer find leaks (and
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]957ignore non-leaks). Some exceptions may be better described using
958the <a href="#attr_method_family"><tt>objc_method_family</tt></a>
959attribute instead.</p>
John McCall630b7ae2011-01-25 04:26:21 +0000[diff] [blame]960
961<p><b>Usage</b>: The <tt>ns_returns_retained</tt>, <tt>ns_returns_not_retained</tt>,
962<tt>ns_returns_autoreleased</tt>, <tt>cf_returns_retained</tt>,
963and <tt>cf_returns_not_retained</tt> attributes can be placed on
964methods and functions that return Objective-C or CoreFoundation
965objects. They are commonly placed at the end of a function prototype
966or method declaration:</p>
967
968<pre>
969 id foo() <b>__attribute__((ns_returns_retained))</b>;
970
971 - (NSString*) bar: (int) x <b>__attribute__((ns_returns_retained))</b>;
972</pre>
973
974<p>The <tt>*_returns_retained</tt> attributes specify that the
975returned object has a +1 retain count.
976The <tt>*_returns_not_retained</tt> attributes specify that the return
977object has a +0 retain count, even if the normal convention for its
978selector would be +1. <tt>ns_returns_autoreleased</tt> specifies that the
979returned object is +0, but is guaranteed to live at least as long as the
980next flush of an autorelease pool.</p>
981
982<p><b>Usage</b>: The <tt>ns_consumed</tt> and <tt>cf_consumed</tt>
983attributes can be placed on an parameter declaration; they specify
984that the argument is expected to have a +1 retain count, which will be
985balanced in some way by the function or method.
986The <tt>ns_consumes_self</tt> attribute can only be placed on an
987Objective-C method; it specifies that the method expects
988its <tt>self</tt> parameter to have a +1 retain count, which it will
989balance in some way.</p>
990
991<pre>
992 void <b>foo(__attribute__((ns_consumed))</b> NSString *string);
993
994 - (void) bar <b>__attribute__((ns_consumes_self))</b>;
995 - (void) baz: (id) <b>__attribute__((ns_consumed))</b> x;
996</pre>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]997
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]998<p>Query for these features with <tt>__has_attribute(ns_consumed)</tt>,
999<tt>__has_attribute(ns_returns_retained)</tt>, etc.</p>
1000
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]1001</div>
1002</body>
1003</html>