Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / clang / a391a4642c5eb98aae606f2a2b7219b1a3619d11 / . / docs / LanguageExtensions.html
blob: f86835a9d1df099a42ffc011b5160849b762da98 [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>
35 <li><a href="#cxx_attributes">C++0x attributes</a></li>
36 <li><a href="#cxx_decltype">C++0x <tt>decltype()</tt></a></li>
Douglas Gregor07508002011-02-05 20:35:30 +0000[diff] [blame]37 <li><a href="#cxx_default_function_template_args">C++0x default template arguments in function templates</a></li>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]38 <li><a href="#cxx_deleted_functions">C++0x deleted functions</a></li>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]39 <li><a href="#cxx_lambdas">C++0x lambdas</a></li>
40 <li><a href="#cxx_nullptr">C++0x nullptr</a></li>
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000[diff] [blame]41 <li><a href="#cxx_override_control">C++0x override control</a></li>
Richard Smitha391a462011-04-15 15:14:40 +0000[diff] [blame^]42 <li><a href="#cxx_range_for">C++0x range-based for loop</a></li>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]43 <li><a href="#cxx_rvalue_references">C++0x rvalue references</a></li>
Douglas Gregor56209ff2011-01-26 21:25:54 +0000[diff] [blame]44 <li><a href="#cxx_reference_qualified_functions">C++0x reference-qualified functions</a></li>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]45 <li><a href="#cxx_static_assert">C++0x <tt>static_assert()</tt></a></li>
46 <li><a href="#cxx_auto_type">C++0x type inference</a></li>
47 <li><a href="#cxx_variadic_templates">C++0x variadic templates</a></li>
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]48 <li><a href="#cxx_inline_namespaces">C++0x inline namespaces</a></li>
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000[diff] [blame]49 <li><a href="#cxx_strong_enums">C++0x strongly-typed enumerations</a></li>
Douglas Gregordab60ad2010-10-01 18:44:50 +0000[diff] [blame]50 <li><a href="#cxx_trailing_return">C++0x trailing return type</a></li>
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000[diff] [blame]51 <li><a href="#cxx_noexcept">C++0x noexcept specification</a></li>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]52 </ul>
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]53<li><a href="#checking_type_traits">Checks for Type Traits</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]54<li><a href="#blocks">Blocks</a></li>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]55<li><a href="#overloading-in-c">Function Overloading in C</a></li>
Peter Collingbournef111d932011-04-15 00:35:48 +0000[diff] [blame]56<li><a href="#generic-selections">Generic Selections</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]57<li><a href="#builtins">Builtin Functions</a>
58 <ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]59 <li><a href="#__builtin_shufflevector">__builtin_shufflevector</a></li>
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]60 <li><a href="#__builtin_unreachable">__builtin_unreachable</a></li>
Chris Lattner23aa9c82011-04-09 03:57:26 +0000[diff] [blame]61 <li><a href="#__sync_swap">__sync_swap</a></li>
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]62 </ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]63</li>
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]64<li><a href="#targetspecific">Target-Specific Extensions</a>
65 <ul>
66 <li><a href="#x86-specific">X86/X86-64 Language Extensions</a></li>
67 </ul>
68</li>
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]69<li><a href="#analyzerspecific">Static Analysis-Specific Extensions</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]70</ul>
71
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]72<!-- ======================================================================= -->
73<h2 id="intro">Introduction</h2>
74<!-- ======================================================================= -->
75
76<p>This document describes the language extensions provided by Clang. In
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]77addition to the language extensions listed here, Clang aims to support a broad
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]78range of GCC extensions. Please see the <a
79href="http://gcc.gnu.org/onlinedocs/gcc/C-Extensions.html">GCC manual</a> for
80more information on these extensions.</p>
81
82<!-- ======================================================================= -->
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]83<h2 id="feature_check">Feature Checking Macros</h2>
84<!-- ======================================================================= -->
85
86<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]87on them. In order to allow fine-grain features checks, we support three builtin
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]88function-like macros. This allows you to directly test for a feature in your
89code without having to resort to something like autoconf or fragile "compiler
90version checks".</p>
91
92<!-- ======================================================================= -->
93<h3 id="__has_builtin">__has_builtin</h3>
94<!-- ======================================================================= -->
95
96<p>This function-like macro takes a single identifier argument that is the name
97of a builtin function. It evaluates to 1 if the builtin is supported or 0 if
98not. It can be used like this:</p>
99
100<blockquote>
101<pre>
102#ifndef __has_builtin // Optional of course.
103 #define __has_builtin(x) 0 // Compatibility with non-clang compilers.
104#endif
105
106...
107#if __has_builtin(__builtin_trap)
108 __builtin_trap();
109#else
110 abort();
111#endif
112...
113</pre>
114</blockquote>
115
116
117<!-- ======================================================================= -->
118<h3 id="__has_feature">__has_feature</h3>
119<!-- ======================================================================= -->
120
121<p>This function-like macro takes a single identifier argument that is the name
122of a feature. It evaluates to 1 if the feature is supported or 0 if not. It
123can be used like this:</p>
124
125<blockquote>
126<pre>
127#ifndef __has_feature // Optional of course.
128 #define __has_feature(x) 0 // Compatibility with non-clang compilers.
129#endif
130
131...
132#if __has_feature(attribute_overloadable) || \
133 __has_feature(blocks)
134...
135#endif
136...
137</pre>
138</blockquote>
139
140<p>The feature tag is described along with the language feature below.</p>
141
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]142<!-- ======================================================================= -->
Anders Carlssoncae50952010-10-20 02:31:43 +0000[diff] [blame]143<h3 id="__has_attribute">__has_attribute</h3>
144<!-- ======================================================================= -->
145
146<p>This function-like macro takes a single identifier argument that is the name
147of an attribute. It evaluates to 1 if the attribute is supported or 0 if not. It
148can be used like this:</p>
149
150<blockquote>
151<pre>
152#ifndef __has_attribute // Optional of course.
153 #define __has_attribute(x) 0 // Compatibility with non-clang compilers.
154#endif
155
156...
Anders Carlsson961003d2011-01-24 03:54:51 +0000[diff] [blame]157#if __has_attribute(always_inline)
158#define ALWAYS_INLINE __attribute__((always_inline))
Anders Carlssoncae50952010-10-20 02:31:43 +0000[diff] [blame]159#else
Anders Carlsson961003d2011-01-24 03:54:51 +0000[diff] [blame]160#define ALWAYS_INLINE
Anders Carlssoncae50952010-10-20 02:31:43 +0000[diff] [blame]161#endif
162...
163</pre>
164</blockquote>
165
166<!-- ======================================================================= -->
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]167<h2 id="has_include">Include File Checking Macros</h2>
168<!-- ======================================================================= -->
169
170<p>Not all developments systems have the same include files.
171The <a href="#__has_include">__has_include</a> and
172<a href="#__has_include_next">__has_include_next</a> macros allow you to
173check for the existence of an include file before doing
174a possibly failing #include directive.</p>
175
176<!-- ======================================================================= -->
177<h3 id="__has_include">__has_include</h3>
178<!-- ======================================================================= -->
179
180<p>This function-like macro takes a single file name string argument that
181is the name of an include file. It evaluates to 1 if the file can
182be found using the include paths, or 0 otherwise:</p>
183
184<blockquote>
185<pre>
186// Note the two possible file name string formats.
187#if __has_include("myinclude.h") && __has_include(&lt;stdint.h&gt;)
188# include "myinclude.h"
189#endif
190
191// To avoid problem with non-clang compilers not having this macro.
192#if defined(__has_include) && __has_include("myinclude.h")
193# include "myinclude.h"
194#endif
195</pre>
196</blockquote>
197
198<p>To test for this feature, use #if defined(__has_include).</p>
199
200<!-- ======================================================================= -->
201<h3 id="__has_include_next">__has_include_next</h3>
202<!-- ======================================================================= -->
203
204<p>This function-like macro takes a single file name string argument that
205is the name of an include file. It is like __has_include except that it
206looks for the second instance of the given file found in the include
207paths. It evaluates to 1 if the second instance of the file can
208be found using the include paths, or 0 otherwise:</p>
209
210<blockquote>
211<pre>
212// Note the two possible file name string formats.
213#if __has_include_next("myinclude.h") && __has_include_next(&lt;stdint.h&gt;)
214# include_next "myinclude.h"
215#endif
216
217// To avoid problem with non-clang compilers not having this macro.
218#if defined(__has_include_next) && __has_include_next("myinclude.h")
219# include_next "myinclude.h"
220#endif
221</pre>
222</blockquote>
223
224<p>Note that __has_include_next, like the GNU extension
225#include_next directive, is intended for use in headers only,
226and will issue a warning if used in the top-level compilation
227file. A warning will also be issued if an absolute path
228is used in the file argument.</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]229
230<!-- ======================================================================= -->
Chris Lattner81edc9f2009-04-13 02:45:46 +0000[diff] [blame]231<h2 id="builtinmacros">Builtin Macros</h2>
232<!-- ======================================================================= -->
233
Douglas Gregor4290fbd2010-04-30 02:51:06 +0000[diff] [blame]234<dl>
235 <dt><code>__BASE_FILE__</code></dt>
236 <dd>Defined to a string that contains the name of the main input
237 file passed to Clang.</dd>
238
239 <dt><code>__COUNTER__</code></dt>
240 <dd>Defined to an integer value that starts at zero and is
241 incremented each time the <code>__COUNTER__</code> macro is
242 expanded.</dd>
243
244 <dt><code>__INCLUDE_LEVEL__</code></dt>
245 <dd>Defined to an integral value that is the include depth of the
246 file currently being translated. For the main file, this value is
247 zero.</dd>
248
249 <dt><code>__TIMESTAMP__</code></dt>
250 <dd>Defined to the date and time of the last modification of the
251 current source file.</dd>
252
253 <dt><code>__clang__</code></dt>
254 <dd>Defined when compiling with Clang</dd>
255
256 <dt><code>__clang_major__</code></dt>
257 <dd>Defined to the major version number of Clang (e.g., the 2 in
258 2.0.1).</dd>
259
260 <dt><code>__clang_minor__</code></dt>
261 <dd>Defined to the minor version number of Clang (e.g., the 0 in
262 2.0.1).</dd>
263
264 <dt><code>__clang_patchlevel__</code></dt>
265 <dd>Defined to the patch level of Clang (e.g., the 1 in 2.0.1).</dd>
266
267 <dt><code>__clang_version__</code></dt>
268 <dd>Defined to a string that captures the Clang version, including
269 the Subversion tag or revision number, e.g., "1.5 (trunk
270 102332)".</dd>
271</dl>
Chris Lattner81edc9f2009-04-13 02:45:46 +0000[diff] [blame]272
273<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]274<h2 id="vectors">Vectors and Extended Vectors</h2>
275<!-- ======================================================================= -->
276
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]277<p>Supports the GCC vector extensions, plus some stuff like V[1].</p>
278
279<p>Also supports <tt>ext_vector</tt>, which additionally support for V.xyzw
280syntax and other tidbits as seen in OpenCL. An example is:</p>
281
282<blockquote>
283<pre>
284typedef float float4 <b>__attribute__((ext_vector_type(4)))</b>;
285typedef float float2 <b>__attribute__((ext_vector_type(2)))</b>;
286
287float4 foo(float2 a, float2 b) {
288 float4 c;
289 c.xz = a;
290 c.yw = b;
291 return c;
292}
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]293</pre>
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]294</blockquote>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]295
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]296<p>Query for this feature with __has_feature(attribute_ext_vector_type).</p>
297
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]298<p>See also <a href="#__builtin_shufflevector">__builtin_shufflevector</a>.</p>
299
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]300<!-- ======================================================================= -->
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]301<h2 id="deprecated">Messages on <tt>deprecated</tt> and <tt>unavailable</tt> Attributes</h2>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000[diff] [blame]302<!-- ======================================================================= -->
303
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]304<p>An optional string message can be added to the <tt>deprecated</tt>
305and <tt>unavailable</tt> attributes. For example:</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000[diff] [blame]306
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]307<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000[diff] [blame]308<pre>void explode(void) __attribute__((deprecated("extremely unsafe, use 'combust' instead!!!")));</pre>
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]309</blockquote>
310
311<p>If the deprecated or unavailable declaration is used, the message
312will be incorporated into the appropriate diagnostic:</p>
313
314<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000[diff] [blame]315<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]316 explode();
317 ^</pre>
318</blockquote>
319
320<p>Query for this feature
321with <tt>__has_feature(attribute_deprecated_with_message)</tt>
322and <tt>__has_feature(attribute_unavailable_with_message)</tt>.</p>
323
324<!-- ======================================================================= -->
325<h2 id="attributes-on-enumerators">Attributes on Enumerators</h2>
326<!-- ======================================================================= -->
327
328<p>Clang allows attributes to be written on individual enumerators.
329This allows enumerators to be deprecated, made unavailable, etc. The
330attribute must appear after the enumerator name and before any
331initializer, like so:</p>
332
333<blockquote>
334<pre>enum OperationMode {
335 OM_Invalid,
336 OM_Normal,
337 OM_Terrified __attribute__((deprecated)),
338 OM_AbortOnError __attribute__((deprecated)) = 4
339};</pre>
340</blockquote>
341
342<p>Attributes on the <tt>enum</tt> declaration do not apply to
343individual enumerators.</p>
344
345<p>Query for this feature with <tt>__has_feature(enumerator_attributes)</tt>.</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000[diff] [blame]346
347<!-- ======================================================================= -->
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]348<h2 id="checking_language_features">Checks for Standard Language Features</h2>
349<!-- ======================================================================= -->
350
351<p>The <tt>__has_feature</tt> macro can be used to query if certain standard language features are
352enabled. Those features are listed here.</p>
353
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]354<h3 id="cxx_exceptions">C++ exceptions</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]355
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]356<p>Use <tt>__has_feature(cxx_exceptions)</tt> to determine if C++ exceptions have been enabled. For
357example, compiling code with <tt>-fexceptions</tt> enables C++ exceptions.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]358
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]359<h3 id="cxx_rtti">C++ RTTI</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]360
Ted Kremenek0eb95602009-12-03 02:06:43 +0000[diff] [blame]361<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]362compiling code with <tt>-fno-rtti</tt> disables the use of RTTI.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]363
364<!-- ======================================================================= -->
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]365<h2 id="checking_upcoming_features">Checks for Upcoming Standard Language Features</h2>
366<!-- ======================================================================= -->
367
368<p>The <tt>__has_feature</tt> macro can be used to query if certain upcoming
369standard language features are enabled. Those features are listed here.</p>
370
371<p>Currently, all features listed here are slated for inclusion in the upcoming
372C++0x standard. As a result, all the features that clang supports are enabled
373with the <tt>-std=c++0x</tt> option when compiling C++ code. Features that are
374not yet implemented will be noted.</p>
375
376<h3 id="cxx_decltype">C++0x <tt>decltype()</tt></h3>
377
378<p>Use <tt>__has_feature(cxx_decltype)</tt> to determine if support for the
379<tt>decltype()</tt> specifier is enabled.</p>
380
381<h3 id="cxx_attributes">C++0x attributes</h3>
382
383<p>Use <tt>__has_feature(cxx_attributes)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]384attribute parsing with C++0x's square bracket notation is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]385
Douglas Gregor07508002011-02-05 20:35:30 +0000[diff] [blame]386<h3 id="cxx_default_function_template_args">C++0x default template arguments in function templates</h3>
387
388<p>Use <tt>__has_feature(cxx_default_function_template_args)</tt> to determine if support for default template arguments in function templates is enabled.</p>
389
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]390<h3 id="cxx_deleted_functions">C++0x deleted functions</tt></h3>
391
392<p>Use <tt>__has_feature(cxx_deleted_functions)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]393deleted function definitions (with <tt>= delete</tt>) is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]394
Douglas Gregor9cc90a32010-01-13 16:27:49 +0000[diff] [blame]395<h3 id="cxx_lambdas">C++0x lambdas</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]396
397<p>Use <tt>__has_feature(cxx_lambdas)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]398lambdas is enabled. clang does not currently implement this feature.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]399
400<h3 id="cxx_nullptr">C++0x <tt>nullptr</tt></h3>
401
402<p>Use <tt>__has_feature(cxx_nullptr)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]403<tt>nullptr</tt> is enabled. clang does not yet fully implement this
404feature.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]405
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000[diff] [blame]406<h3 id="cxx_override_control">C++0x <tt>override control</tt></h3>
407
408<p>Use <tt>__has_feature(cxx_override_control)</tt> to determine if support for
409the override control keywords is enabled.</p>
410
Douglas Gregor56209ff2011-01-26 21:25:54 +0000[diff] [blame]411<h3 id="cxx_reference_qualified_functions">C++0x reference-qualified functions</h3>
412<p>Use <tt>__has_feature(cxx_reference_qualified_functions)</tt> to determine if support for reference-qualified functions (e.g., member functions with <code>&amp;</code> or <code>&amp;&amp;</code> applied to <code>*this</code>) is enabled.</p>
413
Richard Smitha391a462011-04-15 15:14:40 +0000[diff] [blame^]414<h3 id="cxx_range_for">C++0x range-based for loop</tt></h3>
415
416<p>Use <tt>__has_feature(cxx_range_for)</tt> to determine if support for
417the range-based for loop is enabled. </p>
418
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]419<h3 id="cxx_rvalue_references">C++0x rvalue references</tt></h3>
420
421<p>Use <tt>__has_feature(cxx_rvalue_references)</tt> to determine if support for
Douglas Gregor56209ff2011-01-26 21:25:54 +0000[diff] [blame]422rvalue references is enabled. </p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]423
424<h3 id="cxx_static_assert">C++0x <tt>static_assert()</tt></h3>
425
426<p>Use <tt>__has_feature(cxx_static_assert)</tt> to determine if support for
427compile-time assertions using <tt>static_assert</tt> is enabled.</p>
428
429<h3 id="cxx_auto_type">C++0x type inference</h3>
430
431<p>Use <tt>__has_feature(cxx_auto_type)</tt> to determine C++0x type inference
432is supported using the <tt>auto</tt> specifier. If this is disabled,
Richard Smithfd405ef2011-02-23 00:41:16 +0000[diff] [blame]433<tt>auto</tt> will instead be a storage class specifier, as in C or C++98.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]434
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]435<h3 id="cxx_variadic_templates">C++0x variadic templates</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]436
437<p>Use <tt>__has_feature(cxx_variadic_templates)</tt> to determine if support
Douglas Gregor83d77812011-01-19 23:15:20 +0000[diff] [blame]438for variadic templates is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]439
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]440<h3 id="cxx_inline_namespaces">C++0x inline namespaces</h3>
441
442<p>Use <tt>__has_feature(cxx_inline_namespaces)</tt> to determine if support for
443inline namespaces is enabled.</p>
444
Douglas Gregordab60ad2010-10-01 18:44:50 +0000[diff] [blame]445<h3 id="cxx_trailing_return">C++0x trailing return type</h3>
446
447<p>Use <tt>__has_feature(cxx_trailing_return)</tt> to determine if support for
448the alternate function declaration syntax with trailing return type is enabled.</p>
449
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000[diff] [blame]450<h3 id="cxx_noexcept">C++0x noexcept</h3>
451
452<p>Use <tt>__has_feature(cxx_noexcept)</tt> to determine if support for
453noexcept exception specifications is enabled.</p>
454
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000[diff] [blame]455<h3 id="cxx_strong_enums">C++0x strongly typed enumerations</h3>
456
457<p>Use <tt>__has_feature(cxx_strong_enums)</tt> to determine if support for
458strongly typed, scoped enumerations is enabled.</p>
459
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]460<!-- ======================================================================= -->
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]461<h2 id="checking_type_traits">Checks for Type Traits</h2>
462<!-- ======================================================================= -->
463
464<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_feature(X)</code> indicates the presence of the type trait. For example:
465<blockquote>
466<pre>
467#if __has_feature(is_convertible_to)
468template&lt;typename From, typename To&gt;
469struct is_convertible_to {
470 static const bool value = __is_convertible_to(From, To);
471};
472#else
473// Emulate type trait
474#endif
475</pre>
476</blockquote>
477
478<p>The following type traits are supported by Clang:</p>
479<ul>
480 <li><code>__has_nothrow_assign</code> (GNU, Microsoft)</li>
481 <li><code>__has_nothrow_copy</code> (GNU, Microsoft)</li>
482 <li><code>__has_nothrow_constructor</code> (GNU, Microsoft)</li>
483 <li><code>__has_trivial_assign</code> (GNU, Microsoft)</li>
484 <li><code>__has_trivial_copy</code> (GNU, Microsoft)</li>
485 <li><code>__has_trivial_constructor</code> (GNU, Microsoft)</li>
486 <li><code>__has_trivial_destructor</code> (GNU, Microsoft)</li>
487 <li><code>__has_virtual_destructor</code> (GNU, Microsoft)</li>
488 <li><code>__is_abstract</code> (GNU, Microsoft)</li>
489 <li><code>__is_base_of</code> (GNU, Microsoft)</li>
490 <li><code>__is_class</code> (GNU, Microsoft)</li>
491 <li><code>__is_convertible_to</code> (Microsoft)</li>
492 <li><code>__is_empty</code> (GNU, Microsoft)</li>
493 <li><code>__is_enum</code> (GNU, Microsoft)</li>
494 <li><code>__is_pod</code> (GNU, Microsoft)</li>
495 <li><code>__is_polymorphic</code> (GNU, Microsoft)</li>
496 <li><code>__is_union</code> (GNU, Microsoft)</li>
497 <li><code>__is_literal(type)</code>: Determines whether the given type is a literal type</li>
498</ul>
499
500<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]501<h2 id="blocks">Blocks</h2>
502<!-- ======================================================================= -->
503
Chris Lattnera7dbdf52009-03-09 07:03:22 +0000[diff] [blame]504<p>The syntax and high level language feature description is in <a
505href="BlockLanguageSpec.txt">BlockLanguageSpec.txt</a>. Implementation and ABI
506details for the clang implementation are in <a
Chris Lattner5d7650b2010-03-16 21:43:03 +0000[diff] [blame]507href="Block-ABI-Apple.txt">Block-ABI-Apple.txt</a>.</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]508
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]509
510<p>Query for this feature with __has_feature(blocks).</p>
511
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]512<!-- ======================================================================= -->
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]513<h2 id="overloading-in-c">Function Overloading in C</h2>
514<!-- ======================================================================= -->
515
Chris Lattnerf161d412009-02-13 21:51:45 +0000[diff] [blame]516<p>Clang provides support for C++ function overloading in C. Function
517overloading in C is introduced using the <tt>overloadable</tt> attribute. For
518example, one might provide several overloaded versions of a <tt>tgsin</tt>
519function that invokes the appropriate standard function computing the sine of a
520value with <tt>float</tt>, <tt>double</tt>, or <tt>long double</tt>
521precision:</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]522
523<blockquote>
524<pre>
525#include &lt;math.h&gt;
526float <b>__attribute__((overloadable))</b> tgsin(float x) { return sinf(x); }
527double <b>__attribute__((overloadable))</b> tgsin(double x) { return sin(x); }
528long double <b>__attribute__((overloadable))</b> tgsin(long double x) { return sinl(x); }
529</pre>
530</blockquote>
531
532<p>Given these declarations, one can call <tt>tgsin</tt> with a
533<tt>float</tt> value to receive a <tt>float</tt> result, with a
534<tt>double</tt> to receive a <tt>double</tt> result, etc. Function
535overloading in C follows the rules of C++ function overloading to pick
536the best overload given the call arguments, with a few C-specific
537semantics:</p>
538<ul>
539 <li>Conversion from <tt>float</tt> or <tt>double</tt> to <tt>long
540 double</tt> is ranked as a floating-point promotion (per C99) rather
541 than as a floating-point conversion (as in C++).</li>
542
543 <li>A conversion from a pointer of type <tt>T*</tt> to a pointer of type
544 <tt>U*</tt> is considered a pointer conversion (with conversion
545 rank) if <tt>T</tt> and <tt>U</tt> are compatible types.</li>
546
547 <li>A conversion from type <tt>T</tt> to a value of type <tt>U</tt>
548 is permitted if <tt>T</tt> and <tt>U</tt> are compatible types. This
549 conversion is given "conversion" rank.</li>
550</ul>
551
552<p>The declaration of <tt>overloadable</tt> functions is restricted to
553function declarations and definitions. Most importantly, if any
554function with a given name is given the <tt>overloadable</tt>
555attribute, then all function declarations and definitions with that
556name (and in that scope) must have the <tt>overloadable</tt>
Chris Lattnerf161d412009-02-13 21:51:45 +0000[diff] [blame]557attribute. This rule even applies to redeclarations of functions whose original
558declaration had the <tt>overloadable</tt> attribute, e.g.,</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]559
560<blockquote>
561<pre>
562int f(int) __attribute__((overloadable));
563float f(float); <i>// error: declaration of "f" must have the "overloadable" attribute</i>
564
565int g(int) __attribute__((overloadable));
566int g(int) { } <i>// error: redeclaration of "g" must also have the "overloadable" attribute</i>
567</pre>
568</blockquote>
569
Douglas Gregor965acbb2009-02-18 07:07:28 +0000[diff] [blame]570<p>Functions marked <tt>overloadable</tt> must have
571prototypes. Therefore, the following code is ill-formed:</p>
572
573<blockquote>
574<pre>
575int h() __attribute__((overloadable)); <i>// error: h does not have a prototype</i>
576</pre>
577</blockquote>
578
579<p>However, <tt>overloadable</tt> functions are allowed to use a
580ellipsis 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>
581
582<blockquote>
583<pre>
Chris Lattner02246802009-02-18 22:27:46 +0000[diff] [blame]584void honeypot(...) __attribute__((overloadable, unavailable)); <i>// calling me is an error</i>
Douglas Gregor965acbb2009-02-18 07:07:28 +0000[diff] [blame]585</pre>
586</blockquote>
587
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]588<p>Functions declared with the <tt>overloadable</tt> attribute have
589their names mangled according to the same rules as C++ function
590names. For example, the three <tt>tgsin</tt> functions in our
591motivating example get the mangled names <tt>_Z5tgsinf</tt>,
Chris Lattner71b48d62010-11-28 18:19:13 +0000[diff] [blame]592<tt>_Z5tgsind</tt>, and <tt>_Z5tgsine</tt>, respectively. There are two
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]593caveats to this use of name mangling:</p>
594
595<ul>
596
597 <li>Future versions of Clang may change the name mangling of
598 functions overloaded in C, so you should not depend on an specific
599 mangling. To be completely safe, we strongly urge the use of
600 <tt>static inline</tt> with <tt>overloadable</tt> functions.</li>
601
602 <li>The <tt>overloadable</tt> attribute has almost no meaning when
603 used in C++, because names will already be mangled and functions are
604 already overloadable. However, when an <tt>overloadable</tt>
605 function occurs within an <tt>extern "C"</tt> linkage specification,
606 it's name <i>will</i> be mangled in the same way as it would in
607 C.</li>
608</ul>
609
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]610<p>Query for this feature with __has_feature(attribute_overloadable).</p>
611
612
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]613<!-- ======================================================================= -->
Peter Collingbournef111d932011-04-15 00:35:48 +0000[diff] [blame]614<h2 id="generic-selections">Generic Selections</h2>
615<!-- ======================================================================= -->
616
617<p>The C1X generic selection expression is available in all languages
618supported by Clang. The syntax is the same as that given in the C1X draft
619standard.</p>
620
621<p>In C, type compatibility is decided according to the rules given in the
622appropriate standard, but in C++, which lacks the type compatibility rules
623used in C, types are considered compatible only if they are equivalent.</p>
624
625<p>Query for this feature with __has_feature(generic_selections).</p>
626
627<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]628<h2 id="builtins">Builtin Functions</h2>
629<!-- ======================================================================= -->
630
631<p>Clang supports a number of builtin library functions with the same syntax as
632GCC, including things like <tt>__builtin_nan</tt>,
633<tt>__builtin_constant_p</tt>, <tt>__builtin_choose_expr</tt>,
634<tt>__builtin_types_compatible_p</tt>, <tt>__sync_fetch_and_add</tt>, etc. In
635addition to the GCC builtins, Clang supports a number of builtins that GCC does
636not, which are listed here.</p>
637
638<p>Please note that Clang does not and will not support all of the GCC builtins
639for vector operations. Instead of using builtins, you should use the functions
640defined in target-specific header files like <tt>&lt;xmmintrin.h&gt;</tt>, which
641define portable wrappers for these. Many of the Clang versions of these
642functions are implemented directly in terms of <a href="#vectors">extended
643vector support</a> instead of builtins, in order to reduce the number of
644builtins that we need to implement.</p>
645
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]646<!-- ======================================================================= -->
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]647<h3 id="__builtin_shufflevector">__builtin_shufflevector</h3>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]648<!-- ======================================================================= -->
649
Chris Lattneraad826b2009-09-16 18:56:12 +0000[diff] [blame]650<p><tt>__builtin_shufflevector</tt> is used to express generic vector
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]651permutation/shuffle/swizzle operations. This builtin is also very important for
652the implementation of various target-specific header files like
653<tt>&lt;xmmintrin.h&gt;</tt>.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]654</p>
655
656<p><b>Syntax:</b></p>
657
658<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]659__builtin_shufflevector(vec1, vec2, index1, index2, ...)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]660</pre>
661
662<p><b>Examples:</b></p>
663
664<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]665 // Identity operation - return 4-element vector V1.
666 __builtin_shufflevector(V1, V1, 0, 1, 2, 3)
667
668 // "Splat" element 0 of V1 into a 4-element result.
669 __builtin_shufflevector(V1, V1, 0, 0, 0, 0)
670
671 // Reverse 4-element vector V1.
672 __builtin_shufflevector(V1, V1, 3, 2, 1, 0)
673
674 // Concatenate every other element of 4-element vectors V1 and V2.
675 __builtin_shufflevector(V1, V2, 0, 2, 4, 6)
676
677 // Concatenate every other element of 8-element vectors V1 and V2.
678 __builtin_shufflevector(V1, V2, 0, 2, 4, 6, 8, 10, 12, 14)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]679</pre>
680
681<p><b>Description:</b></p>
682
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]683<p>The first two arguments to __builtin_shufflevector are vectors that have the
684same element type. The remaining arguments are a list of integers that specify
685the elements indices of the first two vectors that should be extracted and
686returned in a new vector. These element indices are numbered sequentially
687starting with the first vector, continuing into the second vector. Thus, if
688vec1 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]689</p>
690
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]691<p>The result of __builtin_shufflevector is a vector
692with the same element type as vec1/vec2 but that has an element count equal to
693the number of indices specified.
694</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]695
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]696<p>Query for this feature with __has_builtin(__builtin_shufflevector).</p>
697
698<!-- ======================================================================= -->
699<h3 id="__builtin_unreachable">__builtin_unreachable</h3>
700<!-- ======================================================================= -->
701
702<p><tt>__builtin_unreachable</tt> is used to indicate that a specific point in
703the program cannot be reached, even if the compiler might otherwise think it
704can. This is useful to improve optimization and eliminates certain warnings.
705For example, without the <tt>__builtin_unreachable</tt> in the example below,
706the compiler assumes that the inline asm can fall through and prints a "function
707declared 'noreturn' should not return" warning.
708</p>
709
710<p><b>Syntax:</b></p>
711
712<pre>
713__builtin_unreachable()
714</pre>
715
716<p><b>Example of Use:</b></p>
717
718<pre>
719void myabort(void) __attribute__((noreturn));
720void myabort(void) {
721 asm("int3");
722 __builtin_unreachable();
723}
724</pre>
725
726<p><b>Description:</b></p>
727
728<p>The __builtin_unreachable() builtin has completely undefined behavior. Since
729it has undefined behavior, it is a statement that it is never reached and the
730optimizer can take advantage of this to produce better code. This builtin takes
731no arguments and produces a void result.
732</p>
733
734<p>Query for this feature with __has_builtin(__builtin_unreachable).</p>
735
Chris Lattner23aa9c82011-04-09 03:57:26 +0000[diff] [blame]736<!-- ======================================================================= -->
737<h3 id="__sync_swap">__sync_swap</h3>
738<!-- ======================================================================= -->
739
740<p><tt>__sync_swap</tt> is used to atomically swap integers or pointers in
741memory.
742</p>
743
744<p><b>Syntax:</b></p>
745
746<pre>
747<i>type</i> __sync_swap(<i>type</i> *ptr, <i>type</i> value, ...)
748</pre>
749
750<p><b>Example of Use:</b></p>
751
752<pre>
753int old_value = __sync_swap(&value, new_value);
754</pre>
755
756<p><b>Description:</b></p>
757
758<p>The __sync_swap() builtin extends the existing __sync_*() family of atomic
759intrinsics to allow code to atomically swap the current value with the new
760value. More importantly, it helps developers write more efficient and correct
761code by avoiding expensive loops around __sync_bool_compare_and_swap() or
762relying on the platform specific implementation details of
763__sync_lock_test_and_set(). The __sync_swap() builtin is a full barrier.
764</p>
765
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]766
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]767<!-- ======================================================================= -->
768<h2 id="targetspecific">Target-Specific Extensions</h2>
769<!-- ======================================================================= -->
770
771<p>Clang supports some language features conditionally on some targets.</p>
772
773<!-- ======================================================================= -->
774<h3 id="x86-specific">X86/X86-64 Language Extensions</h3>
775<!-- ======================================================================= -->
776
777<p>The X86 backend has these language extensions:</p>
778
779<!-- ======================================================================= -->
780<h4 id="x86-gs-segment">Memory references off the GS segment</h4>
781<!-- ======================================================================= -->
782
783<p>Annotating a pointer with address space #256 causes it to be code generated
Chris Lattnera021e7c2009-05-05 18:54:47 +0000[diff] [blame]784relative to the X86 GS segment register, and address space #257 causes it to be
785relative to the X86 FS segment. Note that this is a very very low-level
786feature that should only be used if you know what you're doing (for example in
787an OS kernel).</p>
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]788
789<p>Here is an example:</p>
790
791<pre>
792#define GS_RELATIVE __attribute__((address_space(256)))
793int foo(int GS_RELATIVE *P) {
794 return *P;
795}
796</pre>
797
798<p>Which compiles to (on X86-32):</p>
799
800<pre>
801_foo:
802 movl 4(%esp), %eax
803 movl %gs:(%eax), %eax
804 ret
805</pre>
806
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]807<!-- ======================================================================= -->
808<h2 id="analyzerspecific">Static Analysis-Specific Extensions</h2>
809<!-- ======================================================================= -->
810
811<p>Clang supports additional attributes that are useful for documenting program
812invariants and rules for static analysis tools. The extensions documented here
813are used by the <a
814href="http://clang.llvm.org/StaticAnalysis.html">path-sensitive static analyzer
815engine</a> that is part of Clang's Analysis library.</p>
816
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]817<h3 id="attr_analyzer_noreturn">The <tt>analyzer_noreturn</tt> attribute</h3>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]818
819<p>Clang's static analysis engine understands the standard <tt>noreturn</tt>
Ted Kremenek4df21142009-04-10 05:04:22 +0000[diff] [blame]820attribute. This attribute, which is typically affixed to a function prototype,
821indicates that a call to a given function never returns. Function prototypes for
822common functions like <tt>exit</tt> are typically annotated with this attribute,
823as well as a variety of common assertion handlers. Users can educate the static
824analyzer about their own custom assertion handles (thus cutting down on false
825positives due to false paths) by marking their own &quot;panic&quot; functions
826with this attribute.</p>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]827
828<p>While useful, <tt>noreturn</tt> is not applicable in all cases. Sometimes
Nick Lewycky625b5862009-06-14 04:08:08 +0000[diff] [blame]829there are special functions that for all intents and purposes should be
830considered panic functions (i.e., they are only called when an internal program
831error occurs) but may actually return so that the program can fail gracefully.
832The <tt>analyzer_noreturn</tt> attribute allows one to annotate such functions
833as being interpreted as &quot;no return&quot; functions by the analyzer (thus
Chris Lattner28935892009-04-10 05:54:56 +0000[diff] [blame]834pruning bogus paths) but will not affect compilation (as in the case of
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]835<tt>noreturn</tt>).</p>
836
837<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]838same places where the <tt>noreturn</tt> attribute can be placed. It is commonly
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]839placed at the end of function prototypes:</p>
840
841<pre>
842 void foo() <b>__attribute__((analyzer_noreturn))</b>;
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]843</pre>
844
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]845<p>Query for this feature with
846<tt>__has_attribute(analyzer_noreturn)</tt>.</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]847
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]848<h3 id="attr_method_family">The <tt>objc_method_family</tt> attribute</h3>
849
850<p>Many methods in Objective-C have conventional meanings determined
851by their selectors. For the purposes of static analysis, it is
852sometimes useful to be able to mark a method as having a particular
853conventional meaning despite not having the right selector, or as not
854having the conventional meaning that its selector would suggest.
855For these use cases, we provide an attribute to specifically describe
856the <q>method family</q> that a method belongs to.</p>
857
858<p><b>Usage</b>: <tt>__attribute__((objc_method_family(X)))</tt>,
859where <tt>X</tt> is one of <tt>none</tt>, <tt>alloc</tt>, <tt>copy</tt>,
860<tt>init</tt>, <tt>mutableCopy</tt>, or <tt>new</tt>. This attribute
861can only be placed at the end of a method declaration:</p>
862
863<pre>
864 - (NSString*) initMyStringValue <b>__attribute__((objc_method_family(none)))</b>;
865</pre>
866
867<p>Users who do not wish to change the conventional meaning of a
868method, and who merely want to document its non-standard retain and
869release semantics, should use the
870<a href="#attr_retain_release">retaining behavior attributes</a>
871described below.</p>
872
873<p>Query for this feature with
874<tt>__has_attribute(objc_method_family)</tt>.</p>
875
876<h3 id="attr_retain_release">Objective-C retaining behavior attributes</h3>
John McCall630b7ae2011-01-25 04:26:21 +0000[diff] [blame]877
878<p>In Objective-C, functions and methods are generally assumed to take
879and return objects with +0 retain counts, with some exceptions for
880special methods like <tt>+alloc</tt> and <tt>init</tt>. However,
881there are exceptions, and so Clang provides attributes to allow these
882exceptions to be documented, which helps the analyzer find leaks (and
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]883ignore non-leaks). Some exceptions may be better described using
884the <a href="#attr_method_family"><tt>objc_method_family</tt></a>
885attribute instead.</p>
John McCall630b7ae2011-01-25 04:26:21 +0000[diff] [blame]886
887<p><b>Usage</b>: The <tt>ns_returns_retained</tt>, <tt>ns_returns_not_retained</tt>,
888<tt>ns_returns_autoreleased</tt>, <tt>cf_returns_retained</tt>,
889and <tt>cf_returns_not_retained</tt> attributes can be placed on
890methods and functions that return Objective-C or CoreFoundation
891objects. They are commonly placed at the end of a function prototype
892or method declaration:</p>
893
894<pre>
895 id foo() <b>__attribute__((ns_returns_retained))</b>;
896
897 - (NSString*) bar: (int) x <b>__attribute__((ns_returns_retained))</b>;
898</pre>
899
900<p>The <tt>*_returns_retained</tt> attributes specify that the
901returned object has a +1 retain count.
902The <tt>*_returns_not_retained</tt> attributes specify that the return
903object has a +0 retain count, even if the normal convention for its
904selector would be +1. <tt>ns_returns_autoreleased</tt> specifies that the
905returned object is +0, but is guaranteed to live at least as long as the
906next flush of an autorelease pool.</p>
907
908<p><b>Usage</b>: The <tt>ns_consumed</tt> and <tt>cf_consumed</tt>
909attributes can be placed on an parameter declaration; they specify
910that the argument is expected to have a +1 retain count, which will be
911balanced in some way by the function or method.
912The <tt>ns_consumes_self</tt> attribute can only be placed on an
913Objective-C method; it specifies that the method expects
914its <tt>self</tt> parameter to have a +1 retain count, which it will
915balance in some way.</p>
916
917<pre>
918 void <b>foo(__attribute__((ns_consumed))</b> NSString *string);
919
920 - (void) bar <b>__attribute__((ns_consumes_self))</b>;
921 - (void) baz: (id) <b>__attribute__((ns_consumed))</b> x;
922</pre>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]923
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]924<p>Query for these features with <tt>__has_attribute(ns_consumed)</tt>,
925<tt>__has_attribute(ns_returns_retained)</tt>, etc.</p>
926
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]927</div>
928</body>
929</html>