blob: d8a57df4865a6aa7b5cb4e9c749e49b8fe937714 [file] [log] [blame]
Sean Hunt7e98b472011-06-23 01:21:01 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
Chris Lattner5ce933f2009-02-09 08:46:11 +00004<html>
5<head>
Sean Hunt7e98b472011-06-23 01:21:01 +00006 <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
Eli Friedman0c706c22011-09-19 23:17:44 +00007 <title>Clang Language Extensions</title>
Sean Hunt64f857b2011-06-23 01:22:53 +00008 <link type="text/css" rel="stylesheet" href="../menu.css">
9 <link type="text/css" rel="stylesheet" href="../content.css">
Sean Hunt7e98b472011-06-23 01:21:01 +000010 <style type="text/css">
11 td {
12 vertical-align: top;
13 }
14 </style>
Chris Lattner5ce933f2009-02-09 08:46:11 +000015</head>
16<body>
17
18<!--#include virtual="../menu.html.incl"-->
19
20<div id="content">
21
22<h1>Clang Language Extensions</h1>
23
24<ul>
25<li><a href="#intro">Introduction</a></li>
Chris Lattner148772a2009-06-13 07:13:28 +000026<li><a href="#feature_check">Feature Checking Macros</a></li>
John Thompson92bd8c72009-11-02 22:28:12 +000027<li><a href="#has_include">Include File Checking Macros</a></li>
Chris Lattner81edc9f2009-04-13 02:45:46 +000028<li><a href="#builtinmacros">Builtin Macros</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +000029<li><a href="#vectors">Vectors and Extended Vectors</a></li>
John McCall48209082010-11-08 19:48:17 +000030<li><a href="#deprecated">Messages on <tt>deprecated</tt> and <tt>unavailable</tt> attributes</a></li>
31<li><a href="#attributes-on-enumerators">Attributes on enumerators</a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +000032<li><a href="#checking_language_features">Checks for Standard Language Features</a>
Ted Kremenek22c34102009-12-03 02:05:57 +000033 <ul>
34 <li><a href="#cxx_exceptions">C++ exceptions</a></li>
35 <li><a href="#cxx_rtti">C++ RTTI</a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +000036 </ul></li>
37<li><a href="#checking_upcoming_features">Checks for Upcoming Standard Language Features</a>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +000038 <ul>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000039 <li><a href="#cxx0x">C++0x</a>
40 <ul>
Douglas Gregorece38942011-08-29 17:28:38 +000041 <li><a href="#cxx_access_control_sfinae">C++0x SFINAE includes access control</a></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000042 <li><a href="#cxx_alias_templates">C++0x alias templates</a></li>
43 <li><a href="#cxx_attributes">C++0x attributes</a></li>
Douglas Gregorece38942011-08-29 17:28:38 +000044 <li><a href="#cxx_constexpr">C++0x generalized constant expressions</a></li>
45 <li><a href="#cxx_decltype">C++0x <tt>decltype()</tt></a></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000046 <li><a href="#cxx_default_function_template_args">C++0x default template arguments in function templates</a></li>
Sean Huntd9624992011-06-23 06:11:37 +000047 <li><a href="#cxx_delegating_constructor">C++0x delegating constructors</a></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000048 <li><a href="#cxx_deleted_functions">C++0x deleted functions</a></li>
Douglas Gregorece38942011-08-29 17:28:38 +000049 <li><a href="#cxx_explicit_conversions">C++0x explicit conversion functions</a></li>
Sean Hunte1f6dea2011-08-07 00:34:32 +000050 <li><a href="#cxx_generalized_initializers">C++0x generalized initializers</a></li>
Douglas Gregorece38942011-08-29 17:28:38 +000051 <li><a href="#cxx_implicit_moves">C++0x implicit move constructors/assignment operators</a></li>
52 <li><a href="#cxx_inheriting_constructors">C++0x inheriting constructors</a></li>
53 <li><a href="#cxx_inline_namespaces">C++0x inline namespaces</a></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000054 <li><a href="#cxx_lambdas">C++0x lambdas</a></li>
Douglas Gregorece38942011-08-29 17:28:38 +000055 <li><a href="#cxx_noexcept">C++0x noexcept specification</a></li>
56 <li><a href="#cxx_nonstatic_member_init">C++0x in-class non-static data member initialization</a></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000057 <li><a href="#cxx_nullptr">C++0x nullptr</a></li>
58 <li><a href="#cxx_override_control">C++0x override control</a></li>
59 <li><a href="#cxx_range_for">C++0x range-based for loop</a></li>
Douglas Gregorece38942011-08-29 17:28:38 +000060 <li><a href="#cxx_raw_string_literals">C++0x raw string literals</a></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000061 <li><a href="#cxx_rvalue_references">C++0x rvalue references</a></li>
62 <li><a href="#cxx_reference_qualified_functions">C++0x reference-qualified functions</a></li>
63 <li><a href="#cxx_static_assert">C++0x <tt>static_assert()</tt></a></li>
64 <li><a href="#cxx_auto_type">C++0x type inference</a></li>
Douglas Gregorece38942011-08-29 17:28:38 +000065 <li><a href="#cxx_strong_enums">C++0x strongly-typed enumerations</a></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000066 <li><a href="#cxx_trailing_return">C++0x trailing return type</a></li>
Douglas Gregorece38942011-08-29 17:28:38 +000067 <li><a href="#cxx_unicode_literals">C++0x Unicode string literals</a></li>
68 <li><a href="#cxx_unrestricted_unions">C++0x unrestricted unions</a></li>
69 <li><a href="#cxx_user_literals">C++0x user-defined literals</a></li>
70 <li><a href="#cxx_variadic_templates">C++0x variadic templates</a></li>
71 </ul></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +000072 <li><a href="#c1x">C1X</a>
73 <ul>
74 <li><a href="#c_generic_selections">C1X generic selections</a></li>
75 <li><a href="#c_static_assert">C1X <tt>_Static_assert()</tt></a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +000076 </ul></li>
77 </ul> </li>
Douglas Gregorafdf1372011-02-03 21:57:35 +000078<li><a href="#checking_type_traits">Checks for Type Traits</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +000079<li><a href="#blocks">Blocks</a></li>
Douglas Gregor926df6c2011-06-11 01:09:30 +000080<li><a href="#objc_features">Objective-C Features</a>
81 <ul>
82 <li><a href="#objc_instancetype">Related result types</a></li>
John McCallf85e1932011-06-15 23:02:42 +000083 <li><a href="#objc_arc">Automatic reference counting</a></li>
Douglas Gregor5471bc82011-09-08 17:18:35 +000084 <li><a href="#objc_fixed_enum">Enumerations with a fixed underlying type</a></li>
Douglas Gregor926df6c2011-06-11 01:09:30 +000085 </ul>
86</li>
Douglas Gregorcb54d432009-02-13 00:57:04 +000087<li><a href="#overloading-in-c">Function Overloading in C</a></li>
Eli Friedman0c706c22011-09-19 23:17:44 +000088<li><a href="#complex-list-init">Initializer lists for complex numbers in C</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +000089<li><a href="#builtins">Builtin Functions</a>
90 <ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +000091 <li><a href="#__builtin_shufflevector">__builtin_shufflevector</a></li>
Chris Lattner21190d52009-09-21 03:09:59 +000092 <li><a href="#__builtin_unreachable">__builtin_unreachable</a></li>
Chris Lattner23aa9c82011-04-09 03:57:26 +000093 <li><a href="#__sync_swap">__sync_swap</a></li>
Douglas Gregorafdf1372011-02-03 21:57:35 +000094 </ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +000095</li>
Chris Lattner1177f912009-04-09 19:58:15 +000096<li><a href="#targetspecific">Target-Specific Extensions</a>
97 <ul>
98 <li><a href="#x86-specific">X86/X86-64 Language Extensions</a></li>
99 </ul>
100</li>
John McCall87494012011-03-18 03:51:49 +0000101<li><a href="#analyzerspecific">Static Analysis-Specific Extensions</a></li>
Caitlin Sadowskifdde9e72011-07-28 17:21:07 +0000102<li><a href="#threadsafety">Thread Safety Annotation Checking</a></li>
103 <ul>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +0000104 <li><a href="#ts_noanal"><tt>no_thread_safety_analysis</tt></a></li>
105 <li><a href="#ts_lockable"><tt>lockable</tt></a></li>
106 <li><a href="#ts_scopedlockable"><tt>scoped_lockable</tt></a></li>
Caitlin Sadowskifdde9e72011-07-28 17:21:07 +0000107 <li><a href="#ts_guardedvar"><tt>guarded_var</tt></a></li>
108 <li><a href="#ts_ptguardedvar"><tt>pt_guarded_var</tt></a></li>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +0000109 <li><a href="#ts_guardedby"><tt>guarded_by(l)</tt></a></li>
110 <li><a href="#ts_ptguardedby"><tt>pt_guarded_by(l)</tt></a></li>
111 <li><a href="#ts_acquiredbefore"><tt>acquired_before(...)</tt></a></li>
112 <li><a href="#ts_acquiredafter"><tt>acquired_after(...)</tt></a></li>
113 <li><a href="#ts_elf"><tt>exclusive_lock_function(...)</tt></a></li>
114 <li><a href="#ts_slf"><tt>shared_lock_function(...)</tt></a></li>
115 <li><a href="#ts_etf"><tt>exclusive_trylock_function(...)</tt></a></li>
116 <li><a href="#ts_stf"><tt>shared_trylock_function(...)</tt></a></li>
117 <li><a href="#ts_uf"><tt>unlock_function(...)</tt></a></li>
118 <li><a href="#ts_lr"><tt>lock_returned(l)</tt></a></li>
119 <li><a href="#ts_le"><tt>locks_excluded(...)</tt></a></li>
120 <li><a href="#ts_elr"><tt>exclusive_locks_required(...)</tt></a></li>
121 <li><a href="#ts_slr"><tt>shared_locks_required(...)</tt></a></li>
Caitlin Sadowskifdde9e72011-07-28 17:21:07 +0000122 </ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000123</ul>
124
Chris Lattner5ce933f2009-02-09 08:46:11 +0000125<!-- ======================================================================= -->
126<h2 id="intro">Introduction</h2>
127<!-- ======================================================================= -->
128
129<p>This document describes the language extensions provided by Clang. In
Chris Lattner148772a2009-06-13 07:13:28 +0000130addition to the language extensions listed here, Clang aims to support a broad
Chris Lattner5ce933f2009-02-09 08:46:11 +0000131range of GCC extensions. Please see the <a
132href="http://gcc.gnu.org/onlinedocs/gcc/C-Extensions.html">GCC manual</a> for
133more information on these extensions.</p>
134
135<!-- ======================================================================= -->
Chris Lattner148772a2009-06-13 07:13:28 +0000136<h2 id="feature_check">Feature Checking Macros</h2>
137<!-- ======================================================================= -->
138
139<p>Language extensions can be very useful, but only if you know you can depend
Chris Lattnerc70e1932011-03-21 16:25:11 +0000140on them. In order to allow fine-grain features checks, we support three builtin
Chris Lattner148772a2009-06-13 07:13:28 +0000141function-like macros. This allows you to directly test for a feature in your
142code without having to resort to something like autoconf or fragile "compiler
143version checks".</p>
144
145<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000146<h3><a name="__has_builtin">__has_builtin</a></h3>
Chris Lattner148772a2009-06-13 07:13:28 +0000147<!-- ======================================================================= -->
148
149<p>This function-like macro takes a single identifier argument that is the name
150of a builtin function. It evaluates to 1 if the builtin is supported or 0 if
151not. It can be used like this:</p>
152
153<blockquote>
154<pre>
155#ifndef __has_builtin // Optional of course.
156 #define __has_builtin(x) 0 // Compatibility with non-clang compilers.
157#endif
158
159...
160#if __has_builtin(__builtin_trap)
161 __builtin_trap();
162#else
163 abort();
164#endif
165...
166</pre>
167</blockquote>
168
169
170<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000171<h3><a name="__has_feature_extension"> __has_feature and __has_extension</a></h3>
Chris Lattner148772a2009-06-13 07:13:28 +0000172<!-- ======================================================================= -->
173
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000174<p>These function-like macros take a single identifier argument that is the
175name of a feature. <code>__has_feature</code> evaluates to 1 if the feature
176is both supported by Clang and standardized in the current language standard
177or 0 if not (but see <a href="#has_feature_back_compat">below</a>), while
178<code>__has_extension</code> evaluates to 1 if the feature is supported by
179Clang in the current language (either as a language extension or a standard
180language feature) or 0 if not. They can be used like this:</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000181
182<blockquote>
183<pre>
184#ifndef __has_feature // Optional of course.
185 #define __has_feature(x) 0 // Compatibility with non-clang compilers.
186#endif
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000187#ifndef __has_extension
188 #define __has_extension __has_feature // Compatibility with pre-3.0 compilers.
189#endif
Chris Lattner148772a2009-06-13 07:13:28 +0000190
191...
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000192#if __has_feature(cxx_rvalue_references)
193// This code will only be compiled with the -std=c++0x and -std=gnu++0x
194// options, because rvalue references are only standardized in C++0x.
Chris Lattner148772a2009-06-13 07:13:28 +0000195#endif
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000196
197#if __has_extension(cxx_rvalue_references)
198// This code will be compiled with the -std=c++0x, -std=gnu++0x, -std=c++98
199// and -std=gnu++98 options, because rvalue references are supported as a
200// language extension in C++98.
201#endif
Chris Lattner148772a2009-06-13 07:13:28 +0000202</pre>
203</blockquote>
204
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000205<p id="has_feature_back_compat">For backwards compatibility reasons,
206<code>__has_feature</code> can also be used to test for support for
207non-standardized features, i.e. features not prefixed <code>c_</code>,
208<code>cxx_</code> or <code>objc_</code>.</p>
209
210<p>If the <code>-pedantic-errors</code> option is given,
211<code>__has_extension</code> is equivalent to <code>__has_feature</code>.</p>
212
Chris Lattner148772a2009-06-13 07:13:28 +0000213<p>The feature tag is described along with the language feature below.</p>
214
John Thompson92bd8c72009-11-02 22:28:12 +0000215<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000216<h3><a name="__has_attribute">__has_attribute</a></h3>
Anders Carlssoncae50952010-10-20 02:31:43 +0000217<!-- ======================================================================= -->
218
219<p>This function-like macro takes a single identifier argument that is the name
220of an attribute. It evaluates to 1 if the attribute is supported or 0 if not. It
221can be used like this:</p>
222
223<blockquote>
224<pre>
225#ifndef __has_attribute // Optional of course.
226 #define __has_attribute(x) 0 // Compatibility with non-clang compilers.
227#endif
228
229...
Anders Carlsson961003d2011-01-24 03:54:51 +0000230#if __has_attribute(always_inline)
231#define ALWAYS_INLINE __attribute__((always_inline))
Anders Carlssoncae50952010-10-20 02:31:43 +0000232#else
Anders Carlsson961003d2011-01-24 03:54:51 +0000233#define ALWAYS_INLINE
Anders Carlssoncae50952010-10-20 02:31:43 +0000234#endif
235...
236</pre>
237</blockquote>
238
239<!-- ======================================================================= -->
John Thompson92bd8c72009-11-02 22:28:12 +0000240<h2 id="has_include">Include File Checking Macros</h2>
241<!-- ======================================================================= -->
242
243<p>Not all developments systems have the same include files.
244The <a href="#__has_include">__has_include</a> and
245<a href="#__has_include_next">__has_include_next</a> macros allow you to
246check for the existence of an include file before doing
247a possibly failing #include directive.</p>
248
249<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000250<h3><a name="__has_include">__has_include</a></h3>
John Thompson92bd8c72009-11-02 22:28:12 +0000251<!-- ======================================================================= -->
252
253<p>This function-like macro takes a single file name string argument that
254is the name of an include file. It evaluates to 1 if the file can
255be found using the include paths, or 0 otherwise:</p>
256
257<blockquote>
258<pre>
259// Note the two possible file name string formats.
Sean Hunt7e98b472011-06-23 01:21:01 +0000260#if __has_include("myinclude.h") &amp;&amp; __has_include(&lt;stdint.h&gt;)
John Thompson92bd8c72009-11-02 22:28:12 +0000261# include "myinclude.h"
262#endif
263
264// To avoid problem with non-clang compilers not having this macro.
Sean Hunt7e98b472011-06-23 01:21:01 +0000265#if defined(__has_include) &amp;&amp; __has_include("myinclude.h")
John Thompson92bd8c72009-11-02 22:28:12 +0000266# include "myinclude.h"
267#endif
268</pre>
269</blockquote>
270
271<p>To test for this feature, use #if defined(__has_include).</p>
272
273<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000274<h3><a name="__has_include_next">__has_include_next</a></h3>
John Thompson92bd8c72009-11-02 22:28:12 +0000275<!-- ======================================================================= -->
276
277<p>This function-like macro takes a single file name string argument that
278is the name of an include file. It is like __has_include except that it
279looks for the second instance of the given file found in the include
280paths. It evaluates to 1 if the second instance of the file can
281be found using the include paths, or 0 otherwise:</p>
282
283<blockquote>
284<pre>
285// Note the two possible file name string formats.
Sean Hunt7e98b472011-06-23 01:21:01 +0000286#if __has_include_next("myinclude.h") &amp;&amp; __has_include_next(&lt;stdint.h&gt;)
John Thompson92bd8c72009-11-02 22:28:12 +0000287# include_next "myinclude.h"
288#endif
289
290// To avoid problem with non-clang compilers not having this macro.
Sean Hunt7e98b472011-06-23 01:21:01 +0000291#if defined(__has_include_next) &amp;&amp; __has_include_next("myinclude.h")
John Thompson92bd8c72009-11-02 22:28:12 +0000292# include_next "myinclude.h"
293#endif
294</pre>
295</blockquote>
296
297<p>Note that __has_include_next, like the GNU extension
298#include_next directive, is intended for use in headers only,
299and will issue a warning if used in the top-level compilation
300file. A warning will also be issued if an absolute path
301is used in the file argument.</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000302
303<!-- ======================================================================= -->
Chris Lattner81edc9f2009-04-13 02:45:46 +0000304<h2 id="builtinmacros">Builtin Macros</h2>
305<!-- ======================================================================= -->
306
Douglas Gregor4290fbd2010-04-30 02:51:06 +0000307<dl>
308 <dt><code>__BASE_FILE__</code></dt>
309 <dd>Defined to a string that contains the name of the main input
310 file passed to Clang.</dd>
311
312 <dt><code>__COUNTER__</code></dt>
313 <dd>Defined to an integer value that starts at zero and is
314 incremented each time the <code>__COUNTER__</code> macro is
315 expanded.</dd>
316
317 <dt><code>__INCLUDE_LEVEL__</code></dt>
318 <dd>Defined to an integral value that is the include depth of the
319 file currently being translated. For the main file, this value is
320 zero.</dd>
321
322 <dt><code>__TIMESTAMP__</code></dt>
323 <dd>Defined to the date and time of the last modification of the
324 current source file.</dd>
325
326 <dt><code>__clang__</code></dt>
327 <dd>Defined when compiling with Clang</dd>
328
329 <dt><code>__clang_major__</code></dt>
330 <dd>Defined to the major version number of Clang (e.g., the 2 in
331 2.0.1).</dd>
332
333 <dt><code>__clang_minor__</code></dt>
334 <dd>Defined to the minor version number of Clang (e.g., the 0 in
335 2.0.1).</dd>
336
337 <dt><code>__clang_patchlevel__</code></dt>
338 <dd>Defined to the patch level of Clang (e.g., the 1 in 2.0.1).</dd>
339
340 <dt><code>__clang_version__</code></dt>
341 <dd>Defined to a string that captures the Clang version, including
342 the Subversion tag or revision number, e.g., "1.5 (trunk
343 102332)".</dd>
344</dl>
Chris Lattner81edc9f2009-04-13 02:45:46 +0000345
346<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000347<h2 id="vectors">Vectors and Extended Vectors</h2>
348<!-- ======================================================================= -->
349
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000350<p>Supports the GCC vector extensions, plus some stuff like V[1].</p>
351
352<p>Also supports <tt>ext_vector</tt>, which additionally support for V.xyzw
353syntax and other tidbits as seen in OpenCL. An example is:</p>
354
355<blockquote>
356<pre>
357typedef float float4 <b>__attribute__((ext_vector_type(4)))</b>;
358typedef float float2 <b>__attribute__((ext_vector_type(2)))</b>;
359
360float4 foo(float2 a, float2 b) {
361 float4 c;
362 c.xz = a;
363 c.yw = b;
364 return c;
365}
John McCall48209082010-11-08 19:48:17 +0000366</pre>
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000367</blockquote>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000368
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000369<p>Query for this feature with __has_extension(attribute_ext_vector_type).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000370
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000371<p>See also <a href="#__builtin_shufflevector">__builtin_shufflevector</a>.</p>
372
Chris Lattner5ce933f2009-02-09 08:46:11 +0000373<!-- ======================================================================= -->
John McCall48209082010-11-08 19:48:17 +0000374<h2 id="deprecated">Messages on <tt>deprecated</tt> and <tt>unavailable</tt> Attributes</h2>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000375<!-- ======================================================================= -->
376
John McCall48209082010-11-08 19:48:17 +0000377<p>An optional string message can be added to the <tt>deprecated</tt>
378and <tt>unavailable</tt> attributes. For example:</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000379
John McCall48209082010-11-08 19:48:17 +0000380<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000381<pre>void explode(void) __attribute__((deprecated("extremely unsafe, use 'combust' instead!!!")));</pre>
John McCall48209082010-11-08 19:48:17 +0000382</blockquote>
383
384<p>If the deprecated or unavailable declaration is used, the message
385will be incorporated into the appropriate diagnostic:</p>
386
387<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000388<pre>harmless.c:4:3: warning: 'explode' is deprecated: extremely unsafe, use 'combust' instead!!! [-Wdeprecated-declarations]
John McCall48209082010-11-08 19:48:17 +0000389 explode();
390 ^</pre>
391</blockquote>
392
393<p>Query for this feature
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000394with <tt>__has_extension(attribute_deprecated_with_message)</tt>
395and <tt>__has_extension(attribute_unavailable_with_message)</tt>.</p>
John McCall48209082010-11-08 19:48:17 +0000396
397<!-- ======================================================================= -->
398<h2 id="attributes-on-enumerators">Attributes on Enumerators</h2>
399<!-- ======================================================================= -->
400
401<p>Clang allows attributes to be written on individual enumerators.
402This allows enumerators to be deprecated, made unavailable, etc. The
403attribute must appear after the enumerator name and before any
404initializer, like so:</p>
405
406<blockquote>
407<pre>enum OperationMode {
408 OM_Invalid,
409 OM_Normal,
410 OM_Terrified __attribute__((deprecated)),
411 OM_AbortOnError __attribute__((deprecated)) = 4
412};</pre>
413</blockquote>
414
415<p>Attributes on the <tt>enum</tt> declaration do not apply to
416individual enumerators.</p>
417
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000418<p>Query for this feature with <tt>__has_extension(enumerator_attributes)</tt>.</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000419
420<!-- ======================================================================= -->
Ted Kremenek87774fd2009-12-03 02:04:01 +0000421<h2 id="checking_language_features">Checks for Standard Language Features</h2>
422<!-- ======================================================================= -->
423
424<p>The <tt>__has_feature</tt> macro can be used to query if certain standard language features are
425enabled. Those features are listed here.</p>
426
Ted Kremenek22c34102009-12-03 02:05:57 +0000427<h3 id="cxx_exceptions">C++ exceptions</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000428
Ted Kremenek22c34102009-12-03 02:05:57 +0000429<p>Use <tt>__has_feature(cxx_exceptions)</tt> to determine if C++ exceptions have been enabled. For
Sean Hunt647ba1b2011-06-23 00:42:53 +0000430example, compiling code with <tt>-fexceptions</tt> enables C++ exceptions.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000431
Ted Kremenek22c34102009-12-03 02:05:57 +0000432<h3 id="cxx_rtti">C++ RTTI</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000433
Ted Kremenek0eb95602009-12-03 02:06:43 +0000434<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 +0000435compiling code with <tt>-fno-rtti</tt> disables the use of RTTI.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000436
437<!-- ======================================================================= -->
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000438<h2 id="checking_upcoming_features">Checks for Upcoming Standard Language Features</h2>
439<!-- ======================================================================= -->
440
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000441<p>The <tt>__has_feature</tt> or <tt>__has_extension</tt> macros can be used
442to query if certain upcoming standard language features are enabled. Those
443features are listed here. Features that are not yet implemented will be
444noted.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000445
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000446<h3 id="cxx0x">C++0x</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000447
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000448<p>The features listed below are slated for inclusion in the upcoming
449C++0x standard. As a result, all these features are enabled
450with the <tt>-std=c++0x</tt> option when compiling C++ code.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000451
Sean Hunt7e98b472011-06-23 01:21:01 +0000452<h4 id="cxx_access_control_sfinae">C++0x SFINAE includes access control</h4>
Douglas Gregor7822ee32011-05-11 23:45:11 +0000453
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000454<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 +0000455
Sean Hunt7e98b472011-06-23 01:21:01 +0000456<h4 id="cxx_alias_templates">C++0x alias templates</h4>
Richard Smith3e4c6c42011-05-05 21:57:07 +0000457
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000458<p>Use <tt>__has_feature(cxx_alias_templates)</tt> or
459<tt>__has_extension(cxx_alias_templates)</tt> to determine if support for
Richard Smith3e4c6c42011-05-05 21:57:07 +0000460C++0x's alias declarations and alias templates is enabled.</p>
461
Sean Hunt7e98b472011-06-23 01:21:01 +0000462<h4 id="cxx_attributes">C++0x attributes</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000463
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000464<p>Use <tt>__has_feature(cxx_attributes)</tt> or
465<tt>__has_extension(cxx_attributes)</tt> to determine if support for attribute
466parsing with C++0x's square bracket notation is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000467
Douglas Gregorece38942011-08-29 17:28:38 +0000468<h4 id="cxx_constexpr">C++0x generalized constant expressions</h4>
469
470<p>Use <tt>__has_feature(cxx_constexpr)</tt> to determine if support
471for generalized constant expressions (e.g., <tt>constexpr</tt>) is
472enabled. Clang does not currently implement this feature.</p>
473
474<h4 id="cxx_decltype">C++0x <tt>decltype()</tt></h4>
475
476<p>Use <tt>__has_feature(cxx_decltype)</tt> or
477<tt>__has_extension(cxx_decltype)</tt> to determine if support for the
478<tt>decltype()</tt> specifier is enabled.</p>
479
Sean Hunt7e98b472011-06-23 01:21:01 +0000480<h4 id="cxx_default_function_template_args">C++0x default template arguments in function templates</h4>
Douglas Gregor07508002011-02-05 20:35:30 +0000481
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000482<p>Use <tt>__has_feature(cxx_default_function_template_args)</tt> or
483<tt>__has_extension(cxx_default_function_template_args)</tt> to determine
484if support for default template arguments in function templates is enabled.</p>
Douglas Gregor07508002011-02-05 20:35:30 +0000485
Sean Huntd9624992011-06-23 06:11:37 +0000486<h4 id="cxx_delegating_constructors">C++0x delegating constructors</h4>
487
488<p>Use <tt>__has_feature(cxx_delegating_constructors)</tt> to determine if
489support for delegating constructors is enabled.</p>
490
Sean Hunt7e98b472011-06-23 01:21:01 +0000491<h4 id="cxx_deleted_functions">C++0x <tt>delete</tt>d functions</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000492
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000493<p>Use <tt>__has_feature(cxx_deleted_functions)</tt> or
494<tt>__has_extension(cxx_deleted_functions)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000495deleted function definitions (with <tt>= delete</tt>) is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000496
Douglas Gregorece38942011-08-29 17:28:38 +0000497<h4 id="cxx_explicit_conversions">C++0x explicit conversion functions</h3>
498<p>Use <tt>__has_feature(cxx_explicit_conversions)</tt> to determine if support for <tt>explicit</tt> conversion functions is enabled.</p>
499
Sean Hunte1f6dea2011-08-07 00:34:32 +0000500<h4 id="cxx_generalized_initializers">C++0x generalized initializers</h4>
501
502<p>Use <tt>__has_feature(cxx_generalized_initializers)</tt> to determine if
503support for generalized initializers (using braced lists and
Douglas Gregorece38942011-08-29 17:28:38 +0000504<tt>std::initializer_list</tt>) is enabled. Clang does not currently implement
505this feature.</p>
506
507<h4 id="cxx_implicit_moves">C++0x implicit move constructors/assignment operators</h4>
508
Sebastian Redl72a81d22011-10-10 18:10:00 +0000509<p>Use <tt>__has_feature(cxx_implicit_moves)</tt> to determine if Clang will
510implicitly generate move constructors and move assignment operators where needed.</p>
Douglas Gregorece38942011-08-29 17:28:38 +0000511
512<h4 id="cxx_inheriting_constructors">C++0x inheriting constructors</h4>
513
514<p>Use <tt>__has_feature(cxx_inheriting_constructors)</tt> to determine if support for inheriting constructors is enabled. Clang does not currently implement this feature.</p>
515
516<h4 id="cxx_inline_namespaces">C++0x inline namespaces</h4>
517
518<p>Use <tt>__has_feature(cxx_inline_namespaces)</tt> or
519<tt>__has_extension(cxx_inline_namespaces)</tt> to determine if support for
520inline namespaces is enabled.</p>
Sean Hunte1f6dea2011-08-07 00:34:32 +0000521
Sean Hunt7e98b472011-06-23 01:21:01 +0000522<h4 id="cxx_lambdas">C++0x lambdas</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000523
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000524<p>Use <tt>__has_feature(cxx_lambdas)</tt> or
525<tt>__has_extension(cxx_lambdas)</tt> to determine if support for lambdas
Douglas Gregorece38942011-08-29 17:28:38 +0000526is enabled. Clang does not currently implement this feature.</p>
527
528<h4 id="cxx_noexcept">C++0x noexcept</h4>
529
530<p>Use <tt>__has_feature(cxx_noexcept)</tt> or
531<tt>__has_extension(cxx_noexcept)</tt> to determine if support for noexcept
532exception specifications is enabled.</p>
533
534<h4 id="cxx_nonstatic_member_init">C++0x in-class non-static data member initialization</h4>
535
536<p>Use <tt>__has_feature(cxx_nonstatic_member_init)</tt> to determine whether in-class initialization of non-static data members is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000537
Sean Hunt7e98b472011-06-23 01:21:01 +0000538<h4 id="cxx_nullptr">C++0x <tt>nullptr</tt></h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000539
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000540<p>Use <tt>__has_feature(cxx_nullptr)</tt> or
541<tt>__has_extension(cxx_nullptr)</tt> to determine if support for
Douglas Gregor84ee2ee2011-05-21 23:15:46 +0000542<tt>nullptr</tt> is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000543
Sean Hunt7e98b472011-06-23 01:21:01 +0000544<h4 id="cxx_override_control">C++0x <tt>override control</tt></h4>
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000545
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000546<p>Use <tt>__has_feature(cxx_override_control)</tt> or
547<tt>__has_extension(cxx_override_control)</tt> to determine if support for
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000548the override control keywords is enabled.</p>
549
Sean Hunt7e98b472011-06-23 01:21:01 +0000550<h4 id="cxx_reference_qualified_functions">C++0x reference-qualified functions</h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000551<p>Use <tt>__has_feature(cxx_reference_qualified_functions)</tt> or
552<tt>__has_extension(cxx_reference_qualified_functions)</tt> to determine
553if support for reference-qualified functions (e.g., member functions with
554<code>&amp;</code> or <code>&amp;&amp;</code> applied to <code>*this</code>)
555is enabled.</p>
Douglas Gregor56209ff2011-01-26 21:25:54 +0000556
Sean Hunt7e98b472011-06-23 01:21:01 +0000557<h4 id="cxx_range_for">C++0x range-based <tt>for</tt> loop</h4>
Richard Smitha391a462011-04-15 15:14:40 +0000558
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000559<p>Use <tt>__has_feature(cxx_range_for)</tt> or
560<tt>__has_extension(cxx_range_for)</tt> to determine if support for the
561range-based for loop is enabled. </p>
Richard Smitha391a462011-04-15 15:14:40 +0000562
Douglas Gregorece38942011-08-29 17:28:38 +0000563<h4 id="cxx_raw_string_literals">C++0x raw string literals</h4>
564<p>Use <tt>__has_feature(cxx_raw_string_literals)</tt> to determine if support for raw string literals (e.g., <tt>R"foo\bar"</tt>) is enabled.</p>
565
Sean Hunt7e98b472011-06-23 01:21:01 +0000566<h4 id="cxx_rvalue_references">C++0x rvalue references</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000567
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000568<p>Use <tt>__has_feature(cxx_rvalue_references)</tt> or
569<tt>__has_extension(cxx_rvalue_references)</tt> to determine if support for
Douglas Gregor56209ff2011-01-26 21:25:54 +0000570rvalue references is enabled. </p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000571
Sean Hunt7e98b472011-06-23 01:21:01 +0000572<h4 id="cxx_static_assert">C++0x <tt>static_assert()</tt></h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000573
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000574<p>Use <tt>__has_feature(cxx_static_assert)</tt> or
575<tt>__has_extension(cxx_static_assert)</tt> to determine if support for
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000576compile-time assertions using <tt>static_assert</tt> is enabled.</p>
577
Sean Hunt7e98b472011-06-23 01:21:01 +0000578<h4 id="cxx_auto_type">C++0x type inference</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000579
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000580<p>Use <tt>__has_feature(cxx_auto_type)</tt> or
581<tt>__has_extension(cxx_auto_type)</tt> to determine C++0x type inference is
582supported using the <tt>auto</tt> specifier. If this is disabled, <tt>auto</tt>
583will instead be a storage class specifier, as in C or C++98.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000584
Douglas Gregorece38942011-08-29 17:28:38 +0000585<h4 id="cxx_strong_enums">C++0x strongly typed enumerations</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000586
Douglas Gregorece38942011-08-29 17:28:38 +0000587<p>Use <tt>__has_feature(cxx_strong_enums)</tt> or
588<tt>__has_extension(cxx_strong_enums)</tt> to determine if support for
589strongly typed, scoped enumerations is enabled.</p>
Sebastian Redlf6c09772010-08-31 23:28:47 +0000590
Sean Hunt7e98b472011-06-23 01:21:01 +0000591<h4 id="cxx_trailing_return">C++0x trailing return type</h4>
Douglas Gregordab60ad2010-10-01 18:44:50 +0000592
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000593<p>Use <tt>__has_feature(cxx_trailing_return)</tt> or
594<tt>__has_extension(cxx_trailing_return)</tt> to determine if support for the
595alternate function declaration syntax with trailing return type is enabled.</p>
Douglas Gregordab60ad2010-10-01 18:44:50 +0000596
Douglas Gregorece38942011-08-29 17:28:38 +0000597<h4 id="cxx_unicode_literals">C++0x Unicode string literals</h4>
598<p>Use <tt>__has_feature(cxx_unicode_literals)</tt> to determine if
599support for Unicode string literals is enabled.</p>
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000600
Douglas Gregorece38942011-08-29 17:28:38 +0000601<h4 id="cxx_unrestricted_unions">C++0x unrestricted unions</h4>
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000602
Douglas Gregorece38942011-08-29 17:28:38 +0000603<p>Use <tt>__has_feature(cxx_unrestricted_unions)</tt> to determine if support for unrestricted unions is enabled. Clang does not currently support this feature.</p>
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000604
Douglas Gregorece38942011-08-29 17:28:38 +0000605<h4 id="cxx_user_literals">C++0x user-defined literals</h4>
606
607<p>Use <tt>__has_feature(cxx_user_literals)</tt> to determine if support for user-defined literals is enabled. Clang does not currently support this feature.</p>
608
609<h4 id="cxx_variadic_templates">C++0x variadic templates</h4>
610
611<p>Use <tt>__has_feature(cxx_variadic_templates)</tt> or
612<tt>__has_extension(cxx_variadic_templates)</tt> to determine if support
613for variadic templates is enabled.</p>
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000614
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000615<h3 id="c1x">C1X</h3>
616
617<p>The features listed below are slated for inclusion in the upcoming
618C1X standard. As a result, all these features are enabled
619with the <tt>-std=c1x</tt> option when compiling C code.</p>
620
Sean Hunt7e98b472011-06-23 01:21:01 +0000621<h4 id="c_generic_selections">C1X generic selections</h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000622
623<p>Use <tt>__has_feature(c_generic_selections)</tt> or
624<tt>__has_extension(c_generic_selections)</tt> to determine if support for
625generic selections is enabled.</p>
626
627<p>As an extension, the C1X generic selection expression is available in all
628languages supported by Clang. The syntax is the same as that given in the
629C1X draft standard.</p>
630
631<p>In C, type compatibility is decided according to the rules given in the
632appropriate standard, but in C++, which lacks the type compatibility rules
633used in C, types are considered compatible only if they are equivalent.</p>
634
Sean Hunt7e98b472011-06-23 01:21:01 +0000635<h4 id="c_static_assert">C1X <tt>_Static_assert()</tt></h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000636
637<p>Use <tt>__has_feature(c_static_assert)</tt> or
638<tt>__has_extension(c_static_assert)</tt> to determine if support for
639compile-time assertions using <tt>_Static_assert</tt> is enabled.</p>
640
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000641<!-- ======================================================================= -->
Douglas Gregorafdf1372011-02-03 21:57:35 +0000642<h2 id="checking_type_traits">Checks for Type Traits</h2>
643<!-- ======================================================================= -->
644
Sean Hunt7e98b472011-06-23 01:21:01 +0000645<p>Clang supports the <a href="http://gcc.gnu.org/onlinedocs/gcc/Type-Traits.html">GNU C++ type traits</a> and a subset of the <a href="http://msdn.microsoft.com/en-us/library/ms177194(v=VS.100).aspx">Microsoft Visual C++ Type traits</a>. For each supported type trait <code>__X</code>, <code>__has_extension(X)</code> indicates the presence of the type trait. For example:
Douglas Gregorafdf1372011-02-03 21:57:35 +0000646<blockquote>
647<pre>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000648#if __has_extension(is_convertible_to)
Douglas Gregorafdf1372011-02-03 21:57:35 +0000649template&lt;typename From, typename To&gt;
650struct is_convertible_to {
651 static const bool value = __is_convertible_to(From, To);
652};
653#else
654// Emulate type trait
655#endif
656</pre>
657</blockquote>
658
659<p>The following type traits are supported by Clang:</p>
660<ul>
661 <li><code>__has_nothrow_assign</code> (GNU, Microsoft)</li>
662 <li><code>__has_nothrow_copy</code> (GNU, Microsoft)</li>
663 <li><code>__has_nothrow_constructor</code> (GNU, Microsoft)</li>
664 <li><code>__has_trivial_assign</code> (GNU, Microsoft)</li>
665 <li><code>__has_trivial_copy</code> (GNU, Microsoft)</li>
666 <li><code>__has_trivial_constructor</code> (GNU, Microsoft)</li>
667 <li><code>__has_trivial_destructor</code> (GNU, Microsoft)</li>
668 <li><code>__has_virtual_destructor</code> (GNU, Microsoft)</li>
669 <li><code>__is_abstract</code> (GNU, Microsoft)</li>
670 <li><code>__is_base_of</code> (GNU, Microsoft)</li>
671 <li><code>__is_class</code> (GNU, Microsoft)</li>
672 <li><code>__is_convertible_to</code> (Microsoft)</li>
673 <li><code>__is_empty</code> (GNU, Microsoft)</li>
674 <li><code>__is_enum</code> (GNU, Microsoft)</li>
675 <li><code>__is_pod</code> (GNU, Microsoft)</li>
676 <li><code>__is_polymorphic</code> (GNU, Microsoft)</li>
677 <li><code>__is_union</code> (GNU, Microsoft)</li>
678 <li><code>__is_literal(type)</code>: Determines whether the given type is a literal type</li>
Sean Hunt1fba8282011-07-18 17:22:33 +0000679 <li><code>__underlying_type(type)</code>: Retrieves the underlying type for a given <code>enum</code> type. This trait is required to implement the C++0x standard library.</li>
Douglas Gregorafdf1372011-02-03 21:57:35 +0000680</ul>
681
682<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000683<h2 id="blocks">Blocks</h2>
684<!-- ======================================================================= -->
685
Chris Lattnera7dbdf52009-03-09 07:03:22 +0000686<p>The syntax and high level language feature description is in <a
687href="BlockLanguageSpec.txt">BlockLanguageSpec.txt</a>. Implementation and ABI
688details for the clang implementation are in <a
Chris Lattner5d7650b2010-03-16 21:43:03 +0000689href="Block-ABI-Apple.txt">Block-ABI-Apple.txt</a>.</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000690
Chris Lattner148772a2009-06-13 07:13:28 +0000691
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000692<p>Query for this feature with __has_extension(blocks).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000693
Chris Lattner5ce933f2009-02-09 08:46:11 +0000694<!-- ======================================================================= -->
Douglas Gregor926df6c2011-06-11 01:09:30 +0000695<h2 id="objc_features">Objective-C Features</h2>
696<!-- ======================================================================= -->
697
698<h3 id="objc_instancetype">Related result types</h3>
699
700<p>According to Cocoa conventions, Objective-C methods with certain names ("init", "alloc", etc.) always return objects that are an instance of the receiving class's type. Such methods are said to have a "related result type", meaning that a message send to one of these methods will have the same static type as an instance of the receiver class. For example, given the following classes:</p>
701
702<blockquote>
703<pre>
704@interface NSObject
705+ (id)alloc;
706- (id)init;
707@end
708
709@interface NSArray : NSObject
710@end
711</pre>
712</blockquote>
713
714<p>and this common initialization pattern</p>
715
716<blockquote>
717<pre>
718NSArray *array = [[NSArray alloc] init];
719</pre>
720</blockquote>
721
722<p>the type of the expression <code>[NSArray alloc]</code> is
723<code>NSArray*</code> because <code>alloc</code> implicitly has a
724related result type. Similarly, the type of the expression
725<code>[[NSArray alloc] init]</code> is <code>NSArray*</code>, since
726<code>init</code> has a related result type and its receiver is known
727to have the type <code>NSArray *</code>. If neither <code>alloc</code> nor <code>init</code> had a related result type, the expressions would have had type <code>id</code>, as declared in the method signature.</p>
728
Douglas Gregore97179c2011-09-08 01:46:34 +0000729<p>A method with a related result type can be declared by using the
730type <tt>instancetype</tt> as its result type. <tt>instancetype</tt>
731is a contextual keyword that is only permitted in the result type of
732an Objective-C method, e.g.</p>
733
734<pre>
735@interface A
736+ (<b>instancetype</b>)constructAnA;
737@end
738</pre>
739
740<p>The related result type can also be inferred for some methods.
741To determine whether a method has an inferred related result type, the first
Douglas Gregor926df6c2011-06-11 01:09:30 +0000742word in the camel-case selector (e.g., "init" in "initWithObjects") is
743considered, and the method will a related result type if its return
Sean Hunt7e98b472011-06-23 01:21:01 +0000744type is compatible with the type of its class and if</p>
Douglas Gregor926df6c2011-06-11 01:09:30 +0000745
746<ul>
747
748 <li>the first word is "alloc" or "new", and the method is a class
749 method, or</li>
750
751 <li>the first word is "autorelease", "init", "retain", or "self",
752 and the method is an instance method.</li>
753
Sean Hunt7e98b472011-06-23 01:21:01 +0000754</ul>
Douglas Gregor926df6c2011-06-11 01:09:30 +0000755
756<p>If a method with a related result type is overridden by a subclass
757method, the subclass method must also return a type that is compatible
758with the subclass type. For example:</p>
759
760<blockquote>
761<pre>
762@interface NSString : NSObject
763- (NSUnrelated *)init; // incorrect usage: NSUnrelated is not NSString or a superclass of NSString
764@end
765</pre>
766</blockquote>
767
768<p>Related result types only affect the type of a message send or
769property access via the given method. In all other respects, a method
Douglas Gregore97179c2011-09-08 01:46:34 +0000770with a related result type is treated the same way as method that
771returns <tt>id</tt>.</p>
Douglas Gregor926df6c2011-06-11 01:09:30 +0000772
Douglas Gregoraebb6532011-09-08 17:19:31 +0000773<p>Use <tt>__has_feature(objc_instancetype)</tt> to determine whether
774the <tt>instancetype</tt> contextual keyword is available.</p>
775
Douglas Gregor926df6c2011-06-11 01:09:30 +0000776<!-- ======================================================================= -->
John McCallf85e1932011-06-15 23:02:42 +0000777<h2 id="objc_arc">Automatic reference counting </h2>
778<!-- ======================================================================= -->
779
780<p>Clang provides support for <a href="AutomaticReferenceCounting.html">automated reference counting</a> in Objective-C, which eliminates the need for manual retain/release/autorelease message sends. There are two feature macros associated with automatic reference counting: <code>__has_feature(objc_arc)</code> indicates the availability of automated reference counting in general, while <code>__has_feature(objc_arc_weak)</code> indicates that automated reference counting also includes support for <code>__weak</code> pointers to Objective-C objects.</p>
781
782<!-- ======================================================================= -->
Douglas Gregor5471bc82011-09-08 17:18:35 +0000783<h2 id="objc_fixed_enum">Enumerations with a fixed underlying type</h2>
784<!-- ======================================================================= -->
785
786<p>Clang provides support for C++0x enumerations with a fixed
787underlying type within Objective-C. For example, one can write an
788enumeration type as:</p>
789
790<pre>
791typedef enum : unsigned char { Red, Green, Blue } Color;
792</pre>
793
794<p>This specifies that the underlying type, which is used to store the
795enumeration value, is <tt>unsigned char</tt>.</p>
796
797<p>Use <tt>__has_feature(objc_fixed_enum)</tt> to determine whether
798support for fixed underlying types is available in Objective-C.</p>
799
800<!-- ======================================================================= -->
Douglas Gregorcb54d432009-02-13 00:57:04 +0000801<h2 id="overloading-in-c">Function Overloading in C</h2>
802<!-- ======================================================================= -->
803
Chris Lattnerf161d412009-02-13 21:51:45 +0000804<p>Clang provides support for C++ function overloading in C. Function
805overloading in C is introduced using the <tt>overloadable</tt> attribute. For
806example, one might provide several overloaded versions of a <tt>tgsin</tt>
807function that invokes the appropriate standard function computing the sine of a
808value with <tt>float</tt>, <tt>double</tt>, or <tt>long double</tt>
809precision:</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000810
811<blockquote>
812<pre>
813#include &lt;math.h&gt;
814float <b>__attribute__((overloadable))</b> tgsin(float x) { return sinf(x); }
815double <b>__attribute__((overloadable))</b> tgsin(double x) { return sin(x); }
816long double <b>__attribute__((overloadable))</b> tgsin(long double x) { return sinl(x); }
817</pre>
818</blockquote>
819
820<p>Given these declarations, one can call <tt>tgsin</tt> with a
821<tt>float</tt> value to receive a <tt>float</tt> result, with a
822<tt>double</tt> to receive a <tt>double</tt> result, etc. Function
823overloading in C follows the rules of C++ function overloading to pick
824the best overload given the call arguments, with a few C-specific
825semantics:</p>
826<ul>
827 <li>Conversion from <tt>float</tt> or <tt>double</tt> to <tt>long
828 double</tt> is ranked as a floating-point promotion (per C99) rather
829 than as a floating-point conversion (as in C++).</li>
830
831 <li>A conversion from a pointer of type <tt>T*</tt> to a pointer of type
832 <tt>U*</tt> is considered a pointer conversion (with conversion
833 rank) if <tt>T</tt> and <tt>U</tt> are compatible types.</li>
834
835 <li>A conversion from type <tt>T</tt> to a value of type <tt>U</tt>
836 is permitted if <tt>T</tt> and <tt>U</tt> are compatible types. This
837 conversion is given "conversion" rank.</li>
838</ul>
839
840<p>The declaration of <tt>overloadable</tt> functions is restricted to
841function declarations and definitions. Most importantly, if any
842function with a given name is given the <tt>overloadable</tt>
843attribute, then all function declarations and definitions with that
844name (and in that scope) must have the <tt>overloadable</tt>
Chris Lattnerf161d412009-02-13 21:51:45 +0000845attribute. This rule even applies to redeclarations of functions whose original
846declaration had the <tt>overloadable</tt> attribute, e.g.,</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000847
848<blockquote>
849<pre>
850int f(int) __attribute__((overloadable));
851float f(float); <i>// error: declaration of "f" must have the "overloadable" attribute</i>
852
853int g(int) __attribute__((overloadable));
854int g(int) { } <i>// error: redeclaration of "g" must also have the "overloadable" attribute</i>
855</pre>
856</blockquote>
857
Douglas Gregor965acbb2009-02-18 07:07:28 +0000858<p>Functions marked <tt>overloadable</tt> must have
859prototypes. Therefore, the following code is ill-formed:</p>
860
861<blockquote>
862<pre>
863int h() __attribute__((overloadable)); <i>// error: h does not have a prototype</i>
864</pre>
865</blockquote>
866
867<p>However, <tt>overloadable</tt> functions are allowed to use a
868ellipsis 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>
869
870<blockquote>
871<pre>
Chris Lattner02246802009-02-18 22:27:46 +0000872void honeypot(...) __attribute__((overloadable, unavailable)); <i>// calling me is an error</i>
Douglas Gregor965acbb2009-02-18 07:07:28 +0000873</pre>
874</blockquote>
875
Douglas Gregorcb54d432009-02-13 00:57:04 +0000876<p>Functions declared with the <tt>overloadable</tt> attribute have
877their names mangled according to the same rules as C++ function
878names. For example, the three <tt>tgsin</tt> functions in our
879motivating example get the mangled names <tt>_Z5tgsinf</tt>,
Chris Lattner71b48d62010-11-28 18:19:13 +0000880<tt>_Z5tgsind</tt>, and <tt>_Z5tgsine</tt>, respectively. There are two
Douglas Gregorcb54d432009-02-13 00:57:04 +0000881caveats to this use of name mangling:</p>
882
883<ul>
884
885 <li>Future versions of Clang may change the name mangling of
886 functions overloaded in C, so you should not depend on an specific
887 mangling. To be completely safe, we strongly urge the use of
888 <tt>static inline</tt> with <tt>overloadable</tt> functions.</li>
889
890 <li>The <tt>overloadable</tt> attribute has almost no meaning when
891 used in C++, because names will already be mangled and functions are
892 already overloadable. However, when an <tt>overloadable</tt>
893 function occurs within an <tt>extern "C"</tt> linkage specification,
894 it's name <i>will</i> be mangled in the same way as it would in
895 C.</li>
896</ul>
897
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000898<p>Query for this feature with __has_extension(attribute_overloadable).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000899
Eli Friedman0c706c22011-09-19 23:17:44 +0000900<!-- ======================================================================= -->
901<h2 id="complex-list-init">Initializer lists for complex numbers in C</h2>
902<!-- ======================================================================= -->
903
904<p>clang supports an extension which allows the following in C:</p>
905
906<blockquote>
907<pre>
908#include &lt;math.h&gt;
909#include &lt;complex.h&gt;
910complex float x = { 1.0f, INFINITY }; // Init to (1, Inf)
911</pre>
912</blockquote>
913
914<p>This construct is useful because there is no way to separately
915initialize the real and imaginary parts of a complex variable in
916standard C, given that clang does not support <code>_Imaginary</code>.
917(clang also supports the <code>__real__</code> and <code>__imag__</code>
918extensions from gcc, which help in some cases, but are not usable in
919static initializers.)
920
921<p>Note that this extension does not allow eliding the braces; the
922meaning of the following two lines is different:</p>
923
924<blockquote>
925<pre>
926complex float x[] = { { 1.0f, 1.0f } }; // [0] = (1, 1)
927complex float x[] = { 1.0f, 1.0f }; // [0] = (1, 0), [1] = (1, 0)
928</pre>
929</blockquote>
930
931<p>This extension also works in C++ mode, as far as that goes, but does not
932 apply to the C++ <code>std::complex</code>. (In C++11, list
933 initialization allows the same syntax to be used with
934 <code>std::complex</code> with the same meaning.)
Chris Lattner148772a2009-06-13 07:13:28 +0000935
Douglas Gregorcb54d432009-02-13 00:57:04 +0000936<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000937<h2 id="builtins">Builtin Functions</h2>
938<!-- ======================================================================= -->
939
940<p>Clang supports a number of builtin library functions with the same syntax as
941GCC, including things like <tt>__builtin_nan</tt>,
942<tt>__builtin_constant_p</tt>, <tt>__builtin_choose_expr</tt>,
943<tt>__builtin_types_compatible_p</tt>, <tt>__sync_fetch_and_add</tt>, etc. In
944addition to the GCC builtins, Clang supports a number of builtins that GCC does
945not, which are listed here.</p>
946
947<p>Please note that Clang does not and will not support all of the GCC builtins
948for vector operations. Instead of using builtins, you should use the functions
949defined in target-specific header files like <tt>&lt;xmmintrin.h&gt;</tt>, which
950define portable wrappers for these. Many of the Clang versions of these
951functions are implemented directly in terms of <a href="#vectors">extended
952vector support</a> instead of builtins, in order to reduce the number of
953builtins that we need to implement.</p>
954
Chris Lattner5ce933f2009-02-09 08:46:11 +0000955<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000956<h3><a name="__builtin_shufflevector">__builtin_shufflevector</a></h3>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000957<!-- ======================================================================= -->
958
Chris Lattneraad826b2009-09-16 18:56:12 +0000959<p><tt>__builtin_shufflevector</tt> is used to express generic vector
Chris Lattner6f72da52009-02-13 20:00:20 +0000960permutation/shuffle/swizzle operations. This builtin is also very important for
961the implementation of various target-specific header files like
962<tt>&lt;xmmintrin.h&gt;</tt>.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000963</p>
964
965<p><b>Syntax:</b></p>
966
967<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000968__builtin_shufflevector(vec1, vec2, index1, index2, ...)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000969</pre>
970
971<p><b>Examples:</b></p>
972
973<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000974 // Identity operation - return 4-element vector V1.
975 __builtin_shufflevector(V1, V1, 0, 1, 2, 3)
976
977 // "Splat" element 0 of V1 into a 4-element result.
978 __builtin_shufflevector(V1, V1, 0, 0, 0, 0)
979
980 // Reverse 4-element vector V1.
981 __builtin_shufflevector(V1, V1, 3, 2, 1, 0)
982
983 // Concatenate every other element of 4-element vectors V1 and V2.
984 __builtin_shufflevector(V1, V2, 0, 2, 4, 6)
985
986 // Concatenate every other element of 8-element vectors V1 and V2.
987 __builtin_shufflevector(V1, V2, 0, 2, 4, 6, 8, 10, 12, 14)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000988</pre>
989
990<p><b>Description:</b></p>
991
Chris Lattner6f72da52009-02-13 20:00:20 +0000992<p>The first two arguments to __builtin_shufflevector are vectors that have the
993same element type. The remaining arguments are a list of integers that specify
994the elements indices of the first two vectors that should be extracted and
995returned in a new vector. These element indices are numbered sequentially
996starting with the first vector, continuing into the second vector. Thus, if
997vec1 is a 4-element vector, index 5 would refer to the second element of vec2.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000998</p>
999
Chris Lattner6f72da52009-02-13 20:00:20 +00001000<p>The result of __builtin_shufflevector is a vector
1001with the same element type as vec1/vec2 but that has an element count equal to
1002the number of indices specified.
1003</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +00001004
Chris Lattner21190d52009-09-21 03:09:59 +00001005<p>Query for this feature with __has_builtin(__builtin_shufflevector).</p>
1006
1007<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +00001008<h3><a name="__builtin_unreachable">__builtin_unreachable</a></h3>
Chris Lattner21190d52009-09-21 03:09:59 +00001009<!-- ======================================================================= -->
1010
1011<p><tt>__builtin_unreachable</tt> is used to indicate that a specific point in
1012the program cannot be reached, even if the compiler might otherwise think it
1013can. This is useful to improve optimization and eliminates certain warnings.
1014For example, without the <tt>__builtin_unreachable</tt> in the example below,
1015the compiler assumes that the inline asm can fall through and prints a "function
1016declared 'noreturn' should not return" warning.
1017</p>
1018
1019<p><b>Syntax:</b></p>
1020
1021<pre>
1022__builtin_unreachable()
1023</pre>
1024
1025<p><b>Example of Use:</b></p>
1026
1027<pre>
1028void myabort(void) __attribute__((noreturn));
1029void myabort(void) {
1030 asm("int3");
1031 __builtin_unreachable();
1032}
1033</pre>
1034
1035<p><b>Description:</b></p>
1036
1037<p>The __builtin_unreachable() builtin has completely undefined behavior. Since
1038it has undefined behavior, it is a statement that it is never reached and the
1039optimizer can take advantage of this to produce better code. This builtin takes
1040no arguments and produces a void result.
1041</p>
1042
1043<p>Query for this feature with __has_builtin(__builtin_unreachable).</p>
1044
Chris Lattner23aa9c82011-04-09 03:57:26 +00001045<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +00001046<h3><a name="__sync_swap">__sync_swap</a></h3>
Chris Lattner23aa9c82011-04-09 03:57:26 +00001047<!-- ======================================================================= -->
1048
1049<p><tt>__sync_swap</tt> is used to atomically swap integers or pointers in
1050memory.
1051</p>
1052
1053<p><b>Syntax:</b></p>
1054
1055<pre>
1056<i>type</i> __sync_swap(<i>type</i> *ptr, <i>type</i> value, ...)
1057</pre>
1058
1059<p><b>Example of Use:</b></p>
1060
1061<pre>
Sean Hunt7e98b472011-06-23 01:21:01 +00001062int old_value = __sync_swap(&amp;value, new_value);
Chris Lattner23aa9c82011-04-09 03:57:26 +00001063</pre>
1064
1065<p><b>Description:</b></p>
1066
1067<p>The __sync_swap() builtin extends the existing __sync_*() family of atomic
1068intrinsics to allow code to atomically swap the current value with the new
1069value. More importantly, it helps developers write more efficient and correct
1070code by avoiding expensive loops around __sync_bool_compare_and_swap() or
1071relying on the platform specific implementation details of
1072__sync_lock_test_and_set(). The __sync_swap() builtin is a full barrier.
1073</p>
1074
Chris Lattner21190d52009-09-21 03:09:59 +00001075
Chris Lattner1177f912009-04-09 19:58:15 +00001076<!-- ======================================================================= -->
1077<h2 id="targetspecific">Target-Specific Extensions</h2>
1078<!-- ======================================================================= -->
1079
1080<p>Clang supports some language features conditionally on some targets.</p>
1081
1082<!-- ======================================================================= -->
1083<h3 id="x86-specific">X86/X86-64 Language Extensions</h3>
1084<!-- ======================================================================= -->
1085
1086<p>The X86 backend has these language extensions:</p>
1087
1088<!-- ======================================================================= -->
1089<h4 id="x86-gs-segment">Memory references off the GS segment</h4>
1090<!-- ======================================================================= -->
1091
1092<p>Annotating a pointer with address space #256 causes it to be code generated
Chris Lattnera021e7c2009-05-05 18:54:47 +00001093relative to the X86 GS segment register, and address space #257 causes it to be
1094relative to the X86 FS segment. Note that this is a very very low-level
1095feature that should only be used if you know what you're doing (for example in
1096an OS kernel).</p>
Chris Lattner1177f912009-04-09 19:58:15 +00001097
1098<p>Here is an example:</p>
1099
1100<pre>
1101#define GS_RELATIVE __attribute__((address_space(256)))
1102int foo(int GS_RELATIVE *P) {
1103 return *P;
1104}
1105</pre>
1106
1107<p>Which compiles to (on X86-32):</p>
1108
1109<pre>
1110_foo:
1111 movl 4(%esp), %eax
1112 movl %gs:(%eax), %eax
1113 ret
1114</pre>
1115
Ted Kremeneked869312009-04-10 05:03:33 +00001116<!-- ======================================================================= -->
1117<h2 id="analyzerspecific">Static Analysis-Specific Extensions</h2>
1118<!-- ======================================================================= -->
1119
1120<p>Clang supports additional attributes that are useful for documenting program
1121invariants and rules for static analysis tools. The extensions documented here
1122are used by the <a
1123href="http://clang.llvm.org/StaticAnalysis.html">path-sensitive static analyzer
1124engine</a> that is part of Clang's Analysis library.</p>
1125
John McCall87494012011-03-18 03:51:49 +00001126<h3 id="attr_analyzer_noreturn">The <tt>analyzer_noreturn</tt> attribute</h3>
Ted Kremeneked869312009-04-10 05:03:33 +00001127
1128<p>Clang's static analysis engine understands the standard <tt>noreturn</tt>
Ted Kremenek4df21142009-04-10 05:04:22 +00001129attribute. This attribute, which is typically affixed to a function prototype,
1130indicates that a call to a given function never returns. Function prototypes for
1131common functions like <tt>exit</tt> are typically annotated with this attribute,
1132as well as a variety of common assertion handlers. Users can educate the static
1133analyzer about their own custom assertion handles (thus cutting down on false
1134positives due to false paths) by marking their own &quot;panic&quot; functions
1135with this attribute.</p>
Ted Kremeneked869312009-04-10 05:03:33 +00001136
1137<p>While useful, <tt>noreturn</tt> is not applicable in all cases. Sometimes
Nick Lewycky625b5862009-06-14 04:08:08 +00001138there are special functions that for all intents and purposes should be
1139considered panic functions (i.e., they are only called when an internal program
1140error occurs) but may actually return so that the program can fail gracefully.
1141The <tt>analyzer_noreturn</tt> attribute allows one to annotate such functions
1142as being interpreted as &quot;no return&quot; functions by the analyzer (thus
Chris Lattner28935892009-04-10 05:54:56 +00001143pruning bogus paths) but will not affect compilation (as in the case of
Ted Kremeneked869312009-04-10 05:03:33 +00001144<tt>noreturn</tt>).</p>
1145
1146<p><b>Usage</b>: The <tt>analyzer_noreturn</tt> attribute can be placed in the
Chris Lattner28935892009-04-10 05:54:56 +00001147same places where the <tt>noreturn</tt> attribute can be placed. It is commonly
Ted Kremeneked869312009-04-10 05:03:33 +00001148placed at the end of function prototypes:</p>
1149
1150<pre>
1151 void foo() <b>__attribute__((analyzer_noreturn))</b>;
Chris Lattner148772a2009-06-13 07:13:28 +00001152</pre>
1153
John McCall87494012011-03-18 03:51:49 +00001154<p>Query for this feature with
1155<tt>__has_attribute(analyzer_noreturn)</tt>.</p>
Chris Lattner148772a2009-06-13 07:13:28 +00001156
John McCall87494012011-03-18 03:51:49 +00001157<h3 id="attr_method_family">The <tt>objc_method_family</tt> attribute</h3>
1158
1159<p>Many methods in Objective-C have conventional meanings determined
1160by their selectors. For the purposes of static analysis, it is
1161sometimes useful to be able to mark a method as having a particular
1162conventional meaning despite not having the right selector, or as not
1163having the conventional meaning that its selector would suggest.
1164For these use cases, we provide an attribute to specifically describe
1165the <q>method family</q> that a method belongs to.</p>
1166
1167<p><b>Usage</b>: <tt>__attribute__((objc_method_family(X)))</tt>,
1168where <tt>X</tt> is one of <tt>none</tt>, <tt>alloc</tt>, <tt>copy</tt>,
1169<tt>init</tt>, <tt>mutableCopy</tt>, or <tt>new</tt>. This attribute
1170can only be placed at the end of a method declaration:</p>
1171
1172<pre>
1173 - (NSString*) initMyStringValue <b>__attribute__((objc_method_family(none)))</b>;
1174</pre>
1175
1176<p>Users who do not wish to change the conventional meaning of a
1177method, and who merely want to document its non-standard retain and
1178release semantics, should use the
1179<a href="#attr_retain_release">retaining behavior attributes</a>
1180described below.</p>
1181
1182<p>Query for this feature with
1183<tt>__has_attribute(objc_method_family)</tt>.</p>
1184
1185<h3 id="attr_retain_release">Objective-C retaining behavior attributes</h3>
John McCall630b7ae2011-01-25 04:26:21 +00001186
1187<p>In Objective-C, functions and methods are generally assumed to take
1188and return objects with +0 retain counts, with some exceptions for
1189special methods like <tt>+alloc</tt> and <tt>init</tt>. However,
1190there are exceptions, and so Clang provides attributes to allow these
1191exceptions to be documented, which helps the analyzer find leaks (and
John McCall87494012011-03-18 03:51:49 +00001192ignore non-leaks). Some exceptions may be better described using
1193the <a href="#attr_method_family"><tt>objc_method_family</tt></a>
1194attribute instead.</p>
John McCall630b7ae2011-01-25 04:26:21 +00001195
1196<p><b>Usage</b>: The <tt>ns_returns_retained</tt>, <tt>ns_returns_not_retained</tt>,
1197<tt>ns_returns_autoreleased</tt>, <tt>cf_returns_retained</tt>,
1198and <tt>cf_returns_not_retained</tt> attributes can be placed on
1199methods and functions that return Objective-C or CoreFoundation
1200objects. They are commonly placed at the end of a function prototype
1201or method declaration:</p>
1202
1203<pre>
1204 id foo() <b>__attribute__((ns_returns_retained))</b>;
1205
1206 - (NSString*) bar: (int) x <b>__attribute__((ns_returns_retained))</b>;
1207</pre>
1208
1209<p>The <tt>*_returns_retained</tt> attributes specify that the
1210returned object has a +1 retain count.
1211The <tt>*_returns_not_retained</tt> attributes specify that the return
1212object has a +0 retain count, even if the normal convention for its
1213selector would be +1. <tt>ns_returns_autoreleased</tt> specifies that the
1214returned object is +0, but is guaranteed to live at least as long as the
1215next flush of an autorelease pool.</p>
1216
1217<p><b>Usage</b>: The <tt>ns_consumed</tt> and <tt>cf_consumed</tt>
1218attributes can be placed on an parameter declaration; they specify
1219that the argument is expected to have a +1 retain count, which will be
1220balanced in some way by the function or method.
1221The <tt>ns_consumes_self</tt> attribute can only be placed on an
1222Objective-C method; it specifies that the method expects
1223its <tt>self</tt> parameter to have a +1 retain count, which it will
1224balance in some way.</p>
1225
1226<pre>
1227 void <b>foo(__attribute__((ns_consumed))</b> NSString *string);
1228
1229 - (void) bar <b>__attribute__((ns_consumes_self))</b>;
1230 - (void) baz: (id) <b>__attribute__((ns_consumed))</b> x;
1231</pre>
Ted Kremeneked869312009-04-10 05:03:33 +00001232
John McCall87494012011-03-18 03:51:49 +00001233<p>Query for these features with <tt>__has_attribute(ns_consumed)</tt>,
1234<tt>__has_attribute(ns_returns_retained)</tt>, etc.</p>
1235
Caitlin Sadowskifdde9e72011-07-28 17:21:07 +00001236
1237<!-- ======================================================================= -->
Caitlin Sadowski73cbbc82011-07-28 18:38:36 +00001238<h2 id="threadsafety">Thread-Safety Annotation Checking</h2>
Caitlin Sadowskifdde9e72011-07-28 17:21:07 +00001239<!-- ======================================================================= -->
1240
1241<p>Clang supports additional attributes for checking basic locking policies in
1242multithreaded programs.
1243Clang currently parses the following list of attributes, although
1244<b>the implementation for these annotations is currently in development.</b>
1245For more details, see the
1246<a href="http://gcc.gnu.org/wiki/ThreadSafetyAnnotation">GCC implementation</a>.
1247</p>
1248
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001249<h4 id="ts_noanal">no_thread_safety_analysis</h4>
Caitlin Sadowskifdde9e72011-07-28 17:21:07 +00001250
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001251<p>Use <tt>__attribute__((no_thread_safety_analysis))</tt> on a function
1252declaration to specify that the thread safety analysis should not be run on that
1253function. This attribute provides an escape hatch (e.g. for situations when it
1254is difficult to annotate the locking policy). </p>
Caitlin Sadowskifdde9e72011-07-28 17:21:07 +00001255
1256<h4 id="ts_lockable">lockable</h4>
1257
1258<p>Use <tt>__attribute__((lockable))</tt> on a class definition to specify
1259that it has a lockable type (e.g. a Mutex class). This annotation is primarily
1260used to check consistency.</p>
1261
1262<h4 id="ts_scopedlockable">scoped_lockable</h4>
1263
1264<p>Use <tt>__attribute__((scoped_lockable))</tt> on a class definition to
1265specify that it has a "scoped" lockable type. Objects of this type will acquire
1266the lock upon construction and release it upon going out of scope.
1267 This annotation is primarily used to check
1268consistency.</p>
1269
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001270<h4 id="ts_guardedvar">guarded_var</h4>
Caitlin Sadowskifdde9e72011-07-28 17:21:07 +00001271
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001272<p>Use <tt>__attribute__((guarded_var))</tt> on a variable declaration to
1273specify that the variable must be accessed while holding some lock.</p>
1274
1275<h4 id="ts_ptguardedvar">pt_guarded_var</h4>
1276
1277<p>Use <tt>__attribute__((pt_guarded_var))</tt> on a pointer declaration to
1278specify that the pointer must be dereferenced while holding some lock.</p>
1279
1280<h4 id="ts_guardedby">guarded_by(l)</h4>
1281
1282<p>Use <tt>__attribute__((guarded_by(l)))</tt> on a variable declaration to
Caitlin Sadowskib51e0312011-08-09 17:59:31 +00001283specify that the variable must be accessed while holding lock <tt>l</tt>.</p>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001284
1285<h4 id="ts_ptguardedby">pt_guarded_by(l)</h4>
1286
1287<p>Use <tt>__attribute__((pt_guarded_by(l)))</tt> on a pointer declaration to
Caitlin Sadowskib51e0312011-08-09 17:59:31 +00001288specify that the pointer must be dereferenced while holding lock <tt>l</tt>.</p>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001289
1290<h4 id="ts_acquiredbefore">acquired_before(...)</h4>
1291
1292<p>Use <tt>__attribute__((acquired_before(...)))</tt> on a declaration
1293of a lockable variable to specify that the lock must be acquired before all
1294attribute arguments. Arguments must be lockable type, and there must be at
1295least one argument.</p>
1296
1297<h4 id="ts_acquiredafter">acquired_after(...)</h4>
1298
1299<p>Use <tt>__attribute__((acquired_after(...)))</tt> on a declaration
1300of a lockable variable to specify that the lock must be acquired after all
1301attribute arguments. Arguments must be lockable type, and there must be at
1302least one argument.</p>
1303
1304<h4 id="ts_elf">exclusive_lock_function(...)</h4>
1305
1306<p>Use <tt>__attribute__((exclusive_lock_function(...)))</tt> on a function
1307declaration to specify that the function acquires all listed locks
Caitlin Sadowskib51e0312011-08-09 17:59:31 +00001308exclusively. This attribute takes zero or more arguments: either of lockable
1309type or integers indexing into function parameters of lockable type. If no
1310arguments are given, the acquired lock is implicitly <tt>this</tt> of the
1311enclosing object.</p>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001312
1313<h4 id="ts_slf">shared_lock_function(...)</h4>
1314
1315<p>Use <tt>__attribute__((shared_lock_function(...)))</tt> on a function
1316declaration to specify that the function acquires all listed locks, although
Caitlin Sadowskib51e0312011-08-09 17:59:31 +00001317 the locks may be shared (e.g. read locks). This attribute takes zero or more
1318arguments: either of lockable type or integers indexing into function
1319parameters of lockable type. If no arguments are given, the acquired lock is
1320implicitly <tt>this</tt> of the enclosing object.</p>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001321
1322<h4 id="ts_etf">exclusive_trylock_function(...)</h4>
1323
1324<p>Use <tt>__attribute__((exclusive_lock_function(...)))</tt> on a function
1325declaration to specify that the function will try (without blocking) to acquire
Caitlin Sadowskib51e0312011-08-09 17:59:31 +00001326all listed locks exclusively. This attribute takes one or more arguments. The
1327first argument is an integer or boolean value specifying the return value of a
1328successful lock acquisition. The remaining arugments are either of lockable type
1329or integers indexing into function parameters of lockable type. If only one
1330argument is given, the acquired lock is implicitly <tt>this</tt> of the
1331enclosing object.</p>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001332
1333<h4 id="ts_stf">shared_trylock_function(...)</h4>
1334
1335<p>Use <tt>__attribute__((shared_lock_function(...)))</tt> on a function
1336declaration to specify that the function will try (without blocking) to acquire
Caitlin Sadowskib51e0312011-08-09 17:59:31 +00001337all listed locks, although the locks may be shared (e.g. read locks). This
1338attribute takes one or more arguments. The first argument is an integer or
1339boolean value specifying the return value of a successful lock acquisition. The
1340remaining arugments are either of lockable type or integers indexing into
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001341function parameters of lockable type. If only one argument is given, the
1342acquired lock is implicitly <tt>this</tt> of the enclosing object.</p>
1343
1344<h4 id="ts_uf">unlock_function(...)</h4>
1345
1346<p>Use <tt>__attribute__((unlock_function(...)))</tt> on a function
Caitlin Sadowskib51e0312011-08-09 17:59:31 +00001347declaration to specify that the function release all listed locks. This
1348attribute takes zero or more arguments: either of lockable type or integers
1349indexing into function parameters of lockable type. If no arguments are given,
1350the acquired lock is implicitly <tt>this</tt> of the enclosing object.</p>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001351
1352<h4 id="ts_lr">lock_returned(l)</h4>
1353
1354<p>Use <tt>__attribute__((lock_returned(l)))</tt> on a function
Caitlin Sadowskib51e0312011-08-09 17:59:31 +00001355declaration to specify that the function returns lock <tt>l</tt> (<tt>l</tt>
1356must be of lockable type). This annotation is used to aid in resolving lock
1357expressions.</p>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001358
1359<h4 id="ts_le">locks_excluded(...)</h4>
1360
1361<p>Use <tt>__attribute__((locks_excluded(...)))</tt> on a function declaration
Caitlin Sadowskib51e0312011-08-09 17:59:31 +00001362to specify that the function must not be called with the listed locks. Arguments
1363must be lockable type, and there must be at least one argument.</p>
Caitlin Sadowskidb33e142011-07-28 20:12:35 +00001364
1365<h4 id="ts_elr">exclusive_locks_required(...)</h4>
1366
1367<p>Use <tt>__attribute__((exclusive_locks_required(...)))</tt> on a function
1368declaration to specify that the function must be called while holding the listed
1369exclusive locks. Arguments must be lockable type, and there must be at
1370least one argument.</p>
1371
1372<h4 id="ts_slr">shared_locks_required(...)</h4>
1373
1374<p>Use <tt>__attribute__((shared_locks_required(...)))</tt> on a function
1375declaration to specify that the function must be called while holding the listed
1376shared locks. Arguments must be lockable type, and there must be at
1377least one argument.</p>
Caitlin Sadowskifdde9e72011-07-28 17:21:07 +00001378
Chris Lattner5ce933f2009-02-09 08:46:11 +00001379</div>
1380</body>
1381</html>