Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / clang / 9dffe6f51c676c1e423c382c62d1648746e36cd4 / . / docs / LanguageExtensions.html
blob: 9779c3ff441f2e9f3590813a058d7fffcc9def20 [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>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]26<li><a href="#checking_language_features">Checks for Standard Language Features</a></li>
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]27 <ul>
28 <li><a href="#cxx_exceptions">C++ exceptions</a></li>
29 <li><a href="#cxx_rtti">C++ RTTI</a></li>
30 </ul>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]31<li><a href="#checking_upcoming_features">Checks for Upcoming Standard Language Features</a></li>
32 <ul>
33 <li><a href="#cxx_attributes">C++0x attributes</a></li>
34 <li><a href="#cxx_decltype">C++0x <tt>decltype()</tt></a></li>
35 <li><a href="#cxx_deleted_functions">C++0x deleted functions</a></li>
36 <li><a href="#cxx_concepts">C++ TR concepts</a></li>
37 <li><a href="#cxx_lambdas">C++0x lambdas</a></li>
38 <li><a href="#cxx_nullptr">C++0x nullptr</a></li>
39 <li><a href="#cxx_rvalue_references">C++0x rvalue references</a></li>
40 <li><a href="#cxx_static_assert">C++0x <tt>static_assert()</tt></a></li>
41 <li><a href="#cxx_auto_type">C++0x type inference</a></li>
42 <li><a href="#cxx_variadic_templates">C++0x variadic templates</a></li>
43 </ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]44<li><a href="#blocks">Blocks</a></li>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]45<li><a href="#overloading-in-c">Function Overloading in C</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]46<li><a href="#builtins">Builtin Functions</a>
47 <ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]48 <li><a href="#__builtin_shufflevector">__builtin_shufflevector</a></li>
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]49 <li><a href="#__builtin_unreachable">__builtin_unreachable</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]50 </ul>
51</li>
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]52<li><a href="#targetspecific">Target-Specific Extensions</a>
53 <ul>
54 <li><a href="#x86-specific">X86/X86-64 Language Extensions</a></li>
55 </ul>
56</li>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]57<li><a href="#analyzerspecific">Static Analysis-Specific Extensions</a>
58 <ul>
59 <li><a href="#analyzerattributes">Analyzer Attributes</a></li>
60 </ul>
61</li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]62</ul>
63
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]64<!-- ======================================================================= -->
65<h2 id="intro">Introduction</h2>
66<!-- ======================================================================= -->
67
68<p>This document describes the language extensions provided by Clang. In
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]69addition to the language extensions listed here, Clang aims to support a broad
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]70range of GCC extensions. Please see the <a
71href="http://gcc.gnu.org/onlinedocs/gcc/C-Extensions.html">GCC manual</a> for
72more information on these extensions.</p>
73
74<!-- ======================================================================= -->
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]75<h2 id="feature_check">Feature Checking Macros</h2>
76<!-- ======================================================================= -->
77
78<p>Language extensions can be very useful, but only if you know you can depend
79on them. In order to allow fine-grain features checks, we support two builtin
80function-like macros. This allows you to directly test for a feature in your
81code without having to resort to something like autoconf or fragile "compiler
82version checks".</p>
83
84<!-- ======================================================================= -->
85<h3 id="__has_builtin">__has_builtin</h3>
86<!-- ======================================================================= -->
87
88<p>This function-like macro takes a single identifier argument that is the name
89of a builtin function. It evaluates to 1 if the builtin is supported or 0 if
90not. It can be used like this:</p>
91
92<blockquote>
93<pre>
94#ifndef __has_builtin // Optional of course.
95 #define __has_builtin(x) 0 // Compatibility with non-clang compilers.
96#endif
97
98...
99#if __has_builtin(__builtin_trap)
100 __builtin_trap();
101#else
102 abort();
103#endif
104...
105</pre>
106</blockquote>
107
108
109<!-- ======================================================================= -->
110<h3 id="__has_feature">__has_feature</h3>
111<!-- ======================================================================= -->
112
113<p>This function-like macro takes a single identifier argument that is the name
114of a feature. It evaluates to 1 if the feature is supported or 0 if not. It
115can be used like this:</p>
116
117<blockquote>
118<pre>
119#ifndef __has_feature // Optional of course.
120 #define __has_feature(x) 0 // Compatibility with non-clang compilers.
121#endif
122
123...
124#if __has_feature(attribute_overloadable) || \
125 __has_feature(blocks)
126...
127#endif
128...
129</pre>
130</blockquote>
131
132<p>The feature tag is described along with the language feature below.</p>
133
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]134<!-- ======================================================================= -->
135<h2 id="has_include">Include File Checking Macros</h2>
136<!-- ======================================================================= -->
137
138<p>Not all developments systems have the same include files.
139The <a href="#__has_include">__has_include</a> and
140<a href="#__has_include_next">__has_include_next</a> macros allow you to
141check for the existence of an include file before doing
142a possibly failing #include directive.</p>
143
144<!-- ======================================================================= -->
145<h3 id="__has_include">__has_include</h3>
146<!-- ======================================================================= -->
147
148<p>This function-like macro takes a single file name string argument that
149is the name of an include file. It evaluates to 1 if the file can
150be found using the include paths, or 0 otherwise:</p>
151
152<blockquote>
153<pre>
154// Note the two possible file name string formats.
155#if __has_include("myinclude.h") && __has_include(&lt;stdint.h&gt;)
156# include "myinclude.h"
157#endif
158
159// To avoid problem with non-clang compilers not having this macro.
160#if defined(__has_include) && __has_include("myinclude.h")
161# include "myinclude.h"
162#endif
163</pre>
164</blockquote>
165
166<p>To test for this feature, use #if defined(__has_include).</p>
167
168<!-- ======================================================================= -->
169<h3 id="__has_include_next">__has_include_next</h3>
170<!-- ======================================================================= -->
171
172<p>This function-like macro takes a single file name string argument that
173is the name of an include file. It is like __has_include except that it
174looks for the second instance of the given file found in the include
175paths. It evaluates to 1 if the second instance of the file can
176be found using the include paths, or 0 otherwise:</p>
177
178<blockquote>
179<pre>
180// Note the two possible file name string formats.
181#if __has_include_next("myinclude.h") && __has_include_next(&lt;stdint.h&gt;)
182# include_next "myinclude.h"
183#endif
184
185// To avoid problem with non-clang compilers not having this macro.
186#if defined(__has_include_next) && __has_include_next("myinclude.h")
187# include_next "myinclude.h"
188#endif
189</pre>
190</blockquote>
191
192<p>Note that __has_include_next, like the GNU extension
193#include_next directive, is intended for use in headers only,
194and will issue a warning if used in the top-level compilation
195file. A warning will also be issued if an absolute path
196is used in the file argument.</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]197
198<!-- ======================================================================= -->
Chris Lattner81edc9f2009-04-13 02:45:46 +0000[diff] [blame]199<h2 id="builtinmacros">Builtin Macros</h2>
200<!-- ======================================================================= -->
201
202<p>__BASE_FILE__, __INCLUDE_LEVEL__, __TIMESTAMP__, __COUNTER__</p>
203
204<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]205<h2 id="vectors">Vectors and Extended Vectors</h2>
206<!-- ======================================================================= -->
207
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]208<p>Supports the GCC vector extensions, plus some stuff like V[1].</p>
209
210<p>Also supports <tt>ext_vector</tt>, which additionally support for V.xyzw
211syntax and other tidbits as seen in OpenCL. An example is:</p>
212
213<blockquote>
214<pre>
215typedef float float4 <b>__attribute__((ext_vector_type(4)))</b>;
216typedef float float2 <b>__attribute__((ext_vector_type(2)))</b>;
217
218float4 foo(float2 a, float2 b) {
219 float4 c;
220 c.xz = a;
221 c.yw = b;
222 return c;
223}
224</blockquote>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]225
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]226<p>Query for this feature with __has_feature(attribute_ext_vector_type).</p>
227
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]228<p>See also <a href="#__builtin_shufflevector">__builtin_shufflevector</a>.</p>
229
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]230<!-- ======================================================================= -->
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]231<h2 id="checking_language_features">Checks for Standard Language Features</h2>
232<!-- ======================================================================= -->
233
234<p>The <tt>__has_feature</tt> macro can be used to query if certain standard language features are
235enabled. Those features are listed here.</p>
236
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]237<h3 id="cxx_exceptions">C++ exceptions</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]238
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]239<p>Use <tt>__has_feature(cxx_exceptions)</tt> to determine if C++ exceptions have been enabled. For
240example, compiling code with <tt>-fexceptions</tt> enables C++ exceptions.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]241
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]242<h3 id="cxx_rtti">C++ RTTI</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]243
Ted Kremenek0eb95602009-12-03 02:06:43 +0000[diff] [blame]244<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]245compiling code with <tt>-fno-rtti</tt> disables the use of RTTI.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]246
247<!-- ======================================================================= -->
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]248<h2 id="checking_upcoming_features">Checks for Upcoming Standard Language Features</h2>
249<!-- ======================================================================= -->
250
251<p>The <tt>__has_feature</tt> macro can be used to query if certain upcoming
252standard language features are enabled. Those features are listed here.</p>
253
254<p>Currently, all features listed here are slated for inclusion in the upcoming
255C++0x standard. As a result, all the features that clang supports are enabled
256with the <tt>-std=c++0x</tt> option when compiling C++ code. Features that are
257not yet implemented will be noted.</p>
258
259<h3 id="cxx_decltype">C++0x <tt>decltype()</tt></h3>
260
261<p>Use <tt>__has_feature(cxx_decltype)</tt> to determine if support for the
262<tt>decltype()</tt> specifier is enabled.</p>
263
264<h3 id="cxx_attributes">C++0x attributes</h3>
265
266<p>Use <tt>__has_feature(cxx_attributes)</tt> to determine if support for
267attribute parsing with C++0x's square bracket notation is enabled.
268
269<h3 id="cxx_deleted_functions">C++0x deleted functions</tt></h3>
270
271<p>Use <tt>__has_feature(cxx_deleted_functions)</tt> to determine if support for
272deleted function definitions (with <tt>= delete</tt>) is enabled.
273
Nick Lewycky1444aef2010-04-23 06:09:40 +0000[diff] [blame]274<h3 id="cxx_concepts">C++ TR concepts</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]275
Nick Lewycky1444aef2010-04-23 06:09:40 +0000[diff] [blame]276<p>Use <tt>__has_feature(cxx_concepts)</tt> to determine if support for
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]277concepts is enabled. clang does not currently implement this feature.
278
Douglas Gregor9cc90a32010-01-13 16:27:49 +0000[diff] [blame]279<h3 id="cxx_lambdas">C++0x lambdas</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]280
281<p>Use <tt>__has_feature(cxx_lambdas)</tt> to determine if support for
282lambdas is enabled. clang does not currently implement this feature.
283
284<h3 id="cxx_nullptr">C++0x <tt>nullptr</tt></h3>
285
286<p>Use <tt>__has_feature(cxx_nullptr)</tt> to determine if support for
287<tt>nullptr</tt> is enabled. clang does not yet fully implement this feature.
288
289<h3 id="cxx_rvalue_references">C++0x rvalue references</tt></h3>
290
291<p>Use <tt>__has_feature(cxx_rvalue_references)</tt> to determine if support for
292rvalue references is enabled. clang does not yet fully implement this feature.
293
294<h3 id="cxx_static_assert">C++0x <tt>static_assert()</tt></h3>
295
296<p>Use <tt>__has_feature(cxx_static_assert)</tt> to determine if support for
297compile-time assertions using <tt>static_assert</tt> is enabled.</p>
298
299<h3 id="cxx_auto_type">C++0x type inference</h3>
300
301<p>Use <tt>__has_feature(cxx_auto_type)</tt> to determine C++0x type inference
302is supported using the <tt>auto</tt> specifier. If this is disabled,
303<tt>auto</tt> will instead be a storage class specifier, as in C or C++98.</p>
304
305<h3 id="cxx_variadic_templates">C++0x variadic templates</tt></h3>
306
307<p>Use <tt>__has_feature(cxx_variadic_templates)</tt> to determine if support
308for templates taking any number of arguments with the ellipsis notation is
309enabled. clang does not yet fully implement this feature.</p>
310
311<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]312<h2 id="blocks">Blocks</h2>
313<!-- ======================================================================= -->
314
Chris Lattnera7dbdf52009-03-09 07:03:22 +0000[diff] [blame]315<p>The syntax and high level language feature description is in <a
316href="BlockLanguageSpec.txt">BlockLanguageSpec.txt</a>. Implementation and ABI
317details for the clang implementation are in <a
Chris Lattner5d7650b2010-03-16 21:43:03 +0000[diff] [blame]318href="Block-ABI-Apple.txt">Block-ABI-Apple.txt</a>.</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]319
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]320
321<p>Query for this feature with __has_feature(blocks).</p>
322
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]323<!-- ======================================================================= -->
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]324<h2 id="overloading-in-c">Function Overloading in C</h2>
325<!-- ======================================================================= -->
326
Chris Lattnerf161d412009-02-13 21:51:45 +0000[diff] [blame]327<p>Clang provides support for C++ function overloading in C. Function
328overloading in C is introduced using the <tt>overloadable</tt> attribute. For
329example, one might provide several overloaded versions of a <tt>tgsin</tt>
330function that invokes the appropriate standard function computing the sine of a
331value with <tt>float</tt>, <tt>double</tt>, or <tt>long double</tt>
332precision:</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]333
334<blockquote>
335<pre>
336#include &lt;math.h&gt;
337float <b>__attribute__((overloadable))</b> tgsin(float x) { return sinf(x); }
338double <b>__attribute__((overloadable))</b> tgsin(double x) { return sin(x); }
339long double <b>__attribute__((overloadable))</b> tgsin(long double x) { return sinl(x); }
340</pre>
341</blockquote>
342
343<p>Given these declarations, one can call <tt>tgsin</tt> with a
344<tt>float</tt> value to receive a <tt>float</tt> result, with a
345<tt>double</tt> to receive a <tt>double</tt> result, etc. Function
346overloading in C follows the rules of C++ function overloading to pick
347the best overload given the call arguments, with a few C-specific
348semantics:</p>
349<ul>
350 <li>Conversion from <tt>float</tt> or <tt>double</tt> to <tt>long
351 double</tt> is ranked as a floating-point promotion (per C99) rather
352 than as a floating-point conversion (as in C++).</li>
353
354 <li>A conversion from a pointer of type <tt>T*</tt> to a pointer of type
355 <tt>U*</tt> is considered a pointer conversion (with conversion
356 rank) if <tt>T</tt> and <tt>U</tt> are compatible types.</li>
357
358 <li>A conversion from type <tt>T</tt> to a value of type <tt>U</tt>
359 is permitted if <tt>T</tt> and <tt>U</tt> are compatible types. This
360 conversion is given "conversion" rank.</li>
361</ul>
362
363<p>The declaration of <tt>overloadable</tt> functions is restricted to
364function declarations and definitions. Most importantly, if any
365function with a given name is given the <tt>overloadable</tt>
366attribute, then all function declarations and definitions with that
367name (and in that scope) must have the <tt>overloadable</tt>
Chris Lattnerf161d412009-02-13 21:51:45 +0000[diff] [blame]368attribute. This rule even applies to redeclarations of functions whose original
369declaration had the <tt>overloadable</tt> attribute, e.g.,</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]370
371<blockquote>
372<pre>
373int f(int) __attribute__((overloadable));
374float f(float); <i>// error: declaration of "f" must have the "overloadable" attribute</i>
375
376int g(int) __attribute__((overloadable));
377int g(int) { } <i>// error: redeclaration of "g" must also have the "overloadable" attribute</i>
378</pre>
379</blockquote>
380
Douglas Gregor965acbb2009-02-18 07:07:28 +0000[diff] [blame]381<p>Functions marked <tt>overloadable</tt> must have
382prototypes. Therefore, the following code is ill-formed:</p>
383
384<blockquote>
385<pre>
386int h() __attribute__((overloadable)); <i>// error: h does not have a prototype</i>
387</pre>
388</blockquote>
389
390<p>However, <tt>overloadable</tt> functions are allowed to use a
391ellipsis 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>
392
393<blockquote>
394<pre>
Chris Lattner02246802009-02-18 22:27:46 +0000[diff] [blame]395void honeypot(...) __attribute__((overloadable, unavailable)); <i>// calling me is an error</i>
Douglas Gregor965acbb2009-02-18 07:07:28 +0000[diff] [blame]396</pre>
397</blockquote>
398
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]399<p>Functions declared with the <tt>overloadable</tt> attribute have
400their names mangled according to the same rules as C++ function
401names. For example, the three <tt>tgsin</tt> functions in our
402motivating example get the mangled names <tt>_Z5tgsinf</tt>,
403<tt>_Z5tgsind</tt>, and <tt>Z5tgsine</tt>, respectively. There are two
404caveats to this use of name mangling:</p>
405
406<ul>
407
408 <li>Future versions of Clang may change the name mangling of
409 functions overloaded in C, so you should not depend on an specific
410 mangling. To be completely safe, we strongly urge the use of
411 <tt>static inline</tt> with <tt>overloadable</tt> functions.</li>
412
413 <li>The <tt>overloadable</tt> attribute has almost no meaning when
414 used in C++, because names will already be mangled and functions are
415 already overloadable. However, when an <tt>overloadable</tt>
416 function occurs within an <tt>extern "C"</tt> linkage specification,
417 it's name <i>will</i> be mangled in the same way as it would in
418 C.</li>
419</ul>
420
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]421<p>Query for this feature with __has_feature(attribute_overloadable).</p>
422
423
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]424<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]425<h2 id="builtins">Builtin Functions</h2>
426<!-- ======================================================================= -->
427
428<p>Clang supports a number of builtin library functions with the same syntax as
429GCC, including things like <tt>__builtin_nan</tt>,
430<tt>__builtin_constant_p</tt>, <tt>__builtin_choose_expr</tt>,
431<tt>__builtin_types_compatible_p</tt>, <tt>__sync_fetch_and_add</tt>, etc. In
432addition to the GCC builtins, Clang supports a number of builtins that GCC does
433not, which are listed here.</p>
434
435<p>Please note that Clang does not and will not support all of the GCC builtins
436for vector operations. Instead of using builtins, you should use the functions
437defined in target-specific header files like <tt>&lt;xmmintrin.h&gt;</tt>, which
438define portable wrappers for these. Many of the Clang versions of these
439functions are implemented directly in terms of <a href="#vectors">extended
440vector support</a> instead of builtins, in order to reduce the number of
441builtins that we need to implement.</p>
442
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]443<!-- ======================================================================= -->
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]444<h3 id="__builtin_shufflevector">__builtin_shufflevector</h3>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]445<!-- ======================================================================= -->
446
Chris Lattneraad826b2009-09-16 18:56:12 +0000[diff] [blame]447<p><tt>__builtin_shufflevector</tt> is used to express generic vector
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]448permutation/shuffle/swizzle operations. This builtin is also very important for
449the implementation of various target-specific header files like
450<tt>&lt;xmmintrin.h&gt;</tt>.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]451</p>
452
453<p><b>Syntax:</b></p>
454
455<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]456__builtin_shufflevector(vec1, vec2, index1, index2, ...)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]457</pre>
458
459<p><b>Examples:</b></p>
460
461<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]462 // Identity operation - return 4-element vector V1.
463 __builtin_shufflevector(V1, V1, 0, 1, 2, 3)
464
465 // "Splat" element 0 of V1 into a 4-element result.
466 __builtin_shufflevector(V1, V1, 0, 0, 0, 0)
467
468 // Reverse 4-element vector V1.
469 __builtin_shufflevector(V1, V1, 3, 2, 1, 0)
470
471 // Concatenate every other element of 4-element vectors V1 and V2.
472 __builtin_shufflevector(V1, V2, 0, 2, 4, 6)
473
474 // Concatenate every other element of 8-element vectors V1 and V2.
475 __builtin_shufflevector(V1, V2, 0, 2, 4, 6, 8, 10, 12, 14)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]476</pre>
477
478<p><b>Description:</b></p>
479
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]480<p>The first two arguments to __builtin_shufflevector are vectors that have the
481same element type. The remaining arguments are a list of integers that specify
482the elements indices of the first two vectors that should be extracted and
483returned in a new vector. These element indices are numbered sequentially
484starting with the first vector, continuing into the second vector. Thus, if
485vec1 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]486</p>
487
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]488<p>The result of __builtin_shufflevector is a vector
489with the same element type as vec1/vec2 but that has an element count equal to
490the number of indices specified.
491</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]492
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]493<p>Query for this feature with __has_builtin(__builtin_shufflevector).</p>
494
495<!-- ======================================================================= -->
496<h3 id="__builtin_unreachable">__builtin_unreachable</h3>
497<!-- ======================================================================= -->
498
499<p><tt>__builtin_unreachable</tt> is used to indicate that a specific point in
500the program cannot be reached, even if the compiler might otherwise think it
501can. This is useful to improve optimization and eliminates certain warnings.
502For example, without the <tt>__builtin_unreachable</tt> in the example below,
503the compiler assumes that the inline asm can fall through and prints a "function
504declared 'noreturn' should not return" warning.
505</p>
506
507<p><b>Syntax:</b></p>
508
509<pre>
510__builtin_unreachable()
511</pre>
512
513<p><b>Example of Use:</b></p>
514
515<pre>
516void myabort(void) __attribute__((noreturn));
517void myabort(void) {
518 asm("int3");
519 __builtin_unreachable();
520}
521</pre>
522
523<p><b>Description:</b></p>
524
525<p>The __builtin_unreachable() builtin has completely undefined behavior. Since
526it has undefined behavior, it is a statement that it is never reached and the
527optimizer can take advantage of this to produce better code. This builtin takes
528no arguments and produces a void result.
529</p>
530
531<p>Query for this feature with __has_builtin(__builtin_unreachable).</p>
532
533
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]534<!-- ======================================================================= -->
535<h2 id="targetspecific">Target-Specific Extensions</h2>
536<!-- ======================================================================= -->
537
538<p>Clang supports some language features conditionally on some targets.</p>
539
540<!-- ======================================================================= -->
541<h3 id="x86-specific">X86/X86-64 Language Extensions</h3>
542<!-- ======================================================================= -->
543
544<p>The X86 backend has these language extensions:</p>
545
546<!-- ======================================================================= -->
547<h4 id="x86-gs-segment">Memory references off the GS segment</h4>
548<!-- ======================================================================= -->
549
550<p>Annotating a pointer with address space #256 causes it to be code generated
Chris Lattnera021e7c2009-05-05 18:54:47 +0000[diff] [blame]551relative to the X86 GS segment register, and address space #257 causes it to be
552relative to the X86 FS segment. Note that this is a very very low-level
553feature that should only be used if you know what you're doing (for example in
554an OS kernel).</p>
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]555
556<p>Here is an example:</p>
557
558<pre>
559#define GS_RELATIVE __attribute__((address_space(256)))
560int foo(int GS_RELATIVE *P) {
561 return *P;
562}
563</pre>
564
565<p>Which compiles to (on X86-32):</p>
566
567<pre>
568_foo:
569 movl 4(%esp), %eax
570 movl %gs:(%eax), %eax
571 ret
572</pre>
573
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]574<!-- ======================================================================= -->
575<h2 id="analyzerspecific">Static Analysis-Specific Extensions</h2>
576<!-- ======================================================================= -->
577
578<p>Clang supports additional attributes that are useful for documenting program
579invariants and rules for static analysis tools. The extensions documented here
580are used by the <a
581href="http://clang.llvm.org/StaticAnalysis.html">path-sensitive static analyzer
582engine</a> that is part of Clang's Analysis library.</p>
583
584<!-- ======================================================================= -->
585<h3 id="analyzerattributes">Analyzer Attributes</h3>
586<!-- ======================================================================= -->
587
588<h4 id="attr_analyzer_noreturn"><tt>analyzer_noreturn</tt></h4>
589
590<p>Clang's static analysis engine understands the standard <tt>noreturn</tt>
Ted Kremenek4df21142009-04-10 05:04:22 +0000[diff] [blame]591attribute. This attribute, which is typically affixed to a function prototype,
592indicates that a call to a given function never returns. Function prototypes for
593common functions like <tt>exit</tt> are typically annotated with this attribute,
594as well as a variety of common assertion handlers. Users can educate the static
595analyzer about their own custom assertion handles (thus cutting down on false
596positives due to false paths) by marking their own &quot;panic&quot; functions
597with this attribute.</p>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]598
599<p>While useful, <tt>noreturn</tt> is not applicable in all cases. Sometimes
Nick Lewycky625b5862009-06-14 04:08:08 +0000[diff] [blame]600there are special functions that for all intents and purposes should be
601considered panic functions (i.e., they are only called when an internal program
602error occurs) but may actually return so that the program can fail gracefully.
603The <tt>analyzer_noreturn</tt> attribute allows one to annotate such functions
604as being interpreted as &quot;no return&quot; functions by the analyzer (thus
Chris Lattner28935892009-04-10 05:54:56 +0000[diff] [blame]605pruning bogus paths) but will not affect compilation (as in the case of
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]606<tt>noreturn</tt>).</p>
607
608<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]609same places where the <tt>noreturn</tt> attribute can be placed. It is commonly
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]610placed at the end of function prototypes:</p>
611
612<pre>
613 void foo() <b>__attribute__((analyzer_noreturn))</b>;
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]614</pre>
615
616<p>Query for this feature with __has_feature(attribute_analyzer_noreturn).</p>
617
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]618
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]619</div>
620</body>
621</html>