aarch64-linux-gnu-4.9/share/doc/gcc/Function-Attributes.html - toolchains/linux-x86/gcc - Git at Google

 <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
 <html>
 <!-- Copyright (C) 1988-2015 Free Software Foundation, Inc.

 Permission is granted to copy, distribute and/or modify this document
 under the terms of the GNU Free Documentation License, Version 1.3 or
 any later version published by the Free Software Foundation; with the
 Invariant Sections being "Funding Free Software", the Front-Cover
 Texts being (a) (see below), and with the Back-Cover Texts being (b)
 (see below).  A copy of the license is included in the section entitled
 "GNU Free Documentation License".

 (a) The FSF's Front-Cover Text is:

 A GNU Manual

 (b) The FSF's Back-Cover Text is:

 You have freedom to copy and modify this GNU Manual, like GNU
      software.  Copies published by the Free Software Foundation raise
      funds for GNU development. -->
 <!-- Created by GNU Texinfo 5.2, http://www.gnu.org/software/texinfo/ -->
 <head>
 <title>Using the GNU Compiler Collection (GCC): Function Attributes</title>

 <meta name="description" content="Using the GNU Compiler Collection (GCC): Function Attributes">
 <meta name="keywords" content="Using the GNU Compiler Collection (GCC): Function Attributes">
 <meta name="resource-type" content="document">
 <meta name="distribution" content="global">
 <meta name="Generator" content="makeinfo">
 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
 <link href="index.html#Top" rel="start" title="Top">
 <link href="Option-Index.html#Option-Index" rel="index" title="Option Index">
 <link href="index.html#SEC_Contents" rel="contents" title="Table of Contents">
 <link href="C-Extensions.html#C-Extensions" rel="up" title="C Extensions">
 <link href="Attribute-Syntax.html#Attribute-Syntax" rel="next" title="Attribute Syntax">
 <link href="Mixed-Declarations.html#Mixed-Declarations" rel="prev" title="Mixed Declarations">
 <style type="text/css">
 <!--
 a.summary-letter {text-decoration: none}
 blockquote.smallquotation {font-size: smaller}
 div.display {margin-left: 3.2em}
 div.example {margin-left: 3.2em}
 div.indentedblock {margin-left: 3.2em}
 div.lisp {margin-left: 3.2em}
 div.smalldisplay {margin-left: 3.2em}
 div.smallexample {margin-left: 3.2em}
 div.smallindentedblock {margin-left: 3.2em; font-size: smaller}
 div.smalllisp {margin-left: 3.2em}
 kbd {font-style:oblique}
 pre.display {font-family: inherit}
 pre.format {font-family: inherit}
 pre.menu-comment {font-family: serif}
 pre.menu-preformatted {font-family: serif}
 pre.smalldisplay {font-family: inherit; font-size: smaller}
 pre.smallexample {font-size: smaller}
 pre.smallformat {font-family: inherit; font-size: smaller}
 pre.smalllisp {font-size: smaller}
 span.nocodebreak {white-space:nowrap}
 span.nolinebreak {white-space:nowrap}
 span.roman {font-family:serif; font-weight:normal}
 span.sansserif {font-family:sans-serif; font-weight:normal}
 ul.no-bullet {list-style: none}
 -->
 </style>


 </head>

 <body lang="en" bgcolor="#FFFFFF" text="#000000" link="#0000FF" vlink="#800080" alink="#FF0000">
 <a name="Function-Attributes"></a>
 <div class="header">
 <p>
 Next: <a href="Attribute-Syntax.html#Attribute-Syntax" accesskey="n" rel="next">Attribute Syntax</a>, Previous: <a href="Mixed-Declarations.html#Mixed-Declarations" accesskey="p" rel="prev">Mixed Declarations</a>, Up: <a href="C-Extensions.html#C-Extensions" accesskey="u" rel="up">C Extensions</a> &nbsp; [<a href="index.html#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="Option-Index.html#Option-Index" title="Index" rel="index">Index</a>]</p>
 </div>
 <hr>
 <a name="Declaring-Attributes-of-Functions"></a>
 <h3 class="section">6.30 Declaring Attributes of Functions</h3>
 <a name="index-function-attributes"></a>
 <a name="index-declaring-attributes-of-functions"></a>
 <a name="index-functions-that-never-return"></a>
 <a name="index-functions-that-return-more-than-once"></a>
 <a name="index-functions-that-have-no-side-effects"></a>
 <a name="index-functions-in-arbitrary-sections"></a>
 <a name="index-functions-that-behave-like-malloc"></a>
 <a name="index-volatile-applied-to-function"></a>
 <a name="index-const-applied-to-function"></a>
 <a name="index-functions-with-printf_002c-scanf_002c-strftime-or-strfmon-style-arguments"></a>
 <a name="index-functions-with-non_002dnull-pointer-arguments"></a>
 <a name="index-functions-that-are-passed-arguments-in-registers-on-the-386"></a>
 <a name="index-functions-that-pop-the-argument-stack-on-the-386"></a>
 <a name="index-functions-that-do-not-pop-the-argument-stack-on-the-386"></a>
 <a name="index-functions-that-have-different-compilation-options-on-the-386"></a>
 <a name="index-functions-that-have-different-optimization-options"></a>
 <a name="index-functions-that-are-dynamically-resolved"></a>

 <p>In GNU C, you declare certain things about functions called in your program
 which help the compiler optimize function calls and check your code more
 carefully.
 </p>
 <p>The keyword <code>__attribute__</code> allows you to specify special
 attributes when making a declaration.  This keyword is followed by an
 attribute specification inside double parentheses.  The following
 attributes are currently defined for functions on all targets:
 <code>aligned</code>, <code>alloc_size</code>, <code>alloc_align</code>, <code>assume_aligned</code>,
 <code>noreturn</code>, <code>returns_twice</code>, <code>noinline</code>, <code>noclone</code>,
 <code>always_inline</code>, <code>flatten</code>, <code>pure</code>, <code>const</code>,
 <code>nothrow</code>, <code>sentinel</code>, <code>format</code>, <code>format_arg</code>,
 <code>no_instrument_function</code>, <code>no_split_stack</code>,
 <code>section</code>, <code>constructor</code>,
 <code>destructor</code>, <code>used</code>, <code>unused</code>, <code>deprecated</code>,
 <code>weak</code>, <code>malloc</code>, <code>alias</code>, <code>ifunc</code>,
 <code>warn_unused_result</code>, <code>nonnull</code>,
 <code>returns_nonnull</code>, <code>gnu_inline</code>,
 <code>externally_visible</code>, <code>hot</code>, <code>cold</code>, <code>artificial</code>,
 <code>no_sanitize_address</code>, <code>no_address_safety_analysis</code>,
 <code>no_sanitize_undefined</code>,
 <code>error</code> and <code>warning</code>.
 Several other attributes are defined for functions on particular
 target systems.  Other attributes, including <code>section</code> are
 supported for variables declarations (see <a href="Variable-Attributes.html#Variable-Attributes">Variable Attributes</a>)
 and for types (see <a href="Type-Attributes.html#Type-Attributes">Type Attributes</a>).
 </p>
 <p>GCC plugins may provide their own attributes.
 </p>
 <p>You may also specify attributes with &lsquo;<samp>__</samp>&rsquo; preceding and following
 each keyword.  This allows you to use them in header files without
 being concerned about a possible macro of the same name.  For example,
 you may use <code>__noreturn__</code> instead of <code>noreturn</code>.
 </p>
 <p>See <a href="Attribute-Syntax.html#Attribute-Syntax">Attribute Syntax</a>, for details of the exact syntax for using
 attributes.
 </p>
 <dl compact="compact">
 <dt><code>alias (&quot;<var>target</var>&quot;)</code></dt>
 <dd><a name="index-alias-attribute"></a>
 <p>The <code>alias</code> attribute causes the declaration to be emitted as an
 alias for another symbol, which must be specified.  For instance,
 </p>
 <div class="smallexample">
 <pre class="smallexample">void __f () { /* <span class="roman">Do something.</span> */; }
 void f () __attribute__ ((weak, alias (&quot;__f&quot;)));
 </pre></div>

 <p>defines &lsquo;<samp>f</samp>&rsquo; to be a weak alias for &lsquo;<samp>__f</samp>&rsquo;.  In C++, the
 mangled name for the target must be used.  It is an error if &lsquo;<samp>__f</samp>&rsquo;
 is not defined in the same translation unit.
 </p>
 <p>Not all target machines support this attribute.
 </p>
 </dd>
 <dt><code>aligned (<var>alignment</var>)</code></dt>
 <dd><a name="index-aligned-attribute"></a>
 <p>This attribute specifies a minimum alignment for the function,
 measured in bytes.
 </p>
 <p>You cannot use this attribute to decrease the alignment of a function,
 only to increase it.  However, when you explicitly specify a function
 alignment this overrides the effect of the
 <samp>-falign-functions</samp> (see <a href="Optimize-Options.html#Optimize-Options">Optimize Options</a>) option for this
 function.
 </p>
 <p>Note that the effectiveness of <code>aligned</code> attributes may be
 limited by inherent limitations in your linker.  On many systems, the
 linker is only able to arrange for functions to be aligned up to a
 certain maximum alignment.  (For some linkers, the maximum supported
 alignment may be very very small.)  See your linker documentation for
 further information.
 </p>
 <p>The <code>aligned</code> attribute can also be used for variables and fields
 (see <a href="Variable-Attributes.html#Variable-Attributes">Variable Attributes</a>.)
 </p>
 </dd>
 <dt><code>alloc_size</code></dt>
 <dd><a name="index-alloc_005fsize-attribute"></a>
 <p>The <code>alloc_size</code> attribute is used to tell the compiler that the
 function return value points to memory, where the size is given by
 one or two of the functions parameters.  GCC uses this
 information to improve the correctness of <code>__builtin_object_size</code>.
 </p>
 <p>The function parameter(s) denoting the allocated size are specified by
 one or two integer arguments supplied to the attribute.  The allocated size
 is either the value of the single function argument specified or the product
 of the two function arguments specified.  Argument numbering starts at
 one.
 </p>
 <p>For instance,
 </p>
 <div class="smallexample">
 <pre class="smallexample">void* my_calloc(size_t, size_t) __attribute__((alloc_size(1,2)))
 void* my_realloc(void*, size_t) __attribute__((alloc_size(2)))
 </pre></div>

 <p>declares that <code>my_calloc</code> returns memory of the size given by
 the product of parameter 1 and 2 and that <code>my_realloc</code> returns memory
 of the size given by parameter 2.
 </p>
 </dd>
 <dt><code>alloc_align</code></dt>
 <dd><a name="index-alloc_005falign-attribute"></a>
 <p>The <code>alloc_align</code> attribute is used to tell the compiler that the
 function return value points to memory, where the returned pointer minimum
 alignment is given by one of the functions parameters.  GCC uses this
 information to improve pointer alignment analysis.
 </p>
 <p>The function parameter denoting the allocated alignment is specified by
 one integer argument, whose number is the argument of the attribute.
 Argument numbering starts at one.
 </p>
 <p>For instance,
 </p>
 <div class="smallexample">
 <pre class="smallexample">void* my_memalign(size_t, size_t) __attribute__((alloc_align(1)))
 </pre></div>

 <p>declares that <code>my_memalign</code> returns memory with minimum alignment
 given by parameter 1.
 </p>
 </dd>
 <dt><code>assume_aligned</code></dt>
 <dd><a name="index-assume_005faligned-attribute"></a>
 <p>The <code>assume_aligned</code> attribute is used to tell the compiler that the
 function return value points to memory, where the returned pointer minimum
 alignment is given by the first argument.
 If the attribute has two arguments, the second argument is misalignment offset.
 </p>
 <p>For instance
 </p>
 <div class="smallexample">
 <pre class="smallexample">void* my_alloc1(size_t) __attribute__((assume_aligned(16)))
 void* my_alloc2(size_t) __attribute__((assume_aligned(32, 8)))
 </pre></div>

 <p>declares that <code>my_alloc1</code> returns 16-byte aligned pointer and
 that <code>my_alloc2</code> returns a pointer whose value modulo 32 is equal
 to 8.
 </p>
 </dd>
 <dt><code>always_inline</code></dt>
 <dd><a name="index-always_005finline-function-attribute"></a>
 <p>Generally, functions are not inlined unless optimization is specified.
 For functions declared inline, this attribute inlines the function even
 if no optimization level is specified.
 </p>
 </dd>
 <dt><code>gnu_inline</code></dt>
 <dd><a name="index-gnu_005finline-function-attribute"></a>
 <p>This attribute should be used with a function that is also declared
 with the <code>inline</code> keyword.  It directs GCC to treat the function
 as if it were defined in gnu90 mode even when compiling in C99 or
 gnu99 mode.
 </p>
 <p>If the function is declared <code>extern</code>, then this definition of the
 function is used only for inlining.  In no case is the function
 compiled as a standalone function, not even if you take its address
 explicitly.  Such an address becomes an external reference, as if you
 had only declared the function, and had not defined it.  This has
 almost the effect of a macro.  The way to use this is to put a
 function definition in a header file with this attribute, and put
 another copy of the function, without <code>extern</code>, in a library
 file.  The definition in the header file causes most calls to the
 function to be inlined.  If any uses of the function remain, they
 refer to the single copy in the library.  Note that the two
 definitions of the functions need not be precisely the same, although
 if they do not have the same effect your program may behave oddly.
 </p>
 <p>In C, if the function is neither <code>extern</code> nor <code>static</code>, then
 the function is compiled as a standalone function, as well as being
 inlined where possible.
 </p>
 <p>This is how GCC traditionally handled functions declared
 <code>inline</code>.  Since ISO C99 specifies a different semantics for
 <code>inline</code>, this function attribute is provided as a transition
 measure and as a useful feature in its own right.  This attribute is
 available in GCC 4.1.3 and later.  It is available if either of the
 preprocessor macros <code>__GNUC_GNU_INLINE__</code> or
 <code>__GNUC_STDC_INLINE__</code> are defined.  See <a href="Inline.html#Inline">An Inline
 Function is As Fast As a Macro</a>.
 </p>
 <p>In C++, this attribute does not depend on <code>extern</code> in any way,
 but it still requires the <code>inline</code> keyword to enable its special
 behavior.
 </p>
 </dd>
 <dt><code>artificial</code></dt>
 <dd><a name="index-artificial-function-attribute"></a>
 <p>This attribute is useful for small inline wrappers that if possible
 should appear during debugging as a unit.  Depending on the debug
 info format it either means marking the function as artificial
 or using the caller location for all instructions within the inlined
 body.
 </p>
 </dd>
 <dt><code>bank_switch</code></dt>
 <dd><a name="index-interrupt-handler-functions"></a>
 <p>When added to an interrupt handler with the M32C port, causes the
 prologue and epilogue to use bank switching to preserve the registers
 rather than saving them on the stack.
 </p>
 </dd>
 <dt><code>flatten</code></dt>
 <dd><a name="index-flatten-function-attribute"></a>
 <p>Generally, inlining into a function is limited.  For a function marked with
 this attribute, every call inside this function is inlined, if possible.
 Whether the function itself is considered for inlining depends on its size and
 the current inlining parameters.
 </p>
 </dd>
 <dt><code>error (&quot;<var>message</var>&quot;)</code></dt>
 <dd><a name="index-error-function-attribute"></a>
 <p>If this attribute is used on a function declaration and a call to such a function
 is not eliminated through dead code elimination or other optimizations, an error
 that includes <var>message</var> is diagnosed.  This is useful
 for compile-time checking, especially together with <code>__builtin_constant_p</code>
 and inline functions where checking the inline function arguments is not
 possible through <code>extern char [(condition) ? 1 : -1];</code> tricks.
 While it is possible to leave the function undefined and thus invoke
 a link failure, when using this attribute the problem is diagnosed
 earlier and with exact location of the call even in presence of inline
 functions or when not emitting debugging information.
 </p>
 </dd>
 <dt><code>warning (&quot;<var>message</var>&quot;)</code></dt>
 <dd><a name="index-warning-function-attribute"></a>
 <p>If this attribute is used on a function declaration and a call to such a function
 is not eliminated through dead code elimination or other optimizations, a warning
 that includes <var>message</var> is diagnosed.  This is useful
 for compile-time checking, especially together with <code>__builtin_constant_p</code>
 and inline functions.  While it is possible to define the function with
 a message in <code>.gnu.warning*</code> section, when using this attribute the problem
 is diagnosed earlier and with exact location of the call even in presence
 of inline functions or when not emitting debugging information.
 </p>
 </dd>
 <dt><code>cdecl</code></dt>
 <dd><a name="index-functions-that-do-pop-the-argument-stack-on-the-386"></a>
 <a name="index-mrtd-2"></a>
 <p>On the Intel 386, the <code>cdecl</code> attribute causes the compiler to
 assume that the calling function pops off the stack space used to
 pass arguments.  This is
 useful to override the effects of the <samp>-mrtd</samp> switch.
 </p>
 </dd>
 <dt><code>const</code></dt>
 <dd><a name="index-const-function-attribute"></a>
 <p>Many functions do not examine any values except their arguments, and
 have no effects except the return value.  Basically this is just slightly
 more strict class than the <code>pure</code> attribute below, since function is not
 allowed to read global memory.
 </p>
 <a name="index-pointer-arguments"></a>
 <p>Note that a function that has pointer arguments and examines the data
 pointed to must <em>not</em> be declared <code>const</code>.  Likewise, a
 function that calls a non-<code>const</code> function usually must not be
 <code>const</code>.  It does not make sense for a <code>const</code> function to
 return <code>void</code>.
 </p>
 <p>The attribute <code>const</code> is not implemented in GCC versions earlier
 than 2.5.  An alternative way to declare that a function has no side
 effects, which works in the current version and in some older versions,
 is as follows:
 </p>
 <div class="smallexample">
 <pre class="smallexample">typedef int intfn ();

 extern const intfn square;
 </pre></div>

 <p>This approach does not work in GNU C++ from 2.6.0 on, since the language
 specifies that the &lsquo;<samp>const</samp>&rsquo; must be attached to the return value.
 </p>
 </dd>
 <dt><code>constructor</code></dt>
 <dt><code>destructor</code></dt>
 <dt><code>constructor (<var>priority</var>)</code></dt>
 <dt><code>destructor (<var>priority</var>)</code></dt>
 <dd><a name="index-constructor-function-attribute"></a>
 <a name="index-destructor-function-attribute"></a>
 <p>The <code>constructor</code> attribute causes the function to be called
 automatically before execution enters <code>main ()</code>.  Similarly, the
 <code>destructor</code> attribute causes the function to be called
 automatically after <code>main ()</code> completes or <code>exit ()</code> is
 called.  Functions with these attributes are useful for
 initializing data that is used implicitly during the execution of
 the program.
 </p>
 <p>You may provide an optional integer priority to control the order in
 which constructor and destructor functions are run.  A constructor
 with a smaller priority number runs before a constructor with a larger
 priority number; the opposite relationship holds for destructors.  So,
 if you have a constructor that allocates a resource and a destructor
 that deallocates the same resource, both functions typically have the
 same priority.  The priorities for constructor and destructor
 functions are the same as those specified for namespace-scope C++
 objects (see <a href="C_002b_002b-Attributes.html#C_002b_002b-Attributes">C++ Attributes</a>).
 </p>
 <p>These attributes are not currently implemented for Objective-C.
 </p>
 </dd>
 <dt><code>deprecated</code></dt>
 <dt><code>deprecated (<var>msg</var>)</code></dt>
 <dd><a name="index-deprecated-attribute_002e"></a>
 <p>The <code>deprecated</code> attribute results in a warning if the function
 is used anywhere in the source file.  This is useful when identifying
 functions that are expected to be removed in a future version of a
 program.  The warning also includes the location of the declaration
 of the deprecated function, to enable users to easily find further
 information about why the function is deprecated, or what they should
 do instead.  Note that the warnings only occurs for uses:
 </p>
 <div class="smallexample">
 <pre class="smallexample">int old_fn () __attribute__ ((deprecated));
 int old_fn ();
 int (*fn_ptr)() = old_fn;
 </pre></div>

 <p>results in a warning on line 3 but not line 2.  The optional <var>msg</var>
 argument, which must be a string, is printed in the warning if
 present.
 </p>
 <p>The <code>deprecated</code> attribute can also be used for variables and
 types (see <a href="Variable-Attributes.html#Variable-Attributes">Variable Attributes</a>, see <a href="Type-Attributes.html#Type-Attributes">Type Attributes</a>.)
 </p>
 </dd>
 <dt><code>disinterrupt</code></dt>
 <dd><a name="index-disinterrupt-attribute"></a>
 <p>On Epiphany and MeP targets, this attribute causes the compiler to emit
 instructions to disable interrupts for the duration of the given
 function.
 </p>
 </dd>
 <dt><code>dllexport</code></dt>
 <dd><a name="index-_005f_005fdeclspec_0028dllexport_0029"></a>
 <p>On Microsoft Windows targets and Symbian OS targets the
 <code>dllexport</code> attribute causes the compiler to provide a global
 pointer to a pointer in a DLL, so that it can be referenced with the
 <code>dllimport</code> attribute.  On Microsoft Windows targets, the pointer
 name is formed by combining <code>_imp__</code> and the function or variable
 name.
 </p>
 <p>You can use <code>__declspec(dllexport)</code> as a synonym for
 <code>__attribute__ ((dllexport))</code> for compatibility with other
 compilers.
 </p>
 <p>On systems that support the <code>visibility</code> attribute, this
 attribute also implies &ldquo;default&rdquo; visibility.  It is an error to
 explicitly specify any other visibility.
 </p>
 <p>In previous versions of GCC, the <code>dllexport</code> attribute was ignored
 for inlined functions, unless the <samp>-fkeep-inline-functions</samp> flag
 had been used.  The default behavior now is to emit all dllexported
 inline functions; however, this can cause object file-size bloat, in
 which case the old behavior can be restored by using
 <samp>-fno-keep-inline-dllexport</samp>.
 </p>
 <p>The attribute is also ignored for undefined symbols.
 </p>
 <p>When applied to C++ classes, the attribute marks defined non-inlined
 member functions and static data members as exports.  Static consts
 initialized in-class are not marked unless they are also defined
 out-of-class.
 </p>
 <p>For Microsoft Windows targets there are alternative methods for
 including the symbol in the DLL&rsquo;s export table such as using a
 <samp>.def</samp> file with an <code>EXPORTS</code> section or, with GNU ld, using
 the <samp>--export-all</samp> linker flag.
 </p>
 </dd>
 <dt><code>dllimport</code></dt>
 <dd><a name="index-_005f_005fdeclspec_0028dllimport_0029"></a>
 <p>On Microsoft Windows and Symbian OS targets, the <code>dllimport</code>
 attribute causes the compiler to reference a function or variable via
 a global pointer to a pointer that is set up by the DLL exporting the
 symbol.  The attribute implies <code>extern</code>.  On Microsoft Windows
 targets, the pointer name is formed by combining <code>_imp__</code> and the
 function or variable name.
 </p>
 <p>You can use <code>__declspec(dllimport)</code> as a synonym for
 <code>__attribute__ ((dllimport))</code> for compatibility with other
 compilers.
 </p>
 <p>On systems that support the <code>visibility</code> attribute, this
 attribute also implies &ldquo;default&rdquo; visibility.  It is an error to
 explicitly specify any other visibility.
 </p>
 <p>Currently, the attribute is ignored for inlined functions.  If the
 attribute is applied to a symbol <em>definition</em>, an error is reported.
 If a symbol previously declared <code>dllimport</code> is later defined, the
 attribute is ignored in subsequent references, and a warning is emitted.
 The attribute is also overridden by a subsequent declaration as
 <code>dllexport</code>.
 </p>
 <p>When applied to C++ classes, the attribute marks non-inlined
 member functions and static data members as imports.  However, the
 attribute is ignored for virtual methods to allow creation of vtables
 using thunks.
 </p>
 <p>On the SH Symbian OS target the <code>dllimport</code> attribute also has
 another affect&mdash;it can cause the vtable and run-time type information
 for a class to be exported.  This happens when the class has a
 dllimported constructor or a non-inline, non-pure virtual function
 and, for either of those two conditions, the class also has an inline
 constructor or destructor and has a key function that is defined in
 the current translation unit.
 </p>
 <p>For Microsoft Windows targets the use of the <code>dllimport</code>
 attribute on functions is not necessary, but provides a small
 performance benefit by eliminating a thunk in the DLL.  The use of the
 <code>dllimport</code> attribute on imported variables was required on older
 versions of the GNU linker, but can now be avoided by passing the
 <samp>--enable-auto-import</samp> switch to the GNU linker.  As with
 functions, using the attribute for a variable eliminates a thunk in
 the DLL.
 </p>
 <p>One drawback to using this attribute is that a pointer to a
 <em>variable</em> marked as <code>dllimport</code> cannot be used as a constant
 address. However, a pointer to a <em>function</em> with the
 <code>dllimport</code> attribute can be used as a constant initializer; in
 this case, the address of a stub function in the import lib is
 referenced.  On Microsoft Windows targets, the attribute can be disabled
 for functions by setting the <samp>-mnop-fun-dllimport</samp> flag.
 </p>
 </dd>
 <dt><code>eightbit_data</code></dt>
 <dd><a name="index-eight_002dbit-data-on-the-H8_002f300_002c-H8_002f300H_002c-and-H8S"></a>
 <p>Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
 variable should be placed into the eight-bit data section.
 The compiler generates more efficient code for certain operations
 on data in the eight-bit data area.  Note the eight-bit data area is limited to
 256 bytes of data.
 </p>
 <p>You must use GAS and GLD from GNU binutils version 2.7 or later for
 this attribute to work correctly.
 </p>
 </dd>
 <dt><code>exception</code></dt>
 <dd><a name="index-exception-handler-functions"></a>
 <p>Use this attribute on the NDS32 target to indicate that the specified function
 is an exception handler.  The compiler will generate corresponding sections
 for use in an exception handler.
 </p>
 </dd>
 <dt><code>exception_handler</code></dt>
 <dd><a name="index-exception-handler-functions-on-the-Blackfin-processor"></a>
 <p>Use this attribute on the Blackfin to indicate that the specified function
 is an exception handler.  The compiler generates function entry and
 exit sequences suitable for use in an exception handler when this
 attribute is present.
 </p>
 </dd>
 <dt><code>externally_visible</code></dt>
 <dd><a name="index-externally_005fvisible-attribute_002e"></a>
 <p>This attribute, attached to a global variable or function, nullifies
 the effect of the <samp>-fwhole-program</samp> command-line option, so the
 object remains visible outside the current compilation unit.
 </p>
 <p>If <samp>-fwhole-program</samp> is used together with <samp>-flto</samp> and
 <code>gold</code> is used as the linker plugin,
 <code>externally_visible</code> attributes are automatically added to functions
 (not variable yet due to a current <code>gold</code> issue)
 that are accessed outside of LTO objects according to resolution file
 produced by <code>gold</code>.
 For other linkers that cannot generate resolution file,
 explicit <code>externally_visible</code> attributes are still necessary.
 </p>
 </dd>
 <dt><code>far</code></dt>
 <dd><a name="index-functions-that-handle-memory-bank-switching"></a>
 <p>On 68HC11 and 68HC12 the <code>far</code> attribute causes the compiler to
 use a calling convention that takes care of switching memory banks when
 entering and leaving a function.  This calling convention is also the
 default when using the <samp>-mlong-calls</samp> option.
 </p>
 <p>On 68HC12 the compiler uses the <code>call</code> and <code>rtc</code> instructions
 to call and return from a function.
 </p>
 <p>On 68HC11 the compiler generates a sequence of instructions
 to invoke a board-specific routine to switch the memory bank and call the
 real function.  The board-specific routine simulates a <code>call</code>.
 At the end of a function, it jumps to a board-specific routine
 instead of using <code>rts</code>.  The board-specific return routine simulates
 the <code>rtc</code>.
 </p>
 <p>On MeP targets this causes the compiler to use a calling convention
 that assumes the called function is too far away for the built-in
 addressing modes.
 </p>
 </dd>
 <dt><code>fast_interrupt</code></dt>
 <dd><a name="index-interrupt-handler-functions-1"></a>
 <p>Use this attribute on the M32C and RX ports to indicate that the specified
 function is a fast interrupt handler.  This is just like the
 <code>interrupt</code> attribute, except that <code>freit</code> is used to return
 instead of <code>reit</code>.
 </p>
 </dd>
 <dt><code>fastcall</code></dt>
 <dd><a name="index-functions-that-pop-the-argument-stack-on-the-386-1"></a>
 <p>On the Intel 386, the <code>fastcall</code> attribute causes the compiler to
 pass the first argument (if of integral type) in the register ECX and
 the second argument (if of integral type) in the register EDX.  Subsequent
 and other typed arguments are passed on the stack.  The called function
 pops the arguments off the stack.  If the number of arguments is variable all
 arguments are pushed on the stack.
 </p>
 </dd>
 <dt><code>thiscall</code></dt>
 <dd><a name="index-functions-that-pop-the-argument-stack-on-the-386-2"></a>
 <p>On the Intel 386, the <code>thiscall</code> attribute causes the compiler to
 pass the first argument (if of integral type) in the register ECX.
 Subsequent and other typed arguments are passed on the stack. The called
 function pops the arguments off the stack.
 If the number of arguments is variable all arguments are pushed on the
 stack.
 The <code>thiscall</code> attribute is intended for C++ non-static member functions.
 As a GCC extension, this calling convention can be used for C functions
 and for static member methods.
 </p>
 </dd>
 <dt><code>format (<var>archetype</var>, <var>string-index</var>, <var>first-to-check</var>)</code></dt>
 <dd><a name="index-format-function-attribute"></a>
 <a name="index-Wformat-3"></a>
 <p>The <code>format</code> attribute specifies that a function takes <code>printf</code>,
 <code>scanf</code>, <code>strftime</code> or <code>strfmon</code> style arguments that
 should be type-checked against a format string.  For example, the
 declaration:
 </p>
 <div class="smallexample">
 <pre class="smallexample">extern int
 my_printf (void *my_object, const char *my_format, ...)
       __attribute__ ((format (printf, 2, 3)));
 </pre></div>

 <p>causes the compiler to check the arguments in calls to <code>my_printf</code>
 for consistency with the <code>printf</code> style format string argument
 <code>my_format</code>.
 </p>
 <p>The parameter <var>archetype</var> determines how the format string is
 interpreted, and should be <code>printf</code>, <code>scanf</code>, <code>strftime</code>,
 <code>gnu_printf</code>, <code>gnu_scanf</code>, <code>gnu_strftime</code> or
 <code>strfmon</code>.  (You can also use <code>__printf__</code>,
 <code>__scanf__</code>, <code>__strftime__</code> or <code>__strfmon__</code>.)  On
 MinGW targets, <code>ms_printf</code>, <code>ms_scanf</code>, and
 <code>ms_strftime</code> are also present.
 <var>archetype</var> values such as <code>printf</code> refer to the formats accepted
 by the system&rsquo;s C runtime library,
 while values prefixed with &lsquo;<samp>gnu_</samp>&rsquo; always refer
 to the formats accepted by the GNU C Library.  On Microsoft Windows
 targets, values prefixed with &lsquo;<samp>ms_</samp>&rsquo; refer to the formats accepted by the
 <samp>msvcrt.dll</samp> library.
 The parameter <var>string-index</var>
 specifies which argument is the format string argument (starting
 from 1), while <var>first-to-check</var> is the number of the first
 argument to check against the format string.  For functions
 where the arguments are not available to be checked (such as
 <code>vprintf</code>), specify the third parameter as zero.  In this case the
 compiler only checks the format string for consistency.  For
 <code>strftime</code> formats, the third parameter is required to be zero.
 Since non-static C++ methods have an implicit <code>this</code> argument, the
 arguments of such methods should be counted from two, not one, when
 giving values for <var>string-index</var> and <var>first-to-check</var>.
 </p>
 <p>In the example above, the format string (<code>my_format</code>) is the second
 argument of the function <code>my_print</code>, and the arguments to check
 start with the third argument, so the correct parameters for the format
 attribute are 2 and 3.
 </p>
 <a name="index-ffreestanding-3"></a>
 <a name="index-fno_002dbuiltin-2"></a>
 <p>The <code>format</code> attribute allows you to identify your own functions
 that take format strings as arguments, so that GCC can check the
 calls to these functions for errors.  The compiler always (unless
 <samp>-ffreestanding</samp> or <samp>-fno-builtin</samp> is used) checks formats
 for the standard library functions <code>printf</code>, <code>fprintf</code>,
 <code>sprintf</code>, <code>scanf</code>, <code>fscanf</code>, <code>sscanf</code>, <code>strftime</code>,
 <code>vprintf</code>, <code>vfprintf</code> and <code>vsprintf</code> whenever such
 warnings are requested (using <samp>-Wformat</samp>), so there is no need to
 modify the header file <samp>stdio.h</samp>.  In C99 mode, the functions
 <code>snprintf</code>, <code>vsnprintf</code>, <code>vscanf</code>, <code>vfscanf</code> and
 <code>vsscanf</code> are also checked.  Except in strictly conforming C
 standard modes, the X/Open function <code>strfmon</code> is also checked as
 are <code>printf_unlocked</code> and <code>fprintf_unlocked</code>.
 See <a href="C-Dialect-Options.html#C-Dialect-Options">Options Controlling C Dialect</a>.
 </p>
 <p>For Objective-C dialects, <code>NSString</code> (or <code>__NSString__</code>) is
 recognized in the same context.  Declarations including these format attributes
 are parsed for correct syntax, however the result of checking of such format
 strings is not yet defined, and is not carried out by this version of the
 compiler.
 </p>
 <p>The target may also provide additional types of format checks.
 See <a href="Target-Format-Checks.html#Target-Format-Checks">Format Checks Specific to Particular
 Target Machines</a>.
 </p>
 </dd>
 <dt><code>format_arg (<var>string-index</var>)</code></dt>
 <dd><a name="index-format_005farg-function-attribute"></a>
 <a name="index-Wformat_002dnonliteral-1"></a>
 <p>The <code>format_arg</code> attribute specifies that a function takes a format
 string for a <code>printf</code>, <code>scanf</code>, <code>strftime</code> or
 <code>strfmon</code> style function and modifies it (for example, to translate
 it into another language), so the result can be passed to a
 <code>printf</code>, <code>scanf</code>, <code>strftime</code> or <code>strfmon</code> style
 function (with the remaining arguments to the format function the same
 as they would have been for the unmodified string).  For example, the
 declaration:
 </p>
 <div class="smallexample">
 <pre class="smallexample">extern char *
 my_dgettext (char *my_domain, const char *my_format)
       __attribute__ ((format_arg (2)));
 </pre></div>

 <p>causes the compiler to check the arguments in calls to a <code>printf</code>,
 <code>scanf</code>, <code>strftime</code> or <code>strfmon</code> type function, whose
 format string argument is a call to the <code>my_dgettext</code> function, for
 consistency with the format string argument <code>my_format</code>.  If the
 <code>format_arg</code> attribute had not been specified, all the compiler
 could tell in such calls to format functions would be that the format
 string argument is not constant; this would generate a warning when
 <samp>-Wformat-nonliteral</samp> is used, but the calls could not be checked
 without the attribute.
 </p>
 <p>The parameter <var>string-index</var> specifies which argument is the format
 string argument (starting from one).  Since non-static C++ methods have
 an implicit <code>this</code> argument, the arguments of such methods should
 be counted from two.
 </p>
 <p>The <code>format_arg</code> attribute allows you to identify your own
 functions that modify format strings, so that GCC can check the
 calls to <code>printf</code>, <code>scanf</code>, <code>strftime</code> or <code>strfmon</code>
 type function whose operands are a call to one of your own function.
 The compiler always treats <code>gettext</code>, <code>dgettext</code>, and
 <code>dcgettext</code> in this manner except when strict ISO C support is
 requested by <samp>-ansi</samp> or an appropriate <samp>-std</samp> option, or
 <samp>-ffreestanding</samp> or <samp>-fno-builtin</samp>
 is used.  See <a href="C-Dialect-Options.html#C-Dialect-Options">Options
 Controlling C Dialect</a>.
 </p>
 <p>For Objective-C dialects, the <code>format-arg</code> attribute may refer to an
 <code>NSString</code> reference for compatibility with the <code>format</code> attribute
 above.
 </p>
 <p>The target may also allow additional types in <code>format-arg</code> attributes.
 See <a href="Target-Format-Checks.html#Target-Format-Checks">Format Checks Specific to Particular
 Target Machines</a>.
 </p>
 </dd>
 <dt><code>function_vector</code></dt>
 <dd><a name="index-calling-functions-through-the-function-vector-on-H8_002f300_002c-M16C_002c-M32C-and-SH2A-processors"></a>
 <p>Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
 function should be called through the function vector.  Calling a
 function through the function vector reduces code size, however;
 the function vector has a limited size (maximum 128 entries on the H8/300
 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
 </p>
 <p>On SH2A targets, this attribute declares a function to be called using the
 TBR relative addressing mode.  The argument to this attribute is the entry
 number of the same function in a vector table containing all the TBR
 relative addressable functions.  For correct operation the TBR must be setup
 accordingly to point to the start of the vector table before any functions with
 this attribute are invoked.  Usually a good place to do the initialization is
 the startup routine.  The TBR relative vector table can have at max 256 function
 entries.  The jumps to these functions are generated using a SH2A specific,
 non delayed branch instruction JSR/N @(disp8,TBR).  You must use GAS and GLD
 from GNU binutils version 2.7 or later for this attribute to work correctly.
 </p>
 <p>Please refer the example of M16C target, to see the use of this
 attribute while declaring a function,
 </p>
 <p>In an application, for a function being called once, this attribute
 saves at least 8 bytes of code; and if other successive calls are being
 made to the same function, it saves 2 bytes of code per each of these
 calls.
 </p>
 <p>On M16C/M32C targets, the <code>function_vector</code> attribute declares a
 special page subroutine call function. Use of this attribute reduces
 the code size by 2 bytes for each call generated to the
 subroutine. The argument to the attribute is the vector number entry
 from the special page vector table which contains the 16 low-order
 bits of the subroutine&rsquo;s entry address. Each vector table has special
 page number (18 to 255) that is used in <code>jsrs</code> instructions.
 Jump addresses of the routines are generated by adding 0x0F0000 (in
 case of M16C targets) or 0xFF0000 (in case of M32C targets), to the
 2-byte addresses set in the vector table. Therefore you need to ensure
 that all the special page vector routines should get mapped within the
 address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF
 (for M32C).
 </p>
 <p>In the following example 2 bytes are saved for each call to
 function <code>foo</code>.
 </p>
 <div class="smallexample">
 <pre class="smallexample">void foo (void) __attribute__((function_vector(0x18)));
 void foo (void)
 {
 }

 void bar (void)
 {
     foo();
 }
 </pre></div>

 <p>If functions are defined in one file and are called in another file,
 then be sure to write this declaration in both files.
 </p>
 <p>This attribute is ignored for R8C target.
 </p>
 </dd>
 <dt><code>ifunc (&quot;<var>resolver</var>&quot;)</code></dt>
 <dd><a name="index-ifunc-attribute"></a>
 <p>The <code>ifunc</code> attribute is used to mark a function as an indirect
 function using the STT_GNU_IFUNC symbol type extension to the ELF
 standard.  This allows the resolution of the symbol value to be
 determined dynamically at load time, and an optimized version of the
 routine can be selected for the particular processor or other system
 characteristics determined then.  To use this attribute, first define
 the implementation functions available, and a resolver function that
 returns a pointer to the selected implementation function.  The
 implementation functions&rsquo; declarations must match the API of the
 function being implemented, the resolver&rsquo;s declaration is be a
 function returning pointer to void function returning void:
 </p>
 <div class="smallexample">
 <pre class="smallexample">void *my_memcpy (void *dst, const void *src, size_t len)
 {
   &hellip;
 }

 static void (*resolve_memcpy (void)) (void)
 {
   return my_memcpy; // we'll just always select this routine
 }
 </pre></div>

 <p>The exported header file declaring the function the user calls would
 contain:
 </p>
 <div class="smallexample">
 <pre class="smallexample">extern void *memcpy (void *, const void *, size_t);
 </pre></div>

 <p>allowing the user to call this as a regular function, unaware of the
 implementation.  Finally, the indirect function needs to be defined in
 the same translation unit as the resolver function:
 </p>
 <div class="smallexample">
 <pre class="smallexample">void *memcpy (void *, const void *, size_t)
      __attribute__ ((ifunc (&quot;resolve_memcpy&quot;)));
 </pre></div>

 <p>Indirect functions cannot be weak, and require a recent binutils (at
 least version 2.20.1), and GNU C library (at least version 2.11.1).
 </p>
 </dd>
 <dt><code>interrupt</code></dt>
 <dd><a name="index-interrupt-handler-functions-2"></a>
 <p>Use this attribute on the ARC, ARM, AVR, CR16, Epiphany, M32C, M32R/D,
 m68k, MeP, MIPS, MSP430, RL78, RX and Xstormy16 ports to indicate that
 the specified function is an
 interrupt handler.  The compiler generates function entry and exit
 sequences suitable for use in an interrupt handler when this attribute
 is present.  With Epiphany targets it may also generate a special section with
 code to initialize the interrupt vector table.
 </p>
 <p>Note, interrupt handlers for the Blackfin, H8/300, H8/300H, H8S, MicroBlaze,
 and SH processors can be specified via the <code>interrupt_handler</code> attribute.
 </p>
 <p>Note, on the ARC, you must specify the kind of interrupt to be handled
 in a parameter to the interrupt attribute like this:
 </p>
 <div class="smallexample">
 <pre class="smallexample">void f () __attribute__ ((interrupt (&quot;ilink1&quot;)));
 </pre></div>

 <p>Permissible values for this parameter are: <code>ilink1</code><!-- /@w --> and
 <code>ilink2</code><!-- /@w -->.
 </p>
 <p>Note, on the AVR, the hardware globally disables interrupts when an
 interrupt is executed.  The first instruction of an interrupt handler
 declared with this attribute is a <code>SEI</code> instruction to
 re-enable interrupts.  See also the <code>signal</code> function attribute
 that does not insert a <code>SEI</code> instruction.  If both <code>signal</code> and
 <code>interrupt</code> are specified for the same function, <code>signal</code>
 is silently ignored.
 </p>
 <p>Note, for the ARM, you can specify the kind of interrupt to be handled by
 adding an optional parameter to the interrupt attribute like this:
 </p>
 <div class="smallexample">
 <pre class="smallexample">void f () __attribute__ ((interrupt (&quot;IRQ&quot;)));
 </pre></div>

 <p>Permissible values for this parameter are: <code>IRQ</code>, <code>FIQ</code>,
 <code>SWI</code>, <code>ABORT</code> and <code>UNDEF</code>.
 </p>
 <p>On ARMv7-M the interrupt type is ignored, and the attribute means the function
 may be called with a word-aligned stack pointer.
 </p>
 <p>Note, for the MSP430 you can provide an argument to the interrupt
 attribute which specifies a name or number.  If the argument is a
 number it indicates the slot in the interrupt vector table (0 - 31) to
 which this handler should be assigned.  If the argument is a name it
 is treated as a symbolic name for the vector slot.  These names should
 match up with appropriate entries in the linker script.  By default
 the names <code>watchdog</code> for vector 26, <code>nmi</code> for vector 30 and
 <code>reset</code> for vector 31 are recognised.
 </p>
 <p>You can also use the following function attributes to modify how
 normal functions interact with interrupt functions:
 </p>
 <dl compact="compact">
 <dt><code>critical</code></dt>
 <dd><a name="index-critical-attribute"></a>
 <p>Critical functions disable interrupts upon entry and restore the
 previous interrupt state upon exit.  Critical functions cannot also
 have the <code>naked</code> or <code>reentrant</code> attributes.  They can have
 the <code>interrupt</code> attribute.
 </p>
 </dd>
 <dt><code>reentrant</code></dt>
 <dd><a name="index-reentrant-attribute"></a>
 <p>Reentrant functions disable interrupts upon entry and enable them
 upon exit.  Reentrant functions cannot also have the <code>naked</code>
 or <code>critical</code> attributes.  They can have the <code>interrupt</code>
 attribute.
 </p>
 </dd>
 <dt><code>wakeup</code></dt>
 <dd><a name="index-wakeup-attribute"></a>
 <p>This attribute only applies to interrupt functions.  It is silently
 ignored if applied to a non-interrupt function.  A wakeup interrupt
 function will rouse the processor from any low-power state that it
 might be in when the function exits.
 </p>
 </dd>
 </dl>

 <p>On Epiphany targets one or more optional parameters can be added like this:
 </p>
 <div class="smallexample">
 <pre class="smallexample">void __attribute__ ((interrupt (&quot;dma0, dma1&quot;))) universal_dma_handler ();
 </pre></div>

 <p>Permissible values for these parameters are: <code>reset</code><!-- /@w -->,
 <code><span class="nolinebreak">software_exception</span></code><!-- /@w -->, <code><span class="nolinebreak">page_miss</span></code><!-- /@w -->,
 <code>timer0</code><!-- /@w -->, <code>timer1</code><!-- /@w -->, <code>message</code><!-- /@w -->,
 <code>dma0</code><!-- /@w -->, <code>dma1</code><!-- /@w -->, <code>wand</code><!-- /@w --> and <code>swi</code><!-- /@w -->.
 Multiple parameters indicate that multiple entries in the interrupt
 vector table should be initialized for this function, i.e. for each
 parameter <var>name</var><!-- /@w -->, a jump to the function is emitted in
 the section <span class="nolinebreak">ivt_entry_</span><var>name</var><!-- /@w -->.  The parameter(s) may be omitted
 entirely, in which case no interrupt vector table entry is provided.
 </p>
 <p>Note, on Epiphany targets, interrupts are enabled inside the function
 unless the <code>disinterrupt</code> attribute is also specified.
 </p>
 <p>On Epiphany targets, you can also use the following attribute to
 modify the behavior of an interrupt handler:
 </p><dl compact="compact">
 <dt><code>forwarder_section</code></dt>
 <dd><a name="index-forwarder_005fsection-attribute"></a>
 <p>The interrupt handler may be in external memory which cannot be
 reached by a branch instruction, so generate a local memory trampoline
 to transfer control.  The single parameter identifies the section where
 the trampoline is placed.
 </p></dd>
 </dl>

 <p>The following examples are all valid uses of these attributes on
 Epiphany targets:
 </p><div class="smallexample">
 <pre class="smallexample">void __attribute__ ((interrupt)) universal_handler ();
 void __attribute__ ((interrupt (&quot;dma1&quot;))) dma1_handler ();
 void __attribute__ ((interrupt (&quot;dma0, dma1&quot;))) universal_dma_handler ();
 void __attribute__ ((interrupt (&quot;timer0&quot;), disinterrupt))
   fast_timer_handler ();
 void __attribute__ ((interrupt (&quot;dma0, dma1&quot;), forwarder_section (&quot;tramp&quot;)))
   external_dma_handler ();
 </pre></div>

 <p>On MIPS targets, you can use the following attributes to modify the behavior
 of an interrupt handler:
 </p><dl compact="compact">
 <dt><code>use_shadow_register_set</code></dt>
 <dd><a name="index-use_005fshadow_005fregister_005fset-attribute"></a>
 <p>Assume that the handler uses a shadow register set, instead of
 the main general-purpose registers.
 </p>
 </dd>
 <dt><code>keep_interrupts_masked</code></dt>
 <dd><a name="index-keep_005finterrupts_005fmasked-attribute"></a>
 <p>Keep interrupts masked for the whole function.  Without this attribute,
 GCC tries to reenable interrupts for as much of the function as it can.
 </p>
 </dd>
 <dt><code>use_debug_exception_return</code></dt>
 <dd><a name="index-use_005fdebug_005fexception_005freturn-attribute"></a>
 <p>Return using the <code>deret</code> instruction.  Interrupt handlers that don&rsquo;t
 have this attribute return using <code>eret</code> instead.
 </p></dd>
 </dl>

 <p>You can use any combination of these attributes, as shown below:
 </p><div class="smallexample">
 <pre class="smallexample">void __attribute__ ((interrupt)) v0 ();
 void __attribute__ ((interrupt, use_shadow_register_set)) v1 ();
 void __attribute__ ((interrupt, keep_interrupts_masked)) v2 ();
 void __attribute__ ((interrupt, use_debug_exception_return)) v3 ();
 void __attribute__ ((interrupt, use_shadow_register_set,
                      keep_interrupts_masked)) v4 ();
 void __attribute__ ((interrupt, use_shadow_register_set,
                      use_debug_exception_return)) v5 ();
 void __attribute__ ((interrupt, keep_interrupts_masked,
                      use_debug_exception_return)) v6 ();
 void __attribute__ ((interrupt, use_shadow_register_set,
                      keep_interrupts_masked,
                      use_debug_exception_return)) v7 ();
 </pre></div>

 <p>On NDS32 target, this attribute is to indicate that the specified function
 is an interrupt handler.  The compiler will generate corresponding sections
 for use in an interrupt handler.  You can use the following attributes
 to modify the behavior:
 </p><dl compact="compact">
 <dt><code>nested</code></dt>
 <dd><a name="index-nested-attribute"></a>
 <p>This interrupt service routine is interruptible.
 </p></dd>
 <dt><code>not_nested</code></dt>
 <dd><a name="index-not_005fnested-attribute"></a>
 <p>This interrupt service routine is not interruptible.
 </p></dd>
 <dt><code>nested_ready</code></dt>
 <dd><a name="index-nested_005fready-attribute"></a>
 <p>This interrupt service routine is interruptible after <code>PSW.GIE</code>
 (global interrupt enable) is set.  This allows interrupt service routine to
 finish some short critical code before enabling interrupts.
 </p></dd>
 <dt><code>save_all</code></dt>
 <dd><a name="index-save_005fall-attribute"></a>
 <p>The system will help save all registers into stack before entering
 interrupt handler.
 </p></dd>
 <dt><code>partial_save</code></dt>
 <dd><a name="index-partial_005fsave-attribute"></a>
 <p>The system will help save caller registers into stack before entering
 interrupt handler.
 </p></dd>
 </dl>

 <p>On RL78, use <code>brk_interrupt</code> instead of <code>interrupt</code> for
 handlers intended to be used with the <code>BRK</code> opcode (i.e. those
 that must end with <code>RETB</code> instead of <code>RETI</code>).
 </p>
 </dd>
 <dt><code>interrupt_handler</code></dt>
 <dd><a name="index-interrupt-handler-functions-on-the-Blackfin_002c-m68k_002c-H8_002f300-and-SH-processors"></a>
 <p>Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
 indicate that the specified function is an interrupt handler.  The compiler
 generates function entry and exit sequences suitable for use in an
 interrupt handler when this attribute is present.
 </p>
 </dd>
 <dt><code>interrupt_thread</code></dt>
 <dd><a name="index-interrupt-thread-functions-on-fido"></a>
 <p>Use this attribute on fido, a subarchitecture of the m68k, to indicate
 that the specified function is an interrupt handler that is designed
 to run as a thread.  The compiler omits generate prologue/epilogue
 sequences and replaces the return instruction with a <code>sleep</code>
 instruction.  This attribute is available only on fido.
 </p>
 </dd>
 <dt><code>isr</code></dt>
 <dd><a name="index-interrupt-service-routines-on-ARM"></a>
 <p>Use this attribute on ARM to write Interrupt Service Routines. This is an
 alias to the <code>interrupt</code> attribute above.
 </p>
 </dd>
 <dt><code>kspisusp</code></dt>
 <dd><a name="index-User-stack-pointer-in-interrupts-on-the-Blackfin"></a>
 <p>When used together with <code>interrupt_handler</code>, <code>exception_handler</code>
 or <code>nmi_handler</code>, code is generated to load the stack pointer
 from the USP register in the function prologue.
 </p>
 </dd>
 <dt><code>l1_text</code></dt>
 <dd><a name="index-l1_005ftext-function-attribute"></a>
 <p>This attribute specifies a function to be placed into L1 Instruction
 SRAM. The function is put into a specific section named <code>.l1.text</code>.
 With <samp>-mfdpic</samp>, function calls with a such function as the callee
 or caller uses inlined PLT.
 </p>
 </dd>
 <dt><code>l2</code></dt>
 <dd><a name="index-l2-function-attribute"></a>
 <p>On the Blackfin, this attribute specifies a function to be placed into L2
 SRAM. The function is put into a specific section named
 <code>.l1.text</code>. With <samp>-mfdpic</samp>, callers of such functions use
 an inlined PLT.
 </p>
 </dd>
 <dt><code>leaf</code></dt>
 <dd><a name="index-leaf-function-attribute"></a>
 <p>Calls to external functions with this attribute must return to the current
 compilation unit only by return or by exception handling.  In particular, leaf
 functions are not allowed to call callback function passed to it from the current
 compilation unit or directly call functions exported by the unit or longjmp
 into the unit.  Leaf function might still call functions from other compilation
 units and thus they are not necessarily leaf in the sense that they contain no
 function calls at all.
 </p>
 <p>The attribute is intended for library functions to improve dataflow analysis.
 The compiler takes the hint that any data not escaping the current compilation unit can
 not be used or modified by the leaf function.  For example, the <code>sin</code> function
 is a leaf function, but <code>qsort</code> is not.
 </p>
 <p>Note that leaf functions might invoke signals and signal handlers might be
 defined in the current compilation unit and use static variables.  The only
 compliant way to write such a signal handler is to declare such variables
 <code>volatile</code>.
 </p>
 <p>The attribute has no effect on functions defined within the current compilation
 unit.  This is to allow easy merging of multiple compilation units into one,
 for example, by using the link-time optimization.  For this reason the
 attribute is not allowed on types to annotate indirect calls.
 </p>
 </dd>
 <dt><code>long_call/medium_call/short_call</code></dt>
 <dd><a name="index-indirect-calls-on-ARC"></a>
 <a name="index-indirect-calls-on-ARM"></a>
 <a name="index-indirect-calls-on-Epiphany"></a>
 <p>These attributes specify how a particular function is called on
 ARC, ARM and Epiphany - with <code>medium_call</code> being specific to ARC.
 These attributes override the
 <samp>-mlong-calls</samp> (see <a href="ARM-Options.html#ARM-Options">ARM Options</a> and <a href="ARC-Options.html#ARC-Options">ARC Options</a>)
 and <samp>-mmedium-calls</samp> (see <a href="ARC-Options.html#ARC-Options">ARC Options</a>)
 command-line switches and <code>#pragma long_calls</code> settings.  For ARM, the
 <code>long_call</code> attribute indicates that the function might be far
 away from the call site and require a different (more expensive)
 calling sequence.   The <code>short_call</code> attribute always places
 the offset to the function from the call site into the &lsquo;<samp>BL</samp>&rsquo;
 instruction directly.
 </p>
 <p>For ARC, a function marked with the <code>long_call</code> attribute is
 always called using register-indirect jump-and-link instructions,
 thereby enabling the called function to be placed anywhere within the
 32-bit address space.  A function marked with the <code>medium_call</code>
 attribute will always be close enough to be called with an unconditional
 branch-and-link instruction, which has a 25-bit offset from
 the call site.  A function marked with the <code>short_call</code>
 attribute will always be close enough to be called with a conditional
 branch-and-link instruction, which has a 21-bit offset from
 the call site.
 </p>
 </dd>
 <dt><code>longcall/shortcall</code></dt>
 <dd><a name="index-functions-called-via-pointer-on-the-RS_002f6000-and-PowerPC"></a>
 <p>On the Blackfin, RS/6000 and PowerPC, the <code>longcall</code> attribute
 indicates that the function might be far away from the call site and
 require a different (more expensive) calling sequence.  The
 <code>shortcall</code> attribute indicates that the function is always close
 enough for the shorter calling sequence to be used.  These attributes
 override both the <samp>-mlongcall</samp> switch and, on the RS/6000 and
 PowerPC, the <code>#pragma longcall</code> setting.
 </p>
 <p>See <a href="RS_002f6000-and-PowerPC-Options.html#RS_002f6000-and-PowerPC-Options">RS/6000 and PowerPC Options</a>, for more information on whether long
 calls are necessary.
 </p>
 </dd>
 <dt><code>long_call/near/far</code></dt>
 <dd><a name="index-indirect-calls-on-MIPS"></a>
 <p>These attributes specify how a particular function is called on MIPS.
 The attributes override the <samp>-mlong-calls</samp> (see <a href="MIPS-Options.html#MIPS-Options">MIPS Options</a>)
 command-line switch.  The <code>long_call</code> and <code>far</code> attributes are
 synonyms, and cause the compiler to always call
 the function by first loading its address into a register, and then using
 the contents of that register.  The <code>near</code> attribute has the opposite
 effect; it specifies that non-PIC calls should be made using the more
 efficient <code>jal</code> instruction.
 </p>
 </dd>
 <dt><code>malloc</code></dt>
 <dd><a name="index-malloc-attribute"></a>
 <p>The <code>malloc</code> attribute is used to tell the compiler that a function
 may be treated as if any non-<code>NULL</code> pointer it returns cannot
 alias any other pointer valid when the function returns and that the memory
 has undefined content.
 This often improves optimization.
 Standard functions with this property include <code>malloc</code> and
 <code>calloc</code>.  <code>realloc</code>-like functions do not have this
 property as the memory pointed to does not have undefined content.
 </p>
 </dd>
 <dt><code>mips16/nomips16</code></dt>
 <dd><a name="index-mips16-attribute"></a>
 <a name="index-nomips16-attribute"></a>

 <p>On MIPS targets, you can use the <code>mips16</code> and <code>nomips16</code>
 function attributes to locally select or turn off MIPS16 code generation.
 A function with the <code>mips16</code> attribute is emitted as MIPS16 code,
 while MIPS16 code generation is disabled for functions with the
 <code>nomips16</code> attribute.  These attributes override the
 <samp>-mips16</samp> and <samp>-mno-mips16</samp> options on the command line
 (see <a href="MIPS-Options.html#MIPS-Options">MIPS Options</a>).
 </p>
 <p>When compiling files containing mixed MIPS16 and non-MIPS16 code, the
 preprocessor symbol <code>__mips16</code> reflects the setting on the command line,
 not that within individual functions.  Mixed MIPS16 and non-MIPS16 code
 may interact badly with some GCC extensions such as <code>__builtin_apply</code>
 (see <a href="Constructing-Calls.html#Constructing-Calls">Constructing Calls</a>).
 </p>
 </dd>
 <dt><code>micromips/nomicromips</code></dt>
 <dd><a name="index-micromips-attribute"></a>
 <a name="index-nomicromips-attribute"></a>

 <p>On MIPS targets, you can use the <code>micromips</code> and <code>nomicromips</code>
 function attributes to locally select or turn off microMIPS code generation.
 A function with the <code>micromips</code> attribute is emitted as microMIPS code,
 while microMIPS code generation is disabled for functions with the
 <code>nomicromips</code> attribute.  These attributes override the
 <samp>-mmicromips</samp> and <samp>-mno-micromips</samp> options on the command line
 (see <a href="MIPS-Options.html#MIPS-Options">MIPS Options</a>).
 </p>
 <p>When compiling files containing mixed microMIPS and non-microMIPS code, the
 preprocessor symbol <code>__mips_micromips</code> reflects the setting on the
 command line,
 not that within individual functions.  Mixed microMIPS and non-microMIPS code
 may interact badly with some GCC extensions such as <code>__builtin_apply</code>
 (see <a href="Constructing-Calls.html#Constructing-Calls">Constructing Calls</a>).
 </p>
 </dd>
 <dt><code>model (<var>model-name</var>)</code></dt>
 <dd><a name="index-function-addressability-on-the-M32R_002fD"></a>
 <a name="index-variable-addressability-on-the-IA_002d64"></a>

 <p>On the M32R/D, use this attribute to set the addressability of an
 object, and of the code generated for a function.  The identifier
 <var>model-name</var> is one of <code>small</code>, <code>medium</code>, or
 <code>large</code>, representing each of the code models.
 </p>
 <p>Small model objects live in the lower 16MB of memory (so that their
 addresses can be loaded with the <code>ld24</code> instruction), and are
 callable with the <code>bl</code> instruction.
 </p>
 <p>Medium model objects may live anywhere in the 32-bit address space (the
 compiler generates <code>seth/add3</code> instructions to load their addresses),
 and are callable with the <code>bl</code> instruction.
 </p>
 <p>Large model objects may live anywhere in the 32-bit address space (the
 compiler generates <code>seth/add3</code> instructions to load their addresses),
 and may not be reachable with the <code>bl</code> instruction (the compiler
 generates the much slower <code>seth/add3/jl</code> instruction sequence).
 </p>
 <p>On IA-64, use this attribute to set the addressability of an object.
 At present, the only supported identifier for <var>model-name</var> is
 <code>small</code>, indicating addressability via &ldquo;small&rdquo; (22-bit)
 addresses (so that their addresses can be loaded with the <code>addl</code>
 instruction).  Caveat: such addressing is by definition not position
 independent and hence this attribute must not be used for objects
 defined by shared libraries.
 </p>
 </dd>
 <dt><code>ms_abi/sysv_abi</code></dt>
 <dd><a name="index-ms_005fabi-attribute"></a>
 <a name="index-sysv_005fabi-attribute"></a>

 <p>On 32-bit and 64-bit (i?86|x86_64)-*-* targets, you can use an ABI attribute
 to indicate which calling convention should be used for a function.  The
 <code>ms_abi</code> attribute tells the compiler to use the Microsoft ABI,
 while the <code>sysv_abi</code> attribute tells the compiler to use the ABI
 used on GNU/Linux and other systems.  The default is to use the Microsoft ABI
 when targeting Windows.  On all other systems, the default is the x86/AMD ABI.
 </p>
 <p>Note, the <code>ms_abi</code> attribute for Microsoft Windows 64-bit targets currently
 requires the <samp>-maccumulate-outgoing-args</samp> option.
 </p>
 </dd>
 <dt><code>callee_pop_aggregate_return (<var>number</var>)</code></dt>
 <dd><a name="index-callee_005fpop_005faggregate_005freturn-attribute"></a>

 <p>On 32-bit i?86-*-* targets, you can use this attribute to control how
 aggregates are returned in memory.  If the caller is responsible for
 popping the hidden pointer together with the rest of the arguments, specify
 <var>number</var> equal to zero.  If callee is responsible for popping the
 hidden pointer, specify <var>number</var> equal to one.
 </p>
 <p>The default i386 ABI assumes that the callee pops the
 stack for hidden pointer.  However, on 32-bit i386 Microsoft Windows targets,
 the compiler assumes that the
 caller pops the stack for hidden pointer.
 </p>
 </dd>
 <dt><code>ms_hook_prologue</code></dt>
 <dd><a name="index-ms_005fhook_005fprologue-attribute"></a>

 <p>On 32-bit i[34567]86-*-* targets and 64-bit x86_64-*-* targets, you can use
 this function attribute to make GCC generate the &ldquo;hot-patching&rdquo; function
 prologue used in Win32 API functions in Microsoft Windows XP Service Pack 2
 and newer.
 </p>
 </dd>
 <dt><code>hotpatch (<var>halfwords-before-function-label</var>,<var>halfwords-after-function-label</var>)</code></dt>
 <dd><a name="index-hotpatch-attribute"></a>

 <p>On S/390 System z targets, you can use this function attribute to
 make GCC generate a &ldquo;hot-patching&rdquo; function prologue.  If the
 <samp>-mhotpatch=</samp> command-line option is used at the same time,
 the <code>hotpatch</code> attribute takes precedence.  The first of the
 two arguments specifies the number of halfwords to be added before
 the function label.  A second argument can be used to specify the
 number of halfwords to be added after the function label.  For
 both arguments the maximum allowed value is 1000000.
 </p>
 <p>If both ar guments are zero, hotpatching is disabled.
 </p>
 </dd>
 <dt><code>naked</code></dt>
 <dd><a name="index-function-without-a-prologue_002fepilogue-code"></a>
 <p>Use this attribute on the ARM, AVR, MCORE, MSP430, NDS32, RL78, RX and SPU
 ports to indicate that the specified function does not need prologue/epilogue
 sequences generated by the compiler.
 It is up to the programmer to provide these sequences. The
 only statements that can be safely included in naked functions are
 <code>asm</code> statements that do not have operands.  All other statements,
 including declarations of local variables, <code>if</code> statements, and so
 forth, should be avoided.  Naked functions should be used to implement the
 body of an assembly function, while allowing the compiler to construct
 the requisite function declaration for the assembler.
 </p>
 </dd>
 <dt><code>near</code></dt>
 <dd><a name="index-functions-that-do-not-handle-memory-bank-switching-on-68HC11_002f68HC12"></a>
 <p>On 68HC11 and 68HC12 the <code>near</code> attribute causes the compiler to
 use the normal calling convention based on <code>jsr</code> and <code>rts</code>.
 This attribute can be used to cancel the effect of the <samp>-mlong-calls</samp>
 option.
 </p>
 <p>On MeP targets this attribute causes the compiler to assume the called
 function is close enough to use the normal calling convention,
 overriding the <samp>-mtf</samp> command-line option.
 </p>
 </dd>
 <dt><code>nesting</code></dt>
 <dd><a name="index-Allow-nesting-in-an-interrupt-handler-on-the-Blackfin-processor_002e"></a>
 <p>Use this attribute together with <code>interrupt_handler</code>,
 <code>exception_handler</code> or <code>nmi_handler</code> to indicate that the function
 entry code should enable nested interrupts or exceptions.
 </p>
 </dd>
 <dt><code>nmi_handler</code></dt>
 <dd><a name="index-NMI-handler-functions-on-the-Blackfin-processor"></a>
 <p>Use this attribute on the Blackfin to indicate that the specified function
 is an NMI handler.  The compiler generates function entry and
 exit sequences suitable for use in an NMI handler when this
 attribute is present.
 </p>
 </dd>
 <dt><code>nocompression</code></dt>
 <dd><a name="index-nocompression-attribute"></a>
 <p>On MIPS targets, you can use the <code>nocompression</code> function attribute
 to locally turn off MIPS16 and microMIPS code generation.  This attribute
 overrides the <samp>-mips16</samp> and <samp>-mmicromips</samp> options on the
 command line (see <a href="MIPS-Options.html#MIPS-Options">MIPS Options</a>).
 </p>
 </dd>
 <dt><code>no_instrument_function</code></dt>
 <dd><a name="index-no_005finstrument_005ffunction-function-attribute"></a>
 <a name="index-finstrument_002dfunctions-1"></a>
 <p>If <samp>-finstrument-functions</samp> is given, profiling function calls are
 generated at entry and exit of most user-compiled functions.
 Functions with this attribute are not so instrumented.
 </p>
 </dd>
 <dt><code>no_split_stack</code></dt>
 <dd><a name="index-no_005fsplit_005fstack-function-attribute"></a>
 <a name="index-fsplit_002dstack-1"></a>
 <p>If <samp>-fsplit-stack</samp> is given, functions have a small
 prologue which decides whether to split the stack.  Functions with the
 <code>no_split_stack</code> attribute do not have that prologue, and thus
 may run with only a small amount of stack space available.
 </p>
 </dd>
 <dt><code>noinline</code></dt>
 <dd><a name="index-noinline-function-attribute"></a>
 <p>This function attribute prevents a function from being considered for
 inlining.
 If the function does not have side-effects, there are optimizations
 other than inlining that cause function calls to be optimized away,
 although the function call is live.  To keep such calls from being
 optimized away, put
 </p><div class="smallexample">
 <pre class="smallexample">asm (&quot;&quot;);
 </pre></div>

 <p>(see <a href="Extended-Asm.html#Extended-Asm">Extended Asm</a>) in the called function, to serve as a special
 side-effect.
 </p>
 </dd>
 <dt><code>noclone</code></dt>
 <dd><a name="index-noclone-function-attribute"></a>
 <p>This function attribute prevents a function from being considered for
 cloning&mdash;a mechanism that produces specialized copies of functions
 and which is (currently) performed by interprocedural constant
 propagation.
 </p>
 </dd>
 <dt><code>nonnull (<var>arg-index</var>, &hellip;)</code></dt>
 <dd><a name="index-nonnull-function-attribute"></a>
 <p>The <code>nonnull</code> attribute specifies that some function parameters should
 be non-null pointers.  For instance, the declaration:
 </p>
 <div class="smallexample">
 <pre class="smallexample">extern void *
 my_memcpy (void *dest, const void *src, size_t len)
         __attribute__((nonnull (1, 2)));
 </pre></div>

 <p>causes the compiler to check that, in calls to <code>my_memcpy</code>,
 arguments <var>dest</var> and <var>src</var> are non-null.  If the compiler
 determines that a null pointer is passed in an argument slot marked
 as non-null, and the <samp>-Wnonnull</samp> option is enabled, a warning
 is issued.  The compiler may also choose to make optimizations based
 on the knowledge that certain function arguments will never be null.
 </p>
 <p>If no argument index list is given to the <code>nonnull</code> attribute,
 all pointer arguments are marked as non-null.  To illustrate, the
 following declaration is equivalent to the previous example:
 </p>
 <div class="smallexample">
 <pre class="smallexample">extern void *
 my_memcpy (void *dest, const void *src, size_t len)
         __attribute__((nonnull));
 </pre></div>

 </dd>
 <dt><code>returns_nonnull</code></dt>
 <dd><a name="index-returns_005fnonnull-function-attribute"></a>
 <p>The <code>returns_nonnull</code> attribute specifies that the function
 return value should be a non-null pointer.  For instance, the declaration:
 </p>
 <div class="smallexample">
 <pre class="smallexample">extern void *
 mymalloc (size_t len) __attribute__((returns_nonnull));
 </pre></div>

 <p>lets the compiler optimize callers based on the knowledge
 that the return value will never be null.
 </p>
 </dd>
 <dt><code>noreturn</code></dt>
 <dd><a name="index-noreturn-function-attribute"></a>
 <p>A few standard library functions, such as <code>abort</code> and <code>exit</code>,
 cannot return.  GCC knows this automatically.  Some programs define
 their own functions that never return.  You can declare them
 <code>noreturn</code> to tell the compiler this fact.  For example,
 </p>
 <div class="smallexample">
 <pre class="smallexample">void fatal () __attribute__ ((noreturn));

 void
 fatal (/* <span class="roman">&hellip;</span> */)
 {
   /* <span class="roman">&hellip;</span> */ /* <span class="roman">Print error message.</span> */ /* <span class="roman">&hellip;</span> */
   exit (1);
 }
 </pre></div>

 <p>The <code>noreturn</code> keyword tells the compiler to assume that
 <code>fatal</code> cannot return.  It can then optimize without regard to what
 would happen if <code>fatal</code> ever did return.  This makes slightly
 better code.  More importantly, it helps avoid spurious warnings of
 uninitialized variables.
 </p>
 <p>The <code>noreturn</code> keyword does not affect the exceptional path when that
 applies: a <code>noreturn</code>-marked function may still return to the caller
 by throwing an exception or calling <code>longjmp</code>.
 </p>
 <p>Do not assume that registers saved by the calling function are
 restored before calling the <code>noreturn</code> function.
 </p>
 <p>It does not make sense for a <code>noreturn</code> function to have a return
 type other than <code>void</code>.
 </p>
 <p>The attribute <code>noreturn</code> is not implemented in GCC versions
 earlier than 2.5.  An alternative way to declare that a function does
 not return, which works in the current version and in some older
 versions, is as follows:
 </p>
 <div class="smallexample">
 <pre class="smallexample">typedef void voidfn ();

 volatile voidfn fatal;
 </pre></div>

 <p>This approach does not work in GNU C++.
 </p>
 </dd>
 <dt><code>nothrow</code></dt>
 <dd><a name="index-nothrow-function-attribute"></a>
 <p>The <code>nothrow</code> attribute is used to inform the compiler that a
 function cannot throw an exception.  For example, most functions in
 the standard C library can be guaranteed not to throw an exception
 with the notable exceptions of <code>qsort</code> and <code>bsearch</code> that
 take function pointer arguments.  The <code>nothrow</code> attribute is not
 implemented in GCC versions earlier than 3.3.
 </p>
 </dd>
 <dt><code>nosave_low_regs</code></dt>
 <dd><a name="index-nosave_005flow_005fregs-attribute"></a>
 <p>Use this attribute on SH targets to indicate that an <code>interrupt_handler</code>
 function should not save and restore registers R0..R7.  This can be used on SH3*
 and SH4* targets that have a second R0..R7 register bank for non-reentrant
 interrupt handlers.
 </p>
 </dd>
 <dt><code>optimize</code></dt>
 <dd><a name="index-optimize-function-attribute"></a>
 <p>The <code>optimize</code> attribute is used to specify that a function is to
 be compiled with different optimization options than specified on the
 command line.  Arguments can either be numbers or strings.  Numbers
 are assumed to be an optimization level.  Strings that begin with
 <code>O</code> are assumed to be an optimization option, while other options
 are assumed to be used with a <code>-f</code> prefix.  You can also use the
 &lsquo;<samp>#pragma GCC optimize</samp>&rsquo; pragma to set the optimization options
 that affect more than one function.
 See <a href="Function-Specific-Option-Pragmas.html#Function-Specific-Option-Pragmas">Function Specific Option Pragmas</a>, for details about the
 &lsquo;<samp>#pragma GCC optimize</samp>&rsquo; pragma.
 </p>
 <p>This can be used for instance to have frequently-executed functions
 compiled with more aggressive optimization options that produce faster
 and larger code, while other functions can be compiled with less
 aggressive options.
 </p>
 </dd>
 <dt><code>OS_main/OS_task</code></dt>
 <dd><a name="index-OS_005fmain-AVR-function-attribute"></a>
 <a name="index-OS_005ftask-AVR-function-attribute"></a>
 <p>On AVR, functions with the <code>OS_main</code> or <code>OS_task</code> attribute
 do not save/restore any call-saved register in their prologue/epilogue.
 </p>
 <p>The <code>OS_main</code> attribute can be used when there <em>is
 guarantee</em> that interrupts are disabled at the time when the function
 is entered.  This saves resources when the stack pointer has to be
 changed to set up a frame for local variables.
 </p>
 <p>The <code>OS_task</code> attribute can be used when there is <em>no
 guarantee</em> that interrupts are disabled at that time when the function
 is entered like for, e.g. task functions in a multi-threading operating
 system. In that case, changing the stack pointer register is
 guarded by save/clear/restore of the global interrupt enable flag.
 </p>
 <p>The differences to the <code>naked</code> function attribute are:
 </p><ul>
 <li> <code>naked</code> functions do not have a return instruction whereas
 <code>OS_main</code> and <code>OS_task</code> functions have a <code>RET</code> or
 <code>RETI</code> return instruction.
 </li><li> <code>naked</code> functions do not set up a frame for local variables
 or a frame pointer whereas <code>OS_main</code> and <code>OS_task</code> do this
 as needed.
 </li></ul>

 </dd>
 <dt><code>pcs</code></dt>
 <dd><a name="index-pcs-function-attribute"></a>

 <p>The <code>pcs</code> attribute can be used to control the calling convention
 used for a function on ARM.  The attribute takes an argument that specifies
 the calling convention to use.
 </p>
 <p>When compiling using the AAPCS ABI (or a variant of it) then valid
 values for the argument are <code>&quot;aapcs&quot;</code> and <code>&quot;aapcs-vfp&quot;</code>.  In
 order to use a variant other than <code>&quot;aapcs&quot;</code> then the compiler must
 be permitted to use the appropriate co-processor registers (i.e., the
 VFP registers must be available in order to use <code>&quot;aapcs-vfp&quot;</code>).
 For example,
 </p>
 <div class="smallexample">
 <pre class="smallexample">/* Argument passed in r0, and result returned in r0+r1.  */
 double f2d (float) __attribute__((pcs(&quot;aapcs&quot;)));
 </pre></div>

 <p>Variadic functions always use the <code>&quot;aapcs&quot;</code> calling convention and
 the compiler rejects attempts to specify an alternative.
 </p>
 </dd>
 <dt><code>pure</code></dt>
 <dd><a name="index-pure-function-attribute"></a>
 <p>Many functions have no effects except the return value and their
 return value depends only on the parameters and/or global variables.
 Such a function can be subject
 to common subexpression elimination and loop optimization just as an
 arithmetic operator would be.  These functions should be declared
 with the attribute <code>pure</code>.  For example,
 </p>
 <div class="smallexample">
 <pre class="smallexample">int square (int) __attribute__ ((pure));
 </pre></div>

 <p>says that the hypothetical function <code>square</code> is safe to call
 fewer times than the program says.
 </p>
 <p>Some of common examples of pure functions are <code>strlen</code> or <code>memcmp</code>.
 Interesting non-pure functions are functions with infinite loops or those
 depending on volatile memory or other system resource, that may change between
 two consecutive calls (such as <code>feof</code> in a multithreading environment).
 </p>
 <p>The attribute <code>pure</code> is not implemented in GCC versions earlier
 than 2.96.
 </p>
 </dd>
 <dt><code>hot</code></dt>
 <dd><a name="index-hot-function-attribute"></a>
 <p>The <code>hot</code> attribute on a function is used to inform the compiler that
 the function is a hot spot of the compiled program.  The function is
 optimized more aggressively and on many target it is placed into special
 subsection of the text section so all hot functions appears close together
 improving locality.
 </p>
 <p>When profile feedback is available, via <samp>-fprofile-use</samp>, hot functions
 are automatically detected and this attribute is ignored.
 </p>
 <p>The <code>hot</code> attribute on functions is not implemented in GCC versions
 earlier than 4.3.
 </p>
 <a name="index-hot-label-attribute"></a>
 <p>The <code>hot</code> attribute on a label is used to inform the compiler that
 path following the label are more likely than paths that are not so
 annotated.  This attribute is used in cases where <code>__builtin_expect</code>
 cannot be used, for instance with computed goto or <code>asm goto</code>.
 </p>
 <p>The <code>hot</code> attribute on labels is not implemented in GCC versions
 earlier than 4.8.
 </p>
 </dd>
 <dt><code>cold</code></dt>
 <dd><a name="index-cold-function-attribute"></a>
 <p>The <code>cold</code> attribute on functions is used to inform the compiler that
 the function is unlikely to be executed.  The function is optimized for
 size rather than speed and on many targets it is placed into special
 subsection of the text section so all cold functions appears close together
 improving code locality of non-cold parts of program.  The paths leading
 to call of cold functions within code are marked as unlikely by the branch
 prediction mechanism.  It is thus useful to mark functions used to handle
 unlikely conditions, such as <code>perror</code>, as cold to improve optimization
 of hot functions that do call marked functions in rare occasions.
 </p>
 <p>When profile feedback is available, via <samp>-fprofile-use</samp>, cold functions
 are automatically detected and this attribute is ignored.
 </p>
 <p>The <code>cold</code> attribute on functions is not implemented in GCC versions
 earlier than 4.3.
 </p>
 <a name="index-cold-label-attribute"></a>
 <p>The <code>cold</code> attribute on labels is used to inform the compiler that
 the path following the label is unlikely to be executed.  This attribute
 is used in cases where <code>__builtin_expect</code> cannot be used, for instance
 with computed goto or <code>asm goto</code>.
 </p>
 <p>The <code>cold</code> attribute on labels is not implemented in GCC versions
 earlier than 4.8.
 </p>
 </dd>
 <dt><code>no_sanitize_address</code></dt>
 <dt><code>no_address_safety_analysis</code></dt>
 <dd><a name="index-no_005fsanitize_005faddress-function-attribute"></a>
 <p>The <code>no_sanitize_address</code> attribute on functions is used
 to inform the compiler that it should not instrument memory accesses
 in the function when compiling with the <samp>-fsanitize=address</samp> option.
 The <code>no_address_safety_analysis</code> is a deprecated alias of the
 <code>no_sanitize_address</code> attribute, new code should use
 <code>no_sanitize_address</code>.
 </p>
 </dd>
 <dt><code>no_sanitize_undefined</code></dt>
 <dd><a name="index-no_005fsanitize_005fundefined-function-attribute"></a>
 <p>The <code>no_sanitize_undefined</code> attribute on functions is used
 to inform the compiler that it should not check for undefined behavior
 in the function when compiling with the <samp>-fsanitize=undefined</samp> option.
 </p>
 </dd>
 <dt><code>regparm (<var>number</var>)</code></dt>
 <dd><a name="index-regparm-attribute"></a>
 <a name="index-functions-that-are-passed-arguments-in-registers-on-the-386-1"></a>
 <p>On the Intel 386, the <code>regparm</code> attribute causes the compiler to
 pass arguments number one to <var>number</var> if they are of integral type
 in registers EAX, EDX, and ECX instead of on the stack.  Functions that
 take a variable number of arguments continue to be passed all of their
 arguments on the stack.
 </p>
 <p>Beware that on some ELF systems this attribute is unsuitable for
 global functions in shared libraries with lazy binding (which is the
 default).  Lazy binding sends the first call via resolving code in
 the loader, which might assume EAX, EDX and ECX can be clobbered, as
 per the standard calling conventions.  Solaris 8 is affected by this.
 Systems with the GNU C Library version 2.1 or higher
 and FreeBSD are believed to be
 safe since the loaders there save EAX, EDX and ECX.  (Lazy binding can be
 disabled with the linker or the loader if desired, to avoid the
 problem.)
 </p>
 </dd>
 <dt><code>reset</code></dt>
 <dd><a name="index-reset-handler-functions"></a>
 <p>Use this attribute on the NDS32 target to indicate that the specified function
 is a reset handler.  The compiler will generate corresponding sections
 for use in a reset handler.  You can use the following attributes
 to provide extra exception handling:
 </p><dl compact="compact">
 <dt><code>nmi</code></dt>
 <dd><a name="index-nmi-attribute"></a>
 <p>Provide a user-defined function to handle NMI exception.
 </p></dd>
 <dt><code>warm</code></dt>
 <dd><a name="index-warm-attribute"></a>
 <p>Provide a user-defined function to handle warm reset exception.
 </p></dd>
 </dl>

 </dd>
 <dt><code>sseregparm</code></dt>
 <dd><a name="index-sseregparm-attribute"></a>
 <p>On the Intel 386 with SSE support, the <code>sseregparm</code> attribute
 causes the compiler to pass up to 3 floating-point arguments in
 SSE registers instead of on the stack.  Functions that take a
 variable number of arguments continue to pass all of their
 floating-point arguments on the stack.
 </p>
 </dd>
 <dt><code>force_align_arg_pointer</code></dt>
 <dd><a name="index-force_005falign_005farg_005fpointer-attribute"></a>
 <p>On the Intel x86, the <code>force_align_arg_pointer</code> attribute may be
 applied to individual function definitions, generating an alternate
 prologue and epilogue that realigns the run-time stack if necessary.
 This supports mixing legacy codes that run with a 4-byte aligned stack
 with modern codes that keep a 16-byte stack for SSE compatibility.
 </p>
 </dd>
 <dt><code>renesas</code></dt>
 <dd><a name="index-renesas-attribute"></a>
 <p>On SH targets this attribute specifies that the function or struct follows the
 Renesas ABI.
 </p>
 </dd>
 <dt><code>resbank</code></dt>
 <dd><a name="index-resbank-attribute"></a>
 <p>On the SH2A target, this attribute enables the high-speed register
 saving and restoration using a register bank for <code>interrupt_handler</code>
 routines.  Saving to the bank is performed automatically after the CPU
 accepts an interrupt that uses a register bank.
 </p>
 <p>The nineteen 32-bit registers comprising general register R0 to R14,
 control register GBR, and system registers MACH, MACL, and PR and the
 vector table address offset are saved into a register bank.  Register
 banks are stacked in first-in last-out (FILO) sequence.  Restoration
 from the bank is executed by issuing a RESBANK instruction.
 </p>
 </dd>
 <dt><code>returns_twice</code></dt>
 <dd><a name="index-returns_005ftwice-attribute"></a>
 <p>The <code>returns_twice</code> attribute tells the compiler that a function may
 return more than one time.  The compiler ensures that all registers
 are dead before calling such a function and emits a warning about
 the variables that may be clobbered after the second return from the
 function.  Examples of such functions are <code>setjmp</code> and <code>vfork</code>.
 The <code>longjmp</code>-like counterpart of such function, if any, might need
 to be marked with the <code>noreturn</code> attribute.
 </p>
 </dd>
 <dt><code>saveall</code></dt>
 <dd><a name="index-save-all-registers-on-the-Blackfin_002c-H8_002f300_002c-H8_002f300H_002c-and-H8S"></a>
 <p>Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
 all registers except the stack pointer should be saved in the prologue
 regardless of whether they are used or not.
 </p>
 </dd>
 <dt><code>save_volatiles</code></dt>
 <dd><a name="index-save-volatile-registers-on-the-MicroBlaze"></a>
 <p>Use this attribute on the MicroBlaze to indicate that the function is
 an interrupt handler.  All volatile registers (in addition to non-volatile
 registers) are saved in the function prologue.  If the function is a leaf
 function, only volatiles used by the function are saved.  A normal function
 return is generated instead of a return from interrupt.
 </p>
 </dd>
 <dt><code>section (&quot;<var>section-name</var>&quot;)</code></dt>
 <dd><a name="index-section-function-attribute"></a>
 <p>Normally, the compiler places the code it generates in the <code>text</code> section.
 Sometimes, however, you need additional sections, or you need certain
 particular functions to appear in special sections.  The <code>section</code>
 attribute specifies that a function lives in a particular section.
 For example, the declaration:
 </p>
 <div class="smallexample">
 <pre class="smallexample">extern void foobar (void) __attribute__ ((section (&quot;bar&quot;)));
 </pre></div>

 <p>puts the function <code>foobar</code> in the <code>bar</code> section.
 </p>
 <p>Some file formats do not support arbitrary sections so the <code>section</code>
 attribute is not available on all platforms.
 If you need to map the entire contents of a module to a particular
 section, consider using the facilities of the linker instead.
 </p>
 </dd>
 <dt><code>sentinel</code></dt>
 <dd><a name="index-sentinel-function-attribute"></a>
 <p>This function attribute ensures that a parameter in a function call is
 an explicit <code>NULL</code>.  The attribute is only valid on variadic
 functions.  By default, the sentinel is located at position zero, the
 last parameter of the function call.  If an optional integer position
 argument P is supplied to the attribute, the sentinel must be located at
 position P counting backwards from the end of the argument list.
 </p>
 <div class="smallexample">
 <pre class="smallexample">__attribute__ ((sentinel))
 is equivalent to
 __attribute__ ((sentinel(0)))
 </pre></div>

 <p>The attribute is automatically set with a position of 0 for the built-in
 functions <code>execl</code> and <code>execlp</code>.  The built-in function
 <code>execle</code> has the attribute set with a position of 1.
 </p>
 <p>A valid <code>NULL</code> in this context is defined as zero with any pointer
 type.  If your system defines the <code>NULL</code> macro with an integer type
 then you need to add an explicit cast.  GCC replaces <code>stddef.h</code>
 with a copy that redefines NULL appropriately.
 </p>
 <p>The warnings for missing or incorrect sentinels are enabled with
 <samp>-Wformat</samp>.
 </p>
 </dd>
 <dt><code>short_call</code></dt>
 <dd><p>See <code>long_call/short_call</code>.
 </p>
 </dd>
 <dt><code>shortcall</code></dt>
 <dd><p>See <code>longcall/shortcall</code>.
 </p>
 </dd>
 <dt><code>signal</code></dt>
 <dd><a name="index-interrupt-handler-functions-on-the-AVR-processors"></a>
 <p>Use this attribute on the AVR to indicate that the specified
 function is an interrupt handler.  The compiler generates function
 entry and exit sequences suitable for use in an interrupt handler when this
 attribute is present.
 </p>
 <p>See also the <code>interrupt</code> function attribute.
 </p>
 <p>The AVR hardware globally disables interrupts when an interrupt is executed.
 Interrupt handler functions defined with the <code>signal</code> attribute
 do not re-enable interrupts.  It is save to enable interrupts in a
 <code>signal</code> handler.  This &ldquo;save&rdquo; only applies to the code
 generated by the compiler and not to the IRQ layout of the
 application which is responsibility of the application.
 </p>
 <p>If both <code>signal</code> and <code>interrupt</code> are specified for the same
 function, <code>signal</code> is silently ignored.
 </p>
 </dd>
 <dt><code>sp_switch</code></dt>
 <dd><a name="index-sp_005fswitch-attribute"></a>
 <p>Use this attribute on the SH to indicate an <code>interrupt_handler</code>
 function should switch to an alternate stack.  It expects a string
 argument that names a global variable holding the address of the
 alternate stack.
 </p>
 <div class="smallexample">
 <pre class="smallexample">void *alt_stack;
 void f () __attribute__ ((interrupt_handler,
                           sp_switch (&quot;alt_stack&quot;)));
 </pre></div>

 </dd>
 <dt><code>stdcall</code></dt>
 <dd><a name="index-functions-that-pop-the-argument-stack-on-the-386-3"></a>
 <p>On the Intel 386, the <code>stdcall</code> attribute causes the compiler to
 assume that the called function pops off the stack space used to
 pass arguments, unless it takes a variable number of arguments.
 </p>
 </dd>
 <dt><code>syscall_linkage</code></dt>
 <dd><a name="index-syscall_005flinkage-attribute"></a>
 <p>This attribute is used to modify the IA-64 calling convention by marking
 all input registers as live at all function exits.  This makes it possible
 to restart a system call after an interrupt without having to save/restore
 the input registers.  This also prevents kernel data from leaking into
 application code.
 </p>
 </dd>
 <dt><code>target</code></dt>
 <dd><a name="index-target-function-attribute"></a>
 <p>The <code>target</code> attribute is used to specify that a function is to
 be compiled with different target options than specified on the
 command line.  This can be used for instance to have functions
 compiled with a different ISA (instruction set architecture) than the
 default.  You can also use the &lsquo;<samp>#pragma GCC target</samp>&rsquo; pragma to set
 more than one function to be compiled with specific target options.
 See <a href="Function-Specific-Option-Pragmas.html#Function-Specific-Option-Pragmas">Function Specific Option Pragmas</a>, for details about the
 &lsquo;<samp>#pragma GCC target</samp>&rsquo; pragma.
 </p>
 <p>For instance on a 386, you could compile one function with
 <code>target(&quot;sse4.1,arch=core2&quot;)</code> and another with
 <code>target(&quot;sse4a,arch=amdfam10&quot;)</code>.  This is equivalent to
 compiling the first function with <samp>-msse4.1</samp> and
 <samp>-march=core2</samp> options, and the second function with
 <samp>-msse4a</samp> and <samp>-march=amdfam10</samp> options.  It is up to the
 user to make sure that a function is only invoked on a machine that
 supports the particular ISA it is compiled for (for example by using
 <code>cpuid</code> on 386 to determine what feature bits and architecture
 family are used).
 </p>
 <div class="smallexample">
 <pre class="smallexample">int core2_func (void) __attribute__ ((__target__ (&quot;arch=core2&quot;)));
 int sse3_func (void) __attribute__ ((__target__ (&quot;sse3&quot;)));
 </pre></div>

 <p>You can either use multiple
 strings to specify multiple options, or separate the options
 with a comma (&lsquo;<samp>,</samp>&rsquo;).
 </p>
 <p>The <code>target</code> attribute is presently implemented for
 i386/x86_64, PowerPC, and Nios II targets only.
 The options supported are specific to each target.
 </p>
 <p>On the 386, the following options are allowed:
 </p>
 <dl compact="compact">
 <dt>&lsquo;<samp>abm</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-abm</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022abm_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the advanced bit instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>aes</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-aes</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022aes_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the AES instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>default</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022default_0022_0029-attribute"></a>
 <p>See <a href="Function-Multiversioning.html#Function-Multiversioning">Function Multiversioning</a>, where it is used to specify the
 default function version.
 </p>
 </dd>
 <dt>&lsquo;<samp>mmx</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-mmx</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022mmx_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the MMX instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>pclmul</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-pclmul</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022pclmul_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the PCLMUL instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>popcnt</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-popcnt</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022popcnt_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the POPCNT instruction.
 </p>
 </dd>
 <dt>&lsquo;<samp>sse</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-sse</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022sse_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the SSE instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>sse2</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-sse2</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022sse2_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the SSE2 instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>sse3</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-sse3</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022sse3_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the SSE3 instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>sse4</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-sse4</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022sse4_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the SSE4 instructions (both SSE4.1
 and SSE4.2).
 </p>
 </dd>
 <dt>&lsquo;<samp>sse4.1</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-sse4.1</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022sse4_002e1_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the sse4.1 instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>sse4.2</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-sse4.2</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022sse4_002e2_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the sse4.2 instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>sse4a</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-sse4a</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022sse4a_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the SSE4A instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>fma4</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-fma4</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022fma4_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the FMA4 instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>xop</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-xop</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022xop_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the XOP instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>lwp</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-lwp</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022lwp_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the LWP instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>ssse3</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-ssse3</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022ssse3_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the SSSE3 instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>cld</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-cld</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022cld_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the CLD before string moves.
 </p>
 </dd>
 <dt>&lsquo;<samp>fancy-math-387</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-fancy-math-387</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022fancy_002dmath_002d387_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the <code>sin</code>, <code>cos</code>, and
 <code>sqrt</code> instructions on the 387 floating-point unit.
 </p>
 </dd>
 <dt>&lsquo;<samp>fused-madd</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-fused-madd</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022fused_002dmadd_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the fused multiply/add instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>ieee-fp</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-ieee-fp</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022ieee_002dfp_0022_0029-attribute"></a>
 <p>Enable/disable the generation of floating point that depends on IEEE arithmetic.
 </p>
 </dd>
 <dt>&lsquo;<samp>inline-all-stringops</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-inline-all-stringops</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022inline_002dall_002dstringops_0022_0029-attribute"></a>
 <p>Enable/disable inlining of string operations.
 </p>
 </dd>
 <dt>&lsquo;<samp>inline-stringops-dynamically</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-inline-stringops-dynamically</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022inline_002dstringops_002ddynamically_0022_0029-attribute"></a>
 <p>Enable/disable the generation of the inline code to do small string
 operations and calling the library routines for large operations.
 </p>
 </dd>
 <dt>&lsquo;<samp>align-stringops</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-align-stringops</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022align_002dstringops_0022_0029-attribute"></a>
 <p>Do/do not align destination of inlined string operations.
 </p>
 </dd>
 <dt>&lsquo;<samp>recip</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-recip</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022recip_0022_0029-attribute"></a>
 <p>Enable/disable the generation of RCPSS, RCPPS, RSQRTSS and RSQRTPS
 instructions followed an additional Newton-Raphson step instead of
 doing a floating-point division.
 </p>
 </dd>
 <dt>&lsquo;<samp>arch=<var>ARCH</var></samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022arch_003dARCH_0022_0029-attribute"></a>
 <p>Specify the architecture to generate code for in compiling the function.
 </p>
 </dd>
 <dt>&lsquo;<samp>tune=<var>TUNE</var></samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022tune_003dTUNE_0022_0029-attribute"></a>
 <p>Specify the architecture to tune for in compiling the function.
 </p>
 </dd>
 <dt>&lsquo;<samp>fpmath=<var>FPMATH</var></samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022fpmath_003dFPMATH_0022_0029-attribute"></a>
 <p>Specify which floating-point unit to use.  The
 <code>target(&quot;fpmath=sse,387&quot;)</code> option must be specified as
 <code>target(&quot;fpmath=sse+387&quot;)</code> because the comma would separate
 different options.
 </p></dd>
 </dl>

 <p>On the PowerPC, the following options are allowed:
 </p>
 <dl compact="compact">
 <dt>&lsquo;<samp>altivec</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-altivec</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022altivec_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) AltiVec instructions.  In
 32-bit code, you cannot enable AltiVec instructions unless
 <samp>-mabi=altivec</samp> is used on the command line.
 </p>
 </dd>
 <dt>&lsquo;<samp>cmpb</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-cmpb</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022cmpb_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the compare bytes instruction
 implemented on the POWER6 processor and other processors that support
 the PowerPC V2.05 architecture.
 </p>
 </dd>
 <dt>&lsquo;<samp>dlmzb</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-dlmzb</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022dlmzb_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the string-search &lsquo;<samp>dlmzb</samp>&rsquo;
 instruction on the IBM 405, 440, 464 and 476 processors.  This instruction is
 generated by default when targeting those processors.
 </p>
 </dd>
 <dt>&lsquo;<samp>fprnd</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-fprnd</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022fprnd_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the FP round to integer
 instructions implemented on the POWER5+ processor and other processors
 that support the PowerPC V2.03 architecture.
 </p>
 </dd>
 <dt>&lsquo;<samp>hard-dfp</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-hard-dfp</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022hard_002ddfp_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the decimal floating-point
 instructions implemented on some POWER processors.
 </p>
 </dd>
 <dt>&lsquo;<samp>isel</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-isel</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022isel_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) ISEL instruction.
 </p>
 </dd>
 <dt>&lsquo;<samp>mfcrf</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-mfcrf</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022mfcrf_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the move from condition
 register field instruction implemented on the POWER4 processor and
 other processors that support the PowerPC V2.01 architecture.
 </p>
 </dd>
 <dt>&lsquo;<samp>mfpgpr</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-mfpgpr</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022mfpgpr_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the FP move to/from general
 purpose register instructions implemented on the POWER6X processor and
 other processors that support the extended PowerPC V2.05 architecture.
 </p>
 </dd>
 <dt>&lsquo;<samp>mulhw</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-mulhw</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022mulhw_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the half-word multiply and
 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
 These instructions are generated by default when targeting those
 processors.
 </p>
 </dd>
 <dt>&lsquo;<samp>multiple</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-multiple</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022multiple_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the load multiple word
 instructions and the store multiple word instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>update</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-update</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022update_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the load or store instructions
 that update the base register to the address of the calculated memory
 location.
 </p>
 </dd>
 <dt>&lsquo;<samp>popcntb</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-popcntb</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022popcntb_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the popcount and double-precision
 FP reciprocal estimate instruction implemented on the POWER5
 processor and other processors that support the PowerPC V2.02
 architecture.
 </p>
 </dd>
 <dt>&lsquo;<samp>popcntd</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-popcntd</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022popcntd_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the popcount instruction
 implemented on the POWER7 processor and other processors that support
 the PowerPC V2.06 architecture.
 </p>
 </dd>
 <dt>&lsquo;<samp>powerpc-gfxopt</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-powerpc-gfxopt</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022powerpc_002dgfxopt_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the optional PowerPC
 architecture instructions in the Graphics group, including
 floating-point select.
 </p>
 </dd>
 <dt>&lsquo;<samp>powerpc-gpopt</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-powerpc-gpopt</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022powerpc_002dgpopt_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the optional PowerPC
 architecture instructions in the General Purpose group, including
 floating-point square root.
 </p>
 </dd>
 <dt>&lsquo;<samp>recip-precision</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-recip-precision</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022recip_002dprecision_0022_0029-attribute"></a>
 <p>Assume (do not assume) that the reciprocal estimate instructions
 provide higher-precision estimates than is mandated by the powerpc
 ABI.
 </p>
 </dd>
 <dt>&lsquo;<samp>string</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-string</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022string_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the load string instructions
 and the store string word instructions to save multiple registers and
 do small block moves.
 </p>
 </dd>
 <dt>&lsquo;<samp>vsx</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-vsx</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022vsx_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) vector/scalar (VSX)
 instructions, and also enable the use of built-in functions that allow
 more direct access to the VSX instruction set.  In 32-bit code, you
 cannot enable VSX or AltiVec instructions unless
 <samp>-mabi=altivec</samp> is used on the command line.
 </p>
 </dd>
 <dt>&lsquo;<samp>friz</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-friz</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022friz_0022_0029-attribute"></a>
 <p>Generate (do not generate) the <code>friz</code> instruction when the
 <samp>-funsafe-math-optimizations</samp> option is used to optimize
 rounding a floating-point value to 64-bit integer and back to floating
 point.  The <code>friz</code> instruction does not return the same value if
 the floating-point number is too large to fit in an integer.
 </p>
 </dd>
 <dt>&lsquo;<samp>avoid-indexed-addresses</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-avoid-indexed-addresses</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022avoid_002dindexed_002daddresses_0022_0029-attribute"></a>
 <p>Generate code that tries to avoid (not avoid) the use of indexed load
 or store instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>paired</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-paired</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022paired_0022_0029-attribute"></a>
 <p>Generate code that uses (does not use) the generation of PAIRED simd
 instructions.
 </p>
 </dd>
 <dt>&lsquo;<samp>longcall</samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-longcall</samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022longcall_0022_0029-attribute"></a>
 <p>Generate code that assumes (does not assume) that all calls are far
 away so that a longer more expensive calling sequence is required.
 </p>
 </dd>
 <dt>&lsquo;<samp>cpu=<var>CPU</var></samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022cpu_003dCPU_0022_0029-attribute"></a>
 <p>Specify the architecture to generate code for when compiling the
 function.  If you select the <code>target(&quot;cpu=power7&quot;)</code> attribute when
 generating 32-bit code, VSX and AltiVec instructions are not generated
 unless you use the <samp>-mabi=altivec</samp> option on the command line.
 </p>
 </dd>
 <dt>&lsquo;<samp>tune=<var>TUNE</var></samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022tune_003dTUNE_0022_0029-attribute-1"></a>
 <p>Specify the architecture to tune for when compiling the function.  If
 you do not specify the <code>target(&quot;tune=<var>TUNE</var>&quot;)</code> attribute and
 you do specify the <code>target(&quot;cpu=<var>CPU</var>&quot;)</code> attribute,
 compilation tunes for the <var>CPU</var> architecture, and not the
 default tuning specified on the command line.
 </p></dd>
 </dl>

 <p>When compiling for Nios II, the following options are allowed:
 </p>
 <dl compact="compact">
 <dt>&lsquo;<samp>custom-<var>insn</var>=<var>N</var></samp>&rsquo;</dt>
 <dt>&lsquo;<samp>no-custom-<var>insn</var></samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022custom_002dinsn_003dN_0022_0029-attribute"></a>
 <a name="index-target_0028_0022no_002dcustom_002dinsn_0022_0029-attribute"></a>
 <p>Each &lsquo;<samp>custom-<var>insn</var>=<var>N</var></samp>&rsquo; attribute locally enables use of a
 custom instruction with encoding <var>N</var> when generating code that uses
 <var>insn</var>.  Similarly, &lsquo;<samp>no-custom-<var>insn</var></samp>&rsquo; locally inhibits use of
 the custom instruction <var>insn</var>.
 These target attributes correspond to the
 <samp>-mcustom-<var>insn</var>=<var>N</var></samp> and <samp>-mno-custom-<var>insn</var></samp>
 command-line options, and support the same set of <var>insn</var> keywords.
 See <a href="Nios-II-Options.html#Nios-II-Options">Nios II Options</a>, for more information.
 </p>
 </dd>
 <dt>&lsquo;<samp>custom-fpu-cfg=<var>name</var></samp>&rsquo;</dt>
 <dd><a name="index-target_0028_0022custom_002dfpu_002dcfg_003dname_0022_0029-attribute"></a>
 <p>This attribute corresponds to the <samp>-mcustom-fpu-cfg=<var>name</var></samp>
 command-line option, to select a predefined set of custom instructions
 named <var>name</var>.
 See <a href="Nios-II-Options.html#Nios-II-Options">Nios II Options</a>, for more information.
 </p></dd>
 </dl>

 <p>On the 386/x86_64 and PowerPC back ends, the inliner does not inline a
 function that has different target options than the caller, unless the
 callee has a subset of the target options of the caller.  For example
 a function declared with <code>target(&quot;sse3&quot;)</code> can inline a function
 with <code>target(&quot;sse2&quot;)</code>, since <code>-msse3</code> implies <code>-msse2</code>.
 </p>
 </dd>
 <dt><code>tiny_data</code></dt>
 <dd><a name="index-tiny-data-section-on-the-H8_002f300H-and-H8S"></a>
 <p>Use this attribute on the H8/300H and H8S to indicate that the specified
 variable should be placed into the tiny data section.
 The compiler generates more efficient code for loads and stores
 on data in the tiny data section.  Note the tiny data area is limited to
 slightly under 32KB of data.
 </p>
 </dd>
 <dt><code>trap_exit</code></dt>
 <dd><a name="index-trap_005fexit-attribute"></a>
 <p>Use this attribute on the SH for an <code>interrupt_handler</code> to return using
 <code>trapa</code> instead of <code>rte</code>.  This attribute expects an integer
 argument specifying the trap number to be used.
 </p>
 </dd>
 <dt><code>trapa_handler</code></dt>
 <dd><a name="index-trapa_005fhandler-attribute"></a>
 <p>On SH targets this function attribute is similar to <code>interrupt_handler</code>
 but it does not save and restore all registers.
 </p>
 </dd>
 <dt><code>unused</code></dt>
 <dd><a name="index-unused-attribute_002e"></a>
 <p>This attribute, attached to a function, means that the function is meant
 to be possibly unused.  GCC does not produce a warning for this
 function.
 </p>
 </dd>
 <dt><code>used</code></dt>
 <dd><a name="index-used-attribute_002e"></a>
 <p>This attribute, attached to a function, means that code must be emitted
 for the function even if it appears that the function is not referenced.
 This is useful, for example, when the function is referenced only in
 inline assembly.
 </p>
 <p>When applied to a member function of a C++ class template, the
 attribute also means that the function is instantiated if the
 class itself is instantiated.
 </p>
 </dd>
 <dt><code>version_id</code></dt>
 <dd><a name="index-version_005fid-attribute"></a>
 <p>This IA-64 HP-UX attribute, attached to a global variable or function, renames a
 symbol to contain a version string, thus allowing for function level
 versioning.  HP-UX system header files may use function level versioning
 for some system calls.
 </p>
 <div class="smallexample">
 <pre class="smallexample">extern int foo () __attribute__((version_id (&quot;20040821&quot;)));
 </pre></div>

 <p>Calls to <var>foo</var> are mapped to calls to <var>foo{20040821}</var>.
 </p>
 </dd>
 <dt><code>visibility (&quot;<var>visibility_type</var>&quot;)</code></dt>
 <dd><a name="index-visibility-attribute"></a>
 <p>This attribute affects the linkage of the declaration to which it is attached.
 There are four supported <var>visibility_type</var> values: default,
 hidden, protected or internal visibility.
 </p>
 <div class="smallexample">
 <pre class="smallexample">void __attribute__ ((visibility (&quot;protected&quot;)))
 f () { /* <span class="roman">Do something.</span> */; }
 int i __attribute__ ((visibility (&quot;hidden&quot;)));
 </pre></div>

 <p>The possible values of <var>visibility_type</var> correspond to the
 visibility settings in the ELF gABI.
 </p>
 <dl compact="compact">
 <dt><em>default</em></dt>
 <dd><p>Default visibility is the normal case for the object file format.
 This value is available for the visibility attribute to override other
 options that may change the assumed visibility of entities.
 </p>
 <p>On ELF, default visibility means that the declaration is visible to other
 modules and, in shared libraries, means that the declared entity may be
 overridden.
 </p>
 <p>On Darwin, default visibility means that the declaration is visible to
 other modules.
 </p>
 <p>Default visibility corresponds to &ldquo;external linkage&rdquo; in the language.
 </p>
 </dd>
 <dt><em>hidden</em></dt>
 <dd><p>Hidden visibility indicates that the entity declared has a new
 form of linkage, which we call &ldquo;hidden linkage&rdquo;.  Two
 declarations of an object with hidden linkage refer to the same object
 if they are in the same shared object.
 </p>
 </dd>
 <dt><em>internal</em></dt>
 <dd><p>Internal visibility is like hidden visibility, but with additional
 processor specific semantics.  Unless otherwise specified by the
 psABI, GCC defines internal visibility to mean that a function is
 <em>never</em> called from another module.  Compare this with hidden
 functions which, while they cannot be referenced directly by other
 modules, can be referenced indirectly via function pointers.  By
 indicating that a function cannot be called from outside the module,
 GCC may for instance omit the load of a PIC register since it is known
 that the calling function loaded the correct value.
 </p>
 </dd>
 <dt><em>protected</em></dt>
 <dd><p>Protected visibility is like default visibility except that it
 indicates that references within the defining module bind to the
 definition in that module.  That is, the declared entity cannot be
 overridden by another module.
 </p>
 </dd>
 </dl>

 <p>All visibilities are supported on many, but not all, ELF targets
 (supported when the assembler supports the &lsquo;<samp>.visibility</samp>&rsquo;
 pseudo-op).  Default visibility is supported everywhere.  Hidden
 visibility is supported on Darwin targets.
 </p>
 <p>The visibility attribute should be applied only to declarations that
 would otherwise have external linkage.  The attribute should be applied
 consistently, so that the same entity should not be declared with
 different settings of the attribute.
 </p>
 <p>In C++, the visibility attribute applies to types as well as functions
 and objects, because in C++ types have linkage.  A class must not have
 greater visibility than its non-static data member types and bases,
 and class members default to the visibility of their class.  Also, a
 declaration without explicit visibility is limited to the visibility
 of its type.
 </p>
 <p>In C++, you can mark member functions and static member variables of a
 class with the visibility attribute.  This is useful if you know a
 particular method or static member variable should only be used from
 one shared object; then you can mark it hidden while the rest of the
 class has default visibility.  Care must be taken to avoid breaking
 the One Definition Rule; for example, it is usually not useful to mark
 an inline method as hidden without marking the whole class as hidden.
 </p>
 <p>A C++ namespace declaration can also have the visibility attribute.
 </p>
 <div class="smallexample">
 <pre class="smallexample">namespace nspace1 __attribute__ ((visibility (&quot;protected&quot;)))
 { /* <span class="roman">Do something.</span> */; }
 </pre></div>

 <p>This attribute applies only to the particular namespace body, not to
 other definitions of the same namespace; it is equivalent to using
 &lsquo;<samp>#pragma GCC visibility</samp>&rsquo; before and after the namespace
 definition (see <a href="Visibility-Pragmas.html#Visibility-Pragmas">Visibility Pragmas</a>).
 </p>
 <p>In C++, if a template argument has limited visibility, this
 restriction is implicitly propagated to the template instantiation.
 Otherwise, template instantiations and specializations default to the
 visibility of their template.
 </p>
 <p>If both the template and enclosing class have explicit visibility, the
 visibility from the template is used.
 </p>
 </dd>
 <dt><code>vliw</code></dt>
 <dd><a name="index-vliw-attribute"></a>
 <p>On MeP, the <code>vliw</code> attribute tells the compiler to emit
 instructions in VLIW mode instead of core mode.  Note that this
 attribute is not allowed unless a VLIW coprocessor has been configured
 and enabled through command-line options.
 </p>
 </dd>
 <dt><code>warn_unused_result</code></dt>
 <dd><a name="index-warn_005funused_005fresult-attribute"></a>
 <p>The <code>warn_unused_result</code> attribute causes a warning to be emitted
 if a caller of the function with this attribute does not use its
 return value.  This is useful for functions where not checking
 the result is either a security problem or always a bug, such as
 <code>realloc</code>.
 </p>
 <div class="smallexample">
 <pre class="smallexample">int fn () __attribute__ ((warn_unused_result));
 int foo ()
 {
   if (fn () &lt; 0) return -1;
   fn ();
   return 0;
 }
 </pre></div>

 <p>results in warning on line 5.
 </p>
 </dd>
 <dt><code>weak</code></dt>
 <dd><a name="index-weak-attribute"></a>
 <p>The <code>weak</code> attribute causes the declaration to be emitted as a weak
 symbol rather than a global.  This is primarily useful in defining
 library functions that can be overridden in user code, though it can
 also be used with non-function declarations.  Weak symbols are supported
 for ELF targets, and also for a.out targets when using the GNU assembler
 and linker.
 </p>
 </dd>
 <dt><code>weakref</code></dt>
 <dt><code>weakref (&quot;<var>target</var>&quot;)</code></dt>
 <dd><a name="index-weakref-attribute"></a>
 <p>The <code>weakref</code> attribute marks a declaration as a weak reference.
 Without arguments, it should be accompanied by an <code>alias</code> attribute
 naming the target symbol.  Optionally, the <var>target</var> may be given as
 an argument to <code>weakref</code> itself.  In either case, <code>weakref</code>
 implicitly marks the declaration as <code>weak</code>.  Without a
 <var>target</var>, given as an argument to <code>weakref</code> or to <code>alias</code>,
 <code>weakref</code> is equivalent to <code>weak</code>.
 </p>
 <div class="smallexample">
 <pre class="smallexample">static int x() __attribute__ ((weakref (&quot;y&quot;)));
 /* is equivalent to... */
 static int x() __attribute__ ((weak, weakref, alias (&quot;y&quot;)));
 /* and to... */
 static int x() __attribute__ ((weakref));
 static int x() __attribute__ ((alias (&quot;y&quot;)));
 </pre></div>

 <p>A weak reference is an alias that does not by itself require a
 definition to be given for the target symbol.  If the target symbol is
 only referenced through weak references, then it becomes a <code>weak</code>
 undefined symbol.  If it is directly referenced, however, then such
 strong references prevail, and a definition is required for the
 symbol, not necessarily in the same translation unit.
 </p>
 <p>The effect is equivalent to moving all references to the alias to a
 separate translation unit, renaming the alias to the aliased symbol,
 declaring it as weak, compiling the two separate translation units and
 performing a reloadable link on them.
 </p>
 <p>At present, a declaration to which <code>weakref</code> is attached can
 only be <code>static</code>.
 </p>
 </dd>
 </dl>

 <p>You can specify multiple attributes in a declaration by separating them
 by commas within the double parentheses or by immediately following an
 attribute declaration with another attribute declaration.
 </p>
 <a name="index-_0023pragma_002c-reason-for-not-using"></a>
 <a name="index-pragma_002c-reason-for-not-using"></a>
 <p>Some people object to the <code>__attribute__</code> feature, suggesting that
 ISO C&rsquo;s <code>#pragma</code> should be used instead.  At the time
 <code>__attribute__</code> was designed, there were two reasons for not doing
 this.
 </p>
 <ol>
 <li> It is impossible to generate <code>#pragma</code> commands from a macro.

 </li><li> There is no telling what the same <code>#pragma</code> might mean in another
 compiler.
 </li></ol>

 <p>These two reasons applied to almost any application that might have been
 proposed for <code>#pragma</code>.  It was basically a mistake to use
 <code>#pragma</code> for <em>anything</em>.
 </p>
 <p>The ISO C99 standard includes <code>_Pragma</code>, which now allows pragmas
 to be generated from macros.  In addition, a <code>#pragma GCC</code>
 namespace is now in use for GCC-specific pragmas.  However, it has been
 found convenient to use <code>__attribute__</code> to achieve a natural
 attachment of attributes to their corresponding declarations, whereas
 <code>#pragma GCC</code> is of use for constructs that do not naturally form
 part of the grammar.  See <a href="Pragmas.html#Pragmas">Pragmas Accepted by GCC</a>.
 </p>
 <hr>
 <div class="header">
 <p>
 Next: <a href="Attribute-Syntax.html#Attribute-Syntax" accesskey="n" rel="next">Attribute Syntax</a>, Previous: <a href="Mixed-Declarations.html#Mixed-Declarations" accesskey="p" rel="prev">Mixed Declarations</a>, Up: <a href="C-Extensions.html#C-Extensions" accesskey="u" rel="up">C Extensions</a> &nbsp; [<a href="index.html#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="Option-Index.html#Option-Index" title="Index" rel="index">Index</a>]</p>
 </div>


 </body>
 </html>