1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1996, 1998, 1999, 2000, 2001,
2 @c 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010
3 @c Free Software Foundation, Inc.
5 @c This is part of the GCC manual.
6 @c For copying conditions, see the file gcc.texi.
9 @chapter Extensions to the C Language Family
10 @cindex extensions, C language
11 @cindex C language extensions
14 GNU C provides several language features not found in ISO standard C@.
15 (The @option{-pedantic} option directs GCC to print a warning message if
16 any of these features is used.) To test for the availability of these
17 features in conditional compilation, check for a predefined macro
18 @code{__GNUC__}, which is always defined under GCC@.
20 These extensions are available in C and Objective-C@. Most of them are
21 also available in C++. @xref{C++ Extensions,,Extensions to the
22 C++ Language}, for extensions that apply @emph{only} to C++.
24 Some features that are in ISO C99 but not C90 or C++ are also, as
25 extensions, accepted by GCC in C90 mode and in C++.
28 * Statement Exprs:: Putting statements and declarations inside expressions.
29 * Local Labels:: Labels local to a block.
30 * Labels as Values:: Getting pointers to labels, and computed gotos.
31 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
32 * Constructing Calls:: Dispatching a call to another function.
33 * Typeof:: @code{typeof}: referring to the type of an expression.
34 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
35 * Long Long:: Double-word integers---@code{long long int}.
36 * __int128:: 128-bit integers---@code{__int128}.
37 * Complex:: Data types for complex numbers.
38 * Floating Types:: Additional Floating Types.
39 * Half-Precision:: Half-Precision Floating Point.
40 * Decimal Float:: Decimal Floating Types.
41 * Hex Floats:: Hexadecimal floating-point constants.
42 * Fixed-Point:: Fixed-Point Types.
43 * Named Address Spaces::Named address spaces.
44 * Zero Length:: Zero-length arrays.
45 * Variable Length:: Arrays whose length is computed at run time.
46 * Empty Structures:: Structures with no members.
47 * Variadic Macros:: Macros with a variable number of arguments.
48 * Escaped Newlines:: Slightly looser rules for escaped newlines.
49 * Subscripting:: Any array can be subscripted, even if not an lvalue.
50 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
51 * Initializers:: Non-constant initializers.
52 * Compound Literals:: Compound literals give structures, unions
54 * Designated Inits:: Labeling elements of initializers.
55 * Cast to Union:: Casting to union type from any member of the union.
56 * Case Ranges:: `case 1 ... 9' and such.
57 * Mixed Declarations:: Mixing declarations and code.
58 * Function Attributes:: Declaring that functions have no side effects,
59 or that they can never return.
60 * Attribute Syntax:: Formal syntax for attributes.
61 * Function Prototypes:: Prototype declarations and old-style definitions.
62 * C++ Comments:: C++ comments are recognized.
63 * Dollar Signs:: Dollar sign is allowed in identifiers.
64 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
65 * Variable Attributes:: Specifying attributes of variables.
66 * Type Attributes:: Specifying attributes of types.
67 * Alignment:: Inquiring about the alignment of a type or variable.
68 * Inline:: Defining inline functions (as fast as macros).
69 * Volatiles:: What constitutes an access to a volatile object.
70 * Extended Asm:: Assembler instructions with C expressions as operands.
71 (With them you can define ``built-in'' functions.)
72 * Constraints:: Constraints for asm operands
73 * Asm Labels:: Specifying the assembler name to use for a C symbol.
74 * Explicit Reg Vars:: Defining variables residing in specified registers.
75 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
76 * Incomplete Enums:: @code{enum foo;}, with details to follow.
77 * Function Names:: Printable strings which are the name of the current
79 * Return Address:: Getting the return or frame address of a function.
80 * Vector Extensions:: Using vector instructions through built-in functions.
81 * Offsetof:: Special syntax for implementing @code{offsetof}.
82 * Atomic Builtins:: Built-in functions for atomic memory access.
83 * Object Size Checking:: Built-in functions for limited buffer overflow
85 * Other Builtins:: Other built-in functions.
86 * Target Builtins:: Built-in functions specific to particular targets.
87 * Target Format Checks:: Format checks specific to particular targets.
88 * Pragmas:: Pragmas accepted by GCC.
89 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
90 * Thread-Local:: Per-thread variables.
91 * Binary constants:: Binary constants using the @samp{0b} prefix.
95 @section Statements and Declarations in Expressions
96 @cindex statements inside expressions
97 @cindex declarations inside expressions
98 @cindex expressions containing statements
99 @cindex macros, statements in expressions
101 @c the above section title wrapped and causes an underfull hbox.. i
102 @c changed it from "within" to "in". --mew 4feb93
103 A compound statement enclosed in parentheses may appear as an expression
104 in GNU C@. This allows you to use loops, switches, and local variables
105 within an expression.
107 Recall that a compound statement is a sequence of statements surrounded
108 by braces; in this construct, parentheses go around the braces. For
112 (@{ int y = foo (); int z;
119 is a valid (though slightly more complex than necessary) expression
120 for the absolute value of @code{foo ()}.
122 The last thing in the compound statement should be an expression
123 followed by a semicolon; the value of this subexpression serves as the
124 value of the entire construct. (If you use some other kind of statement
125 last within the braces, the construct has type @code{void}, and thus
126 effectively no value.)
128 This feature is especially useful in making macro definitions ``safe'' (so
129 that they evaluate each operand exactly once). For example, the
130 ``maximum'' function is commonly defined as a macro in standard C as
134 #define max(a,b) ((a) > (b) ? (a) : (b))
138 @cindex side effects, macro argument
139 But this definition computes either @var{a} or @var{b} twice, with bad
140 results if the operand has side effects. In GNU C, if you know the
141 type of the operands (here taken as @code{int}), you can define
142 the macro safely as follows:
145 #define maxint(a,b) \
146 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
149 Embedded statements are not allowed in constant expressions, such as
150 the value of an enumeration constant, the width of a bit-field, or
151 the initial value of a static variable.
153 If you don't know the type of the operand, you can still do this, but you
154 must use @code{typeof} (@pxref{Typeof}).
156 In G++, the result value of a statement expression undergoes array and
157 function pointer decay, and is returned by value to the enclosing
158 expression. For instance, if @code{A} is a class, then
167 will construct a temporary @code{A} object to hold the result of the
168 statement expression, and that will be used to invoke @code{Foo}.
169 Therefore the @code{this} pointer observed by @code{Foo} will not be the
172 Any temporaries created within a statement within a statement expression
173 will be destroyed at the statement's end. This makes statement
174 expressions inside macros slightly different from function calls. In
175 the latter case temporaries introduced during argument evaluation will
176 be destroyed at the end of the statement that includes the function
177 call. In the statement expression case they will be destroyed during
178 the statement expression. For instance,
181 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
182 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
192 will have different places where temporaries are destroyed. For the
193 @code{macro} case, the temporary @code{X} will be destroyed just after
194 the initialization of @code{b}. In the @code{function} case that
195 temporary will be destroyed when the function returns.
197 These considerations mean that it is probably a bad idea to use
198 statement-expressions of this form in header files that are designed to
199 work with C++. (Note that some versions of the GNU C Library contained
200 header files using statement-expression that lead to precisely this
203 Jumping into a statement expression with @code{goto} or using a
204 @code{switch} statement outside the statement expression with a
205 @code{case} or @code{default} label inside the statement expression is
206 not permitted. Jumping into a statement expression with a computed
207 @code{goto} (@pxref{Labels as Values}) yields undefined behavior.
208 Jumping out of a statement expression is permitted, but if the
209 statement expression is part of a larger expression then it is
210 unspecified which other subexpressions of that expression have been
211 evaluated except where the language definition requires certain
212 subexpressions to be evaluated before or after the statement
213 expression. In any case, as with a function call the evaluation of a
214 statement expression is not interleaved with the evaluation of other
215 parts of the containing expression. For example,
218 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
222 will call @code{foo} and @code{bar1} and will not call @code{baz} but
223 may or may not call @code{bar2}. If @code{bar2} is called, it will be
224 called after @code{foo} and before @code{bar1}
227 @section Locally Declared Labels
229 @cindex macros, local labels
231 GCC allows you to declare @dfn{local labels} in any nested block
232 scope. A local label is just like an ordinary label, but you can
233 only reference it (with a @code{goto} statement, or by taking its
234 address) within the block in which it was declared.
236 A local label declaration looks like this:
239 __label__ @var{label};
246 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
249 Local label declarations must come at the beginning of the block,
250 before any ordinary declarations or statements.
252 The label declaration defines the label @emph{name}, but does not define
253 the label itself. You must do this in the usual way, with
254 @code{@var{label}:}, within the statements of the statement expression.
256 The local label feature is useful for complex macros. If a macro
257 contains nested loops, a @code{goto} can be useful for breaking out of
258 them. However, an ordinary label whose scope is the whole function
259 cannot be used: if the macro can be expanded several times in one
260 function, the label will be multiply defined in that function. A
261 local label avoids this problem. For example:
264 #define SEARCH(value, array, target) \
267 typeof (target) _SEARCH_target = (target); \
268 typeof (*(array)) *_SEARCH_array = (array); \
271 for (i = 0; i < max; i++) \
272 for (j = 0; j < max; j++) \
273 if (_SEARCH_array[i][j] == _SEARCH_target) \
274 @{ (value) = i; goto found; @} \
280 This could also be written using a statement-expression:
283 #define SEARCH(array, target) \
286 typeof (target) _SEARCH_target = (target); \
287 typeof (*(array)) *_SEARCH_array = (array); \
290 for (i = 0; i < max; i++) \
291 for (j = 0; j < max; j++) \
292 if (_SEARCH_array[i][j] == _SEARCH_target) \
293 @{ value = i; goto found; @} \
300 Local label declarations also make the labels they declare visible to
301 nested functions, if there are any. @xref{Nested Functions}, for details.
303 @node Labels as Values
304 @section Labels as Values
305 @cindex labels as values
306 @cindex computed gotos
307 @cindex goto with computed label
308 @cindex address of a label
310 You can get the address of a label defined in the current function
311 (or a containing function) with the unary operator @samp{&&}. The
312 value has type @code{void *}. This value is a constant and can be used
313 wherever a constant of that type is valid. For example:
321 To use these values, you need to be able to jump to one. This is done
322 with the computed goto statement@footnote{The analogous feature in
323 Fortran is called an assigned goto, but that name seems inappropriate in
324 C, where one can do more than simply store label addresses in label
325 variables.}, @code{goto *@var{exp};}. For example,
332 Any expression of type @code{void *} is allowed.
334 One way of using these constants is in initializing a static array that
335 will serve as a jump table:
338 static void *array[] = @{ &&foo, &&bar, &&hack @};
341 Then you can select a label with indexing, like this:
348 Note that this does not check whether the subscript is in bounds---array
349 indexing in C never does that.
351 Such an array of label values serves a purpose much like that of the
352 @code{switch} statement. The @code{switch} statement is cleaner, so
353 use that rather than an array unless the problem does not fit a
354 @code{switch} statement very well.
356 Another use of label values is in an interpreter for threaded code.
357 The labels within the interpreter function can be stored in the
358 threaded code for super-fast dispatching.
360 You may not use this mechanism to jump to code in a different function.
361 If you do that, totally unpredictable things will happen. The best way to
362 avoid this is to store the label address only in automatic variables and
363 never pass it as an argument.
365 An alternate way to write the above example is
368 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
370 goto *(&&foo + array[i]);
374 This is more friendly to code living in shared libraries, as it reduces
375 the number of dynamic relocations that are needed, and by consequence,
376 allows the data to be read-only.
378 The @code{&&foo} expressions for the same label might have different
379 values if the containing function is inlined or cloned. If a program
380 relies on them being always the same,
381 @code{__attribute__((__noinline__,__noclone__))} should be used to
382 prevent inlining and cloning. If @code{&&foo} is used in a static
383 variable initializer, inlining and cloning is forbidden.
385 @node Nested Functions
386 @section Nested Functions
387 @cindex nested functions
388 @cindex downward funargs
391 A @dfn{nested function} is a function defined inside another function.
392 (Nested functions are not supported for GNU C++.) The nested function's
393 name is local to the block where it is defined. For example, here we
394 define a nested function named @code{square}, and call it twice:
398 foo (double a, double b)
400 double square (double z) @{ return z * z; @}
402 return square (a) + square (b);
407 The nested function can access all the variables of the containing
408 function that are visible at the point of its definition. This is
409 called @dfn{lexical scoping}. For example, here we show a nested
410 function which uses an inherited variable named @code{offset}:
414 bar (int *array, int offset, int size)
416 int access (int *array, int index)
417 @{ return array[index + offset]; @}
420 for (i = 0; i < size; i++)
421 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
426 Nested function definitions are permitted within functions in the places
427 where variable definitions are allowed; that is, in any block, mixed
428 with the other declarations and statements in the block.
430 It is possible to call the nested function from outside the scope of its
431 name by storing its address or passing the address to another function:
434 hack (int *array, int size)
436 void store (int index, int value)
437 @{ array[index] = value; @}
439 intermediate (store, size);
443 Here, the function @code{intermediate} receives the address of
444 @code{store} as an argument. If @code{intermediate} calls @code{store},
445 the arguments given to @code{store} are used to store into @code{array}.
446 But this technique works only so long as the containing function
447 (@code{hack}, in this example) does not exit.
449 If you try to call the nested function through its address after the
450 containing function has exited, all hell will break loose. If you try
451 to call it after a containing scope level has exited, and if it refers
452 to some of the variables that are no longer in scope, you may be lucky,
453 but it's not wise to take the risk. If, however, the nested function
454 does not refer to anything that has gone out of scope, you should be
457 GCC implements taking the address of a nested function using a technique
458 called @dfn{trampolines}. This technique was described in
459 @cite{Lexical Closures for C++} (Thomas M. Breuel, USENIX
460 C++ Conference Proceedings, October 17-21, 1988).
462 A nested function can jump to a label inherited from a containing
463 function, provided the label was explicitly declared in the containing
464 function (@pxref{Local Labels}). Such a jump returns instantly to the
465 containing function, exiting the nested function which did the
466 @code{goto} and any intermediate functions as well. Here is an example:
470 bar (int *array, int offset, int size)
473 int access (int *array, int index)
477 return array[index + offset];
481 for (i = 0; i < size; i++)
482 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
486 /* @r{Control comes here from @code{access}
487 if it detects an error.} */
494 A nested function always has no linkage. Declaring one with
495 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
496 before its definition, use @code{auto} (which is otherwise meaningless
497 for function declarations).
500 bar (int *array, int offset, int size)
503 auto int access (int *, int);
505 int access (int *array, int index)
509 return array[index + offset];
515 @node Constructing Calls
516 @section Constructing Function Calls
517 @cindex constructing calls
518 @cindex forwarding calls
520 Using the built-in functions described below, you can record
521 the arguments a function received, and call another function
522 with the same arguments, without knowing the number or types
525 You can also record the return value of that function call,
526 and later return that value, without knowing what data type
527 the function tried to return (as long as your caller expects
530 However, these built-in functions may interact badly with some
531 sophisticated features or other extensions of the language. It
532 is, therefore, not recommended to use them outside very simple
533 functions acting as mere forwarders for their arguments.
535 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
536 This built-in function returns a pointer to data
537 describing how to perform a call with the same arguments as were passed
538 to the current function.
540 The function saves the arg pointer register, structure value address,
541 and all registers that might be used to pass arguments to a function
542 into a block of memory allocated on the stack. Then it returns the
543 address of that block.
546 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
547 This built-in function invokes @var{function}
548 with a copy of the parameters described by @var{arguments}
551 The value of @var{arguments} should be the value returned by
552 @code{__builtin_apply_args}. The argument @var{size} specifies the size
553 of the stack argument data, in bytes.
555 This function returns a pointer to data describing
556 how to return whatever value was returned by @var{function}. The data
557 is saved in a block of memory allocated on the stack.
559 It is not always simple to compute the proper value for @var{size}. The
560 value is used by @code{__builtin_apply} to compute the amount of data
561 that should be pushed on the stack and copied from the incoming argument
565 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
566 This built-in function returns the value described by @var{result} from
567 the containing function. You should specify, for @var{result}, a value
568 returned by @code{__builtin_apply}.
571 @deftypefn {Built-in Function} {} __builtin_va_arg_pack ()
572 This built-in function represents all anonymous arguments of an inline
573 function. It can be used only in inline functions which will be always
574 inlined, never compiled as a separate function, such as those using
575 @code{__attribute__ ((__always_inline__))} or
576 @code{__attribute__ ((__gnu_inline__))} extern inline functions.
577 It must be only passed as last argument to some other function
578 with variable arguments. This is useful for writing small wrapper
579 inlines for variable argument functions, when using preprocessor
580 macros is undesirable. For example:
582 extern int myprintf (FILE *f, const char *format, ...);
583 extern inline __attribute__ ((__gnu_inline__)) int
584 myprintf (FILE *f, const char *format, ...)
586 int r = fprintf (f, "myprintf: ");
589 int s = fprintf (f, format, __builtin_va_arg_pack ());
597 @deftypefn {Built-in Function} {size_t} __builtin_va_arg_pack_len ()
598 This built-in function returns the number of anonymous arguments of
599 an inline function. It can be used only in inline functions which
600 will be always inlined, never compiled as a separate function, such
601 as those using @code{__attribute__ ((__always_inline__))} or
602 @code{__attribute__ ((__gnu_inline__))} extern inline functions.
603 For example following will do link or runtime checking of open
604 arguments for optimized code:
607 extern inline __attribute__((__gnu_inline__)) int
608 myopen (const char *path, int oflag, ...)
610 if (__builtin_va_arg_pack_len () > 1)
611 warn_open_too_many_arguments ();
613 if (__builtin_constant_p (oflag))
615 if ((oflag & O_CREAT) != 0 && __builtin_va_arg_pack_len () < 1)
617 warn_open_missing_mode ();
618 return __open_2 (path, oflag);
620 return open (path, oflag, __builtin_va_arg_pack ());
623 if (__builtin_va_arg_pack_len () < 1)
624 return __open_2 (path, oflag);
626 return open (path, oflag, __builtin_va_arg_pack ());
633 @section Referring to a Type with @code{typeof}
636 @cindex macros, types of arguments
638 Another way to refer to the type of an expression is with @code{typeof}.
639 The syntax of using of this keyword looks like @code{sizeof}, but the
640 construct acts semantically like a type name defined with @code{typedef}.
642 There are two ways of writing the argument to @code{typeof}: with an
643 expression or with a type. Here is an example with an expression:
650 This assumes that @code{x} is an array of pointers to functions;
651 the type described is that of the values of the functions.
653 Here is an example with a typename as the argument:
660 Here the type described is that of pointers to @code{int}.
662 If you are writing a header file that must work when included in ISO C
663 programs, write @code{__typeof__} instead of @code{typeof}.
664 @xref{Alternate Keywords}.
666 A @code{typeof}-construct can be used anywhere a typedef name could be
667 used. For example, you can use it in a declaration, in a cast, or inside
668 of @code{sizeof} or @code{typeof}.
670 The operand of @code{typeof} is evaluated for its side effects if and
671 only if it is an expression of variably modified type or the name of
674 @code{typeof} is often useful in conjunction with the
675 statements-within-expressions feature. Here is how the two together can
676 be used to define a safe ``maximum'' macro that operates on any
677 arithmetic type and evaluates each of its arguments exactly once:
681 (@{ typeof (a) _a = (a); \
682 typeof (b) _b = (b); \
683 _a > _b ? _a : _b; @})
686 @cindex underscores in variables in macros
687 @cindex @samp{_} in variables in macros
688 @cindex local variables in macros
689 @cindex variables, local, in macros
690 @cindex macros, local variables in
692 The reason for using names that start with underscores for the local
693 variables is to avoid conflicts with variable names that occur within the
694 expressions that are substituted for @code{a} and @code{b}. Eventually we
695 hope to design a new form of declaration syntax that allows you to declare
696 variables whose scopes start only after their initializers; this will be a
697 more reliable way to prevent such conflicts.
700 Some more examples of the use of @code{typeof}:
704 This declares @code{y} with the type of what @code{x} points to.
711 This declares @code{y} as an array of such values.
718 This declares @code{y} as an array of pointers to characters:
721 typeof (typeof (char *)[4]) y;
725 It is equivalent to the following traditional C declaration:
731 To see the meaning of the declaration using @code{typeof}, and why it
732 might be a useful way to write, rewrite it with these macros:
735 #define pointer(T) typeof(T *)
736 #define array(T, N) typeof(T [N])
740 Now the declaration can be rewritten this way:
743 array (pointer (char), 4) y;
747 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
748 pointers to @code{char}.
751 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
752 a more limited extension which permitted one to write
755 typedef @var{T} = @var{expr};
759 with the effect of declaring @var{T} to have the type of the expression
760 @var{expr}. This extension does not work with GCC 3 (versions between
761 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
762 relies on it should be rewritten to use @code{typeof}:
765 typedef typeof(@var{expr}) @var{T};
769 This will work with all versions of GCC@.
772 @section Conditionals with Omitted Operands
773 @cindex conditional expressions, extensions
774 @cindex omitted middle-operands
775 @cindex middle-operands, omitted
776 @cindex extensions, @code{?:}
777 @cindex @code{?:} extensions
779 The middle operand in a conditional expression may be omitted. Then
780 if the first operand is nonzero, its value is the value of the conditional
783 Therefore, the expression
790 has the value of @code{x} if that is nonzero; otherwise, the value of
793 This example is perfectly equivalent to
799 @cindex side effect in @code{?:}
800 @cindex @code{?:} side effect
802 In this simple case, the ability to omit the middle operand is not
803 especially useful. When it becomes useful is when the first operand does,
804 or may (if it is a macro argument), contain a side effect. Then repeating
805 the operand in the middle would perform the side effect twice. Omitting
806 the middle operand uses the value already computed without the undesirable
807 effects of recomputing it.
810 @section 128-bits integers
811 @cindex @code{__int128} data types
813 As an extension the integer scalar type @code{__int128} is supported for
814 targets having an integer mode wide enough to hold 128-bit.
815 Simply write @code{__int128} for a signed 128-bit integer, or
816 @code{unsigned __int128} for an unsigned 128-bit integer. There is no
817 support in GCC to express an integer constant of type @code{__int128}
818 for targets having @code{long long} integer with less then 128 bit width.
821 @section Double-Word Integers
822 @cindex @code{long long} data types
823 @cindex double-word arithmetic
824 @cindex multiprecision arithmetic
825 @cindex @code{LL} integer suffix
826 @cindex @code{ULL} integer suffix
828 ISO C99 supports data types for integers that are at least 64 bits wide,
829 and as an extension GCC supports them in C90 mode and in C++.
830 Simply write @code{long long int} for a signed integer, or
831 @code{unsigned long long int} for an unsigned integer. To make an
832 integer constant of type @code{long long int}, add the suffix @samp{LL}
833 to the integer. To make an integer constant of type @code{unsigned long
834 long int}, add the suffix @samp{ULL} to the integer.
836 You can use these types in arithmetic like any other integer types.
837 Addition, subtraction, and bitwise boolean operations on these types
838 are open-coded on all types of machines. Multiplication is open-coded
839 if the machine supports fullword-to-doubleword a widening multiply
840 instruction. Division and shifts are open-coded only on machines that
841 provide special support. The operations that are not open-coded use
842 special library routines that come with GCC@.
844 There may be pitfalls when you use @code{long long} types for function
845 arguments, unless you declare function prototypes. If a function
846 expects type @code{int} for its argument, and you pass a value of type
847 @code{long long int}, confusion will result because the caller and the
848 subroutine will disagree about the number of bytes for the argument.
849 Likewise, if the function expects @code{long long int} and you pass
850 @code{int}. The best way to avoid such problems is to use prototypes.
853 @section Complex Numbers
854 @cindex complex numbers
855 @cindex @code{_Complex} keyword
856 @cindex @code{__complex__} keyword
858 ISO C99 supports complex floating data types, and as an extension GCC
859 supports them in C90 mode and in C++, and supports complex integer data
860 types which are not part of ISO C99. You can declare complex types
861 using the keyword @code{_Complex}. As an extension, the older GNU
862 keyword @code{__complex__} is also supported.
864 For example, @samp{_Complex double x;} declares @code{x} as a
865 variable whose real part and imaginary part are both of type
866 @code{double}. @samp{_Complex short int y;} declares @code{y} to
867 have real and imaginary parts of type @code{short int}; this is not
868 likely to be useful, but it shows that the set of complex types is
871 To write a constant with a complex data type, use the suffix @samp{i} or
872 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
873 has type @code{_Complex float} and @code{3i} has type
874 @code{_Complex int}. Such a constant always has a pure imaginary
875 value, but you can form any complex value you like by adding one to a
876 real constant. This is a GNU extension; if you have an ISO C99
877 conforming C library (such as GNU libc), and want to construct complex
878 constants of floating type, you should include @code{<complex.h>} and
879 use the macros @code{I} or @code{_Complex_I} instead.
881 @cindex @code{__real__} keyword
882 @cindex @code{__imag__} keyword
883 To extract the real part of a complex-valued expression @var{exp}, write
884 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
885 extract the imaginary part. This is a GNU extension; for values of
886 floating type, you should use the ISO C99 functions @code{crealf},
887 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
888 @code{cimagl}, declared in @code{<complex.h>} and also provided as
889 built-in functions by GCC@.
891 @cindex complex conjugation
892 The operator @samp{~} performs complex conjugation when used on a value
893 with a complex type. This is a GNU extension; for values of
894 floating type, you should use the ISO C99 functions @code{conjf},
895 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
896 provided as built-in functions by GCC@.
898 GCC can allocate complex automatic variables in a noncontiguous
899 fashion; it's even possible for the real part to be in a register while
900 the imaginary part is on the stack (or vice-versa). Only the DWARF2
901 debug info format can represent this, so use of DWARF2 is recommended.
902 If you are using the stabs debug info format, GCC describes a noncontiguous
903 complex variable as if it were two separate variables of noncomplex type.
904 If the variable's actual name is @code{foo}, the two fictitious
905 variables are named @code{foo$real} and @code{foo$imag}. You can
906 examine and set these two fictitious variables with your debugger.
909 @section Additional Floating Types
910 @cindex additional floating types
911 @cindex @code{__float80} data type
912 @cindex @code{__float128} data type
913 @cindex @code{w} floating point suffix
914 @cindex @code{q} floating point suffix
915 @cindex @code{W} floating point suffix
916 @cindex @code{Q} floating point suffix
918 As an extension, the GNU C compiler supports additional floating
919 types, @code{__float80} and @code{__float128} to support 80bit
920 (@code{XFmode}) and 128 bit (@code{TFmode}) floating types.
921 Support for additional types includes the arithmetic operators:
922 add, subtract, multiply, divide; unary arithmetic operators;
923 relational operators; equality operators; and conversions to and from
924 integer and other floating types. Use a suffix @samp{w} or @samp{W}
925 in a literal constant of type @code{__float80} and @samp{q} or @samp{Q}
926 for @code{_float128}. You can declare complex types using the
927 corresponding internal complex type, @code{XCmode} for @code{__float80}
928 type and @code{TCmode} for @code{__float128} type:
931 typedef _Complex float __attribute__((mode(TC))) _Complex128;
932 typedef _Complex float __attribute__((mode(XC))) _Complex80;
935 Not all targets support additional floating point types. @code{__float80}
936 and @code{__float128} types are supported on i386, x86_64 and ia64 targets.
939 @section Half-Precision Floating Point
940 @cindex half-precision floating point
941 @cindex @code{__fp16} data type
943 On ARM targets, GCC supports half-precision (16-bit) floating point via
944 the @code{__fp16} type. You must enable this type explicitly
945 with the @option{-mfp16-format} command-line option in order to use it.
947 ARM supports two incompatible representations for half-precision
948 floating-point values. You must choose one of the representations and
949 use it consistently in your program.
951 Specifying @option{-mfp16-format=ieee} selects the IEEE 754-2008 format.
952 This format can represent normalized values in the range of @math{2^{-14}} to 65504.
953 There are 11 bits of significand precision, approximately 3
956 Specifying @option{-mfp16-format=alternative} selects the ARM
957 alternative format. This representation is similar to the IEEE
958 format, but does not support infinities or NaNs. Instead, the range
959 of exponents is extended, so that this format can represent normalized
960 values in the range of @math{2^{-14}} to 131008.
962 The @code{__fp16} type is a storage format only. For purposes
963 of arithmetic and other operations, @code{__fp16} values in C or C++
964 expressions are automatically promoted to @code{float}. In addition,
965 you cannot declare a function with a return value or parameters
966 of type @code{__fp16}.
968 Note that conversions from @code{double} to @code{__fp16}
969 involve an intermediate conversion to @code{float}. Because
970 of rounding, this can sometimes produce a different result than a
973 ARM provides hardware support for conversions between
974 @code{__fp16} and @code{float} values
975 as an extension to VFP and NEON (Advanced SIMD). GCC generates
976 code using these hardware instructions if you compile with
977 options to select an FPU that provides them;
978 for example, @option{-mfpu=neon-fp16 -mfloat-abi=softfp},
979 in addition to the @option{-mfp16-format} option to select
980 a half-precision format.
982 Language-level support for the @code{__fp16} data type is
983 independent of whether GCC generates code using hardware floating-point
984 instructions. In cases where hardware support is not specified, GCC
985 implements conversions between @code{__fp16} and @code{float} values
989 @section Decimal Floating Types
990 @cindex decimal floating types
991 @cindex @code{_Decimal32} data type
992 @cindex @code{_Decimal64} data type
993 @cindex @code{_Decimal128} data type
994 @cindex @code{df} integer suffix
995 @cindex @code{dd} integer suffix
996 @cindex @code{dl} integer suffix
997 @cindex @code{DF} integer suffix
998 @cindex @code{DD} integer suffix
999 @cindex @code{DL} integer suffix
1001 As an extension, the GNU C compiler supports decimal floating types as
1002 defined in the N1312 draft of ISO/IEC WDTR24732. Support for decimal
1003 floating types in GCC will evolve as the draft technical report changes.
1004 Calling conventions for any target might also change. Not all targets
1005 support decimal floating types.
1007 The decimal floating types are @code{_Decimal32}, @code{_Decimal64}, and
1008 @code{_Decimal128}. They use a radix of ten, unlike the floating types
1009 @code{float}, @code{double}, and @code{long double} whose radix is not
1010 specified by the C standard but is usually two.
1012 Support for decimal floating types includes the arithmetic operators
1013 add, subtract, multiply, divide; unary arithmetic operators;
1014 relational operators; equality operators; and conversions to and from
1015 integer and other floating types. Use a suffix @samp{df} or
1016 @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd}
1017 or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for
1020 GCC support of decimal float as specified by the draft technical report
1025 When the value of a decimal floating type cannot be represented in the
1026 integer type to which it is being converted, the result is undefined
1027 rather than the result value specified by the draft technical report.
1030 GCC does not provide the C library functionality associated with
1031 @file{math.h}, @file{fenv.h}, @file{stdio.h}, @file{stdlib.h}, and
1032 @file{wchar.h}, which must come from a separate C library implementation.
1033 Because of this the GNU C compiler does not define macro
1034 @code{__STDC_DEC_FP__} to indicate that the implementation conforms to
1035 the technical report.
1038 Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128}
1039 are supported by the DWARF2 debug information format.
1045 ISO C99 supports floating-point numbers written not only in the usual
1046 decimal notation, such as @code{1.55e1}, but also numbers such as
1047 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
1048 supports this in C90 mode (except in some cases when strictly
1049 conforming) and in C++. In that format the
1050 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
1051 mandatory. The exponent is a decimal number that indicates the power of
1052 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
1059 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
1060 is the same as @code{1.55e1}.
1062 Unlike for floating-point numbers in the decimal notation the exponent
1063 is always required in the hexadecimal notation. Otherwise the compiler
1064 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
1065 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
1066 extension for floating-point constants of type @code{float}.
1069 @section Fixed-Point Types
1070 @cindex fixed-point types
1071 @cindex @code{_Fract} data type
1072 @cindex @code{_Accum} data type
1073 @cindex @code{_Sat} data type
1074 @cindex @code{hr} fixed-suffix
1075 @cindex @code{r} fixed-suffix
1076 @cindex @code{lr} fixed-suffix
1077 @cindex @code{llr} fixed-suffix
1078 @cindex @code{uhr} fixed-suffix
1079 @cindex @code{ur} fixed-suffix
1080 @cindex @code{ulr} fixed-suffix
1081 @cindex @code{ullr} fixed-suffix
1082 @cindex @code{hk} fixed-suffix
1083 @cindex @code{k} fixed-suffix
1084 @cindex @code{lk} fixed-suffix
1085 @cindex @code{llk} fixed-suffix
1086 @cindex @code{uhk} fixed-suffix
1087 @cindex @code{uk} fixed-suffix
1088 @cindex @code{ulk} fixed-suffix
1089 @cindex @code{ullk} fixed-suffix
1090 @cindex @code{HR} fixed-suffix
1091 @cindex @code{R} fixed-suffix
1092 @cindex @code{LR} fixed-suffix
1093 @cindex @code{LLR} fixed-suffix
1094 @cindex @code{UHR} fixed-suffix
1095 @cindex @code{UR} fixed-suffix
1096 @cindex @code{ULR} fixed-suffix
1097 @cindex @code{ULLR} fixed-suffix
1098 @cindex @code{HK} fixed-suffix
1099 @cindex @code{K} fixed-suffix
1100 @cindex @code{LK} fixed-suffix
1101 @cindex @code{LLK} fixed-suffix
1102 @cindex @code{UHK} fixed-suffix
1103 @cindex @code{UK} fixed-suffix
1104 @cindex @code{ULK} fixed-suffix
1105 @cindex @code{ULLK} fixed-suffix
1107 As an extension, the GNU C compiler supports fixed-point types as
1108 defined in the N1169 draft of ISO/IEC DTR 18037. Support for fixed-point
1109 types in GCC will evolve as the draft technical report changes.
1110 Calling conventions for any target might also change. Not all targets
1111 support fixed-point types.
1113 The fixed-point types are
1114 @code{short _Fract},
1117 @code{long long _Fract},
1118 @code{unsigned short _Fract},
1119 @code{unsigned _Fract},
1120 @code{unsigned long _Fract},
1121 @code{unsigned long long _Fract},
1122 @code{_Sat short _Fract},
1124 @code{_Sat long _Fract},
1125 @code{_Sat long long _Fract},
1126 @code{_Sat unsigned short _Fract},
1127 @code{_Sat unsigned _Fract},
1128 @code{_Sat unsigned long _Fract},
1129 @code{_Sat unsigned long long _Fract},
1130 @code{short _Accum},
1133 @code{long long _Accum},
1134 @code{unsigned short _Accum},
1135 @code{unsigned _Accum},
1136 @code{unsigned long _Accum},
1137 @code{unsigned long long _Accum},
1138 @code{_Sat short _Accum},
1140 @code{_Sat long _Accum},
1141 @code{_Sat long long _Accum},
1142 @code{_Sat unsigned short _Accum},
1143 @code{_Sat unsigned _Accum},
1144 @code{_Sat unsigned long _Accum},
1145 @code{_Sat unsigned long long _Accum}.
1147 Fixed-point data values contain fractional and optional integral parts.
1148 The format of fixed-point data varies and depends on the target machine.
1150 Support for fixed-point types includes:
1153 prefix and postfix increment and decrement operators (@code{++}, @code{--})
1155 unary arithmetic operators (@code{+}, @code{-}, @code{!})
1157 binary arithmetic operators (@code{+}, @code{-}, @code{*}, @code{/})
1159 binary shift operators (@code{<<}, @code{>>})
1161 relational operators (@code{<}, @code{<=}, @code{>=}, @code{>})
1163 equality operators (@code{==}, @code{!=})
1165 assignment operators (@code{+=}, @code{-=}, @code{*=}, @code{/=},
1166 @code{<<=}, @code{>>=})
1168 conversions to and from integer, floating-point, or fixed-point types
1171 Use a suffix in a fixed-point literal constant:
1173 @item @samp{hr} or @samp{HR} for @code{short _Fract} and
1174 @code{_Sat short _Fract}
1175 @item @samp{r} or @samp{R} for @code{_Fract} and @code{_Sat _Fract}
1176 @item @samp{lr} or @samp{LR} for @code{long _Fract} and
1177 @code{_Sat long _Fract}
1178 @item @samp{llr} or @samp{LLR} for @code{long long _Fract} and
1179 @code{_Sat long long _Fract}
1180 @item @samp{uhr} or @samp{UHR} for @code{unsigned short _Fract} and
1181 @code{_Sat unsigned short _Fract}
1182 @item @samp{ur} or @samp{UR} for @code{unsigned _Fract} and
1183 @code{_Sat unsigned _Fract}
1184 @item @samp{ulr} or @samp{ULR} for @code{unsigned long _Fract} and
1185 @code{_Sat unsigned long _Fract}
1186 @item @samp{ullr} or @samp{ULLR} for @code{unsigned long long _Fract}
1187 and @code{_Sat unsigned long long _Fract}
1188 @item @samp{hk} or @samp{HK} for @code{short _Accum} and
1189 @code{_Sat short _Accum}
1190 @item @samp{k} or @samp{K} for @code{_Accum} and @code{_Sat _Accum}
1191 @item @samp{lk} or @samp{LK} for @code{long _Accum} and
1192 @code{_Sat long _Accum}
1193 @item @samp{llk} or @samp{LLK} for @code{long long _Accum} and
1194 @code{_Sat long long _Accum}
1195 @item @samp{uhk} or @samp{UHK} for @code{unsigned short _Accum} and
1196 @code{_Sat unsigned short _Accum}
1197 @item @samp{uk} or @samp{UK} for @code{unsigned _Accum} and
1198 @code{_Sat unsigned _Accum}
1199 @item @samp{ulk} or @samp{ULK} for @code{unsigned long _Accum} and
1200 @code{_Sat unsigned long _Accum}
1201 @item @samp{ullk} or @samp{ULLK} for @code{unsigned long long _Accum}
1202 and @code{_Sat unsigned long long _Accum}
1205 GCC support of fixed-point types as specified by the draft technical report
1210 Pragmas to control overflow and rounding behaviors are not implemented.
1213 Fixed-point types are supported by the DWARF2 debug information format.
1215 @node Named Address Spaces
1216 @section Named address spaces
1217 @cindex named address spaces
1219 As an extension, the GNU C compiler supports named address spaces as
1220 defined in the N1275 draft of ISO/IEC DTR 18037. Support for named
1221 address spaces in GCC will evolve as the draft technical report changes.
1222 Calling conventions for any target might also change. At present, only
1223 the SPU and M32C targets support other address spaces. On the SPU target, for
1224 example, variables may be declared as belonging to another address space
1225 by qualifying the type with the @code{__ea} address space identifier:
1231 When the variable @code{i} is accessed, the compiler will generate
1232 special code to access this variable. It may use runtime library
1233 support, or generate special machine instructions to access that address
1236 The @code{__ea} identifier may be used exactly like any other C type
1237 qualifier (e.g., @code{const} or @code{volatile}). See the N1275
1238 document for more details.
1240 On the M32C target, with the R8C and M16C cpu variants, variables
1241 qualified with @code{__far} are accessed using 32-bit addresses in
1242 order to access memory beyond the first 64k bytes. If @code{__far} is
1243 used with the M32CM or M32C cpu variants, it has no effect.
1246 @section Arrays of Length Zero
1247 @cindex arrays of length zero
1248 @cindex zero-length arrays
1249 @cindex length-zero arrays
1250 @cindex flexible array members
1252 Zero-length arrays are allowed in GNU C@. They are very useful as the
1253 last element of a structure which is really a header for a variable-length
1262 struct line *thisline = (struct line *)
1263 malloc (sizeof (struct line) + this_length);
1264 thisline->length = this_length;
1267 In ISO C90, you would have to give @code{contents} a length of 1, which
1268 means either you waste space or complicate the argument to @code{malloc}.
1270 In ISO C99, you would use a @dfn{flexible array member}, which is
1271 slightly different in syntax and semantics:
1275 Flexible array members are written as @code{contents[]} without
1279 Flexible array members have incomplete type, and so the @code{sizeof}
1280 operator may not be applied. As a quirk of the original implementation
1281 of zero-length arrays, @code{sizeof} evaluates to zero.
1284 Flexible array members may only appear as the last member of a
1285 @code{struct} that is otherwise non-empty.
1288 A structure containing a flexible array member, or a union containing
1289 such a structure (possibly recursively), may not be a member of a
1290 structure or an element of an array. (However, these uses are
1291 permitted by GCC as extensions.)
1294 GCC versions before 3.0 allowed zero-length arrays to be statically
1295 initialized, as if they were flexible arrays. In addition to those
1296 cases that were useful, it also allowed initializations in situations
1297 that would corrupt later data. Non-empty initialization of zero-length
1298 arrays is now treated like any case where there are more initializer
1299 elements than the array holds, in that a suitable warning about "excess
1300 elements in array" is given, and the excess elements (all of them, in
1301 this case) are ignored.
1303 Instead GCC allows static initialization of flexible array members.
1304 This is equivalent to defining a new structure containing the original
1305 structure followed by an array of sufficient size to contain the data.
1306 I.e.@: in the following, @code{f1} is constructed as if it were declared
1312 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
1315 struct f1 f1; int data[3];
1316 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1320 The convenience of this extension is that @code{f1} has the desired
1321 type, eliminating the need to consistently refer to @code{f2.f1}.
1323 This has symmetry with normal static arrays, in that an array of
1324 unknown size is also written with @code{[]}.
1326 Of course, this extension only makes sense if the extra data comes at
1327 the end of a top-level object, as otherwise we would be overwriting
1328 data at subsequent offsets. To avoid undue complication and confusion
1329 with initialization of deeply nested arrays, we simply disallow any
1330 non-empty initialization except when the structure is the top-level
1331 object. For example:
1334 struct foo @{ int x; int y[]; @};
1335 struct bar @{ struct foo z; @};
1337 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1338 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1339 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1340 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1343 @node Empty Structures
1344 @section Structures With No Members
1345 @cindex empty structures
1346 @cindex zero-size structures
1348 GCC permits a C structure to have no members:
1355 The structure will have size zero. In C++, empty structures are part
1356 of the language. G++ treats empty structures as if they had a single
1357 member of type @code{char}.
1359 @node Variable Length
1360 @section Arrays of Variable Length
1361 @cindex variable-length arrays
1362 @cindex arrays of variable length
1365 Variable-length automatic arrays are allowed in ISO C99, and as an
1366 extension GCC accepts them in C90 mode and in C++. These arrays are
1367 declared like any other automatic arrays, but with a length that is not
1368 a constant expression. The storage is allocated at the point of
1369 declaration and deallocated when the brace-level is exited. For
1374 concat_fopen (char *s1, char *s2, char *mode)
1376 char str[strlen (s1) + strlen (s2) + 1];
1379 return fopen (str, mode);
1383 @cindex scope of a variable length array
1384 @cindex variable-length array scope
1385 @cindex deallocating variable length arrays
1386 Jumping or breaking out of the scope of the array name deallocates the
1387 storage. Jumping into the scope is not allowed; you get an error
1390 @cindex @code{alloca} vs variable-length arrays
1391 You can use the function @code{alloca} to get an effect much like
1392 variable-length arrays. The function @code{alloca} is available in
1393 many other C implementations (but not in all). On the other hand,
1394 variable-length arrays are more elegant.
1396 There are other differences between these two methods. Space allocated
1397 with @code{alloca} exists until the containing @emph{function} returns.
1398 The space for a variable-length array is deallocated as soon as the array
1399 name's scope ends. (If you use both variable-length arrays and
1400 @code{alloca} in the same function, deallocation of a variable-length array
1401 will also deallocate anything more recently allocated with @code{alloca}.)
1403 You can also use variable-length arrays as arguments to functions:
1407 tester (int len, char data[len][len])
1413 The length of an array is computed once when the storage is allocated
1414 and is remembered for the scope of the array in case you access it with
1417 If you want to pass the array first and the length afterward, you can
1418 use a forward declaration in the parameter list---another GNU extension.
1422 tester (int len; char data[len][len], int len)
1428 @cindex parameter forward declaration
1429 The @samp{int len} before the semicolon is a @dfn{parameter forward
1430 declaration}, and it serves the purpose of making the name @code{len}
1431 known when the declaration of @code{data} is parsed.
1433 You can write any number of such parameter forward declarations in the
1434 parameter list. They can be separated by commas or semicolons, but the
1435 last one must end with a semicolon, which is followed by the ``real''
1436 parameter declarations. Each forward declaration must match a ``real''
1437 declaration in parameter name and data type. ISO C99 does not support
1438 parameter forward declarations.
1440 @node Variadic Macros
1441 @section Macros with a Variable Number of Arguments.
1442 @cindex variable number of arguments
1443 @cindex macro with variable arguments
1444 @cindex rest argument (in macro)
1445 @cindex variadic macros
1447 In the ISO C standard of 1999, a macro can be declared to accept a
1448 variable number of arguments much as a function can. The syntax for
1449 defining the macro is similar to that of a function. Here is an
1453 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1456 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1457 such a macro, it represents the zero or more tokens until the closing
1458 parenthesis that ends the invocation, including any commas. This set of
1459 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1460 wherever it appears. See the CPP manual for more information.
1462 GCC has long supported variadic macros, and used a different syntax that
1463 allowed you to give a name to the variable arguments just like any other
1464 argument. Here is an example:
1467 #define debug(format, args...) fprintf (stderr, format, args)
1470 This is in all ways equivalent to the ISO C example above, but arguably
1471 more readable and descriptive.
1473 GNU CPP has two further variadic macro extensions, and permits them to
1474 be used with either of the above forms of macro definition.
1476 In standard C, you are not allowed to leave the variable argument out
1477 entirely; but you are allowed to pass an empty argument. For example,
1478 this invocation is invalid in ISO C, because there is no comma after
1485 GNU CPP permits you to completely omit the variable arguments in this
1486 way. In the above examples, the compiler would complain, though since
1487 the expansion of the macro still has the extra comma after the format
1490 To help solve this problem, CPP behaves specially for variable arguments
1491 used with the token paste operator, @samp{##}. If instead you write
1494 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1497 and if the variable arguments are omitted or empty, the @samp{##}
1498 operator causes the preprocessor to remove the comma before it. If you
1499 do provide some variable arguments in your macro invocation, GNU CPP
1500 does not complain about the paste operation and instead places the
1501 variable arguments after the comma. Just like any other pasted macro
1502 argument, these arguments are not macro expanded.
1504 @node Escaped Newlines
1505 @section Slightly Looser Rules for Escaped Newlines
1506 @cindex escaped newlines
1507 @cindex newlines (escaped)
1509 Recently, the preprocessor has relaxed its treatment of escaped
1510 newlines. Previously, the newline had to immediately follow a
1511 backslash. The current implementation allows whitespace in the form
1512 of spaces, horizontal and vertical tabs, and form feeds between the
1513 backslash and the subsequent newline. The preprocessor issues a
1514 warning, but treats it as a valid escaped newline and combines the two
1515 lines to form a single logical line. This works within comments and
1516 tokens, as well as between tokens. Comments are @emph{not} treated as
1517 whitespace for the purposes of this relaxation, since they have not
1518 yet been replaced with spaces.
1521 @section Non-Lvalue Arrays May Have Subscripts
1522 @cindex subscripting
1523 @cindex arrays, non-lvalue
1525 @cindex subscripting and function values
1526 In ISO C99, arrays that are not lvalues still decay to pointers, and
1527 may be subscripted, although they may not be modified or used after
1528 the next sequence point and the unary @samp{&} operator may not be
1529 applied to them. As an extension, GCC allows such arrays to be
1530 subscripted in C90 mode, though otherwise they do not decay to
1531 pointers outside C99 mode. For example,
1532 this is valid in GNU C though not valid in C90:
1536 struct foo @{int a[4];@};
1542 return f().a[index];
1548 @section Arithmetic on @code{void}- and Function-Pointers
1549 @cindex void pointers, arithmetic
1550 @cindex void, size of pointer to
1551 @cindex function pointers, arithmetic
1552 @cindex function, size of pointer to
1554 In GNU C, addition and subtraction operations are supported on pointers to
1555 @code{void} and on pointers to functions. This is done by treating the
1556 size of a @code{void} or of a function as 1.
1558 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1559 and on function types, and returns 1.
1561 @opindex Wpointer-arith
1562 The option @option{-Wpointer-arith} requests a warning if these extensions
1566 @section Non-Constant Initializers
1567 @cindex initializers, non-constant
1568 @cindex non-constant initializers
1570 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1571 automatic variable are not required to be constant expressions in GNU C@.
1572 Here is an example of an initializer with run-time varying elements:
1575 foo (float f, float g)
1577 float beat_freqs[2] = @{ f-g, f+g @};
1582 @node Compound Literals
1583 @section Compound Literals
1584 @cindex constructor expressions
1585 @cindex initializations in expressions
1586 @cindex structures, constructor expression
1587 @cindex expressions, constructor
1588 @cindex compound literals
1589 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1591 ISO C99 supports compound literals. A compound literal looks like
1592 a cast containing an initializer. Its value is an object of the
1593 type specified in the cast, containing the elements specified in
1594 the initializer; it is an lvalue. As an extension, GCC supports
1595 compound literals in C90 mode and in C++.
1597 Usually, the specified type is a structure. Assume that
1598 @code{struct foo} and @code{structure} are declared as shown:
1601 struct foo @{int a; char b[2];@} structure;
1605 Here is an example of constructing a @code{struct foo} with a compound literal:
1608 structure = ((struct foo) @{x + y, 'a', 0@});
1612 This is equivalent to writing the following:
1616 struct foo temp = @{x + y, 'a', 0@};
1621 You can also construct an array. If all the elements of the compound literal
1622 are (made up of) simple constant expressions, suitable for use in
1623 initializers of objects of static storage duration, then the compound
1624 literal can be coerced to a pointer to its first element and used in
1625 such an initializer, as shown here:
1628 char **foo = (char *[]) @{ "x", "y", "z" @};
1631 Compound literals for scalar types and union types are is
1632 also allowed, but then the compound literal is equivalent
1635 As a GNU extension, GCC allows initialization of objects with static storage
1636 duration by compound literals (which is not possible in ISO C99, because
1637 the initializer is not a constant).
1638 It is handled as if the object was initialized only with the bracket
1639 enclosed list if the types of the compound literal and the object match.
1640 The initializer list of the compound literal must be constant.
1641 If the object being initialized has array type of unknown size, the size is
1642 determined by compound literal size.
1645 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1646 static int y[] = (int []) @{1, 2, 3@};
1647 static int z[] = (int [3]) @{1@};
1651 The above lines are equivalent to the following:
1653 static struct foo x = @{1, 'a', 'b'@};
1654 static int y[] = @{1, 2, 3@};
1655 static int z[] = @{1, 0, 0@};
1658 @node Designated Inits
1659 @section Designated Initializers
1660 @cindex initializers with labeled elements
1661 @cindex labeled elements in initializers
1662 @cindex case labels in initializers
1663 @cindex designated initializers
1665 Standard C90 requires the elements of an initializer to appear in a fixed
1666 order, the same as the order of the elements in the array or structure
1669 In ISO C99 you can give the elements in any order, specifying the array
1670 indices or structure field names they apply to, and GNU C allows this as
1671 an extension in C90 mode as well. This extension is not
1672 implemented in GNU C++.
1674 To specify an array index, write
1675 @samp{[@var{index}] =} before the element value. For example,
1678 int a[6] = @{ [4] = 29, [2] = 15 @};
1685 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1689 The index values must be constant expressions, even if the array being
1690 initialized is automatic.
1692 An alternative syntax for this which has been obsolete since GCC 2.5 but
1693 GCC still accepts is to write @samp{[@var{index}]} before the element
1694 value, with no @samp{=}.
1696 To initialize a range of elements to the same value, write
1697 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1698 extension. For example,
1701 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1705 If the value in it has side-effects, the side-effects will happen only once,
1706 not for each initialized field by the range initializer.
1709 Note that the length of the array is the highest value specified
1712 In a structure initializer, specify the name of a field to initialize
1713 with @samp{.@var{fieldname} =} before the element value. For example,
1714 given the following structure,
1717 struct point @{ int x, y; @};
1721 the following initialization
1724 struct point p = @{ .y = yvalue, .x = xvalue @};
1731 struct point p = @{ xvalue, yvalue @};
1734 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1735 @samp{@var{fieldname}:}, as shown here:
1738 struct point p = @{ y: yvalue, x: xvalue @};
1742 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1743 @dfn{designator}. You can also use a designator (or the obsolete colon
1744 syntax) when initializing a union, to specify which element of the union
1745 should be used. For example,
1748 union foo @{ int i; double d; @};
1750 union foo f = @{ .d = 4 @};
1754 will convert 4 to a @code{double} to store it in the union using
1755 the second element. By contrast, casting 4 to type @code{union foo}
1756 would store it into the union as the integer @code{i}, since it is
1757 an integer. (@xref{Cast to Union}.)
1759 You can combine this technique of naming elements with ordinary C
1760 initialization of successive elements. Each initializer element that
1761 does not have a designator applies to the next consecutive element of the
1762 array or structure. For example,
1765 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1772 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1775 Labeling the elements of an array initializer is especially useful
1776 when the indices are characters or belong to an @code{enum} type.
1781 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1782 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1785 @cindex designator lists
1786 You can also write a series of @samp{.@var{fieldname}} and
1787 @samp{[@var{index}]} designators before an @samp{=} to specify a
1788 nested subobject to initialize; the list is taken relative to the
1789 subobject corresponding to the closest surrounding brace pair. For
1790 example, with the @samp{struct point} declaration above:
1793 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1797 If the same field is initialized multiple times, it will have value from
1798 the last initialization. If any such overridden initialization has
1799 side-effect, it is unspecified whether the side-effect happens or not.
1800 Currently, GCC will discard them and issue a warning.
1803 @section Case Ranges
1805 @cindex ranges in case statements
1807 You can specify a range of consecutive values in a single @code{case} label,
1811 case @var{low} ... @var{high}:
1815 This has the same effect as the proper number of individual @code{case}
1816 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1818 This feature is especially useful for ranges of ASCII character codes:
1824 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1825 it may be parsed wrong when you use it with integer values. For example,
1840 @section Cast to a Union Type
1841 @cindex cast to a union
1842 @cindex union, casting to a
1844 A cast to union type is similar to other casts, except that the type
1845 specified is a union type. You can specify the type either with
1846 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1847 a constructor though, not a cast, and hence does not yield an lvalue like
1848 normal casts. (@xref{Compound Literals}.)
1850 The types that may be cast to the union type are those of the members
1851 of the union. Thus, given the following union and variables:
1854 union foo @{ int i; double d; @};
1860 both @code{x} and @code{y} can be cast to type @code{union foo}.
1862 Using the cast as the right-hand side of an assignment to a variable of
1863 union type is equivalent to storing in a member of the union:
1868 u = (union foo) x @equiv{} u.i = x
1869 u = (union foo) y @equiv{} u.d = y
1872 You can also use the union cast as a function argument:
1875 void hack (union foo);
1877 hack ((union foo) x);
1880 @node Mixed Declarations
1881 @section Mixed Declarations and Code
1882 @cindex mixed declarations and code
1883 @cindex declarations, mixed with code
1884 @cindex code, mixed with declarations
1886 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1887 within compound statements. As an extension, GCC also allows this in
1888 C90 mode. For example, you could do:
1897 Each identifier is visible from where it is declared until the end of
1898 the enclosing block.
1900 @node Function Attributes
1901 @section Declaring Attributes of Functions
1902 @cindex function attributes
1903 @cindex declaring attributes of functions
1904 @cindex functions that never return
1905 @cindex functions that return more than once
1906 @cindex functions that have no side effects
1907 @cindex functions in arbitrary sections
1908 @cindex functions that behave like malloc
1909 @cindex @code{volatile} applied to function
1910 @cindex @code{const} applied to function
1911 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1912 @cindex functions with non-null pointer arguments
1913 @cindex functions that are passed arguments in registers on the 386
1914 @cindex functions that pop the argument stack on the 386
1915 @cindex functions that do not pop the argument stack on the 386
1916 @cindex functions that have different compilation options on the 386
1917 @cindex functions that have different optimization options
1918 @cindex functions that are dynamically resolved
1920 In GNU C, you declare certain things about functions called in your program
1921 which help the compiler optimize function calls and check your code more
1924 The keyword @code{__attribute__} allows you to specify special
1925 attributes when making a declaration. This keyword is followed by an
1926 attribute specification inside double parentheses. The following
1927 attributes are currently defined for functions on all targets:
1928 @code{aligned}, @code{alloc_size}, @code{noreturn},
1929 @code{returns_twice}, @code{noinline}, @code{noclone},
1930 @code{always_inline}, @code{flatten}, @code{pure}, @code{const},
1931 @code{nothrow}, @code{sentinel}, @code{format}, @code{format_arg},
1932 @code{no_instrument_function}, @code{no_split_stack},
1933 @code{section}, @code{constructor},
1934 @code{destructor}, @code{used}, @code{unused}, @code{deprecated},
1935 @code{weak}, @code{malloc}, @code{alias}, @code{ifunc},
1936 @code{warn_unused_result}, @code{nonnull}, @code{gnu_inline},
1937 @code{externally_visible}, @code{hot}, @code{cold}, @code{artificial},
1938 @code{error} and @code{warning}. Several other attributes are defined
1939 for functions on particular target systems. Other attributes,
1940 including @code{section} are supported for variables declarations
1941 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1943 GCC plugins may provide their own attributes.
1945 You may also specify attributes with @samp{__} preceding and following
1946 each keyword. This allows you to use them in header files without
1947 being concerned about a possible macro of the same name. For example,
1948 you may use @code{__noreturn__} instead of @code{noreturn}.
1950 @xref{Attribute Syntax}, for details of the exact syntax for using
1954 @c Keep this table alphabetized by attribute name. Treat _ as space.
1956 @item alias ("@var{target}")
1957 @cindex @code{alias} attribute
1958 The @code{alias} attribute causes the declaration to be emitted as an
1959 alias for another symbol, which must be specified. For instance,
1962 void __f () @{ /* @r{Do something.} */; @}
1963 void f () __attribute__ ((weak, alias ("__f")));
1966 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
1967 mangled name for the target must be used. It is an error if @samp{__f}
1968 is not defined in the same translation unit.
1970 Not all target machines support this attribute.
1972 @item aligned (@var{alignment})
1973 @cindex @code{aligned} attribute
1974 This attribute specifies a minimum alignment for the function,
1977 You cannot use this attribute to decrease the alignment of a function,
1978 only to increase it. However, when you explicitly specify a function
1979 alignment this will override the effect of the
1980 @option{-falign-functions} (@pxref{Optimize Options}) option for this
1983 Note that the effectiveness of @code{aligned} attributes may be
1984 limited by inherent limitations in your linker. On many systems, the
1985 linker is only able to arrange for functions to be aligned up to a
1986 certain maximum alignment. (For some linkers, the maximum supported
1987 alignment may be very very small.) See your linker documentation for
1988 further information.
1990 The @code{aligned} attribute can also be used for variables and fields
1991 (@pxref{Variable Attributes}.)
1994 @cindex @code{alloc_size} attribute
1995 The @code{alloc_size} attribute is used to tell the compiler that the
1996 function return value points to memory, where the size is given by
1997 one or two of the functions parameters. GCC uses this
1998 information to improve the correctness of @code{__builtin_object_size}.
2000 The function parameter(s) denoting the allocated size are specified by
2001 one or two integer arguments supplied to the attribute. The allocated size
2002 is either the value of the single function argument specified or the product
2003 of the two function arguments specified. Argument numbering starts at
2009 void* my_calloc(size_t, size_t) __attribute__((alloc_size(1,2)))
2010 void my_realloc(void*, size_t) __attribute__((alloc_size(2)))
2013 declares that my_calloc will return memory of the size given by
2014 the product of parameter 1 and 2 and that my_realloc will return memory
2015 of the size given by parameter 2.
2018 @cindex @code{always_inline} function attribute
2019 Generally, functions are not inlined unless optimization is specified.
2020 For functions declared inline, this attribute inlines the function even
2021 if no optimization level was specified.
2024 @cindex @code{gnu_inline} function attribute
2025 This attribute should be used with a function which is also declared
2026 with the @code{inline} keyword. It directs GCC to treat the function
2027 as if it were defined in gnu90 mode even when compiling in C99 or
2030 If the function is declared @code{extern}, then this definition of the
2031 function is used only for inlining. In no case is the function
2032 compiled as a standalone function, not even if you take its address
2033 explicitly. Such an address becomes an external reference, as if you
2034 had only declared the function, and had not defined it. This has
2035 almost the effect of a macro. The way to use this is to put a
2036 function definition in a header file with this attribute, and put
2037 another copy of the function, without @code{extern}, in a library
2038 file. The definition in the header file will cause most calls to the
2039 function to be inlined. If any uses of the function remain, they will
2040 refer to the single copy in the library. Note that the two
2041 definitions of the functions need not be precisely the same, although
2042 if they do not have the same effect your program may behave oddly.
2044 In C, if the function is neither @code{extern} nor @code{static}, then
2045 the function is compiled as a standalone function, as well as being
2046 inlined where possible.
2048 This is how GCC traditionally handled functions declared
2049 @code{inline}. Since ISO C99 specifies a different semantics for
2050 @code{inline}, this function attribute is provided as a transition
2051 measure and as a useful feature in its own right. This attribute is
2052 available in GCC 4.1.3 and later. It is available if either of the
2053 preprocessor macros @code{__GNUC_GNU_INLINE__} or
2054 @code{__GNUC_STDC_INLINE__} are defined. @xref{Inline,,An Inline
2055 Function is As Fast As a Macro}.
2057 In C++, this attribute does not depend on @code{extern} in any way,
2058 but it still requires the @code{inline} keyword to enable its special
2062 @cindex @code{artificial} function attribute
2063 This attribute is useful for small inline wrappers which if possible
2064 should appear during debugging as a unit, depending on the debug
2065 info format it will either mean marking the function as artificial
2066 or using the caller location for all instructions within the inlined
2070 @cindex interrupt handler functions
2071 When added to an interrupt handler with the M32C port, causes the
2072 prologue and epilogue to use bank switching to preserve the registers
2073 rather than saving them on the stack.
2076 @cindex @code{flatten} function attribute
2077 Generally, inlining into a function is limited. For a function marked with
2078 this attribute, every call inside this function will be inlined, if possible.
2079 Whether the function itself is considered for inlining depends on its size and
2080 the current inlining parameters.
2082 @item error ("@var{message}")
2083 @cindex @code{error} function attribute
2084 If this attribute is used on a function declaration and a call to such a function
2085 is not eliminated through dead code elimination or other optimizations, an error
2086 which will include @var{message} will be diagnosed. This is useful
2087 for compile time checking, especially together with @code{__builtin_constant_p}
2088 and inline functions where checking the inline function arguments is not
2089 possible through @code{extern char [(condition) ? 1 : -1];} tricks.
2090 While it is possible to leave the function undefined and thus invoke
2091 a link failure, when using this attribute the problem will be diagnosed
2092 earlier and with exact location of the call even in presence of inline
2093 functions or when not emitting debugging information.
2095 @item warning ("@var{message}")
2096 @cindex @code{warning} function attribute
2097 If this attribute is used on a function declaration and a call to such a function
2098 is not eliminated through dead code elimination or other optimizations, a warning
2099 which will include @var{message} will be diagnosed. This is useful
2100 for compile time checking, especially together with @code{__builtin_constant_p}
2101 and inline functions. While it is possible to define the function with
2102 a message in @code{.gnu.warning*} section, when using this attribute the problem
2103 will be diagnosed earlier and with exact location of the call even in presence
2104 of inline functions or when not emitting debugging information.
2107 @cindex functions that do pop the argument stack on the 386
2109 On the Intel 386, the @code{cdecl} attribute causes the compiler to
2110 assume that the calling function will pop off the stack space used to
2111 pass arguments. This is
2112 useful to override the effects of the @option{-mrtd} switch.
2115 @cindex @code{const} function attribute
2116 Many functions do not examine any values except their arguments, and
2117 have no effects except the return value. Basically this is just slightly
2118 more strict class than the @code{pure} attribute below, since function is not
2119 allowed to read global memory.
2121 @cindex pointer arguments
2122 Note that a function that has pointer arguments and examines the data
2123 pointed to must @emph{not} be declared @code{const}. Likewise, a
2124 function that calls a non-@code{const} function usually must not be
2125 @code{const}. It does not make sense for a @code{const} function to
2128 The attribute @code{const} is not implemented in GCC versions earlier
2129 than 2.5. An alternative way to declare that a function has no side
2130 effects, which works in the current version and in some older versions,
2134 typedef int intfn ();
2136 extern const intfn square;
2139 This approach does not work in GNU C++ from 2.6.0 on, since the language
2140 specifies that the @samp{const} must be attached to the return value.
2144 @itemx constructor (@var{priority})
2145 @itemx destructor (@var{priority})
2146 @cindex @code{constructor} function attribute
2147 @cindex @code{destructor} function attribute
2148 The @code{constructor} attribute causes the function to be called
2149 automatically before execution enters @code{main ()}. Similarly, the
2150 @code{destructor} attribute causes the function to be called
2151 automatically after @code{main ()} has completed or @code{exit ()} has
2152 been called. Functions with these attributes are useful for
2153 initializing data that will be used implicitly during the execution of
2156 You may provide an optional integer priority to control the order in
2157 which constructor and destructor functions are run. A constructor
2158 with a smaller priority number runs before a constructor with a larger
2159 priority number; the opposite relationship holds for destructors. So,
2160 if you have a constructor that allocates a resource and a destructor
2161 that deallocates the same resource, both functions typically have the
2162 same priority. The priorities for constructor and destructor
2163 functions are the same as those specified for namespace-scope C++
2164 objects (@pxref{C++ Attributes}).
2166 These attributes are not currently implemented for Objective-C@.
2169 @itemx deprecated (@var{msg})
2170 @cindex @code{deprecated} attribute.
2171 The @code{deprecated} attribute results in a warning if the function
2172 is used anywhere in the source file. This is useful when identifying
2173 functions that are expected to be removed in a future version of a
2174 program. The warning also includes the location of the declaration
2175 of the deprecated function, to enable users to easily find further
2176 information about why the function is deprecated, or what they should
2177 do instead. Note that the warnings only occurs for uses:
2180 int old_fn () __attribute__ ((deprecated));
2182 int (*fn_ptr)() = old_fn;
2185 results in a warning on line 3 but not line 2. The optional msg
2186 argument, which must be a string, will be printed in the warning if
2189 The @code{deprecated} attribute can also be used for variables and
2190 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2193 @cindex @code{disinterrupt} attribute
2194 On MeP targets, this attribute causes the compiler to emit
2195 instructions to disable interrupts for the duration of the given
2199 @cindex @code{__declspec(dllexport)}
2200 On Microsoft Windows targets and Symbian OS targets the
2201 @code{dllexport} attribute causes the compiler to provide a global
2202 pointer to a pointer in a DLL, so that it can be referenced with the
2203 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
2204 name is formed by combining @code{_imp__} and the function or variable
2207 You can use @code{__declspec(dllexport)} as a synonym for
2208 @code{__attribute__ ((dllexport))} for compatibility with other
2211 On systems that support the @code{visibility} attribute, this
2212 attribute also implies ``default'' visibility. It is an error to
2213 explicitly specify any other visibility.
2215 Currently, the @code{dllexport} attribute is ignored for inlined
2216 functions, unless the @option{-fkeep-inline-functions} flag has been
2217 used. The attribute is also ignored for undefined symbols.
2219 When applied to C++ classes, the attribute marks defined non-inlined
2220 member functions and static data members as exports. Static consts
2221 initialized in-class are not marked unless they are also defined
2224 For Microsoft Windows targets there are alternative methods for
2225 including the symbol in the DLL's export table such as using a
2226 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
2227 the @option{--export-all} linker flag.
2230 @cindex @code{__declspec(dllimport)}
2231 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
2232 attribute causes the compiler to reference a function or variable via
2233 a global pointer to a pointer that is set up by the DLL exporting the
2234 symbol. The attribute implies @code{extern}. On Microsoft Windows
2235 targets, the pointer name is formed by combining @code{_imp__} and the
2236 function or variable name.
2238 You can use @code{__declspec(dllimport)} as a synonym for
2239 @code{__attribute__ ((dllimport))} for compatibility with other
2242 On systems that support the @code{visibility} attribute, this
2243 attribute also implies ``default'' visibility. It is an error to
2244 explicitly specify any other visibility.
2246 Currently, the attribute is ignored for inlined functions. If the
2247 attribute is applied to a symbol @emph{definition}, an error is reported.
2248 If a symbol previously declared @code{dllimport} is later defined, the
2249 attribute is ignored in subsequent references, and a warning is emitted.
2250 The attribute is also overridden by a subsequent declaration as
2253 When applied to C++ classes, the attribute marks non-inlined
2254 member functions and static data members as imports. However, the
2255 attribute is ignored for virtual methods to allow creation of vtables
2258 On the SH Symbian OS target the @code{dllimport} attribute also has
2259 another affect---it can cause the vtable and run-time type information
2260 for a class to be exported. This happens when the class has a
2261 dllimport'ed constructor or a non-inline, non-pure virtual function
2262 and, for either of those two conditions, the class also has an inline
2263 constructor or destructor and has a key function that is defined in
2264 the current translation unit.
2266 For Microsoft Windows based targets the use of the @code{dllimport}
2267 attribute on functions is not necessary, but provides a small
2268 performance benefit by eliminating a thunk in the DLL@. The use of the
2269 @code{dllimport} attribute on imported variables was required on older
2270 versions of the GNU linker, but can now be avoided by passing the
2271 @option{--enable-auto-import} switch to the GNU linker. As with
2272 functions, using the attribute for a variable eliminates a thunk in
2275 One drawback to using this attribute is that a pointer to a
2276 @emph{variable} marked as @code{dllimport} cannot be used as a constant
2277 address. However, a pointer to a @emph{function} with the
2278 @code{dllimport} attribute can be used as a constant initializer; in
2279 this case, the address of a stub function in the import lib is
2280 referenced. On Microsoft Windows targets, the attribute can be disabled
2281 for functions by setting the @option{-mnop-fun-dllimport} flag.
2284 @cindex eight bit data on the H8/300, H8/300H, and H8S
2285 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2286 variable should be placed into the eight bit data section.
2287 The compiler will generate more efficient code for certain operations
2288 on data in the eight bit data area. Note the eight bit data area is limited to
2291 You must use GAS and GLD from GNU binutils version 2.7 or later for
2292 this attribute to work correctly.
2294 @item exception_handler
2295 @cindex exception handler functions on the Blackfin processor
2296 Use this attribute on the Blackfin to indicate that the specified function
2297 is an exception handler. The compiler will generate function entry and
2298 exit sequences suitable for use in an exception handler when this
2299 attribute is present.
2301 @item externally_visible
2302 @cindex @code{externally_visible} attribute.
2303 This attribute, attached to a global variable or function, nullifies
2304 the effect of the @option{-fwhole-program} command-line option, so the
2305 object remains visible outside the current compilation unit. If @option{-fwhole-program} is used together with @option{-flto} and @command{gold} is used as the linker plugin, @code{externally_visible} attributes are automatically added to functions (not variable yet due to a current @command{gold} issue) that are accessed outside of LTO objects according to resolution file produced by @command{gold}. For other linkers that cannot generate resolution file, explicit @code{externally_visible} attributes are still necessary.
2308 @cindex functions which handle memory bank switching
2309 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
2310 use a calling convention that takes care of switching memory banks when
2311 entering and leaving a function. This calling convention is also the
2312 default when using the @option{-mlong-calls} option.
2314 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
2315 to call and return from a function.
2317 On 68HC11 the compiler will generate a sequence of instructions
2318 to invoke a board-specific routine to switch the memory bank and call the
2319 real function. The board-specific routine simulates a @code{call}.
2320 At the end of a function, it will jump to a board-specific routine
2321 instead of using @code{rts}. The board-specific return routine simulates
2324 On MeP targets this causes the compiler to use a calling convention
2325 which assumes the called function is too far away for the built-in
2328 @item fast_interrupt
2329 @cindex interrupt handler functions
2330 Use this attribute on the M32C and RX ports to indicate that the specified
2331 function is a fast interrupt handler. This is just like the
2332 @code{interrupt} attribute, except that @code{freit} is used to return
2333 instead of @code{reit}.
2336 @cindex functions that pop the argument stack on the 386
2337 On the Intel 386, the @code{fastcall} attribute causes the compiler to
2338 pass the first argument (if of integral type) in the register ECX and
2339 the second argument (if of integral type) in the register EDX@. Subsequent
2340 and other typed arguments are passed on the stack. The called function will
2341 pop the arguments off the stack. If the number of arguments is variable all
2342 arguments are pushed on the stack.
2345 @cindex functions that pop the argument stack on the 386
2346 On the Intel 386, the @code{thiscall} attribute causes the compiler to
2347 pass the first argument (if of integral type) in the register ECX.
2348 Subsequent and other typed arguments are passed on the stack. The called
2349 function will pop the arguments off the stack.
2350 If the number of arguments is variable all arguments are pushed on the
2352 The @code{thiscall} attribute is intended for C++ non-static member functions.
2353 As gcc extension this calling convention can be used for C-functions
2354 and for static member methods.
2356 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2357 @cindex @code{format} function attribute
2359 The @code{format} attribute specifies that a function takes @code{printf},
2360 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
2361 should be type-checked against a format string. For example, the
2366 my_printf (void *my_object, const char *my_format, ...)
2367 __attribute__ ((format (printf, 2, 3)));
2371 causes the compiler to check the arguments in calls to @code{my_printf}
2372 for consistency with the @code{printf} style format string argument
2375 The parameter @var{archetype} determines how the format string is
2376 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime},
2377 @code{gnu_printf}, @code{gnu_scanf}, @code{gnu_strftime} or
2378 @code{strfmon}. (You can also use @code{__printf__},
2379 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) On
2380 MinGW targets, @code{ms_printf}, @code{ms_scanf}, and
2381 @code{ms_strftime} are also present.
2382 @var{archtype} values such as @code{printf} refer to the formats accepted
2383 by the system's C run-time library, while @code{gnu_} values always refer
2384 to the formats accepted by the GNU C Library. On Microsoft Windows
2385 targets, @code{ms_} values refer to the formats accepted by the
2386 @file{msvcrt.dll} library.
2387 The parameter @var{string-index}
2388 specifies which argument is the format string argument (starting
2389 from 1), while @var{first-to-check} is the number of the first
2390 argument to check against the format string. For functions
2391 where the arguments are not available to be checked (such as
2392 @code{vprintf}), specify the third parameter as zero. In this case the
2393 compiler only checks the format string for consistency. For
2394 @code{strftime} formats, the third parameter is required to be zero.
2395 Since non-static C++ methods have an implicit @code{this} argument, the
2396 arguments of such methods should be counted from two, not one, when
2397 giving values for @var{string-index} and @var{first-to-check}.
2399 In the example above, the format string (@code{my_format}) is the second
2400 argument of the function @code{my_print}, and the arguments to check
2401 start with the third argument, so the correct parameters for the format
2402 attribute are 2 and 3.
2404 @opindex ffreestanding
2405 @opindex fno-builtin
2406 The @code{format} attribute allows you to identify your own functions
2407 which take format strings as arguments, so that GCC can check the
2408 calls to these functions for errors. The compiler always (unless
2409 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
2410 for the standard library functions @code{printf}, @code{fprintf},
2411 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2412 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2413 warnings are requested (using @option{-Wformat}), so there is no need to
2414 modify the header file @file{stdio.h}. In C99 mode, the functions
2415 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2416 @code{vsscanf} are also checked. Except in strictly conforming C
2417 standard modes, the X/Open function @code{strfmon} is also checked as
2418 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2419 @xref{C Dialect Options,,Options Controlling C Dialect}.
2421 The target may provide additional types of format checks.
2422 @xref{Target Format Checks,,Format Checks Specific to Particular
2425 @item format_arg (@var{string-index})
2426 @cindex @code{format_arg} function attribute
2427 @opindex Wformat-nonliteral
2428 The @code{format_arg} attribute specifies that a function takes a format
2429 string for a @code{printf}, @code{scanf}, @code{strftime} or
2430 @code{strfmon} style function and modifies it (for example, to translate
2431 it into another language), so the result can be passed to a
2432 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2433 function (with the remaining arguments to the format function the same
2434 as they would have been for the unmodified string). For example, the
2439 my_dgettext (char *my_domain, const char *my_format)
2440 __attribute__ ((format_arg (2)));
2444 causes the compiler to check the arguments in calls to a @code{printf},
2445 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2446 format string argument is a call to the @code{my_dgettext} function, for
2447 consistency with the format string argument @code{my_format}. If the
2448 @code{format_arg} attribute had not been specified, all the compiler
2449 could tell in such calls to format functions would be that the format
2450 string argument is not constant; this would generate a warning when
2451 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2452 without the attribute.
2454 The parameter @var{string-index} specifies which argument is the format
2455 string argument (starting from one). Since non-static C++ methods have
2456 an implicit @code{this} argument, the arguments of such methods should
2457 be counted from two.
2459 The @code{format-arg} attribute allows you to identify your own
2460 functions which modify format strings, so that GCC can check the
2461 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2462 type function whose operands are a call to one of your own function.
2463 The compiler always treats @code{gettext}, @code{dgettext}, and
2464 @code{dcgettext} in this manner except when strict ISO C support is
2465 requested by @option{-ansi} or an appropriate @option{-std} option, or
2466 @option{-ffreestanding} or @option{-fno-builtin}
2467 is used. @xref{C Dialect Options,,Options
2468 Controlling C Dialect}.
2470 @item function_vector
2471 @cindex calling functions through the function vector on H8/300, M16C, M32C and SH2A processors
2472 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2473 function should be called through the function vector. Calling a
2474 function through the function vector will reduce code size, however;
2475 the function vector has a limited size (maximum 128 entries on the H8/300
2476 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2478 In SH2A target, this attribute declares a function to be called using the
2479 TBR relative addressing mode. The argument to this attribute is the entry
2480 number of the same function in a vector table containing all the TBR
2481 relative addressable functions. For the successful jump, register TBR
2482 should contain the start address of this TBR relative vector table.
2483 In the startup routine of the user application, user needs to care of this
2484 TBR register initialization. The TBR relative vector table can have at
2485 max 256 function entries. The jumps to these functions will be generated
2486 using a SH2A specific, non delayed branch instruction JSR/N @@(disp8,TBR).
2487 You must use GAS and GLD from GNU binutils version 2.7 or later for
2488 this attribute to work correctly.
2490 Please refer the example of M16C target, to see the use of this
2491 attribute while declaring a function,
2493 In an application, for a function being called once, this attribute will
2494 save at least 8 bytes of code; and if other successive calls are being
2495 made to the same function, it will save 2 bytes of code per each of these
2498 On M16C/M32C targets, the @code{function_vector} attribute declares a
2499 special page subroutine call function. Use of this attribute reduces
2500 the code size by 2 bytes for each call generated to the
2501 subroutine. The argument to the attribute is the vector number entry
2502 from the special page vector table which contains the 16 low-order
2503 bits of the subroutine's entry address. Each vector table has special
2504 page number (18 to 255) which are used in @code{jsrs} instruction.
2505 Jump addresses of the routines are generated by adding 0x0F0000 (in
2506 case of M16C targets) or 0xFF0000 (in case of M32C targets), to the 2
2507 byte addresses set in the vector table. Therefore you need to ensure
2508 that all the special page vector routines should get mapped within the
2509 address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF
2512 In the following example 2 bytes will be saved for each call to
2513 function @code{foo}.
2516 void foo (void) __attribute__((function_vector(0x18)));
2527 If functions are defined in one file and are called in another file,
2528 then be sure to write this declaration in both files.
2530 This attribute is ignored for R8C target.
2533 @cindex interrupt handler functions
2534 Use this attribute on the ARM, AVR, CRX, M32C, M32R/D, m68k, MeP, MIPS,
2535 RX and Xstormy16 ports to indicate that the specified function is an
2536 interrupt handler. The compiler will generate function entry and exit
2537 sequences suitable for use in an interrupt handler when this attribute
2540 Note, interrupt handlers for the Blackfin, H8/300, H8/300H, H8S, MicroBlaze,
2541 and SH processors can be specified via the @code{interrupt_handler} attribute.
2543 Note, on the AVR, interrupts will be enabled inside the function.
2545 Note, for the ARM, you can specify the kind of interrupt to be handled by
2546 adding an optional parameter to the interrupt attribute like this:
2549 void f () __attribute__ ((interrupt ("IRQ")));
2552 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2554 On ARMv7-M the interrupt type is ignored, and the attribute means the function
2555 may be called with a word aligned stack pointer.
2557 On MIPS targets, you can use the following attributes to modify the behavior
2558 of an interrupt handler:
2560 @item use_shadow_register_set
2561 @cindex @code{use_shadow_register_set} attribute
2562 Assume that the handler uses a shadow register set, instead of
2563 the main general-purpose registers.
2565 @item keep_interrupts_masked
2566 @cindex @code{keep_interrupts_masked} attribute
2567 Keep interrupts masked for the whole function. Without this attribute,
2568 GCC tries to reenable interrupts for as much of the function as it can.
2570 @item use_debug_exception_return
2571 @cindex @code{use_debug_exception_return} attribute
2572 Return using the @code{deret} instruction. Interrupt handlers that don't
2573 have this attribute return using @code{eret} instead.
2576 You can use any combination of these attributes, as shown below:
2578 void __attribute__ ((interrupt)) v0 ();
2579 void __attribute__ ((interrupt, use_shadow_register_set)) v1 ();
2580 void __attribute__ ((interrupt, keep_interrupts_masked)) v2 ();
2581 void __attribute__ ((interrupt, use_debug_exception_return)) v3 ();
2582 void __attribute__ ((interrupt, use_shadow_register_set,
2583 keep_interrupts_masked)) v4 ();
2584 void __attribute__ ((interrupt, use_shadow_register_set,
2585 use_debug_exception_return)) v5 ();
2586 void __attribute__ ((interrupt, keep_interrupts_masked,
2587 use_debug_exception_return)) v6 ();
2588 void __attribute__ ((interrupt, use_shadow_register_set,
2589 keep_interrupts_masked,
2590 use_debug_exception_return)) v7 ();
2593 @item ifunc ("@var{resolver}")
2594 @cindex @code{ifunc} attribute
2595 The @code{ifunc} attribute is used to mark a function as an indirect
2596 function using the STT_GNU_IFUNC symbol type extension to the ELF
2597 standard. This allows the resolution of the symbol value to be
2598 determined dynamically at load time, and an optimized version of the
2599 routine can be selected for the particular processor or other system
2600 characteristics determined then. To use this attribute, first define
2601 the implementation functions available, and a resolver function that
2602 returns a pointer to the selected implementation function. The
2603 implementation functions' declarations must match the API of the
2604 function being implemented, the resolver's declaration is be a
2605 function returning pointer to void function returning void:
2608 void *my_memcpy (void *dst, const void *src, size_t len)
2613 static void (*resolve_memcpy (void)) (void)
2615 return my_memcpy; // we'll just always select this routine
2619 The exported header file declaring the function the user calls would
2623 extern void *memcpy (void *, const void *, size_t);
2626 allowing the user to call this as a regular function, unaware of the
2627 implementation. Finally, the indirect function needs to be defined in
2628 the same translation unit as the resolver function:
2631 void *memcpy (void *, const void *, size_t)
2632 __attribute__ ((ifunc ("resolve_memcpy")));
2635 Indirect functions cannot be weak, and require a recent binutils (at
2636 least version 2.20.1), and GNU C library (at least version 2.11.1).
2638 @item interrupt_handler
2639 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
2640 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
2641 indicate that the specified function is an interrupt handler. The compiler
2642 will generate function entry and exit sequences suitable for use in an
2643 interrupt handler when this attribute is present.
2645 @item interrupt_thread
2646 @cindex interrupt thread functions on fido
2647 Use this attribute on fido, a subarchitecture of the m68k, to indicate
2648 that the specified function is an interrupt handler that is designed
2649 to run as a thread. The compiler omits generate prologue/epilogue
2650 sequences and replaces the return instruction with a @code{sleep}
2651 instruction. This attribute is available only on fido.
2654 @cindex interrupt service routines on ARM
2655 Use this attribute on ARM to write Interrupt Service Routines. This is an
2656 alias to the @code{interrupt} attribute above.
2659 @cindex User stack pointer in interrupts on the Blackfin
2660 When used together with @code{interrupt_handler}, @code{exception_handler}
2661 or @code{nmi_handler}, code will be generated to load the stack pointer
2662 from the USP register in the function prologue.
2665 @cindex @code{l1_text} function attribute
2666 This attribute specifies a function to be placed into L1 Instruction
2667 SRAM@. The function will be put into a specific section named @code{.l1.text}.
2668 With @option{-mfdpic}, function calls with a such function as the callee
2669 or caller will use inlined PLT.
2672 @cindex @code{l2} function attribute
2673 On the Blackfin, this attribute specifies a function to be placed into L2
2674 SRAM. The function will be put into a specific section named
2675 @code{.l1.text}. With @option{-mfdpic}, callers of such functions will use
2679 @cindex @code{leaf} function attribute
2680 Calls to external functions with this attribute must return to the current
2681 compilation unit only by return or by exception handling. In particular, leaf
2682 functions are not allowed to call callback function passed to it from current
2683 compilation unit or directly call functions exported by the unit or longjmp
2684 into the unit. Still leaf function might call functions from other complation
2685 units and thus they are not neccesarily leaf in the sense that they contains no
2686 function calls at all.
2688 The attribute is intended for library functions to improve dataflow analysis.
2689 Compiler takes the hint that any data not escaping current compilation unit can
2690 not be used or modified by the leaf function. For example, function @code{sin}
2691 is leaf, function @code{qsort} is not.
2693 Note that the leaf functions might invoke signals and signal handlers might be
2694 defined in the current compilation unit and use static variables. Only
2695 compliant way to write such a signal handler is to declare such variables
2698 The attribute has no effect on functions defined within current compilation
2699 unit. This is to allow easy merging of multiple compilation units into one,
2700 for example, by using the link time optimization. For this reason the
2701 attribute is not allowed on types to annotate indirect calls.
2703 @item long_call/short_call
2704 @cindex indirect calls on ARM
2705 This attribute specifies how a particular function is called on
2706 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2707 command-line switch and @code{#pragma long_calls} settings. The
2708 @code{long_call} attribute indicates that the function might be far
2709 away from the call site and require a different (more expensive)
2710 calling sequence. The @code{short_call} attribute always places
2711 the offset to the function from the call site into the @samp{BL}
2712 instruction directly.
2714 @item longcall/shortcall
2715 @cindex functions called via pointer on the RS/6000 and PowerPC
2716 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
2717 indicates that the function might be far away from the call site and
2718 require a different (more expensive) calling sequence. The
2719 @code{shortcall} attribute indicates that the function is always close
2720 enough for the shorter calling sequence to be used. These attributes
2721 override both the @option{-mlongcall} switch and, on the RS/6000 and
2722 PowerPC, the @code{#pragma longcall} setting.
2724 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2725 calls are necessary.
2727 @item long_call/near/far
2728 @cindex indirect calls on MIPS
2729 These attributes specify how a particular function is called on MIPS@.
2730 The attributes override the @option{-mlong-calls} (@pxref{MIPS Options})
2731 command-line switch. The @code{long_call} and @code{far} attributes are
2732 synonyms, and cause the compiler to always call
2733 the function by first loading its address into a register, and then using
2734 the contents of that register. The @code{near} attribute has the opposite
2735 effect; it specifies that non-PIC calls should be made using the more
2736 efficient @code{jal} instruction.
2739 @cindex @code{malloc} attribute
2740 The @code{malloc} attribute is used to tell the compiler that a function
2741 may be treated as if any non-@code{NULL} pointer it returns cannot
2742 alias any other pointer valid when the function returns.
2743 This will often improve optimization.
2744 Standard functions with this property include @code{malloc} and
2745 @code{calloc}. @code{realloc}-like functions have this property as
2746 long as the old pointer is never referred to (including comparing it
2747 to the new pointer) after the function returns a non-@code{NULL}
2750 @item mips16/nomips16
2751 @cindex @code{mips16} attribute
2752 @cindex @code{nomips16} attribute
2754 On MIPS targets, you can use the @code{mips16} and @code{nomips16}
2755 function attributes to locally select or turn off MIPS16 code generation.
2756 A function with the @code{mips16} attribute is emitted as MIPS16 code,
2757 while MIPS16 code generation is disabled for functions with the
2758 @code{nomips16} attribute. These attributes override the
2759 @option{-mips16} and @option{-mno-mips16} options on the command line
2760 (@pxref{MIPS Options}).
2762 When compiling files containing mixed MIPS16 and non-MIPS16 code, the
2763 preprocessor symbol @code{__mips16} reflects the setting on the command line,
2764 not that within individual functions. Mixed MIPS16 and non-MIPS16 code
2765 may interact badly with some GCC extensions such as @code{__builtin_apply}
2766 (@pxref{Constructing Calls}).
2768 @item model (@var{model-name})
2769 @cindex function addressability on the M32R/D
2770 @cindex variable addressability on the IA-64
2772 On the M32R/D, use this attribute to set the addressability of an
2773 object, and of the code generated for a function. The identifier
2774 @var{model-name} is one of @code{small}, @code{medium}, or
2775 @code{large}, representing each of the code models.
2777 Small model objects live in the lower 16MB of memory (so that their
2778 addresses can be loaded with the @code{ld24} instruction), and are
2779 callable with the @code{bl} instruction.
2781 Medium model objects may live anywhere in the 32-bit address space (the
2782 compiler will generate @code{seth/add3} instructions to load their addresses),
2783 and are callable with the @code{bl} instruction.
2785 Large model objects may live anywhere in the 32-bit address space (the
2786 compiler will generate @code{seth/add3} instructions to load their addresses),
2787 and may not be reachable with the @code{bl} instruction (the compiler will
2788 generate the much slower @code{seth/add3/jl} instruction sequence).
2790 On IA-64, use this attribute to set the addressability of an object.
2791 At present, the only supported identifier for @var{model-name} is
2792 @code{small}, indicating addressability via ``small'' (22-bit)
2793 addresses (so that their addresses can be loaded with the @code{addl}
2794 instruction). Caveat: such addressing is by definition not position
2795 independent and hence this attribute must not be used for objects
2796 defined by shared libraries.
2798 @item ms_abi/sysv_abi
2799 @cindex @code{ms_abi} attribute
2800 @cindex @code{sysv_abi} attribute
2802 On 64-bit x86_64-*-* targets, you can use an ABI attribute to indicate
2803 which calling convention should be used for a function. The @code{ms_abi}
2804 attribute tells the compiler to use the Microsoft ABI, while the
2805 @code{sysv_abi} attribute tells the compiler to use the ABI used on
2806 GNU/Linux and other systems. The default is to use the Microsoft ABI
2807 when targeting Windows. On all other systems, the default is the AMD ABI.
2809 Note, the @code{ms_abi} attribute for Windows targets currently requires
2810 the @option{-maccumulate-outgoing-args} option.
2812 @item ms_hook_prologue
2813 @cindex @code{ms_hook_prologue} attribute
2815 On 32 bit i[34567]86-*-* targets and 64 bit x86_64-*-* targets, you can use
2816 this function attribute to make gcc generate the "hot-patching" function
2817 prologue used in Win32 API functions in Microsoft Windows XP Service Pack 2
2821 @cindex function without a prologue/epilogue code
2822 Use this attribute on the ARM, AVR, MCORE, RX and SPU ports to indicate that
2823 the specified function does not need prologue/epilogue sequences generated by
2824 the compiler. It is up to the programmer to provide these sequences. The
2825 only statements that can be safely included in naked functions are
2826 @code{asm} statements that do not have operands. All other statements,
2827 including declarations of local variables, @code{if} statements, and so
2828 forth, should be avoided. Naked functions should be used to implement the
2829 body of an assembly function, while allowing the compiler to construct
2830 the requisite function declaration for the assembler.
2833 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2834 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2835 use the normal calling convention based on @code{jsr} and @code{rts}.
2836 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2839 On MeP targets this attribute causes the compiler to assume the called
2840 function is close enough to use the normal calling convention,
2841 overriding the @code{-mtf} command line option.
2844 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2845 Use this attribute together with @code{interrupt_handler},
2846 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2847 entry code should enable nested interrupts or exceptions.
2850 @cindex NMI handler functions on the Blackfin processor
2851 Use this attribute on the Blackfin to indicate that the specified function
2852 is an NMI handler. The compiler will generate function entry and
2853 exit sequences suitable for use in an NMI handler when this
2854 attribute is present.
2856 @item no_instrument_function
2857 @cindex @code{no_instrument_function} function attribute
2858 @opindex finstrument-functions
2859 If @option{-finstrument-functions} is given, profiling function calls will
2860 be generated at entry and exit of most user-compiled functions.
2861 Functions with this attribute will not be so instrumented.
2863 @item no_split_stack
2864 @cindex @code{no_split_stack} function attribute
2865 @opindex fsplit-stack
2866 If @option{-fsplit-stack} is given, functions will have a small
2867 prologue which decides whether to split the stack. Functions with the
2868 @code{no_split_stack} attribute will not have that prologue, and thus
2869 may run with only a small amount of stack space available.
2872 @cindex @code{noinline} function attribute
2873 This function attribute prevents a function from being considered for
2875 @c Don't enumerate the optimizations by name here; we try to be
2876 @c future-compatible with this mechanism.
2877 If the function does not have side-effects, there are optimizations
2878 other than inlining that causes function calls to be optimized away,
2879 although the function call is live. To keep such calls from being
2884 (@pxref{Extended Asm}) in the called function, to serve as a special
2888 @cindex @code{noclone} function attribute
2889 This function attribute prevents a function from being considered for
2890 cloning - a mechanism which produces specialized copies of functions
2891 and which is (currently) performed by interprocedural constant
2894 @item nonnull (@var{arg-index}, @dots{})
2895 @cindex @code{nonnull} function attribute
2896 The @code{nonnull} attribute specifies that some function parameters should
2897 be non-null pointers. For instance, the declaration:
2901 my_memcpy (void *dest, const void *src, size_t len)
2902 __attribute__((nonnull (1, 2)));
2906 causes the compiler to check that, in calls to @code{my_memcpy},
2907 arguments @var{dest} and @var{src} are non-null. If the compiler
2908 determines that a null pointer is passed in an argument slot marked
2909 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2910 is issued. The compiler may also choose to make optimizations based
2911 on the knowledge that certain function arguments will not be null.
2913 If no argument index list is given to the @code{nonnull} attribute,
2914 all pointer arguments are marked as non-null. To illustrate, the
2915 following declaration is equivalent to the previous example:
2919 my_memcpy (void *dest, const void *src, size_t len)
2920 __attribute__((nonnull));
2924 @cindex @code{noreturn} function attribute
2925 A few standard library functions, such as @code{abort} and @code{exit},
2926 cannot return. GCC knows this automatically. Some programs define
2927 their own functions that never return. You can declare them
2928 @code{noreturn} to tell the compiler this fact. For example,
2932 void fatal () __attribute__ ((noreturn));
2935 fatal (/* @r{@dots{}} */)
2937 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2943 The @code{noreturn} keyword tells the compiler to assume that
2944 @code{fatal} cannot return. It can then optimize without regard to what
2945 would happen if @code{fatal} ever did return. This makes slightly
2946 better code. More importantly, it helps avoid spurious warnings of
2947 uninitialized variables.
2949 The @code{noreturn} keyword does not affect the exceptional path when that
2950 applies: a @code{noreturn}-marked function may still return to the caller
2951 by throwing an exception or calling @code{longjmp}.
2953 Do not assume that registers saved by the calling function are
2954 restored before calling the @code{noreturn} function.
2956 It does not make sense for a @code{noreturn} function to have a return
2957 type other than @code{void}.
2959 The attribute @code{noreturn} is not implemented in GCC versions
2960 earlier than 2.5. An alternative way to declare that a function does
2961 not return, which works in the current version and in some older
2962 versions, is as follows:
2965 typedef void voidfn ();
2967 volatile voidfn fatal;
2970 This approach does not work in GNU C++.
2973 @cindex @code{nothrow} function attribute
2974 The @code{nothrow} attribute is used to inform the compiler that a
2975 function cannot throw an exception. For example, most functions in
2976 the standard C library can be guaranteed not to throw an exception
2977 with the notable exceptions of @code{qsort} and @code{bsearch} that
2978 take function pointer arguments. The @code{nothrow} attribute is not
2979 implemented in GCC versions earlier than 3.3.
2982 @cindex @code{optimize} function attribute
2983 The @code{optimize} attribute is used to specify that a function is to
2984 be compiled with different optimization options than specified on the
2985 command line. Arguments can either be numbers or strings. Numbers
2986 are assumed to be an optimization level. Strings that begin with
2987 @code{O} are assumed to be an optimization option, while other options
2988 are assumed to be used with a @code{-f} prefix. You can also use the
2989 @samp{#pragma GCC optimize} pragma to set the optimization options
2990 that affect more than one function.
2991 @xref{Function Specific Option Pragmas}, for details about the
2992 @samp{#pragma GCC optimize} pragma.
2994 This can be used for instance to have frequently executed functions
2995 compiled with more aggressive optimization options that produce faster
2996 and larger code, while other functions can be called with less
3000 @cindex @code{pcs} function attribute
3002 The @code{pcs} attribute can be used to control the calling convention
3003 used for a function on ARM. The attribute takes an argument that specifies
3004 the calling convention to use.
3006 When compiling using the AAPCS ABI (or a variant of that) then valid
3007 values for the argument are @code{"aapcs"} and @code{"aapcs-vfp"}. In
3008 order to use a variant other than @code{"aapcs"} then the compiler must
3009 be permitted to use the appropriate co-processor registers (i.e., the
3010 VFP registers must be available in order to use @code{"aapcs-vfp"}).
3014 /* Argument passed in r0, and result returned in r0+r1. */
3015 double f2d (float) __attribute__((pcs("aapcs")));
3018 Variadic functions always use the @code{"aapcs"} calling convention and
3019 the compiler will reject attempts to specify an alternative.
3022 @cindex @code{pure} function attribute
3023 Many functions have no effects except the return value and their
3024 return value depends only on the parameters and/or global variables.
3025 Such a function can be subject
3026 to common subexpression elimination and loop optimization just as an
3027 arithmetic operator would be. These functions should be declared
3028 with the attribute @code{pure}. For example,
3031 int square (int) __attribute__ ((pure));
3035 says that the hypothetical function @code{square} is safe to call
3036 fewer times than the program says.
3038 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
3039 Interesting non-pure functions are functions with infinite loops or those
3040 depending on volatile memory or other system resource, that may change between
3041 two consecutive calls (such as @code{feof} in a multithreading environment).
3043 The attribute @code{pure} is not implemented in GCC versions earlier
3047 @cindex @code{hot} function attribute
3048 The @code{hot} attribute is used to inform the compiler that a function is a
3049 hot spot of the compiled program. The function is optimized more aggressively
3050 and on many target it is placed into special subsection of the text section so
3051 all hot functions appears close together improving locality.
3053 When profile feedback is available, via @option{-fprofile-use}, hot functions
3054 are automatically detected and this attribute is ignored.
3056 The @code{hot} attribute is not implemented in GCC versions earlier
3060 @cindex @code{cold} function attribute
3061 The @code{cold} attribute is used to inform the compiler that a function is
3062 unlikely executed. The function is optimized for size rather than speed and on
3063 many targets it is placed into special subsection of the text section so all
3064 cold functions appears close together improving code locality of non-cold parts
3065 of program. The paths leading to call of cold functions within code are marked
3066 as unlikely by the branch prediction mechanism. It is thus useful to mark
3067 functions used to handle unlikely conditions, such as @code{perror}, as cold to
3068 improve optimization of hot functions that do call marked functions in rare
3071 When profile feedback is available, via @option{-fprofile-use}, hot functions
3072 are automatically detected and this attribute is ignored.
3074 The @code{cold} attribute is not implemented in GCC versions earlier than 4.3.
3076 @item regparm (@var{number})
3077 @cindex @code{regparm} attribute
3078 @cindex functions that are passed arguments in registers on the 386
3079 On the Intel 386, the @code{regparm} attribute causes the compiler to
3080 pass arguments number one to @var{number} if they are of integral type
3081 in registers EAX, EDX, and ECX instead of on the stack. Functions that
3082 take a variable number of arguments will continue to be passed all of their
3083 arguments on the stack.
3085 Beware that on some ELF systems this attribute is unsuitable for
3086 global functions in shared libraries with lazy binding (which is the
3087 default). Lazy binding will send the first call via resolving code in
3088 the loader, which might assume EAX, EDX and ECX can be clobbered, as
3089 per the standard calling conventions. Solaris 8 is affected by this.
3090 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
3091 safe since the loaders there save EAX, EDX and ECX. (Lazy binding can be
3092 disabled with the linker or the loader if desired, to avoid the
3096 @cindex @code{sseregparm} attribute
3097 On the Intel 386 with SSE support, the @code{sseregparm} attribute
3098 causes the compiler to pass up to 3 floating point arguments in
3099 SSE registers instead of on the stack. Functions that take a
3100 variable number of arguments will continue to pass all of their
3101 floating point arguments on the stack.
3103 @item force_align_arg_pointer
3104 @cindex @code{force_align_arg_pointer} attribute
3105 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
3106 applied to individual function definitions, generating an alternate
3107 prologue and epilogue that realigns the runtime stack if necessary.
3108 This supports mixing legacy codes that run with a 4-byte aligned stack
3109 with modern codes that keep a 16-byte stack for SSE compatibility.
3112 @cindex @code{resbank} attribute
3113 On the SH2A target, this attribute enables the high-speed register
3114 saving and restoration using a register bank for @code{interrupt_handler}
3115 routines. Saving to the bank is performed automatically after the CPU
3116 accepts an interrupt that uses a register bank.
3118 The nineteen 32-bit registers comprising general register R0 to R14,
3119 control register GBR, and system registers MACH, MACL, and PR and the
3120 vector table address offset are saved into a register bank. Register
3121 banks are stacked in first-in last-out (FILO) sequence. Restoration
3122 from the bank is executed by issuing a RESBANK instruction.
3125 @cindex @code{returns_twice} attribute
3126 The @code{returns_twice} attribute tells the compiler that a function may
3127 return more than one time. The compiler will ensure that all registers
3128 are dead before calling such a function and will emit a warning about
3129 the variables that may be clobbered after the second return from the
3130 function. Examples of such functions are @code{setjmp} and @code{vfork}.
3131 The @code{longjmp}-like counterpart of such function, if any, might need
3132 to be marked with the @code{noreturn} attribute.
3135 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
3136 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
3137 all registers except the stack pointer should be saved in the prologue
3138 regardless of whether they are used or not.
3140 @item save_volatiles
3141 @cindex save volatile registers on the MicroBlaze
3142 Use this attribute on the MicroBlaze to indicate that the function is
3143 an interrupt handler. All volatile registers (in addition to non-volatile
3144 registers) will be saved in the function prologue. If the function is a leaf
3145 function, only volatiles used by the function are saved. A normal function
3146 return is generated instead of a return from interrupt.
3148 @item section ("@var{section-name}")
3149 @cindex @code{section} function attribute
3150 Normally, the compiler places the code it generates in the @code{text} section.
3151 Sometimes, however, you need additional sections, or you need certain
3152 particular functions to appear in special sections. The @code{section}
3153 attribute specifies that a function lives in a particular section.
3154 For example, the declaration:
3157 extern void foobar (void) __attribute__ ((section ("bar")));
3161 puts the function @code{foobar} in the @code{bar} section.
3163 Some file formats do not support arbitrary sections so the @code{section}
3164 attribute is not available on all platforms.
3165 If you need to map the entire contents of a module to a particular
3166 section, consider using the facilities of the linker instead.
3169 @cindex @code{sentinel} function attribute
3170 This function attribute ensures that a parameter in a function call is
3171 an explicit @code{NULL}. The attribute is only valid on variadic
3172 functions. By default, the sentinel is located at position zero, the
3173 last parameter of the function call. If an optional integer position
3174 argument P is supplied to the attribute, the sentinel must be located at
3175 position P counting backwards from the end of the argument list.
3178 __attribute__ ((sentinel))
3180 __attribute__ ((sentinel(0)))
3183 The attribute is automatically set with a position of 0 for the built-in
3184 functions @code{execl} and @code{execlp}. The built-in function
3185 @code{execle} has the attribute set with a position of 1.
3187 A valid @code{NULL} in this context is defined as zero with any pointer
3188 type. If your system defines the @code{NULL} macro with an integer type
3189 then you need to add an explicit cast. GCC replaces @code{stddef.h}
3190 with a copy that redefines NULL appropriately.
3192 The warnings for missing or incorrect sentinels are enabled with
3196 See long_call/short_call.
3199 See longcall/shortcall.
3202 @cindex signal handler functions on the AVR processors
3203 Use this attribute on the AVR to indicate that the specified
3204 function is a signal handler. The compiler will generate function
3205 entry and exit sequences suitable for use in a signal handler when this
3206 attribute is present. Interrupts will be disabled inside the function.
3209 Use this attribute on the SH to indicate an @code{interrupt_handler}
3210 function should switch to an alternate stack. It expects a string
3211 argument that names a global variable holding the address of the
3216 void f () __attribute__ ((interrupt_handler,
3217 sp_switch ("alt_stack")));
3221 @cindex functions that pop the argument stack on the 386
3222 On the Intel 386, the @code{stdcall} attribute causes the compiler to
3223 assume that the called function will pop off the stack space used to
3224 pass arguments, unless it takes a variable number of arguments.
3226 @item syscall_linkage
3227 @cindex @code{syscall_linkage} attribute
3228 This attribute is used to modify the IA64 calling convention by marking
3229 all input registers as live at all function exits. This makes it possible
3230 to restart a system call after an interrupt without having to save/restore
3231 the input registers. This also prevents kernel data from leaking into
3235 @cindex @code{target} function attribute
3236 The @code{target} attribute is used to specify that a function is to
3237 be compiled with different target options than specified on the
3238 command line. This can be used for instance to have functions
3239 compiled with a different ISA (instruction set architecture) than the
3240 default. You can also use the @samp{#pragma GCC target} pragma to set
3241 more than one function to be compiled with specific target options.
3242 @xref{Function Specific Option Pragmas}, for details about the
3243 @samp{#pragma GCC target} pragma.
3245 For instance on a 386, you could compile one function with
3246 @code{target("sse4.1,arch=core2")} and another with
3247 @code{target("sse4a,arch=amdfam10")} that would be equivalent to
3248 compiling the first function with @option{-msse4.1} and
3249 @option{-march=core2} options, and the second function with
3250 @option{-msse4a} and @option{-march=amdfam10} options. It is up to the
3251 user to make sure that a function is only invoked on a machine that
3252 supports the particular ISA it was compiled for (for example by using
3253 @code{cpuid} on 386 to determine what feature bits and architecture
3257 int core2_func (void) __attribute__ ((__target__ ("arch=core2")));
3258 int sse3_func (void) __attribute__ ((__target__ ("sse3")));
3261 On the 386, the following options are allowed:
3266 @cindex @code{target("abm")} attribute
3267 Enable/disable the generation of the advanced bit instructions.
3271 @cindex @code{target("aes")} attribute
3272 Enable/disable the generation of the AES instructions.
3276 @cindex @code{target("mmx")} attribute
3277 Enable/disable the generation of the MMX instructions.
3281 @cindex @code{target("pclmul")} attribute
3282 Enable/disable the generation of the PCLMUL instructions.
3286 @cindex @code{target("popcnt")} attribute
3287 Enable/disable the generation of the POPCNT instruction.
3291 @cindex @code{target("sse")} attribute
3292 Enable/disable the generation of the SSE instructions.
3296 @cindex @code{target("sse2")} attribute
3297 Enable/disable the generation of the SSE2 instructions.
3301 @cindex @code{target("sse3")} attribute
3302 Enable/disable the generation of the SSE3 instructions.
3306 @cindex @code{target("sse4")} attribute
3307 Enable/disable the generation of the SSE4 instructions (both SSE4.1
3312 @cindex @code{target("sse4.1")} attribute
3313 Enable/disable the generation of the sse4.1 instructions.
3317 @cindex @code{target("sse4.2")} attribute
3318 Enable/disable the generation of the sse4.2 instructions.
3322 @cindex @code{target("sse4a")} attribute
3323 Enable/disable the generation of the SSE4A instructions.
3327 @cindex @code{target("fma4")} attribute
3328 Enable/disable the generation of the FMA4 instructions.
3332 @cindex @code{target("xop")} attribute
3333 Enable/disable the generation of the XOP instructions.
3337 @cindex @code{target("lwp")} attribute
3338 Enable/disable the generation of the LWP instructions.
3342 @cindex @code{target("ssse3")} attribute
3343 Enable/disable the generation of the SSSE3 instructions.
3347 @cindex @code{target("cld")} attribute
3348 Enable/disable the generation of the CLD before string moves.
3350 @item fancy-math-387
3351 @itemx no-fancy-math-387
3352 @cindex @code{target("fancy-math-387")} attribute
3353 Enable/disable the generation of the @code{sin}, @code{cos}, and
3354 @code{sqrt} instructions on the 387 floating point unit.
3357 @itemx no-fused-madd
3358 @cindex @code{target("fused-madd")} attribute
3359 Enable/disable the generation of the fused multiply/add instructions.
3363 @cindex @code{target("ieee-fp")} attribute
3364 Enable/disable the generation of floating point that depends on IEEE arithmetic.
3366 @item inline-all-stringops
3367 @itemx no-inline-all-stringops
3368 @cindex @code{target("inline-all-stringops")} attribute
3369 Enable/disable inlining of string operations.
3371 @item inline-stringops-dynamically
3372 @itemx no-inline-stringops-dynamically
3373 @cindex @code{target("inline-stringops-dynamically")} attribute
3374 Enable/disable the generation of the inline code to do small string
3375 operations and calling the library routines for large operations.
3377 @item align-stringops
3378 @itemx no-align-stringops
3379 @cindex @code{target("align-stringops")} attribute
3380 Do/do not align destination of inlined string operations.
3384 @cindex @code{target("recip")} attribute
3385 Enable/disable the generation of RCPSS, RCPPS, RSQRTSS and RSQRTPS
3386 instructions followed an additional Newton-Raphson step instead of
3387 doing a floating point division.
3389 @item arch=@var{ARCH}
3390 @cindex @code{target("arch=@var{ARCH}")} attribute
3391 Specify the architecture to generate code for in compiling the function.
3393 @item tune=@var{TUNE}
3394 @cindex @code{target("tune=@var{TUNE}")} attribute
3395 Specify the architecture to tune for in compiling the function.
3397 @item fpmath=@var{FPMATH}
3398 @cindex @code{target("fpmath=@var{FPMATH}")} attribute
3399 Specify which floating point unit to use. The
3400 @code{target("fpmath=sse,387")} option must be specified as
3401 @code{target("fpmath=sse+387")} because the comma would separate
3405 On the 386, you can use either multiple strings to specify multiple
3406 options, or you can separate the option with a comma (@code{,}).
3408 On the 386, the inliner will not inline a function that has different
3409 target options than the caller, unless the callee has a subset of the
3410 target options of the caller. For example a function declared with
3411 @code{target("sse3")} can inline a function with
3412 @code{target("sse2")}, since @code{-msse3} implies @code{-msse2}.
3414 The @code{target} attribute is not implemented in GCC versions earlier
3415 than 4.4, and at present only the 386 uses it.
3418 @cindex tiny data section on the H8/300H and H8S
3419 Use this attribute on the H8/300H and H8S to indicate that the specified
3420 variable should be placed into the tiny data section.
3421 The compiler will generate more efficient code for loads and stores
3422 on data in the tiny data section. Note the tiny data area is limited to
3423 slightly under 32kbytes of data.
3426 Use this attribute on the SH for an @code{interrupt_handler} to return using
3427 @code{trapa} instead of @code{rte}. This attribute expects an integer
3428 argument specifying the trap number to be used.
3431 @cindex @code{unused} attribute.
3432 This attribute, attached to a function, means that the function is meant
3433 to be possibly unused. GCC will not produce a warning for this
3437 @cindex @code{used} attribute.
3438 This attribute, attached to a function, means that code must be emitted
3439 for the function even if it appears that the function is not referenced.
3440 This is useful, for example, when the function is referenced only in
3444 @cindex @code{version_id} attribute
3445 This IA64 HP-UX attribute, attached to a global variable or function, renames a
3446 symbol to contain a version string, thus allowing for function level
3447 versioning. HP-UX system header files may use version level functioning
3448 for some system calls.
3451 extern int foo () __attribute__((version_id ("20040821")));
3454 Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}.
3456 @item visibility ("@var{visibility_type}")
3457 @cindex @code{visibility} attribute
3458 This attribute affects the linkage of the declaration to which it is attached.
3459 There are four supported @var{visibility_type} values: default,
3460 hidden, protected or internal visibility.
3463 void __attribute__ ((visibility ("protected")))
3464 f () @{ /* @r{Do something.} */; @}
3465 int i __attribute__ ((visibility ("hidden")));
3468 The possible values of @var{visibility_type} correspond to the
3469 visibility settings in the ELF gABI.
3472 @c keep this list of visibilities in alphabetical order.
3475 Default visibility is the normal case for the object file format.
3476 This value is available for the visibility attribute to override other
3477 options that may change the assumed visibility of entities.
3479 On ELF, default visibility means that the declaration is visible to other
3480 modules and, in shared libraries, means that the declared entity may be
3483 On Darwin, default visibility means that the declaration is visible to
3486 Default visibility corresponds to ``external linkage'' in the language.
3489 Hidden visibility indicates that the entity declared will have a new
3490 form of linkage, which we'll call ``hidden linkage''. Two
3491 declarations of an object with hidden linkage refer to the same object
3492 if they are in the same shared object.
3495 Internal visibility is like hidden visibility, but with additional
3496 processor specific semantics. Unless otherwise specified by the
3497 psABI, GCC defines internal visibility to mean that a function is
3498 @emph{never} called from another module. Compare this with hidden
3499 functions which, while they cannot be referenced directly by other
3500 modules, can be referenced indirectly via function pointers. By
3501 indicating that a function cannot be called from outside the module,
3502 GCC may for instance omit the load of a PIC register since it is known
3503 that the calling function loaded the correct value.
3506 Protected visibility is like default visibility except that it
3507 indicates that references within the defining module will bind to the
3508 definition in that module. That is, the declared entity cannot be
3509 overridden by another module.
3513 All visibilities are supported on many, but not all, ELF targets
3514 (supported when the assembler supports the @samp{.visibility}
3515 pseudo-op). Default visibility is supported everywhere. Hidden
3516 visibility is supported on Darwin targets.
3518 The visibility attribute should be applied only to declarations which
3519 would otherwise have external linkage. The attribute should be applied
3520 consistently, so that the same entity should not be declared with
3521 different settings of the attribute.
3523 In C++, the visibility attribute applies to types as well as functions
3524 and objects, because in C++ types have linkage. A class must not have
3525 greater visibility than its non-static data member types and bases,
3526 and class members default to the visibility of their class. Also, a
3527 declaration without explicit visibility is limited to the visibility
3530 In C++, you can mark member functions and static member variables of a
3531 class with the visibility attribute. This is useful if you know a
3532 particular method or static member variable should only be used from
3533 one shared object; then you can mark it hidden while the rest of the
3534 class has default visibility. Care must be taken to avoid breaking
3535 the One Definition Rule; for example, it is usually not useful to mark
3536 an inline method as hidden without marking the whole class as hidden.
3538 A C++ namespace declaration can also have the visibility attribute.
3539 This attribute applies only to the particular namespace body, not to
3540 other definitions of the same namespace; it is equivalent to using
3541 @samp{#pragma GCC visibility} before and after the namespace
3542 definition (@pxref{Visibility Pragmas}).
3544 In C++, if a template argument has limited visibility, this
3545 restriction is implicitly propagated to the template instantiation.
3546 Otherwise, template instantiations and specializations default to the
3547 visibility of their template.
3549 If both the template and enclosing class have explicit visibility, the
3550 visibility from the template is used.
3553 @cindex @code{vliw} attribute
3554 On MeP, the @code{vliw} attribute tells the compiler to emit
3555 instructions in VLIW mode instead of core mode. Note that this
3556 attribute is not allowed unless a VLIW coprocessor has been configured
3557 and enabled through command line options.
3559 @item warn_unused_result
3560 @cindex @code{warn_unused_result} attribute
3561 The @code{warn_unused_result} attribute causes a warning to be emitted
3562 if a caller of the function with this attribute does not use its
3563 return value. This is useful for functions where not checking
3564 the result is either a security problem or always a bug, such as
3568 int fn () __attribute__ ((warn_unused_result));
3571 if (fn () < 0) return -1;
3577 results in warning on line 5.
3580 @cindex @code{weak} attribute
3581 The @code{weak} attribute causes the declaration to be emitted as a weak
3582 symbol rather than a global. This is primarily useful in defining
3583 library functions which can be overridden in user code, though it can
3584 also be used with non-function declarations. Weak symbols are supported
3585 for ELF targets, and also for a.out targets when using the GNU assembler
3589 @itemx weakref ("@var{target}")
3590 @cindex @code{weakref} attribute
3591 The @code{weakref} attribute marks a declaration as a weak reference.
3592 Without arguments, it should be accompanied by an @code{alias} attribute
3593 naming the target symbol. Optionally, the @var{target} may be given as
3594 an argument to @code{weakref} itself. In either case, @code{weakref}
3595 implicitly marks the declaration as @code{weak}. Without a
3596 @var{target}, given as an argument to @code{weakref} or to @code{alias},
3597 @code{weakref} is equivalent to @code{weak}.
3600 static int x() __attribute__ ((weakref ("y")));
3601 /* is equivalent to... */
3602 static int x() __attribute__ ((weak, weakref, alias ("y")));
3604 static int x() __attribute__ ((weakref));
3605 static int x() __attribute__ ((alias ("y")));
3608 A weak reference is an alias that does not by itself require a
3609 definition to be given for the target symbol. If the target symbol is
3610 only referenced through weak references, then it becomes a @code{weak}
3611 undefined symbol. If it is directly referenced, however, then such
3612 strong references prevail, and a definition will be required for the
3613 symbol, not necessarily in the same translation unit.
3615 The effect is equivalent to moving all references to the alias to a
3616 separate translation unit, renaming the alias to the aliased symbol,
3617 declaring it as weak, compiling the two separate translation units and
3618 performing a reloadable link on them.
3620 At present, a declaration to which @code{weakref} is attached can
3621 only be @code{static}.
3625 You can specify multiple attributes in a declaration by separating them
3626 by commas within the double parentheses or by immediately following an
3627 attribute declaration with another attribute declaration.
3629 @cindex @code{#pragma}, reason for not using
3630 @cindex pragma, reason for not using
3631 Some people object to the @code{__attribute__} feature, suggesting that
3632 ISO C's @code{#pragma} should be used instead. At the time
3633 @code{__attribute__} was designed, there were two reasons for not doing
3638 It is impossible to generate @code{#pragma} commands from a macro.
3641 There is no telling what the same @code{#pragma} might mean in another
3645 These two reasons applied to almost any application that might have been
3646 proposed for @code{#pragma}. It was basically a mistake to use
3647 @code{#pragma} for @emph{anything}.
3649 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
3650 to be generated from macros. In addition, a @code{#pragma GCC}
3651 namespace is now in use for GCC-specific pragmas. However, it has been
3652 found convenient to use @code{__attribute__} to achieve a natural
3653 attachment of attributes to their corresponding declarations, whereas
3654 @code{#pragma GCC} is of use for constructs that do not naturally form
3655 part of the grammar. @xref{Other Directives,,Miscellaneous
3656 Preprocessing Directives, cpp, The GNU C Preprocessor}.
3658 @node Attribute Syntax
3659 @section Attribute Syntax
3660 @cindex attribute syntax
3662 This section describes the syntax with which @code{__attribute__} may be
3663 used, and the constructs to which attribute specifiers bind, for the C
3664 language. Some details may vary for C++ and Objective-C@. Because of
3665 infelicities in the grammar for attributes, some forms described here
3666 may not be successfully parsed in all cases.
3668 There are some problems with the semantics of attributes in C++. For
3669 example, there are no manglings for attributes, although they may affect
3670 code generation, so problems may arise when attributed types are used in
3671 conjunction with templates or overloading. Similarly, @code{typeid}
3672 does not distinguish between types with different attributes. Support
3673 for attributes in C++ may be restricted in future to attributes on
3674 declarations only, but not on nested declarators.
3676 @xref{Function Attributes}, for details of the semantics of attributes
3677 applying to functions. @xref{Variable Attributes}, for details of the
3678 semantics of attributes applying to variables. @xref{Type Attributes},
3679 for details of the semantics of attributes applying to structure, union
3680 and enumerated types.
3682 An @dfn{attribute specifier} is of the form
3683 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
3684 is a possibly empty comma-separated sequence of @dfn{attributes}, where
3685 each attribute is one of the following:
3689 Empty. Empty attributes are ignored.
3692 A word (which may be an identifier such as @code{unused}, or a reserved
3693 word such as @code{const}).
3696 A word, followed by, in parentheses, parameters for the attribute.
3697 These parameters take one of the following forms:
3701 An identifier. For example, @code{mode} attributes use this form.
3704 An identifier followed by a comma and a non-empty comma-separated list
3705 of expressions. For example, @code{format} attributes use this form.
3708 A possibly empty comma-separated list of expressions. For example,
3709 @code{format_arg} attributes use this form with the list being a single
3710 integer constant expression, and @code{alias} attributes use this form
3711 with the list being a single string constant.
3715 An @dfn{attribute specifier list} is a sequence of one or more attribute
3716 specifiers, not separated by any other tokens.
3718 In GNU C, an attribute specifier list may appear after the colon following a
3719 label, other than a @code{case} or @code{default} label. The only
3720 attribute it makes sense to use after a label is @code{unused}. This
3721 feature is intended for code generated by programs which contains labels
3722 that may be unused but which is compiled with @option{-Wall}. It would
3723 not normally be appropriate to use in it human-written code, though it
3724 could be useful in cases where the code that jumps to the label is
3725 contained within an @code{#ifdef} conditional. GNU C++ only permits
3726 attributes on labels if the attribute specifier is immediately
3727 followed by a semicolon (i.e., the label applies to an empty
3728 statement). If the semicolon is missing, C++ label attributes are
3729 ambiguous, as it is permissible for a declaration, which could begin
3730 with an attribute list, to be labelled in C++. Declarations cannot be
3731 labelled in C90 or C99, so the ambiguity does not arise there.
3733 An attribute specifier list may appear as part of a @code{struct},
3734 @code{union} or @code{enum} specifier. It may go either immediately
3735 after the @code{struct}, @code{union} or @code{enum} keyword, or after
3736 the closing brace. The former syntax is preferred.
3737 Where attribute specifiers follow the closing brace, they are considered
3738 to relate to the structure, union or enumerated type defined, not to any
3739 enclosing declaration the type specifier appears in, and the type
3740 defined is not complete until after the attribute specifiers.
3741 @c Otherwise, there would be the following problems: a shift/reduce
3742 @c conflict between attributes binding the struct/union/enum and
3743 @c binding to the list of specifiers/qualifiers; and "aligned"
3744 @c attributes could use sizeof for the structure, but the size could be
3745 @c changed later by "packed" attributes.
3747 Otherwise, an attribute specifier appears as part of a declaration,
3748 counting declarations of unnamed parameters and type names, and relates
3749 to that declaration (which may be nested in another declaration, for
3750 example in the case of a parameter declaration), or to a particular declarator
3751 within a declaration. Where an
3752 attribute specifier is applied to a parameter declared as a function or
3753 an array, it should apply to the function or array rather than the
3754 pointer to which the parameter is implicitly converted, but this is not
3755 yet correctly implemented.
3757 Any list of specifiers and qualifiers at the start of a declaration may
3758 contain attribute specifiers, whether or not such a list may in that
3759 context contain storage class specifiers. (Some attributes, however,
3760 are essentially in the nature of storage class specifiers, and only make
3761 sense where storage class specifiers may be used; for example,
3762 @code{section}.) There is one necessary limitation to this syntax: the
3763 first old-style parameter declaration in a function definition cannot
3764 begin with an attribute specifier, because such an attribute applies to
3765 the function instead by syntax described below (which, however, is not
3766 yet implemented in this case). In some other cases, attribute
3767 specifiers are permitted by this grammar but not yet supported by the
3768 compiler. All attribute specifiers in this place relate to the
3769 declaration as a whole. In the obsolescent usage where a type of
3770 @code{int} is implied by the absence of type specifiers, such a list of
3771 specifiers and qualifiers may be an attribute specifier list with no
3772 other specifiers or qualifiers.
3774 At present, the first parameter in a function prototype must have some
3775 type specifier which is not an attribute specifier; this resolves an
3776 ambiguity in the interpretation of @code{void f(int
3777 (__attribute__((foo)) x))}, but is subject to change. At present, if
3778 the parentheses of a function declarator contain only attributes then
3779 those attributes are ignored, rather than yielding an error or warning
3780 or implying a single parameter of type int, but this is subject to
3783 An attribute specifier list may appear immediately before a declarator
3784 (other than the first) in a comma-separated list of declarators in a
3785 declaration of more than one identifier using a single list of
3786 specifiers and qualifiers. Such attribute specifiers apply
3787 only to the identifier before whose declarator they appear. For
3791 __attribute__((noreturn)) void d0 (void),
3792 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
3797 the @code{noreturn} attribute applies to all the functions
3798 declared; the @code{format} attribute only applies to @code{d1}.
3800 An attribute specifier list may appear immediately before the comma,
3801 @code{=} or semicolon terminating the declaration of an identifier other
3802 than a function definition. Such attribute specifiers apply
3803 to the declared object or function. Where an
3804 assembler name for an object or function is specified (@pxref{Asm
3805 Labels}), the attribute must follow the @code{asm}
3808 An attribute specifier list may, in future, be permitted to appear after
3809 the declarator in a function definition (before any old-style parameter
3810 declarations or the function body).
3812 Attribute specifiers may be mixed with type qualifiers appearing inside
3813 the @code{[]} of a parameter array declarator, in the C99 construct by
3814 which such qualifiers are applied to the pointer to which the array is
3815 implicitly converted. Such attribute specifiers apply to the pointer,
3816 not to the array, but at present this is not implemented and they are
3819 An attribute specifier list may appear at the start of a nested
3820 declarator. At present, there are some limitations in this usage: the
3821 attributes correctly apply to the declarator, but for most individual
3822 attributes the semantics this implies are not implemented.
3823 When attribute specifiers follow the @code{*} of a pointer
3824 declarator, they may be mixed with any type qualifiers present.
3825 The following describes the formal semantics of this syntax. It will make the
3826 most sense if you are familiar with the formal specification of
3827 declarators in the ISO C standard.
3829 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
3830 D1}, where @code{T} contains declaration specifiers that specify a type
3831 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
3832 contains an identifier @var{ident}. The type specified for @var{ident}
3833 for derived declarators whose type does not include an attribute
3834 specifier is as in the ISO C standard.
3836 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
3837 and the declaration @code{T D} specifies the type
3838 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3839 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3840 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
3842 If @code{D1} has the form @code{*
3843 @var{type-qualifier-and-attribute-specifier-list} D}, and the
3844 declaration @code{T D} specifies the type
3845 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3846 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3847 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
3853 void (__attribute__((noreturn)) ****f) (void);
3857 specifies the type ``pointer to pointer to pointer to pointer to
3858 non-returning function returning @code{void}''. As another example,
3861 char *__attribute__((aligned(8))) *f;
3865 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
3866 Note again that this does not work with most attributes; for example,
3867 the usage of @samp{aligned} and @samp{noreturn} attributes given above
3868 is not yet supported.
3870 For compatibility with existing code written for compiler versions that
3871 did not implement attributes on nested declarators, some laxity is
3872 allowed in the placing of attributes. If an attribute that only applies
3873 to types is applied to a declaration, it will be treated as applying to
3874 the type of that declaration. If an attribute that only applies to
3875 declarations is applied to the type of a declaration, it will be treated
3876 as applying to that declaration; and, for compatibility with code
3877 placing the attributes immediately before the identifier declared, such
3878 an attribute applied to a function return type will be treated as
3879 applying to the function type, and such an attribute applied to an array
3880 element type will be treated as applying to the array type. If an
3881 attribute that only applies to function types is applied to a
3882 pointer-to-function type, it will be treated as applying to the pointer
3883 target type; if such an attribute is applied to a function return type
3884 that is not a pointer-to-function type, it will be treated as applying
3885 to the function type.
3887 @node Function Prototypes
3888 @section Prototypes and Old-Style Function Definitions
3889 @cindex function prototype declarations
3890 @cindex old-style function definitions
3891 @cindex promotion of formal parameters
3893 GNU C extends ISO C to allow a function prototype to override a later
3894 old-style non-prototype definition. Consider the following example:
3897 /* @r{Use prototypes unless the compiler is old-fashioned.} */
3904 /* @r{Prototype function declaration.} */
3905 int isroot P((uid_t));
3907 /* @r{Old-style function definition.} */
3909 isroot (x) /* @r{??? lossage here ???} */
3916 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
3917 not allow this example, because subword arguments in old-style
3918 non-prototype definitions are promoted. Therefore in this example the
3919 function definition's argument is really an @code{int}, which does not
3920 match the prototype argument type of @code{short}.
3922 This restriction of ISO C makes it hard to write code that is portable
3923 to traditional C compilers, because the programmer does not know
3924 whether the @code{uid_t} type is @code{short}, @code{int}, or
3925 @code{long}. Therefore, in cases like these GNU C allows a prototype
3926 to override a later old-style definition. More precisely, in GNU C, a
3927 function prototype argument type overrides the argument type specified
3928 by a later old-style definition if the former type is the same as the
3929 latter type before promotion. Thus in GNU C the above example is
3930 equivalent to the following:
3943 GNU C++ does not support old-style function definitions, so this
3944 extension is irrelevant.
3947 @section C++ Style Comments
3949 @cindex C++ comments
3950 @cindex comments, C++ style
3952 In GNU C, you may use C++ style comments, which start with @samp{//} and
3953 continue until the end of the line. Many other C implementations allow
3954 such comments, and they are included in the 1999 C standard. However,
3955 C++ style comments are not recognized if you specify an @option{-std}
3956 option specifying a version of ISO C before C99, or @option{-ansi}
3957 (equivalent to @option{-std=c90}).
3960 @section Dollar Signs in Identifier Names
3962 @cindex dollar signs in identifier names
3963 @cindex identifier names, dollar signs in
3965 In GNU C, you may normally use dollar signs in identifier names.
3966 This is because many traditional C implementations allow such identifiers.
3967 However, dollar signs in identifiers are not supported on a few target
3968 machines, typically because the target assembler does not allow them.
3970 @node Character Escapes
3971 @section The Character @key{ESC} in Constants
3973 You can use the sequence @samp{\e} in a string or character constant to
3974 stand for the ASCII character @key{ESC}.
3977 @section Inquiring on Alignment of Types or Variables
3979 @cindex type alignment
3980 @cindex variable alignment
3982 The keyword @code{__alignof__} allows you to inquire about how an object
3983 is aligned, or the minimum alignment usually required by a type. Its
3984 syntax is just like @code{sizeof}.
3986 For example, if the target machine requires a @code{double} value to be
3987 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
3988 This is true on many RISC machines. On more traditional machine
3989 designs, @code{__alignof__ (double)} is 4 or even 2.
3991 Some machines never actually require alignment; they allow reference to any
3992 data type even at an odd address. For these machines, @code{__alignof__}
3993 reports the smallest alignment that GCC will give the data type, usually as
3994 mandated by the target ABI.
3996 If the operand of @code{__alignof__} is an lvalue rather than a type,
3997 its value is the required alignment for its type, taking into account
3998 any minimum alignment specified with GCC's @code{__attribute__}
3999 extension (@pxref{Variable Attributes}). For example, after this
4003 struct foo @{ int x; char y; @} foo1;
4007 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
4008 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
4010 It is an error to ask for the alignment of an incomplete type.
4012 @node Variable Attributes
4013 @section Specifying Attributes of Variables
4014 @cindex attribute of variables
4015 @cindex variable attributes
4017 The keyword @code{__attribute__} allows you to specify special
4018 attributes of variables or structure fields. This keyword is followed
4019 by an attribute specification inside double parentheses. Some
4020 attributes are currently defined generically for variables.
4021 Other attributes are defined for variables on particular target
4022 systems. Other attributes are available for functions
4023 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
4024 Other front ends might define more attributes
4025 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
4027 You may also specify attributes with @samp{__} preceding and following
4028 each keyword. This allows you to use them in header files without
4029 being concerned about a possible macro of the same name. For example,
4030 you may use @code{__aligned__} instead of @code{aligned}.
4032 @xref{Attribute Syntax}, for details of the exact syntax for using
4036 @cindex @code{aligned} attribute
4037 @item aligned (@var{alignment})
4038 This attribute specifies a minimum alignment for the variable or
4039 structure field, measured in bytes. For example, the declaration:
4042 int x __attribute__ ((aligned (16))) = 0;
4046 causes the compiler to allocate the global variable @code{x} on a
4047 16-byte boundary. On a 68040, this could be used in conjunction with
4048 an @code{asm} expression to access the @code{move16} instruction which
4049 requires 16-byte aligned operands.
4051 You can also specify the alignment of structure fields. For example, to
4052 create a double-word aligned @code{int} pair, you could write:
4055 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
4059 This is an alternative to creating a union with a @code{double} member
4060 that forces the union to be double-word aligned.
4062 As in the preceding examples, you can explicitly specify the alignment
4063 (in bytes) that you wish the compiler to use for a given variable or
4064 structure field. Alternatively, you can leave out the alignment factor
4065 and just ask the compiler to align a variable or field to the
4066 default alignment for the target architecture you are compiling for.
4067 The default alignment is sufficient for all scalar types, but may not be
4068 enough for all vector types on a target which supports vector operations.
4069 The default alignment is fixed for a particular target ABI.
4071 Gcc also provides a target specific macro @code{__BIGGEST_ALIGNMENT__},
4072 which is the largest alignment ever used for any data type on the
4073 target machine you are compiling for. For example, you could write:
4076 short array[3] __attribute__ ((aligned (__BIGGEST_ALIGNMENT__)));
4079 The compiler automatically sets the alignment for the declared
4080 variable or field to @code{__BIGGEST_ALIGNMENT__}. Doing this can
4081 often make copy operations more efficient, because the compiler can
4082 use whatever instructions copy the biggest chunks of memory when
4083 performing copies to or from the variables or fields that you have
4084 aligned this way. Note that the value of @code{__BIGGEST_ALIGNMENT__}
4085 may change depending on command line options.
4087 When used on a struct, or struct member, the @code{aligned} attribute can
4088 only increase the alignment; in order to decrease it, the @code{packed}
4089 attribute must be specified as well. When used as part of a typedef, the
4090 @code{aligned} attribute can both increase and decrease alignment, and
4091 specifying the @code{packed} attribute will generate a warning.
4093 Note that the effectiveness of @code{aligned} attributes may be limited
4094 by inherent limitations in your linker. On many systems, the linker is
4095 only able to arrange for variables to be aligned up to a certain maximum
4096 alignment. (For some linkers, the maximum supported alignment may
4097 be very very small.) If your linker is only able to align variables
4098 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4099 in an @code{__attribute__} will still only provide you with 8 byte
4100 alignment. See your linker documentation for further information.
4102 The @code{aligned} attribute can also be used for functions
4103 (@pxref{Function Attributes}.)
4105 @item cleanup (@var{cleanup_function})
4106 @cindex @code{cleanup} attribute
4107 The @code{cleanup} attribute runs a function when the variable goes
4108 out of scope. This attribute can only be applied to auto function
4109 scope variables; it may not be applied to parameters or variables
4110 with static storage duration. The function must take one parameter,
4111 a pointer to a type compatible with the variable. The return value
4112 of the function (if any) is ignored.
4114 If @option{-fexceptions} is enabled, then @var{cleanup_function}
4115 will be run during the stack unwinding that happens during the
4116 processing of the exception. Note that the @code{cleanup} attribute
4117 does not allow the exception to be caught, only to perform an action.
4118 It is undefined what happens if @var{cleanup_function} does not
4123 @cindex @code{common} attribute
4124 @cindex @code{nocommon} attribute
4127 The @code{common} attribute requests GCC to place a variable in
4128 ``common'' storage. The @code{nocommon} attribute requests the
4129 opposite---to allocate space for it directly.
4131 These attributes override the default chosen by the
4132 @option{-fno-common} and @option{-fcommon} flags respectively.
4135 @itemx deprecated (@var{msg})
4136 @cindex @code{deprecated} attribute
4137 The @code{deprecated} attribute results in a warning if the variable
4138 is used anywhere in the source file. This is useful when identifying
4139 variables that are expected to be removed in a future version of a
4140 program. The warning also includes the location of the declaration
4141 of the deprecated variable, to enable users to easily find further
4142 information about why the variable is deprecated, or what they should
4143 do instead. Note that the warning only occurs for uses:
4146 extern int old_var __attribute__ ((deprecated));
4148 int new_fn () @{ return old_var; @}
4151 results in a warning on line 3 but not line 2. The optional msg
4152 argument, which must be a string, will be printed in the warning if
4155 The @code{deprecated} attribute can also be used for functions and
4156 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
4158 @item mode (@var{mode})
4159 @cindex @code{mode} attribute
4160 This attribute specifies the data type for the declaration---whichever
4161 type corresponds to the mode @var{mode}. This in effect lets you
4162 request an integer or floating point type according to its width.
4164 You may also specify a mode of @samp{byte} or @samp{__byte__} to
4165 indicate the mode corresponding to a one-byte integer, @samp{word} or
4166 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
4167 or @samp{__pointer__} for the mode used to represent pointers.
4170 @cindex @code{packed} attribute
4171 The @code{packed} attribute specifies that a variable or structure field
4172 should have the smallest possible alignment---one byte for a variable,
4173 and one bit for a field, unless you specify a larger value with the
4174 @code{aligned} attribute.
4176 Here is a structure in which the field @code{x} is packed, so that it
4177 immediately follows @code{a}:
4183 int x[2] __attribute__ ((packed));
4187 @emph{Note:} The 4.1, 4.2 and 4.3 series of GCC ignore the
4188 @code{packed} attribute on bit-fields of type @code{char}. This has
4189 been fixed in GCC 4.4 but the change can lead to differences in the
4190 structure layout. See the documentation of
4191 @option{-Wpacked-bitfield-compat} for more information.
4193 @item section ("@var{section-name}")
4194 @cindex @code{section} variable attribute
4195 Normally, the compiler places the objects it generates in sections like
4196 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
4197 or you need certain particular variables to appear in special sections,
4198 for example to map to special hardware. The @code{section}
4199 attribute specifies that a variable (or function) lives in a particular
4200 section. For example, this small program uses several specific section names:
4203 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
4204 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
4205 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
4206 int init_data __attribute__ ((section ("INITDATA")));
4210 /* @r{Initialize stack pointer} */
4211 init_sp (stack + sizeof (stack));
4213 /* @r{Initialize initialized data} */
4214 memcpy (&init_data, &data, &edata - &data);
4216 /* @r{Turn on the serial ports} */
4223 Use the @code{section} attribute with
4224 @emph{global} variables and not @emph{local} variables,
4225 as shown in the example.
4227 You may use the @code{section} attribute with initialized or
4228 uninitialized global variables but the linker requires
4229 each object be defined once, with the exception that uninitialized
4230 variables tentatively go in the @code{common} (or @code{bss}) section
4231 and can be multiply ``defined''. Using the @code{section} attribute
4232 will change what section the variable goes into and may cause the
4233 linker to issue an error if an uninitialized variable has multiple
4234 definitions. You can force a variable to be initialized with the
4235 @option{-fno-common} flag or the @code{nocommon} attribute.
4237 Some file formats do not support arbitrary sections so the @code{section}
4238 attribute is not available on all platforms.
4239 If you need to map the entire contents of a module to a particular
4240 section, consider using the facilities of the linker instead.
4243 @cindex @code{shared} variable attribute
4244 On Microsoft Windows, in addition to putting variable definitions in a named
4245 section, the section can also be shared among all running copies of an
4246 executable or DLL@. For example, this small program defines shared data
4247 by putting it in a named section @code{shared} and marking the section
4251 int foo __attribute__((section ("shared"), shared)) = 0;
4256 /* @r{Read and write foo. All running
4257 copies see the same value.} */
4263 You may only use the @code{shared} attribute along with @code{section}
4264 attribute with a fully initialized global definition because of the way
4265 linkers work. See @code{section} attribute for more information.
4267 The @code{shared} attribute is only available on Microsoft Windows@.
4269 @item tls_model ("@var{tls_model}")
4270 @cindex @code{tls_model} attribute
4271 The @code{tls_model} attribute sets thread-local storage model
4272 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
4273 overriding @option{-ftls-model=} command-line switch on a per-variable
4275 The @var{tls_model} argument should be one of @code{global-dynamic},
4276 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
4278 Not all targets support this attribute.
4281 This attribute, attached to a variable, means that the variable is meant
4282 to be possibly unused. GCC will not produce a warning for this
4286 This attribute, attached to a variable, means that the variable must be
4287 emitted even if it appears that the variable is not referenced.
4289 @item vector_size (@var{bytes})
4290 This attribute specifies the vector size for the variable, measured in
4291 bytes. For example, the declaration:
4294 int foo __attribute__ ((vector_size (16)));
4298 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
4299 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
4300 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
4302 This attribute is only applicable to integral and float scalars,
4303 although arrays, pointers, and function return values are allowed in
4304 conjunction with this construct.
4306 Aggregates with this attribute are invalid, even if they are of the same
4307 size as a corresponding scalar. For example, the declaration:
4310 struct S @{ int a; @};
4311 struct S __attribute__ ((vector_size (16))) foo;
4315 is invalid even if the size of the structure is the same as the size of
4319 The @code{selectany} attribute causes an initialized global variable to
4320 have link-once semantics. When multiple definitions of the variable are
4321 encountered by the linker, the first is selected and the remainder are
4322 discarded. Following usage by the Microsoft compiler, the linker is told
4323 @emph{not} to warn about size or content differences of the multiple
4326 Although the primary usage of this attribute is for POD types, the
4327 attribute can also be applied to global C++ objects that are initialized
4328 by a constructor. In this case, the static initialization and destruction
4329 code for the object is emitted in each translation defining the object,
4330 but the calls to the constructor and destructor are protected by a
4331 link-once guard variable.
4333 The @code{selectany} attribute is only available on Microsoft Windows
4334 targets. You can use @code{__declspec (selectany)} as a synonym for
4335 @code{__attribute__ ((selectany))} for compatibility with other
4339 The @code{weak} attribute is described in @ref{Function Attributes}.
4342 The @code{dllimport} attribute is described in @ref{Function Attributes}.
4345 The @code{dllexport} attribute is described in @ref{Function Attributes}.
4349 @subsection Blackfin Variable Attributes
4351 Three attributes are currently defined for the Blackfin.
4357 @cindex @code{l1_data} variable attribute
4358 @cindex @code{l1_data_A} variable attribute
4359 @cindex @code{l1_data_B} variable attribute
4360 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
4361 Variables with @code{l1_data} attribute will be put into the specific section
4362 named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into
4363 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
4364 attribute will be put into the specific section named @code{.l1.data.B}.
4367 @cindex @code{l2} variable attribute
4368 Use this attribute on the Blackfin to place the variable into L2 SRAM.
4369 Variables with @code{l2} attribute will be put into the specific section
4370 named @code{.l2.data}.
4373 @subsection M32R/D Variable Attributes
4375 One attribute is currently defined for the M32R/D@.
4378 @item model (@var{model-name})
4379 @cindex variable addressability on the M32R/D
4380 Use this attribute on the M32R/D to set the addressability of an object.
4381 The identifier @var{model-name} is one of @code{small}, @code{medium},
4382 or @code{large}, representing each of the code models.
4384 Small model objects live in the lower 16MB of memory (so that their
4385 addresses can be loaded with the @code{ld24} instruction).
4387 Medium and large model objects may live anywhere in the 32-bit address space
4388 (the compiler will generate @code{seth/add3} instructions to load their
4392 @anchor{MeP Variable Attributes}
4393 @subsection MeP Variable Attributes
4395 The MeP target has a number of addressing modes and busses. The
4396 @code{near} space spans the standard memory space's first 16 megabytes
4397 (24 bits). The @code{far} space spans the entire 32-bit memory space.
4398 The @code{based} space is a 128 byte region in the memory space which
4399 is addressed relative to the @code{$tp} register. The @code{tiny}
4400 space is a 65536 byte region relative to the @code{$gp} register. In
4401 addition to these memory regions, the MeP target has a separate 16-bit
4402 control bus which is specified with @code{cb} attributes.
4407 Any variable with the @code{based} attribute will be assigned to the
4408 @code{.based} section, and will be accessed with relative to the
4409 @code{$tp} register.
4412 Likewise, the @code{tiny} attribute assigned variables to the
4413 @code{.tiny} section, relative to the @code{$gp} register.
4416 Variables with the @code{near} attribute are assumed to have addresses
4417 that fit in a 24-bit addressing mode. This is the default for large
4418 variables (@code{-mtiny=4} is the default) but this attribute can
4419 override @code{-mtiny=} for small variables, or override @code{-ml}.
4422 Variables with the @code{far} attribute are addressed using a full
4423 32-bit address. Since this covers the entire memory space, this
4424 allows modules to make no assumptions about where variables might be
4428 @itemx io (@var{addr})
4429 Variables with the @code{io} attribute are used to address
4430 memory-mapped peripherals. If an address is specified, the variable
4431 is assigned that address, else it is not assigned an address (it is
4432 assumed some other module will assign an address). Example:
4435 int timer_count __attribute__((io(0x123)));
4439 @itemx cb (@var{addr})
4440 Variables with the @code{cb} attribute are used to access the control
4441 bus, using special instructions. @code{addr} indicates the control bus
4445 int cpu_clock __attribute__((cb(0x123)));
4450 @anchor{i386 Variable Attributes}
4451 @subsection i386 Variable Attributes
4453 Two attributes are currently defined for i386 configurations:
4454 @code{ms_struct} and @code{gcc_struct}
4459 @cindex @code{ms_struct} attribute
4460 @cindex @code{gcc_struct} attribute
4462 If @code{packed} is used on a structure, or if bit-fields are used
4463 it may be that the Microsoft ABI packs them differently
4464 than GCC would normally pack them. Particularly when moving packed
4465 data between functions compiled with GCC and the native Microsoft compiler
4466 (either via function call or as data in a file), it may be necessary to access
4469 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4470 compilers to match the native Microsoft compiler.
4472 The Microsoft structure layout algorithm is fairly simple with the exception
4473 of the bitfield packing:
4475 The padding and alignment of members of structures and whether a bit field
4476 can straddle a storage-unit boundary
4479 @item Structure members are stored sequentially in the order in which they are
4480 declared: the first member has the lowest memory address and the last member
4483 @item Every data object has an alignment-requirement. The alignment-requirement
4484 for all data except structures, unions, and arrays is either the size of the
4485 object or the current packing size (specified with either the aligned attribute
4486 or the pack pragma), whichever is less. For structures, unions, and arrays,
4487 the alignment-requirement is the largest alignment-requirement of its members.
4488 Every object is allocated an offset so that:
4490 offset % alignment-requirement == 0
4492 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
4493 unit if the integral types are the same size and if the next bit field fits
4494 into the current allocation unit without crossing the boundary imposed by the
4495 common alignment requirements of the bit fields.
4498 Handling of zero-length bitfields:
4500 MSVC interprets zero-length bitfields in the following ways:
4503 @item If a zero-length bitfield is inserted between two bitfields that would
4504 normally be coalesced, the bitfields will not be coalesced.
4511 unsigned long bf_1 : 12;
4513 unsigned long bf_2 : 12;
4517 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
4518 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
4520 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
4521 alignment of the zero-length bitfield is greater than the member that follows it,
4522 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
4542 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
4543 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
4544 bitfield will not affect the alignment of @code{bar} or, as a result, the size
4547 Taking this into account, it is important to note the following:
4550 @item If a zero-length bitfield follows a normal bitfield, the type of the
4551 zero-length bitfield may affect the alignment of the structure as whole. For
4552 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
4553 normal bitfield, and is of type short.
4555 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
4556 still affect the alignment of the structure:
4566 Here, @code{t4} will take up 4 bytes.
4569 @item Zero-length bitfields following non-bitfield members are ignored:
4580 Here, @code{t5} will take up 2 bytes.
4584 @subsection PowerPC Variable Attributes
4586 Three attributes currently are defined for PowerPC configurations:
4587 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4589 For full documentation of the struct attributes please see the
4590 documentation in @ref{i386 Variable Attributes}.
4592 For documentation of @code{altivec} attribute please see the
4593 documentation in @ref{PowerPC Type Attributes}.
4595 @subsection SPU Variable Attributes
4597 The SPU supports the @code{spu_vector} attribute for variables. For
4598 documentation of this attribute please see the documentation in
4599 @ref{SPU Type Attributes}.
4601 @subsection Xstormy16 Variable Attributes
4603 One attribute is currently defined for xstormy16 configurations:
4608 @cindex @code{below100} attribute
4610 If a variable has the @code{below100} attribute (@code{BELOW100} is
4611 allowed also), GCC will place the variable in the first 0x100 bytes of
4612 memory and use special opcodes to access it. Such variables will be
4613 placed in either the @code{.bss_below100} section or the
4614 @code{.data_below100} section.
4618 @subsection AVR Variable Attributes
4622 @cindex @code{progmem} variable attribute
4623 The @code{progmem} attribute is used on the AVR to place data in the Program
4624 Memory address space. The AVR is a Harvard Architecture processor and data
4625 normally resides in the Data Memory address space.
4628 @node Type Attributes
4629 @section Specifying Attributes of Types
4630 @cindex attribute of types
4631 @cindex type attributes
4633 The keyword @code{__attribute__} allows you to specify special
4634 attributes of @code{struct} and @code{union} types when you define
4635 such types. This keyword is followed by an attribute specification
4636 inside double parentheses. Seven attributes are currently defined for
4637 types: @code{aligned}, @code{packed}, @code{transparent_union},
4638 @code{unused}, @code{deprecated}, @code{visibility}, and
4639 @code{may_alias}. Other attributes are defined for functions
4640 (@pxref{Function Attributes}) and for variables (@pxref{Variable
4643 You may also specify any one of these attributes with @samp{__}
4644 preceding and following its keyword. This allows you to use these
4645 attributes in header files without being concerned about a possible
4646 macro of the same name. For example, you may use @code{__aligned__}
4647 instead of @code{aligned}.
4649 You may specify type attributes in an enum, struct or union type
4650 declaration or definition, or for other types in a @code{typedef}
4653 For an enum, struct or union type, you may specify attributes either
4654 between the enum, struct or union tag and the name of the type, or
4655 just past the closing curly brace of the @emph{definition}. The
4656 former syntax is preferred.
4658 @xref{Attribute Syntax}, for details of the exact syntax for using
4662 @cindex @code{aligned} attribute
4663 @item aligned (@var{alignment})
4664 This attribute specifies a minimum alignment (in bytes) for variables
4665 of the specified type. For example, the declarations:
4668 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
4669 typedef int more_aligned_int __attribute__ ((aligned (8)));
4673 force the compiler to insure (as far as it can) that each variable whose
4674 type is @code{struct S} or @code{more_aligned_int} will be allocated and
4675 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
4676 variables of type @code{struct S} aligned to 8-byte boundaries allows
4677 the compiler to use the @code{ldd} and @code{std} (doubleword load and
4678 store) instructions when copying one variable of type @code{struct S} to
4679 another, thus improving run-time efficiency.
4681 Note that the alignment of any given @code{struct} or @code{union} type
4682 is required by the ISO C standard to be at least a perfect multiple of
4683 the lowest common multiple of the alignments of all of the members of
4684 the @code{struct} or @code{union} in question. This means that you @emph{can}
4685 effectively adjust the alignment of a @code{struct} or @code{union}
4686 type by attaching an @code{aligned} attribute to any one of the members
4687 of such a type, but the notation illustrated in the example above is a
4688 more obvious, intuitive, and readable way to request the compiler to
4689 adjust the alignment of an entire @code{struct} or @code{union} type.
4691 As in the preceding example, you can explicitly specify the alignment
4692 (in bytes) that you wish the compiler to use for a given @code{struct}
4693 or @code{union} type. Alternatively, you can leave out the alignment factor
4694 and just ask the compiler to align a type to the maximum
4695 useful alignment for the target machine you are compiling for. For
4696 example, you could write:
4699 struct S @{ short f[3]; @} __attribute__ ((aligned));
4702 Whenever you leave out the alignment factor in an @code{aligned}
4703 attribute specification, the compiler automatically sets the alignment
4704 for the type to the largest alignment which is ever used for any data
4705 type on the target machine you are compiling for. Doing this can often
4706 make copy operations more efficient, because the compiler can use
4707 whatever instructions copy the biggest chunks of memory when performing
4708 copies to or from the variables which have types that you have aligned
4711 In the example above, if the size of each @code{short} is 2 bytes, then
4712 the size of the entire @code{struct S} type is 6 bytes. The smallest
4713 power of two which is greater than or equal to that is 8, so the
4714 compiler sets the alignment for the entire @code{struct S} type to 8
4717 Note that although you can ask the compiler to select a time-efficient
4718 alignment for a given type and then declare only individual stand-alone
4719 objects of that type, the compiler's ability to select a time-efficient
4720 alignment is primarily useful only when you plan to create arrays of
4721 variables having the relevant (efficiently aligned) type. If you
4722 declare or use arrays of variables of an efficiently-aligned type, then
4723 it is likely that your program will also be doing pointer arithmetic (or
4724 subscripting, which amounts to the same thing) on pointers to the
4725 relevant type, and the code that the compiler generates for these
4726 pointer arithmetic operations will often be more efficient for
4727 efficiently-aligned types than for other types.
4729 The @code{aligned} attribute can only increase the alignment; but you
4730 can decrease it by specifying @code{packed} as well. See below.
4732 Note that the effectiveness of @code{aligned} attributes may be limited
4733 by inherent limitations in your linker. On many systems, the linker is
4734 only able to arrange for variables to be aligned up to a certain maximum
4735 alignment. (For some linkers, the maximum supported alignment may
4736 be very very small.) If your linker is only able to align variables
4737 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4738 in an @code{__attribute__} will still only provide you with 8 byte
4739 alignment. See your linker documentation for further information.
4742 This attribute, attached to @code{struct} or @code{union} type
4743 definition, specifies that each member (other than zero-width bitfields)
4744 of the structure or union is placed to minimize the memory required. When
4745 attached to an @code{enum} definition, it indicates that the smallest
4746 integral type should be used.
4748 @opindex fshort-enums
4749 Specifying this attribute for @code{struct} and @code{union} types is
4750 equivalent to specifying the @code{packed} attribute on each of the
4751 structure or union members. Specifying the @option{-fshort-enums}
4752 flag on the line is equivalent to specifying the @code{packed}
4753 attribute on all @code{enum} definitions.
4755 In the following example @code{struct my_packed_struct}'s members are
4756 packed closely together, but the internal layout of its @code{s} member
4757 is not packed---to do that, @code{struct my_unpacked_struct} would need to
4761 struct my_unpacked_struct
4767 struct __attribute__ ((__packed__)) my_packed_struct
4771 struct my_unpacked_struct s;
4775 You may only specify this attribute on the definition of an @code{enum},
4776 @code{struct} or @code{union}, not on a @code{typedef} which does not
4777 also define the enumerated type, structure or union.
4779 @item transparent_union
4780 This attribute, attached to a @code{union} type definition, indicates
4781 that any function parameter having that union type causes calls to that
4782 function to be treated in a special way.
4784 First, the argument corresponding to a transparent union type can be of
4785 any type in the union; no cast is required. Also, if the union contains
4786 a pointer type, the corresponding argument can be a null pointer
4787 constant or a void pointer expression; and if the union contains a void
4788 pointer type, the corresponding argument can be any pointer expression.
4789 If the union member type is a pointer, qualifiers like @code{const} on
4790 the referenced type must be respected, just as with normal pointer
4793 Second, the argument is passed to the function using the calling
4794 conventions of the first member of the transparent union, not the calling
4795 conventions of the union itself. All members of the union must have the
4796 same machine representation; this is necessary for this argument passing
4799 Transparent unions are designed for library functions that have multiple
4800 interfaces for compatibility reasons. For example, suppose the
4801 @code{wait} function must accept either a value of type @code{int *} to
4802 comply with Posix, or a value of type @code{union wait *} to comply with
4803 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
4804 @code{wait} would accept both kinds of arguments, but it would also
4805 accept any other pointer type and this would make argument type checking
4806 less useful. Instead, @code{<sys/wait.h>} might define the interface
4810 typedef union __attribute__ ((__transparent_union__))
4814 @} wait_status_ptr_t;
4816 pid_t wait (wait_status_ptr_t);
4819 This interface allows either @code{int *} or @code{union wait *}
4820 arguments to be passed, using the @code{int *} calling convention.
4821 The program can call @code{wait} with arguments of either type:
4824 int w1 () @{ int w; return wait (&w); @}
4825 int w2 () @{ union wait w; return wait (&w); @}
4828 With this interface, @code{wait}'s implementation might look like this:
4831 pid_t wait (wait_status_ptr_t p)
4833 return waitpid (-1, p.__ip, 0);
4838 When attached to a type (including a @code{union} or a @code{struct}),
4839 this attribute means that variables of that type are meant to appear
4840 possibly unused. GCC will not produce a warning for any variables of
4841 that type, even if the variable appears to do nothing. This is often
4842 the case with lock or thread classes, which are usually defined and then
4843 not referenced, but contain constructors and destructors that have
4844 nontrivial bookkeeping functions.
4847 @itemx deprecated (@var{msg})
4848 The @code{deprecated} attribute results in a warning if the type
4849 is used anywhere in the source file. This is useful when identifying
4850 types that are expected to be removed in a future version of a program.
4851 If possible, the warning also includes the location of the declaration
4852 of the deprecated type, to enable users to easily find further
4853 information about why the type is deprecated, or what they should do
4854 instead. Note that the warnings only occur for uses and then only
4855 if the type is being applied to an identifier that itself is not being
4856 declared as deprecated.
4859 typedef int T1 __attribute__ ((deprecated));
4863 typedef T1 T3 __attribute__ ((deprecated));
4864 T3 z __attribute__ ((deprecated));
4867 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
4868 warning is issued for line 4 because T2 is not explicitly
4869 deprecated. Line 5 has no warning because T3 is explicitly
4870 deprecated. Similarly for line 6. The optional msg
4871 argument, which must be a string, will be printed in the warning if
4874 The @code{deprecated} attribute can also be used for functions and
4875 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
4878 Accesses through pointers to types with this attribute are not subject
4879 to type-based alias analysis, but are instead assumed to be able to alias
4880 any other type of objects. In the context of 6.5/7 an lvalue expression
4881 dereferencing such a pointer is treated like having a character type.
4882 See @option{-fstrict-aliasing} for more information on aliasing issues.
4883 This extension exists to support some vector APIs, in which pointers to
4884 one vector type are permitted to alias pointers to a different vector type.
4886 Note that an object of a type with this attribute does not have any
4892 typedef short __attribute__((__may_alias__)) short_a;
4898 short_a *b = (short_a *) &a;
4902 if (a == 0x12345678)
4909 If you replaced @code{short_a} with @code{short} in the variable
4910 declaration, the above program would abort when compiled with
4911 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
4912 above in recent GCC versions.
4915 In C++, attribute visibility (@pxref{Function Attributes}) can also be
4916 applied to class, struct, union and enum types. Unlike other type
4917 attributes, the attribute must appear between the initial keyword and
4918 the name of the type; it cannot appear after the body of the type.
4920 Note that the type visibility is applied to vague linkage entities
4921 associated with the class (vtable, typeinfo node, etc.). In
4922 particular, if a class is thrown as an exception in one shared object
4923 and caught in another, the class must have default visibility.
4924 Otherwise the two shared objects will be unable to use the same
4925 typeinfo node and exception handling will break.
4929 @subsection ARM Type Attributes
4931 On those ARM targets that support @code{dllimport} (such as Symbian
4932 OS), you can use the @code{notshared} attribute to indicate that the
4933 virtual table and other similar data for a class should not be
4934 exported from a DLL@. For example:
4937 class __declspec(notshared) C @{
4939 __declspec(dllimport) C();
4943 __declspec(dllexport)
4947 In this code, @code{C::C} is exported from the current DLL, but the
4948 virtual table for @code{C} is not exported. (You can use
4949 @code{__attribute__} instead of @code{__declspec} if you prefer, but
4950 most Symbian OS code uses @code{__declspec}.)
4952 @anchor{MeP Type Attributes}
4953 @subsection MeP Type Attributes
4955 Many of the MeP variable attributes may be applied to types as well.
4956 Specifically, the @code{based}, @code{tiny}, @code{near}, and
4957 @code{far} attributes may be applied to either. The @code{io} and
4958 @code{cb} attributes may not be applied to types.
4960 @anchor{i386 Type Attributes}
4961 @subsection i386 Type Attributes
4963 Two attributes are currently defined for i386 configurations:
4964 @code{ms_struct} and @code{gcc_struct}.
4970 @cindex @code{ms_struct}
4971 @cindex @code{gcc_struct}
4973 If @code{packed} is used on a structure, or if bit-fields are used
4974 it may be that the Microsoft ABI packs them differently
4975 than GCC would normally pack them. Particularly when moving packed
4976 data between functions compiled with GCC and the native Microsoft compiler
4977 (either via function call or as data in a file), it may be necessary to access
4980 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4981 compilers to match the native Microsoft compiler.
4984 To specify multiple attributes, separate them by commas within the
4985 double parentheses: for example, @samp{__attribute__ ((aligned (16),
4988 @anchor{PowerPC Type Attributes}
4989 @subsection PowerPC Type Attributes
4991 Three attributes currently are defined for PowerPC configurations:
4992 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4994 For full documentation of the @code{ms_struct} and @code{gcc_struct}
4995 attributes please see the documentation in @ref{i386 Type Attributes}.
4997 The @code{altivec} attribute allows one to declare AltiVec vector data
4998 types supported by the AltiVec Programming Interface Manual. The
4999 attribute requires an argument to specify one of three vector types:
5000 @code{vector__}, @code{pixel__} (always followed by unsigned short),
5001 and @code{bool__} (always followed by unsigned).
5004 __attribute__((altivec(vector__)))
5005 __attribute__((altivec(pixel__))) unsigned short
5006 __attribute__((altivec(bool__))) unsigned
5009 These attributes mainly are intended to support the @code{__vector},
5010 @code{__pixel}, and @code{__bool} AltiVec keywords.
5012 @anchor{SPU Type Attributes}
5013 @subsection SPU Type Attributes
5015 The SPU supports the @code{spu_vector} attribute for types. This attribute
5016 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
5017 Language Extensions Specification. It is intended to support the
5018 @code{__vector} keyword.
5022 @section An Inline Function is As Fast As a Macro
5023 @cindex inline functions
5024 @cindex integrating function code
5026 @cindex macros, inline alternative
5028 By declaring a function inline, you can direct GCC to make
5029 calls to that function faster. One way GCC can achieve this is to
5030 integrate that function's code into the code for its callers. This
5031 makes execution faster by eliminating the function-call overhead; in
5032 addition, if any of the actual argument values are constant, their
5033 known values may permit simplifications at compile time so that not
5034 all of the inline function's code needs to be included. The effect on
5035 code size is less predictable; object code may be larger or smaller
5036 with function inlining, depending on the particular case. You can
5037 also direct GCC to try to integrate all ``simple enough'' functions
5038 into their callers with the option @option{-finline-functions}.
5040 GCC implements three different semantics of declaring a function
5041 inline. One is available with @option{-std=gnu89} or
5042 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
5043 on all inline declarations, another when
5044 @option{-std=c99}, @option{-std=c1x},
5045 @option{-std=gnu99} or @option{-std=gnu1x}
5046 (without @option{-fgnu89-inline}), and the third
5047 is used when compiling C++.
5049 To declare a function inline, use the @code{inline} keyword in its
5050 declaration, like this:
5060 If you are writing a header file to be included in ISO C90 programs, write
5061 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
5063 The three types of inlining behave similarly in two important cases:
5064 when the @code{inline} keyword is used on a @code{static} function,
5065 like the example above, and when a function is first declared without
5066 using the @code{inline} keyword and then is defined with
5067 @code{inline}, like this:
5070 extern int inc (int *a);
5078 In both of these common cases, the program behaves the same as if you
5079 had not used the @code{inline} keyword, except for its speed.
5081 @cindex inline functions, omission of
5082 @opindex fkeep-inline-functions
5083 When a function is both inline and @code{static}, if all calls to the
5084 function are integrated into the caller, and the function's address is
5085 never used, then the function's own assembler code is never referenced.
5086 In this case, GCC does not actually output assembler code for the
5087 function, unless you specify the option @option{-fkeep-inline-functions}.
5088 Some calls cannot be integrated for various reasons (in particular,
5089 calls that precede the function's definition cannot be integrated, and
5090 neither can recursive calls within the definition). If there is a
5091 nonintegrated call, then the function is compiled to assembler code as
5092 usual. The function must also be compiled as usual if the program
5093 refers to its address, because that can't be inlined.
5096 Note that certain usages in a function definition can make it unsuitable
5097 for inline substitution. Among these usages are: use of varargs, use of
5098 alloca, use of variable sized data types (@pxref{Variable Length}),
5099 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
5100 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
5101 will warn when a function marked @code{inline} could not be substituted,
5102 and will give the reason for the failure.
5104 @cindex automatic @code{inline} for C++ member fns
5105 @cindex @code{inline} automatic for C++ member fns
5106 @cindex member fns, automatically @code{inline}
5107 @cindex C++ member fns, automatically @code{inline}
5108 @opindex fno-default-inline
5109 As required by ISO C++, GCC considers member functions defined within
5110 the body of a class to be marked inline even if they are
5111 not explicitly declared with the @code{inline} keyword. You can
5112 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
5113 Options,,Options Controlling C++ Dialect}.
5115 GCC does not inline any functions when not optimizing unless you specify
5116 the @samp{always_inline} attribute for the function, like this:
5119 /* @r{Prototype.} */
5120 inline void foo (const char) __attribute__((always_inline));
5123 The remainder of this section is specific to GNU C90 inlining.
5125 @cindex non-static inline function
5126 When an inline function is not @code{static}, then the compiler must assume
5127 that there may be calls from other source files; since a global symbol can
5128 be defined only once in any program, the function must not be defined in
5129 the other source files, so the calls therein cannot be integrated.
5130 Therefore, a non-@code{static} inline function is always compiled on its
5131 own in the usual fashion.
5133 If you specify both @code{inline} and @code{extern} in the function
5134 definition, then the definition is used only for inlining. In no case
5135 is the function compiled on its own, not even if you refer to its
5136 address explicitly. Such an address becomes an external reference, as
5137 if you had only declared the function, and had not defined it.
5139 This combination of @code{inline} and @code{extern} has almost the
5140 effect of a macro. The way to use it is to put a function definition in
5141 a header file with these keywords, and put another copy of the
5142 definition (lacking @code{inline} and @code{extern}) in a library file.
5143 The definition in the header file will cause most calls to the function
5144 to be inlined. If any uses of the function remain, they will refer to
5145 the single copy in the library.
5148 @section When is a Volatile Object Accessed?
5149 @cindex accessing volatiles
5150 @cindex volatile read
5151 @cindex volatile write
5152 @cindex volatile access
5154 C has the concept of volatile objects. These are normally accessed by
5155 pointers and used for accessing hardware or inter-thread
5156 communication. The standard encourage compilers to refrain from
5157 optimizations concerning accesses to volatile objects, but leaves it
5158 implementation defined as to what constitutes a volatile access. The
5159 minimum requirement is that at a sequence point all previous accesses
5160 to volatile objects have stabilized and no subsequent accesses have
5161 occurred. Thus an implementation is free to reorder and combine
5162 volatile accesses which occur between sequence points, but cannot do
5163 so for accesses across a sequence point. The use of volatiles does
5164 not allow you to violate the restriction on updating objects multiple
5165 times between two sequence points.
5167 Accesses to non-volatile objects are not ordered with respect to
5168 volatile accesses. You cannot use a volatile object as a memory
5169 barrier to order a sequence of writes to non-volatile memory. For
5173 int *ptr = @var{something};
5175 *ptr = @var{something};
5179 Unless @var{*ptr} and @var{vobj} can be aliased, it is not guaranteed
5180 that the write to @var{*ptr} will have occurred by the time the update
5181 of @var{vobj} has happened. If you need this guarantee, you must use
5182 a stronger memory barrier such as:
5185 int *ptr = @var{something};
5187 *ptr = @var{something};
5188 asm volatile ("" : : : "memory");
5192 A scalar volatile object is read, when it is accessed in a void context:
5195 volatile int *src = @var{somevalue};
5199 Such expressions are rvalues, and GCC implements this as a
5200 read of the volatile object being pointed to.
5202 Assignments are also expressions and have an rvalue. However when
5203 assigning to a scalar volatile, the volatile object is not reread,
5204 regardless of whether the assignment expression's rvalue is used or
5205 not. If the assignment's rvalue is used, the value is that assigned
5206 to the volatile object. For instance, there is no read of @var{vobj}
5207 in all the following cases:
5212 vobj = @var{something};
5213 obj = vobj = @var{something};
5214 obj ? vobj = @var{onething} : vobj = @var{anotherthing};
5215 obj = (@var{something}, vobj = @var{anotherthing});
5218 If you need to read the volatile object after an assignment has
5219 occurred, you must use a separate expression with an intervening
5222 As bitfields are not individually addressable, volatile bitfields may
5223 be implicitly read when written to, or when adjacent bitfields are
5224 accessed. Bitfield operations may be optimized such that adjacent
5225 bitfields are only partially accessed, if they straddle a storage unit
5226 boundary. For these reasons it is unwise to use volatile bitfields to
5230 @section Assembler Instructions with C Expression Operands
5231 @cindex extended @code{asm}
5232 @cindex @code{asm} expressions
5233 @cindex assembler instructions
5236 In an assembler instruction using @code{asm}, you can specify the
5237 operands of the instruction using C expressions. This means you need not
5238 guess which registers or memory locations will contain the data you want
5241 You must specify an assembler instruction template much like what
5242 appears in a machine description, plus an operand constraint string for
5245 For example, here is how to use the 68881's @code{fsinx} instruction:
5248 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
5252 Here @code{angle} is the C expression for the input operand while
5253 @code{result} is that of the output operand. Each has @samp{"f"} as its
5254 operand constraint, saying that a floating point register is required.
5255 The @samp{=} in @samp{=f} indicates that the operand is an output; all
5256 output operands' constraints must use @samp{=}. The constraints use the
5257 same language used in the machine description (@pxref{Constraints}).
5259 Each operand is described by an operand-constraint string followed by
5260 the C expression in parentheses. A colon separates the assembler
5261 template from the first output operand and another separates the last
5262 output operand from the first input, if any. Commas separate the
5263 operands within each group. The total number of operands is currently
5264 limited to 30; this limitation may be lifted in some future version of
5267 If there are no output operands but there are input operands, you must
5268 place two consecutive colons surrounding the place where the output
5271 As of GCC version 3.1, it is also possible to specify input and output
5272 operands using symbolic names which can be referenced within the
5273 assembler code. These names are specified inside square brackets
5274 preceding the constraint string, and can be referenced inside the
5275 assembler code using @code{%[@var{name}]} instead of a percentage sign
5276 followed by the operand number. Using named operands the above example
5280 asm ("fsinx %[angle],%[output]"
5281 : [output] "=f" (result)
5282 : [angle] "f" (angle));
5286 Note that the symbolic operand names have no relation whatsoever to
5287 other C identifiers. You may use any name you like, even those of
5288 existing C symbols, but you must ensure that no two operands within the same
5289 assembler construct use the same symbolic name.
5291 Output operand expressions must be lvalues; the compiler can check this.
5292 The input operands need not be lvalues. The compiler cannot check
5293 whether the operands have data types that are reasonable for the
5294 instruction being executed. It does not parse the assembler instruction
5295 template and does not know what it means or even whether it is valid
5296 assembler input. The extended @code{asm} feature is most often used for
5297 machine instructions the compiler itself does not know exist. If
5298 the output expression cannot be directly addressed (for example, it is a
5299 bit-field), your constraint must allow a register. In that case, GCC
5300 will use the register as the output of the @code{asm}, and then store
5301 that register into the output.
5303 The ordinary output operands must be write-only; GCC will assume that
5304 the values in these operands before the instruction are dead and need
5305 not be generated. Extended asm supports input-output or read-write
5306 operands. Use the constraint character @samp{+} to indicate such an
5307 operand and list it with the output operands. You should only use
5308 read-write operands when the constraints for the operand (or the
5309 operand in which only some of the bits are to be changed) allow a
5312 You may, as an alternative, logically split its function into two
5313 separate operands, one input operand and one write-only output
5314 operand. The connection between them is expressed by constraints
5315 which say they need to be in the same location when the instruction
5316 executes. You can use the same C expression for both operands, or
5317 different expressions. For example, here we write the (fictitious)
5318 @samp{combine} instruction with @code{bar} as its read-only source
5319 operand and @code{foo} as its read-write destination:
5322 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
5326 The constraint @samp{"0"} for operand 1 says that it must occupy the
5327 same location as operand 0. A number in constraint is allowed only in
5328 an input operand and it must refer to an output operand.
5330 Only a number in the constraint can guarantee that one operand will be in
5331 the same place as another. The mere fact that @code{foo} is the value
5332 of both operands is not enough to guarantee that they will be in the
5333 same place in the generated assembler code. The following would not
5337 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
5340 Various optimizations or reloading could cause operands 0 and 1 to be in
5341 different registers; GCC knows no reason not to do so. For example, the
5342 compiler might find a copy of the value of @code{foo} in one register and
5343 use it for operand 1, but generate the output operand 0 in a different
5344 register (copying it afterward to @code{foo}'s own address). Of course,
5345 since the register for operand 1 is not even mentioned in the assembler
5346 code, the result will not work, but GCC can't tell that.
5348 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
5349 the operand number for a matching constraint. For example:
5352 asm ("cmoveq %1,%2,%[result]"
5353 : [result] "=r"(result)
5354 : "r" (test), "r"(new), "[result]"(old));
5357 Sometimes you need to make an @code{asm} operand be a specific register,
5358 but there's no matching constraint letter for that register @emph{by
5359 itself}. To force the operand into that register, use a local variable
5360 for the operand and specify the register in the variable declaration.
5361 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
5362 register constraint letter that matches the register:
5365 register int *p1 asm ("r0") = @dots{};
5366 register int *p2 asm ("r1") = @dots{};
5367 register int *result asm ("r0");
5368 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5371 @anchor{Example of asm with clobbered asm reg}
5372 In the above example, beware that a register that is call-clobbered by
5373 the target ABI will be overwritten by any function call in the
5374 assignment, including library calls for arithmetic operators.
5375 Also a register may be clobbered when generating some operations,
5376 like variable shift, memory copy or memory move on x86.
5377 Assuming it is a call-clobbered register, this may happen to @code{r0}
5378 above by the assignment to @code{p2}. If you have to use such a
5379 register, use temporary variables for expressions between the register
5384 register int *p1 asm ("r0") = @dots{};
5385 register int *p2 asm ("r1") = t1;
5386 register int *result asm ("r0");
5387 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5390 Some instructions clobber specific hard registers. To describe this,
5391 write a third colon after the input operands, followed by the names of
5392 the clobbered hard registers (given as strings). Here is a realistic
5393 example for the VAX:
5396 asm volatile ("movc3 %0,%1,%2"
5397 : /* @r{no outputs} */
5398 : "g" (from), "g" (to), "g" (count)
5399 : "r0", "r1", "r2", "r3", "r4", "r5");
5402 You may not write a clobber description in a way that overlaps with an
5403 input or output operand. For example, you may not have an operand
5404 describing a register class with one member if you mention that register
5405 in the clobber list. Variables declared to live in specific registers
5406 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
5407 have no part mentioned in the clobber description.
5408 There is no way for you to specify that an input
5409 operand is modified without also specifying it as an output
5410 operand. Note that if all the output operands you specify are for this
5411 purpose (and hence unused), you will then also need to specify
5412 @code{volatile} for the @code{asm} construct, as described below, to
5413 prevent GCC from deleting the @code{asm} statement as unused.
5415 If you refer to a particular hardware register from the assembler code,
5416 you will probably have to list the register after the third colon to
5417 tell the compiler the register's value is modified. In some assemblers,
5418 the register names begin with @samp{%}; to produce one @samp{%} in the
5419 assembler code, you must write @samp{%%} in the input.
5421 If your assembler instruction can alter the condition code register, add
5422 @samp{cc} to the list of clobbered registers. GCC on some machines
5423 represents the condition codes as a specific hardware register;
5424 @samp{cc} serves to name this register. On other machines, the
5425 condition code is handled differently, and specifying @samp{cc} has no
5426 effect. But it is valid no matter what the machine.
5428 If your assembler instructions access memory in an unpredictable
5429 fashion, add @samp{memory} to the list of clobbered registers. This
5430 will cause GCC to not keep memory values cached in registers across the
5431 assembler instruction and not optimize stores or loads to that memory.
5432 You will also want to add the @code{volatile} keyword if the memory
5433 affected is not listed in the inputs or outputs of the @code{asm}, as
5434 the @samp{memory} clobber does not count as a side-effect of the
5435 @code{asm}. If you know how large the accessed memory is, you can add
5436 it as input or output but if this is not known, you should add
5437 @samp{memory}. As an example, if you access ten bytes of a string, you
5438 can use a memory input like:
5441 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
5444 Note that in the following example the memory input is necessary,
5445 otherwise GCC might optimize the store to @code{x} away:
5452 asm ("magic stuff accessing an 'int' pointed to by '%1'"
5453 "=&d" (r) : "a" (y), "m" (*y));
5458 You can put multiple assembler instructions together in a single
5459 @code{asm} template, separated by the characters normally used in assembly
5460 code for the system. A combination that works in most places is a newline
5461 to break the line, plus a tab character to move to the instruction field
5462 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
5463 assembler allows semicolons as a line-breaking character. Note that some
5464 assembler dialects use semicolons to start a comment.
5465 The input operands are guaranteed not to use any of the clobbered
5466 registers, and neither will the output operands' addresses, so you can
5467 read and write the clobbered registers as many times as you like. Here
5468 is an example of multiple instructions in a template; it assumes the
5469 subroutine @code{_foo} accepts arguments in registers 9 and 10:
5472 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
5474 : "g" (from), "g" (to)
5478 Unless an output operand has the @samp{&} constraint modifier, GCC
5479 may allocate it in the same register as an unrelated input operand, on
5480 the assumption the inputs are consumed before the outputs are produced.
5481 This assumption may be false if the assembler code actually consists of
5482 more than one instruction. In such a case, use @samp{&} for each output
5483 operand that may not overlap an input. @xref{Modifiers}.
5485 If you want to test the condition code produced by an assembler
5486 instruction, you must include a branch and a label in the @code{asm}
5487 construct, as follows:
5490 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
5496 This assumes your assembler supports local labels, as the GNU assembler
5497 and most Unix assemblers do.
5499 Speaking of labels, jumps from one @code{asm} to another are not
5500 supported. The compiler's optimizers do not know about these jumps, and
5501 therefore they cannot take account of them when deciding how to
5502 optimize. @xref{Extended asm with goto}.
5504 @cindex macros containing @code{asm}
5505 Usually the most convenient way to use these @code{asm} instructions is to
5506 encapsulate them in macros that look like functions. For example,
5510 (@{ double __value, __arg = (x); \
5511 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
5516 Here the variable @code{__arg} is used to make sure that the instruction
5517 operates on a proper @code{double} value, and to accept only those
5518 arguments @code{x} which can convert automatically to a @code{double}.
5520 Another way to make sure the instruction operates on the correct data
5521 type is to use a cast in the @code{asm}. This is different from using a
5522 variable @code{__arg} in that it converts more different types. For
5523 example, if the desired type were @code{int}, casting the argument to
5524 @code{int} would accept a pointer with no complaint, while assigning the
5525 argument to an @code{int} variable named @code{__arg} would warn about
5526 using a pointer unless the caller explicitly casts it.
5528 If an @code{asm} has output operands, GCC assumes for optimization
5529 purposes the instruction has no side effects except to change the output
5530 operands. This does not mean instructions with a side effect cannot be
5531 used, but you must be careful, because the compiler may eliminate them
5532 if the output operands aren't used, or move them out of loops, or
5533 replace two with one if they constitute a common subexpression. Also,
5534 if your instruction does have a side effect on a variable that otherwise
5535 appears not to change, the old value of the variable may be reused later
5536 if it happens to be found in a register.
5538 You can prevent an @code{asm} instruction from being deleted
5539 by writing the keyword @code{volatile} after
5540 the @code{asm}. For example:
5543 #define get_and_set_priority(new) \
5545 asm volatile ("get_and_set_priority %0, %1" \
5546 : "=g" (__old) : "g" (new)); \
5551 The @code{volatile} keyword indicates that the instruction has
5552 important side-effects. GCC will not delete a volatile @code{asm} if
5553 it is reachable. (The instruction can still be deleted if GCC can
5554 prove that control-flow will never reach the location of the
5555 instruction.) Note that even a volatile @code{asm} instruction
5556 can be moved relative to other code, including across jump
5557 instructions. For example, on many targets there is a system
5558 register which can be set to control the rounding mode of
5559 floating point operations. You might try
5560 setting it with a volatile @code{asm}, like this PowerPC example:
5563 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
5568 This will not work reliably, as the compiler may move the addition back
5569 before the volatile @code{asm}. To make it work you need to add an
5570 artificial dependency to the @code{asm} referencing a variable in the code
5571 you don't want moved, for example:
5574 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
5578 Similarly, you can't expect a
5579 sequence of volatile @code{asm} instructions to remain perfectly
5580 consecutive. If you want consecutive output, use a single @code{asm}.
5581 Also, GCC will perform some optimizations across a volatile @code{asm}
5582 instruction; GCC does not ``forget everything'' when it encounters
5583 a volatile @code{asm} instruction the way some other compilers do.
5585 An @code{asm} instruction without any output operands will be treated
5586 identically to a volatile @code{asm} instruction.
5588 It is a natural idea to look for a way to give access to the condition
5589 code left by the assembler instruction. However, when we attempted to
5590 implement this, we found no way to make it work reliably. The problem
5591 is that output operands might need reloading, which would result in
5592 additional following ``store'' instructions. On most machines, these
5593 instructions would alter the condition code before there was time to
5594 test it. This problem doesn't arise for ordinary ``test'' and
5595 ``compare'' instructions because they don't have any output operands.
5597 For reasons similar to those described above, it is not possible to give
5598 an assembler instruction access to the condition code left by previous
5601 @anchor{Extended asm with goto}
5602 As of GCC version 4.5, @code{asm goto} may be used to have the assembly
5603 jump to one or more C labels. In this form, a fifth section after the
5604 clobber list contains a list of all C labels to which the assembly may jump.
5605 Each label operand is implicitly self-named. The @code{asm} is also assumed
5606 to fall through to the next statement.
5608 This form of @code{asm} is restricted to not have outputs. This is due
5609 to a internal restriction in the compiler that control transfer instructions
5610 cannot have outputs. This restriction on @code{asm goto} may be lifted
5611 in some future version of the compiler. In the mean time, @code{asm goto}
5612 may include a memory clobber, and so leave outputs in memory.
5618 asm goto ("frob %%r5, %1; jc %l[error]; mov (%2), %%r5"
5619 : : "r"(x), "r"(&y) : "r5", "memory" : error);
5626 In this (inefficient) example, the @code{frob} instruction sets the
5627 carry bit to indicate an error. The @code{jc} instruction detects
5628 this and branches to the @code{error} label. Finally, the output
5629 of the @code{frob} instruction (@code{%r5}) is stored into the memory
5630 for variable @code{y}, which is later read by the @code{return} statement.
5636 asm goto ("mfsr %%r1, 123; jmp %%r1;"
5637 ".pushsection doit_table;"
5638 ".long %l0, %l1, %l2, %l3;"
5640 : : : "r1" : label1, label2, label3, label4);
5641 __builtin_unreachable ();
5656 In this (also inefficient) example, the @code{mfsr} instruction reads
5657 an address from some out-of-band machine register, and the following
5658 @code{jmp} instruction branches to that address. The address read by
5659 the @code{mfsr} instruction is assumed to have been previously set via
5660 some application-specific mechanism to be one of the four values stored
5661 in the @code{doit_table} section. Finally, the @code{asm} is followed
5662 by a call to @code{__builtin_unreachable} to indicate that the @code{asm}
5663 does not in fact fall through.
5666 #define TRACE1(NUM) \
5668 asm goto ("0: nop;" \
5669 ".pushsection trace_table;" \
5672 : : : : trace#NUM); \
5673 if (0) @{ trace#NUM: trace(); @} \
5675 #define TRACE TRACE1(__COUNTER__)
5678 In this example (which in fact inspired the @code{asm goto} feature)
5679 we want on rare occasions to call the @code{trace} function; on other
5680 occasions we'd like to keep the overhead to the absolute minimum.
5681 The normal code path consists of a single @code{nop} instruction.
5682 However, we record the address of this @code{nop} together with the
5683 address of a label that calls the @code{trace} function. This allows
5684 the @code{nop} instruction to be patched at runtime to be an
5685 unconditional branch to the stored label. It is assumed that an
5686 optimizing compiler will move the labeled block out of line, to
5687 optimize the fall through path from the @code{asm}.
5689 If you are writing a header file that should be includable in ISO C
5690 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
5693 @subsection Size of an @code{asm}
5695 Some targets require that GCC track the size of each instruction used in
5696 order to generate correct code. Because the final length of an
5697 @code{asm} is only known by the assembler, GCC must make an estimate as
5698 to how big it will be. The estimate is formed by counting the number of
5699 statements in the pattern of the @code{asm} and multiplying that by the
5700 length of the longest instruction on that processor. Statements in the
5701 @code{asm} are identified by newline characters and whatever statement
5702 separator characters are supported by the assembler; on most processors
5703 this is the `@code{;}' character.
5705 Normally, GCC's estimate is perfectly adequate to ensure that correct
5706 code is generated, but it is possible to confuse the compiler if you use
5707 pseudo instructions or assembler macros that expand into multiple real
5708 instructions or if you use assembler directives that expand to more
5709 space in the object file than would be needed for a single instruction.
5710 If this happens then the assembler will produce a diagnostic saying that
5711 a label is unreachable.
5713 @subsection i386 floating point asm operands
5715 There are several rules on the usage of stack-like regs in
5716 asm_operands insns. These rules apply only to the operands that are
5721 Given a set of input regs that die in an asm_operands, it is
5722 necessary to know which are implicitly popped by the asm, and
5723 which must be explicitly popped by gcc.
5725 An input reg that is implicitly popped by the asm must be
5726 explicitly clobbered, unless it is constrained to match an
5730 For any input reg that is implicitly popped by an asm, it is
5731 necessary to know how to adjust the stack to compensate for the pop.
5732 If any non-popped input is closer to the top of the reg-stack than
5733 the implicitly popped reg, it would not be possible to know what the
5734 stack looked like---it's not clear how the rest of the stack ``slides
5737 All implicitly popped input regs must be closer to the top of
5738 the reg-stack than any input that is not implicitly popped.
5740 It is possible that if an input dies in an insn, reload might
5741 use the input reg for an output reload. Consider this example:
5744 asm ("foo" : "=t" (a) : "f" (b));
5747 This asm says that input B is not popped by the asm, and that
5748 the asm pushes a result onto the reg-stack, i.e., the stack is one
5749 deeper after the asm than it was before. But, it is possible that
5750 reload will think that it can use the same reg for both the input and
5751 the output, if input B dies in this insn.
5753 If any input operand uses the @code{f} constraint, all output reg
5754 constraints must use the @code{&} earlyclobber.
5756 The asm above would be written as
5759 asm ("foo" : "=&t" (a) : "f" (b));
5763 Some operands need to be in particular places on the stack. All
5764 output operands fall in this category---there is no other way to
5765 know which regs the outputs appear in unless the user indicates
5766 this in the constraints.
5768 Output operands must specifically indicate which reg an output
5769 appears in after an asm. @code{=f} is not allowed: the operand
5770 constraints must select a class with a single reg.
5773 Output operands may not be ``inserted'' between existing stack regs.
5774 Since no 387 opcode uses a read/write operand, all output operands
5775 are dead before the asm_operands, and are pushed by the asm_operands.
5776 It makes no sense to push anywhere but the top of the reg-stack.
5778 Output operands must start at the top of the reg-stack: output
5779 operands may not ``skip'' a reg.
5782 Some asm statements may need extra stack space for internal
5783 calculations. This can be guaranteed by clobbering stack registers
5784 unrelated to the inputs and outputs.
5788 Here are a couple of reasonable asms to want to write. This asm
5789 takes one input, which is internally popped, and produces two outputs.
5792 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
5795 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
5796 and replaces them with one output. The user must code the @code{st(1)}
5797 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
5800 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
5806 @section Controlling Names Used in Assembler Code
5807 @cindex assembler names for identifiers
5808 @cindex names used in assembler code
5809 @cindex identifiers, names in assembler code
5811 You can specify the name to be used in the assembler code for a C
5812 function or variable by writing the @code{asm} (or @code{__asm__})
5813 keyword after the declarator as follows:
5816 int foo asm ("myfoo") = 2;
5820 This specifies that the name to be used for the variable @code{foo} in
5821 the assembler code should be @samp{myfoo} rather than the usual
5824 On systems where an underscore is normally prepended to the name of a C
5825 function or variable, this feature allows you to define names for the
5826 linker that do not start with an underscore.
5828 It does not make sense to use this feature with a non-static local
5829 variable since such variables do not have assembler names. If you are
5830 trying to put the variable in a particular register, see @ref{Explicit
5831 Reg Vars}. GCC presently accepts such code with a warning, but will
5832 probably be changed to issue an error, rather than a warning, in the
5835 You cannot use @code{asm} in this way in a function @emph{definition}; but
5836 you can get the same effect by writing a declaration for the function
5837 before its definition and putting @code{asm} there, like this:
5840 extern func () asm ("FUNC");
5847 It is up to you to make sure that the assembler names you choose do not
5848 conflict with any other assembler symbols. Also, you must not use a
5849 register name; that would produce completely invalid assembler code. GCC
5850 does not as yet have the ability to store static variables in registers.
5851 Perhaps that will be added.
5853 @node Explicit Reg Vars
5854 @section Variables in Specified Registers
5855 @cindex explicit register variables
5856 @cindex variables in specified registers
5857 @cindex specified registers
5858 @cindex registers, global allocation
5860 GNU C allows you to put a few global variables into specified hardware
5861 registers. You can also specify the register in which an ordinary
5862 register variable should be allocated.
5866 Global register variables reserve registers throughout the program.
5867 This may be useful in programs such as programming language
5868 interpreters which have a couple of global variables that are accessed
5872 Local register variables in specific registers do not reserve the
5873 registers, except at the point where they are used as input or output
5874 operands in an @code{asm} statement and the @code{asm} statement itself is
5875 not deleted. The compiler's data flow analysis is capable of determining
5876 where the specified registers contain live values, and where they are
5877 available for other uses. Stores into local register variables may be deleted
5878 when they appear to be dead according to dataflow analysis. References
5879 to local register variables may be deleted or moved or simplified.
5881 These local variables are sometimes convenient for use with the extended
5882 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
5883 output of the assembler instruction directly into a particular register.
5884 (This will work provided the register you specify fits the constraints
5885 specified for that operand in the @code{asm}.)
5893 @node Global Reg Vars
5894 @subsection Defining Global Register Variables
5895 @cindex global register variables
5896 @cindex registers, global variables in
5898 You can define a global register variable in GNU C like this:
5901 register int *foo asm ("a5");
5905 Here @code{a5} is the name of the register which should be used. Choose a
5906 register which is normally saved and restored by function calls on your
5907 machine, so that library routines will not clobber it.
5909 Naturally the register name is cpu-dependent, so you would need to
5910 conditionalize your program according to cpu type. The register
5911 @code{a5} would be a good choice on a 68000 for a variable of pointer
5912 type. On machines with register windows, be sure to choose a ``global''
5913 register that is not affected magically by the function call mechanism.
5915 In addition, operating systems on one type of cpu may differ in how they
5916 name the registers; then you would need additional conditionals. For
5917 example, some 68000 operating systems call this register @code{%a5}.
5919 Eventually there may be a way of asking the compiler to choose a register
5920 automatically, but first we need to figure out how it should choose and
5921 how to enable you to guide the choice. No solution is evident.
5923 Defining a global register variable in a certain register reserves that
5924 register entirely for this use, at least within the current compilation.
5925 The register will not be allocated for any other purpose in the functions
5926 in the current compilation. The register will not be saved and restored by
5927 these functions. Stores into this register are never deleted even if they
5928 would appear to be dead, but references may be deleted or moved or
5931 It is not safe to access the global register variables from signal
5932 handlers, or from more than one thread of control, because the system
5933 library routines may temporarily use the register for other things (unless
5934 you recompile them specially for the task at hand).
5936 @cindex @code{qsort}, and global register variables
5937 It is not safe for one function that uses a global register variable to
5938 call another such function @code{foo} by way of a third function
5939 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
5940 different source file in which the variable wasn't declared). This is
5941 because @code{lose} might save the register and put some other value there.
5942 For example, you can't expect a global register variable to be available in
5943 the comparison-function that you pass to @code{qsort}, since @code{qsort}
5944 might have put something else in that register. (If you are prepared to
5945 recompile @code{qsort} with the same global register variable, you can
5946 solve this problem.)
5948 If you want to recompile @code{qsort} or other source files which do not
5949 actually use your global register variable, so that they will not use that
5950 register for any other purpose, then it suffices to specify the compiler
5951 option @option{-ffixed-@var{reg}}. You need not actually add a global
5952 register declaration to their source code.
5954 A function which can alter the value of a global register variable cannot
5955 safely be called from a function compiled without this variable, because it
5956 could clobber the value the caller expects to find there on return.
5957 Therefore, the function which is the entry point into the part of the
5958 program that uses the global register variable must explicitly save and
5959 restore the value which belongs to its caller.
5961 @cindex register variable after @code{longjmp}
5962 @cindex global register after @code{longjmp}
5963 @cindex value after @code{longjmp}
5966 On most machines, @code{longjmp} will restore to each global register
5967 variable the value it had at the time of the @code{setjmp}. On some
5968 machines, however, @code{longjmp} will not change the value of global
5969 register variables. To be portable, the function that called @code{setjmp}
5970 should make other arrangements to save the values of the global register
5971 variables, and to restore them in a @code{longjmp}. This way, the same
5972 thing will happen regardless of what @code{longjmp} does.
5974 All global register variable declarations must precede all function
5975 definitions. If such a declaration could appear after function
5976 definitions, the declaration would be too late to prevent the register from
5977 being used for other purposes in the preceding functions.
5979 Global register variables may not have initial values, because an
5980 executable file has no means to supply initial contents for a register.
5982 On the SPARC, there are reports that g3 @dots{} g7 are suitable
5983 registers, but certain library functions, such as @code{getwd}, as well
5984 as the subroutines for division and remainder, modify g3 and g4. g1 and
5985 g2 are local temporaries.
5987 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
5988 Of course, it will not do to use more than a few of those.
5990 @node Local Reg Vars
5991 @subsection Specifying Registers for Local Variables
5992 @cindex local variables, specifying registers
5993 @cindex specifying registers for local variables
5994 @cindex registers for local variables
5996 You can define a local register variable with a specified register
6000 register int *foo asm ("a5");
6004 Here @code{a5} is the name of the register which should be used. Note
6005 that this is the same syntax used for defining global register
6006 variables, but for a local variable it would appear within a function.
6008 Naturally the register name is cpu-dependent, but this is not a
6009 problem, since specific registers are most often useful with explicit
6010 assembler instructions (@pxref{Extended Asm}). Both of these things
6011 generally require that you conditionalize your program according to
6014 In addition, operating systems on one type of cpu may differ in how they
6015 name the registers; then you would need additional conditionals. For
6016 example, some 68000 operating systems call this register @code{%a5}.
6018 Defining such a register variable does not reserve the register; it
6019 remains available for other uses in places where flow control determines
6020 the variable's value is not live.
6022 This option does not guarantee that GCC will generate code that has
6023 this variable in the register you specify at all times. You may not
6024 code an explicit reference to this register in the @emph{assembler
6025 instruction template} part of an @code{asm} statement and assume it will
6026 always refer to this variable. However, using the variable as an
6027 @code{asm} @emph{operand} guarantees that the specified register is used
6030 Stores into local register variables may be deleted when they appear to be dead
6031 according to dataflow analysis. References to local register variables may
6032 be deleted or moved or simplified.
6034 As for global register variables, it's recommended that you choose a
6035 register which is normally saved and restored by function calls on
6036 your machine, so that library routines will not clobber it. A common
6037 pitfall is to initialize multiple call-clobbered registers with
6038 arbitrary expressions, where a function call or library call for an
6039 arithmetic operator will overwrite a register value from a previous
6040 assignment, for example @code{r0} below:
6042 register int *p1 asm ("r0") = @dots{};
6043 register int *p2 asm ("r1") = @dots{};
6045 In those cases, a solution is to use a temporary variable for
6046 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
6048 @node Alternate Keywords
6049 @section Alternate Keywords
6050 @cindex alternate keywords
6051 @cindex keywords, alternate
6053 @option{-ansi} and the various @option{-std} options disable certain
6054 keywords. This causes trouble when you want to use GNU C extensions, or
6055 a general-purpose header file that should be usable by all programs,
6056 including ISO C programs. The keywords @code{asm}, @code{typeof} and
6057 @code{inline} are not available in programs compiled with
6058 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
6059 program compiled with @option{-std=c99} or @option{-std=c1x}). The
6061 @code{restrict} is only available when @option{-std=gnu99} (which will
6062 eventually be the default) or @option{-std=c99} (or the equivalent
6063 @option{-std=iso9899:1999}), or an option for a later standard
6066 The way to solve these problems is to put @samp{__} at the beginning and
6067 end of each problematical keyword. For example, use @code{__asm__}
6068 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
6070 Other C compilers won't accept these alternative keywords; if you want to
6071 compile with another compiler, you can define the alternate keywords as
6072 macros to replace them with the customary keywords. It looks like this:
6080 @findex __extension__
6082 @option{-pedantic} and other options cause warnings for many GNU C extensions.
6084 prevent such warnings within one expression by writing
6085 @code{__extension__} before the expression. @code{__extension__} has no
6086 effect aside from this.
6088 @node Incomplete Enums
6089 @section Incomplete @code{enum} Types
6091 You can define an @code{enum} tag without specifying its possible values.
6092 This results in an incomplete type, much like what you get if you write
6093 @code{struct foo} without describing the elements. A later declaration
6094 which does specify the possible values completes the type.
6096 You can't allocate variables or storage using the type while it is
6097 incomplete. However, you can work with pointers to that type.
6099 This extension may not be very useful, but it makes the handling of
6100 @code{enum} more consistent with the way @code{struct} and @code{union}
6103 This extension is not supported by GNU C++.
6105 @node Function Names
6106 @section Function Names as Strings
6107 @cindex @code{__func__} identifier
6108 @cindex @code{__FUNCTION__} identifier
6109 @cindex @code{__PRETTY_FUNCTION__} identifier
6111 GCC provides three magic variables which hold the name of the current
6112 function, as a string. The first of these is @code{__func__}, which
6113 is part of the C99 standard:
6115 The identifier @code{__func__} is implicitly declared by the translator
6116 as if, immediately following the opening brace of each function
6117 definition, the declaration
6120 static const char __func__[] = "function-name";
6124 appeared, where function-name is the name of the lexically-enclosing
6125 function. This name is the unadorned name of the function.
6127 @code{__FUNCTION__} is another name for @code{__func__}. Older
6128 versions of GCC recognize only this name. However, it is not
6129 standardized. For maximum portability, we recommend you use
6130 @code{__func__}, but provide a fallback definition with the
6134 #if __STDC_VERSION__ < 199901L
6136 # define __func__ __FUNCTION__
6138 # define __func__ "<unknown>"
6143 In C, @code{__PRETTY_FUNCTION__} is yet another name for
6144 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
6145 the type signature of the function as well as its bare name. For
6146 example, this program:
6150 extern int printf (char *, ...);
6157 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
6158 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
6176 __PRETTY_FUNCTION__ = void a::sub(int)
6179 These identifiers are not preprocessor macros. In GCC 3.3 and
6180 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
6181 were treated as string literals; they could be used to initialize
6182 @code{char} arrays, and they could be concatenated with other string
6183 literals. GCC 3.4 and later treat them as variables, like
6184 @code{__func__}. In C++, @code{__FUNCTION__} and
6185 @code{__PRETTY_FUNCTION__} have always been variables.
6187 @node Return Address
6188 @section Getting the Return or Frame Address of a Function
6190 These functions may be used to get information about the callers of a
6193 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
6194 This function returns the return address of the current function, or of
6195 one of its callers. The @var{level} argument is number of frames to
6196 scan up the call stack. A value of @code{0} yields the return address
6197 of the current function, a value of @code{1} yields the return address
6198 of the caller of the current function, and so forth. When inlining
6199 the expected behavior is that the function will return the address of
6200 the function that will be returned to. To work around this behavior use
6201 the @code{noinline} function attribute.
6203 The @var{level} argument must be a constant integer.
6205 On some machines it may be impossible to determine the return address of
6206 any function other than the current one; in such cases, or when the top
6207 of the stack has been reached, this function will return @code{0} or a
6208 random value. In addition, @code{__builtin_frame_address} may be used
6209 to determine if the top of the stack has been reached.
6211 Additional post-processing of the returned value may be needed, see
6212 @code{__builtin_extract_return_address}.
6214 This function should only be used with a nonzero argument for debugging
6218 @deftypefn {Built-in Function} {void *} __builtin_extract_return_address (void *@var{addr})
6219 The address as returned by @code{__builtin_return_address} may have to be fed
6220 through this function to get the actual encoded address. For example, on the
6221 31-bit S/390 platform the highest bit has to be masked out, or on SPARC
6222 platforms an offset has to be added for the true next instruction to be
6225 If no fixup is needed, this function simply passes through @var{addr}.
6228 @deftypefn {Built-in Function} {void *} __builtin_frob_return_address (void *@var{addr})
6229 This function does the reverse of @code{__builtin_extract_return_address}.
6232 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
6233 This function is similar to @code{__builtin_return_address}, but it
6234 returns the address of the function frame rather than the return address
6235 of the function. Calling @code{__builtin_frame_address} with a value of
6236 @code{0} yields the frame address of the current function, a value of
6237 @code{1} yields the frame address of the caller of the current function,
6240 The frame is the area on the stack which holds local variables and saved
6241 registers. The frame address is normally the address of the first word
6242 pushed on to the stack by the function. However, the exact definition
6243 depends upon the processor and the calling convention. If the processor
6244 has a dedicated frame pointer register, and the function has a frame,
6245 then @code{__builtin_frame_address} will return the value of the frame
6248 On some machines it may be impossible to determine the frame address of
6249 any function other than the current one; in such cases, or when the top
6250 of the stack has been reached, this function will return @code{0} if
6251 the first frame pointer is properly initialized by the startup code.
6253 This function should only be used with a nonzero argument for debugging
6257 @node Vector Extensions
6258 @section Using vector instructions through built-in functions
6260 On some targets, the instruction set contains SIMD vector instructions that
6261 operate on multiple values contained in one large register at the same time.
6262 For example, on the i386 the MMX, 3DNow!@: and SSE extensions can be used
6265 The first step in using these extensions is to provide the necessary data
6266 types. This should be done using an appropriate @code{typedef}:
6269 typedef int v4si __attribute__ ((vector_size (16)));
6272 The @code{int} type specifies the base type, while the attribute specifies
6273 the vector size for the variable, measured in bytes. For example, the
6274 declaration above causes the compiler to set the mode for the @code{v4si}
6275 type to be 16 bytes wide and divided into @code{int} sized units. For
6276 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
6277 corresponding mode of @code{foo} will be @acronym{V4SI}.
6279 The @code{vector_size} attribute is only applicable to integral and
6280 float scalars, although arrays, pointers, and function return values
6281 are allowed in conjunction with this construct.
6283 All the basic integer types can be used as base types, both as signed
6284 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
6285 @code{long long}. In addition, @code{float} and @code{double} can be
6286 used to build floating-point vector types.
6288 Specifying a combination that is not valid for the current architecture
6289 will cause GCC to synthesize the instructions using a narrower mode.
6290 For example, if you specify a variable of type @code{V4SI} and your
6291 architecture does not allow for this specific SIMD type, GCC will
6292 produce code that uses 4 @code{SIs}.
6294 The types defined in this manner can be used with a subset of normal C
6295 operations. Currently, GCC will allow using the following operators
6296 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~, %}@.
6298 The operations behave like C++ @code{valarrays}. Addition is defined as
6299 the addition of the corresponding elements of the operands. For
6300 example, in the code below, each of the 4 elements in @var{a} will be
6301 added to the corresponding 4 elements in @var{b} and the resulting
6302 vector will be stored in @var{c}.
6305 typedef int v4si __attribute__ ((vector_size (16)));
6312 Subtraction, multiplication, division, and the logical operations
6313 operate in a similar manner. Likewise, the result of using the unary
6314 minus or complement operators on a vector type is a vector whose
6315 elements are the negative or complemented values of the corresponding
6316 elements in the operand.
6318 In C it is possible to use shifting operators @code{<<}, @code{>>} on
6319 integer-type vectors. The operation is defined as following: @code{@{a0,
6320 a1, @dots{}, an@} >> @{b0, b1, @dots{}, bn@} == @{a0 >> b0, a1 >> b1,
6321 @dots{}, an >> bn@}}@. Vector operands must have the same number of
6322 elements. Additionally second operands can be a scalar integer in which
6323 case the scalar is converted to the type used by the vector operand (with
6324 possible truncation) and each element of this new vector is the scalar's
6326 Consider the following code.
6329 typedef int v4si __attribute__ ((vector_size (16)));
6333 b = a >> 1; /* b = a >> @{1,1,1,1@}; */
6336 In C vectors can be subscripted as if the vector were an array with
6337 the same number of elements and base type. Out of bound accesses
6338 invoke undefined behavior at runtime. Warnings for out of bound
6339 accesses for vector subscription can be enabled with
6340 @option{-Warray-bounds}.
6342 You can declare variables and use them in function calls and returns, as
6343 well as in assignments and some casts. You can specify a vector type as
6344 a return type for a function. Vector types can also be used as function
6345 arguments. It is possible to cast from one vector type to another,
6346 provided they are of the same size (in fact, you can also cast vectors
6347 to and from other datatypes of the same size).
6349 You cannot operate between vectors of different lengths or different
6350 signedness without a cast.
6352 A port that supports hardware vector operations, usually provides a set
6353 of built-in functions that can be used to operate on vectors. For
6354 example, a function to add two vectors and multiply the result by a
6355 third could look like this:
6358 v4si f (v4si a, v4si b, v4si c)
6360 v4si tmp = __builtin_addv4si (a, b);
6361 return __builtin_mulv4si (tmp, c);
6368 @findex __builtin_offsetof
6370 GCC implements for both C and C++ a syntactic extension to implement
6371 the @code{offsetof} macro.
6375 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
6377 offsetof_member_designator:
6379 | offsetof_member_designator "." @code{identifier}
6380 | offsetof_member_designator "[" @code{expr} "]"
6383 This extension is sufficient such that
6386 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
6389 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
6390 may be dependent. In either case, @var{member} may consist of a single
6391 identifier, or a sequence of member accesses and array references.
6393 @node Atomic Builtins
6394 @section Built-in functions for atomic memory access
6396 The following builtins are intended to be compatible with those described
6397 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
6398 section 7.4. As such, they depart from the normal GCC practice of using
6399 the ``__builtin_'' prefix, and further that they are overloaded such that
6400 they work on multiple types.
6402 The definition given in the Intel documentation allows only for the use of
6403 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
6404 counterparts. GCC will allow any integral scalar or pointer type that is
6405 1, 2, 4 or 8 bytes in length.
6407 Not all operations are supported by all target processors. If a particular
6408 operation cannot be implemented on the target processor, a warning will be
6409 generated and a call an external function will be generated. The external
6410 function will carry the same name as the builtin, with an additional suffix
6411 @samp{_@var{n}} where @var{n} is the size of the data type.
6413 @c ??? Should we have a mechanism to suppress this warning? This is almost
6414 @c useful for implementing the operation under the control of an external
6417 In most cases, these builtins are considered a @dfn{full barrier}. That is,
6418 no memory operand will be moved across the operation, either forward or
6419 backward. Further, instructions will be issued as necessary to prevent the
6420 processor from speculating loads across the operation and from queuing stores
6421 after the operation.
6423 All of the routines are described in the Intel documentation to take
6424 ``an optional list of variables protected by the memory barrier''. It's
6425 not clear what is meant by that; it could mean that @emph{only} the
6426 following variables are protected, or it could mean that these variables
6427 should in addition be protected. At present GCC ignores this list and
6428 protects all variables which are globally accessible. If in the future
6429 we make some use of this list, an empty list will continue to mean all
6430 globally accessible variables.
6433 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
6434 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
6435 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
6436 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
6437 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
6438 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
6439 @findex __sync_fetch_and_add
6440 @findex __sync_fetch_and_sub
6441 @findex __sync_fetch_and_or
6442 @findex __sync_fetch_and_and
6443 @findex __sync_fetch_and_xor
6444 @findex __sync_fetch_and_nand
6445 These builtins perform the operation suggested by the name, and
6446 returns the value that had previously been in memory. That is,
6449 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
6450 @{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; @} // nand
6453 @emph{Note:} GCC 4.4 and later implement @code{__sync_fetch_and_nand}
6454 builtin as @code{*ptr = ~(tmp & value)} instead of @code{*ptr = ~tmp & value}.
6456 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
6457 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
6458 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
6459 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
6460 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
6461 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
6462 @findex __sync_add_and_fetch
6463 @findex __sync_sub_and_fetch
6464 @findex __sync_or_and_fetch
6465 @findex __sync_and_and_fetch
6466 @findex __sync_xor_and_fetch
6467 @findex __sync_nand_and_fetch
6468 These builtins perform the operation suggested by the name, and
6469 return the new value. That is,
6472 @{ *ptr @var{op}= value; return *ptr; @}
6473 @{ *ptr = ~(*ptr & value); return *ptr; @} // nand
6476 @emph{Note:} GCC 4.4 and later implement @code{__sync_nand_and_fetch}
6477 builtin as @code{*ptr = ~(*ptr & value)} instead of
6478 @code{*ptr = ~*ptr & value}.
6480 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6481 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6482 @findex __sync_bool_compare_and_swap
6483 @findex __sync_val_compare_and_swap
6484 These builtins perform an atomic compare and swap. That is, if the current
6485 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
6488 The ``bool'' version returns true if the comparison is successful and
6489 @var{newval} was written. The ``val'' version returns the contents
6490 of @code{*@var{ptr}} before the operation.
6492 @item __sync_synchronize (...)
6493 @findex __sync_synchronize
6494 This builtin issues a full memory barrier.
6496 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
6497 @findex __sync_lock_test_and_set
6498 This builtin, as described by Intel, is not a traditional test-and-set
6499 operation, but rather an atomic exchange operation. It writes @var{value}
6500 into @code{*@var{ptr}}, and returns the previous contents of
6503 Many targets have only minimal support for such locks, and do not support
6504 a full exchange operation. In this case, a target may support reduced
6505 functionality here by which the @emph{only} valid value to store is the
6506 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
6507 is implementation defined.
6509 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
6510 This means that references after the builtin cannot move to (or be
6511 speculated to) before the builtin, but previous memory stores may not
6512 be globally visible yet, and previous memory loads may not yet be
6515 @item void __sync_lock_release (@var{type} *ptr, ...)
6516 @findex __sync_lock_release
6517 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
6518 Normally this means writing the constant 0 to @code{*@var{ptr}}.
6520 This builtin is not a full barrier, but rather a @dfn{release barrier}.
6521 This means that all previous memory stores are globally visible, and all
6522 previous memory loads have been satisfied, but following memory reads
6523 are not prevented from being speculated to before the barrier.
6526 @node Object Size Checking
6527 @section Object Size Checking Builtins
6528 @findex __builtin_object_size
6529 @findex __builtin___memcpy_chk
6530 @findex __builtin___mempcpy_chk
6531 @findex __builtin___memmove_chk
6532 @findex __builtin___memset_chk
6533 @findex __builtin___strcpy_chk
6534 @findex __builtin___stpcpy_chk
6535 @findex __builtin___strncpy_chk
6536 @findex __builtin___strcat_chk
6537 @findex __builtin___strncat_chk
6538 @findex __builtin___sprintf_chk
6539 @findex __builtin___snprintf_chk
6540 @findex __builtin___vsprintf_chk
6541 @findex __builtin___vsnprintf_chk
6542 @findex __builtin___printf_chk
6543 @findex __builtin___vprintf_chk
6544 @findex __builtin___fprintf_chk
6545 @findex __builtin___vfprintf_chk
6547 GCC implements a limited buffer overflow protection mechanism
6548 that can prevent some buffer overflow attacks.
6550 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
6551 is a built-in construct that returns a constant number of bytes from
6552 @var{ptr} to the end of the object @var{ptr} pointer points to
6553 (if known at compile time). @code{__builtin_object_size} never evaluates
6554 its arguments for side-effects. If there are any side-effects in them, it
6555 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6556 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
6557 point to and all of them are known at compile time, the returned number
6558 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
6559 0 and minimum if nonzero. If it is not possible to determine which objects
6560 @var{ptr} points to at compile time, @code{__builtin_object_size} should
6561 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6562 for @var{type} 2 or 3.
6564 @var{type} is an integer constant from 0 to 3. If the least significant
6565 bit is clear, objects are whole variables, if it is set, a closest
6566 surrounding subobject is considered the object a pointer points to.
6567 The second bit determines if maximum or minimum of remaining bytes
6571 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
6572 char *p = &var.buf1[1], *q = &var.b;
6574 /* Here the object p points to is var. */
6575 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
6576 /* The subobject p points to is var.buf1. */
6577 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
6578 /* The object q points to is var. */
6579 assert (__builtin_object_size (q, 0)
6580 == (char *) (&var + 1) - (char *) &var.b);
6581 /* The subobject q points to is var.b. */
6582 assert (__builtin_object_size (q, 1) == sizeof (var.b));
6586 There are built-in functions added for many common string operation
6587 functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk}
6588 built-in is provided. This built-in has an additional last argument,
6589 which is the number of bytes remaining in object the @var{dest}
6590 argument points to or @code{(size_t) -1} if the size is not known.
6592 The built-in functions are optimized into the normal string functions
6593 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
6594 it is known at compile time that the destination object will not
6595 be overflown. If the compiler can determine at compile time the
6596 object will be always overflown, it issues a warning.
6598 The intended use can be e.g.
6602 #define bos0(dest) __builtin_object_size (dest, 0)
6603 #define memcpy(dest, src, n) \
6604 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
6608 /* It is unknown what object p points to, so this is optimized
6609 into plain memcpy - no checking is possible. */
6610 memcpy (p, "abcde", n);
6611 /* Destination is known and length too. It is known at compile
6612 time there will be no overflow. */
6613 memcpy (&buf[5], "abcde", 5);
6614 /* Destination is known, but the length is not known at compile time.
6615 This will result in __memcpy_chk call that can check for overflow
6617 memcpy (&buf[5], "abcde", n);
6618 /* Destination is known and it is known at compile time there will
6619 be overflow. There will be a warning and __memcpy_chk call that
6620 will abort the program at runtime. */
6621 memcpy (&buf[6], "abcde", 5);
6624 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
6625 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
6626 @code{strcat} and @code{strncat}.
6628 There are also checking built-in functions for formatted output functions.
6630 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
6631 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6632 const char *fmt, ...);
6633 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
6635 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6636 const char *fmt, va_list ap);
6639 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
6640 etc.@: functions and can contain implementation specific flags on what
6641 additional security measures the checking function might take, such as
6642 handling @code{%n} differently.
6644 The @var{os} argument is the object size @var{s} points to, like in the
6645 other built-in functions. There is a small difference in the behavior
6646 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
6647 optimized into the non-checking functions only if @var{flag} is 0, otherwise
6648 the checking function is called with @var{os} argument set to
6651 In addition to this, there are checking built-in functions
6652 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
6653 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
6654 These have just one additional argument, @var{flag}, right before
6655 format string @var{fmt}. If the compiler is able to optimize them to
6656 @code{fputc} etc.@: functions, it will, otherwise the checking function
6657 should be called and the @var{flag} argument passed to it.
6659 @node Other Builtins
6660 @section Other built-in functions provided by GCC
6661 @cindex built-in functions
6662 @findex __builtin_fpclassify
6663 @findex __builtin_isfinite
6664 @findex __builtin_isnormal
6665 @findex __builtin_isgreater
6666 @findex __builtin_isgreaterequal
6667 @findex __builtin_isinf_sign
6668 @findex __builtin_isless
6669 @findex __builtin_islessequal
6670 @findex __builtin_islessgreater
6671 @findex __builtin_isunordered
6672 @findex __builtin_powi
6673 @findex __builtin_powif
6674 @findex __builtin_powil
6832 @findex fprintf_unlocked
6834 @findex fputs_unlocked
6951 @findex printf_unlocked
6983 @findex significandf
6984 @findex significandl
7055 GCC provides a large number of built-in functions other than the ones
7056 mentioned above. Some of these are for internal use in the processing
7057 of exceptions or variable-length argument lists and will not be
7058 documented here because they may change from time to time; we do not
7059 recommend general use of these functions.
7061 The remaining functions are provided for optimization purposes.
7063 @opindex fno-builtin
7064 GCC includes built-in versions of many of the functions in the standard
7065 C library. The versions prefixed with @code{__builtin_} will always be
7066 treated as having the same meaning as the C library function even if you
7067 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
7068 Many of these functions are only optimized in certain cases; if they are
7069 not optimized in a particular case, a call to the library function will
7074 Outside strict ISO C mode (@option{-ansi}, @option{-std=c90},
7075 @option{-std=c99} or @option{-std=c1x}), the functions
7076 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
7077 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
7078 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
7079 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
7080 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
7081 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
7082 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
7083 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
7084 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
7085 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
7086 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
7087 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
7088 @code{signbitd64}, @code{signbitd128}, @code{significandf},
7089 @code{significandl}, @code{significand}, @code{sincosf},
7090 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
7091 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
7092 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
7093 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
7095 may be handled as built-in functions.
7096 All these functions have corresponding versions
7097 prefixed with @code{__builtin_}, which may be used even in strict C90
7100 The ISO C99 functions
7101 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
7102 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
7103 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
7104 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
7105 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
7106 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
7107 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
7108 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
7109 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
7110 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
7111 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
7112 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
7113 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
7114 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
7115 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
7116 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
7117 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
7118 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
7119 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
7120 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
7121 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
7122 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
7123 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
7124 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
7125 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
7126 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
7127 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
7128 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
7129 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
7130 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
7131 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
7132 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
7133 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
7134 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
7135 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
7136 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
7137 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
7138 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
7139 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
7140 are handled as built-in functions
7141 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
7143 There are also built-in versions of the ISO C99 functions
7144 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
7145 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
7146 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
7147 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
7148 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
7149 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
7150 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
7151 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
7152 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
7153 that are recognized in any mode since ISO C90 reserves these names for
7154 the purpose to which ISO C99 puts them. All these functions have
7155 corresponding versions prefixed with @code{__builtin_}.
7157 The ISO C94 functions
7158 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
7159 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
7160 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
7162 are handled as built-in functions
7163 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
7165 The ISO C90 functions
7166 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
7167 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
7168 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
7169 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
7170 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
7171 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
7172 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
7173 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
7174 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
7175 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
7176 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
7177 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
7178 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
7179 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
7180 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
7181 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
7182 are all recognized as built-in functions unless
7183 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
7184 is specified for an individual function). All of these functions have
7185 corresponding versions prefixed with @code{__builtin_}.
7187 GCC provides built-in versions of the ISO C99 floating point comparison
7188 macros that avoid raising exceptions for unordered operands. They have
7189 the same names as the standard macros ( @code{isgreater},
7190 @code{isgreaterequal}, @code{isless}, @code{islessequal},
7191 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
7192 prefixed. We intend for a library implementor to be able to simply
7193 @code{#define} each standard macro to its built-in equivalent.
7194 In the same fashion, GCC provides @code{fpclassify}, @code{isfinite},
7195 @code{isinf_sign} and @code{isnormal} built-ins used with
7196 @code{__builtin_} prefixed. The @code{isinf} and @code{isnan}
7197 builtins appear both with and without the @code{__builtin_} prefix.
7199 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
7201 You can use the built-in function @code{__builtin_types_compatible_p} to
7202 determine whether two types are the same.
7204 This built-in function returns 1 if the unqualified versions of the
7205 types @var{type1} and @var{type2} (which are types, not expressions) are
7206 compatible, 0 otherwise. The result of this built-in function can be
7207 used in integer constant expressions.
7209 This built-in function ignores top level qualifiers (e.g., @code{const},
7210 @code{volatile}). For example, @code{int} is equivalent to @code{const
7213 The type @code{int[]} and @code{int[5]} are compatible. On the other
7214 hand, @code{int} and @code{char *} are not compatible, even if the size
7215 of their types, on the particular architecture are the same. Also, the
7216 amount of pointer indirection is taken into account when determining
7217 similarity. Consequently, @code{short *} is not similar to
7218 @code{short **}. Furthermore, two types that are typedefed are
7219 considered compatible if their underlying types are compatible.
7221 An @code{enum} type is not considered to be compatible with another
7222 @code{enum} type even if both are compatible with the same integer
7223 type; this is what the C standard specifies.
7224 For example, @code{enum @{foo, bar@}} is not similar to
7225 @code{enum @{hot, dog@}}.
7227 You would typically use this function in code whose execution varies
7228 depending on the arguments' types. For example:
7233 typeof (x) tmp = (x); \
7234 if (__builtin_types_compatible_p (typeof (x), long double)) \
7235 tmp = foo_long_double (tmp); \
7236 else if (__builtin_types_compatible_p (typeof (x), double)) \
7237 tmp = foo_double (tmp); \
7238 else if (__builtin_types_compatible_p (typeof (x), float)) \
7239 tmp = foo_float (tmp); \
7246 @emph{Note:} This construct is only available for C@.
7250 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
7252 You can use the built-in function @code{__builtin_choose_expr} to
7253 evaluate code depending on the value of a constant expression. This
7254 built-in function returns @var{exp1} if @var{const_exp}, which is an
7255 integer constant expression, is nonzero. Otherwise it returns @var{exp2}.
7257 This built-in function is analogous to the @samp{? :} operator in C,
7258 except that the expression returned has its type unaltered by promotion
7259 rules. Also, the built-in function does not evaluate the expression
7260 that was not chosen. For example, if @var{const_exp} evaluates to true,
7261 @var{exp2} is not evaluated even if it has side-effects.
7263 This built-in function can return an lvalue if the chosen argument is an
7266 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
7267 type. Similarly, if @var{exp2} is returned, its return type is the same
7274 __builtin_choose_expr ( \
7275 __builtin_types_compatible_p (typeof (x), double), \
7277 __builtin_choose_expr ( \
7278 __builtin_types_compatible_p (typeof (x), float), \
7280 /* @r{The void expression results in a compile-time error} \
7281 @r{when assigning the result to something.} */ \
7285 @emph{Note:} This construct is only available for C@. Furthermore, the
7286 unused expression (@var{exp1} or @var{exp2} depending on the value of
7287 @var{const_exp}) may still generate syntax errors. This may change in
7292 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
7293 You can use the built-in function @code{__builtin_constant_p} to
7294 determine if a value is known to be constant at compile-time and hence
7295 that GCC can perform constant-folding on expressions involving that
7296 value. The argument of the function is the value to test. The function
7297 returns the integer 1 if the argument is known to be a compile-time
7298 constant and 0 if it is not known to be a compile-time constant. A
7299 return of 0 does not indicate that the value is @emph{not} a constant,
7300 but merely that GCC cannot prove it is a constant with the specified
7301 value of the @option{-O} option.
7303 You would typically use this function in an embedded application where
7304 memory was a critical resource. If you have some complex calculation,
7305 you may want it to be folded if it involves constants, but need to call
7306 a function if it does not. For example:
7309 #define Scale_Value(X) \
7310 (__builtin_constant_p (X) \
7311 ? ((X) * SCALE + OFFSET) : Scale (X))
7314 You may use this built-in function in either a macro or an inline
7315 function. However, if you use it in an inlined function and pass an
7316 argument of the function as the argument to the built-in, GCC will
7317 never return 1 when you call the inline function with a string constant
7318 or compound literal (@pxref{Compound Literals}) and will not return 1
7319 when you pass a constant numeric value to the inline function unless you
7320 specify the @option{-O} option.
7322 You may also use @code{__builtin_constant_p} in initializers for static
7323 data. For instance, you can write
7326 static const int table[] = @{
7327 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
7333 This is an acceptable initializer even if @var{EXPRESSION} is not a
7334 constant expression, including the case where
7335 @code{__builtin_constant_p} returns 1 because @var{EXPRESSION} can be
7336 folded to a constant but @var{EXPRESSION} contains operands that would
7337 not otherwise be permitted in a static initializer (for example,
7338 @code{0 && foo ()}). GCC must be more conservative about evaluating the
7339 built-in in this case, because it has no opportunity to perform
7342 Previous versions of GCC did not accept this built-in in data
7343 initializers. The earliest version where it is completely safe is
7347 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
7348 @opindex fprofile-arcs
7349 You may use @code{__builtin_expect} to provide the compiler with
7350 branch prediction information. In general, you should prefer to
7351 use actual profile feedback for this (@option{-fprofile-arcs}), as
7352 programmers are notoriously bad at predicting how their programs
7353 actually perform. However, there are applications in which this
7354 data is hard to collect.
7356 The return value is the value of @var{exp}, which should be an integral
7357 expression. The semantics of the built-in are that it is expected that
7358 @var{exp} == @var{c}. For example:
7361 if (__builtin_expect (x, 0))
7366 would indicate that we do not expect to call @code{foo}, since
7367 we expect @code{x} to be zero. Since you are limited to integral
7368 expressions for @var{exp}, you should use constructions such as
7371 if (__builtin_expect (ptr != NULL, 1))
7376 when testing pointer or floating-point values.
7379 @deftypefn {Built-in Function} void __builtin_trap (void)
7380 This function causes the program to exit abnormally. GCC implements
7381 this function by using a target-dependent mechanism (such as
7382 intentionally executing an illegal instruction) or by calling
7383 @code{abort}. The mechanism used may vary from release to release so
7384 you should not rely on any particular implementation.
7387 @deftypefn {Built-in Function} void __builtin_unreachable (void)
7388 If control flow reaches the point of the @code{__builtin_unreachable},
7389 the program is undefined. It is useful in situations where the
7390 compiler cannot deduce the unreachability of the code.
7392 One such case is immediately following an @code{asm} statement that
7393 will either never terminate, or one that transfers control elsewhere
7394 and never returns. In this example, without the
7395 @code{__builtin_unreachable}, GCC would issue a warning that control
7396 reaches the end of a non-void function. It would also generate code
7397 to return after the @code{asm}.
7400 int f (int c, int v)
7408 asm("jmp error_handler");
7409 __builtin_unreachable ();
7414 Because the @code{asm} statement unconditionally transfers control out
7415 of the function, control will never reach the end of the function
7416 body. The @code{__builtin_unreachable} is in fact unreachable and
7417 communicates this fact to the compiler.
7419 Another use for @code{__builtin_unreachable} is following a call a
7420 function that never returns but that is not declared
7421 @code{__attribute__((noreturn))}, as in this example:
7424 void function_that_never_returns (void);
7434 function_that_never_returns ();
7435 __builtin_unreachable ();
7442 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
7443 This function is used to flush the processor's instruction cache for
7444 the region of memory between @var{begin} inclusive and @var{end}
7445 exclusive. Some targets require that the instruction cache be
7446 flushed, after modifying memory containing code, in order to obtain
7447 deterministic behavior.
7449 If the target does not require instruction cache flushes,
7450 @code{__builtin___clear_cache} has no effect. Otherwise either
7451 instructions are emitted in-line to clear the instruction cache or a
7452 call to the @code{__clear_cache} function in libgcc is made.
7455 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
7456 This function is used to minimize cache-miss latency by moving data into
7457 a cache before it is accessed.
7458 You can insert calls to @code{__builtin_prefetch} into code for which
7459 you know addresses of data in memory that is likely to be accessed soon.
7460 If the target supports them, data prefetch instructions will be generated.
7461 If the prefetch is done early enough before the access then the data will
7462 be in the cache by the time it is accessed.
7464 The value of @var{addr} is the address of the memory to prefetch.
7465 There are two optional arguments, @var{rw} and @var{locality}.
7466 The value of @var{rw} is a compile-time constant one or zero; one
7467 means that the prefetch is preparing for a write to the memory address
7468 and zero, the default, means that the prefetch is preparing for a read.
7469 The value @var{locality} must be a compile-time constant integer between
7470 zero and three. A value of zero means that the data has no temporal
7471 locality, so it need not be left in the cache after the access. A value
7472 of three means that the data has a high degree of temporal locality and
7473 should be left in all levels of cache possible. Values of one and two
7474 mean, respectively, a low or moderate degree of temporal locality. The
7478 for (i = 0; i < n; i++)
7481 __builtin_prefetch (&a[i+j], 1, 1);
7482 __builtin_prefetch (&b[i+j], 0, 1);
7487 Data prefetch does not generate faults if @var{addr} is invalid, but
7488 the address expression itself must be valid. For example, a prefetch
7489 of @code{p->next} will not fault if @code{p->next} is not a valid
7490 address, but evaluation will fault if @code{p} is not a valid address.
7492 If the target does not support data prefetch, the address expression
7493 is evaluated if it includes side effects but no other code is generated
7494 and GCC does not issue a warning.
7497 @deftypefn {Built-in Function} double __builtin_huge_val (void)
7498 Returns a positive infinity, if supported by the floating-point format,
7499 else @code{DBL_MAX}. This function is suitable for implementing the
7500 ISO C macro @code{HUGE_VAL}.
7503 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
7504 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
7507 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
7508 Similar to @code{__builtin_huge_val}, except the return
7509 type is @code{long double}.
7512 @deftypefn {Built-in Function} int __builtin_fpclassify (int, int, int, int, int, ...)
7513 This built-in implements the C99 fpclassify functionality. The first
7514 five int arguments should be the target library's notion of the
7515 possible FP classes and are used for return values. They must be
7516 constant values and they must appear in this order: @code{FP_NAN},
7517 @code{FP_INFINITE}, @code{FP_NORMAL}, @code{FP_SUBNORMAL} and
7518 @code{FP_ZERO}. The ellipsis is for exactly one floating point value
7519 to classify. GCC treats the last argument as type-generic, which
7520 means it does not do default promotion from float to double.
7523 @deftypefn {Built-in Function} double __builtin_inf (void)
7524 Similar to @code{__builtin_huge_val}, except a warning is generated
7525 if the target floating-point format does not support infinities.
7528 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
7529 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
7532 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
7533 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
7536 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
7537 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
7540 @deftypefn {Built-in Function} float __builtin_inff (void)
7541 Similar to @code{__builtin_inf}, except the return type is @code{float}.
7542 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
7545 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
7546 Similar to @code{__builtin_inf}, except the return
7547 type is @code{long double}.
7550 @deftypefn {Built-in Function} int __builtin_isinf_sign (...)
7551 Similar to @code{isinf}, except the return value will be negative for
7552 an argument of @code{-Inf}. Note while the parameter list is an
7553 ellipsis, this function only accepts exactly one floating point
7554 argument. GCC treats this parameter as type-generic, which means it
7555 does not do default promotion from float to double.
7558 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
7559 This is an implementation of the ISO C99 function @code{nan}.
7561 Since ISO C99 defines this function in terms of @code{strtod}, which we
7562 do not implement, a description of the parsing is in order. The string
7563 is parsed as by @code{strtol}; that is, the base is recognized by
7564 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
7565 in the significand such that the least significant bit of the number
7566 is at the least significant bit of the significand. The number is
7567 truncated to fit the significand field provided. The significand is
7568 forced to be a quiet NaN@.
7570 This function, if given a string literal all of which would have been
7571 consumed by strtol, is evaluated early enough that it is considered a
7572 compile-time constant.
7575 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
7576 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
7579 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
7580 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
7583 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
7584 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
7587 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
7588 Similar to @code{__builtin_nan}, except the return type is @code{float}.
7591 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
7592 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
7595 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
7596 Similar to @code{__builtin_nan}, except the significand is forced
7597 to be a signaling NaN@. The @code{nans} function is proposed by
7598 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
7601 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
7602 Similar to @code{__builtin_nans}, except the return type is @code{float}.
7605 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
7606 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
7609 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
7610 Returns one plus the index of the least significant 1-bit of @var{x}, or
7611 if @var{x} is zero, returns zero.
7614 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
7615 Returns the number of leading 0-bits in @var{x}, starting at the most
7616 significant bit position. If @var{x} is 0, the result is undefined.
7619 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
7620 Returns the number of trailing 0-bits in @var{x}, starting at the least
7621 significant bit position. If @var{x} is 0, the result is undefined.
7624 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
7625 Returns the number of 1-bits in @var{x}.
7628 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
7629 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
7633 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
7634 Similar to @code{__builtin_ffs}, except the argument type is
7635 @code{unsigned long}.
7638 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
7639 Similar to @code{__builtin_clz}, except the argument type is
7640 @code{unsigned long}.
7643 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
7644 Similar to @code{__builtin_ctz}, except the argument type is
7645 @code{unsigned long}.
7648 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
7649 Similar to @code{__builtin_popcount}, except the argument type is
7650 @code{unsigned long}.
7653 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
7654 Similar to @code{__builtin_parity}, except the argument type is
7655 @code{unsigned long}.
7658 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
7659 Similar to @code{__builtin_ffs}, except the argument type is
7660 @code{unsigned long long}.
7663 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
7664 Similar to @code{__builtin_clz}, except the argument type is
7665 @code{unsigned long long}.
7668 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
7669 Similar to @code{__builtin_ctz}, except the argument type is
7670 @code{unsigned long long}.
7673 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
7674 Similar to @code{__builtin_popcount}, except the argument type is
7675 @code{unsigned long long}.
7678 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
7679 Similar to @code{__builtin_parity}, except the argument type is
7680 @code{unsigned long long}.
7683 @deftypefn {Built-in Function} double __builtin_powi (double, int)
7684 Returns the first argument raised to the power of the second. Unlike the
7685 @code{pow} function no guarantees about precision and rounding are made.
7688 @deftypefn {Built-in Function} float __builtin_powif (float, int)
7689 Similar to @code{__builtin_powi}, except the argument and return types
7693 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
7694 Similar to @code{__builtin_powi}, except the argument and return types
7695 are @code{long double}.
7698 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
7699 Returns @var{x} with the order of the bytes reversed; for example,
7700 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
7704 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
7705 Similar to @code{__builtin_bswap32}, except the argument and return types
7709 @node Target Builtins
7710 @section Built-in Functions Specific to Particular Target Machines
7712 On some target machines, GCC supports many built-in functions specific
7713 to those machines. Generally these generate calls to specific machine
7714 instructions, but allow the compiler to schedule those calls.
7717 * Alpha Built-in Functions::
7718 * ARM iWMMXt Built-in Functions::
7719 * ARM NEON Intrinsics::
7720 * Blackfin Built-in Functions::
7721 * FR-V Built-in Functions::
7722 * X86 Built-in Functions::
7723 * MIPS DSP Built-in Functions::
7724 * MIPS Paired-Single Support::
7725 * MIPS Loongson Built-in Functions::
7726 * Other MIPS Built-in Functions::
7727 * picoChip Built-in Functions::
7728 * PowerPC AltiVec/VSX Built-in Functions::
7729 * RX Built-in Functions::
7730 * SPARC VIS Built-in Functions::
7731 * SPU Built-in Functions::
7734 @node Alpha Built-in Functions
7735 @subsection Alpha Built-in Functions
7737 These built-in functions are available for the Alpha family of
7738 processors, depending on the command-line switches used.
7740 The following built-in functions are always available. They
7741 all generate the machine instruction that is part of the name.
7744 long __builtin_alpha_implver (void)
7745 long __builtin_alpha_rpcc (void)
7746 long __builtin_alpha_amask (long)
7747 long __builtin_alpha_cmpbge (long, long)
7748 long __builtin_alpha_extbl (long, long)
7749 long __builtin_alpha_extwl (long, long)
7750 long __builtin_alpha_extll (long, long)
7751 long __builtin_alpha_extql (long, long)
7752 long __builtin_alpha_extwh (long, long)
7753 long __builtin_alpha_extlh (long, long)
7754 long __builtin_alpha_extqh (long, long)
7755 long __builtin_alpha_insbl (long, long)
7756 long __builtin_alpha_inswl (long, long)
7757 long __builtin_alpha_insll (long, long)
7758 long __builtin_alpha_insql (long, long)
7759 long __builtin_alpha_inswh (long, long)
7760 long __builtin_alpha_inslh (long, long)
7761 long __builtin_alpha_insqh (long, long)
7762 long __builtin_alpha_mskbl (long, long)
7763 long __builtin_alpha_mskwl (long, long)
7764 long __builtin_alpha_mskll (long, long)
7765 long __builtin_alpha_mskql (long, long)
7766 long __builtin_alpha_mskwh (long, long)
7767 long __builtin_alpha_msklh (long, long)
7768 long __builtin_alpha_mskqh (long, long)
7769 long __builtin_alpha_umulh (long, long)
7770 long __builtin_alpha_zap (long, long)
7771 long __builtin_alpha_zapnot (long, long)
7774 The following built-in functions are always with @option{-mmax}
7775 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
7776 later. They all generate the machine instruction that is part
7780 long __builtin_alpha_pklb (long)
7781 long __builtin_alpha_pkwb (long)
7782 long __builtin_alpha_unpkbl (long)
7783 long __builtin_alpha_unpkbw (long)
7784 long __builtin_alpha_minub8 (long, long)
7785 long __builtin_alpha_minsb8 (long, long)
7786 long __builtin_alpha_minuw4 (long, long)
7787 long __builtin_alpha_minsw4 (long, long)
7788 long __builtin_alpha_maxub8 (long, long)
7789 long __builtin_alpha_maxsb8 (long, long)
7790 long __builtin_alpha_maxuw4 (long, long)
7791 long __builtin_alpha_maxsw4 (long, long)
7792 long __builtin_alpha_perr (long, long)
7795 The following built-in functions are always with @option{-mcix}
7796 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
7797 later. They all generate the machine instruction that is part
7801 long __builtin_alpha_cttz (long)
7802 long __builtin_alpha_ctlz (long)
7803 long __builtin_alpha_ctpop (long)
7806 The following builtins are available on systems that use the OSF/1
7807 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
7808 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
7809 @code{rdval} and @code{wrval}.
7812 void *__builtin_thread_pointer (void)
7813 void __builtin_set_thread_pointer (void *)
7816 @node ARM iWMMXt Built-in Functions
7817 @subsection ARM iWMMXt Built-in Functions
7819 These built-in functions are available for the ARM family of
7820 processors when the @option{-mcpu=iwmmxt} switch is used:
7823 typedef int v2si __attribute__ ((vector_size (8)));
7824 typedef short v4hi __attribute__ ((vector_size (8)));
7825 typedef char v8qi __attribute__ ((vector_size (8)));
7827 int __builtin_arm_getwcx (int)
7828 void __builtin_arm_setwcx (int, int)
7829 int __builtin_arm_textrmsb (v8qi, int)
7830 int __builtin_arm_textrmsh (v4hi, int)
7831 int __builtin_arm_textrmsw (v2si, int)
7832 int __builtin_arm_textrmub (v8qi, int)
7833 int __builtin_arm_textrmuh (v4hi, int)
7834 int __builtin_arm_textrmuw (v2si, int)
7835 v8qi __builtin_arm_tinsrb (v8qi, int)
7836 v4hi __builtin_arm_tinsrh (v4hi, int)
7837 v2si __builtin_arm_tinsrw (v2si, int)
7838 long long __builtin_arm_tmia (long long, int, int)
7839 long long __builtin_arm_tmiabb (long long, int, int)
7840 long long __builtin_arm_tmiabt (long long, int, int)
7841 long long __builtin_arm_tmiaph (long long, int, int)
7842 long long __builtin_arm_tmiatb (long long, int, int)
7843 long long __builtin_arm_tmiatt (long long, int, int)
7844 int __builtin_arm_tmovmskb (v8qi)
7845 int __builtin_arm_tmovmskh (v4hi)
7846 int __builtin_arm_tmovmskw (v2si)
7847 long long __builtin_arm_waccb (v8qi)
7848 long long __builtin_arm_wacch (v4hi)
7849 long long __builtin_arm_waccw (v2si)
7850 v8qi __builtin_arm_waddb (v8qi, v8qi)
7851 v8qi __builtin_arm_waddbss (v8qi, v8qi)
7852 v8qi __builtin_arm_waddbus (v8qi, v8qi)
7853 v4hi __builtin_arm_waddh (v4hi, v4hi)
7854 v4hi __builtin_arm_waddhss (v4hi, v4hi)
7855 v4hi __builtin_arm_waddhus (v4hi, v4hi)
7856 v2si __builtin_arm_waddw (v2si, v2si)
7857 v2si __builtin_arm_waddwss (v2si, v2si)
7858 v2si __builtin_arm_waddwus (v2si, v2si)
7859 v8qi __builtin_arm_walign (v8qi, v8qi, int)
7860 long long __builtin_arm_wand(long long, long long)
7861 long long __builtin_arm_wandn (long long, long long)
7862 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
7863 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
7864 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
7865 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
7866 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
7867 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
7868 v2si __builtin_arm_wcmpeqw (v2si, v2si)
7869 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
7870 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
7871 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
7872 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
7873 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
7874 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
7875 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
7876 long long __builtin_arm_wmacsz (v4hi, v4hi)
7877 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
7878 long long __builtin_arm_wmacuz (v4hi, v4hi)
7879 v4hi __builtin_arm_wmadds (v4hi, v4hi)
7880 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
7881 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
7882 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
7883 v2si __builtin_arm_wmaxsw (v2si, v2si)
7884 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
7885 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
7886 v2si __builtin_arm_wmaxuw (v2si, v2si)
7887 v8qi __builtin_arm_wminsb (v8qi, v8qi)
7888 v4hi __builtin_arm_wminsh (v4hi, v4hi)
7889 v2si __builtin_arm_wminsw (v2si, v2si)
7890 v8qi __builtin_arm_wminub (v8qi, v8qi)
7891 v4hi __builtin_arm_wminuh (v4hi, v4hi)
7892 v2si __builtin_arm_wminuw (v2si, v2si)
7893 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
7894 v4hi __builtin_arm_wmulul (v4hi, v4hi)
7895 v4hi __builtin_arm_wmulum (v4hi, v4hi)
7896 long long __builtin_arm_wor (long long, long long)
7897 v2si __builtin_arm_wpackdss (long long, long long)
7898 v2si __builtin_arm_wpackdus (long long, long long)
7899 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
7900 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
7901 v4hi __builtin_arm_wpackwss (v2si, v2si)
7902 v4hi __builtin_arm_wpackwus (v2si, v2si)
7903 long long __builtin_arm_wrord (long long, long long)
7904 long long __builtin_arm_wrordi (long long, int)
7905 v4hi __builtin_arm_wrorh (v4hi, long long)
7906 v4hi __builtin_arm_wrorhi (v4hi, int)
7907 v2si __builtin_arm_wrorw (v2si, long long)
7908 v2si __builtin_arm_wrorwi (v2si, int)
7909 v2si __builtin_arm_wsadb (v8qi, v8qi)
7910 v2si __builtin_arm_wsadbz (v8qi, v8qi)
7911 v2si __builtin_arm_wsadh (v4hi, v4hi)
7912 v2si __builtin_arm_wsadhz (v4hi, v4hi)
7913 v4hi __builtin_arm_wshufh (v4hi, int)
7914 long long __builtin_arm_wslld (long long, long long)
7915 long long __builtin_arm_wslldi (long long, int)
7916 v4hi __builtin_arm_wsllh (v4hi, long long)
7917 v4hi __builtin_arm_wsllhi (v4hi, int)
7918 v2si __builtin_arm_wsllw (v2si, long long)
7919 v2si __builtin_arm_wsllwi (v2si, int)
7920 long long __builtin_arm_wsrad (long long, long long)
7921 long long __builtin_arm_wsradi (long long, int)
7922 v4hi __builtin_arm_wsrah (v4hi, long long)
7923 v4hi __builtin_arm_wsrahi (v4hi, int)
7924 v2si __builtin_arm_wsraw (v2si, long long)
7925 v2si __builtin_arm_wsrawi (v2si, int)
7926 long long __builtin_arm_wsrld (long long, long long)
7927 long long __builtin_arm_wsrldi (long long, int)
7928 v4hi __builtin_arm_wsrlh (v4hi, long long)
7929 v4hi __builtin_arm_wsrlhi (v4hi, int)
7930 v2si __builtin_arm_wsrlw (v2si, long long)
7931 v2si __builtin_arm_wsrlwi (v2si, int)
7932 v8qi __builtin_arm_wsubb (v8qi, v8qi)
7933 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
7934 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
7935 v4hi __builtin_arm_wsubh (v4hi, v4hi)
7936 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
7937 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
7938 v2si __builtin_arm_wsubw (v2si, v2si)
7939 v2si __builtin_arm_wsubwss (v2si, v2si)
7940 v2si __builtin_arm_wsubwus (v2si, v2si)
7941 v4hi __builtin_arm_wunpckehsb (v8qi)
7942 v2si __builtin_arm_wunpckehsh (v4hi)
7943 long long __builtin_arm_wunpckehsw (v2si)
7944 v4hi __builtin_arm_wunpckehub (v8qi)
7945 v2si __builtin_arm_wunpckehuh (v4hi)
7946 long long __builtin_arm_wunpckehuw (v2si)
7947 v4hi __builtin_arm_wunpckelsb (v8qi)
7948 v2si __builtin_arm_wunpckelsh (v4hi)
7949 long long __builtin_arm_wunpckelsw (v2si)
7950 v4hi __builtin_arm_wunpckelub (v8qi)
7951 v2si __builtin_arm_wunpckeluh (v4hi)
7952 long long __builtin_arm_wunpckeluw (v2si)
7953 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
7954 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
7955 v2si __builtin_arm_wunpckihw (v2si, v2si)
7956 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
7957 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
7958 v2si __builtin_arm_wunpckilw (v2si, v2si)
7959 long long __builtin_arm_wxor (long long, long long)
7960 long long __builtin_arm_wzero ()
7963 @node ARM NEON Intrinsics
7964 @subsection ARM NEON Intrinsics
7966 These built-in intrinsics for the ARM Advanced SIMD extension are available
7967 when the @option{-mfpu=neon} switch is used:
7969 @include arm-neon-intrinsics.texi
7971 @node Blackfin Built-in Functions
7972 @subsection Blackfin Built-in Functions
7974 Currently, there are two Blackfin-specific built-in functions. These are
7975 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
7976 using inline assembly; by using these built-in functions the compiler can
7977 automatically add workarounds for hardware errata involving these
7978 instructions. These functions are named as follows:
7981 void __builtin_bfin_csync (void)
7982 void __builtin_bfin_ssync (void)
7985 @node FR-V Built-in Functions
7986 @subsection FR-V Built-in Functions
7988 GCC provides many FR-V-specific built-in functions. In general,
7989 these functions are intended to be compatible with those described
7990 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
7991 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
7992 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
7993 pointer rather than by value.
7995 Most of the functions are named after specific FR-V instructions.
7996 Such functions are said to be ``directly mapped'' and are summarized
7997 here in tabular form.
8001 * Directly-mapped Integer Functions::
8002 * Directly-mapped Media Functions::
8003 * Raw read/write Functions::
8004 * Other Built-in Functions::
8007 @node Argument Types
8008 @subsubsection Argument Types
8010 The arguments to the built-in functions can be divided into three groups:
8011 register numbers, compile-time constants and run-time values. In order
8012 to make this classification clear at a glance, the arguments and return
8013 values are given the following pseudo types:
8015 @multitable @columnfractions .20 .30 .15 .35
8016 @item Pseudo type @tab Real C type @tab Constant? @tab Description
8017 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
8018 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
8019 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
8020 @item @code{uw2} @tab @code{unsigned long long} @tab No
8021 @tab an unsigned doubleword
8022 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
8023 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
8024 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
8025 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
8028 These pseudo types are not defined by GCC, they are simply a notational
8029 convenience used in this manual.
8031 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
8032 and @code{sw2} are evaluated at run time. They correspond to
8033 register operands in the underlying FR-V instructions.
8035 @code{const} arguments represent immediate operands in the underlying
8036 FR-V instructions. They must be compile-time constants.
8038 @code{acc} arguments are evaluated at compile time and specify the number
8039 of an accumulator register. For example, an @code{acc} argument of 2
8040 will select the ACC2 register.
8042 @code{iacc} arguments are similar to @code{acc} arguments but specify the
8043 number of an IACC register. See @pxref{Other Built-in Functions}
8046 @node Directly-mapped Integer Functions
8047 @subsubsection Directly-mapped Integer Functions
8049 The functions listed below map directly to FR-V I-type instructions.
8051 @multitable @columnfractions .45 .32 .23
8052 @item Function prototype @tab Example usage @tab Assembly output
8053 @item @code{sw1 __ADDSS (sw1, sw1)}
8054 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
8055 @tab @code{ADDSS @var{a},@var{b},@var{c}}
8056 @item @code{sw1 __SCAN (sw1, sw1)}
8057 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
8058 @tab @code{SCAN @var{a},@var{b},@var{c}}
8059 @item @code{sw1 __SCUTSS (sw1)}
8060 @tab @code{@var{b} = __SCUTSS (@var{a})}
8061 @tab @code{SCUTSS @var{a},@var{b}}
8062 @item @code{sw1 __SLASS (sw1, sw1)}
8063 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
8064 @tab @code{SLASS @var{a},@var{b},@var{c}}
8065 @item @code{void __SMASS (sw1, sw1)}
8066 @tab @code{__SMASS (@var{a}, @var{b})}
8067 @tab @code{SMASS @var{a},@var{b}}
8068 @item @code{void __SMSSS (sw1, sw1)}
8069 @tab @code{__SMSSS (@var{a}, @var{b})}
8070 @tab @code{SMSSS @var{a},@var{b}}
8071 @item @code{void __SMU (sw1, sw1)}
8072 @tab @code{__SMU (@var{a}, @var{b})}
8073 @tab @code{SMU @var{a},@var{b}}
8074 @item @code{sw2 __SMUL (sw1, sw1)}
8075 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
8076 @tab @code{SMUL @var{a},@var{b},@var{c}}
8077 @item @code{sw1 __SUBSS (sw1, sw1)}
8078 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
8079 @tab @code{SUBSS @var{a},@var{b},@var{c}}
8080 @item @code{uw2 __UMUL (uw1, uw1)}
8081 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
8082 @tab @code{UMUL @var{a},@var{b},@var{c}}
8085 @node Directly-mapped Media Functions
8086 @subsubsection Directly-mapped Media Functions
8088 The functions listed below map directly to FR-V M-type instructions.
8090 @multitable @columnfractions .45 .32 .23
8091 @item Function prototype @tab Example usage @tab Assembly output
8092 @item @code{uw1 __MABSHS (sw1)}
8093 @tab @code{@var{b} = __MABSHS (@var{a})}
8094 @tab @code{MABSHS @var{a},@var{b}}
8095 @item @code{void __MADDACCS (acc, acc)}
8096 @tab @code{__MADDACCS (@var{b}, @var{a})}
8097 @tab @code{MADDACCS @var{a},@var{b}}
8098 @item @code{sw1 __MADDHSS (sw1, sw1)}
8099 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
8100 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
8101 @item @code{uw1 __MADDHUS (uw1, uw1)}
8102 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
8103 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
8104 @item @code{uw1 __MAND (uw1, uw1)}
8105 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
8106 @tab @code{MAND @var{a},@var{b},@var{c}}
8107 @item @code{void __MASACCS (acc, acc)}
8108 @tab @code{__MASACCS (@var{b}, @var{a})}
8109 @tab @code{MASACCS @var{a},@var{b}}
8110 @item @code{uw1 __MAVEH (uw1, uw1)}
8111 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
8112 @tab @code{MAVEH @var{a},@var{b},@var{c}}
8113 @item @code{uw2 __MBTOH (uw1)}
8114 @tab @code{@var{b} = __MBTOH (@var{a})}
8115 @tab @code{MBTOH @var{a},@var{b}}
8116 @item @code{void __MBTOHE (uw1 *, uw1)}
8117 @tab @code{__MBTOHE (&@var{b}, @var{a})}
8118 @tab @code{MBTOHE @var{a},@var{b}}
8119 @item @code{void __MCLRACC (acc)}
8120 @tab @code{__MCLRACC (@var{a})}
8121 @tab @code{MCLRACC @var{a}}
8122 @item @code{void __MCLRACCA (void)}
8123 @tab @code{__MCLRACCA ()}
8124 @tab @code{MCLRACCA}
8125 @item @code{uw1 __Mcop1 (uw1, uw1)}
8126 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
8127 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
8128 @item @code{uw1 __Mcop2 (uw1, uw1)}
8129 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
8130 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
8131 @item @code{uw1 __MCPLHI (uw2, const)}
8132 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
8133 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
8134 @item @code{uw1 __MCPLI (uw2, const)}
8135 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
8136 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
8137 @item @code{void __MCPXIS (acc, sw1, sw1)}
8138 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
8139 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
8140 @item @code{void __MCPXIU (acc, uw1, uw1)}
8141 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
8142 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
8143 @item @code{void __MCPXRS (acc, sw1, sw1)}
8144 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
8145 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
8146 @item @code{void __MCPXRU (acc, uw1, uw1)}
8147 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
8148 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
8149 @item @code{uw1 __MCUT (acc, uw1)}
8150 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
8151 @tab @code{MCUT @var{a},@var{b},@var{c}}
8152 @item @code{uw1 __MCUTSS (acc, sw1)}
8153 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
8154 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
8155 @item @code{void __MDADDACCS (acc, acc)}
8156 @tab @code{__MDADDACCS (@var{b}, @var{a})}
8157 @tab @code{MDADDACCS @var{a},@var{b}}
8158 @item @code{void __MDASACCS (acc, acc)}
8159 @tab @code{__MDASACCS (@var{b}, @var{a})}
8160 @tab @code{MDASACCS @var{a},@var{b}}
8161 @item @code{uw2 __MDCUTSSI (acc, const)}
8162 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
8163 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
8164 @item @code{uw2 __MDPACKH (uw2, uw2)}
8165 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
8166 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
8167 @item @code{uw2 __MDROTLI (uw2, const)}
8168 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
8169 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
8170 @item @code{void __MDSUBACCS (acc, acc)}
8171 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
8172 @tab @code{MDSUBACCS @var{a},@var{b}}
8173 @item @code{void __MDUNPACKH (uw1 *, uw2)}
8174 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
8175 @tab @code{MDUNPACKH @var{a},@var{b}}
8176 @item @code{uw2 __MEXPDHD (uw1, const)}
8177 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
8178 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
8179 @item @code{uw1 __MEXPDHW (uw1, const)}
8180 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
8181 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
8182 @item @code{uw1 __MHDSETH (uw1, const)}
8183 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
8184 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
8185 @item @code{sw1 __MHDSETS (const)}
8186 @tab @code{@var{b} = __MHDSETS (@var{a})}
8187 @tab @code{MHDSETS #@var{a},@var{b}}
8188 @item @code{uw1 __MHSETHIH (uw1, const)}
8189 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
8190 @tab @code{MHSETHIH #@var{a},@var{b}}
8191 @item @code{sw1 __MHSETHIS (sw1, const)}
8192 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
8193 @tab @code{MHSETHIS #@var{a},@var{b}}
8194 @item @code{uw1 __MHSETLOH (uw1, const)}
8195 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
8196 @tab @code{MHSETLOH #@var{a},@var{b}}
8197 @item @code{sw1 __MHSETLOS (sw1, const)}
8198 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
8199 @tab @code{MHSETLOS #@var{a},@var{b}}
8200 @item @code{uw1 __MHTOB (uw2)}
8201 @tab @code{@var{b} = __MHTOB (@var{a})}
8202 @tab @code{MHTOB @var{a},@var{b}}
8203 @item @code{void __MMACHS (acc, sw1, sw1)}
8204 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
8205 @tab @code{MMACHS @var{a},@var{b},@var{c}}
8206 @item @code{void __MMACHU (acc, uw1, uw1)}
8207 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
8208 @tab @code{MMACHU @var{a},@var{b},@var{c}}
8209 @item @code{void __MMRDHS (acc, sw1, sw1)}
8210 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
8211 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
8212 @item @code{void __MMRDHU (acc, uw1, uw1)}
8213 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
8214 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
8215 @item @code{void __MMULHS (acc, sw1, sw1)}
8216 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
8217 @tab @code{MMULHS @var{a},@var{b},@var{c}}
8218 @item @code{void __MMULHU (acc, uw1, uw1)}
8219 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
8220 @tab @code{MMULHU @var{a},@var{b},@var{c}}
8221 @item @code{void __MMULXHS (acc, sw1, sw1)}
8222 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
8223 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
8224 @item @code{void __MMULXHU (acc, uw1, uw1)}
8225 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
8226 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
8227 @item @code{uw1 __MNOT (uw1)}
8228 @tab @code{@var{b} = __MNOT (@var{a})}
8229 @tab @code{MNOT @var{a},@var{b}}
8230 @item @code{uw1 __MOR (uw1, uw1)}
8231 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
8232 @tab @code{MOR @var{a},@var{b},@var{c}}
8233 @item @code{uw1 __MPACKH (uh, uh)}
8234 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
8235 @tab @code{MPACKH @var{a},@var{b},@var{c}}
8236 @item @code{sw2 __MQADDHSS (sw2, sw2)}
8237 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
8238 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
8239 @item @code{uw2 __MQADDHUS (uw2, uw2)}
8240 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
8241 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
8242 @item @code{void __MQCPXIS (acc, sw2, sw2)}
8243 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
8244 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
8245 @item @code{void __MQCPXIU (acc, uw2, uw2)}
8246 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
8247 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
8248 @item @code{void __MQCPXRS (acc, sw2, sw2)}
8249 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
8250 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
8251 @item @code{void __MQCPXRU (acc, uw2, uw2)}
8252 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
8253 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
8254 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
8255 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
8256 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
8257 @item @code{sw2 __MQLMTHS (sw2, sw2)}
8258 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
8259 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
8260 @item @code{void __MQMACHS (acc, sw2, sw2)}
8261 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
8262 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
8263 @item @code{void __MQMACHU (acc, uw2, uw2)}
8264 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
8265 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
8266 @item @code{void __MQMACXHS (acc, sw2, sw2)}
8267 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
8268 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
8269 @item @code{void __MQMULHS (acc, sw2, sw2)}
8270 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
8271 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
8272 @item @code{void __MQMULHU (acc, uw2, uw2)}
8273 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
8274 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
8275 @item @code{void __MQMULXHS (acc, sw2, sw2)}
8276 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
8277 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
8278 @item @code{void __MQMULXHU (acc, uw2, uw2)}
8279 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
8280 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
8281 @item @code{sw2 __MQSATHS (sw2, sw2)}
8282 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
8283 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
8284 @item @code{uw2 __MQSLLHI (uw2, int)}
8285 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
8286 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
8287 @item @code{sw2 __MQSRAHI (sw2, int)}
8288 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
8289 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
8290 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
8291 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
8292 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
8293 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
8294 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
8295 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
8296 @item @code{void __MQXMACHS (acc, sw2, sw2)}
8297 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
8298 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
8299 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
8300 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
8301 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
8302 @item @code{uw1 __MRDACC (acc)}
8303 @tab @code{@var{b} = __MRDACC (@var{a})}
8304 @tab @code{MRDACC @var{a},@var{b}}
8305 @item @code{uw1 __MRDACCG (acc)}
8306 @tab @code{@var{b} = __MRDACCG (@var{a})}
8307 @tab @code{MRDACCG @var{a},@var{b}}
8308 @item @code{uw1 __MROTLI (uw1, const)}
8309 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
8310 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
8311 @item @code{uw1 __MROTRI (uw1, const)}
8312 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
8313 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
8314 @item @code{sw1 __MSATHS (sw1, sw1)}
8315 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
8316 @tab @code{MSATHS @var{a},@var{b},@var{c}}
8317 @item @code{uw1 __MSATHU (uw1, uw1)}
8318 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
8319 @tab @code{MSATHU @var{a},@var{b},@var{c}}
8320 @item @code{uw1 __MSLLHI (uw1, const)}
8321 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
8322 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
8323 @item @code{sw1 __MSRAHI (sw1, const)}
8324 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
8325 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
8326 @item @code{uw1 __MSRLHI (uw1, const)}
8327 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
8328 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
8329 @item @code{void __MSUBACCS (acc, acc)}
8330 @tab @code{__MSUBACCS (@var{b}, @var{a})}
8331 @tab @code{MSUBACCS @var{a},@var{b}}
8332 @item @code{sw1 __MSUBHSS (sw1, sw1)}
8333 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
8334 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
8335 @item @code{uw1 __MSUBHUS (uw1, uw1)}
8336 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
8337 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
8338 @item @code{void __MTRAP (void)}
8339 @tab @code{__MTRAP ()}
8341 @item @code{uw2 __MUNPACKH (uw1)}
8342 @tab @code{@var{b} = __MUNPACKH (@var{a})}
8343 @tab @code{MUNPACKH @var{a},@var{b}}
8344 @item @code{uw1 __MWCUT (uw2, uw1)}
8345 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
8346 @tab @code{MWCUT @var{a},@var{b},@var{c}}
8347 @item @code{void __MWTACC (acc, uw1)}
8348 @tab @code{__MWTACC (@var{b}, @var{a})}
8349 @tab @code{MWTACC @var{a},@var{b}}
8350 @item @code{void __MWTACCG (acc, uw1)}
8351 @tab @code{__MWTACCG (@var{b}, @var{a})}
8352 @tab @code{MWTACCG @var{a},@var{b}}
8353 @item @code{uw1 __MXOR (uw1, uw1)}
8354 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
8355 @tab @code{MXOR @var{a},@var{b},@var{c}}
8358 @node Raw read/write Functions
8359 @subsubsection Raw read/write Functions
8361 This sections describes built-in functions related to read and write
8362 instructions to access memory. These functions generate
8363 @code{membar} instructions to flush the I/O load and stores where
8364 appropriate, as described in Fujitsu's manual described above.
8368 @item unsigned char __builtin_read8 (void *@var{data})
8369 @item unsigned short __builtin_read16 (void *@var{data})
8370 @item unsigned long __builtin_read32 (void *@var{data})
8371 @item unsigned long long __builtin_read64 (void *@var{data})
8373 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
8374 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
8375 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
8376 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
8379 @node Other Built-in Functions
8380 @subsubsection Other Built-in Functions
8382 This section describes built-in functions that are not named after
8383 a specific FR-V instruction.
8386 @item sw2 __IACCreadll (iacc @var{reg})
8387 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
8388 for future expansion and must be 0.
8390 @item sw1 __IACCreadl (iacc @var{reg})
8391 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
8392 Other values of @var{reg} are rejected as invalid.
8394 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
8395 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
8396 is reserved for future expansion and must be 0.
8398 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
8399 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
8400 is 1. Other values of @var{reg} are rejected as invalid.
8402 @item void __data_prefetch0 (const void *@var{x})
8403 Use the @code{dcpl} instruction to load the contents of address @var{x}
8404 into the data cache.
8406 @item void __data_prefetch (const void *@var{x})
8407 Use the @code{nldub} instruction to load the contents of address @var{x}
8408 into the data cache. The instruction will be issued in slot I1@.
8411 @node X86 Built-in Functions
8412 @subsection X86 Built-in Functions
8414 These built-in functions are available for the i386 and x86-64 family
8415 of computers, depending on the command-line switches used.
8417 Note that, if you specify command-line switches such as @option{-msse},
8418 the compiler could use the extended instruction sets even if the built-ins
8419 are not used explicitly in the program. For this reason, applications
8420 which perform runtime CPU detection must compile separate files for each
8421 supported architecture, using the appropriate flags. In particular,
8422 the file containing the CPU detection code should be compiled without
8425 The following machine modes are available for use with MMX built-in functions
8426 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
8427 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
8428 vector of eight 8-bit integers. Some of the built-in functions operate on
8429 MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode.
8431 If 3DNow!@: extensions are enabled, @code{V2SF} is used as a mode for a vector
8432 of two 32-bit floating point values.
8434 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
8435 floating point values. Some instructions use a vector of four 32-bit
8436 integers, these use @code{V4SI}. Finally, some instructions operate on an
8437 entire vector register, interpreting it as a 128-bit integer, these use mode
8440 In 64-bit mode, the x86-64 family of processors uses additional built-in
8441 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
8442 floating point and @code{TC} 128-bit complex floating point values.
8444 The following floating point built-in functions are available in 64-bit
8445 mode. All of them implement the function that is part of the name.
8448 __float128 __builtin_fabsq (__float128)
8449 __float128 __builtin_copysignq (__float128, __float128)
8452 The following floating point built-in functions are made available in the
8456 @item __float128 __builtin_infq (void)
8457 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
8458 @findex __builtin_infq
8460 @item __float128 __builtin_huge_valq (void)
8461 Similar to @code{__builtin_huge_val}, except the return type is @code{__float128}.
8462 @findex __builtin_huge_valq
8465 The following built-in functions are made available by @option{-mmmx}.
8466 All of them generate the machine instruction that is part of the name.
8469 v8qi __builtin_ia32_paddb (v8qi, v8qi)
8470 v4hi __builtin_ia32_paddw (v4hi, v4hi)
8471 v2si __builtin_ia32_paddd (v2si, v2si)
8472 v8qi __builtin_ia32_psubb (v8qi, v8qi)
8473 v4hi __builtin_ia32_psubw (v4hi, v4hi)
8474 v2si __builtin_ia32_psubd (v2si, v2si)
8475 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
8476 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
8477 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
8478 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
8479 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
8480 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
8481 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
8482 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
8483 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
8484 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
8485 di __builtin_ia32_pand (di, di)
8486 di __builtin_ia32_pandn (di,di)
8487 di __builtin_ia32_por (di, di)
8488 di __builtin_ia32_pxor (di, di)
8489 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
8490 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
8491 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
8492 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
8493 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
8494 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
8495 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
8496 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
8497 v2si __builtin_ia32_punpckhdq (v2si, v2si)
8498 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
8499 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
8500 v2si __builtin_ia32_punpckldq (v2si, v2si)
8501 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
8502 v4hi __builtin_ia32_packssdw (v2si, v2si)
8503 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
8505 v4hi __builtin_ia32_psllw (v4hi, v4hi)
8506 v2si __builtin_ia32_pslld (v2si, v2si)
8507 v1di __builtin_ia32_psllq (v1di, v1di)
8508 v4hi __builtin_ia32_psrlw (v4hi, v4hi)
8509 v2si __builtin_ia32_psrld (v2si, v2si)
8510 v1di __builtin_ia32_psrlq (v1di, v1di)
8511 v4hi __builtin_ia32_psraw (v4hi, v4hi)
8512 v2si __builtin_ia32_psrad (v2si, v2si)
8513 v4hi __builtin_ia32_psllwi (v4hi, int)
8514 v2si __builtin_ia32_pslldi (v2si, int)
8515 v1di __builtin_ia32_psllqi (v1di, int)
8516 v4hi __builtin_ia32_psrlwi (v4hi, int)
8517 v2si __builtin_ia32_psrldi (v2si, int)
8518 v1di __builtin_ia32_psrlqi (v1di, int)
8519 v4hi __builtin_ia32_psrawi (v4hi, int)
8520 v2si __builtin_ia32_psradi (v2si, int)
8524 The following built-in functions are made available either with
8525 @option{-msse}, or with a combination of @option{-m3dnow} and
8526 @option{-march=athlon}. All of them generate the machine
8527 instruction that is part of the name.
8530 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
8531 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
8532 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
8533 v1di __builtin_ia32_psadbw (v8qi, v8qi)
8534 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
8535 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
8536 v8qi __builtin_ia32_pminub (v8qi, v8qi)
8537 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
8538 int __builtin_ia32_pextrw (v4hi, int)
8539 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
8540 int __builtin_ia32_pmovmskb (v8qi)
8541 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
8542 void __builtin_ia32_movntq (di *, di)
8543 void __builtin_ia32_sfence (void)
8546 The following built-in functions are available when @option{-msse} is used.
8547 All of them generate the machine instruction that is part of the name.
8550 int __builtin_ia32_comieq (v4sf, v4sf)
8551 int __builtin_ia32_comineq (v4sf, v4sf)
8552 int __builtin_ia32_comilt (v4sf, v4sf)
8553 int __builtin_ia32_comile (v4sf, v4sf)
8554 int __builtin_ia32_comigt (v4sf, v4sf)
8555 int __builtin_ia32_comige (v4sf, v4sf)
8556 int __builtin_ia32_ucomieq (v4sf, v4sf)
8557 int __builtin_ia32_ucomineq (v4sf, v4sf)
8558 int __builtin_ia32_ucomilt (v4sf, v4sf)
8559 int __builtin_ia32_ucomile (v4sf, v4sf)
8560 int __builtin_ia32_ucomigt (v4sf, v4sf)
8561 int __builtin_ia32_ucomige (v4sf, v4sf)
8562 v4sf __builtin_ia32_addps (v4sf, v4sf)
8563 v4sf __builtin_ia32_subps (v4sf, v4sf)
8564 v4sf __builtin_ia32_mulps (v4sf, v4sf)
8565 v4sf __builtin_ia32_divps (v4sf, v4sf)
8566 v4sf __builtin_ia32_addss (v4sf, v4sf)
8567 v4sf __builtin_ia32_subss (v4sf, v4sf)
8568 v4sf __builtin_ia32_mulss (v4sf, v4sf)
8569 v4sf __builtin_ia32_divss (v4sf, v4sf)
8570 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
8571 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
8572 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
8573 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
8574 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
8575 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
8576 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
8577 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
8578 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
8579 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
8580 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
8581 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
8582 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
8583 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
8584 v4si __builtin_ia32_cmpless (v4sf, v4sf)
8585 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
8586 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
8587 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
8588 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
8589 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
8590 v4sf __builtin_ia32_maxps (v4sf, v4sf)
8591 v4sf __builtin_ia32_maxss (v4sf, v4sf)
8592 v4sf __builtin_ia32_minps (v4sf, v4sf)
8593 v4sf __builtin_ia32_minss (v4sf, v4sf)
8594 v4sf __builtin_ia32_andps (v4sf, v4sf)
8595 v4sf __builtin_ia32_andnps (v4sf, v4sf)
8596 v4sf __builtin_ia32_orps (v4sf, v4sf)
8597 v4sf __builtin_ia32_xorps (v4sf, v4sf)
8598 v4sf __builtin_ia32_movss (v4sf, v4sf)
8599 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
8600 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
8601 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
8602 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
8603 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
8604 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
8605 v2si __builtin_ia32_cvtps2pi (v4sf)
8606 int __builtin_ia32_cvtss2si (v4sf)
8607 v2si __builtin_ia32_cvttps2pi (v4sf)
8608 int __builtin_ia32_cvttss2si (v4sf)
8609 v4sf __builtin_ia32_rcpps (v4sf)
8610 v4sf __builtin_ia32_rsqrtps (v4sf)
8611 v4sf __builtin_ia32_sqrtps (v4sf)
8612 v4sf __builtin_ia32_rcpss (v4sf)
8613 v4sf __builtin_ia32_rsqrtss (v4sf)
8614 v4sf __builtin_ia32_sqrtss (v4sf)
8615 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
8616 void __builtin_ia32_movntps (float *, v4sf)
8617 int __builtin_ia32_movmskps (v4sf)
8620 The following built-in functions are available when @option{-msse} is used.
8623 @item v4sf __builtin_ia32_loadaps (float *)
8624 Generates the @code{movaps} machine instruction as a load from memory.
8625 @item void __builtin_ia32_storeaps (float *, v4sf)
8626 Generates the @code{movaps} machine instruction as a store to memory.
8627 @item v4sf __builtin_ia32_loadups (float *)
8628 Generates the @code{movups} machine instruction as a load from memory.
8629 @item void __builtin_ia32_storeups (float *, v4sf)
8630 Generates the @code{movups} machine instruction as a store to memory.
8631 @item v4sf __builtin_ia32_loadsss (float *)
8632 Generates the @code{movss} machine instruction as a load from memory.
8633 @item void __builtin_ia32_storess (float *, v4sf)
8634 Generates the @code{movss} machine instruction as a store to memory.
8635 @item v4sf __builtin_ia32_loadhps (v4sf, const v2sf *)
8636 Generates the @code{movhps} machine instruction as a load from memory.
8637 @item v4sf __builtin_ia32_loadlps (v4sf, const v2sf *)
8638 Generates the @code{movlps} machine instruction as a load from memory
8639 @item void __builtin_ia32_storehps (v2sf *, v4sf)
8640 Generates the @code{movhps} machine instruction as a store to memory.
8641 @item void __builtin_ia32_storelps (v2sf *, v4sf)
8642 Generates the @code{movlps} machine instruction as a store to memory.
8645 The following built-in functions are available when @option{-msse2} is used.
8646 All of them generate the machine instruction that is part of the name.
8649 int __builtin_ia32_comisdeq (v2df, v2df)
8650 int __builtin_ia32_comisdlt (v2df, v2df)
8651 int __builtin_ia32_comisdle (v2df, v2df)
8652 int __builtin_ia32_comisdgt (v2df, v2df)
8653 int __builtin_ia32_comisdge (v2df, v2df)
8654 int __builtin_ia32_comisdneq (v2df, v2df)
8655 int __builtin_ia32_ucomisdeq (v2df, v2df)
8656 int __builtin_ia32_ucomisdlt (v2df, v2df)
8657 int __builtin_ia32_ucomisdle (v2df, v2df)
8658 int __builtin_ia32_ucomisdgt (v2df, v2df)
8659 int __builtin_ia32_ucomisdge (v2df, v2df)
8660 int __builtin_ia32_ucomisdneq (v2df, v2df)
8661 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
8662 v2df __builtin_ia32_cmpltpd (v2df, v2df)
8663 v2df __builtin_ia32_cmplepd (v2df, v2df)
8664 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
8665 v2df __builtin_ia32_cmpgepd (v2df, v2df)
8666 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
8667 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
8668 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
8669 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
8670 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
8671 v2df __builtin_ia32_cmpngepd (v2df, v2df)
8672 v2df __builtin_ia32_cmpordpd (v2df, v2df)
8673 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
8674 v2df __builtin_ia32_cmpltsd (v2df, v2df)
8675 v2df __builtin_ia32_cmplesd (v2df, v2df)
8676 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
8677 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
8678 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
8679 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
8680 v2df __builtin_ia32_cmpordsd (v2df, v2df)
8681 v2di __builtin_ia32_paddq (v2di, v2di)
8682 v2di __builtin_ia32_psubq (v2di, v2di)
8683 v2df __builtin_ia32_addpd (v2df, v2df)
8684 v2df __builtin_ia32_subpd (v2df, v2df)
8685 v2df __builtin_ia32_mulpd (v2df, v2df)
8686 v2df __builtin_ia32_divpd (v2df, v2df)
8687 v2df __builtin_ia32_addsd (v2df, v2df)
8688 v2df __builtin_ia32_subsd (v2df, v2df)
8689 v2df __builtin_ia32_mulsd (v2df, v2df)
8690 v2df __builtin_ia32_divsd (v2df, v2df)
8691 v2df __builtin_ia32_minpd (v2df, v2df)
8692 v2df __builtin_ia32_maxpd (v2df, v2df)
8693 v2df __builtin_ia32_minsd (v2df, v2df)
8694 v2df __builtin_ia32_maxsd (v2df, v2df)
8695 v2df __builtin_ia32_andpd (v2df, v2df)
8696 v2df __builtin_ia32_andnpd (v2df, v2df)
8697 v2df __builtin_ia32_orpd (v2df, v2df)
8698 v2df __builtin_ia32_xorpd (v2df, v2df)
8699 v2df __builtin_ia32_movsd (v2df, v2df)
8700 v2df __builtin_ia32_unpckhpd (v2df, v2df)
8701 v2df __builtin_ia32_unpcklpd (v2df, v2df)
8702 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
8703 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
8704 v4si __builtin_ia32_paddd128 (v4si, v4si)
8705 v2di __builtin_ia32_paddq128 (v2di, v2di)
8706 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
8707 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
8708 v4si __builtin_ia32_psubd128 (v4si, v4si)
8709 v2di __builtin_ia32_psubq128 (v2di, v2di)
8710 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
8711 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
8712 v2di __builtin_ia32_pand128 (v2di, v2di)
8713 v2di __builtin_ia32_pandn128 (v2di, v2di)
8714 v2di __builtin_ia32_por128 (v2di, v2di)
8715 v2di __builtin_ia32_pxor128 (v2di, v2di)
8716 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
8717 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
8718 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
8719 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
8720 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
8721 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
8722 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
8723 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
8724 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
8725 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
8726 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
8727 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
8728 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
8729 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
8730 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
8731 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
8732 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
8733 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
8734 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
8735 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
8736 v16qi __builtin_ia32_packsswb128 (v8hi, v8hi)
8737 v8hi __builtin_ia32_packssdw128 (v4si, v4si)
8738 v16qi __builtin_ia32_packuswb128 (v8hi, v8hi)
8739 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
8740 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
8741 v2df __builtin_ia32_loadupd (double *)
8742 void __builtin_ia32_storeupd (double *, v2df)
8743 v2df __builtin_ia32_loadhpd (v2df, double const *)
8744 v2df __builtin_ia32_loadlpd (v2df, double const *)
8745 int __builtin_ia32_movmskpd (v2df)
8746 int __builtin_ia32_pmovmskb128 (v16qi)
8747 void __builtin_ia32_movnti (int *, int)
8748 void __builtin_ia32_movntpd (double *, v2df)
8749 void __builtin_ia32_movntdq (v2df *, v2df)
8750 v4si __builtin_ia32_pshufd (v4si, int)
8751 v8hi __builtin_ia32_pshuflw (v8hi, int)
8752 v8hi __builtin_ia32_pshufhw (v8hi, int)
8753 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
8754 v2df __builtin_ia32_sqrtpd (v2df)
8755 v2df __builtin_ia32_sqrtsd (v2df)
8756 v2df __builtin_ia32_shufpd (v2df, v2df, int)
8757 v2df __builtin_ia32_cvtdq2pd (v4si)
8758 v4sf __builtin_ia32_cvtdq2ps (v4si)
8759 v4si __builtin_ia32_cvtpd2dq (v2df)
8760 v2si __builtin_ia32_cvtpd2pi (v2df)
8761 v4sf __builtin_ia32_cvtpd2ps (v2df)
8762 v4si __builtin_ia32_cvttpd2dq (v2df)
8763 v2si __builtin_ia32_cvttpd2pi (v2df)
8764 v2df __builtin_ia32_cvtpi2pd (v2si)
8765 int __builtin_ia32_cvtsd2si (v2df)
8766 int __builtin_ia32_cvttsd2si (v2df)
8767 long long __builtin_ia32_cvtsd2si64 (v2df)
8768 long long __builtin_ia32_cvttsd2si64 (v2df)
8769 v4si __builtin_ia32_cvtps2dq (v4sf)
8770 v2df __builtin_ia32_cvtps2pd (v4sf)
8771 v4si __builtin_ia32_cvttps2dq (v4sf)
8772 v2df __builtin_ia32_cvtsi2sd (v2df, int)
8773 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
8774 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
8775 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
8776 void __builtin_ia32_clflush (const void *)
8777 void __builtin_ia32_lfence (void)
8778 void __builtin_ia32_mfence (void)
8779 v16qi __builtin_ia32_loaddqu (const char *)
8780 void __builtin_ia32_storedqu (char *, v16qi)
8781 v1di __builtin_ia32_pmuludq (v2si, v2si)
8782 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
8783 v8hi __builtin_ia32_psllw128 (v8hi, v8hi)
8784 v4si __builtin_ia32_pslld128 (v4si, v4si)
8785 v2di __builtin_ia32_psllq128 (v2di, v2di)
8786 v8hi __builtin_ia32_psrlw128 (v8hi, v8hi)
8787 v4si __builtin_ia32_psrld128 (v4si, v4si)
8788 v2di __builtin_ia32_psrlq128 (v2di, v2di)
8789 v8hi __builtin_ia32_psraw128 (v8hi, v8hi)
8790 v4si __builtin_ia32_psrad128 (v4si, v4si)
8791 v2di __builtin_ia32_pslldqi128 (v2di, int)
8792 v8hi __builtin_ia32_psllwi128 (v8hi, int)
8793 v4si __builtin_ia32_pslldi128 (v4si, int)
8794 v2di __builtin_ia32_psllqi128 (v2di, int)
8795 v2di __builtin_ia32_psrldqi128 (v2di, int)
8796 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
8797 v4si __builtin_ia32_psrldi128 (v4si, int)
8798 v2di __builtin_ia32_psrlqi128 (v2di, int)
8799 v8hi __builtin_ia32_psrawi128 (v8hi, int)
8800 v4si __builtin_ia32_psradi128 (v4si, int)
8801 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
8802 v2di __builtin_ia32_movq128 (v2di)
8805 The following built-in functions are available when @option{-msse3} is used.
8806 All of them generate the machine instruction that is part of the name.
8809 v2df __builtin_ia32_addsubpd (v2df, v2df)
8810 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
8811 v2df __builtin_ia32_haddpd (v2df, v2df)
8812 v4sf __builtin_ia32_haddps (v4sf, v4sf)
8813 v2df __builtin_ia32_hsubpd (v2df, v2df)
8814 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
8815 v16qi __builtin_ia32_lddqu (char const *)
8816 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
8817 v2df __builtin_ia32_movddup (v2df)
8818 v4sf __builtin_ia32_movshdup (v4sf)
8819 v4sf __builtin_ia32_movsldup (v4sf)
8820 void __builtin_ia32_mwait (unsigned int, unsigned int)
8823 The following built-in functions are available when @option{-msse3} is used.
8826 @item v2df __builtin_ia32_loadddup (double const *)
8827 Generates the @code{movddup} machine instruction as a load from memory.
8830 The following built-in functions are available when @option{-mssse3} is used.
8831 All of them generate the machine instruction that is part of the name
8835 v2si __builtin_ia32_phaddd (v2si, v2si)
8836 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
8837 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
8838 v2si __builtin_ia32_phsubd (v2si, v2si)
8839 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
8840 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
8841 v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi)
8842 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
8843 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
8844 v8qi __builtin_ia32_psignb (v8qi, v8qi)
8845 v2si __builtin_ia32_psignd (v2si, v2si)
8846 v4hi __builtin_ia32_psignw (v4hi, v4hi)
8847 v1di __builtin_ia32_palignr (v1di, v1di, int)
8848 v8qi __builtin_ia32_pabsb (v8qi)
8849 v2si __builtin_ia32_pabsd (v2si)
8850 v4hi __builtin_ia32_pabsw (v4hi)
8853 The following built-in functions are available when @option{-mssse3} is used.
8854 All of them generate the machine instruction that is part of the name
8858 v4si __builtin_ia32_phaddd128 (v4si, v4si)
8859 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
8860 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
8861 v4si __builtin_ia32_phsubd128 (v4si, v4si)
8862 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
8863 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
8864 v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
8865 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
8866 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
8867 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
8868 v4si __builtin_ia32_psignd128 (v4si, v4si)
8869 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
8870 v2di __builtin_ia32_palignr128 (v2di, v2di, int)
8871 v16qi __builtin_ia32_pabsb128 (v16qi)
8872 v4si __builtin_ia32_pabsd128 (v4si)
8873 v8hi __builtin_ia32_pabsw128 (v8hi)
8876 The following built-in functions are available when @option{-msse4.1} is
8877 used. All of them generate the machine instruction that is part of the
8881 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
8882 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
8883 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
8884 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
8885 v2df __builtin_ia32_dppd (v2df, v2df, const int)
8886 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
8887 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
8888 v2di __builtin_ia32_movntdqa (v2di *);
8889 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
8890 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
8891 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
8892 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
8893 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
8894 v8hi __builtin_ia32_phminposuw128 (v8hi)
8895 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
8896 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
8897 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
8898 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
8899 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
8900 v4si __builtin_ia32_pminsd128 (v4si, v4si)
8901 v4si __builtin_ia32_pminud128 (v4si, v4si)
8902 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
8903 v4si __builtin_ia32_pmovsxbd128 (v16qi)
8904 v2di __builtin_ia32_pmovsxbq128 (v16qi)
8905 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
8906 v2di __builtin_ia32_pmovsxdq128 (v4si)
8907 v4si __builtin_ia32_pmovsxwd128 (v8hi)
8908 v2di __builtin_ia32_pmovsxwq128 (v8hi)
8909 v4si __builtin_ia32_pmovzxbd128 (v16qi)
8910 v2di __builtin_ia32_pmovzxbq128 (v16qi)
8911 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
8912 v2di __builtin_ia32_pmovzxdq128 (v4si)
8913 v4si __builtin_ia32_pmovzxwd128 (v8hi)
8914 v2di __builtin_ia32_pmovzxwq128 (v8hi)
8915 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
8916 v4si __builtin_ia32_pmulld128 (v4si, v4si)
8917 int __builtin_ia32_ptestc128 (v2di, v2di)
8918 int __builtin_ia32_ptestnzc128 (v2di, v2di)
8919 int __builtin_ia32_ptestz128 (v2di, v2di)
8920 v2df __builtin_ia32_roundpd (v2df, const int)
8921 v4sf __builtin_ia32_roundps (v4sf, const int)
8922 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
8923 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
8926 The following built-in functions are available when @option{-msse4.1} is
8930 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
8931 Generates the @code{insertps} machine instruction.
8932 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
8933 Generates the @code{pextrb} machine instruction.
8934 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
8935 Generates the @code{pinsrb} machine instruction.
8936 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
8937 Generates the @code{pinsrd} machine instruction.
8938 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
8939 Generates the @code{pinsrq} machine instruction in 64bit mode.
8942 The following built-in functions are changed to generate new SSE4.1
8943 instructions when @option{-msse4.1} is used.
8946 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
8947 Generates the @code{extractps} machine instruction.
8948 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
8949 Generates the @code{pextrd} machine instruction.
8950 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
8951 Generates the @code{pextrq} machine instruction in 64bit mode.
8954 The following built-in functions are available when @option{-msse4.2} is
8955 used. All of them generate the machine instruction that is part of the
8959 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
8960 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
8961 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
8962 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
8963 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
8964 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
8965 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
8966 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
8967 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
8968 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
8969 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
8970 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
8971 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
8972 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
8973 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
8976 The following built-in functions are available when @option{-msse4.2} is
8980 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
8981 Generates the @code{crc32b} machine instruction.
8982 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
8983 Generates the @code{crc32w} machine instruction.
8984 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
8985 Generates the @code{crc32l} machine instruction.
8986 @item unsigned long long __builtin_ia32_crc32di (unsigned long long, unsigned long long)
8987 Generates the @code{crc32q} machine instruction.
8990 The following built-in functions are changed to generate new SSE4.2
8991 instructions when @option{-msse4.2} is used.
8994 @item int __builtin_popcount (unsigned int)
8995 Generates the @code{popcntl} machine instruction.
8996 @item int __builtin_popcountl (unsigned long)
8997 Generates the @code{popcntl} or @code{popcntq} machine instruction,
8998 depending on the size of @code{unsigned long}.
8999 @item int __builtin_popcountll (unsigned long long)
9000 Generates the @code{popcntq} machine instruction.
9003 The following built-in functions are available when @option{-mavx} is
9004 used. All of them generate the machine instruction that is part of the
9008 v4df __builtin_ia32_addpd256 (v4df,v4df)
9009 v8sf __builtin_ia32_addps256 (v8sf,v8sf)
9010 v4df __builtin_ia32_addsubpd256 (v4df,v4df)
9011 v8sf __builtin_ia32_addsubps256 (v8sf,v8sf)
9012 v4df __builtin_ia32_andnpd256 (v4df,v4df)
9013 v8sf __builtin_ia32_andnps256 (v8sf,v8sf)
9014 v4df __builtin_ia32_andpd256 (v4df,v4df)
9015 v8sf __builtin_ia32_andps256 (v8sf,v8sf)
9016 v4df __builtin_ia32_blendpd256 (v4df,v4df,int)
9017 v8sf __builtin_ia32_blendps256 (v8sf,v8sf,int)
9018 v4df __builtin_ia32_blendvpd256 (v4df,v4df,v4df)
9019 v8sf __builtin_ia32_blendvps256 (v8sf,v8sf,v8sf)
9020 v2df __builtin_ia32_cmppd (v2df,v2df,int)
9021 v4df __builtin_ia32_cmppd256 (v4df,v4df,int)
9022 v4sf __builtin_ia32_cmpps (v4sf,v4sf,int)
9023 v8sf __builtin_ia32_cmpps256 (v8sf,v8sf,int)
9024 v2df __builtin_ia32_cmpsd (v2df,v2df,int)
9025 v4sf __builtin_ia32_cmpss (v4sf,v4sf,int)
9026 v4df __builtin_ia32_cvtdq2pd256 (v4si)
9027 v8sf __builtin_ia32_cvtdq2ps256 (v8si)
9028 v4si __builtin_ia32_cvtpd2dq256 (v4df)
9029 v4sf __builtin_ia32_cvtpd2ps256 (v4df)
9030 v8si __builtin_ia32_cvtps2dq256 (v8sf)
9031 v4df __builtin_ia32_cvtps2pd256 (v4sf)
9032 v4si __builtin_ia32_cvttpd2dq256 (v4df)
9033 v8si __builtin_ia32_cvttps2dq256 (v8sf)
9034 v4df __builtin_ia32_divpd256 (v4df,v4df)
9035 v8sf __builtin_ia32_divps256 (v8sf,v8sf)
9036 v8sf __builtin_ia32_dpps256 (v8sf,v8sf,int)
9037 v4df __builtin_ia32_haddpd256 (v4df,v4df)
9038 v8sf __builtin_ia32_haddps256 (v8sf,v8sf)
9039 v4df __builtin_ia32_hsubpd256 (v4df,v4df)
9040 v8sf __builtin_ia32_hsubps256 (v8sf,v8sf)
9041 v32qi __builtin_ia32_lddqu256 (pcchar)
9042 v32qi __builtin_ia32_loaddqu256 (pcchar)
9043 v4df __builtin_ia32_loadupd256 (pcdouble)
9044 v8sf __builtin_ia32_loadups256 (pcfloat)
9045 v2df __builtin_ia32_maskloadpd (pcv2df,v2df)
9046 v4df __builtin_ia32_maskloadpd256 (pcv4df,v4df)
9047 v4sf __builtin_ia32_maskloadps (pcv4sf,v4sf)
9048 v8sf __builtin_ia32_maskloadps256 (pcv8sf,v8sf)
9049 void __builtin_ia32_maskstorepd (pv2df,v2df,v2df)
9050 void __builtin_ia32_maskstorepd256 (pv4df,v4df,v4df)
9051 void __builtin_ia32_maskstoreps (pv4sf,v4sf,v4sf)
9052 void __builtin_ia32_maskstoreps256 (pv8sf,v8sf,v8sf)
9053 v4df __builtin_ia32_maxpd256 (v4df,v4df)
9054 v8sf __builtin_ia32_maxps256 (v8sf,v8sf)
9055 v4df __builtin_ia32_minpd256 (v4df,v4df)
9056 v8sf __builtin_ia32_minps256 (v8sf,v8sf)
9057 v4df __builtin_ia32_movddup256 (v4df)
9058 int __builtin_ia32_movmskpd256 (v4df)
9059 int __builtin_ia32_movmskps256 (v8sf)
9060 v8sf __builtin_ia32_movshdup256 (v8sf)
9061 v8sf __builtin_ia32_movsldup256 (v8sf)
9062 v4df __builtin_ia32_mulpd256 (v4df,v4df)
9063 v8sf __builtin_ia32_mulps256 (v8sf,v8sf)
9064 v4df __builtin_ia32_orpd256 (v4df,v4df)
9065 v8sf __builtin_ia32_orps256 (v8sf,v8sf)
9066 v2df __builtin_ia32_pd_pd256 (v4df)
9067 v4df __builtin_ia32_pd256_pd (v2df)
9068 v4sf __builtin_ia32_ps_ps256 (v8sf)
9069 v8sf __builtin_ia32_ps256_ps (v4sf)
9070 int __builtin_ia32_ptestc256 (v4di,v4di,ptest)
9071 int __builtin_ia32_ptestnzc256 (v4di,v4di,ptest)
9072 int __builtin_ia32_ptestz256 (v4di,v4di,ptest)
9073 v8sf __builtin_ia32_rcpps256 (v8sf)
9074 v4df __builtin_ia32_roundpd256 (v4df,int)
9075 v8sf __builtin_ia32_roundps256 (v8sf,int)
9076 v8sf __builtin_ia32_rsqrtps_nr256 (v8sf)
9077 v8sf __builtin_ia32_rsqrtps256 (v8sf)
9078 v4df __builtin_ia32_shufpd256 (v4df,v4df,int)
9079 v8sf __builtin_ia32_shufps256 (v8sf,v8sf,int)
9080 v4si __builtin_ia32_si_si256 (v8si)
9081 v8si __builtin_ia32_si256_si (v4si)
9082 v4df __builtin_ia32_sqrtpd256 (v4df)
9083 v8sf __builtin_ia32_sqrtps_nr256 (v8sf)
9084 v8sf __builtin_ia32_sqrtps256 (v8sf)
9085 void __builtin_ia32_storedqu256 (pchar,v32qi)
9086 void __builtin_ia32_storeupd256 (pdouble,v4df)
9087 void __builtin_ia32_storeups256 (pfloat,v8sf)
9088 v4df __builtin_ia32_subpd256 (v4df,v4df)
9089 v8sf __builtin_ia32_subps256 (v8sf,v8sf)
9090 v4df __builtin_ia32_unpckhpd256 (v4df,v4df)
9091 v8sf __builtin_ia32_unpckhps256 (v8sf,v8sf)
9092 v4df __builtin_ia32_unpcklpd256 (v4df,v4df)
9093 v8sf __builtin_ia32_unpcklps256 (v8sf,v8sf)
9094 v4df __builtin_ia32_vbroadcastf128_pd256 (pcv2df)
9095 v8sf __builtin_ia32_vbroadcastf128_ps256 (pcv4sf)
9096 v4df __builtin_ia32_vbroadcastsd256 (pcdouble)
9097 v4sf __builtin_ia32_vbroadcastss (pcfloat)
9098 v8sf __builtin_ia32_vbroadcastss256 (pcfloat)
9099 v2df __builtin_ia32_vextractf128_pd256 (v4df,int)
9100 v4sf __builtin_ia32_vextractf128_ps256 (v8sf,int)
9101 v4si __builtin_ia32_vextractf128_si256 (v8si,int)
9102 v4df __builtin_ia32_vinsertf128_pd256 (v4df,v2df,int)
9103 v8sf __builtin_ia32_vinsertf128_ps256 (v8sf,v4sf,int)
9104 v8si __builtin_ia32_vinsertf128_si256 (v8si,v4si,int)
9105 v4df __builtin_ia32_vperm2f128_pd256 (v4df,v4df,int)
9106 v8sf __builtin_ia32_vperm2f128_ps256 (v8sf,v8sf,int)
9107 v8si __builtin_ia32_vperm2f128_si256 (v8si,v8si,int)
9108 v2df __builtin_ia32_vpermil2pd (v2df,v2df,v2di,int)
9109 v4df __builtin_ia32_vpermil2pd256 (v4df,v4df,v4di,int)
9110 v4sf __builtin_ia32_vpermil2ps (v4sf,v4sf,v4si,int)
9111 v8sf __builtin_ia32_vpermil2ps256 (v8sf,v8sf,v8si,int)
9112 v2df __builtin_ia32_vpermilpd (v2df,int)
9113 v4df __builtin_ia32_vpermilpd256 (v4df,int)
9114 v4sf __builtin_ia32_vpermilps (v4sf,int)
9115 v8sf __builtin_ia32_vpermilps256 (v8sf,int)
9116 v2df __builtin_ia32_vpermilvarpd (v2df,v2di)
9117 v4df __builtin_ia32_vpermilvarpd256 (v4df,v4di)
9118 v4sf __builtin_ia32_vpermilvarps (v4sf,v4si)
9119 v8sf __builtin_ia32_vpermilvarps256 (v8sf,v8si)
9120 int __builtin_ia32_vtestcpd (v2df,v2df,ptest)
9121 int __builtin_ia32_vtestcpd256 (v4df,v4df,ptest)
9122 int __builtin_ia32_vtestcps (v4sf,v4sf,ptest)
9123 int __builtin_ia32_vtestcps256 (v8sf,v8sf,ptest)
9124 int __builtin_ia32_vtestnzcpd (v2df,v2df,ptest)
9125 int __builtin_ia32_vtestnzcpd256 (v4df,v4df,ptest)
9126 int __builtin_ia32_vtestnzcps (v4sf,v4sf,ptest)
9127 int __builtin_ia32_vtestnzcps256 (v8sf,v8sf,ptest)
9128 int __builtin_ia32_vtestzpd (v2df,v2df,ptest)
9129 int __builtin_ia32_vtestzpd256 (v4df,v4df,ptest)
9130 int __builtin_ia32_vtestzps (v4sf,v4sf,ptest)
9131 int __builtin_ia32_vtestzps256 (v8sf,v8sf,ptest)
9132 void __builtin_ia32_vzeroall (void)
9133 void __builtin_ia32_vzeroupper (void)
9134 v4df __builtin_ia32_xorpd256 (v4df,v4df)
9135 v8sf __builtin_ia32_xorps256 (v8sf,v8sf)
9138 The following built-in functions are available when @option{-maes} is
9139 used. All of them generate the machine instruction that is part of the
9143 v2di __builtin_ia32_aesenc128 (v2di, v2di)
9144 v2di __builtin_ia32_aesenclast128 (v2di, v2di)
9145 v2di __builtin_ia32_aesdec128 (v2di, v2di)
9146 v2di __builtin_ia32_aesdeclast128 (v2di, v2di)
9147 v2di __builtin_ia32_aeskeygenassist128 (v2di, const int)
9148 v2di __builtin_ia32_aesimc128 (v2di)
9151 The following built-in function is available when @option{-mpclmul} is
9155 @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int)
9156 Generates the @code{pclmulqdq} machine instruction.
9159 The following built-in function is available when @option{-mfsgsbase} is
9160 used. All of them generate the machine instruction that is part of the
9164 unsigned int __builtin_ia32_rdfsbase32 (void)
9165 unsigned long long __builtin_ia32_rdfsbase64 (void)
9166 unsigned int __builtin_ia32_rdgsbase32 (void)
9167 unsigned long long __builtin_ia32_rdgsbase64 (void)
9168 void _writefsbase_u32 (unsigned int)
9169 void _writefsbase_u64 (unsigned long long)
9170 void _writegsbase_u32 (unsigned int)
9171 void _writegsbase_u64 (unsigned long long)
9174 The following built-in function is available when @option{-mrdrnd} is
9175 used. All of them generate the machine instruction that is part of the
9179 unsigned short __builtin_ia32_rdrand16 (void)
9180 unsigned int __builtin_ia32_rdrand32 (void)
9181 unsigned long long __builtin_ia32_rdrand64 (void)
9184 The following built-in functions are available when @option{-msse4a} is used.
9185 All of them generate the machine instruction that is part of the name.
9188 void __builtin_ia32_movntsd (double *, v2df)
9189 void __builtin_ia32_movntss (float *, v4sf)
9190 v2di __builtin_ia32_extrq (v2di, v16qi)
9191 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
9192 v2di __builtin_ia32_insertq (v2di, v2di)
9193 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
9196 The following built-in functions are available when @option{-mxop} is used.
9198 v2df __builtin_ia32_vfrczpd (v2df)
9199 v4sf __builtin_ia32_vfrczps (v4sf)
9200 v2df __builtin_ia32_vfrczsd (v2df, v2df)
9201 v4sf __builtin_ia32_vfrczss (v4sf, v4sf)
9202 v4df __builtin_ia32_vfrczpd256 (v4df)
9203 v8sf __builtin_ia32_vfrczps256 (v8sf)
9204 v2di __builtin_ia32_vpcmov (v2di, v2di, v2di)
9205 v2di __builtin_ia32_vpcmov_v2di (v2di, v2di, v2di)
9206 v4si __builtin_ia32_vpcmov_v4si (v4si, v4si, v4si)
9207 v8hi __builtin_ia32_vpcmov_v8hi (v8hi, v8hi, v8hi)
9208 v16qi __builtin_ia32_vpcmov_v16qi (v16qi, v16qi, v16qi)
9209 v2df __builtin_ia32_vpcmov_v2df (v2df, v2df, v2df)
9210 v4sf __builtin_ia32_vpcmov_v4sf (v4sf, v4sf, v4sf)
9211 v4di __builtin_ia32_vpcmov_v4di256 (v4di, v4di, v4di)
9212 v8si __builtin_ia32_vpcmov_v8si256 (v8si, v8si, v8si)
9213 v16hi __builtin_ia32_vpcmov_v16hi256 (v16hi, v16hi, v16hi)
9214 v32qi __builtin_ia32_vpcmov_v32qi256 (v32qi, v32qi, v32qi)
9215 v4df __builtin_ia32_vpcmov_v4df256 (v4df, v4df, v4df)
9216 v8sf __builtin_ia32_vpcmov_v8sf256 (v8sf, v8sf, v8sf)
9217 v16qi __builtin_ia32_vpcomeqb (v16qi, v16qi)
9218 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
9219 v4si __builtin_ia32_vpcomeqd (v4si, v4si)
9220 v2di __builtin_ia32_vpcomeqq (v2di, v2di)
9221 v16qi __builtin_ia32_vpcomequb (v16qi, v16qi)
9222 v4si __builtin_ia32_vpcomequd (v4si, v4si)
9223 v2di __builtin_ia32_vpcomequq (v2di, v2di)
9224 v8hi __builtin_ia32_vpcomequw (v8hi, v8hi)
9225 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
9226 v16qi __builtin_ia32_vpcomfalseb (v16qi, v16qi)
9227 v4si __builtin_ia32_vpcomfalsed (v4si, v4si)
9228 v2di __builtin_ia32_vpcomfalseq (v2di, v2di)
9229 v16qi __builtin_ia32_vpcomfalseub (v16qi, v16qi)
9230 v4si __builtin_ia32_vpcomfalseud (v4si, v4si)
9231 v2di __builtin_ia32_vpcomfalseuq (v2di, v2di)
9232 v8hi __builtin_ia32_vpcomfalseuw (v8hi, v8hi)
9233 v8hi __builtin_ia32_vpcomfalsew (v8hi, v8hi)
9234 v16qi __builtin_ia32_vpcomgeb (v16qi, v16qi)
9235 v4si __builtin_ia32_vpcomged (v4si, v4si)
9236 v2di __builtin_ia32_vpcomgeq (v2di, v2di)
9237 v16qi __builtin_ia32_vpcomgeub (v16qi, v16qi)
9238 v4si __builtin_ia32_vpcomgeud (v4si, v4si)
9239 v2di __builtin_ia32_vpcomgeuq (v2di, v2di)
9240 v8hi __builtin_ia32_vpcomgeuw (v8hi, v8hi)
9241 v8hi __builtin_ia32_vpcomgew (v8hi, v8hi)
9242 v16qi __builtin_ia32_vpcomgtb (v16qi, v16qi)
9243 v4si __builtin_ia32_vpcomgtd (v4si, v4si)
9244 v2di __builtin_ia32_vpcomgtq (v2di, v2di)
9245 v16qi __builtin_ia32_vpcomgtub (v16qi, v16qi)
9246 v4si __builtin_ia32_vpcomgtud (v4si, v4si)
9247 v2di __builtin_ia32_vpcomgtuq (v2di, v2di)
9248 v8hi __builtin_ia32_vpcomgtuw (v8hi, v8hi)
9249 v8hi __builtin_ia32_vpcomgtw (v8hi, v8hi)
9250 v16qi __builtin_ia32_vpcomleb (v16qi, v16qi)
9251 v4si __builtin_ia32_vpcomled (v4si, v4si)
9252 v2di __builtin_ia32_vpcomleq (v2di, v2di)
9253 v16qi __builtin_ia32_vpcomleub (v16qi, v16qi)
9254 v4si __builtin_ia32_vpcomleud (v4si, v4si)
9255 v2di __builtin_ia32_vpcomleuq (v2di, v2di)
9256 v8hi __builtin_ia32_vpcomleuw (v8hi, v8hi)
9257 v8hi __builtin_ia32_vpcomlew (v8hi, v8hi)
9258 v16qi __builtin_ia32_vpcomltb (v16qi, v16qi)
9259 v4si __builtin_ia32_vpcomltd (v4si, v4si)
9260 v2di __builtin_ia32_vpcomltq (v2di, v2di)
9261 v16qi __builtin_ia32_vpcomltub (v16qi, v16qi)
9262 v4si __builtin_ia32_vpcomltud (v4si, v4si)
9263 v2di __builtin_ia32_vpcomltuq (v2di, v2di)
9264 v8hi __builtin_ia32_vpcomltuw (v8hi, v8hi)
9265 v8hi __builtin_ia32_vpcomltw (v8hi, v8hi)
9266 v16qi __builtin_ia32_vpcomneb (v16qi, v16qi)
9267 v4si __builtin_ia32_vpcomned (v4si, v4si)
9268 v2di __builtin_ia32_vpcomneq (v2di, v2di)
9269 v16qi __builtin_ia32_vpcomneub (v16qi, v16qi)
9270 v4si __builtin_ia32_vpcomneud (v4si, v4si)
9271 v2di __builtin_ia32_vpcomneuq (v2di, v2di)
9272 v8hi __builtin_ia32_vpcomneuw (v8hi, v8hi)
9273 v8hi __builtin_ia32_vpcomnew (v8hi, v8hi)
9274 v16qi __builtin_ia32_vpcomtrueb (v16qi, v16qi)
9275 v4si __builtin_ia32_vpcomtrued (v4si, v4si)
9276 v2di __builtin_ia32_vpcomtrueq (v2di, v2di)
9277 v16qi __builtin_ia32_vpcomtrueub (v16qi, v16qi)
9278 v4si __builtin_ia32_vpcomtrueud (v4si, v4si)
9279 v2di __builtin_ia32_vpcomtrueuq (v2di, v2di)
9280 v8hi __builtin_ia32_vpcomtrueuw (v8hi, v8hi)
9281 v8hi __builtin_ia32_vpcomtruew (v8hi, v8hi)
9282 v4si __builtin_ia32_vphaddbd (v16qi)
9283 v2di __builtin_ia32_vphaddbq (v16qi)
9284 v8hi __builtin_ia32_vphaddbw (v16qi)
9285 v2di __builtin_ia32_vphadddq (v4si)
9286 v4si __builtin_ia32_vphaddubd (v16qi)
9287 v2di __builtin_ia32_vphaddubq (v16qi)
9288 v8hi __builtin_ia32_vphaddubw (v16qi)
9289 v2di __builtin_ia32_vphaddudq (v4si)
9290 v4si __builtin_ia32_vphadduwd (v8hi)
9291 v2di __builtin_ia32_vphadduwq (v8hi)
9292 v4si __builtin_ia32_vphaddwd (v8hi)
9293 v2di __builtin_ia32_vphaddwq (v8hi)
9294 v8hi __builtin_ia32_vphsubbw (v16qi)
9295 v2di __builtin_ia32_vphsubdq (v4si)
9296 v4si __builtin_ia32_vphsubwd (v8hi)
9297 v4si __builtin_ia32_vpmacsdd (v4si, v4si, v4si)
9298 v2di __builtin_ia32_vpmacsdqh (v4si, v4si, v2di)
9299 v2di __builtin_ia32_vpmacsdql (v4si, v4si, v2di)
9300 v4si __builtin_ia32_vpmacssdd (v4si, v4si, v4si)
9301 v2di __builtin_ia32_vpmacssdqh (v4si, v4si, v2di)
9302 v2di __builtin_ia32_vpmacssdql (v4si, v4si, v2di)
9303 v4si __builtin_ia32_vpmacsswd (v8hi, v8hi, v4si)
9304 v8hi __builtin_ia32_vpmacssww (v8hi, v8hi, v8hi)
9305 v4si __builtin_ia32_vpmacswd (v8hi, v8hi, v4si)
9306 v8hi __builtin_ia32_vpmacsww (v8hi, v8hi, v8hi)
9307 v4si __builtin_ia32_vpmadcsswd (v8hi, v8hi, v4si)
9308 v4si __builtin_ia32_vpmadcswd (v8hi, v8hi, v4si)
9309 v16qi __builtin_ia32_vpperm (v16qi, v16qi, v16qi)
9310 v16qi __builtin_ia32_vprotb (v16qi, v16qi)
9311 v4si __builtin_ia32_vprotd (v4si, v4si)
9312 v2di __builtin_ia32_vprotq (v2di, v2di)
9313 v8hi __builtin_ia32_vprotw (v8hi, v8hi)
9314 v16qi __builtin_ia32_vpshab (v16qi, v16qi)
9315 v4si __builtin_ia32_vpshad (v4si, v4si)
9316 v2di __builtin_ia32_vpshaq (v2di, v2di)
9317 v8hi __builtin_ia32_vpshaw (v8hi, v8hi)
9318 v16qi __builtin_ia32_vpshlb (v16qi, v16qi)
9319 v4si __builtin_ia32_vpshld (v4si, v4si)
9320 v2di __builtin_ia32_vpshlq (v2di, v2di)
9321 v8hi __builtin_ia32_vpshlw (v8hi, v8hi)
9324 The following built-in functions are available when @option{-mfma4} is used.
9325 All of them generate the machine instruction that is part of the name
9329 v2df __builtin_ia32_fmaddpd (v2df, v2df, v2df)
9330 v4sf __builtin_ia32_fmaddps (v4sf, v4sf, v4sf)
9331 v2df __builtin_ia32_fmaddsd (v2df, v2df, v2df)
9332 v4sf __builtin_ia32_fmaddss (v4sf, v4sf, v4sf)
9333 v2df __builtin_ia32_fmsubpd (v2df, v2df, v2df)
9334 v4sf __builtin_ia32_fmsubps (v4sf, v4sf, v4sf)
9335 v2df __builtin_ia32_fmsubsd (v2df, v2df, v2df)
9336 v4sf __builtin_ia32_fmsubss (v4sf, v4sf, v4sf)
9337 v2df __builtin_ia32_fnmaddpd (v2df, v2df, v2df)
9338 v4sf __builtin_ia32_fnmaddps (v4sf, v4sf, v4sf)
9339 v2df __builtin_ia32_fnmaddsd (v2df, v2df, v2df)
9340 v4sf __builtin_ia32_fnmaddss (v4sf, v4sf, v4sf)
9341 v2df __builtin_ia32_fnmsubpd (v2df, v2df, v2df)
9342 v4sf __builtin_ia32_fnmsubps (v4sf, v4sf, v4sf)
9343 v2df __builtin_ia32_fnmsubsd (v2df, v2df, v2df)
9344 v4sf __builtin_ia32_fnmsubss (v4sf, v4sf, v4sf)
9345 v2df __builtin_ia32_fmaddsubpd (v2df, v2df, v2df)
9346 v4sf __builtin_ia32_fmaddsubps (v4sf, v4sf, v4sf)
9347 v2df __builtin_ia32_fmsubaddpd (v2df, v2df, v2df)
9348 v4sf __builtin_ia32_fmsubaddps (v4sf, v4sf, v4sf)
9349 v4df __builtin_ia32_fmaddpd256 (v4df, v4df, v4df)
9350 v8sf __builtin_ia32_fmaddps256 (v8sf, v8sf, v8sf)
9351 v4df __builtin_ia32_fmsubpd256 (v4df, v4df, v4df)
9352 v8sf __builtin_ia32_fmsubps256 (v8sf, v8sf, v8sf)
9353 v4df __builtin_ia32_fnmaddpd256 (v4df, v4df, v4df)
9354 v8sf __builtin_ia32_fnmaddps256 (v8sf, v8sf, v8sf)
9355 v4df __builtin_ia32_fnmsubpd256 (v4df, v4df, v4df)
9356 v8sf __builtin_ia32_fnmsubps256 (v8sf, v8sf, v8sf)
9357 v4df __builtin_ia32_fmaddsubpd256 (v4df, v4df, v4df)
9358 v8sf __builtin_ia32_fmaddsubps256 (v8sf, v8sf, v8sf)
9359 v4df __builtin_ia32_fmsubaddpd256 (v4df, v4df, v4df)
9360 v8sf __builtin_ia32_fmsubaddps256 (v8sf, v8sf, v8sf)
9364 The following built-in functions are available when @option{-mlwp} is used.
9367 void __builtin_ia32_llwpcb16 (void *);
9368 void __builtin_ia32_llwpcb32 (void *);
9369 void __builtin_ia32_llwpcb64 (void *);
9370 void * __builtin_ia32_llwpcb16 (void);
9371 void * __builtin_ia32_llwpcb32 (void);
9372 void * __builtin_ia32_llwpcb64 (void);
9373 void __builtin_ia32_lwpval16 (unsigned short, unsigned int, unsigned short)
9374 void __builtin_ia32_lwpval32 (unsigned int, unsigned int, unsigned int)
9375 void __builtin_ia32_lwpval64 (unsigned __int64, unsigned int, unsigned int)
9376 unsigned char __builtin_ia32_lwpins16 (unsigned short, unsigned int, unsigned short)
9377 unsigned char __builtin_ia32_lwpins32 (unsigned int, unsigned int, unsigned int)
9378 unsigned char __builtin_ia32_lwpins64 (unsigned __int64, unsigned int, unsigned int)
9381 The following built-in functions are available when @option{-m3dnow} is used.
9382 All of them generate the machine instruction that is part of the name.
9385 void __builtin_ia32_femms (void)
9386 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
9387 v2si __builtin_ia32_pf2id (v2sf)
9388 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
9389 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
9390 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
9391 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
9392 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
9393 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
9394 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
9395 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
9396 v2sf __builtin_ia32_pfrcp (v2sf)
9397 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
9398 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
9399 v2sf __builtin_ia32_pfrsqrt (v2sf)
9400 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
9401 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
9402 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
9403 v2sf __builtin_ia32_pi2fd (v2si)
9404 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
9407 The following built-in functions are available when both @option{-m3dnow}
9408 and @option{-march=athlon} are used. All of them generate the machine
9409 instruction that is part of the name.
9412 v2si __builtin_ia32_pf2iw (v2sf)
9413 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
9414 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
9415 v2sf __builtin_ia32_pi2fw (v2si)
9416 v2sf __builtin_ia32_pswapdsf (v2sf)
9417 v2si __builtin_ia32_pswapdsi (v2si)
9420 @node MIPS DSP Built-in Functions
9421 @subsection MIPS DSP Built-in Functions
9423 The MIPS DSP Application-Specific Extension (ASE) includes new
9424 instructions that are designed to improve the performance of DSP and
9425 media applications. It provides instructions that operate on packed
9426 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
9428 GCC supports MIPS DSP operations using both the generic
9429 vector extensions (@pxref{Vector Extensions}) and a collection of
9430 MIPS-specific built-in functions. Both kinds of support are
9431 enabled by the @option{-mdsp} command-line option.
9433 Revision 2 of the ASE was introduced in the second half of 2006.
9434 This revision adds extra instructions to the original ASE, but is
9435 otherwise backwards-compatible with it. You can select revision 2
9436 using the command-line option @option{-mdspr2}; this option implies
9439 The SCOUNT and POS bits of the DSP control register are global. The
9440 WRDSP, EXTPDP, EXTPDPV and MTHLIP instructions modify the SCOUNT and
9441 POS bits. During optimization, the compiler will not delete these
9442 instructions and it will not delete calls to functions containing
9445 At present, GCC only provides support for operations on 32-bit
9446 vectors. The vector type associated with 8-bit integer data is
9447 usually called @code{v4i8}, the vector type associated with Q7
9448 is usually called @code{v4q7}, the vector type associated with 16-bit
9449 integer data is usually called @code{v2i16}, and the vector type
9450 associated with Q15 is usually called @code{v2q15}. They can be
9451 defined in C as follows:
9454 typedef signed char v4i8 __attribute__ ((vector_size(4)));
9455 typedef signed char v4q7 __attribute__ ((vector_size(4)));
9456 typedef short v2i16 __attribute__ ((vector_size(4)));
9457 typedef short v2q15 __attribute__ ((vector_size(4)));
9460 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
9461 initialized in the same way as aggregates. For example:
9464 v4i8 a = @{1, 2, 3, 4@};
9466 b = (v4i8) @{5, 6, 7, 8@};
9468 v2q15 c = @{0x0fcb, 0x3a75@};
9470 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
9473 @emph{Note:} The CPU's endianness determines the order in which values
9474 are packed. On little-endian targets, the first value is the least
9475 significant and the last value is the most significant. The opposite
9476 order applies to big-endian targets. For example, the code above will
9477 set the lowest byte of @code{a} to @code{1} on little-endian targets
9478 and @code{4} on big-endian targets.
9480 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
9481 representation. As shown in this example, the integer representation
9482 of a Q7 value can be obtained by multiplying the fractional value by
9483 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
9484 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
9487 The table below lists the @code{v4i8} and @code{v2q15} operations for which
9488 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
9489 and @code{c} and @code{d} are @code{v2q15} values.
9491 @multitable @columnfractions .50 .50
9492 @item C code @tab MIPS instruction
9493 @item @code{a + b} @tab @code{addu.qb}
9494 @item @code{c + d} @tab @code{addq.ph}
9495 @item @code{a - b} @tab @code{subu.qb}
9496 @item @code{c - d} @tab @code{subq.ph}
9499 The table below lists the @code{v2i16} operation for which
9500 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
9501 @code{v2i16} values.
9503 @multitable @columnfractions .50 .50
9504 @item C code @tab MIPS instruction
9505 @item @code{e * f} @tab @code{mul.ph}
9508 It is easier to describe the DSP built-in functions if we first define
9509 the following types:
9514 typedef unsigned int ui32;
9515 typedef long long a64;
9518 @code{q31} and @code{i32} are actually the same as @code{int}, but we
9519 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
9520 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
9521 @code{long long}, but we use @code{a64} to indicate values that will
9522 be placed in one of the four DSP accumulators (@code{$ac0},
9523 @code{$ac1}, @code{$ac2} or @code{$ac3}).
9525 Also, some built-in functions prefer or require immediate numbers as
9526 parameters, because the corresponding DSP instructions accept both immediate
9527 numbers and register operands, or accept immediate numbers only. The
9528 immediate parameters are listed as follows.
9537 imm_n32_31: -32 to 31.
9538 imm_n512_511: -512 to 511.
9541 The following built-in functions map directly to a particular MIPS DSP
9542 instruction. Please refer to the architecture specification
9543 for details on what each instruction does.
9546 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
9547 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
9548 q31 __builtin_mips_addq_s_w (q31, q31)
9549 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
9550 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
9551 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
9552 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
9553 q31 __builtin_mips_subq_s_w (q31, q31)
9554 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
9555 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
9556 i32 __builtin_mips_addsc (i32, i32)
9557 i32 __builtin_mips_addwc (i32, i32)
9558 i32 __builtin_mips_modsub (i32, i32)
9559 i32 __builtin_mips_raddu_w_qb (v4i8)
9560 v2q15 __builtin_mips_absq_s_ph (v2q15)
9561 q31 __builtin_mips_absq_s_w (q31)
9562 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
9563 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
9564 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
9565 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
9566 q31 __builtin_mips_preceq_w_phl (v2q15)
9567 q31 __builtin_mips_preceq_w_phr (v2q15)
9568 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
9569 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
9570 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
9571 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
9572 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
9573 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
9574 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
9575 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
9576 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
9577 v4i8 __builtin_mips_shll_qb (v4i8, i32)
9578 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
9579 v2q15 __builtin_mips_shll_ph (v2q15, i32)
9580 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
9581 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
9582 q31 __builtin_mips_shll_s_w (q31, imm0_31)
9583 q31 __builtin_mips_shll_s_w (q31, i32)
9584 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
9585 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
9586 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
9587 v2q15 __builtin_mips_shra_ph (v2q15, i32)
9588 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
9589 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
9590 q31 __builtin_mips_shra_r_w (q31, imm0_31)
9591 q31 __builtin_mips_shra_r_w (q31, i32)
9592 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
9593 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
9594 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
9595 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
9596 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
9597 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
9598 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
9599 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
9600 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
9601 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
9602 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
9603 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
9604 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
9605 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
9606 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
9607 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
9608 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
9609 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
9610 i32 __builtin_mips_bitrev (i32)
9611 i32 __builtin_mips_insv (i32, i32)
9612 v4i8 __builtin_mips_repl_qb (imm0_255)
9613 v4i8 __builtin_mips_repl_qb (i32)
9614 v2q15 __builtin_mips_repl_ph (imm_n512_511)
9615 v2q15 __builtin_mips_repl_ph (i32)
9616 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
9617 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
9618 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
9619 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
9620 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
9621 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
9622 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
9623 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
9624 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
9625 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
9626 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
9627 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
9628 i32 __builtin_mips_extr_w (a64, imm0_31)
9629 i32 __builtin_mips_extr_w (a64, i32)
9630 i32 __builtin_mips_extr_r_w (a64, imm0_31)
9631 i32 __builtin_mips_extr_s_h (a64, i32)
9632 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
9633 i32 __builtin_mips_extr_rs_w (a64, i32)
9634 i32 __builtin_mips_extr_s_h (a64, imm0_31)
9635 i32 __builtin_mips_extr_r_w (a64, i32)
9636 i32 __builtin_mips_extp (a64, imm0_31)
9637 i32 __builtin_mips_extp (a64, i32)
9638 i32 __builtin_mips_extpdp (a64, imm0_31)
9639 i32 __builtin_mips_extpdp (a64, i32)
9640 a64 __builtin_mips_shilo (a64, imm_n32_31)
9641 a64 __builtin_mips_shilo (a64, i32)
9642 a64 __builtin_mips_mthlip (a64, i32)
9643 void __builtin_mips_wrdsp (i32, imm0_63)
9644 i32 __builtin_mips_rddsp (imm0_63)
9645 i32 __builtin_mips_lbux (void *, i32)
9646 i32 __builtin_mips_lhx (void *, i32)
9647 i32 __builtin_mips_lwx (void *, i32)
9648 i32 __builtin_mips_bposge32 (void)
9651 The following built-in functions map directly to a particular MIPS DSP REV 2
9652 instruction. Please refer to the architecture specification
9653 for details on what each instruction does.
9656 v4q7 __builtin_mips_absq_s_qb (v4q7);
9657 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
9658 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
9659 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
9660 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
9661 i32 __builtin_mips_append (i32, i32, imm0_31);
9662 i32 __builtin_mips_balign (i32, i32, imm0_3);
9663 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
9664 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
9665 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
9666 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
9667 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
9668 a64 __builtin_mips_madd (a64, i32, i32);
9669 a64 __builtin_mips_maddu (a64, ui32, ui32);
9670 a64 __builtin_mips_msub (a64, i32, i32);
9671 a64 __builtin_mips_msubu (a64, ui32, ui32);
9672 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
9673 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
9674 q31 __builtin_mips_mulq_rs_w (q31, q31);
9675 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
9676 q31 __builtin_mips_mulq_s_w (q31, q31);
9677 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
9678 a64 __builtin_mips_mult (i32, i32);
9679 a64 __builtin_mips_multu (ui32, ui32);
9680 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
9681 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
9682 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
9683 i32 __builtin_mips_prepend (i32, i32, imm0_31);
9684 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
9685 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
9686 v4i8 __builtin_mips_shra_qb (v4i8, i32);
9687 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
9688 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
9689 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
9690 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
9691 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
9692 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
9693 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
9694 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
9695 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
9696 q31 __builtin_mips_addqh_w (q31, q31);
9697 q31 __builtin_mips_addqh_r_w (q31, q31);
9698 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
9699 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
9700 q31 __builtin_mips_subqh_w (q31, q31);
9701 q31 __builtin_mips_subqh_r_w (q31, q31);
9702 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
9703 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
9704 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
9705 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
9706 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
9707 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
9711 @node MIPS Paired-Single Support
9712 @subsection MIPS Paired-Single Support
9714 The MIPS64 architecture includes a number of instructions that
9715 operate on pairs of single-precision floating-point values.
9716 Each pair is packed into a 64-bit floating-point register,
9717 with one element being designated the ``upper half'' and
9718 the other being designated the ``lower half''.
9720 GCC supports paired-single operations using both the generic
9721 vector extensions (@pxref{Vector Extensions}) and a collection of
9722 MIPS-specific built-in functions. Both kinds of support are
9723 enabled by the @option{-mpaired-single} command-line option.
9725 The vector type associated with paired-single values is usually
9726 called @code{v2sf}. It can be defined in C as follows:
9729 typedef float v2sf __attribute__ ((vector_size (8)));
9732 @code{v2sf} values are initialized in the same way as aggregates.
9736 v2sf a = @{1.5, 9.1@};
9739 b = (v2sf) @{e, f@};
9742 @emph{Note:} The CPU's endianness determines which value is stored in
9743 the upper half of a register and which value is stored in the lower half.
9744 On little-endian targets, the first value is the lower one and the second
9745 value is the upper one. The opposite order applies to big-endian targets.
9746 For example, the code above will set the lower half of @code{a} to
9747 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
9749 @node MIPS Loongson Built-in Functions
9750 @subsection MIPS Loongson Built-in Functions
9752 GCC provides intrinsics to access the SIMD instructions provided by the
9753 ST Microelectronics Loongson-2E and -2F processors. These intrinsics,
9754 available after inclusion of the @code{loongson.h} header file,
9755 operate on the following 64-bit vector types:
9758 @item @code{uint8x8_t}, a vector of eight unsigned 8-bit integers;
9759 @item @code{uint16x4_t}, a vector of four unsigned 16-bit integers;
9760 @item @code{uint32x2_t}, a vector of two unsigned 32-bit integers;
9761 @item @code{int8x8_t}, a vector of eight signed 8-bit integers;
9762 @item @code{int16x4_t}, a vector of four signed 16-bit integers;
9763 @item @code{int32x2_t}, a vector of two signed 32-bit integers.
9766 The intrinsics provided are listed below; each is named after the
9767 machine instruction to which it corresponds, with suffixes added as
9768 appropriate to distinguish intrinsics that expand to the same machine
9769 instruction yet have different argument types. Refer to the architecture
9770 documentation for a description of the functionality of each
9774 int16x4_t packsswh (int32x2_t s, int32x2_t t);
9775 int8x8_t packsshb (int16x4_t s, int16x4_t t);
9776 uint8x8_t packushb (uint16x4_t s, uint16x4_t t);
9777 uint32x2_t paddw_u (uint32x2_t s, uint32x2_t t);
9778 uint16x4_t paddh_u (uint16x4_t s, uint16x4_t t);
9779 uint8x8_t paddb_u (uint8x8_t s, uint8x8_t t);
9780 int32x2_t paddw_s (int32x2_t s, int32x2_t t);
9781 int16x4_t paddh_s (int16x4_t s, int16x4_t t);
9782 int8x8_t paddb_s (int8x8_t s, int8x8_t t);
9783 uint64_t paddd_u (uint64_t s, uint64_t t);
9784 int64_t paddd_s (int64_t s, int64_t t);
9785 int16x4_t paddsh (int16x4_t s, int16x4_t t);
9786 int8x8_t paddsb (int8x8_t s, int8x8_t t);
9787 uint16x4_t paddush (uint16x4_t s, uint16x4_t t);
9788 uint8x8_t paddusb (uint8x8_t s, uint8x8_t t);
9789 uint64_t pandn_ud (uint64_t s, uint64_t t);
9790 uint32x2_t pandn_uw (uint32x2_t s, uint32x2_t t);
9791 uint16x4_t pandn_uh (uint16x4_t s, uint16x4_t t);
9792 uint8x8_t pandn_ub (uint8x8_t s, uint8x8_t t);
9793 int64_t pandn_sd (int64_t s, int64_t t);
9794 int32x2_t pandn_sw (int32x2_t s, int32x2_t t);
9795 int16x4_t pandn_sh (int16x4_t s, int16x4_t t);
9796 int8x8_t pandn_sb (int8x8_t s, int8x8_t t);
9797 uint16x4_t pavgh (uint16x4_t s, uint16x4_t t);
9798 uint8x8_t pavgb (uint8x8_t s, uint8x8_t t);
9799 uint32x2_t pcmpeqw_u (uint32x2_t s, uint32x2_t t);
9800 uint16x4_t pcmpeqh_u (uint16x4_t s, uint16x4_t t);
9801 uint8x8_t pcmpeqb_u (uint8x8_t s, uint8x8_t t);
9802 int32x2_t pcmpeqw_s (int32x2_t s, int32x2_t t);
9803 int16x4_t pcmpeqh_s (int16x4_t s, int16x4_t t);
9804 int8x8_t pcmpeqb_s (int8x8_t s, int8x8_t t);
9805 uint32x2_t pcmpgtw_u (uint32x2_t s, uint32x2_t t);
9806 uint16x4_t pcmpgth_u (uint16x4_t s, uint16x4_t t);
9807 uint8x8_t pcmpgtb_u (uint8x8_t s, uint8x8_t t);
9808 int32x2_t pcmpgtw_s (int32x2_t s, int32x2_t t);
9809 int16x4_t pcmpgth_s (int16x4_t s, int16x4_t t);
9810 int8x8_t pcmpgtb_s (int8x8_t s, int8x8_t t);
9811 uint16x4_t pextrh_u (uint16x4_t s, int field);
9812 int16x4_t pextrh_s (int16x4_t s, int field);
9813 uint16x4_t pinsrh_0_u (uint16x4_t s, uint16x4_t t);
9814 uint16x4_t pinsrh_1_u (uint16x4_t s, uint16x4_t t);
9815 uint16x4_t pinsrh_2_u (uint16x4_t s, uint16x4_t t);
9816 uint16x4_t pinsrh_3_u (uint16x4_t s, uint16x4_t t);
9817 int16x4_t pinsrh_0_s (int16x4_t s, int16x4_t t);
9818 int16x4_t pinsrh_1_s (int16x4_t s, int16x4_t t);
9819 int16x4_t pinsrh_2_s (int16x4_t s, int16x4_t t);
9820 int16x4_t pinsrh_3_s (int16x4_t s, int16x4_t t);
9821 int32x2_t pmaddhw (int16x4_t s, int16x4_t t);
9822 int16x4_t pmaxsh (int16x4_t s, int16x4_t t);
9823 uint8x8_t pmaxub (uint8x8_t s, uint8x8_t t);
9824 int16x4_t pminsh (int16x4_t s, int16x4_t t);
9825 uint8x8_t pminub (uint8x8_t s, uint8x8_t t);
9826 uint8x8_t pmovmskb_u (uint8x8_t s);
9827 int8x8_t pmovmskb_s (int8x8_t s);
9828 uint16x4_t pmulhuh (uint16x4_t s, uint16x4_t t);
9829 int16x4_t pmulhh (int16x4_t s, int16x4_t t);
9830 int16x4_t pmullh (int16x4_t s, int16x4_t t);
9831 int64_t pmuluw (uint32x2_t s, uint32x2_t t);
9832 uint8x8_t pasubub (uint8x8_t s, uint8x8_t t);
9833 uint16x4_t biadd (uint8x8_t s);
9834 uint16x4_t psadbh (uint8x8_t s, uint8x8_t t);
9835 uint16x4_t pshufh_u (uint16x4_t dest, uint16x4_t s, uint8_t order);
9836 int16x4_t pshufh_s (int16x4_t dest, int16x4_t s, uint8_t order);
9837 uint16x4_t psllh_u (uint16x4_t s, uint8_t amount);
9838 int16x4_t psllh_s (int16x4_t s, uint8_t amount);
9839 uint32x2_t psllw_u (uint32x2_t s, uint8_t amount);
9840 int32x2_t psllw_s (int32x2_t s, uint8_t amount);
9841 uint16x4_t psrlh_u (uint16x4_t s, uint8_t amount);
9842 int16x4_t psrlh_s (int16x4_t s, uint8_t amount);
9843 uint32x2_t psrlw_u (uint32x2_t s, uint8_t amount);
9844 int32x2_t psrlw_s (int32x2_t s, uint8_t amount);
9845 uint16x4_t psrah_u (uint16x4_t s, uint8_t amount);
9846 int16x4_t psrah_s (int16x4_t s, uint8_t amount);
9847 uint32x2_t psraw_u (uint32x2_t s, uint8_t amount);
9848 int32x2_t psraw_s (int32x2_t s, uint8_t amount);
9849 uint32x2_t psubw_u (uint32x2_t s, uint32x2_t t);
9850 uint16x4_t psubh_u (uint16x4_t s, uint16x4_t t);
9851 uint8x8_t psubb_u (uint8x8_t s, uint8x8_t t);
9852 int32x2_t psubw_s (int32x2_t s, int32x2_t t);
9853 int16x4_t psubh_s (int16x4_t s, int16x4_t t);
9854 int8x8_t psubb_s (int8x8_t s, int8x8_t t);
9855 uint64_t psubd_u (uint64_t s, uint64_t t);
9856 int64_t psubd_s (int64_t s, int64_t t);
9857 int16x4_t psubsh (int16x4_t s, int16x4_t t);
9858 int8x8_t psubsb (int8x8_t s, int8x8_t t);
9859 uint16x4_t psubush (uint16x4_t s, uint16x4_t t);
9860 uint8x8_t psubusb (uint8x8_t s, uint8x8_t t);
9861 uint32x2_t punpckhwd_u (uint32x2_t s, uint32x2_t t);
9862 uint16x4_t punpckhhw_u (uint16x4_t s, uint16x4_t t);
9863 uint8x8_t punpckhbh_u (uint8x8_t s, uint8x8_t t);
9864 int32x2_t punpckhwd_s (int32x2_t s, int32x2_t t);
9865 int16x4_t punpckhhw_s (int16x4_t s, int16x4_t t);
9866 int8x8_t punpckhbh_s (int8x8_t s, int8x8_t t);
9867 uint32x2_t punpcklwd_u (uint32x2_t s, uint32x2_t t);
9868 uint16x4_t punpcklhw_u (uint16x4_t s, uint16x4_t t);
9869 uint8x8_t punpcklbh_u (uint8x8_t s, uint8x8_t t);
9870 int32x2_t punpcklwd_s (int32x2_t s, int32x2_t t);
9871 int16x4_t punpcklhw_s (int16x4_t s, int16x4_t t);
9872 int8x8_t punpcklbh_s (int8x8_t s, int8x8_t t);
9876 * Paired-Single Arithmetic::
9877 * Paired-Single Built-in Functions::
9878 * MIPS-3D Built-in Functions::
9881 @node Paired-Single Arithmetic
9882 @subsubsection Paired-Single Arithmetic
9884 The table below lists the @code{v2sf} operations for which hardware
9885 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
9886 values and @code{x} is an integral value.
9888 @multitable @columnfractions .50 .50
9889 @item C code @tab MIPS instruction
9890 @item @code{a + b} @tab @code{add.ps}
9891 @item @code{a - b} @tab @code{sub.ps}
9892 @item @code{-a} @tab @code{neg.ps}
9893 @item @code{a * b} @tab @code{mul.ps}
9894 @item @code{a * b + c} @tab @code{madd.ps}
9895 @item @code{a * b - c} @tab @code{msub.ps}
9896 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
9897 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
9898 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
9901 Note that the multiply-accumulate instructions can be disabled
9902 using the command-line option @code{-mno-fused-madd}.
9904 @node Paired-Single Built-in Functions
9905 @subsubsection Paired-Single Built-in Functions
9907 The following paired-single functions map directly to a particular
9908 MIPS instruction. Please refer to the architecture specification
9909 for details on what each instruction does.
9912 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
9913 Pair lower lower (@code{pll.ps}).
9915 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
9916 Pair upper lower (@code{pul.ps}).
9918 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
9919 Pair lower upper (@code{plu.ps}).
9921 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
9922 Pair upper upper (@code{puu.ps}).
9924 @item v2sf __builtin_mips_cvt_ps_s (float, float)
9925 Convert pair to paired single (@code{cvt.ps.s}).
9927 @item float __builtin_mips_cvt_s_pl (v2sf)
9928 Convert pair lower to single (@code{cvt.s.pl}).
9930 @item float __builtin_mips_cvt_s_pu (v2sf)
9931 Convert pair upper to single (@code{cvt.s.pu}).
9933 @item v2sf __builtin_mips_abs_ps (v2sf)
9934 Absolute value (@code{abs.ps}).
9936 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
9937 Align variable (@code{alnv.ps}).
9939 @emph{Note:} The value of the third parameter must be 0 or 4
9940 modulo 8, otherwise the result will be unpredictable. Please read the
9941 instruction description for details.
9944 The following multi-instruction functions are also available.
9945 In each case, @var{cond} can be any of the 16 floating-point conditions:
9946 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
9947 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
9948 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
9951 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9952 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9953 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
9954 @code{movt.ps}/@code{movf.ps}).
9956 The @code{movt} functions return the value @var{x} computed by:
9959 c.@var{cond}.ps @var{cc},@var{a},@var{b}
9960 mov.ps @var{x},@var{c}
9961 movt.ps @var{x},@var{d},@var{cc}
9964 The @code{movf} functions are similar but use @code{movf.ps} instead
9967 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9968 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9969 Comparison of two paired-single values (@code{c.@var{cond}.ps},
9970 @code{bc1t}/@code{bc1f}).
9972 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
9973 and return either the upper or lower half of the result. For example:
9977 if (__builtin_mips_upper_c_eq_ps (a, b))
9978 upper_halves_are_equal ();
9980 upper_halves_are_unequal ();
9982 if (__builtin_mips_lower_c_eq_ps (a, b))
9983 lower_halves_are_equal ();
9985 lower_halves_are_unequal ();
9989 @node MIPS-3D Built-in Functions
9990 @subsubsection MIPS-3D Built-in Functions
9992 The MIPS-3D Application-Specific Extension (ASE) includes additional
9993 paired-single instructions that are designed to improve the performance
9994 of 3D graphics operations. Support for these instructions is controlled
9995 by the @option{-mips3d} command-line option.
9997 The functions listed below map directly to a particular MIPS-3D
9998 instruction. Please refer to the architecture specification for
9999 more details on what each instruction does.
10002 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
10003 Reduction add (@code{addr.ps}).
10005 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
10006 Reduction multiply (@code{mulr.ps}).
10008 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
10009 Convert paired single to paired word (@code{cvt.pw.ps}).
10011 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
10012 Convert paired word to paired single (@code{cvt.ps.pw}).
10014 @item float __builtin_mips_recip1_s (float)
10015 @itemx double __builtin_mips_recip1_d (double)
10016 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
10017 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
10019 @item float __builtin_mips_recip2_s (float, float)
10020 @itemx double __builtin_mips_recip2_d (double, double)
10021 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
10022 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
10024 @item float __builtin_mips_rsqrt1_s (float)
10025 @itemx double __builtin_mips_rsqrt1_d (double)
10026 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
10027 Reduced precision reciprocal square root (sequence step 1)
10028 (@code{rsqrt1.@var{fmt}}).
10030 @item float __builtin_mips_rsqrt2_s (float, float)
10031 @itemx double __builtin_mips_rsqrt2_d (double, double)
10032 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
10033 Reduced precision reciprocal square root (sequence step 2)
10034 (@code{rsqrt2.@var{fmt}}).
10037 The following multi-instruction functions are also available.
10038 In each case, @var{cond} can be any of the 16 floating-point conditions:
10039 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
10040 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
10041 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
10044 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
10045 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
10046 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
10047 @code{bc1t}/@code{bc1f}).
10049 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
10050 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
10055 if (__builtin_mips_cabs_eq_s (a, b))
10061 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10062 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10063 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
10064 @code{bc1t}/@code{bc1f}).
10066 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
10067 and return either the upper or lower half of the result. For example:
10071 if (__builtin_mips_upper_cabs_eq_ps (a, b))
10072 upper_halves_are_equal ();
10074 upper_halves_are_unequal ();
10076 if (__builtin_mips_lower_cabs_eq_ps (a, b))
10077 lower_halves_are_equal ();
10079 lower_halves_are_unequal ();
10082 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10083 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10084 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
10085 @code{movt.ps}/@code{movf.ps}).
10087 The @code{movt} functions return the value @var{x} computed by:
10090 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
10091 mov.ps @var{x},@var{c}
10092 movt.ps @var{x},@var{d},@var{cc}
10095 The @code{movf} functions are similar but use @code{movf.ps} instead
10098 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10099 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10100 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10101 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10102 Comparison of two paired-single values
10103 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
10104 @code{bc1any2t}/@code{bc1any2f}).
10106 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
10107 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
10108 result is true and the @code{all} forms return true if both results are true.
10113 if (__builtin_mips_any_c_eq_ps (a, b))
10118 if (__builtin_mips_all_c_eq_ps (a, b))
10124 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10125 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10126 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10127 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10128 Comparison of four paired-single values
10129 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
10130 @code{bc1any4t}/@code{bc1any4f}).
10132 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
10133 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
10134 The @code{any} forms return true if any of the four results are true
10135 and the @code{all} forms return true if all four results are true.
10140 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
10145 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
10152 @node picoChip Built-in Functions
10153 @subsection picoChip Built-in Functions
10155 GCC provides an interface to selected machine instructions from the
10156 picoChip instruction set.
10159 @item int __builtin_sbc (int @var{value})
10160 Sign bit count. Return the number of consecutive bits in @var{value}
10161 which have the same value as the sign-bit. The result is the number of
10162 leading sign bits minus one, giving the number of redundant sign bits in
10165 @item int __builtin_byteswap (int @var{value})
10166 Byte swap. Return the result of swapping the upper and lower bytes of
10169 @item int __builtin_brev (int @var{value})
10170 Bit reversal. Return the result of reversing the bits in
10171 @var{value}. Bit 15 is swapped with bit 0, bit 14 is swapped with bit 1,
10174 @item int __builtin_adds (int @var{x}, int @var{y})
10175 Saturating addition. Return the result of adding @var{x} and @var{y},
10176 storing the value 32767 if the result overflows.
10178 @item int __builtin_subs (int @var{x}, int @var{y})
10179 Saturating subtraction. Return the result of subtracting @var{y} from
10180 @var{x}, storing the value @minus{}32768 if the result overflows.
10182 @item void __builtin_halt (void)
10183 Halt. The processor will stop execution. This built-in is useful for
10184 implementing assertions.
10188 @node Other MIPS Built-in Functions
10189 @subsection Other MIPS Built-in Functions
10191 GCC provides other MIPS-specific built-in functions:
10194 @item void __builtin_mips_cache (int @var{op}, const volatile void *@var{addr})
10195 Insert a @samp{cache} instruction with operands @var{op} and @var{addr}.
10196 GCC defines the preprocessor macro @code{___GCC_HAVE_BUILTIN_MIPS_CACHE}
10197 when this function is available.
10200 @node PowerPC AltiVec/VSX Built-in Functions
10201 @subsection PowerPC AltiVec Built-in Functions
10203 GCC provides an interface for the PowerPC family of processors to access
10204 the AltiVec operations described in Motorola's AltiVec Programming
10205 Interface Manual. The interface is made available by including
10206 @code{<altivec.h>} and using @option{-maltivec} and
10207 @option{-mabi=altivec}. The interface supports the following vector
10211 vector unsigned char
10215 vector unsigned short
10216 vector signed short
10220 vector unsigned int
10226 If @option{-mvsx} is used the following additional vector types are
10230 vector unsigned long
10235 The long types are only implemented for 64-bit code generation, and
10236 the long type is only used in the floating point/integer conversion
10239 GCC's implementation of the high-level language interface available from
10240 C and C++ code differs from Motorola's documentation in several ways.
10245 A vector constant is a list of constant expressions within curly braces.
10248 A vector initializer requires no cast if the vector constant is of the
10249 same type as the variable it is initializing.
10252 If @code{signed} or @code{unsigned} is omitted, the signedness of the
10253 vector type is the default signedness of the base type. The default
10254 varies depending on the operating system, so a portable program should
10255 always specify the signedness.
10258 Compiling with @option{-maltivec} adds keywords @code{__vector},
10259 @code{vector}, @code{__pixel}, @code{pixel}, @code{__bool} and
10260 @code{bool}. When compiling ISO C, the context-sensitive substitution
10261 of the keywords @code{vector}, @code{pixel} and @code{bool} is
10262 disabled. To use them, you must include @code{<altivec.h>} instead.
10265 GCC allows using a @code{typedef} name as the type specifier for a
10269 For C, overloaded functions are implemented with macros so the following
10273 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
10276 Since @code{vec_add} is a macro, the vector constant in the example
10277 is treated as four separate arguments. Wrap the entire argument in
10278 parentheses for this to work.
10281 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
10282 Internally, GCC uses built-in functions to achieve the functionality in
10283 the aforementioned header file, but they are not supported and are
10284 subject to change without notice.
10286 The following interfaces are supported for the generic and specific
10287 AltiVec operations and the AltiVec predicates. In cases where there
10288 is a direct mapping between generic and specific operations, only the
10289 generic names are shown here, although the specific operations can also
10292 Arguments that are documented as @code{const int} require literal
10293 integral values within the range required for that operation.
10296 vector signed char vec_abs (vector signed char);
10297 vector signed short vec_abs (vector signed short);
10298 vector signed int vec_abs (vector signed int);
10299 vector float vec_abs (vector float);
10301 vector signed char vec_abss (vector signed char);
10302 vector signed short vec_abss (vector signed short);
10303 vector signed int vec_abss (vector signed int);
10305 vector signed char vec_add (vector bool char, vector signed char);
10306 vector signed char vec_add (vector signed char, vector bool char);
10307 vector signed char vec_add (vector signed char, vector signed char);
10308 vector unsigned char vec_add (vector bool char, vector unsigned char);
10309 vector unsigned char vec_add (vector unsigned char, vector bool char);
10310 vector unsigned char vec_add (vector unsigned char,
10311 vector unsigned char);
10312 vector signed short vec_add (vector bool short, vector signed short);
10313 vector signed short vec_add (vector signed short, vector bool short);
10314 vector signed short vec_add (vector signed short, vector signed short);
10315 vector unsigned short vec_add (vector bool short,
10316 vector unsigned short);
10317 vector unsigned short vec_add (vector unsigned short,
10318 vector bool short);
10319 vector unsigned short vec_add (vector unsigned short,
10320 vector unsigned short);
10321 vector signed int vec_add (vector bool int, vector signed int);
10322 vector signed int vec_add (vector signed int, vector bool int);
10323 vector signed int vec_add (vector signed int, vector signed int);
10324 vector unsigned int vec_add (vector bool int, vector unsigned int);
10325 vector unsigned int vec_add (vector unsigned int, vector bool int);
10326 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
10327 vector float vec_add (vector float, vector float);
10329 vector float vec_vaddfp (vector float, vector float);
10331 vector signed int vec_vadduwm (vector bool int, vector signed int);
10332 vector signed int vec_vadduwm (vector signed int, vector bool int);
10333 vector signed int vec_vadduwm (vector signed int, vector signed int);
10334 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
10335 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
10336 vector unsigned int vec_vadduwm (vector unsigned int,
10337 vector unsigned int);
10339 vector signed short vec_vadduhm (vector bool short,
10340 vector signed short);
10341 vector signed short vec_vadduhm (vector signed short,
10342 vector bool short);
10343 vector signed short vec_vadduhm (vector signed short,
10344 vector signed short);
10345 vector unsigned short vec_vadduhm (vector bool short,
10346 vector unsigned short);
10347 vector unsigned short vec_vadduhm (vector unsigned short,
10348 vector bool short);
10349 vector unsigned short vec_vadduhm (vector unsigned short,
10350 vector unsigned short);
10352 vector signed char vec_vaddubm (vector bool char, vector signed char);
10353 vector signed char vec_vaddubm (vector signed char, vector bool char);
10354 vector signed char vec_vaddubm (vector signed char, vector signed char);
10355 vector unsigned char vec_vaddubm (vector bool char,
10356 vector unsigned char);
10357 vector unsigned char vec_vaddubm (vector unsigned char,
10359 vector unsigned char vec_vaddubm (vector unsigned char,
10360 vector unsigned char);
10362 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
10364 vector unsigned char vec_adds (vector bool char, vector unsigned char);
10365 vector unsigned char vec_adds (vector unsigned char, vector bool char);
10366 vector unsigned char vec_adds (vector unsigned char,
10367 vector unsigned char);
10368 vector signed char vec_adds (vector bool char, vector signed char);
10369 vector signed char vec_adds (vector signed char, vector bool char);
10370 vector signed char vec_adds (vector signed char, vector signed char);
10371 vector unsigned short vec_adds (vector bool short,
10372 vector unsigned short);
10373 vector unsigned short vec_adds (vector unsigned short,
10374 vector bool short);
10375 vector unsigned short vec_adds (vector unsigned short,
10376 vector unsigned short);
10377 vector signed short vec_adds (vector bool short, vector signed short);
10378 vector signed short vec_adds (vector signed short, vector bool short);
10379 vector signed short vec_adds (vector signed short, vector signed short);
10380 vector unsigned int vec_adds (vector bool int, vector unsigned int);
10381 vector unsigned int vec_adds (vector unsigned int, vector bool int);
10382 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
10383 vector signed int vec_adds (vector bool int, vector signed int);
10384 vector signed int vec_adds (vector signed int, vector bool int);
10385 vector signed int vec_adds (vector signed int, vector signed int);
10387 vector signed int vec_vaddsws (vector bool int, vector signed int);
10388 vector signed int vec_vaddsws (vector signed int, vector bool int);
10389 vector signed int vec_vaddsws (vector signed int, vector signed int);
10391 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
10392 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
10393 vector unsigned int vec_vadduws (vector unsigned int,
10394 vector unsigned int);
10396 vector signed short vec_vaddshs (vector bool short,
10397 vector signed short);
10398 vector signed short vec_vaddshs (vector signed short,
10399 vector bool short);
10400 vector signed short vec_vaddshs (vector signed short,
10401 vector signed short);
10403 vector unsigned short vec_vadduhs (vector bool short,
10404 vector unsigned short);
10405 vector unsigned short vec_vadduhs (vector unsigned short,
10406 vector bool short);
10407 vector unsigned short vec_vadduhs (vector unsigned short,
10408 vector unsigned short);
10410 vector signed char vec_vaddsbs (vector bool char, vector signed char);
10411 vector signed char vec_vaddsbs (vector signed char, vector bool char);
10412 vector signed char vec_vaddsbs (vector signed char, vector signed char);
10414 vector unsigned char vec_vaddubs (vector bool char,
10415 vector unsigned char);
10416 vector unsigned char vec_vaddubs (vector unsigned char,
10418 vector unsigned char vec_vaddubs (vector unsigned char,
10419 vector unsigned char);
10421 vector float vec_and (vector float, vector float);
10422 vector float vec_and (vector float, vector bool int);
10423 vector float vec_and (vector bool int, vector float);
10424 vector bool int vec_and (vector bool int, vector bool int);
10425 vector signed int vec_and (vector bool int, vector signed int);
10426 vector signed int vec_and (vector signed int, vector bool int);
10427 vector signed int vec_and (vector signed int, vector signed int);
10428 vector unsigned int vec_and (vector bool int, vector unsigned int);
10429 vector unsigned int vec_and (vector unsigned int, vector bool int);
10430 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
10431 vector bool short vec_and (vector bool short, vector bool short);
10432 vector signed short vec_and (vector bool short, vector signed short);
10433 vector signed short vec_and (vector signed short, vector bool short);
10434 vector signed short vec_and (vector signed short, vector signed short);
10435 vector unsigned short vec_and (vector bool short,
10436 vector unsigned short);
10437 vector unsigned short vec_and (vector unsigned short,
10438 vector bool short);
10439 vector unsigned short vec_and (vector unsigned short,
10440 vector unsigned short);
10441 vector signed char vec_and (vector bool char, vector signed char);
10442 vector bool char vec_and (vector bool char, vector bool char);
10443 vector signed char vec_and (vector signed char, vector bool char);
10444 vector signed char vec_and (vector signed char, vector signed char);
10445 vector unsigned char vec_and (vector bool char, vector unsigned char);
10446 vector unsigned char vec_and (vector unsigned char, vector bool char);
10447 vector unsigned char vec_and (vector unsigned char,
10448 vector unsigned char);
10450 vector float vec_andc (vector float, vector float);
10451 vector float vec_andc (vector float, vector bool int);
10452 vector float vec_andc (vector bool int, vector float);
10453 vector bool int vec_andc (vector bool int, vector bool int);
10454 vector signed int vec_andc (vector bool int, vector signed int);
10455 vector signed int vec_andc (vector signed int, vector bool int);
10456 vector signed int vec_andc (vector signed int, vector signed int);
10457 vector unsigned int vec_andc (vector bool int, vector unsigned int);
10458 vector unsigned int vec_andc (vector unsigned int, vector bool int);
10459 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
10460 vector bool short vec_andc (vector bool short, vector bool short);
10461 vector signed short vec_andc (vector bool short, vector signed short);
10462 vector signed short vec_andc (vector signed short, vector bool short);
10463 vector signed short vec_andc (vector signed short, vector signed short);
10464 vector unsigned short vec_andc (vector bool short,
10465 vector unsigned short);
10466 vector unsigned short vec_andc (vector unsigned short,
10467 vector bool short);
10468 vector unsigned short vec_andc (vector unsigned short,
10469 vector unsigned short);
10470 vector signed char vec_andc (vector bool char, vector signed char);
10471 vector bool char vec_andc (vector bool char, vector bool char);
10472 vector signed char vec_andc (vector signed char, vector bool char);
10473 vector signed char vec_andc (vector signed char, vector signed char);
10474 vector unsigned char vec_andc (vector bool char, vector unsigned char);
10475 vector unsigned char vec_andc (vector unsigned char, vector bool char);
10476 vector unsigned char vec_andc (vector unsigned char,
10477 vector unsigned char);
10479 vector unsigned char vec_avg (vector unsigned char,
10480 vector unsigned char);
10481 vector signed char vec_avg (vector signed char, vector signed char);
10482 vector unsigned short vec_avg (vector unsigned short,
10483 vector unsigned short);
10484 vector signed short vec_avg (vector signed short, vector signed short);
10485 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
10486 vector signed int vec_avg (vector signed int, vector signed int);
10488 vector signed int vec_vavgsw (vector signed int, vector signed int);
10490 vector unsigned int vec_vavguw (vector unsigned int,
10491 vector unsigned int);
10493 vector signed short vec_vavgsh (vector signed short,
10494 vector signed short);
10496 vector unsigned short vec_vavguh (vector unsigned short,
10497 vector unsigned short);
10499 vector signed char vec_vavgsb (vector signed char, vector signed char);
10501 vector unsigned char vec_vavgub (vector unsigned char,
10502 vector unsigned char);
10504 vector float vec_copysign (vector float);
10506 vector float vec_ceil (vector float);
10508 vector signed int vec_cmpb (vector float, vector float);
10510 vector bool char vec_cmpeq (vector signed char, vector signed char);
10511 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
10512 vector bool short vec_cmpeq (vector signed short, vector signed short);
10513 vector bool short vec_cmpeq (vector unsigned short,
10514 vector unsigned short);
10515 vector bool int vec_cmpeq (vector signed int, vector signed int);
10516 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
10517 vector bool int vec_cmpeq (vector float, vector float);
10519 vector bool int vec_vcmpeqfp (vector float, vector float);
10521 vector bool int vec_vcmpequw (vector signed int, vector signed int);
10522 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
10524 vector bool short vec_vcmpequh (vector signed short,
10525 vector signed short);
10526 vector bool short vec_vcmpequh (vector unsigned short,
10527 vector unsigned short);
10529 vector bool char vec_vcmpequb (vector signed char, vector signed char);
10530 vector bool char vec_vcmpequb (vector unsigned char,
10531 vector unsigned char);
10533 vector bool int vec_cmpge (vector float, vector float);
10535 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
10536 vector bool char vec_cmpgt (vector signed char, vector signed char);
10537 vector bool short vec_cmpgt (vector unsigned short,
10538 vector unsigned short);
10539 vector bool short vec_cmpgt (vector signed short, vector signed short);
10540 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
10541 vector bool int vec_cmpgt (vector signed int, vector signed int);
10542 vector bool int vec_cmpgt (vector float, vector float);
10544 vector bool int vec_vcmpgtfp (vector float, vector float);
10546 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
10548 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
10550 vector bool short vec_vcmpgtsh (vector signed short,
10551 vector signed short);
10553 vector bool short vec_vcmpgtuh (vector unsigned short,
10554 vector unsigned short);
10556 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
10558 vector bool char vec_vcmpgtub (vector unsigned char,
10559 vector unsigned char);
10561 vector bool int vec_cmple (vector float, vector float);
10563 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
10564 vector bool char vec_cmplt (vector signed char, vector signed char);
10565 vector bool short vec_cmplt (vector unsigned short,
10566 vector unsigned short);
10567 vector bool short vec_cmplt (vector signed short, vector signed short);
10568 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
10569 vector bool int vec_cmplt (vector signed int, vector signed int);
10570 vector bool int vec_cmplt (vector float, vector float);
10572 vector float vec_ctf (vector unsigned int, const int);
10573 vector float vec_ctf (vector signed int, const int);
10575 vector float vec_vcfsx (vector signed int, const int);
10577 vector float vec_vcfux (vector unsigned int, const int);
10579 vector signed int vec_cts (vector float, const int);
10581 vector unsigned int vec_ctu (vector float, const int);
10583 void vec_dss (const int);
10585 void vec_dssall (void);
10587 void vec_dst (const vector unsigned char *, int, const int);
10588 void vec_dst (const vector signed char *, int, const int);
10589 void vec_dst (const vector bool char *, int, const int);
10590 void vec_dst (const vector unsigned short *, int, const int);
10591 void vec_dst (const vector signed short *, int, const int);
10592 void vec_dst (const vector bool short *, int, const int);
10593 void vec_dst (const vector pixel *, int, const int);
10594 void vec_dst (const vector unsigned int *, int, const int);
10595 void vec_dst (const vector signed int *, int, const int);
10596 void vec_dst (const vector bool int *, int, const int);
10597 void vec_dst (const vector float *, int, const int);
10598 void vec_dst (const unsigned char *, int, const int);
10599 void vec_dst (const signed char *, int, const int);
10600 void vec_dst (const unsigned short *, int, const int);
10601 void vec_dst (const short *, int, const int);
10602 void vec_dst (const unsigned int *, int, const int);
10603 void vec_dst (const int *, int, const int);
10604 void vec_dst (const unsigned long *, int, const int);
10605 void vec_dst (const long *, int, const int);
10606 void vec_dst (const float *, int, const int);
10608 void vec_dstst (const vector unsigned char *, int, const int);
10609 void vec_dstst (const vector signed char *, int, const int);
10610 void vec_dstst (const vector bool char *, int, const int);
10611 void vec_dstst (const vector unsigned short *, int, const int);
10612 void vec_dstst (const vector signed short *, int, const int);
10613 void vec_dstst (const vector bool short *, int, const int);
10614 void vec_dstst (const vector pixel *, int, const int);
10615 void vec_dstst (const vector unsigned int *, int, const int);
10616 void vec_dstst (const vector signed int *, int, const int);
10617 void vec_dstst (const vector bool int *, int, const int);
10618 void vec_dstst (const vector float *, int, const int);
10619 void vec_dstst (const unsigned char *, int, const int);
10620 void vec_dstst (const signed char *, int, const int);
10621 void vec_dstst (const unsigned short *, int, const int);
10622 void vec_dstst (const short *, int, const int);
10623 void vec_dstst (const unsigned int *, int, const int);
10624 void vec_dstst (const int *, int, const int);
10625 void vec_dstst (const unsigned long *, int, const int);
10626 void vec_dstst (const long *, int, const int);
10627 void vec_dstst (const float *, int, const int);
10629 void vec_dststt (const vector unsigned char *, int, const int);
10630 void vec_dststt (const vector signed char *, int, const int);
10631 void vec_dststt (const vector bool char *, int, const int);
10632 void vec_dststt (const vector unsigned short *, int, const int);
10633 void vec_dststt (const vector signed short *, int, const int);
10634 void vec_dststt (const vector bool short *, int, const int);
10635 void vec_dststt (const vector pixel *, int, const int);
10636 void vec_dststt (const vector unsigned int *, int, const int);
10637 void vec_dststt (const vector signed int *, int, const int);
10638 void vec_dststt (const vector bool int *, int, const int);
10639 void vec_dststt (const vector float *, int, const int);
10640 void vec_dststt (const unsigned char *, int, const int);
10641 void vec_dststt (const signed char *, int, const int);
10642 void vec_dststt (const unsigned short *, int, const int);
10643 void vec_dststt (const short *, int, const int);
10644 void vec_dststt (const unsigned int *, int, const int);
10645 void vec_dststt (const int *, int, const int);
10646 void vec_dststt (const unsigned long *, int, const int);
10647 void vec_dststt (const long *, int, const int);
10648 void vec_dststt (const float *, int, const int);
10650 void vec_dstt (const vector unsigned char *, int, const int);
10651 void vec_dstt (const vector signed char *, int, const int);
10652 void vec_dstt (const vector bool char *, int, const int);
10653 void vec_dstt (const vector unsigned short *, int, const int);
10654 void vec_dstt (const vector signed short *, int, const int);
10655 void vec_dstt (const vector bool short *, int, const int);
10656 void vec_dstt (const vector pixel *, int, const int);
10657 void vec_dstt (const vector unsigned int *, int, const int);
10658 void vec_dstt (const vector signed int *, int, const int);
10659 void vec_dstt (const vector bool int *, int, const int);
10660 void vec_dstt (const vector float *, int, const int);
10661 void vec_dstt (const unsigned char *, int, const int);
10662 void vec_dstt (const signed char *, int, const int);
10663 void vec_dstt (const unsigned short *, int, const int);
10664 void vec_dstt (const short *, int, const int);
10665 void vec_dstt (const unsigned int *, int, const int);
10666 void vec_dstt (const int *, int, const int);
10667 void vec_dstt (const unsigned long *, int, const int);
10668 void vec_dstt (const long *, int, const int);
10669 void vec_dstt (const float *, int, const int);
10671 vector float vec_expte (vector float);
10673 vector float vec_floor (vector float);
10675 vector float vec_ld (int, const vector float *);
10676 vector float vec_ld (int, const float *);
10677 vector bool int vec_ld (int, const vector bool int *);
10678 vector signed int vec_ld (int, const vector signed int *);
10679 vector signed int vec_ld (int, const int *);
10680 vector signed int vec_ld (int, const long *);
10681 vector unsigned int vec_ld (int, const vector unsigned int *);
10682 vector unsigned int vec_ld (int, const unsigned int *);
10683 vector unsigned int vec_ld (int, const unsigned long *);
10684 vector bool short vec_ld (int, const vector bool short *);
10685 vector pixel vec_ld (int, const vector pixel *);
10686 vector signed short vec_ld (int, const vector signed short *);
10687 vector signed short vec_ld (int, const short *);
10688 vector unsigned short vec_ld (int, const vector unsigned short *);
10689 vector unsigned short vec_ld (int, const unsigned short *);
10690 vector bool char vec_ld (int, const vector bool char *);
10691 vector signed char vec_ld (int, const vector signed char *);
10692 vector signed char vec_ld (int, const signed char *);
10693 vector unsigned char vec_ld (int, const vector unsigned char *);
10694 vector unsigned char vec_ld (int, const unsigned char *);
10696 vector signed char vec_lde (int, const signed char *);
10697 vector unsigned char vec_lde (int, const unsigned char *);
10698 vector signed short vec_lde (int, const short *);
10699 vector unsigned short vec_lde (int, const unsigned short *);
10700 vector float vec_lde (int, const float *);
10701 vector signed int vec_lde (int, const int *);
10702 vector unsigned int vec_lde (int, const unsigned int *);
10703 vector signed int vec_lde (int, const long *);
10704 vector unsigned int vec_lde (int, const unsigned long *);
10706 vector float vec_lvewx (int, float *);
10707 vector signed int vec_lvewx (int, int *);
10708 vector unsigned int vec_lvewx (int, unsigned int *);
10709 vector signed int vec_lvewx (int, long *);
10710 vector unsigned int vec_lvewx (int, unsigned long *);
10712 vector signed short vec_lvehx (int, short *);
10713 vector unsigned short vec_lvehx (int, unsigned short *);
10715 vector signed char vec_lvebx (int, char *);
10716 vector unsigned char vec_lvebx (int, unsigned char *);
10718 vector float vec_ldl (int, const vector float *);
10719 vector float vec_ldl (int, const float *);
10720 vector bool int vec_ldl (int, const vector bool int *);
10721 vector signed int vec_ldl (int, const vector signed int *);
10722 vector signed int vec_ldl (int, const int *);
10723 vector signed int vec_ldl (int, const long *);
10724 vector unsigned int vec_ldl (int, const vector unsigned int *);
10725 vector unsigned int vec_ldl (int, const unsigned int *);
10726 vector unsigned int vec_ldl (int, const unsigned long *);
10727 vector bool short vec_ldl (int, const vector bool short *);
10728 vector pixel vec_ldl (int, const vector pixel *);
10729 vector signed short vec_ldl (int, const vector signed short *);
10730 vector signed short vec_ldl (int, const short *);
10731 vector unsigned short vec_ldl (int, const vector unsigned short *);
10732 vector unsigned short vec_ldl (int, const unsigned short *);
10733 vector bool char vec_ldl (int, const vector bool char *);
10734 vector signed char vec_ldl (int, const vector signed char *);
10735 vector signed char vec_ldl (int, const signed char *);
10736 vector unsigned char vec_ldl (int, const vector unsigned char *);
10737 vector unsigned char vec_ldl (int, const unsigned char *);
10739 vector float vec_loge (vector float);
10741 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
10742 vector unsigned char vec_lvsl (int, const volatile signed char *);
10743 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
10744 vector unsigned char vec_lvsl (int, const volatile short *);
10745 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
10746 vector unsigned char vec_lvsl (int, const volatile int *);
10747 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
10748 vector unsigned char vec_lvsl (int, const volatile long *);
10749 vector unsigned char vec_lvsl (int, const volatile float *);
10751 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
10752 vector unsigned char vec_lvsr (int, const volatile signed char *);
10753 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
10754 vector unsigned char vec_lvsr (int, const volatile short *);
10755 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
10756 vector unsigned char vec_lvsr (int, const volatile int *);
10757 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
10758 vector unsigned char vec_lvsr (int, const volatile long *);
10759 vector unsigned char vec_lvsr (int, const volatile float *);
10761 vector float vec_madd (vector float, vector float, vector float);
10763 vector signed short vec_madds (vector signed short,
10764 vector signed short,
10765 vector signed short);
10767 vector unsigned char vec_max (vector bool char, vector unsigned char);
10768 vector unsigned char vec_max (vector unsigned char, vector bool char);
10769 vector unsigned char vec_max (vector unsigned char,
10770 vector unsigned char);
10771 vector signed char vec_max (vector bool char, vector signed char);
10772 vector signed char vec_max (vector signed char, vector bool char);
10773 vector signed char vec_max (vector signed char, vector signed char);
10774 vector unsigned short vec_max (vector bool short,
10775 vector unsigned short);
10776 vector unsigned short vec_max (vector unsigned short,
10777 vector bool short);
10778 vector unsigned short vec_max (vector unsigned short,
10779 vector unsigned short);
10780 vector signed short vec_max (vector bool short, vector signed short);
10781 vector signed short vec_max (vector signed short, vector bool short);
10782 vector signed short vec_max (vector signed short, vector signed short);
10783 vector unsigned int vec_max (vector bool int, vector unsigned int);
10784 vector unsigned int vec_max (vector unsigned int, vector bool int);
10785 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
10786 vector signed int vec_max (vector bool int, vector signed int);
10787 vector signed int vec_max (vector signed int, vector bool int);
10788 vector signed int vec_max (vector signed int, vector signed int);
10789 vector float vec_max (vector float, vector float);
10791 vector float vec_vmaxfp (vector float, vector float);
10793 vector signed int vec_vmaxsw (vector bool int, vector signed int);
10794 vector signed int vec_vmaxsw (vector signed int, vector bool int);
10795 vector signed int vec_vmaxsw (vector signed int, vector signed int);
10797 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
10798 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
10799 vector unsigned int vec_vmaxuw (vector unsigned int,
10800 vector unsigned int);
10802 vector signed short vec_vmaxsh (vector bool short, vector signed short);
10803 vector signed short vec_vmaxsh (vector signed short, vector bool short);
10804 vector signed short vec_vmaxsh (vector signed short,
10805 vector signed short);
10807 vector unsigned short vec_vmaxuh (vector bool short,
10808 vector unsigned short);
10809 vector unsigned short vec_vmaxuh (vector unsigned short,
10810 vector bool short);
10811 vector unsigned short vec_vmaxuh (vector unsigned short,
10812 vector unsigned short);
10814 vector signed char vec_vmaxsb (vector bool char, vector signed char);
10815 vector signed char vec_vmaxsb (vector signed char, vector bool char);
10816 vector signed char vec_vmaxsb (vector signed char, vector signed char);
10818 vector unsigned char vec_vmaxub (vector bool char,
10819 vector unsigned char);
10820 vector unsigned char vec_vmaxub (vector unsigned char,
10822 vector unsigned char vec_vmaxub (vector unsigned char,
10823 vector unsigned char);
10825 vector bool char vec_mergeh (vector bool char, vector bool char);
10826 vector signed char vec_mergeh (vector signed char, vector signed char);
10827 vector unsigned char vec_mergeh (vector unsigned char,
10828 vector unsigned char);
10829 vector bool short vec_mergeh (vector bool short, vector bool short);
10830 vector pixel vec_mergeh (vector pixel, vector pixel);
10831 vector signed short vec_mergeh (vector signed short,
10832 vector signed short);
10833 vector unsigned short vec_mergeh (vector unsigned short,
10834 vector unsigned short);
10835 vector float vec_mergeh (vector float, vector float);
10836 vector bool int vec_mergeh (vector bool int, vector bool int);
10837 vector signed int vec_mergeh (vector signed int, vector signed int);
10838 vector unsigned int vec_mergeh (vector unsigned int,
10839 vector unsigned int);
10841 vector float vec_vmrghw (vector float, vector float);
10842 vector bool int vec_vmrghw (vector bool int, vector bool int);
10843 vector signed int vec_vmrghw (vector signed int, vector signed int);
10844 vector unsigned int vec_vmrghw (vector unsigned int,
10845 vector unsigned int);
10847 vector bool short vec_vmrghh (vector bool short, vector bool short);
10848 vector signed short vec_vmrghh (vector signed short,
10849 vector signed short);
10850 vector unsigned short vec_vmrghh (vector unsigned short,
10851 vector unsigned short);
10852 vector pixel vec_vmrghh (vector pixel, vector pixel);
10854 vector bool char vec_vmrghb (vector bool char, vector bool char);
10855 vector signed char vec_vmrghb (vector signed char, vector signed char);
10856 vector unsigned char vec_vmrghb (vector unsigned char,
10857 vector unsigned char);
10859 vector bool char vec_mergel (vector bool char, vector bool char);
10860 vector signed char vec_mergel (vector signed char, vector signed char);
10861 vector unsigned char vec_mergel (vector unsigned char,
10862 vector unsigned char);
10863 vector bool short vec_mergel (vector bool short, vector bool short);
10864 vector pixel vec_mergel (vector pixel, vector pixel);
10865 vector signed short vec_mergel (vector signed short,
10866 vector signed short);
10867 vector unsigned short vec_mergel (vector unsigned short,
10868 vector unsigned short);
10869 vector float vec_mergel (vector float, vector float);
10870 vector bool int vec_mergel (vector bool int, vector bool int);
10871 vector signed int vec_mergel (vector signed int, vector signed int);
10872 vector unsigned int vec_mergel (vector unsigned int,
10873 vector unsigned int);
10875 vector float vec_vmrglw (vector float, vector float);
10876 vector signed int vec_vmrglw (vector signed int, vector signed int);
10877 vector unsigned int vec_vmrglw (vector unsigned int,
10878 vector unsigned int);
10879 vector bool int vec_vmrglw (vector bool int, vector bool int);
10881 vector bool short vec_vmrglh (vector bool short, vector bool short);
10882 vector signed short vec_vmrglh (vector signed short,
10883 vector signed short);
10884 vector unsigned short vec_vmrglh (vector unsigned short,
10885 vector unsigned short);
10886 vector pixel vec_vmrglh (vector pixel, vector pixel);
10888 vector bool char vec_vmrglb (vector bool char, vector bool char);
10889 vector signed char vec_vmrglb (vector signed char, vector signed char);
10890 vector unsigned char vec_vmrglb (vector unsigned char,
10891 vector unsigned char);
10893 vector unsigned short vec_mfvscr (void);
10895 vector unsigned char vec_min (vector bool char, vector unsigned char);
10896 vector unsigned char vec_min (vector unsigned char, vector bool char);
10897 vector unsigned char vec_min (vector unsigned char,
10898 vector unsigned char);
10899 vector signed char vec_min (vector bool char, vector signed char);
10900 vector signed char vec_min (vector signed char, vector bool char);
10901 vector signed char vec_min (vector signed char, vector signed char);
10902 vector unsigned short vec_min (vector bool short,
10903 vector unsigned short);
10904 vector unsigned short vec_min (vector unsigned short,
10905 vector bool short);
10906 vector unsigned short vec_min (vector unsigned short,
10907 vector unsigned short);
10908 vector signed short vec_min (vector bool short, vector signed short);
10909 vector signed short vec_min (vector signed short, vector bool short);
10910 vector signed short vec_min (vector signed short, vector signed short);
10911 vector unsigned int vec_min (vector bool int, vector unsigned int);
10912 vector unsigned int vec_min (vector unsigned int, vector bool int);
10913 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
10914 vector signed int vec_min (vector bool int, vector signed int);
10915 vector signed int vec_min (vector signed int, vector bool int);
10916 vector signed int vec_min (vector signed int, vector signed int);
10917 vector float vec_min (vector float, vector float);
10919 vector float vec_vminfp (vector float, vector float);
10921 vector signed int vec_vminsw (vector bool int, vector signed int);
10922 vector signed int vec_vminsw (vector signed int, vector bool int);
10923 vector signed int vec_vminsw (vector signed int, vector signed int);
10925 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
10926 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
10927 vector unsigned int vec_vminuw (vector unsigned int,
10928 vector unsigned int);
10930 vector signed short vec_vminsh (vector bool short, vector signed short);
10931 vector signed short vec_vminsh (vector signed short, vector bool short);
10932 vector signed short vec_vminsh (vector signed short,
10933 vector signed short);
10935 vector unsigned short vec_vminuh (vector bool short,
10936 vector unsigned short);
10937 vector unsigned short vec_vminuh (vector unsigned short,
10938 vector bool short);
10939 vector unsigned short vec_vminuh (vector unsigned short,
10940 vector unsigned short);
10942 vector signed char vec_vminsb (vector bool char, vector signed char);
10943 vector signed char vec_vminsb (vector signed char, vector bool char);
10944 vector signed char vec_vminsb (vector signed char, vector signed char);
10946 vector unsigned char vec_vminub (vector bool char,
10947 vector unsigned char);
10948 vector unsigned char vec_vminub (vector unsigned char,
10950 vector unsigned char vec_vminub (vector unsigned char,
10951 vector unsigned char);
10953 vector signed short vec_mladd (vector signed short,
10954 vector signed short,
10955 vector signed short);
10956 vector signed short vec_mladd (vector signed short,
10957 vector unsigned short,
10958 vector unsigned short);
10959 vector signed short vec_mladd (vector unsigned short,
10960 vector signed short,
10961 vector signed short);
10962 vector unsigned short vec_mladd (vector unsigned short,
10963 vector unsigned short,
10964 vector unsigned short);
10966 vector signed short vec_mradds (vector signed short,
10967 vector signed short,
10968 vector signed short);
10970 vector unsigned int vec_msum (vector unsigned char,
10971 vector unsigned char,
10972 vector unsigned int);
10973 vector signed int vec_msum (vector signed char,
10974 vector unsigned char,
10975 vector signed int);
10976 vector unsigned int vec_msum (vector unsigned short,
10977 vector unsigned short,
10978 vector unsigned int);
10979 vector signed int vec_msum (vector signed short,
10980 vector signed short,
10981 vector signed int);
10983 vector signed int vec_vmsumshm (vector signed short,
10984 vector signed short,
10985 vector signed int);
10987 vector unsigned int vec_vmsumuhm (vector unsigned short,
10988 vector unsigned short,
10989 vector unsigned int);
10991 vector signed int vec_vmsummbm (vector signed char,
10992 vector unsigned char,
10993 vector signed int);
10995 vector unsigned int vec_vmsumubm (vector unsigned char,
10996 vector unsigned char,
10997 vector unsigned int);
10999 vector unsigned int vec_msums (vector unsigned short,
11000 vector unsigned short,
11001 vector unsigned int);
11002 vector signed int vec_msums (vector signed short,
11003 vector signed short,
11004 vector signed int);
11006 vector signed int vec_vmsumshs (vector signed short,
11007 vector signed short,
11008 vector signed int);
11010 vector unsigned int vec_vmsumuhs (vector unsigned short,
11011 vector unsigned short,
11012 vector unsigned int);
11014 void vec_mtvscr (vector signed int);
11015 void vec_mtvscr (vector unsigned int);
11016 void vec_mtvscr (vector bool int);
11017 void vec_mtvscr (vector signed short);
11018 void vec_mtvscr (vector unsigned short);
11019 void vec_mtvscr (vector bool short);
11020 void vec_mtvscr (vector pixel);
11021 void vec_mtvscr (vector signed char);
11022 void vec_mtvscr (vector unsigned char);
11023 void vec_mtvscr (vector bool char);
11025 vector unsigned short vec_mule (vector unsigned char,
11026 vector unsigned char);
11027 vector signed short vec_mule (vector signed char,
11028 vector signed char);
11029 vector unsigned int vec_mule (vector unsigned short,
11030 vector unsigned short);
11031 vector signed int vec_mule (vector signed short, vector signed short);
11033 vector signed int vec_vmulesh (vector signed short,
11034 vector signed short);
11036 vector unsigned int vec_vmuleuh (vector unsigned short,
11037 vector unsigned short);
11039 vector signed short vec_vmulesb (vector signed char,
11040 vector signed char);
11042 vector unsigned short vec_vmuleub (vector unsigned char,
11043 vector unsigned char);
11045 vector unsigned short vec_mulo (vector unsigned char,
11046 vector unsigned char);
11047 vector signed short vec_mulo (vector signed char, vector signed char);
11048 vector unsigned int vec_mulo (vector unsigned short,
11049 vector unsigned short);
11050 vector signed int vec_mulo (vector signed short, vector signed short);
11052 vector signed int vec_vmulosh (vector signed short,
11053 vector signed short);
11055 vector unsigned int vec_vmulouh (vector unsigned short,
11056 vector unsigned short);
11058 vector signed short vec_vmulosb (vector signed char,
11059 vector signed char);
11061 vector unsigned short vec_vmuloub (vector unsigned char,
11062 vector unsigned char);
11064 vector float vec_nmsub (vector float, vector float, vector float);
11066 vector float vec_nor (vector float, vector float);
11067 vector signed int vec_nor (vector signed int, vector signed int);
11068 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
11069 vector bool int vec_nor (vector bool int, vector bool int);
11070 vector signed short vec_nor (vector signed short, vector signed short);
11071 vector unsigned short vec_nor (vector unsigned short,
11072 vector unsigned short);
11073 vector bool short vec_nor (vector bool short, vector bool short);
11074 vector signed char vec_nor (vector signed char, vector signed char);
11075 vector unsigned char vec_nor (vector unsigned char,
11076 vector unsigned char);
11077 vector bool char vec_nor (vector bool char, vector bool char);
11079 vector float vec_or (vector float, vector float);
11080 vector float vec_or (vector float, vector bool int);
11081 vector float vec_or (vector bool int, vector float);
11082 vector bool int vec_or (vector bool int, vector bool int);
11083 vector signed int vec_or (vector bool int, vector signed int);
11084 vector signed int vec_or (vector signed int, vector bool int);
11085 vector signed int vec_or (vector signed int, vector signed int);
11086 vector unsigned int vec_or (vector bool int, vector unsigned int);
11087 vector unsigned int vec_or (vector unsigned int, vector bool int);
11088 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
11089 vector bool short vec_or (vector bool short, vector bool short);
11090 vector signed short vec_or (vector bool short, vector signed short);
11091 vector signed short vec_or (vector signed short, vector bool short);
11092 vector signed short vec_or (vector signed short, vector signed short);
11093 vector unsigned short vec_or (vector bool short, vector unsigned short);
11094 vector unsigned short vec_or (vector unsigned short, vector bool short);
11095 vector unsigned short vec_or (vector unsigned short,
11096 vector unsigned short);
11097 vector signed char vec_or (vector bool char, vector signed char);
11098 vector bool char vec_or (vector bool char, vector bool char);
11099 vector signed char vec_or (vector signed char, vector bool char);
11100 vector signed char vec_or (vector signed char, vector signed char);
11101 vector unsigned char vec_or (vector bool char, vector unsigned char);
11102 vector unsigned char vec_or (vector unsigned char, vector bool char);
11103 vector unsigned char vec_or (vector unsigned char,
11104 vector unsigned char);
11106 vector signed char vec_pack (vector signed short, vector signed short);
11107 vector unsigned char vec_pack (vector unsigned short,
11108 vector unsigned short);
11109 vector bool char vec_pack (vector bool short, vector bool short);
11110 vector signed short vec_pack (vector signed int, vector signed int);
11111 vector unsigned short vec_pack (vector unsigned int,
11112 vector unsigned int);
11113 vector bool short vec_pack (vector bool int, vector bool int);
11115 vector bool short vec_vpkuwum (vector bool int, vector bool int);
11116 vector signed short vec_vpkuwum (vector signed int, vector signed int);
11117 vector unsigned short vec_vpkuwum (vector unsigned int,
11118 vector unsigned int);
11120 vector bool char vec_vpkuhum (vector bool short, vector bool short);
11121 vector signed char vec_vpkuhum (vector signed short,
11122 vector signed short);
11123 vector unsigned char vec_vpkuhum (vector unsigned short,
11124 vector unsigned short);
11126 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
11128 vector unsigned char vec_packs (vector unsigned short,
11129 vector unsigned short);
11130 vector signed char vec_packs (vector signed short, vector signed short);
11131 vector unsigned short vec_packs (vector unsigned int,
11132 vector unsigned int);
11133 vector signed short vec_packs (vector signed int, vector signed int);
11135 vector signed short vec_vpkswss (vector signed int, vector signed int);
11137 vector unsigned short vec_vpkuwus (vector unsigned int,
11138 vector unsigned int);
11140 vector signed char vec_vpkshss (vector signed short,
11141 vector signed short);
11143 vector unsigned char vec_vpkuhus (vector unsigned short,
11144 vector unsigned short);
11146 vector unsigned char vec_packsu (vector unsigned short,
11147 vector unsigned short);
11148 vector unsigned char vec_packsu (vector signed short,
11149 vector signed short);
11150 vector unsigned short vec_packsu (vector unsigned int,
11151 vector unsigned int);
11152 vector unsigned short vec_packsu (vector signed int, vector signed int);
11154 vector unsigned short vec_vpkswus (vector signed int,
11155 vector signed int);
11157 vector unsigned char vec_vpkshus (vector signed short,
11158 vector signed short);
11160 vector float vec_perm (vector float,
11162 vector unsigned char);
11163 vector signed int vec_perm (vector signed int,
11165 vector unsigned char);
11166 vector unsigned int vec_perm (vector unsigned int,
11167 vector unsigned int,
11168 vector unsigned char);
11169 vector bool int vec_perm (vector bool int,
11171 vector unsigned char);
11172 vector signed short vec_perm (vector signed short,
11173 vector signed short,
11174 vector unsigned char);
11175 vector unsigned short vec_perm (vector unsigned short,
11176 vector unsigned short,
11177 vector unsigned char);
11178 vector bool short vec_perm (vector bool short,
11180 vector unsigned char);
11181 vector pixel vec_perm (vector pixel,
11183 vector unsigned char);
11184 vector signed char vec_perm (vector signed char,
11185 vector signed char,
11186 vector unsigned char);
11187 vector unsigned char vec_perm (vector unsigned char,
11188 vector unsigned char,
11189 vector unsigned char);
11190 vector bool char vec_perm (vector bool char,
11192 vector unsigned char);
11194 vector float vec_re (vector float);
11196 vector signed char vec_rl (vector signed char,
11197 vector unsigned char);
11198 vector unsigned char vec_rl (vector unsigned char,
11199 vector unsigned char);
11200 vector signed short vec_rl (vector signed short, vector unsigned short);
11201 vector unsigned short vec_rl (vector unsigned short,
11202 vector unsigned short);
11203 vector signed int vec_rl (vector signed int, vector unsigned int);
11204 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
11206 vector signed int vec_vrlw (vector signed int, vector unsigned int);
11207 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
11209 vector signed short vec_vrlh (vector signed short,
11210 vector unsigned short);
11211 vector unsigned short vec_vrlh (vector unsigned short,
11212 vector unsigned short);
11214 vector signed char vec_vrlb (vector signed char, vector unsigned char);
11215 vector unsigned char vec_vrlb (vector unsigned char,
11216 vector unsigned char);
11218 vector float vec_round (vector float);
11220 vector float vec_recip (vector float, vector float);
11222 vector float vec_rsqrt (vector float);
11224 vector float vec_rsqrte (vector float);
11226 vector float vec_sel (vector float, vector float, vector bool int);
11227 vector float vec_sel (vector float, vector float, vector unsigned int);
11228 vector signed int vec_sel (vector signed int,
11231 vector signed int vec_sel (vector signed int,
11233 vector unsigned int);
11234 vector unsigned int vec_sel (vector unsigned int,
11235 vector unsigned int,
11237 vector unsigned int vec_sel (vector unsigned int,
11238 vector unsigned int,
11239 vector unsigned int);
11240 vector bool int vec_sel (vector bool int,
11243 vector bool int vec_sel (vector bool int,
11245 vector unsigned int);
11246 vector signed short vec_sel (vector signed short,
11247 vector signed short,
11248 vector bool short);
11249 vector signed short vec_sel (vector signed short,
11250 vector signed short,
11251 vector unsigned short);
11252 vector unsigned short vec_sel (vector unsigned short,
11253 vector unsigned short,
11254 vector bool short);
11255 vector unsigned short vec_sel (vector unsigned short,
11256 vector unsigned short,
11257 vector unsigned short);
11258 vector bool short vec_sel (vector bool short,
11260 vector bool short);
11261 vector bool short vec_sel (vector bool short,
11263 vector unsigned short);
11264 vector signed char vec_sel (vector signed char,
11265 vector signed char,
11267 vector signed char vec_sel (vector signed char,
11268 vector signed char,
11269 vector unsigned char);
11270 vector unsigned char vec_sel (vector unsigned char,
11271 vector unsigned char,
11273 vector unsigned char vec_sel (vector unsigned char,
11274 vector unsigned char,
11275 vector unsigned char);
11276 vector bool char vec_sel (vector bool char,
11279 vector bool char vec_sel (vector bool char,
11281 vector unsigned char);
11283 vector signed char vec_sl (vector signed char,
11284 vector unsigned char);
11285 vector unsigned char vec_sl (vector unsigned char,
11286 vector unsigned char);
11287 vector signed short vec_sl (vector signed short, vector unsigned short);
11288 vector unsigned short vec_sl (vector unsigned short,
11289 vector unsigned short);
11290 vector signed int vec_sl (vector signed int, vector unsigned int);
11291 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
11293 vector signed int vec_vslw (vector signed int, vector unsigned int);
11294 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
11296 vector signed short vec_vslh (vector signed short,
11297 vector unsigned short);
11298 vector unsigned short vec_vslh (vector unsigned short,
11299 vector unsigned short);
11301 vector signed char vec_vslb (vector signed char, vector unsigned char);
11302 vector unsigned char vec_vslb (vector unsigned char,
11303 vector unsigned char);
11305 vector float vec_sld (vector float, vector float, const int);
11306 vector signed int vec_sld (vector signed int,
11309 vector unsigned int vec_sld (vector unsigned int,
11310 vector unsigned int,
11312 vector bool int vec_sld (vector bool int,
11315 vector signed short vec_sld (vector signed short,
11316 vector signed short,
11318 vector unsigned short vec_sld (vector unsigned short,
11319 vector unsigned short,
11321 vector bool short vec_sld (vector bool short,
11324 vector pixel vec_sld (vector pixel,
11327 vector signed char vec_sld (vector signed char,
11328 vector signed char,
11330 vector unsigned char vec_sld (vector unsigned char,
11331 vector unsigned char,
11333 vector bool char vec_sld (vector bool char,
11337 vector signed int vec_sll (vector signed int,
11338 vector unsigned int);
11339 vector signed int vec_sll (vector signed int,
11340 vector unsigned short);
11341 vector signed int vec_sll (vector signed int,
11342 vector unsigned char);
11343 vector unsigned int vec_sll (vector unsigned int,
11344 vector unsigned int);
11345 vector unsigned int vec_sll (vector unsigned int,
11346 vector unsigned short);
11347 vector unsigned int vec_sll (vector unsigned int,
11348 vector unsigned char);
11349 vector bool int vec_sll (vector bool int,
11350 vector unsigned int);
11351 vector bool int vec_sll (vector bool int,
11352 vector unsigned short);
11353 vector bool int vec_sll (vector bool int,
11354 vector unsigned char);
11355 vector signed short vec_sll (vector signed short,
11356 vector unsigned int);
11357 vector signed short vec_sll (vector signed short,
11358 vector unsigned short);
11359 vector signed short vec_sll (vector signed short,
11360 vector unsigned char);
11361 vector unsigned short vec_sll (vector unsigned short,
11362 vector unsigned int);
11363 vector unsigned short vec_sll (vector unsigned short,
11364 vector unsigned short);
11365 vector unsigned short vec_sll (vector unsigned short,
11366 vector unsigned char);
11367 vector bool short vec_sll (vector bool short, vector unsigned int);
11368 vector bool short vec_sll (vector bool short, vector unsigned short);
11369 vector bool short vec_sll (vector bool short, vector unsigned char);
11370 vector pixel vec_sll (vector pixel, vector unsigned int);
11371 vector pixel vec_sll (vector pixel, vector unsigned short);
11372 vector pixel vec_sll (vector pixel, vector unsigned char);
11373 vector signed char vec_sll (vector signed char, vector unsigned int);
11374 vector signed char vec_sll (vector signed char, vector unsigned short);
11375 vector signed char vec_sll (vector signed char, vector unsigned char);
11376 vector unsigned char vec_sll (vector unsigned char,
11377 vector unsigned int);
11378 vector unsigned char vec_sll (vector unsigned char,
11379 vector unsigned short);
11380 vector unsigned char vec_sll (vector unsigned char,
11381 vector unsigned char);
11382 vector bool char vec_sll (vector bool char, vector unsigned int);
11383 vector bool char vec_sll (vector bool char, vector unsigned short);
11384 vector bool char vec_sll (vector bool char, vector unsigned char);
11386 vector float vec_slo (vector float, vector signed char);
11387 vector float vec_slo (vector float, vector unsigned char);
11388 vector signed int vec_slo (vector signed int, vector signed char);
11389 vector signed int vec_slo (vector signed int, vector unsigned char);
11390 vector unsigned int vec_slo (vector unsigned int, vector signed char);
11391 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
11392 vector signed short vec_slo (vector signed short, vector signed char);
11393 vector signed short vec_slo (vector signed short, vector unsigned char);
11394 vector unsigned short vec_slo (vector unsigned short,
11395 vector signed char);
11396 vector unsigned short vec_slo (vector unsigned short,
11397 vector unsigned char);
11398 vector pixel vec_slo (vector pixel, vector signed char);
11399 vector pixel vec_slo (vector pixel, vector unsigned char);
11400 vector signed char vec_slo (vector signed char, vector signed char);
11401 vector signed char vec_slo (vector signed char, vector unsigned char);
11402 vector unsigned char vec_slo (vector unsigned char, vector signed char);
11403 vector unsigned char vec_slo (vector unsigned char,
11404 vector unsigned char);
11406 vector signed char vec_splat (vector signed char, const int);
11407 vector unsigned char vec_splat (vector unsigned char, const int);
11408 vector bool char vec_splat (vector bool char, const int);
11409 vector signed short vec_splat (vector signed short, const int);
11410 vector unsigned short vec_splat (vector unsigned short, const int);
11411 vector bool short vec_splat (vector bool short, const int);
11412 vector pixel vec_splat (vector pixel, const int);
11413 vector float vec_splat (vector float, const int);
11414 vector signed int vec_splat (vector signed int, const int);
11415 vector unsigned int vec_splat (vector unsigned int, const int);
11416 vector bool int vec_splat (vector bool int, const int);
11418 vector float vec_vspltw (vector float, const int);
11419 vector signed int vec_vspltw (vector signed int, const int);
11420 vector unsigned int vec_vspltw (vector unsigned int, const int);
11421 vector bool int vec_vspltw (vector bool int, const int);
11423 vector bool short vec_vsplth (vector bool short, const int);
11424 vector signed short vec_vsplth (vector signed short, const int);
11425 vector unsigned short vec_vsplth (vector unsigned short, const int);
11426 vector pixel vec_vsplth (vector pixel, const int);
11428 vector signed char vec_vspltb (vector signed char, const int);
11429 vector unsigned char vec_vspltb (vector unsigned char, const int);
11430 vector bool char vec_vspltb (vector bool char, const int);
11432 vector signed char vec_splat_s8 (const int);
11434 vector signed short vec_splat_s16 (const int);
11436 vector signed int vec_splat_s32 (const int);
11438 vector unsigned char vec_splat_u8 (const int);
11440 vector unsigned short vec_splat_u16 (const int);
11442 vector unsigned int vec_splat_u32 (const int);
11444 vector signed char vec_sr (vector signed char, vector unsigned char);
11445 vector unsigned char vec_sr (vector unsigned char,
11446 vector unsigned char);
11447 vector signed short vec_sr (vector signed short,
11448 vector unsigned short);
11449 vector unsigned short vec_sr (vector unsigned short,
11450 vector unsigned short);
11451 vector signed int vec_sr (vector signed int, vector unsigned int);
11452 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
11454 vector signed int vec_vsrw (vector signed int, vector unsigned int);
11455 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
11457 vector signed short vec_vsrh (vector signed short,
11458 vector unsigned short);
11459 vector unsigned short vec_vsrh (vector unsigned short,
11460 vector unsigned short);
11462 vector signed char vec_vsrb (vector signed char, vector unsigned char);
11463 vector unsigned char vec_vsrb (vector unsigned char,
11464 vector unsigned char);
11466 vector signed char vec_sra (vector signed char, vector unsigned char);
11467 vector unsigned char vec_sra (vector unsigned char,
11468 vector unsigned char);
11469 vector signed short vec_sra (vector signed short,
11470 vector unsigned short);
11471 vector unsigned short vec_sra (vector unsigned short,
11472 vector unsigned short);
11473 vector signed int vec_sra (vector signed int, vector unsigned int);
11474 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
11476 vector signed int vec_vsraw (vector signed int, vector unsigned int);
11477 vector unsigned int vec_vsraw (vector unsigned int,
11478 vector unsigned int);
11480 vector signed short vec_vsrah (vector signed short,
11481 vector unsigned short);
11482 vector unsigned short vec_vsrah (vector unsigned short,
11483 vector unsigned short);
11485 vector signed char vec_vsrab (vector signed char, vector unsigned char);
11486 vector unsigned char vec_vsrab (vector unsigned char,
11487 vector unsigned char);
11489 vector signed int vec_srl (vector signed int, vector unsigned int);
11490 vector signed int vec_srl (vector signed int, vector unsigned short);
11491 vector signed int vec_srl (vector signed int, vector unsigned char);
11492 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
11493 vector unsigned int vec_srl (vector unsigned int,
11494 vector unsigned short);
11495 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
11496 vector bool int vec_srl (vector bool int, vector unsigned int);
11497 vector bool int vec_srl (vector bool int, vector unsigned short);
11498 vector bool int vec_srl (vector bool int, vector unsigned char);
11499 vector signed short vec_srl (vector signed short, vector unsigned int);
11500 vector signed short vec_srl (vector signed short,
11501 vector unsigned short);
11502 vector signed short vec_srl (vector signed short, vector unsigned char);
11503 vector unsigned short vec_srl (vector unsigned short,
11504 vector unsigned int);
11505 vector unsigned short vec_srl (vector unsigned short,
11506 vector unsigned short);
11507 vector unsigned short vec_srl (vector unsigned short,
11508 vector unsigned char);
11509 vector bool short vec_srl (vector bool short, vector unsigned int);
11510 vector bool short vec_srl (vector bool short, vector unsigned short);
11511 vector bool short vec_srl (vector bool short, vector unsigned char);
11512 vector pixel vec_srl (vector pixel, vector unsigned int);
11513 vector pixel vec_srl (vector pixel, vector unsigned short);
11514 vector pixel vec_srl (vector pixel, vector unsigned char);
11515 vector signed char vec_srl (vector signed char, vector unsigned int);
11516 vector signed char vec_srl (vector signed char, vector unsigned short);
11517 vector signed char vec_srl (vector signed char, vector unsigned char);
11518 vector unsigned char vec_srl (vector unsigned char,
11519 vector unsigned int);
11520 vector unsigned char vec_srl (vector unsigned char,
11521 vector unsigned short);
11522 vector unsigned char vec_srl (vector unsigned char,
11523 vector unsigned char);
11524 vector bool char vec_srl (vector bool char, vector unsigned int);
11525 vector bool char vec_srl (vector bool char, vector unsigned short);
11526 vector bool char vec_srl (vector bool char, vector unsigned char);
11528 vector float vec_sro (vector float, vector signed char);
11529 vector float vec_sro (vector float, vector unsigned char);
11530 vector signed int vec_sro (vector signed int, vector signed char);
11531 vector signed int vec_sro (vector signed int, vector unsigned char);
11532 vector unsigned int vec_sro (vector unsigned int, vector signed char);
11533 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
11534 vector signed short vec_sro (vector signed short, vector signed char);
11535 vector signed short vec_sro (vector signed short, vector unsigned char);
11536 vector unsigned short vec_sro (vector unsigned short,
11537 vector signed char);
11538 vector unsigned short vec_sro (vector unsigned short,
11539 vector unsigned char);
11540 vector pixel vec_sro (vector pixel, vector signed char);
11541 vector pixel vec_sro (vector pixel, vector unsigned char);
11542 vector signed char vec_sro (vector signed char, vector signed char);
11543 vector signed char vec_sro (vector signed char, vector unsigned char);
11544 vector unsigned char vec_sro (vector unsigned char, vector signed char);
11545 vector unsigned char vec_sro (vector unsigned char,
11546 vector unsigned char);
11548 void vec_st (vector float, int, vector float *);
11549 void vec_st (vector float, int, float *);
11550 void vec_st (vector signed int, int, vector signed int *);
11551 void vec_st (vector signed int, int, int *);
11552 void vec_st (vector unsigned int, int, vector unsigned int *);
11553 void vec_st (vector unsigned int, int, unsigned int *);
11554 void vec_st (vector bool int, int, vector bool int *);
11555 void vec_st (vector bool int, int, unsigned int *);
11556 void vec_st (vector bool int, int, int *);
11557 void vec_st (vector signed short, int, vector signed short *);
11558 void vec_st (vector signed short, int, short *);
11559 void vec_st (vector unsigned short, int, vector unsigned short *);
11560 void vec_st (vector unsigned short, int, unsigned short *);
11561 void vec_st (vector bool short, int, vector bool short *);
11562 void vec_st (vector bool short, int, unsigned short *);
11563 void vec_st (vector pixel, int, vector pixel *);
11564 void vec_st (vector pixel, int, unsigned short *);
11565 void vec_st (vector pixel, int, short *);
11566 void vec_st (vector bool short, int, short *);
11567 void vec_st (vector signed char, int, vector signed char *);
11568 void vec_st (vector signed char, int, signed char *);
11569 void vec_st (vector unsigned char, int, vector unsigned char *);
11570 void vec_st (vector unsigned char, int, unsigned char *);
11571 void vec_st (vector bool char, int, vector bool char *);
11572 void vec_st (vector bool char, int, unsigned char *);
11573 void vec_st (vector bool char, int, signed char *);
11575 void vec_ste (vector signed char, int, signed char *);
11576 void vec_ste (vector unsigned char, int, unsigned char *);
11577 void vec_ste (vector bool char, int, signed char *);
11578 void vec_ste (vector bool char, int, unsigned char *);
11579 void vec_ste (vector signed short, int, short *);
11580 void vec_ste (vector unsigned short, int, unsigned short *);
11581 void vec_ste (vector bool short, int, short *);
11582 void vec_ste (vector bool short, int, unsigned short *);
11583 void vec_ste (vector pixel, int, short *);
11584 void vec_ste (vector pixel, int, unsigned short *);
11585 void vec_ste (vector float, int, float *);
11586 void vec_ste (vector signed int, int, int *);
11587 void vec_ste (vector unsigned int, int, unsigned int *);
11588 void vec_ste (vector bool int, int, int *);
11589 void vec_ste (vector bool int, int, unsigned int *);
11591 void vec_stvewx (vector float, int, float *);
11592 void vec_stvewx (vector signed int, int, int *);
11593 void vec_stvewx (vector unsigned int, int, unsigned int *);
11594 void vec_stvewx (vector bool int, int, int *);
11595 void vec_stvewx (vector bool int, int, unsigned int *);
11597 void vec_stvehx (vector signed short, int, short *);
11598 void vec_stvehx (vector unsigned short, int, unsigned short *);
11599 void vec_stvehx (vector bool short, int, short *);
11600 void vec_stvehx (vector bool short, int, unsigned short *);
11601 void vec_stvehx (vector pixel, int, short *);
11602 void vec_stvehx (vector pixel, int, unsigned short *);
11604 void vec_stvebx (vector signed char, int, signed char *);
11605 void vec_stvebx (vector unsigned char, int, unsigned char *);
11606 void vec_stvebx (vector bool char, int, signed char *);
11607 void vec_stvebx (vector bool char, int, unsigned char *);
11609 void vec_stl (vector float, int, vector float *);
11610 void vec_stl (vector float, int, float *);
11611 void vec_stl (vector signed int, int, vector signed int *);
11612 void vec_stl (vector signed int, int, int *);
11613 void vec_stl (vector unsigned int, int, vector unsigned int *);
11614 void vec_stl (vector unsigned int, int, unsigned int *);
11615 void vec_stl (vector bool int, int, vector bool int *);
11616 void vec_stl (vector bool int, int, unsigned int *);
11617 void vec_stl (vector bool int, int, int *);
11618 void vec_stl (vector signed short, int, vector signed short *);
11619 void vec_stl (vector signed short, int, short *);
11620 void vec_stl (vector unsigned short, int, vector unsigned short *);
11621 void vec_stl (vector unsigned short, int, unsigned short *);
11622 void vec_stl (vector bool short, int, vector bool short *);
11623 void vec_stl (vector bool short, int, unsigned short *);
11624 void vec_stl (vector bool short, int, short *);
11625 void vec_stl (vector pixel, int, vector pixel *);
11626 void vec_stl (vector pixel, int, unsigned short *);
11627 void vec_stl (vector pixel, int, short *);
11628 void vec_stl (vector signed char, int, vector signed char *);
11629 void vec_stl (vector signed char, int, signed char *);
11630 void vec_stl (vector unsigned char, int, vector unsigned char *);
11631 void vec_stl (vector unsigned char, int, unsigned char *);
11632 void vec_stl (vector bool char, int, vector bool char *);
11633 void vec_stl (vector bool char, int, unsigned char *);
11634 void vec_stl (vector bool char, int, signed char *);
11636 vector signed char vec_sub (vector bool char, vector signed char);
11637 vector signed char vec_sub (vector signed char, vector bool char);
11638 vector signed char vec_sub (vector signed char, vector signed char);
11639 vector unsigned char vec_sub (vector bool char, vector unsigned char);
11640 vector unsigned char vec_sub (vector unsigned char, vector bool char);
11641 vector unsigned char vec_sub (vector unsigned char,
11642 vector unsigned char);
11643 vector signed short vec_sub (vector bool short, vector signed short);
11644 vector signed short vec_sub (vector signed short, vector bool short);
11645 vector signed short vec_sub (vector signed short, vector signed short);
11646 vector unsigned short vec_sub (vector bool short,
11647 vector unsigned short);
11648 vector unsigned short vec_sub (vector unsigned short,
11649 vector bool short);
11650 vector unsigned short vec_sub (vector unsigned short,
11651 vector unsigned short);
11652 vector signed int vec_sub (vector bool int, vector signed int);
11653 vector signed int vec_sub (vector signed int, vector bool int);
11654 vector signed int vec_sub (vector signed int, vector signed int);
11655 vector unsigned int vec_sub (vector bool int, vector unsigned int);
11656 vector unsigned int vec_sub (vector unsigned int, vector bool int);
11657 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
11658 vector float vec_sub (vector float, vector float);
11660 vector float vec_vsubfp (vector float, vector float);
11662 vector signed int vec_vsubuwm (vector bool int, vector signed int);
11663 vector signed int vec_vsubuwm (vector signed int, vector bool int);
11664 vector signed int vec_vsubuwm (vector signed int, vector signed int);
11665 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
11666 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
11667 vector unsigned int vec_vsubuwm (vector unsigned int,
11668 vector unsigned int);
11670 vector signed short vec_vsubuhm (vector bool short,
11671 vector signed short);
11672 vector signed short vec_vsubuhm (vector signed short,
11673 vector bool short);
11674 vector signed short vec_vsubuhm (vector signed short,
11675 vector signed short);
11676 vector unsigned short vec_vsubuhm (vector bool short,
11677 vector unsigned short);
11678 vector unsigned short vec_vsubuhm (vector unsigned short,
11679 vector bool short);
11680 vector unsigned short vec_vsubuhm (vector unsigned short,
11681 vector unsigned short);
11683 vector signed char vec_vsububm (vector bool char, vector signed char);
11684 vector signed char vec_vsububm (vector signed char, vector bool char);
11685 vector signed char vec_vsububm (vector signed char, vector signed char);
11686 vector unsigned char vec_vsububm (vector bool char,
11687 vector unsigned char);
11688 vector unsigned char vec_vsububm (vector unsigned char,
11690 vector unsigned char vec_vsububm (vector unsigned char,
11691 vector unsigned char);
11693 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
11695 vector unsigned char vec_subs (vector bool char, vector unsigned char);
11696 vector unsigned char vec_subs (vector unsigned char, vector bool char);
11697 vector unsigned char vec_subs (vector unsigned char,
11698 vector unsigned char);
11699 vector signed char vec_subs (vector bool char, vector signed char);
11700 vector signed char vec_subs (vector signed char, vector bool char);
11701 vector signed char vec_subs (vector signed char, vector signed char);
11702 vector unsigned short vec_subs (vector bool short,
11703 vector unsigned short);
11704 vector unsigned short vec_subs (vector unsigned short,
11705 vector bool short);
11706 vector unsigned short vec_subs (vector unsigned short,
11707 vector unsigned short);
11708 vector signed short vec_subs (vector bool short, vector signed short);
11709 vector signed short vec_subs (vector signed short, vector bool short);
11710 vector signed short vec_subs (vector signed short, vector signed short);
11711 vector unsigned int vec_subs (vector bool int, vector unsigned int);
11712 vector unsigned int vec_subs (vector unsigned int, vector bool int);
11713 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
11714 vector signed int vec_subs (vector bool int, vector signed int);
11715 vector signed int vec_subs (vector signed int, vector bool int);
11716 vector signed int vec_subs (vector signed int, vector signed int);
11718 vector signed int vec_vsubsws (vector bool int, vector signed int);
11719 vector signed int vec_vsubsws (vector signed int, vector bool int);
11720 vector signed int vec_vsubsws (vector signed int, vector signed int);
11722 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
11723 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
11724 vector unsigned int vec_vsubuws (vector unsigned int,
11725 vector unsigned int);
11727 vector signed short vec_vsubshs (vector bool short,
11728 vector signed short);
11729 vector signed short vec_vsubshs (vector signed short,
11730 vector bool short);
11731 vector signed short vec_vsubshs (vector signed short,
11732 vector signed short);
11734 vector unsigned short vec_vsubuhs (vector bool short,
11735 vector unsigned short);
11736 vector unsigned short vec_vsubuhs (vector unsigned short,
11737 vector bool short);
11738 vector unsigned short vec_vsubuhs (vector unsigned short,
11739 vector unsigned short);
11741 vector signed char vec_vsubsbs (vector bool char, vector signed char);
11742 vector signed char vec_vsubsbs (vector signed char, vector bool char);
11743 vector signed char vec_vsubsbs (vector signed char, vector signed char);
11745 vector unsigned char vec_vsububs (vector bool char,
11746 vector unsigned char);
11747 vector unsigned char vec_vsububs (vector unsigned char,
11749 vector unsigned char vec_vsububs (vector unsigned char,
11750 vector unsigned char);
11752 vector unsigned int vec_sum4s (vector unsigned char,
11753 vector unsigned int);
11754 vector signed int vec_sum4s (vector signed char, vector signed int);
11755 vector signed int vec_sum4s (vector signed short, vector signed int);
11757 vector signed int vec_vsum4shs (vector signed short, vector signed int);
11759 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
11761 vector unsigned int vec_vsum4ubs (vector unsigned char,
11762 vector unsigned int);
11764 vector signed int vec_sum2s (vector signed int, vector signed int);
11766 vector signed int vec_sums (vector signed int, vector signed int);
11768 vector float vec_trunc (vector float);
11770 vector signed short vec_unpackh (vector signed char);
11771 vector bool short vec_unpackh (vector bool char);
11772 vector signed int vec_unpackh (vector signed short);
11773 vector bool int vec_unpackh (vector bool short);
11774 vector unsigned int vec_unpackh (vector pixel);
11776 vector bool int vec_vupkhsh (vector bool short);
11777 vector signed int vec_vupkhsh (vector signed short);
11779 vector unsigned int vec_vupkhpx (vector pixel);
11781 vector bool short vec_vupkhsb (vector bool char);
11782 vector signed short vec_vupkhsb (vector signed char);
11784 vector signed short vec_unpackl (vector signed char);
11785 vector bool short vec_unpackl (vector bool char);
11786 vector unsigned int vec_unpackl (vector pixel);
11787 vector signed int vec_unpackl (vector signed short);
11788 vector bool int vec_unpackl (vector bool short);
11790 vector unsigned int vec_vupklpx (vector pixel);
11792 vector bool int vec_vupklsh (vector bool short);
11793 vector signed int vec_vupklsh (vector signed short);
11795 vector bool short vec_vupklsb (vector bool char);
11796 vector signed short vec_vupklsb (vector signed char);
11798 vector float vec_xor (vector float, vector float);
11799 vector float vec_xor (vector float, vector bool int);
11800 vector float vec_xor (vector bool int, vector float);
11801 vector bool int vec_xor (vector bool int, vector bool int);
11802 vector signed int vec_xor (vector bool int, vector signed int);
11803 vector signed int vec_xor (vector signed int, vector bool int);
11804 vector signed int vec_xor (vector signed int, vector signed int);
11805 vector unsigned int vec_xor (vector bool int, vector unsigned int);
11806 vector unsigned int vec_xor (vector unsigned int, vector bool int);
11807 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
11808 vector bool short vec_xor (vector bool short, vector bool short);
11809 vector signed short vec_xor (vector bool short, vector signed short);
11810 vector signed short vec_xor (vector signed short, vector bool short);
11811 vector signed short vec_xor (vector signed short, vector signed short);
11812 vector unsigned short vec_xor (vector bool short,
11813 vector unsigned short);
11814 vector unsigned short vec_xor (vector unsigned short,
11815 vector bool short);
11816 vector unsigned short vec_xor (vector unsigned short,
11817 vector unsigned short);
11818 vector signed char vec_xor (vector bool char, vector signed char);
11819 vector bool char vec_xor (vector bool char, vector bool char);
11820 vector signed char vec_xor (vector signed char, vector bool char);
11821 vector signed char vec_xor (vector signed char, vector signed char);
11822 vector unsigned char vec_xor (vector bool char, vector unsigned char);
11823 vector unsigned char vec_xor (vector unsigned char, vector bool char);
11824 vector unsigned char vec_xor (vector unsigned char,
11825 vector unsigned char);
11827 int vec_all_eq (vector signed char, vector bool char);
11828 int vec_all_eq (vector signed char, vector signed char);
11829 int vec_all_eq (vector unsigned char, vector bool char);
11830 int vec_all_eq (vector unsigned char, vector unsigned char);
11831 int vec_all_eq (vector bool char, vector bool char);
11832 int vec_all_eq (vector bool char, vector unsigned char);
11833 int vec_all_eq (vector bool char, vector signed char);
11834 int vec_all_eq (vector signed short, vector bool short);
11835 int vec_all_eq (vector signed short, vector signed short);
11836 int vec_all_eq (vector unsigned short, vector bool short);
11837 int vec_all_eq (vector unsigned short, vector unsigned short);
11838 int vec_all_eq (vector bool short, vector bool short);
11839 int vec_all_eq (vector bool short, vector unsigned short);
11840 int vec_all_eq (vector bool short, vector signed short);
11841 int vec_all_eq (vector pixel, vector pixel);
11842 int vec_all_eq (vector signed int, vector bool int);
11843 int vec_all_eq (vector signed int, vector signed int);
11844 int vec_all_eq (vector unsigned int, vector bool int);
11845 int vec_all_eq (vector unsigned int, vector unsigned int);
11846 int vec_all_eq (vector bool int, vector bool int);
11847 int vec_all_eq (vector bool int, vector unsigned int);
11848 int vec_all_eq (vector bool int, vector signed int);
11849 int vec_all_eq (vector float, vector float);
11851 int vec_all_ge (vector bool char, vector unsigned char);
11852 int vec_all_ge (vector unsigned char, vector bool char);
11853 int vec_all_ge (vector unsigned char, vector unsigned char);
11854 int vec_all_ge (vector bool char, vector signed char);
11855 int vec_all_ge (vector signed char, vector bool char);
11856 int vec_all_ge (vector signed char, vector signed char);
11857 int vec_all_ge (vector bool short, vector unsigned short);
11858 int vec_all_ge (vector unsigned short, vector bool short);
11859 int vec_all_ge (vector unsigned short, vector unsigned short);
11860 int vec_all_ge (vector signed short, vector signed short);
11861 int vec_all_ge (vector bool short, vector signed short);
11862 int vec_all_ge (vector signed short, vector bool short);
11863 int vec_all_ge (vector bool int, vector unsigned int);
11864 int vec_all_ge (vector unsigned int, vector bool int);
11865 int vec_all_ge (vector unsigned int, vector unsigned int);
11866 int vec_all_ge (vector bool int, vector signed int);
11867 int vec_all_ge (vector signed int, vector bool int);
11868 int vec_all_ge (vector signed int, vector signed int);
11869 int vec_all_ge (vector float, vector float);
11871 int vec_all_gt (vector bool char, vector unsigned char);
11872 int vec_all_gt (vector unsigned char, vector bool char);
11873 int vec_all_gt (vector unsigned char, vector unsigned char);
11874 int vec_all_gt (vector bool char, vector signed char);
11875 int vec_all_gt (vector signed char, vector bool char);
11876 int vec_all_gt (vector signed char, vector signed char);
11877 int vec_all_gt (vector bool short, vector unsigned short);
11878 int vec_all_gt (vector unsigned short, vector bool short);
11879 int vec_all_gt (vector unsigned short, vector unsigned short);
11880 int vec_all_gt (vector bool short, vector signed short);
11881 int vec_all_gt (vector signed short, vector bool short);
11882 int vec_all_gt (vector signed short, vector signed short);
11883 int vec_all_gt (vector bool int, vector unsigned int);
11884 int vec_all_gt (vector unsigned int, vector bool int);
11885 int vec_all_gt (vector unsigned int, vector unsigned int);
11886 int vec_all_gt (vector bool int, vector signed int);
11887 int vec_all_gt (vector signed int, vector bool int);
11888 int vec_all_gt (vector signed int, vector signed int);
11889 int vec_all_gt (vector float, vector float);
11891 int vec_all_in (vector float, vector float);
11893 int vec_all_le (vector bool char, vector unsigned char);
11894 int vec_all_le (vector unsigned char, vector bool char);
11895 int vec_all_le (vector unsigned char, vector unsigned char);
11896 int vec_all_le (vector bool char, vector signed char);
11897 int vec_all_le (vector signed char, vector bool char);
11898 int vec_all_le (vector signed char, vector signed char);
11899 int vec_all_le (vector bool short, vector unsigned short);
11900 int vec_all_le (vector unsigned short, vector bool short);
11901 int vec_all_le (vector unsigned short, vector unsigned short);
11902 int vec_all_le (vector bool short, vector signed short);
11903 int vec_all_le (vector signed short, vector bool short);
11904 int vec_all_le (vector signed short, vector signed short);
11905 int vec_all_le (vector bool int, vector unsigned int);
11906 int vec_all_le (vector unsigned int, vector bool int);
11907 int vec_all_le (vector unsigned int, vector unsigned int);
11908 int vec_all_le (vector bool int, vector signed int);
11909 int vec_all_le (vector signed int, vector bool int);
11910 int vec_all_le (vector signed int, vector signed int);
11911 int vec_all_le (vector float, vector float);
11913 int vec_all_lt (vector bool char, vector unsigned char);
11914 int vec_all_lt (vector unsigned char, vector bool char);
11915 int vec_all_lt (vector unsigned char, vector unsigned char);
11916 int vec_all_lt (vector bool char, vector signed char);
11917 int vec_all_lt (vector signed char, vector bool char);
11918 int vec_all_lt (vector signed char, vector signed char);
11919 int vec_all_lt (vector bool short, vector unsigned short);
11920 int vec_all_lt (vector unsigned short, vector bool short);
11921 int vec_all_lt (vector unsigned short, vector unsigned short);
11922 int vec_all_lt (vector bool short, vector signed short);
11923 int vec_all_lt (vector signed short, vector bool short);
11924 int vec_all_lt (vector signed short, vector signed short);
11925 int vec_all_lt (vector bool int, vector unsigned int);
11926 int vec_all_lt (vector unsigned int, vector bool int);
11927 int vec_all_lt (vector unsigned int, vector unsigned int);
11928 int vec_all_lt (vector bool int, vector signed int);
11929 int vec_all_lt (vector signed int, vector bool int);
11930 int vec_all_lt (vector signed int, vector signed int);
11931 int vec_all_lt (vector float, vector float);
11933 int vec_all_nan (vector float);
11935 int vec_all_ne (vector signed char, vector bool char);
11936 int vec_all_ne (vector signed char, vector signed char);
11937 int vec_all_ne (vector unsigned char, vector bool char);
11938 int vec_all_ne (vector unsigned char, vector unsigned char);
11939 int vec_all_ne (vector bool char, vector bool char);
11940 int vec_all_ne (vector bool char, vector unsigned char);
11941 int vec_all_ne (vector bool char, vector signed char);
11942 int vec_all_ne (vector signed short, vector bool short);
11943 int vec_all_ne (vector signed short, vector signed short);
11944 int vec_all_ne (vector unsigned short, vector bool short);
11945 int vec_all_ne (vector unsigned short, vector unsigned short);
11946 int vec_all_ne (vector bool short, vector bool short);
11947 int vec_all_ne (vector bool short, vector unsigned short);
11948 int vec_all_ne (vector bool short, vector signed short);
11949 int vec_all_ne (vector pixel, vector pixel);
11950 int vec_all_ne (vector signed int, vector bool int);
11951 int vec_all_ne (vector signed int, vector signed int);
11952 int vec_all_ne (vector unsigned int, vector bool int);
11953 int vec_all_ne (vector unsigned int, vector unsigned int);
11954 int vec_all_ne (vector bool int, vector bool int);
11955 int vec_all_ne (vector bool int, vector unsigned int);
11956 int vec_all_ne (vector bool int, vector signed int);
11957 int vec_all_ne (vector float, vector float);
11959 int vec_all_nge (vector float, vector float);
11961 int vec_all_ngt (vector float, vector float);
11963 int vec_all_nle (vector float, vector float);
11965 int vec_all_nlt (vector float, vector float);
11967 int vec_all_numeric (vector float);
11969 int vec_any_eq (vector signed char, vector bool char);
11970 int vec_any_eq (vector signed char, vector signed char);
11971 int vec_any_eq (vector unsigned char, vector bool char);
11972 int vec_any_eq (vector unsigned char, vector unsigned char);
11973 int vec_any_eq (vector bool char, vector bool char);
11974 int vec_any_eq (vector bool char, vector unsigned char);
11975 int vec_any_eq (vector bool char, vector signed char);
11976 int vec_any_eq (vector signed short, vector bool short);
11977 int vec_any_eq (vector signed short, vector signed short);
11978 int vec_any_eq (vector unsigned short, vector bool short);
11979 int vec_any_eq (vector unsigned short, vector unsigned short);
11980 int vec_any_eq (vector bool short, vector bool short);
11981 int vec_any_eq (vector bool short, vector unsigned short);
11982 int vec_any_eq (vector bool short, vector signed short);
11983 int vec_any_eq (vector pixel, vector pixel);
11984 int vec_any_eq (vector signed int, vector bool int);
11985 int vec_any_eq (vector signed int, vector signed int);
11986 int vec_any_eq (vector unsigned int, vector bool int);
11987 int vec_any_eq (vector unsigned int, vector unsigned int);
11988 int vec_any_eq (vector bool int, vector bool int);
11989 int vec_any_eq (vector bool int, vector unsigned int);
11990 int vec_any_eq (vector bool int, vector signed int);
11991 int vec_any_eq (vector float, vector float);
11993 int vec_any_ge (vector signed char, vector bool char);
11994 int vec_any_ge (vector unsigned char, vector bool char);
11995 int vec_any_ge (vector unsigned char, vector unsigned char);
11996 int vec_any_ge (vector signed char, vector signed char);
11997 int vec_any_ge (vector bool char, vector unsigned char);
11998 int vec_any_ge (vector bool char, vector signed char);
11999 int vec_any_ge (vector unsigned short, vector bool short);
12000 int vec_any_ge (vector unsigned short, vector unsigned short);
12001 int vec_any_ge (vector signed short, vector signed short);
12002 int vec_any_ge (vector signed short, vector bool short);
12003 int vec_any_ge (vector bool short, vector unsigned short);
12004 int vec_any_ge (vector bool short, vector signed short);
12005 int vec_any_ge (vector signed int, vector bool int);
12006 int vec_any_ge (vector unsigned int, vector bool int);
12007 int vec_any_ge (vector unsigned int, vector unsigned int);
12008 int vec_any_ge (vector signed int, vector signed int);
12009 int vec_any_ge (vector bool int, vector unsigned int);
12010 int vec_any_ge (vector bool int, vector signed int);
12011 int vec_any_ge (vector float, vector float);
12013 int vec_any_gt (vector bool char, vector unsigned char);
12014 int vec_any_gt (vector unsigned char, vector bool char);
12015 int vec_any_gt (vector unsigned char, vector unsigned char);
12016 int vec_any_gt (vector bool char, vector signed char);
12017 int vec_any_gt (vector signed char, vector bool char);
12018 int vec_any_gt (vector signed char, vector signed char);
12019 int vec_any_gt (vector bool short, vector unsigned short);
12020 int vec_any_gt (vector unsigned short, vector bool short);
12021 int vec_any_gt (vector unsigned short, vector unsigned short);
12022 int vec_any_gt (vector bool short, vector signed short);
12023 int vec_any_gt (vector signed short, vector bool short);
12024 int vec_any_gt (vector signed short, vector signed short);
12025 int vec_any_gt (vector bool int, vector unsigned int);
12026 int vec_any_gt (vector unsigned int, vector bool int);
12027 int vec_any_gt (vector unsigned int, vector unsigned int);
12028 int vec_any_gt (vector bool int, vector signed int);
12029 int vec_any_gt (vector signed int, vector bool int);
12030 int vec_any_gt (vector signed int, vector signed int);
12031 int vec_any_gt (vector float, vector float);
12033 int vec_any_le (vector bool char, vector unsigned char);
12034 int vec_any_le (vector unsigned char, vector bool char);
12035 int vec_any_le (vector unsigned char, vector unsigned char);
12036 int vec_any_le (vector bool char, vector signed char);
12037 int vec_any_le (vector signed char, vector bool char);
12038 int vec_any_le (vector signed char, vector signed char);
12039 int vec_any_le (vector bool short, vector unsigned short);
12040 int vec_any_le (vector unsigned short, vector bool short);
12041 int vec_any_le (vector unsigned short, vector unsigned short);
12042 int vec_any_le (vector bool short, vector signed short);
12043 int vec_any_le (vector signed short, vector bool short);
12044 int vec_any_le (vector signed short, vector signed short);
12045 int vec_any_le (vector bool int, vector unsigned int);
12046 int vec_any_le (vector unsigned int, vector bool int);
12047 int vec_any_le (vector unsigned int, vector unsigned int);
12048 int vec_any_le (vector bool int, vector signed int);
12049 int vec_any_le (vector signed int, vector bool int);
12050 int vec_any_le (vector signed int, vector signed int);
12051 int vec_any_le (vector float, vector float);
12053 int vec_any_lt (vector bool char, vector unsigned char);
12054 int vec_any_lt (vector unsigned char, vector bool char);
12055 int vec_any_lt (vector unsigned char, vector unsigned char);
12056 int vec_any_lt (vector bool char, vector signed char);
12057 int vec_any_lt (vector signed char, vector bool char);
12058 int vec_any_lt (vector signed char, vector signed char);
12059 int vec_any_lt (vector bool short, vector unsigned short);
12060 int vec_any_lt (vector unsigned short, vector bool short);
12061 int vec_any_lt (vector unsigned short, vector unsigned short);
12062 int vec_any_lt (vector bool short, vector signed short);
12063 int vec_any_lt (vector signed short, vector bool short);
12064 int vec_any_lt (vector signed short, vector signed short);
12065 int vec_any_lt (vector bool int, vector unsigned int);
12066 int vec_any_lt (vector unsigned int, vector bool int);
12067 int vec_any_lt (vector unsigned int, vector unsigned int);
12068 int vec_any_lt (vector bool int, vector signed int);
12069 int vec_any_lt (vector signed int, vector bool int);
12070 int vec_any_lt (vector signed int, vector signed int);
12071 int vec_any_lt (vector float, vector float);
12073 int vec_any_nan (vector float);
12075 int vec_any_ne (vector signed char, vector bool char);
12076 int vec_any_ne (vector signed char, vector signed char);
12077 int vec_any_ne (vector unsigned char, vector bool char);
12078 int vec_any_ne (vector unsigned char, vector unsigned char);
12079 int vec_any_ne (vector bool char, vector bool char);
12080 int vec_any_ne (vector bool char, vector unsigned char);
12081 int vec_any_ne (vector bool char, vector signed char);
12082 int vec_any_ne (vector signed short, vector bool short);
12083 int vec_any_ne (vector signed short, vector signed short);
12084 int vec_any_ne (vector unsigned short, vector bool short);
12085 int vec_any_ne (vector unsigned short, vector unsigned short);
12086 int vec_any_ne (vector bool short, vector bool short);
12087 int vec_any_ne (vector bool short, vector unsigned short);
12088 int vec_any_ne (vector bool short, vector signed short);
12089 int vec_any_ne (vector pixel, vector pixel);
12090 int vec_any_ne (vector signed int, vector bool int);
12091 int vec_any_ne (vector signed int, vector signed int);
12092 int vec_any_ne (vector unsigned int, vector bool int);
12093 int vec_any_ne (vector unsigned int, vector unsigned int);
12094 int vec_any_ne (vector bool int, vector bool int);
12095 int vec_any_ne (vector bool int, vector unsigned int);
12096 int vec_any_ne (vector bool int, vector signed int);
12097 int vec_any_ne (vector float, vector float);
12099 int vec_any_nge (vector float, vector float);
12101 int vec_any_ngt (vector float, vector float);
12103 int vec_any_nle (vector float, vector float);
12105 int vec_any_nlt (vector float, vector float);
12107 int vec_any_numeric (vector float);
12109 int vec_any_out (vector float, vector float);
12112 If the vector/scalar (VSX) instruction set is available, the following
12113 additional functions are available:
12116 vector double vec_abs (vector double);
12117 vector double vec_add (vector double, vector double);
12118 vector double vec_and (vector double, vector double);
12119 vector double vec_and (vector double, vector bool long);
12120 vector double vec_and (vector bool long, vector double);
12121 vector double vec_andc (vector double, vector double);
12122 vector double vec_andc (vector double, vector bool long);
12123 vector double vec_andc (vector bool long, vector double);
12124 vector double vec_ceil (vector double);
12125 vector bool long vec_cmpeq (vector double, vector double);
12126 vector bool long vec_cmpge (vector double, vector double);
12127 vector bool long vec_cmpgt (vector double, vector double);
12128 vector bool long vec_cmple (vector double, vector double);
12129 vector bool long vec_cmplt (vector double, vector double);
12130 vector float vec_div (vector float, vector float);
12131 vector double vec_div (vector double, vector double);
12132 vector double vec_floor (vector double);
12133 vector double vec_madd (vector double, vector double, vector double);
12134 vector double vec_max (vector double, vector double);
12135 vector double vec_min (vector double, vector double);
12136 vector float vec_msub (vector float, vector float, vector float);
12137 vector double vec_msub (vector double, vector double, vector double);
12138 vector float vec_mul (vector float, vector float);
12139 vector double vec_mul (vector double, vector double);
12140 vector float vec_nearbyint (vector float);
12141 vector double vec_nearbyint (vector double);
12142 vector float vec_nmadd (vector float, vector float, vector float);
12143 vector double vec_nmadd (vector double, vector double, vector double);
12144 vector double vec_nmsub (vector double, vector double, vector double);
12145 vector double vec_nor (vector double, vector double);
12146 vector double vec_or (vector double, vector double);
12147 vector double vec_or (vector double, vector bool long);
12148 vector double vec_or (vector bool long, vector double);
12149 vector double vec_perm (vector double,
12151 vector unsigned char);
12152 vector double vec_rint (vector double);
12153 vector double vec_recip (vector double, vector double);
12154 vector double vec_rsqrt (vector double);
12155 vector double vec_rsqrte (vector double);
12156 vector double vec_sel (vector double, vector double, vector bool long);
12157 vector double vec_sel (vector double, vector double, vector unsigned long);
12158 vector double vec_sub (vector double, vector double);
12159 vector float vec_sqrt (vector float);
12160 vector double vec_sqrt (vector double);
12161 vector double vec_trunc (vector double);
12162 vector double vec_xor (vector double, vector double);
12163 vector double vec_xor (vector double, vector bool long);
12164 vector double vec_xor (vector bool long, vector double);
12165 int vec_all_eq (vector double, vector double);
12166 int vec_all_ge (vector double, vector double);
12167 int vec_all_gt (vector double, vector double);
12168 int vec_all_le (vector double, vector double);
12169 int vec_all_lt (vector double, vector double);
12170 int vec_all_nan (vector double);
12171 int vec_all_ne (vector double, vector double);
12172 int vec_all_nge (vector double, vector double);
12173 int vec_all_ngt (vector double, vector double);
12174 int vec_all_nle (vector double, vector double);
12175 int vec_all_nlt (vector double, vector double);
12176 int vec_all_numeric (vector double);
12177 int vec_any_eq (vector double, vector double);
12178 int vec_any_ge (vector double, vector double);
12179 int vec_any_gt (vector double, vector double);
12180 int vec_any_le (vector double, vector double);
12181 int vec_any_lt (vector double, vector double);
12182 int vec_any_nan (vector double);
12183 int vec_any_ne (vector double, vector double);
12184 int vec_any_nge (vector double, vector double);
12185 int vec_any_ngt (vector double, vector double);
12186 int vec_any_nle (vector double, vector double);
12187 int vec_any_nlt (vector double, vector double);
12188 int vec_any_numeric (vector double);
12191 GCC provides a few other builtins on Powerpc to access certain instructions:
12193 float __builtin_recipdivf (float, float);
12194 float __builtin_rsqrtf (float);
12195 double __builtin_recipdiv (double, double);
12196 double __builtin_rsqrt (double);
12197 long __builtin_bpermd (long, long);
12198 int __builtin_bswap16 (int);
12201 The @code{vec_rsqrt}, @code{__builtin_rsqrt}, and
12202 @code{__builtin_rsqrtf} functions generate multiple instructions to
12203 implement the reciprocal sqrt functionality using reciprocal sqrt
12204 estimate instructions.
12206 The @code{__builtin_recipdiv}, and @code{__builtin_recipdivf}
12207 functions generate multiple instructions to implement division using
12208 the reciprocal estimate instructions.
12210 @node RX Built-in Functions
12211 @subsection RX Built-in Functions
12212 GCC supports some of the RX instructions which cannot be expressed in
12213 the C programming language via the use of built-in functions. The
12214 following functions are supported:
12216 @deftypefn {Built-in Function} void __builtin_rx_brk (void)
12217 Generates the @code{brk} machine instruction.
12220 @deftypefn {Built-in Function} void __builtin_rx_clrpsw (int)
12221 Generates the @code{clrpsw} machine instruction to clear the specified
12222 bit in the processor status word.
12225 @deftypefn {Built-in Function} void __builtin_rx_int (int)
12226 Generates the @code{int} machine instruction to generate an interrupt
12227 with the specified value.
12230 @deftypefn {Built-in Function} void __builtin_rx_machi (int, int)
12231 Generates the @code{machi} machine instruction to add the result of
12232 multiplying the top 16-bits of the two arguments into the
12236 @deftypefn {Built-in Function} void __builtin_rx_maclo (int, int)
12237 Generates the @code{maclo} machine instruction to add the result of
12238 multiplying the bottom 16-bits of the two arguments into the
12242 @deftypefn {Built-in Function} void __builtin_rx_mulhi (int, int)
12243 Generates the @code{mulhi} machine instruction to place the result of
12244 multiplying the top 16-bits of the two arguments into the
12248 @deftypefn {Built-in Function} void __builtin_rx_mullo (int, int)
12249 Generates the @code{mullo} machine instruction to place the result of
12250 multiplying the bottom 16-bits of the two arguments into the
12254 @deftypefn {Built-in Function} int __builtin_rx_mvfachi (void)
12255 Generates the @code{mvfachi} machine instruction to read the top
12256 32-bits of the accumulator.
12259 @deftypefn {Built-in Function} int __builtin_rx_mvfacmi (void)
12260 Generates the @code{mvfacmi} machine instruction to read the middle
12261 32-bits of the accumulator.
12264 @deftypefn {Built-in Function} int __builtin_rx_mvfc (int)
12265 Generates the @code{mvfc} machine instruction which reads the control
12266 register specified in its argument and returns its value.
12269 @deftypefn {Built-in Function} void __builtin_rx_mvtachi (int)
12270 Generates the @code{mvtachi} machine instruction to set the top
12271 32-bits of the accumulator.
12274 @deftypefn {Built-in Function} void __builtin_rx_mvtaclo (int)
12275 Generates the @code{mvtaclo} machine instruction to set the bottom
12276 32-bits of the accumulator.
12279 @deftypefn {Built-in Function} void __builtin_rx_mvtc (int reg, int val)
12280 Generates the @code{mvtc} machine instruction which sets control
12281 register number @code{reg} to @code{val}.
12284 @deftypefn {Built-in Function} void __builtin_rx_mvtipl (int)
12285 Generates the @code{mvtipl} machine instruction set the interrupt
12289 @deftypefn {Built-in Function} void __builtin_rx_racw (int)
12290 Generates the @code{racw} machine instruction to round the accumulator
12291 according to the specified mode.
12294 @deftypefn {Built-in Function} int __builtin_rx_revw (int)
12295 Generates the @code{revw} machine instruction which swaps the bytes in
12296 the argument so that bits 0--7 now occupy bits 8--15 and vice versa,
12297 and also bits 16--23 occupy bits 24--31 and vice versa.
12300 @deftypefn {Built-in Function} void __builtin_rx_rmpa (void)
12301 Generates the @code{rmpa} machine instruction which initiates a
12302 repeated multiply and accumulate sequence.
12305 @deftypefn {Built-in Function} void __builtin_rx_round (float)
12306 Generates the @code{round} machine instruction which returns the
12307 floating point argument rounded according to the current rounding mode
12308 set in the floating point status word register.
12311 @deftypefn {Built-in Function} int __builtin_rx_sat (int)
12312 Generates the @code{sat} machine instruction which returns the
12313 saturated value of the argument.
12316 @deftypefn {Built-in Function} void __builtin_rx_setpsw (int)
12317 Generates the @code{setpsw} machine instruction to set the specified
12318 bit in the processor status word.
12321 @deftypefn {Built-in Function} void __builtin_rx_wait (void)
12322 Generates the @code{wait} machine instruction.
12325 @node SPARC VIS Built-in Functions
12326 @subsection SPARC VIS Built-in Functions
12328 GCC supports SIMD operations on the SPARC using both the generic vector
12329 extensions (@pxref{Vector Extensions}) as well as built-in functions for
12330 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
12331 switch, the VIS extension is exposed as the following built-in functions:
12334 typedef int v2si __attribute__ ((vector_size (8)));
12335 typedef short v4hi __attribute__ ((vector_size (8)));
12336 typedef short v2hi __attribute__ ((vector_size (4)));
12337 typedef char v8qi __attribute__ ((vector_size (8)));
12338 typedef char v4qi __attribute__ ((vector_size (4)));
12340 void * __builtin_vis_alignaddr (void *, long);
12341 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
12342 v2si __builtin_vis_faligndatav2si (v2si, v2si);
12343 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
12344 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
12346 v4hi __builtin_vis_fexpand (v4qi);
12348 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
12349 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
12350 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
12351 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
12352 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
12353 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
12354 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
12356 v4qi __builtin_vis_fpack16 (v4hi);
12357 v8qi __builtin_vis_fpack32 (v2si, v2si);
12358 v2hi __builtin_vis_fpackfix (v2si);
12359 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
12361 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
12364 @node SPU Built-in Functions
12365 @subsection SPU Built-in Functions
12367 GCC provides extensions for the SPU processor as described in the
12368 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
12369 found at @uref{http://cell.scei.co.jp/} or
12370 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
12371 implementation differs in several ways.
12376 The optional extension of specifying vector constants in parentheses is
12380 A vector initializer requires no cast if the vector constant is of the
12381 same type as the variable it is initializing.
12384 If @code{signed} or @code{unsigned} is omitted, the signedness of the
12385 vector type is the default signedness of the base type. The default
12386 varies depending on the operating system, so a portable program should
12387 always specify the signedness.
12390 By default, the keyword @code{__vector} is added. The macro
12391 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
12395 GCC allows using a @code{typedef} name as the type specifier for a
12399 For C, overloaded functions are implemented with macros so the following
12403 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
12406 Since @code{spu_add} is a macro, the vector constant in the example
12407 is treated as four separate arguments. Wrap the entire argument in
12408 parentheses for this to work.
12411 The extended version of @code{__builtin_expect} is not supported.
12415 @emph{Note:} Only the interface described in the aforementioned
12416 specification is supported. Internally, GCC uses built-in functions to
12417 implement the required functionality, but these are not supported and
12418 are subject to change without notice.
12420 @node Target Format Checks
12421 @section Format Checks Specific to Particular Target Machines
12423 For some target machines, GCC supports additional options to the
12425 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
12428 * Solaris Format Checks::
12431 @node Solaris Format Checks
12432 @subsection Solaris Format Checks
12434 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
12435 check. @code{cmn_err} accepts a subset of the standard @code{printf}
12436 conversions, and the two-argument @code{%b} conversion for displaying
12437 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
12440 @section Pragmas Accepted by GCC
12442 @cindex @code{#pragma}
12444 GCC supports several types of pragmas, primarily in order to compile
12445 code originally written for other compilers. Note that in general
12446 we do not recommend the use of pragmas; @xref{Function Attributes},
12447 for further explanation.
12453 * RS/6000 and PowerPC Pragmas::
12455 * Solaris Pragmas::
12456 * Symbol-Renaming Pragmas::
12457 * Structure-Packing Pragmas::
12459 * Diagnostic Pragmas::
12460 * Visibility Pragmas::
12461 * Push/Pop Macro Pragmas::
12462 * Function Specific Option Pragmas::
12466 @subsection ARM Pragmas
12468 The ARM target defines pragmas for controlling the default addition of
12469 @code{long_call} and @code{short_call} attributes to functions.
12470 @xref{Function Attributes}, for information about the effects of these
12475 @cindex pragma, long_calls
12476 Set all subsequent functions to have the @code{long_call} attribute.
12478 @item no_long_calls
12479 @cindex pragma, no_long_calls
12480 Set all subsequent functions to have the @code{short_call} attribute.
12482 @item long_calls_off
12483 @cindex pragma, long_calls_off
12484 Do not affect the @code{long_call} or @code{short_call} attributes of
12485 subsequent functions.
12489 @subsection M32C Pragmas
12492 @item GCC memregs @var{number}
12493 @cindex pragma, memregs
12494 Overrides the command-line option @code{-memregs=} for the current
12495 file. Use with care! This pragma must be before any function in the
12496 file, and mixing different memregs values in different objects may
12497 make them incompatible. This pragma is useful when a
12498 performance-critical function uses a memreg for temporary values,
12499 as it may allow you to reduce the number of memregs used.
12501 @item ADDRESS @var{name} @var{address}
12502 @cindex pragma, address
12503 For any declared symbols matching @var{name}, this does three things
12504 to that symbol: it forces the symbol to be located at the given
12505 address (a number), it forces the symbol to be volatile, and it
12506 changes the symbol's scope to be static. This pragma exists for
12507 compatibility with other compilers, but note that the common
12508 @code{1234H} numeric syntax is not supported (use @code{0x1234}
12512 #pragma ADDRESS port3 0x103
12519 @subsection MeP Pragmas
12523 @item custom io_volatile (on|off)
12524 @cindex pragma, custom io_volatile
12525 Overrides the command line option @code{-mio-volatile} for the current
12526 file. Note that for compatibility with future GCC releases, this
12527 option should only be used once before any @code{io} variables in each
12530 @item GCC coprocessor available @var{registers}
12531 @cindex pragma, coprocessor available
12532 Specifies which coprocessor registers are available to the register
12533 allocator. @var{registers} may be a single register, register range
12534 separated by ellipses, or comma-separated list of those. Example:
12537 #pragma GCC coprocessor available $c0...$c10, $c28
12540 @item GCC coprocessor call_saved @var{registers}
12541 @cindex pragma, coprocessor call_saved
12542 Specifies which coprocessor registers are to be saved and restored by
12543 any function using them. @var{registers} may be a single register,
12544 register range separated by ellipses, or comma-separated list of
12548 #pragma GCC coprocessor call_saved $c4...$c6, $c31
12551 @item GCC coprocessor subclass '(A|B|C|D)' = @var{registers}
12552 @cindex pragma, coprocessor subclass
12553 Creates and defines a register class. These register classes can be
12554 used by inline @code{asm} constructs. @var{registers} may be a single
12555 register, register range separated by ellipses, or comma-separated
12556 list of those. Example:
12559 #pragma GCC coprocessor subclass 'B' = $c2, $c4, $c6
12561 asm ("cpfoo %0" : "=B" (x));
12564 @item GCC disinterrupt @var{name} , @var{name} @dots{}
12565 @cindex pragma, disinterrupt
12566 For the named functions, the compiler adds code to disable interrupts
12567 for the duration of those functions. Any functions so named, which
12568 are not encountered in the source, cause a warning that the pragma was
12569 not used. Examples:
12572 #pragma disinterrupt foo
12573 #pragma disinterrupt bar, grill
12574 int foo () @{ @dots{} @}
12577 @item GCC call @var{name} , @var{name} @dots{}
12578 @cindex pragma, call
12579 For the named functions, the compiler always uses a register-indirect
12580 call model when calling the named functions. Examples:
12589 @node RS/6000 and PowerPC Pragmas
12590 @subsection RS/6000 and PowerPC Pragmas
12592 The RS/6000 and PowerPC targets define one pragma for controlling
12593 whether or not the @code{longcall} attribute is added to function
12594 declarations by default. This pragma overrides the @option{-mlongcall}
12595 option, but not the @code{longcall} and @code{shortcall} attributes.
12596 @xref{RS/6000 and PowerPC Options}, for more information about when long
12597 calls are and are not necessary.
12601 @cindex pragma, longcall
12602 Apply the @code{longcall} attribute to all subsequent function
12606 Do not apply the @code{longcall} attribute to subsequent function
12610 @c Describe h8300 pragmas here.
12611 @c Describe sh pragmas here.
12612 @c Describe v850 pragmas here.
12614 @node Darwin Pragmas
12615 @subsection Darwin Pragmas
12617 The following pragmas are available for all architectures running the
12618 Darwin operating system. These are useful for compatibility with other
12622 @item mark @var{tokens}@dots{}
12623 @cindex pragma, mark
12624 This pragma is accepted, but has no effect.
12626 @item options align=@var{alignment}
12627 @cindex pragma, options align
12628 This pragma sets the alignment of fields in structures. The values of
12629 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
12630 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
12631 properly; to restore the previous setting, use @code{reset} for the
12634 @item segment @var{tokens}@dots{}
12635 @cindex pragma, segment
12636 This pragma is accepted, but has no effect.
12638 @item unused (@var{var} [, @var{var}]@dots{})
12639 @cindex pragma, unused
12640 This pragma declares variables to be possibly unused. GCC will not
12641 produce warnings for the listed variables. The effect is similar to
12642 that of the @code{unused} attribute, except that this pragma may appear
12643 anywhere within the variables' scopes.
12646 @node Solaris Pragmas
12647 @subsection Solaris Pragmas
12649 The Solaris target supports @code{#pragma redefine_extname}
12650 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
12651 @code{#pragma} directives for compatibility with the system compiler.
12654 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
12655 @cindex pragma, align
12657 Increase the minimum alignment of each @var{variable} to @var{alignment}.
12658 This is the same as GCC's @code{aligned} attribute @pxref{Variable
12659 Attributes}). Macro expansion occurs on the arguments to this pragma
12660 when compiling C and Objective-C@. It does not currently occur when
12661 compiling C++, but this is a bug which may be fixed in a future
12664 @item fini (@var{function} [, @var{function}]...)
12665 @cindex pragma, fini
12667 This pragma causes each listed @var{function} to be called after
12668 main, or during shared module unloading, by adding a call to the
12669 @code{.fini} section.
12671 @item init (@var{function} [, @var{function}]...)
12672 @cindex pragma, init
12674 This pragma causes each listed @var{function} to be called during
12675 initialization (before @code{main}) or during shared module loading, by
12676 adding a call to the @code{.init} section.
12680 @node Symbol-Renaming Pragmas
12681 @subsection Symbol-Renaming Pragmas
12683 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
12684 supports two @code{#pragma} directives which change the name used in
12685 assembly for a given declaration. @code{#pragma extern_prefix} is only
12686 available on platforms whose system headers need it. To get this effect
12687 on all platforms supported by GCC, use the asm labels extension (@pxref{Asm
12691 @item redefine_extname @var{oldname} @var{newname}
12692 @cindex pragma, redefine_extname
12694 This pragma gives the C function @var{oldname} the assembly symbol
12695 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
12696 will be defined if this pragma is available (currently on all platforms).
12698 @item extern_prefix @var{string}
12699 @cindex pragma, extern_prefix
12701 This pragma causes all subsequent external function and variable
12702 declarations to have @var{string} prepended to their assembly symbols.
12703 This effect may be terminated with another @code{extern_prefix} pragma
12704 whose argument is an empty string. The preprocessor macro
12705 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
12706 available (currently only on Tru64 UNIX)@.
12709 These pragmas and the asm labels extension interact in a complicated
12710 manner. Here are some corner cases you may want to be aware of.
12713 @item Both pragmas silently apply only to declarations with external
12714 linkage. Asm labels do not have this restriction.
12716 @item In C++, both pragmas silently apply only to declarations with
12717 ``C'' linkage. Again, asm labels do not have this restriction.
12719 @item If any of the three ways of changing the assembly name of a
12720 declaration is applied to a declaration whose assembly name has
12721 already been determined (either by a previous use of one of these
12722 features, or because the compiler needed the assembly name in order to
12723 generate code), and the new name is different, a warning issues and
12724 the name does not change.
12726 @item The @var{oldname} used by @code{#pragma redefine_extname} is
12727 always the C-language name.
12729 @item If @code{#pragma extern_prefix} is in effect, and a declaration
12730 occurs with an asm label attached, the prefix is silently ignored for
12733 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
12734 apply to the same declaration, whichever triggered first wins, and a
12735 warning issues if they contradict each other. (We would like to have
12736 @code{#pragma redefine_extname} always win, for consistency with asm
12737 labels, but if @code{#pragma extern_prefix} triggers first we have no
12738 way of knowing that that happened.)
12741 @node Structure-Packing Pragmas
12742 @subsection Structure-Packing Pragmas
12744 For compatibility with Microsoft Windows compilers, GCC supports a
12745 set of @code{#pragma} directives which change the maximum alignment of
12746 members of structures (other than zero-width bitfields), unions, and
12747 classes subsequently defined. The @var{n} value below always is required
12748 to be a small power of two and specifies the new alignment in bytes.
12751 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
12752 @item @code{#pragma pack()} sets the alignment to the one that was in
12753 effect when compilation started (see also command-line option
12754 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
12755 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
12756 setting on an internal stack and then optionally sets the new alignment.
12757 @item @code{#pragma pack(pop)} restores the alignment setting to the one
12758 saved at the top of the internal stack (and removes that stack entry).
12759 Note that @code{#pragma pack([@var{n}])} does not influence this internal
12760 stack; thus it is possible to have @code{#pragma pack(push)} followed by
12761 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
12762 @code{#pragma pack(pop)}.
12765 Some targets, e.g.@: i386 and powerpc, support the @code{ms_struct}
12766 @code{#pragma} which lays out a structure as the documented
12767 @code{__attribute__ ((ms_struct))}.
12769 @item @code{#pragma ms_struct on} turns on the layout for structures
12771 @item @code{#pragma ms_struct off} turns off the layout for structures
12773 @item @code{#pragma ms_struct reset} goes back to the default layout.
12777 @subsection Weak Pragmas
12779 For compatibility with SVR4, GCC supports a set of @code{#pragma}
12780 directives for declaring symbols to be weak, and defining weak
12784 @item #pragma weak @var{symbol}
12785 @cindex pragma, weak
12786 This pragma declares @var{symbol} to be weak, as if the declaration
12787 had the attribute of the same name. The pragma may appear before
12788 or after the declaration of @var{symbol}, but must appear before
12789 either its first use or its definition. It is not an error for
12790 @var{symbol} to never be defined at all.
12792 @item #pragma weak @var{symbol1} = @var{symbol2}
12793 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
12794 It is an error if @var{symbol2} is not defined in the current
12798 @node Diagnostic Pragmas
12799 @subsection Diagnostic Pragmas
12801 GCC allows the user to selectively enable or disable certain types of
12802 diagnostics, and change the kind of the diagnostic. For example, a
12803 project's policy might require that all sources compile with
12804 @option{-Werror} but certain files might have exceptions allowing
12805 specific types of warnings. Or, a project might selectively enable
12806 diagnostics and treat them as errors depending on which preprocessor
12807 macros are defined.
12810 @item #pragma GCC diagnostic @var{kind} @var{option}
12811 @cindex pragma, diagnostic
12813 Modifies the disposition of a diagnostic. Note that not all
12814 diagnostics are modifiable; at the moment only warnings (normally
12815 controlled by @samp{-W@dots{}}) can be controlled, and not all of them.
12816 Use @option{-fdiagnostics-show-option} to determine which diagnostics
12817 are controllable and which option controls them.
12819 @var{kind} is @samp{error} to treat this diagnostic as an error,
12820 @samp{warning} to treat it like a warning (even if @option{-Werror} is
12821 in effect), or @samp{ignored} if the diagnostic is to be ignored.
12822 @var{option} is a double quoted string which matches the command-line
12826 #pragma GCC diagnostic warning "-Wformat"
12827 #pragma GCC diagnostic error "-Wformat"
12828 #pragma GCC diagnostic ignored "-Wformat"
12831 Note that these pragmas override any command-line options. GCC keeps
12832 track of the location of each pragma, and issues diagnostics according
12833 to the state as of that point in the source file. Thus, pragmas occurring
12834 after a line do not affect diagnostics caused by that line.
12836 @item #pragma GCC diagnostic push
12837 @itemx #pragma GCC diagnostic pop
12839 Causes GCC to remember the state of the diagnostics as of each
12840 @code{push}, and restore to that point at each @code{pop}. If a
12841 @code{pop} has no matching @code{push}, the command line options are
12845 #pragma GCC diagnostic error "-Wuninitialized"
12846 foo(a); /* error is given for this one */
12847 #pragma GCC diagnostic push
12848 #pragma GCC diagnostic ignored "-Wuninitialized"
12849 foo(b); /* no diagnostic for this one */
12850 #pragma GCC diagnostic pop
12851 foo(c); /* error is given for this one */
12852 #pragma GCC diagnostic pop
12853 foo(d); /* depends on command line options */
12858 GCC also offers a simple mechanism for printing messages during
12862 @item #pragma message @var{string}
12863 @cindex pragma, diagnostic
12865 Prints @var{string} as a compiler message on compilation. The message
12866 is informational only, and is neither a compilation warning nor an error.
12869 #pragma message "Compiling " __FILE__ "..."
12872 @var{string} may be parenthesized, and is printed with location
12873 information. For example,
12876 #define DO_PRAGMA(x) _Pragma (#x)
12877 #define TODO(x) DO_PRAGMA(message ("TODO - " #x))
12879 TODO(Remember to fix this)
12882 prints @samp{/tmp/file.c:4: note: #pragma message:
12883 TODO - Remember to fix this}.
12887 @node Visibility Pragmas
12888 @subsection Visibility Pragmas
12891 @item #pragma GCC visibility push(@var{visibility})
12892 @itemx #pragma GCC visibility pop
12893 @cindex pragma, visibility
12895 This pragma allows the user to set the visibility for multiple
12896 declarations without having to give each a visibility attribute
12897 @xref{Function Attributes}, for more information about visibility and
12898 the attribute syntax.
12900 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
12901 declarations. Class members and template specializations are not
12902 affected; if you want to override the visibility for a particular
12903 member or instantiation, you must use an attribute.
12908 @node Push/Pop Macro Pragmas
12909 @subsection Push/Pop Macro Pragmas
12911 For compatibility with Microsoft Windows compilers, GCC supports
12912 @samp{#pragma push_macro(@var{"macro_name"})}
12913 and @samp{#pragma pop_macro(@var{"macro_name"})}.
12916 @item #pragma push_macro(@var{"macro_name"})
12917 @cindex pragma, push_macro
12918 This pragma saves the value of the macro named as @var{macro_name} to
12919 the top of the stack for this macro.
12921 @item #pragma pop_macro(@var{"macro_name"})
12922 @cindex pragma, pop_macro
12923 This pragma sets the value of the macro named as @var{macro_name} to
12924 the value on top of the stack for this macro. If the stack for
12925 @var{macro_name} is empty, the value of the macro remains unchanged.
12932 #pragma push_macro("X")
12935 #pragma pop_macro("X")
12939 In this example, the definition of X as 1 is saved by @code{#pragma
12940 push_macro} and restored by @code{#pragma pop_macro}.
12942 @node Function Specific Option Pragmas
12943 @subsection Function Specific Option Pragmas
12946 @item #pragma GCC target (@var{"string"}...)
12947 @cindex pragma GCC target
12949 This pragma allows you to set target specific options for functions
12950 defined later in the source file. One or more strings can be
12951 specified. Each function that is defined after this point will be as
12952 if @code{attribute((target("STRING")))} was specified for that
12953 function. The parenthesis around the options is optional.
12954 @xref{Function Attributes}, for more information about the
12955 @code{target} attribute and the attribute syntax.
12957 The @samp{#pragma GCC target} pragma is not implemented in GCC
12958 versions earlier than 4.4, and is currently only implemented for the
12959 386 and x86_64 backends.
12963 @item #pragma GCC optimize (@var{"string"}...)
12964 @cindex pragma GCC optimize
12966 This pragma allows you to set global optimization options for functions
12967 defined later in the source file. One or more strings can be
12968 specified. Each function that is defined after this point will be as
12969 if @code{attribute((optimize("STRING")))} was specified for that
12970 function. The parenthesis around the options is optional.
12971 @xref{Function Attributes}, for more information about the
12972 @code{optimize} attribute and the attribute syntax.
12974 The @samp{#pragma GCC optimize} pragma is not implemented in GCC
12975 versions earlier than 4.4.
12979 @item #pragma GCC push_options
12980 @itemx #pragma GCC pop_options
12981 @cindex pragma GCC push_options
12982 @cindex pragma GCC pop_options
12984 These pragmas maintain a stack of the current target and optimization
12985 options. It is intended for include files where you temporarily want
12986 to switch to using a different @samp{#pragma GCC target} or
12987 @samp{#pragma GCC optimize} and then to pop back to the previous
12990 The @samp{#pragma GCC push_options} and @samp{#pragma GCC pop_options}
12991 pragmas are not implemented in GCC versions earlier than 4.4.
12995 @item #pragma GCC reset_options
12996 @cindex pragma GCC reset_options
12998 This pragma clears the current @code{#pragma GCC target} and
12999 @code{#pragma GCC optimize} to use the default switches as specified
13000 on the command line.
13002 The @samp{#pragma GCC reset_options} pragma is not implemented in GCC
13003 versions earlier than 4.4.
13006 @node Unnamed Fields
13007 @section Unnamed struct/union fields within structs/unions
13008 @cindex @code{struct}
13009 @cindex @code{union}
13011 As permitted by ISO C1X and for compatibility with other compilers,
13012 GCC allows you to define
13013 a structure or union that contains, as fields, structures and unions
13014 without names. For example:
13027 In this example, the user would be able to access members of the unnamed
13028 union with code like @samp{foo.b}. Note that only unnamed structs and
13029 unions are allowed, you may not have, for example, an unnamed
13032 You must never create such structures that cause ambiguous field definitions.
13033 For example, this structure:
13044 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
13045 The compiler gives errors for such constructs.
13047 @opindex fms-extensions
13048 Unless @option{-fms-extensions} is used, the unnamed field must be a
13049 structure or union definition without a tag (for example, @samp{struct
13050 @{ int a; @};}), or a @code{typedef} name for such a structure or
13051 union. If @option{-fms-extensions} is used, the field may
13052 also be a definition with a tag such as @samp{struct foo @{ int a;
13053 @};}, a reference to a previously defined structure or union such as
13054 @samp{struct foo;}, or a reference to a @code{typedef} name for a
13055 previously defined structure or union type with a tag.
13057 @opindex fplan9-extensions
13058 The option @option{-fplan9-extensions} enables
13059 @option{-fms-extensions} as well as two other extensions. First, a
13060 pointer to a structure is automatically converted to a pointer to an
13061 anonymous field for assignments and function calls. For example:
13064 struct s1 @{ int a; @};
13065 struct s2 @{ struct s1; @};
13066 extern void f1 (struct s1 *);
13067 void f2 (struct s2 *p) @{ f1 (p); @}
13070 In the call to @code{f1} inside @code{f2}, the pointer @code{p} is
13071 converted into a pointer to the anonymous field.
13073 Second, when the type of an anonymous field is a @code{typedef} for a
13074 @code{struct} or @code{union}, code may refer to the field using the
13075 name of the @code{typedef}.
13078 typedef struct @{ int a; @} s1;
13079 struct s2 @{ s1; @};
13080 s1 f1 (struct s2 *p) @{ return p->s1; @}
13083 These usages are only permitted when they are not ambiguous.
13086 @section Thread-Local Storage
13087 @cindex Thread-Local Storage
13088 @cindex @acronym{TLS}
13089 @cindex @code{__thread}
13091 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
13092 are allocated such that there is one instance of the variable per extant
13093 thread. The run-time model GCC uses to implement this originates
13094 in the IA-64 processor-specific ABI, but has since been migrated
13095 to other processors as well. It requires significant support from
13096 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
13097 system libraries (@file{libc.so} and @file{libpthread.so}), so it
13098 is not available everywhere.
13100 At the user level, the extension is visible with a new storage
13101 class keyword: @code{__thread}. For example:
13105 extern __thread struct state s;
13106 static __thread char *p;
13109 The @code{__thread} specifier may be used alone, with the @code{extern}
13110 or @code{static} specifiers, but with no other storage class specifier.
13111 When used with @code{extern} or @code{static}, @code{__thread} must appear
13112 immediately after the other storage class specifier.
13114 The @code{__thread} specifier may be applied to any global, file-scoped
13115 static, function-scoped static, or static data member of a class. It may
13116 not be applied to block-scoped automatic or non-static data member.
13118 When the address-of operator is applied to a thread-local variable, it is
13119 evaluated at run-time and returns the address of the current thread's
13120 instance of that variable. An address so obtained may be used by any
13121 thread. When a thread terminates, any pointers to thread-local variables
13122 in that thread become invalid.
13124 No static initialization may refer to the address of a thread-local variable.
13126 In C++, if an initializer is present for a thread-local variable, it must
13127 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
13130 See @uref{http://people.redhat.com/drepper/tls.pdf,
13131 ELF Handling For Thread-Local Storage} for a detailed explanation of
13132 the four thread-local storage addressing models, and how the run-time
13133 is expected to function.
13136 * C99 Thread-Local Edits::
13137 * C++98 Thread-Local Edits::
13140 @node C99 Thread-Local Edits
13141 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
13143 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
13144 that document the exact semantics of the language extension.
13148 @cite{5.1.2 Execution environments}
13150 Add new text after paragraph 1
13153 Within either execution environment, a @dfn{thread} is a flow of
13154 control within a program. It is implementation defined whether
13155 or not there may be more than one thread associated with a program.
13156 It is implementation defined how threads beyond the first are
13157 created, the name and type of the function called at thread
13158 startup, and how threads may be terminated. However, objects
13159 with thread storage duration shall be initialized before thread
13164 @cite{6.2.4 Storage durations of objects}
13166 Add new text before paragraph 3
13169 An object whose identifier is declared with the storage-class
13170 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
13171 Its lifetime is the entire execution of the thread, and its
13172 stored value is initialized only once, prior to thread startup.
13176 @cite{6.4.1 Keywords}
13178 Add @code{__thread}.
13181 @cite{6.7.1 Storage-class specifiers}
13183 Add @code{__thread} to the list of storage class specifiers in
13186 Change paragraph 2 to
13189 With the exception of @code{__thread}, at most one storage-class
13190 specifier may be given [@dots{}]. The @code{__thread} specifier may
13191 be used alone, or immediately following @code{extern} or
13195 Add new text after paragraph 6
13198 The declaration of an identifier for a variable that has
13199 block scope that specifies @code{__thread} shall also
13200 specify either @code{extern} or @code{static}.
13202 The @code{__thread} specifier shall be used only with
13207 @node C++98 Thread-Local Edits
13208 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
13210 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
13211 that document the exact semantics of the language extension.
13215 @b{[intro.execution]}
13217 New text after paragraph 4
13220 A @dfn{thread} is a flow of control within the abstract machine.
13221 It is implementation defined whether or not there may be more than
13225 New text after paragraph 7
13228 It is unspecified whether additional action must be taken to
13229 ensure when and whether side effects are visible to other threads.
13235 Add @code{__thread}.
13238 @b{[basic.start.main]}
13240 Add after paragraph 5
13243 The thread that begins execution at the @code{main} function is called
13244 the @dfn{main thread}. It is implementation defined how functions
13245 beginning threads other than the main thread are designated or typed.
13246 A function so designated, as well as the @code{main} function, is called
13247 a @dfn{thread startup function}. It is implementation defined what
13248 happens if a thread startup function returns. It is implementation
13249 defined what happens to other threads when any thread calls @code{exit}.
13253 @b{[basic.start.init]}
13255 Add after paragraph 4
13258 The storage for an object of thread storage duration shall be
13259 statically initialized before the first statement of the thread startup
13260 function. An object of thread storage duration shall not require
13261 dynamic initialization.
13265 @b{[basic.start.term]}
13267 Add after paragraph 3
13270 The type of an object with thread storage duration shall not have a
13271 non-trivial destructor, nor shall it be an array type whose elements
13272 (directly or indirectly) have non-trivial destructors.
13278 Add ``thread storage duration'' to the list in paragraph 1.
13283 Thread, static, and automatic storage durations are associated with
13284 objects introduced by declarations [@dots{}].
13287 Add @code{__thread} to the list of specifiers in paragraph 3.
13290 @b{[basic.stc.thread]}
13292 New section before @b{[basic.stc.static]}
13295 The keyword @code{__thread} applied to a non-local object gives the
13296 object thread storage duration.
13298 A local variable or class data member declared both @code{static}
13299 and @code{__thread} gives the variable or member thread storage
13304 @b{[basic.stc.static]}
13309 All objects which have neither thread storage duration, dynamic
13310 storage duration nor are local [@dots{}].
13316 Add @code{__thread} to the list in paragraph 1.
13321 With the exception of @code{__thread}, at most one
13322 @var{storage-class-specifier} shall appear in a given
13323 @var{decl-specifier-seq}. The @code{__thread} specifier may
13324 be used alone, or immediately following the @code{extern} or
13325 @code{static} specifiers. [@dots{}]
13328 Add after paragraph 5
13331 The @code{__thread} specifier can be applied only to the names of objects
13332 and to anonymous unions.
13338 Add after paragraph 6
13341 Non-@code{static} members shall not be @code{__thread}.
13345 @node Binary constants
13346 @section Binary constants using the @samp{0b} prefix
13347 @cindex Binary constants using the @samp{0b} prefix
13349 Integer constants can be written as binary constants, consisting of a
13350 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
13351 @samp{0B}. This is particularly useful in environments that operate a
13352 lot on the bit-level (like microcontrollers).
13354 The following statements are identical:
13363 The type of these constants follows the same rules as for octal or
13364 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
13367 @node C++ Extensions
13368 @chapter Extensions to the C++ Language
13369 @cindex extensions, C++ language
13370 @cindex C++ language extensions
13372 The GNU compiler provides these extensions to the C++ language (and you
13373 can also use most of the C language extensions in your C++ programs). If you
13374 want to write code that checks whether these features are available, you can
13375 test for the GNU compiler the same way as for C programs: check for a
13376 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
13377 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
13378 Predefined Macros,cpp,The GNU C Preprocessor}).
13381 * C++ Volatiles:: What constitutes an access to a volatile object.
13382 * Restricted Pointers:: C99 restricted pointers and references.
13383 * Vague Linkage:: Where G++ puts inlines, vtables and such.
13384 * C++ Interface:: You can use a single C++ header file for both
13385 declarations and definitions.
13386 * Template Instantiation:: Methods for ensuring that exactly one copy of
13387 each needed template instantiation is emitted.
13388 * Bound member functions:: You can extract a function pointer to the
13389 method denoted by a @samp{->*} or @samp{.*} expression.
13390 * C++ Attributes:: Variable, function, and type attributes for C++ only.
13391 * Namespace Association:: Strong using-directives for namespace association.
13392 * Type Traits:: Compiler support for type traits
13393 * Java Exceptions:: Tweaking exception handling to work with Java.
13394 * Deprecated Features:: Things will disappear from g++.
13395 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
13398 @node C++ Volatiles
13399 @section When is a Volatile C++ Object Accessed?
13400 @cindex accessing volatiles
13401 @cindex volatile read
13402 @cindex volatile write
13403 @cindex volatile access
13405 The C++ standard differs from the C standard in its treatment of
13406 volatile objects. It fails to specify what constitutes a volatile
13407 access, except to say that C++ should behave in a similar manner to C
13408 with respect to volatiles, where possible. However, the different
13409 lvalueness of expressions between C and C++ complicate the behaviour.
13410 G++ behaves the same as GCC for volatile access, @xref{C
13411 Extensions,,Volatiles}, for a description of GCC's behaviour.
13413 The C and C++ language specifications differ when an object is
13414 accessed in a void context:
13417 volatile int *src = @var{somevalue};
13421 The C++ standard specifies that such expressions do not undergo lvalue
13422 to rvalue conversion, and that the type of the dereferenced object may
13423 be incomplete. The C++ standard does not specify explicitly that it
13424 is lvalue to rvalue conversion which is responsible for causing an
13425 access. There is reason to believe that it is, because otherwise
13426 certain simple expressions become undefined. However, because it
13427 would surprise most programmers, G++ treats dereferencing a pointer to
13428 volatile object of complete type as GCC would do for an equivalent
13429 type in C@. When the object has incomplete type, G++ issues a
13430 warning; if you wish to force an error, you must force a conversion to
13431 rvalue with, for instance, a static cast.
13433 When using a reference to volatile, G++ does not treat equivalent
13434 expressions as accesses to volatiles, but instead issues a warning that
13435 no volatile is accessed. The rationale for this is that otherwise it
13436 becomes difficult to determine where volatile access occur, and not
13437 possible to ignore the return value from functions returning volatile
13438 references. Again, if you wish to force a read, cast the reference to
13441 G++ implements the same behaviour as GCC does when assigning to a
13442 volatile object -- there is no reread of the assigned-to object, the
13443 assigned rvalue is reused. Note that in C++ assignment expressions
13444 are lvalues, and if used as an lvalue, the volatile object will be
13445 referred to. For instance, @var{vref} will refer to @var{vobj}, as
13446 expected, in the following example:
13450 volatile int &vref = vobj = @var{something};
13453 @node Restricted Pointers
13454 @section Restricting Pointer Aliasing
13455 @cindex restricted pointers
13456 @cindex restricted references
13457 @cindex restricted this pointer
13459 As with the C front end, G++ understands the C99 feature of restricted pointers,
13460 specified with the @code{__restrict__}, or @code{__restrict} type
13461 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
13462 language flag, @code{restrict} is not a keyword in C++.
13464 In addition to allowing restricted pointers, you can specify restricted
13465 references, which indicate that the reference is not aliased in the local
13469 void fn (int *__restrict__ rptr, int &__restrict__ rref)
13476 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
13477 @var{rref} refers to a (different) unaliased integer.
13479 You may also specify whether a member function's @var{this} pointer is
13480 unaliased by using @code{__restrict__} as a member function qualifier.
13483 void T::fn () __restrict__
13490 Within the body of @code{T::fn}, @var{this} will have the effective
13491 definition @code{T *__restrict__ const this}. Notice that the
13492 interpretation of a @code{__restrict__} member function qualifier is
13493 different to that of @code{const} or @code{volatile} qualifier, in that it
13494 is applied to the pointer rather than the object. This is consistent with
13495 other compilers which implement restricted pointers.
13497 As with all outermost parameter qualifiers, @code{__restrict__} is
13498 ignored in function definition matching. This means you only need to
13499 specify @code{__restrict__} in a function definition, rather than
13500 in a function prototype as well.
13502 @node Vague Linkage
13503 @section Vague Linkage
13504 @cindex vague linkage
13506 There are several constructs in C++ which require space in the object
13507 file but are not clearly tied to a single translation unit. We say that
13508 these constructs have ``vague linkage''. Typically such constructs are
13509 emitted wherever they are needed, though sometimes we can be more
13513 @item Inline Functions
13514 Inline functions are typically defined in a header file which can be
13515 included in many different compilations. Hopefully they can usually be
13516 inlined, but sometimes an out-of-line copy is necessary, if the address
13517 of the function is taken or if inlining fails. In general, we emit an
13518 out-of-line copy in all translation units where one is needed. As an
13519 exception, we only emit inline virtual functions with the vtable, since
13520 it will always require a copy.
13522 Local static variables and string constants used in an inline function
13523 are also considered to have vague linkage, since they must be shared
13524 between all inlined and out-of-line instances of the function.
13528 C++ virtual functions are implemented in most compilers using a lookup
13529 table, known as a vtable. The vtable contains pointers to the virtual
13530 functions provided by a class, and each object of the class contains a
13531 pointer to its vtable (or vtables, in some multiple-inheritance
13532 situations). If the class declares any non-inline, non-pure virtual
13533 functions, the first one is chosen as the ``key method'' for the class,
13534 and the vtable is only emitted in the translation unit where the key
13537 @emph{Note:} If the chosen key method is later defined as inline, the
13538 vtable will still be emitted in every translation unit which defines it.
13539 Make sure that any inline virtuals are declared inline in the class
13540 body, even if they are not defined there.
13542 @item @code{type_info} objects
13543 @cindex @code{type_info}
13545 C++ requires information about types to be written out in order to
13546 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
13547 For polymorphic classes (classes with virtual functions), the @samp{type_info}
13548 object is written out along with the vtable so that @samp{dynamic_cast}
13549 can determine the dynamic type of a class object at runtime. For all
13550 other types, we write out the @samp{type_info} object when it is used: when
13551 applying @samp{typeid} to an expression, throwing an object, or
13552 referring to a type in a catch clause or exception specification.
13554 @item Template Instantiations
13555 Most everything in this section also applies to template instantiations,
13556 but there are other options as well.
13557 @xref{Template Instantiation,,Where's the Template?}.
13561 When used with GNU ld version 2.8 or later on an ELF system such as
13562 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
13563 these constructs will be discarded at link time. This is known as
13566 On targets that don't support COMDAT, but do support weak symbols, GCC
13567 will use them. This way one copy will override all the others, but
13568 the unused copies will still take up space in the executable.
13570 For targets which do not support either COMDAT or weak symbols,
13571 most entities with vague linkage will be emitted as local symbols to
13572 avoid duplicate definition errors from the linker. This will not happen
13573 for local statics in inlines, however, as having multiple copies will
13574 almost certainly break things.
13576 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
13577 another way to control placement of these constructs.
13579 @node C++ Interface
13580 @section #pragma interface and implementation
13582 @cindex interface and implementation headers, C++
13583 @cindex C++ interface and implementation headers
13584 @cindex pragmas, interface and implementation
13586 @code{#pragma interface} and @code{#pragma implementation} provide the
13587 user with a way of explicitly directing the compiler to emit entities
13588 with vague linkage (and debugging information) in a particular
13591 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
13592 most cases, because of COMDAT support and the ``key method'' heuristic
13593 mentioned in @ref{Vague Linkage}. Using them can actually cause your
13594 program to grow due to unnecessary out-of-line copies of inline
13595 functions. Currently (3.4) the only benefit of these
13596 @code{#pragma}s is reduced duplication of debugging information, and
13597 that should be addressed soon on DWARF 2 targets with the use of
13601 @item #pragma interface
13602 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
13603 @kindex #pragma interface
13604 Use this directive in @emph{header files} that define object classes, to save
13605 space in most of the object files that use those classes. Normally,
13606 local copies of certain information (backup copies of inline member
13607 functions, debugging information, and the internal tables that implement
13608 virtual functions) must be kept in each object file that includes class
13609 definitions. You can use this pragma to avoid such duplication. When a
13610 header file containing @samp{#pragma interface} is included in a
13611 compilation, this auxiliary information will not be generated (unless
13612 the main input source file itself uses @samp{#pragma implementation}).
13613 Instead, the object files will contain references to be resolved at link
13616 The second form of this directive is useful for the case where you have
13617 multiple headers with the same name in different directories. If you
13618 use this form, you must specify the same string to @samp{#pragma
13621 @item #pragma implementation
13622 @itemx #pragma implementation "@var{objects}.h"
13623 @kindex #pragma implementation
13624 Use this pragma in a @emph{main input file}, when you want full output from
13625 included header files to be generated (and made globally visible). The
13626 included header file, in turn, should use @samp{#pragma interface}.
13627 Backup copies of inline member functions, debugging information, and the
13628 internal tables used to implement virtual functions are all generated in
13629 implementation files.
13631 @cindex implied @code{#pragma implementation}
13632 @cindex @code{#pragma implementation}, implied
13633 @cindex naming convention, implementation headers
13634 If you use @samp{#pragma implementation} with no argument, it applies to
13635 an include file with the same basename@footnote{A file's @dfn{basename}
13636 was the name stripped of all leading path information and of trailing
13637 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
13638 file. For example, in @file{allclass.cc}, giving just
13639 @samp{#pragma implementation}
13640 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
13642 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
13643 an implementation file whenever you would include it from
13644 @file{allclass.cc} even if you never specified @samp{#pragma
13645 implementation}. This was deemed to be more trouble than it was worth,
13646 however, and disabled.
13648 Use the string argument if you want a single implementation file to
13649 include code from multiple header files. (You must also use
13650 @samp{#include} to include the header file; @samp{#pragma
13651 implementation} only specifies how to use the file---it doesn't actually
13654 There is no way to split up the contents of a single header file into
13655 multiple implementation files.
13658 @cindex inlining and C++ pragmas
13659 @cindex C++ pragmas, effect on inlining
13660 @cindex pragmas in C++, effect on inlining
13661 @samp{#pragma implementation} and @samp{#pragma interface} also have an
13662 effect on function inlining.
13664 If you define a class in a header file marked with @samp{#pragma
13665 interface}, the effect on an inline function defined in that class is
13666 similar to an explicit @code{extern} declaration---the compiler emits
13667 no code at all to define an independent version of the function. Its
13668 definition is used only for inlining with its callers.
13670 @opindex fno-implement-inlines
13671 Conversely, when you include the same header file in a main source file
13672 that declares it as @samp{#pragma implementation}, the compiler emits
13673 code for the function itself; this defines a version of the function
13674 that can be found via pointers (or by callers compiled without
13675 inlining). If all calls to the function can be inlined, you can avoid
13676 emitting the function by compiling with @option{-fno-implement-inlines}.
13677 If any calls were not inlined, you will get linker errors.
13679 @node Template Instantiation
13680 @section Where's the Template?
13681 @cindex template instantiation
13683 C++ templates are the first language feature to require more
13684 intelligence from the environment than one usually finds on a UNIX
13685 system. Somehow the compiler and linker have to make sure that each
13686 template instance occurs exactly once in the executable if it is needed,
13687 and not at all otherwise. There are two basic approaches to this
13688 problem, which are referred to as the Borland model and the Cfront model.
13691 @item Borland model
13692 Borland C++ solved the template instantiation problem by adding the code
13693 equivalent of common blocks to their linker; the compiler emits template
13694 instances in each translation unit that uses them, and the linker
13695 collapses them together. The advantage of this model is that the linker
13696 only has to consider the object files themselves; there is no external
13697 complexity to worry about. This disadvantage is that compilation time
13698 is increased because the template code is being compiled repeatedly.
13699 Code written for this model tends to include definitions of all
13700 templates in the header file, since they must be seen to be
13704 The AT&T C++ translator, Cfront, solved the template instantiation
13705 problem by creating the notion of a template repository, an
13706 automatically maintained place where template instances are stored. A
13707 more modern version of the repository works as follows: As individual
13708 object files are built, the compiler places any template definitions and
13709 instantiations encountered in the repository. At link time, the link
13710 wrapper adds in the objects in the repository and compiles any needed
13711 instances that were not previously emitted. The advantages of this
13712 model are more optimal compilation speed and the ability to use the
13713 system linker; to implement the Borland model a compiler vendor also
13714 needs to replace the linker. The disadvantages are vastly increased
13715 complexity, and thus potential for error; for some code this can be
13716 just as transparent, but in practice it can been very difficult to build
13717 multiple programs in one directory and one program in multiple
13718 directories. Code written for this model tends to separate definitions
13719 of non-inline member templates into a separate file, which should be
13720 compiled separately.
13723 When used with GNU ld version 2.8 or later on an ELF system such as
13724 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
13725 Borland model. On other systems, G++ implements neither automatic
13728 A future version of G++ will support a hybrid model whereby the compiler
13729 will emit any instantiations for which the template definition is
13730 included in the compile, and store template definitions and
13731 instantiation context information into the object file for the rest.
13732 The link wrapper will extract that information as necessary and invoke
13733 the compiler to produce the remaining instantiations. The linker will
13734 then combine duplicate instantiations.
13736 In the mean time, you have the following options for dealing with
13737 template instantiations:
13742 Compile your template-using code with @option{-frepo}. The compiler will
13743 generate files with the extension @samp{.rpo} listing all of the
13744 template instantiations used in the corresponding object files which
13745 could be instantiated there; the link wrapper, @samp{collect2}, will
13746 then update the @samp{.rpo} files to tell the compiler where to place
13747 those instantiations and rebuild any affected object files. The
13748 link-time overhead is negligible after the first pass, as the compiler
13749 will continue to place the instantiations in the same files.
13751 This is your best option for application code written for the Borland
13752 model, as it will just work. Code written for the Cfront model will
13753 need to be modified so that the template definitions are available at
13754 one or more points of instantiation; usually this is as simple as adding
13755 @code{#include <tmethods.cc>} to the end of each template header.
13757 For library code, if you want the library to provide all of the template
13758 instantiations it needs, just try to link all of its object files
13759 together; the link will fail, but cause the instantiations to be
13760 generated as a side effect. Be warned, however, that this may cause
13761 conflicts if multiple libraries try to provide the same instantiations.
13762 For greater control, use explicit instantiation as described in the next
13766 @opindex fno-implicit-templates
13767 Compile your code with @option{-fno-implicit-templates} to disable the
13768 implicit generation of template instances, and explicitly instantiate
13769 all the ones you use. This approach requires more knowledge of exactly
13770 which instances you need than do the others, but it's less
13771 mysterious and allows greater control. You can scatter the explicit
13772 instantiations throughout your program, perhaps putting them in the
13773 translation units where the instances are used or the translation units
13774 that define the templates themselves; you can put all of the explicit
13775 instantiations you need into one big file; or you can create small files
13782 template class Foo<int>;
13783 template ostream& operator <<
13784 (ostream&, const Foo<int>&);
13787 for each of the instances you need, and create a template instantiation
13788 library from those.
13790 If you are using Cfront-model code, you can probably get away with not
13791 using @option{-fno-implicit-templates} when compiling files that don't
13792 @samp{#include} the member template definitions.
13794 If you use one big file to do the instantiations, you may want to
13795 compile it without @option{-fno-implicit-templates} so you get all of the
13796 instances required by your explicit instantiations (but not by any
13797 other files) without having to specify them as well.
13799 G++ has extended the template instantiation syntax given in the ISO
13800 standard to allow forward declaration of explicit instantiations
13801 (with @code{extern}), instantiation of the compiler support data for a
13802 template class (i.e.@: the vtable) without instantiating any of its
13803 members (with @code{inline}), and instantiation of only the static data
13804 members of a template class, without the support data or member
13805 functions (with (@code{static}):
13808 extern template int max (int, int);
13809 inline template class Foo<int>;
13810 static template class Foo<int>;
13814 Do nothing. Pretend G++ does implement automatic instantiation
13815 management. Code written for the Borland model will work fine, but
13816 each translation unit will contain instances of each of the templates it
13817 uses. In a large program, this can lead to an unacceptable amount of code
13821 @node Bound member functions
13822 @section Extracting the function pointer from a bound pointer to member function
13824 @cindex pointer to member function
13825 @cindex bound pointer to member function
13827 In C++, pointer to member functions (PMFs) are implemented using a wide
13828 pointer of sorts to handle all the possible call mechanisms; the PMF
13829 needs to store information about how to adjust the @samp{this} pointer,
13830 and if the function pointed to is virtual, where to find the vtable, and
13831 where in the vtable to look for the member function. If you are using
13832 PMFs in an inner loop, you should really reconsider that decision. If
13833 that is not an option, you can extract the pointer to the function that
13834 would be called for a given object/PMF pair and call it directly inside
13835 the inner loop, to save a bit of time.
13837 Note that you will still be paying the penalty for the call through a
13838 function pointer; on most modern architectures, such a call defeats the
13839 branch prediction features of the CPU@. This is also true of normal
13840 virtual function calls.
13842 The syntax for this extension is
13846 extern int (A::*fp)();
13847 typedef int (*fptr)(A *);
13849 fptr p = (fptr)(a.*fp);
13852 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
13853 no object is needed to obtain the address of the function. They can be
13854 converted to function pointers directly:
13857 fptr p1 = (fptr)(&A::foo);
13860 @opindex Wno-pmf-conversions
13861 You must specify @option{-Wno-pmf-conversions} to use this extension.
13863 @node C++ Attributes
13864 @section C++-Specific Variable, Function, and Type Attributes
13866 Some attributes only make sense for C++ programs.
13869 @item init_priority (@var{priority})
13870 @cindex @code{init_priority} attribute
13873 In Standard C++, objects defined at namespace scope are guaranteed to be
13874 initialized in an order in strict accordance with that of their definitions
13875 @emph{in a given translation unit}. No guarantee is made for initializations
13876 across translation units. However, GNU C++ allows users to control the
13877 order of initialization of objects defined at namespace scope with the
13878 @code{init_priority} attribute by specifying a relative @var{priority},
13879 a constant integral expression currently bounded between 101 and 65535
13880 inclusive. Lower numbers indicate a higher priority.
13882 In the following example, @code{A} would normally be created before
13883 @code{B}, but the @code{init_priority} attribute has reversed that order:
13886 Some_Class A __attribute__ ((init_priority (2000)));
13887 Some_Class B __attribute__ ((init_priority (543)));
13891 Note that the particular values of @var{priority} do not matter; only their
13894 @item java_interface
13895 @cindex @code{java_interface} attribute
13897 This type attribute informs C++ that the class is a Java interface. It may
13898 only be applied to classes declared within an @code{extern "Java"} block.
13899 Calls to methods declared in this interface will be dispatched using GCJ's
13900 interface table mechanism, instead of regular virtual table dispatch.
13904 See also @ref{Namespace Association}.
13906 @node Namespace Association
13907 @section Namespace Association
13909 @strong{Caution:} The semantics of this extension are not fully
13910 defined. Users should refrain from using this extension as its
13911 semantics may change subtly over time. It is possible that this
13912 extension will be removed in future versions of G++.
13914 A using-directive with @code{__attribute ((strong))} is stronger
13915 than a normal using-directive in two ways:
13919 Templates from the used namespace can be specialized and explicitly
13920 instantiated as though they were members of the using namespace.
13923 The using namespace is considered an associated namespace of all
13924 templates in the used namespace for purposes of argument-dependent
13928 The used namespace must be nested within the using namespace so that
13929 normal unqualified lookup works properly.
13931 This is useful for composing a namespace transparently from
13932 implementation namespaces. For example:
13937 template <class T> struct A @{ @};
13939 using namespace debug __attribute ((__strong__));
13940 template <> struct A<int> @{ @}; // @r{ok to specialize}
13942 template <class T> void f (A<T>);
13947 f (std::A<float>()); // @r{lookup finds} std::f
13953 @section Type Traits
13955 The C++ front-end implements syntactic extensions that allow to
13956 determine at compile time various characteristics of a type (or of a
13960 @item __has_nothrow_assign (type)
13961 If @code{type} is const qualified or is a reference type then the trait is
13962 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
13963 is true, else if @code{type} is a cv class or union type with copy assignment
13964 operators that are known not to throw an exception then the trait is true,
13965 else it is false. Requires: @code{type} shall be a complete type, an array
13966 type of unknown bound, or is a @code{void} type.
13968 @item __has_nothrow_copy (type)
13969 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
13970 @code{type} is a cv class or union type with copy constructors that
13971 are known not to throw an exception then the trait is true, else it is false.
13972 Requires: @code{type} shall be a complete type, an array type of
13973 unknown bound, or is a @code{void} type.
13975 @item __has_nothrow_constructor (type)
13976 If @code{__has_trivial_constructor (type)} is true then the trait is
13977 true, else if @code{type} is a cv class or union type (or array
13978 thereof) with a default constructor that is known not to throw an
13979 exception then the trait is true, else it is false. Requires:
13980 @code{type} shall be a complete type, an array type of unknown bound,
13981 or is a @code{void} type.
13983 @item __has_trivial_assign (type)
13984 If @code{type} is const qualified or is a reference type then the trait is
13985 false. Otherwise if @code{__is_pod (type)} is true then the trait is
13986 true, else if @code{type} is a cv class or union type with a trivial
13987 copy assignment ([class.copy]) then the trait is true, else it is
13988 false. Requires: @code{type} shall be a complete type, an array type
13989 of unknown bound, or is a @code{void} type.
13991 @item __has_trivial_copy (type)
13992 If @code{__is_pod (type)} is true or @code{type} is a reference type
13993 then the trait is true, else if @code{type} is a cv class or union type
13994 with a trivial copy constructor ([class.copy]) then the trait
13995 is true, else it is false. Requires: @code{type} shall be a complete
13996 type, an array type of unknown bound, or is a @code{void} type.
13998 @item __has_trivial_constructor (type)
13999 If @code{__is_pod (type)} is true then the trait is true, else if
14000 @code{type} is a cv class or union type (or array thereof) with a
14001 trivial default constructor ([class.ctor]) then the trait is true,
14002 else it is false. Requires: @code{type} shall be a complete type, an
14003 array type of unknown bound, or is a @code{void} type.
14005 @item __has_trivial_destructor (type)
14006 If @code{__is_pod (type)} is true or @code{type} is a reference type then
14007 the trait is true, else if @code{type} is a cv class or union type (or
14008 array thereof) with a trivial destructor ([class.dtor]) then the trait
14009 is true, else it is false. Requires: @code{type} shall be a complete
14010 type, an array type of unknown bound, or is a @code{void} type.
14012 @item __has_virtual_destructor (type)
14013 If @code{type} is a class type with a virtual destructor
14014 ([class.dtor]) then the trait is true, else it is false. Requires:
14015 @code{type} shall be a complete type, an array type of unknown bound,
14016 or is a @code{void} type.
14018 @item __is_abstract (type)
14019 If @code{type} is an abstract class ([class.abstract]) then the trait
14020 is true, else it is false. Requires: @code{type} shall be a complete
14021 type, an array type of unknown bound, or is a @code{void} type.
14023 @item __is_base_of (base_type, derived_type)
14024 If @code{base_type} is a base class of @code{derived_type}
14025 ([class.derived]) then the trait is true, otherwise it is false.
14026 Top-level cv qualifications of @code{base_type} and
14027 @code{derived_type} are ignored. For the purposes of this trait, a
14028 class type is considered is own base. Requires: if @code{__is_class
14029 (base_type)} and @code{__is_class (derived_type)} are true and
14030 @code{base_type} and @code{derived_type} are not the same type
14031 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
14032 type. Diagnostic is produced if this requirement is not met.
14034 @item __is_class (type)
14035 If @code{type} is a cv class type, and not a union type
14036 ([basic.compound]) the trait is true, else it is false.
14038 @item __is_empty (type)
14039 If @code{__is_class (type)} is false then the trait is false.
14040 Otherwise @code{type} is considered empty if and only if: @code{type}
14041 has no non-static data members, or all non-static data members, if
14042 any, are bit-fields of length 0, and @code{type} has no virtual
14043 members, and @code{type} has no virtual base classes, and @code{type}
14044 has no base classes @code{base_type} for which
14045 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
14046 be a complete type, an array type of unknown bound, or is a
14049 @item __is_enum (type)
14050 If @code{type} is a cv enumeration type ([basic.compound]) the trait is
14051 true, else it is false.
14053 @item __is_pod (type)
14054 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
14055 else it is false. Requires: @code{type} shall be a complete type,
14056 an array type of unknown bound, or is a @code{void} type.
14058 @item __is_polymorphic (type)
14059 If @code{type} is a polymorphic class ([class.virtual]) then the trait
14060 is true, else it is false. Requires: @code{type} shall be a complete
14061 type, an array type of unknown bound, or is a @code{void} type.
14063 @item __is_union (type)
14064 If @code{type} is a cv union type ([basic.compound]) the trait is
14065 true, else it is false.
14069 @node Java Exceptions
14070 @section Java Exceptions
14072 The Java language uses a slightly different exception handling model
14073 from C++. Normally, GNU C++ will automatically detect when you are
14074 writing C++ code that uses Java exceptions, and handle them
14075 appropriately. However, if C++ code only needs to execute destructors
14076 when Java exceptions are thrown through it, GCC will guess incorrectly.
14077 Sample problematic code is:
14080 struct S @{ ~S(); @};
14081 extern void bar(); // @r{is written in Java, and may throw exceptions}
14090 The usual effect of an incorrect guess is a link failure, complaining of
14091 a missing routine called @samp{__gxx_personality_v0}.
14093 You can inform the compiler that Java exceptions are to be used in a
14094 translation unit, irrespective of what it might think, by writing
14095 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
14096 @samp{#pragma} must appear before any functions that throw or catch
14097 exceptions, or run destructors when exceptions are thrown through them.
14099 You cannot mix Java and C++ exceptions in the same translation unit. It
14100 is believed to be safe to throw a C++ exception from one file through
14101 another file compiled for the Java exception model, or vice versa, but
14102 there may be bugs in this area.
14104 @node Deprecated Features
14105 @section Deprecated Features
14107 In the past, the GNU C++ compiler was extended to experiment with new
14108 features, at a time when the C++ language was still evolving. Now that
14109 the C++ standard is complete, some of those features are superseded by
14110 superior alternatives. Using the old features might cause a warning in
14111 some cases that the feature will be dropped in the future. In other
14112 cases, the feature might be gone already.
14114 While the list below is not exhaustive, it documents some of the options
14115 that are now deprecated:
14118 @item -fexternal-templates
14119 @itemx -falt-external-templates
14120 These are two of the many ways for G++ to implement template
14121 instantiation. @xref{Template Instantiation}. The C++ standard clearly
14122 defines how template definitions have to be organized across
14123 implementation units. G++ has an implicit instantiation mechanism that
14124 should work just fine for standard-conforming code.
14126 @item -fstrict-prototype
14127 @itemx -fno-strict-prototype
14128 Previously it was possible to use an empty prototype parameter list to
14129 indicate an unspecified number of parameters (like C), rather than no
14130 parameters, as C++ demands. This feature has been removed, except where
14131 it is required for backwards compatibility. @xref{Backwards Compatibility}.
14134 G++ allows a virtual function returning @samp{void *} to be overridden
14135 by one returning a different pointer type. This extension to the
14136 covariant return type rules is now deprecated and will be removed from a
14139 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
14140 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
14141 and are now removed from G++. Code using these operators should be
14142 modified to use @code{std::min} and @code{std::max} instead.
14144 The named return value extension has been deprecated, and is now
14147 The use of initializer lists with new expressions has been deprecated,
14148 and is now removed from G++.
14150 Floating and complex non-type template parameters have been deprecated,
14151 and are now removed from G++.
14153 The implicit typename extension has been deprecated and is now
14156 The use of default arguments in function pointers, function typedefs
14157 and other places where they are not permitted by the standard is
14158 deprecated and will be removed from a future version of G++.
14160 G++ allows floating-point literals to appear in integral constant expressions,
14161 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
14162 This extension is deprecated and will be removed from a future version.
14164 G++ allows static data members of const floating-point type to be declared
14165 with an initializer in a class definition. The standard only allows
14166 initializers for static members of const integral types and const
14167 enumeration types so this extension has been deprecated and will be removed
14168 from a future version.
14170 @node Backwards Compatibility
14171 @section Backwards Compatibility
14172 @cindex Backwards Compatibility
14173 @cindex ARM [Annotated C++ Reference Manual]
14175 Now that there is a definitive ISO standard C++, G++ has a specification
14176 to adhere to. The C++ language evolved over time, and features that
14177 used to be acceptable in previous drafts of the standard, such as the ARM
14178 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
14179 compilation of C++ written to such drafts, G++ contains some backwards
14180 compatibilities. @emph{All such backwards compatibility features are
14181 liable to disappear in future versions of G++.} They should be considered
14182 deprecated. @xref{Deprecated Features}.
14186 If a variable is declared at for scope, it used to remain in scope until
14187 the end of the scope which contained the for statement (rather than just
14188 within the for scope). G++ retains this, but issues a warning, if such a
14189 variable is accessed outside the for scope.
14191 @item Implicit C language
14192 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
14193 scope to set the language. On such systems, all header files are
14194 implicitly scoped inside a C language scope. Also, an empty prototype
14195 @code{()} will be treated as an unspecified number of arguments, rather
14196 than no arguments, as C++ demands.