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 For Objective-C dialects, @code{NSString} (or @code{__NSString__}) is
2422 recognized in the same context. Declarations including these format attributes
2423 will be parsed for correct syntax, however the result of checking of such format
2424 strings is not yet defined, and will not be carried out by this version of the
2427 The target may also provide additional types of format checks.
2428 @xref{Target Format Checks,,Format Checks Specific to Particular
2431 @item format_arg (@var{string-index})
2432 @cindex @code{format_arg} function attribute
2433 @opindex Wformat-nonliteral
2434 The @code{format_arg} attribute specifies that a function takes a format
2435 string for a @code{printf}, @code{scanf}, @code{strftime} or
2436 @code{strfmon} style function and modifies it (for example, to translate
2437 it into another language), so the result can be passed to a
2438 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2439 function (with the remaining arguments to the format function the same
2440 as they would have been for the unmodified string). For example, the
2445 my_dgettext (char *my_domain, const char *my_format)
2446 __attribute__ ((format_arg (2)));
2450 causes the compiler to check the arguments in calls to a @code{printf},
2451 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2452 format string argument is a call to the @code{my_dgettext} function, for
2453 consistency with the format string argument @code{my_format}. If the
2454 @code{format_arg} attribute had not been specified, all the compiler
2455 could tell in such calls to format functions would be that the format
2456 string argument is not constant; this would generate a warning when
2457 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2458 without the attribute.
2460 The parameter @var{string-index} specifies which argument is the format
2461 string argument (starting from one). Since non-static C++ methods have
2462 an implicit @code{this} argument, the arguments of such methods should
2463 be counted from two.
2465 The @code{format-arg} attribute allows you to identify your own
2466 functions which modify format strings, so that GCC can check the
2467 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2468 type function whose operands are a call to one of your own function.
2469 The compiler always treats @code{gettext}, @code{dgettext}, and
2470 @code{dcgettext} in this manner except when strict ISO C support is
2471 requested by @option{-ansi} or an appropriate @option{-std} option, or
2472 @option{-ffreestanding} or @option{-fno-builtin}
2473 is used. @xref{C Dialect Options,,Options
2474 Controlling C Dialect}.
2476 For Objective-C dialects, the @code{format-arg} attribute may refer to an
2477 @code{NSString} reference for compatibility with the @code{format} attribute
2480 The target may also allow additional types in @code{format-arg} attributes.
2481 @xref{Target Format Checks,,Format Checks Specific to Particular
2484 @item function_vector
2485 @cindex calling functions through the function vector on H8/300, M16C, M32C and SH2A processors
2486 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2487 function should be called through the function vector. Calling a
2488 function through the function vector will reduce code size, however;
2489 the function vector has a limited size (maximum 128 entries on the H8/300
2490 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2492 In SH2A target, this attribute declares a function to be called using the
2493 TBR relative addressing mode. The argument to this attribute is the entry
2494 number of the same function in a vector table containing all the TBR
2495 relative addressable functions. For the successful jump, register TBR
2496 should contain the start address of this TBR relative vector table.
2497 In the startup routine of the user application, user needs to care of this
2498 TBR register initialization. The TBR relative vector table can have at
2499 max 256 function entries. The jumps to these functions will be generated
2500 using a SH2A specific, non delayed branch instruction JSR/N @@(disp8,TBR).
2501 You must use GAS and GLD from GNU binutils version 2.7 or later for
2502 this attribute to work correctly.
2504 Please refer the example of M16C target, to see the use of this
2505 attribute while declaring a function,
2507 In an application, for a function being called once, this attribute will
2508 save at least 8 bytes of code; and if other successive calls are being
2509 made to the same function, it will save 2 bytes of code per each of these
2512 On M16C/M32C targets, the @code{function_vector} attribute declares a
2513 special page subroutine call function. Use of this attribute reduces
2514 the code size by 2 bytes for each call generated to the
2515 subroutine. The argument to the attribute is the vector number entry
2516 from the special page vector table which contains the 16 low-order
2517 bits of the subroutine's entry address. Each vector table has special
2518 page number (18 to 255) which are used in @code{jsrs} instruction.
2519 Jump addresses of the routines are generated by adding 0x0F0000 (in
2520 case of M16C targets) or 0xFF0000 (in case of M32C targets), to the 2
2521 byte addresses set in the vector table. Therefore you need to ensure
2522 that all the special page vector routines should get mapped within the
2523 address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF
2526 In the following example 2 bytes will be saved for each call to
2527 function @code{foo}.
2530 void foo (void) __attribute__((function_vector(0x18)));
2541 If functions are defined in one file and are called in another file,
2542 then be sure to write this declaration in both files.
2544 This attribute is ignored for R8C target.
2547 @cindex interrupt handler functions
2548 Use this attribute on the ARM, AVR, CRX, M32C, M32R/D, m68k, MeP, MIPS,
2549 RX and Xstormy16 ports to indicate that the specified function is an
2550 interrupt handler. The compiler will generate function entry and exit
2551 sequences suitable for use in an interrupt handler when this attribute
2554 Note, interrupt handlers for the Blackfin, H8/300, H8/300H, H8S, MicroBlaze,
2555 and SH processors can be specified via the @code{interrupt_handler} attribute.
2557 Note, on the AVR, interrupts will be enabled inside the function.
2559 Note, for the ARM, you can specify the kind of interrupt to be handled by
2560 adding an optional parameter to the interrupt attribute like this:
2563 void f () __attribute__ ((interrupt ("IRQ")));
2566 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2568 On ARMv7-M the interrupt type is ignored, and the attribute means the function
2569 may be called with a word aligned stack pointer.
2571 On MIPS targets, you can use the following attributes to modify the behavior
2572 of an interrupt handler:
2574 @item use_shadow_register_set
2575 @cindex @code{use_shadow_register_set} attribute
2576 Assume that the handler uses a shadow register set, instead of
2577 the main general-purpose registers.
2579 @item keep_interrupts_masked
2580 @cindex @code{keep_interrupts_masked} attribute
2581 Keep interrupts masked for the whole function. Without this attribute,
2582 GCC tries to reenable interrupts for as much of the function as it can.
2584 @item use_debug_exception_return
2585 @cindex @code{use_debug_exception_return} attribute
2586 Return using the @code{deret} instruction. Interrupt handlers that don't
2587 have this attribute return using @code{eret} instead.
2590 You can use any combination of these attributes, as shown below:
2592 void __attribute__ ((interrupt)) v0 ();
2593 void __attribute__ ((interrupt, use_shadow_register_set)) v1 ();
2594 void __attribute__ ((interrupt, keep_interrupts_masked)) v2 ();
2595 void __attribute__ ((interrupt, use_debug_exception_return)) v3 ();
2596 void __attribute__ ((interrupt, use_shadow_register_set,
2597 keep_interrupts_masked)) v4 ();
2598 void __attribute__ ((interrupt, use_shadow_register_set,
2599 use_debug_exception_return)) v5 ();
2600 void __attribute__ ((interrupt, keep_interrupts_masked,
2601 use_debug_exception_return)) v6 ();
2602 void __attribute__ ((interrupt, use_shadow_register_set,
2603 keep_interrupts_masked,
2604 use_debug_exception_return)) v7 ();
2607 @item ifunc ("@var{resolver}")
2608 @cindex @code{ifunc} attribute
2609 The @code{ifunc} attribute is used to mark a function as an indirect
2610 function using the STT_GNU_IFUNC symbol type extension to the ELF
2611 standard. This allows the resolution of the symbol value to be
2612 determined dynamically at load time, and an optimized version of the
2613 routine can be selected for the particular processor or other system
2614 characteristics determined then. To use this attribute, first define
2615 the implementation functions available, and a resolver function that
2616 returns a pointer to the selected implementation function. The
2617 implementation functions' declarations must match the API of the
2618 function being implemented, the resolver's declaration is be a
2619 function returning pointer to void function returning void:
2622 void *my_memcpy (void *dst, const void *src, size_t len)
2627 static void (*resolve_memcpy (void)) (void)
2629 return my_memcpy; // we'll just always select this routine
2633 The exported header file declaring the function the user calls would
2637 extern void *memcpy (void *, const void *, size_t);
2640 allowing the user to call this as a regular function, unaware of the
2641 implementation. Finally, the indirect function needs to be defined in
2642 the same translation unit as the resolver function:
2645 void *memcpy (void *, const void *, size_t)
2646 __attribute__ ((ifunc ("resolve_memcpy")));
2649 Indirect functions cannot be weak, and require a recent binutils (at
2650 least version 2.20.1), and GNU C library (at least version 2.11.1).
2652 @item interrupt_handler
2653 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
2654 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
2655 indicate that the specified function is an interrupt handler. The compiler
2656 will generate function entry and exit sequences suitable for use in an
2657 interrupt handler when this attribute is present.
2659 @item interrupt_thread
2660 @cindex interrupt thread functions on fido
2661 Use this attribute on fido, a subarchitecture of the m68k, to indicate
2662 that the specified function is an interrupt handler that is designed
2663 to run as a thread. The compiler omits generate prologue/epilogue
2664 sequences and replaces the return instruction with a @code{sleep}
2665 instruction. This attribute is available only on fido.
2668 @cindex interrupt service routines on ARM
2669 Use this attribute on ARM to write Interrupt Service Routines. This is an
2670 alias to the @code{interrupt} attribute above.
2673 @cindex User stack pointer in interrupts on the Blackfin
2674 When used together with @code{interrupt_handler}, @code{exception_handler}
2675 or @code{nmi_handler}, code will be generated to load the stack pointer
2676 from the USP register in the function prologue.
2679 @cindex @code{l1_text} function attribute
2680 This attribute specifies a function to be placed into L1 Instruction
2681 SRAM@. The function will be put into a specific section named @code{.l1.text}.
2682 With @option{-mfdpic}, function calls with a such function as the callee
2683 or caller will use inlined PLT.
2686 @cindex @code{l2} function attribute
2687 On the Blackfin, this attribute specifies a function to be placed into L2
2688 SRAM. The function will be put into a specific section named
2689 @code{.l1.text}. With @option{-mfdpic}, callers of such functions will use
2693 @cindex @code{leaf} function attribute
2694 Calls to external functions with this attribute must return to the current
2695 compilation unit only by return or by exception handling. In particular, leaf
2696 functions are not allowed to call callback function passed to it from current
2697 compilation unit or directly call functions exported by the unit or longjmp
2698 into the unit. Still leaf function might call functions from other complation
2699 units and thus they are not neccesarily leaf in the sense that they contains no
2700 function calls at all.
2702 The attribute is intended for library functions to improve dataflow analysis.
2703 Compiler takes the hint that any data not escaping current compilation unit can
2704 not be used or modified by the leaf function. For example, function @code{sin}
2705 is leaf, function @code{qsort} is not.
2707 Note that the leaf functions might invoke signals and signal handlers might be
2708 defined in the current compilation unit and use static variables. Only
2709 compliant way to write such a signal handler is to declare such variables
2712 The attribute has no effect on functions defined within current compilation
2713 unit. This is to allow easy merging of multiple compilation units into one,
2714 for example, by using the link time optimization. For this reason the
2715 attribute is not allowed on types to annotate indirect calls.
2717 @item long_call/short_call
2718 @cindex indirect calls on ARM
2719 This attribute specifies how a particular function is called on
2720 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2721 command-line switch and @code{#pragma long_calls} settings. The
2722 @code{long_call} attribute indicates that the function might be far
2723 away from the call site and require a different (more expensive)
2724 calling sequence. The @code{short_call} attribute always places
2725 the offset to the function from the call site into the @samp{BL}
2726 instruction directly.
2728 @item longcall/shortcall
2729 @cindex functions called via pointer on the RS/6000 and PowerPC
2730 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
2731 indicates that the function might be far away from the call site and
2732 require a different (more expensive) calling sequence. The
2733 @code{shortcall} attribute indicates that the function is always close
2734 enough for the shorter calling sequence to be used. These attributes
2735 override both the @option{-mlongcall} switch and, on the RS/6000 and
2736 PowerPC, the @code{#pragma longcall} setting.
2738 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2739 calls are necessary.
2741 @item long_call/near/far
2742 @cindex indirect calls on MIPS
2743 These attributes specify how a particular function is called on MIPS@.
2744 The attributes override the @option{-mlong-calls} (@pxref{MIPS Options})
2745 command-line switch. The @code{long_call} and @code{far} attributes are
2746 synonyms, and cause the compiler to always call
2747 the function by first loading its address into a register, and then using
2748 the contents of that register. The @code{near} attribute has the opposite
2749 effect; it specifies that non-PIC calls should be made using the more
2750 efficient @code{jal} instruction.
2753 @cindex @code{malloc} attribute
2754 The @code{malloc} attribute is used to tell the compiler that a function
2755 may be treated as if any non-@code{NULL} pointer it returns cannot
2756 alias any other pointer valid when the function returns.
2757 This will often improve optimization.
2758 Standard functions with this property include @code{malloc} and
2759 @code{calloc}. @code{realloc}-like functions have this property as
2760 long as the old pointer is never referred to (including comparing it
2761 to the new pointer) after the function returns a non-@code{NULL}
2764 @item mips16/nomips16
2765 @cindex @code{mips16} attribute
2766 @cindex @code{nomips16} attribute
2768 On MIPS targets, you can use the @code{mips16} and @code{nomips16}
2769 function attributes to locally select or turn off MIPS16 code generation.
2770 A function with the @code{mips16} attribute is emitted as MIPS16 code,
2771 while MIPS16 code generation is disabled for functions with the
2772 @code{nomips16} attribute. These attributes override the
2773 @option{-mips16} and @option{-mno-mips16} options on the command line
2774 (@pxref{MIPS Options}).
2776 When compiling files containing mixed MIPS16 and non-MIPS16 code, the
2777 preprocessor symbol @code{__mips16} reflects the setting on the command line,
2778 not that within individual functions. Mixed MIPS16 and non-MIPS16 code
2779 may interact badly with some GCC extensions such as @code{__builtin_apply}
2780 (@pxref{Constructing Calls}).
2782 @item model (@var{model-name})
2783 @cindex function addressability on the M32R/D
2784 @cindex variable addressability on the IA-64
2786 On the M32R/D, use this attribute to set the addressability of an
2787 object, and of the code generated for a function. The identifier
2788 @var{model-name} is one of @code{small}, @code{medium}, or
2789 @code{large}, representing each of the code models.
2791 Small model objects live in the lower 16MB of memory (so that their
2792 addresses can be loaded with the @code{ld24} instruction), and are
2793 callable with the @code{bl} instruction.
2795 Medium model objects may live anywhere in the 32-bit address space (the
2796 compiler will generate @code{seth/add3} instructions to load their addresses),
2797 and are callable with the @code{bl} instruction.
2799 Large model objects may live anywhere in the 32-bit address space (the
2800 compiler will generate @code{seth/add3} instructions to load their addresses),
2801 and may not be reachable with the @code{bl} instruction (the compiler will
2802 generate the much slower @code{seth/add3/jl} instruction sequence).
2804 On IA-64, use this attribute to set the addressability of an object.
2805 At present, the only supported identifier for @var{model-name} is
2806 @code{small}, indicating addressability via ``small'' (22-bit)
2807 addresses (so that their addresses can be loaded with the @code{addl}
2808 instruction). Caveat: such addressing is by definition not position
2809 independent and hence this attribute must not be used for objects
2810 defined by shared libraries.
2812 @item ms_abi/sysv_abi
2813 @cindex @code{ms_abi} attribute
2814 @cindex @code{sysv_abi} attribute
2816 On 64-bit x86_64-*-* targets, you can use an ABI attribute to indicate
2817 which calling convention should be used for a function. The @code{ms_abi}
2818 attribute tells the compiler to use the Microsoft ABI, while the
2819 @code{sysv_abi} attribute tells the compiler to use the ABI used on
2820 GNU/Linux and other systems. The default is to use the Microsoft ABI
2821 when targeting Windows. On all other systems, the default is the AMD ABI.
2823 Note, the @code{ms_abi} attribute for Windows targets currently requires
2824 the @option{-maccumulate-outgoing-args} option.
2826 @item ms_hook_prologue
2827 @cindex @code{ms_hook_prologue} attribute
2829 On 32 bit i[34567]86-*-* targets and 64 bit x86_64-*-* targets, you can use
2830 this function attribute to make gcc generate the "hot-patching" function
2831 prologue used in Win32 API functions in Microsoft Windows XP Service Pack 2
2835 @cindex function without a prologue/epilogue code
2836 Use this attribute on the ARM, AVR, MCORE, RX and SPU ports to indicate that
2837 the specified function does not need prologue/epilogue sequences generated by
2838 the compiler. It is up to the programmer to provide these sequences. The
2839 only statements that can be safely included in naked functions are
2840 @code{asm} statements that do not have operands. All other statements,
2841 including declarations of local variables, @code{if} statements, and so
2842 forth, should be avoided. Naked functions should be used to implement the
2843 body of an assembly function, while allowing the compiler to construct
2844 the requisite function declaration for the assembler.
2847 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2848 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2849 use the normal calling convention based on @code{jsr} and @code{rts}.
2850 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2853 On MeP targets this attribute causes the compiler to assume the called
2854 function is close enough to use the normal calling convention,
2855 overriding the @code{-mtf} command line option.
2858 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2859 Use this attribute together with @code{interrupt_handler},
2860 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2861 entry code should enable nested interrupts or exceptions.
2864 @cindex NMI handler functions on the Blackfin processor
2865 Use this attribute on the Blackfin to indicate that the specified function
2866 is an NMI handler. The compiler will generate function entry and
2867 exit sequences suitable for use in an NMI handler when this
2868 attribute is present.
2870 @item no_instrument_function
2871 @cindex @code{no_instrument_function} function attribute
2872 @opindex finstrument-functions
2873 If @option{-finstrument-functions} is given, profiling function calls will
2874 be generated at entry and exit of most user-compiled functions.
2875 Functions with this attribute will not be so instrumented.
2877 @item no_split_stack
2878 @cindex @code{no_split_stack} function attribute
2879 @opindex fsplit-stack
2880 If @option{-fsplit-stack} is given, functions will have a small
2881 prologue which decides whether to split the stack. Functions with the
2882 @code{no_split_stack} attribute will not have that prologue, and thus
2883 may run with only a small amount of stack space available.
2886 @cindex @code{noinline} function attribute
2887 This function attribute prevents a function from being considered for
2889 @c Don't enumerate the optimizations by name here; we try to be
2890 @c future-compatible with this mechanism.
2891 If the function does not have side-effects, there are optimizations
2892 other than inlining that causes function calls to be optimized away,
2893 although the function call is live. To keep such calls from being
2898 (@pxref{Extended Asm}) in the called function, to serve as a special
2902 @cindex @code{noclone} function attribute
2903 This function attribute prevents a function from being considered for
2904 cloning - a mechanism which produces specialized copies of functions
2905 and which is (currently) performed by interprocedural constant
2908 @item nonnull (@var{arg-index}, @dots{})
2909 @cindex @code{nonnull} function attribute
2910 The @code{nonnull} attribute specifies that some function parameters should
2911 be non-null pointers. For instance, the declaration:
2915 my_memcpy (void *dest, const void *src, size_t len)
2916 __attribute__((nonnull (1, 2)));
2920 causes the compiler to check that, in calls to @code{my_memcpy},
2921 arguments @var{dest} and @var{src} are non-null. If the compiler
2922 determines that a null pointer is passed in an argument slot marked
2923 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2924 is issued. The compiler may also choose to make optimizations based
2925 on the knowledge that certain function arguments will not be null.
2927 If no argument index list is given to the @code{nonnull} attribute,
2928 all pointer arguments are marked as non-null. To illustrate, the
2929 following declaration is equivalent to the previous example:
2933 my_memcpy (void *dest, const void *src, size_t len)
2934 __attribute__((nonnull));
2938 @cindex @code{noreturn} function attribute
2939 A few standard library functions, such as @code{abort} and @code{exit},
2940 cannot return. GCC knows this automatically. Some programs define
2941 their own functions that never return. You can declare them
2942 @code{noreturn} to tell the compiler this fact. For example,
2946 void fatal () __attribute__ ((noreturn));
2949 fatal (/* @r{@dots{}} */)
2951 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2957 The @code{noreturn} keyword tells the compiler to assume that
2958 @code{fatal} cannot return. It can then optimize without regard to what
2959 would happen if @code{fatal} ever did return. This makes slightly
2960 better code. More importantly, it helps avoid spurious warnings of
2961 uninitialized variables.
2963 The @code{noreturn} keyword does not affect the exceptional path when that
2964 applies: a @code{noreturn}-marked function may still return to the caller
2965 by throwing an exception or calling @code{longjmp}.
2967 Do not assume that registers saved by the calling function are
2968 restored before calling the @code{noreturn} function.
2970 It does not make sense for a @code{noreturn} function to have a return
2971 type other than @code{void}.
2973 The attribute @code{noreturn} is not implemented in GCC versions
2974 earlier than 2.5. An alternative way to declare that a function does
2975 not return, which works in the current version and in some older
2976 versions, is as follows:
2979 typedef void voidfn ();
2981 volatile voidfn fatal;
2984 This approach does not work in GNU C++.
2987 @cindex @code{nothrow} function attribute
2988 The @code{nothrow} attribute is used to inform the compiler that a
2989 function cannot throw an exception. For example, most functions in
2990 the standard C library can be guaranteed not to throw an exception
2991 with the notable exceptions of @code{qsort} and @code{bsearch} that
2992 take function pointer arguments. The @code{nothrow} attribute is not
2993 implemented in GCC versions earlier than 3.3.
2996 @cindex @code{optimize} function attribute
2997 The @code{optimize} attribute is used to specify that a function is to
2998 be compiled with different optimization options than specified on the
2999 command line. Arguments can either be numbers or strings. Numbers
3000 are assumed to be an optimization level. Strings that begin with
3001 @code{O} are assumed to be an optimization option, while other options
3002 are assumed to be used with a @code{-f} prefix. You can also use the
3003 @samp{#pragma GCC optimize} pragma to set the optimization options
3004 that affect more than one function.
3005 @xref{Function Specific Option Pragmas}, for details about the
3006 @samp{#pragma GCC optimize} pragma.
3008 This can be used for instance to have frequently executed functions
3009 compiled with more aggressive optimization options that produce faster
3010 and larger code, while other functions can be called with less
3014 @cindex @code{pcs} function attribute
3016 The @code{pcs} attribute can be used to control the calling convention
3017 used for a function on ARM. The attribute takes an argument that specifies
3018 the calling convention to use.
3020 When compiling using the AAPCS ABI (or a variant of that) then valid
3021 values for the argument are @code{"aapcs"} and @code{"aapcs-vfp"}. In
3022 order to use a variant other than @code{"aapcs"} then the compiler must
3023 be permitted to use the appropriate co-processor registers (i.e., the
3024 VFP registers must be available in order to use @code{"aapcs-vfp"}).
3028 /* Argument passed in r0, and result returned in r0+r1. */
3029 double f2d (float) __attribute__((pcs("aapcs")));
3032 Variadic functions always use the @code{"aapcs"} calling convention and
3033 the compiler will reject attempts to specify an alternative.
3036 @cindex @code{pure} function attribute
3037 Many functions have no effects except the return value and their
3038 return value depends only on the parameters and/or global variables.
3039 Such a function can be subject
3040 to common subexpression elimination and loop optimization just as an
3041 arithmetic operator would be. These functions should be declared
3042 with the attribute @code{pure}. For example,
3045 int square (int) __attribute__ ((pure));
3049 says that the hypothetical function @code{square} is safe to call
3050 fewer times than the program says.
3052 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
3053 Interesting non-pure functions are functions with infinite loops or those
3054 depending on volatile memory or other system resource, that may change between
3055 two consecutive calls (such as @code{feof} in a multithreading environment).
3057 The attribute @code{pure} is not implemented in GCC versions earlier
3061 @cindex @code{hot} function attribute
3062 The @code{hot} attribute is used to inform the compiler that a function is a
3063 hot spot of the compiled program. The function is optimized more aggressively
3064 and on many target it is placed into special subsection of the text section so
3065 all hot functions appears close together improving locality.
3067 When profile feedback is available, via @option{-fprofile-use}, hot functions
3068 are automatically detected and this attribute is ignored.
3070 The @code{hot} attribute is not implemented in GCC versions earlier
3074 @cindex @code{cold} function attribute
3075 The @code{cold} attribute is used to inform the compiler that a function is
3076 unlikely executed. The function is optimized for size rather than speed and on
3077 many targets it is placed into special subsection of the text section so all
3078 cold functions appears close together improving code locality of non-cold parts
3079 of program. The paths leading to call of cold functions within code are marked
3080 as unlikely by the branch prediction mechanism. It is thus useful to mark
3081 functions used to handle unlikely conditions, such as @code{perror}, as cold to
3082 improve optimization of hot functions that do call marked functions in rare
3085 When profile feedback is available, via @option{-fprofile-use}, hot functions
3086 are automatically detected and this attribute is ignored.
3088 The @code{cold} attribute is not implemented in GCC versions earlier than 4.3.
3090 @item regparm (@var{number})
3091 @cindex @code{regparm} attribute
3092 @cindex functions that are passed arguments in registers on the 386
3093 On the Intel 386, the @code{regparm} attribute causes the compiler to
3094 pass arguments number one to @var{number} if they are of integral type
3095 in registers EAX, EDX, and ECX instead of on the stack. Functions that
3096 take a variable number of arguments will continue to be passed all of their
3097 arguments on the stack.
3099 Beware that on some ELF systems this attribute is unsuitable for
3100 global functions in shared libraries with lazy binding (which is the
3101 default). Lazy binding will send the first call via resolving code in
3102 the loader, which might assume EAX, EDX and ECX can be clobbered, as
3103 per the standard calling conventions. Solaris 8 is affected by this.
3104 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
3105 safe since the loaders there save EAX, EDX and ECX. (Lazy binding can be
3106 disabled with the linker or the loader if desired, to avoid the
3110 @cindex @code{sseregparm} attribute
3111 On the Intel 386 with SSE support, the @code{sseregparm} attribute
3112 causes the compiler to pass up to 3 floating point arguments in
3113 SSE registers instead of on the stack. Functions that take a
3114 variable number of arguments will continue to pass all of their
3115 floating point arguments on the stack.
3117 @item force_align_arg_pointer
3118 @cindex @code{force_align_arg_pointer} attribute
3119 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
3120 applied to individual function definitions, generating an alternate
3121 prologue and epilogue that realigns the runtime stack if necessary.
3122 This supports mixing legacy codes that run with a 4-byte aligned stack
3123 with modern codes that keep a 16-byte stack for SSE compatibility.
3126 @cindex @code{resbank} attribute
3127 On the SH2A target, this attribute enables the high-speed register
3128 saving and restoration using a register bank for @code{interrupt_handler}
3129 routines. Saving to the bank is performed automatically after the CPU
3130 accepts an interrupt that uses a register bank.
3132 The nineteen 32-bit registers comprising general register R0 to R14,
3133 control register GBR, and system registers MACH, MACL, and PR and the
3134 vector table address offset are saved into a register bank. Register
3135 banks are stacked in first-in last-out (FILO) sequence. Restoration
3136 from the bank is executed by issuing a RESBANK instruction.
3139 @cindex @code{returns_twice} attribute
3140 The @code{returns_twice} attribute tells the compiler that a function may
3141 return more than one time. The compiler will ensure that all registers
3142 are dead before calling such a function and will emit a warning about
3143 the variables that may be clobbered after the second return from the
3144 function. Examples of such functions are @code{setjmp} and @code{vfork}.
3145 The @code{longjmp}-like counterpart of such function, if any, might need
3146 to be marked with the @code{noreturn} attribute.
3149 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
3150 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
3151 all registers except the stack pointer should be saved in the prologue
3152 regardless of whether they are used or not.
3154 @item save_volatiles
3155 @cindex save volatile registers on the MicroBlaze
3156 Use this attribute on the MicroBlaze to indicate that the function is
3157 an interrupt handler. All volatile registers (in addition to non-volatile
3158 registers) will be saved in the function prologue. If the function is a leaf
3159 function, only volatiles used by the function are saved. A normal function
3160 return is generated instead of a return from interrupt.
3162 @item section ("@var{section-name}")
3163 @cindex @code{section} function attribute
3164 Normally, the compiler places the code it generates in the @code{text} section.
3165 Sometimes, however, you need additional sections, or you need certain
3166 particular functions to appear in special sections. The @code{section}
3167 attribute specifies that a function lives in a particular section.
3168 For example, the declaration:
3171 extern void foobar (void) __attribute__ ((section ("bar")));
3175 puts the function @code{foobar} in the @code{bar} section.
3177 Some file formats do not support arbitrary sections so the @code{section}
3178 attribute is not available on all platforms.
3179 If you need to map the entire contents of a module to a particular
3180 section, consider using the facilities of the linker instead.
3183 @cindex @code{sentinel} function attribute
3184 This function attribute ensures that a parameter in a function call is
3185 an explicit @code{NULL}. The attribute is only valid on variadic
3186 functions. By default, the sentinel is located at position zero, the
3187 last parameter of the function call. If an optional integer position
3188 argument P is supplied to the attribute, the sentinel must be located at
3189 position P counting backwards from the end of the argument list.
3192 __attribute__ ((sentinel))
3194 __attribute__ ((sentinel(0)))
3197 The attribute is automatically set with a position of 0 for the built-in
3198 functions @code{execl} and @code{execlp}. The built-in function
3199 @code{execle} has the attribute set with a position of 1.
3201 A valid @code{NULL} in this context is defined as zero with any pointer
3202 type. If your system defines the @code{NULL} macro with an integer type
3203 then you need to add an explicit cast. GCC replaces @code{stddef.h}
3204 with a copy that redefines NULL appropriately.
3206 The warnings for missing or incorrect sentinels are enabled with
3210 See long_call/short_call.
3213 See longcall/shortcall.
3216 @cindex signal handler functions on the AVR processors
3217 Use this attribute on the AVR to indicate that the specified
3218 function is a signal handler. The compiler will generate function
3219 entry and exit sequences suitable for use in a signal handler when this
3220 attribute is present. Interrupts will be disabled inside the function.
3223 Use this attribute on the SH to indicate an @code{interrupt_handler}
3224 function should switch to an alternate stack. It expects a string
3225 argument that names a global variable holding the address of the
3230 void f () __attribute__ ((interrupt_handler,
3231 sp_switch ("alt_stack")));
3235 @cindex functions that pop the argument stack on the 386
3236 On the Intel 386, the @code{stdcall} attribute causes the compiler to
3237 assume that the called function will pop off the stack space used to
3238 pass arguments, unless it takes a variable number of arguments.
3240 @item syscall_linkage
3241 @cindex @code{syscall_linkage} attribute
3242 This attribute is used to modify the IA64 calling convention by marking
3243 all input registers as live at all function exits. This makes it possible
3244 to restart a system call after an interrupt without having to save/restore
3245 the input registers. This also prevents kernel data from leaking into
3249 @cindex @code{target} function attribute
3250 The @code{target} attribute is used to specify that a function is to
3251 be compiled with different target options than specified on the
3252 command line. This can be used for instance to have functions
3253 compiled with a different ISA (instruction set architecture) than the
3254 default. You can also use the @samp{#pragma GCC target} pragma to set
3255 more than one function to be compiled with specific target options.
3256 @xref{Function Specific Option Pragmas}, for details about the
3257 @samp{#pragma GCC target} pragma.
3259 For instance on a 386, you could compile one function with
3260 @code{target("sse4.1,arch=core2")} and another with
3261 @code{target("sse4a,arch=amdfam10")} that would be equivalent to
3262 compiling the first function with @option{-msse4.1} and
3263 @option{-march=core2} options, and the second function with
3264 @option{-msse4a} and @option{-march=amdfam10} options. It is up to the
3265 user to make sure that a function is only invoked on a machine that
3266 supports the particular ISA it was compiled for (for example by using
3267 @code{cpuid} on 386 to determine what feature bits and architecture
3271 int core2_func (void) __attribute__ ((__target__ ("arch=core2")));
3272 int sse3_func (void) __attribute__ ((__target__ ("sse3")));
3275 On the 386, the following options are allowed:
3280 @cindex @code{target("abm")} attribute
3281 Enable/disable the generation of the advanced bit instructions.
3285 @cindex @code{target("aes")} attribute
3286 Enable/disable the generation of the AES instructions.
3290 @cindex @code{target("mmx")} attribute
3291 Enable/disable the generation of the MMX instructions.
3295 @cindex @code{target("pclmul")} attribute
3296 Enable/disable the generation of the PCLMUL instructions.
3300 @cindex @code{target("popcnt")} attribute
3301 Enable/disable the generation of the POPCNT instruction.
3305 @cindex @code{target("sse")} attribute
3306 Enable/disable the generation of the SSE instructions.
3310 @cindex @code{target("sse2")} attribute
3311 Enable/disable the generation of the SSE2 instructions.
3315 @cindex @code{target("sse3")} attribute
3316 Enable/disable the generation of the SSE3 instructions.
3320 @cindex @code{target("sse4")} attribute
3321 Enable/disable the generation of the SSE4 instructions (both SSE4.1
3326 @cindex @code{target("sse4.1")} attribute
3327 Enable/disable the generation of the sse4.1 instructions.
3331 @cindex @code{target("sse4.2")} attribute
3332 Enable/disable the generation of the sse4.2 instructions.
3336 @cindex @code{target("sse4a")} attribute
3337 Enable/disable the generation of the SSE4A instructions.
3341 @cindex @code{target("fma4")} attribute
3342 Enable/disable the generation of the FMA4 instructions.
3346 @cindex @code{target("xop")} attribute
3347 Enable/disable the generation of the XOP instructions.
3351 @cindex @code{target("lwp")} attribute
3352 Enable/disable the generation of the LWP instructions.
3356 @cindex @code{target("ssse3")} attribute
3357 Enable/disable the generation of the SSSE3 instructions.
3361 @cindex @code{target("cld")} attribute
3362 Enable/disable the generation of the CLD before string moves.
3364 @item fancy-math-387
3365 @itemx no-fancy-math-387
3366 @cindex @code{target("fancy-math-387")} attribute
3367 Enable/disable the generation of the @code{sin}, @code{cos}, and
3368 @code{sqrt} instructions on the 387 floating point unit.
3371 @itemx no-fused-madd
3372 @cindex @code{target("fused-madd")} attribute
3373 Enable/disable the generation of the fused multiply/add instructions.
3377 @cindex @code{target("ieee-fp")} attribute
3378 Enable/disable the generation of floating point that depends on IEEE arithmetic.
3380 @item inline-all-stringops
3381 @itemx no-inline-all-stringops
3382 @cindex @code{target("inline-all-stringops")} attribute
3383 Enable/disable inlining of string operations.
3385 @item inline-stringops-dynamically
3386 @itemx no-inline-stringops-dynamically
3387 @cindex @code{target("inline-stringops-dynamically")} attribute
3388 Enable/disable the generation of the inline code to do small string
3389 operations and calling the library routines for large operations.
3391 @item align-stringops
3392 @itemx no-align-stringops
3393 @cindex @code{target("align-stringops")} attribute
3394 Do/do not align destination of inlined string operations.
3398 @cindex @code{target("recip")} attribute
3399 Enable/disable the generation of RCPSS, RCPPS, RSQRTSS and RSQRTPS
3400 instructions followed an additional Newton-Raphson step instead of
3401 doing a floating point division.
3403 @item arch=@var{ARCH}
3404 @cindex @code{target("arch=@var{ARCH}")} attribute
3405 Specify the architecture to generate code for in compiling the function.
3407 @item tune=@var{TUNE}
3408 @cindex @code{target("tune=@var{TUNE}")} attribute
3409 Specify the architecture to tune for in compiling the function.
3411 @item fpmath=@var{FPMATH}
3412 @cindex @code{target("fpmath=@var{FPMATH}")} attribute
3413 Specify which floating point unit to use. The
3414 @code{target("fpmath=sse,387")} option must be specified as
3415 @code{target("fpmath=sse+387")} because the comma would separate
3419 On the 386, you can use either multiple strings to specify multiple
3420 options, or you can separate the option with a comma (@code{,}).
3422 On the 386, the inliner will not inline a function that has different
3423 target options than the caller, unless the callee has a subset of the
3424 target options of the caller. For example a function declared with
3425 @code{target("sse3")} can inline a function with
3426 @code{target("sse2")}, since @code{-msse3} implies @code{-msse2}.
3428 The @code{target} attribute is not implemented in GCC versions earlier
3429 than 4.4, and at present only the 386 uses it.
3432 @cindex tiny data section on the H8/300H and H8S
3433 Use this attribute on the H8/300H and H8S to indicate that the specified
3434 variable should be placed into the tiny data section.
3435 The compiler will generate more efficient code for loads and stores
3436 on data in the tiny data section. Note the tiny data area is limited to
3437 slightly under 32kbytes of data.
3440 Use this attribute on the SH for an @code{interrupt_handler} to return using
3441 @code{trapa} instead of @code{rte}. This attribute expects an integer
3442 argument specifying the trap number to be used.
3445 @cindex @code{unused} attribute.
3446 This attribute, attached to a function, means that the function is meant
3447 to be possibly unused. GCC will not produce a warning for this
3451 @cindex @code{used} attribute.
3452 This attribute, attached to a function, means that code must be emitted
3453 for the function even if it appears that the function is not referenced.
3454 This is useful, for example, when the function is referenced only in
3458 @cindex @code{version_id} attribute
3459 This IA64 HP-UX attribute, attached to a global variable or function, renames a
3460 symbol to contain a version string, thus allowing for function level
3461 versioning. HP-UX system header files may use version level functioning
3462 for some system calls.
3465 extern int foo () __attribute__((version_id ("20040821")));
3468 Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}.
3470 @item visibility ("@var{visibility_type}")
3471 @cindex @code{visibility} attribute
3472 This attribute affects the linkage of the declaration to which it is attached.
3473 There are four supported @var{visibility_type} values: default,
3474 hidden, protected or internal visibility.
3477 void __attribute__ ((visibility ("protected")))
3478 f () @{ /* @r{Do something.} */; @}
3479 int i __attribute__ ((visibility ("hidden")));
3482 The possible values of @var{visibility_type} correspond to the
3483 visibility settings in the ELF gABI.
3486 @c keep this list of visibilities in alphabetical order.
3489 Default visibility is the normal case for the object file format.
3490 This value is available for the visibility attribute to override other
3491 options that may change the assumed visibility of entities.
3493 On ELF, default visibility means that the declaration is visible to other
3494 modules and, in shared libraries, means that the declared entity may be
3497 On Darwin, default visibility means that the declaration is visible to
3500 Default visibility corresponds to ``external linkage'' in the language.
3503 Hidden visibility indicates that the entity declared will have a new
3504 form of linkage, which we'll call ``hidden linkage''. Two
3505 declarations of an object with hidden linkage refer to the same object
3506 if they are in the same shared object.
3509 Internal visibility is like hidden visibility, but with additional
3510 processor specific semantics. Unless otherwise specified by the
3511 psABI, GCC defines internal visibility to mean that a function is
3512 @emph{never} called from another module. Compare this with hidden
3513 functions which, while they cannot be referenced directly by other
3514 modules, can be referenced indirectly via function pointers. By
3515 indicating that a function cannot be called from outside the module,
3516 GCC may for instance omit the load of a PIC register since it is known
3517 that the calling function loaded the correct value.
3520 Protected visibility is like default visibility except that it
3521 indicates that references within the defining module will bind to the
3522 definition in that module. That is, the declared entity cannot be
3523 overridden by another module.
3527 All visibilities are supported on many, but not all, ELF targets
3528 (supported when the assembler supports the @samp{.visibility}
3529 pseudo-op). Default visibility is supported everywhere. Hidden
3530 visibility is supported on Darwin targets.
3532 The visibility attribute should be applied only to declarations which
3533 would otherwise have external linkage. The attribute should be applied
3534 consistently, so that the same entity should not be declared with
3535 different settings of the attribute.
3537 In C++, the visibility attribute applies to types as well as functions
3538 and objects, because in C++ types have linkage. A class must not have
3539 greater visibility than its non-static data member types and bases,
3540 and class members default to the visibility of their class. Also, a
3541 declaration without explicit visibility is limited to the visibility
3544 In C++, you can mark member functions and static member variables of a
3545 class with the visibility attribute. This is useful if you know a
3546 particular method or static member variable should only be used from
3547 one shared object; then you can mark it hidden while the rest of the
3548 class has default visibility. Care must be taken to avoid breaking
3549 the One Definition Rule; for example, it is usually not useful to mark
3550 an inline method as hidden without marking the whole class as hidden.
3552 A C++ namespace declaration can also have the visibility attribute.
3553 This attribute applies only to the particular namespace body, not to
3554 other definitions of the same namespace; it is equivalent to using
3555 @samp{#pragma GCC visibility} before and after the namespace
3556 definition (@pxref{Visibility Pragmas}).
3558 In C++, if a template argument has limited visibility, this
3559 restriction is implicitly propagated to the template instantiation.
3560 Otherwise, template instantiations and specializations default to the
3561 visibility of their template.
3563 If both the template and enclosing class have explicit visibility, the
3564 visibility from the template is used.
3567 @cindex @code{vliw} attribute
3568 On MeP, the @code{vliw} attribute tells the compiler to emit
3569 instructions in VLIW mode instead of core mode. Note that this
3570 attribute is not allowed unless a VLIW coprocessor has been configured
3571 and enabled through command line options.
3573 @item warn_unused_result
3574 @cindex @code{warn_unused_result} attribute
3575 The @code{warn_unused_result} attribute causes a warning to be emitted
3576 if a caller of the function with this attribute does not use its
3577 return value. This is useful for functions where not checking
3578 the result is either a security problem or always a bug, such as
3582 int fn () __attribute__ ((warn_unused_result));
3585 if (fn () < 0) return -1;
3591 results in warning on line 5.
3594 @cindex @code{weak} attribute
3595 The @code{weak} attribute causes the declaration to be emitted as a weak
3596 symbol rather than a global. This is primarily useful in defining
3597 library functions which can be overridden in user code, though it can
3598 also be used with non-function declarations. Weak symbols are supported
3599 for ELF targets, and also for a.out targets when using the GNU assembler
3603 @itemx weakref ("@var{target}")
3604 @cindex @code{weakref} attribute
3605 The @code{weakref} attribute marks a declaration as a weak reference.
3606 Without arguments, it should be accompanied by an @code{alias} attribute
3607 naming the target symbol. Optionally, the @var{target} may be given as
3608 an argument to @code{weakref} itself. In either case, @code{weakref}
3609 implicitly marks the declaration as @code{weak}. Without a
3610 @var{target}, given as an argument to @code{weakref} or to @code{alias},
3611 @code{weakref} is equivalent to @code{weak}.
3614 static int x() __attribute__ ((weakref ("y")));
3615 /* is equivalent to... */
3616 static int x() __attribute__ ((weak, weakref, alias ("y")));
3618 static int x() __attribute__ ((weakref));
3619 static int x() __attribute__ ((alias ("y")));
3622 A weak reference is an alias that does not by itself require a
3623 definition to be given for the target symbol. If the target symbol is
3624 only referenced through weak references, then it becomes a @code{weak}
3625 undefined symbol. If it is directly referenced, however, then such
3626 strong references prevail, and a definition will be required for the
3627 symbol, not necessarily in the same translation unit.
3629 The effect is equivalent to moving all references to the alias to a
3630 separate translation unit, renaming the alias to the aliased symbol,
3631 declaring it as weak, compiling the two separate translation units and
3632 performing a reloadable link on them.
3634 At present, a declaration to which @code{weakref} is attached can
3635 only be @code{static}.
3639 You can specify multiple attributes in a declaration by separating them
3640 by commas within the double parentheses or by immediately following an
3641 attribute declaration with another attribute declaration.
3643 @cindex @code{#pragma}, reason for not using
3644 @cindex pragma, reason for not using
3645 Some people object to the @code{__attribute__} feature, suggesting that
3646 ISO C's @code{#pragma} should be used instead. At the time
3647 @code{__attribute__} was designed, there were two reasons for not doing
3652 It is impossible to generate @code{#pragma} commands from a macro.
3655 There is no telling what the same @code{#pragma} might mean in another
3659 These two reasons applied to almost any application that might have been
3660 proposed for @code{#pragma}. It was basically a mistake to use
3661 @code{#pragma} for @emph{anything}.
3663 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
3664 to be generated from macros. In addition, a @code{#pragma GCC}
3665 namespace is now in use for GCC-specific pragmas. However, it has been
3666 found convenient to use @code{__attribute__} to achieve a natural
3667 attachment of attributes to their corresponding declarations, whereas
3668 @code{#pragma GCC} is of use for constructs that do not naturally form
3669 part of the grammar. @xref{Other Directives,,Miscellaneous
3670 Preprocessing Directives, cpp, The GNU C Preprocessor}.
3672 @node Attribute Syntax
3673 @section Attribute Syntax
3674 @cindex attribute syntax
3676 This section describes the syntax with which @code{__attribute__} may be
3677 used, and the constructs to which attribute specifiers bind, for the C
3678 language. Some details may vary for C++ and Objective-C@. Because of
3679 infelicities in the grammar for attributes, some forms described here
3680 may not be successfully parsed in all cases.
3682 There are some problems with the semantics of attributes in C++. For
3683 example, there are no manglings for attributes, although they may affect
3684 code generation, so problems may arise when attributed types are used in
3685 conjunction with templates or overloading. Similarly, @code{typeid}
3686 does not distinguish between types with different attributes. Support
3687 for attributes in C++ may be restricted in future to attributes on
3688 declarations only, but not on nested declarators.
3690 @xref{Function Attributes}, for details of the semantics of attributes
3691 applying to functions. @xref{Variable Attributes}, for details of the
3692 semantics of attributes applying to variables. @xref{Type Attributes},
3693 for details of the semantics of attributes applying to structure, union
3694 and enumerated types.
3696 An @dfn{attribute specifier} is of the form
3697 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
3698 is a possibly empty comma-separated sequence of @dfn{attributes}, where
3699 each attribute is one of the following:
3703 Empty. Empty attributes are ignored.
3706 A word (which may be an identifier such as @code{unused}, or a reserved
3707 word such as @code{const}).
3710 A word, followed by, in parentheses, parameters for the attribute.
3711 These parameters take one of the following forms:
3715 An identifier. For example, @code{mode} attributes use this form.
3718 An identifier followed by a comma and a non-empty comma-separated list
3719 of expressions. For example, @code{format} attributes use this form.
3722 A possibly empty comma-separated list of expressions. For example,
3723 @code{format_arg} attributes use this form with the list being a single
3724 integer constant expression, and @code{alias} attributes use this form
3725 with the list being a single string constant.
3729 An @dfn{attribute specifier list} is a sequence of one or more attribute
3730 specifiers, not separated by any other tokens.
3732 In GNU C, an attribute specifier list may appear after the colon following a
3733 label, other than a @code{case} or @code{default} label. The only
3734 attribute it makes sense to use after a label is @code{unused}. This
3735 feature is intended for code generated by programs which contains labels
3736 that may be unused but which is compiled with @option{-Wall}. It would
3737 not normally be appropriate to use in it human-written code, though it
3738 could be useful in cases where the code that jumps to the label is
3739 contained within an @code{#ifdef} conditional. GNU C++ only permits
3740 attributes on labels if the attribute specifier is immediately
3741 followed by a semicolon (i.e., the label applies to an empty
3742 statement). If the semicolon is missing, C++ label attributes are
3743 ambiguous, as it is permissible for a declaration, which could begin
3744 with an attribute list, to be labelled in C++. Declarations cannot be
3745 labelled in C90 or C99, so the ambiguity does not arise there.
3747 An attribute specifier list may appear as part of a @code{struct},
3748 @code{union} or @code{enum} specifier. It may go either immediately
3749 after the @code{struct}, @code{union} or @code{enum} keyword, or after
3750 the closing brace. The former syntax is preferred.
3751 Where attribute specifiers follow the closing brace, they are considered
3752 to relate to the structure, union or enumerated type defined, not to any
3753 enclosing declaration the type specifier appears in, and the type
3754 defined is not complete until after the attribute specifiers.
3755 @c Otherwise, there would be the following problems: a shift/reduce
3756 @c conflict between attributes binding the struct/union/enum and
3757 @c binding to the list of specifiers/qualifiers; and "aligned"
3758 @c attributes could use sizeof for the structure, but the size could be
3759 @c changed later by "packed" attributes.
3761 Otherwise, an attribute specifier appears as part of a declaration,
3762 counting declarations of unnamed parameters and type names, and relates
3763 to that declaration (which may be nested in another declaration, for
3764 example in the case of a parameter declaration), or to a particular declarator
3765 within a declaration. Where an
3766 attribute specifier is applied to a parameter declared as a function or
3767 an array, it should apply to the function or array rather than the
3768 pointer to which the parameter is implicitly converted, but this is not
3769 yet correctly implemented.
3771 Any list of specifiers and qualifiers at the start of a declaration may
3772 contain attribute specifiers, whether or not such a list may in that
3773 context contain storage class specifiers. (Some attributes, however,
3774 are essentially in the nature of storage class specifiers, and only make
3775 sense where storage class specifiers may be used; for example,
3776 @code{section}.) There is one necessary limitation to this syntax: the
3777 first old-style parameter declaration in a function definition cannot
3778 begin with an attribute specifier, because such an attribute applies to
3779 the function instead by syntax described below (which, however, is not
3780 yet implemented in this case). In some other cases, attribute
3781 specifiers are permitted by this grammar but not yet supported by the
3782 compiler. All attribute specifiers in this place relate to the
3783 declaration as a whole. In the obsolescent usage where a type of
3784 @code{int} is implied by the absence of type specifiers, such a list of
3785 specifiers and qualifiers may be an attribute specifier list with no
3786 other specifiers or qualifiers.
3788 At present, the first parameter in a function prototype must have some
3789 type specifier which is not an attribute specifier; this resolves an
3790 ambiguity in the interpretation of @code{void f(int
3791 (__attribute__((foo)) x))}, but is subject to change. At present, if
3792 the parentheses of a function declarator contain only attributes then
3793 those attributes are ignored, rather than yielding an error or warning
3794 or implying a single parameter of type int, but this is subject to
3797 An attribute specifier list may appear immediately before a declarator
3798 (other than the first) in a comma-separated list of declarators in a
3799 declaration of more than one identifier using a single list of
3800 specifiers and qualifiers. Such attribute specifiers apply
3801 only to the identifier before whose declarator they appear. For
3805 __attribute__((noreturn)) void d0 (void),
3806 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
3811 the @code{noreturn} attribute applies to all the functions
3812 declared; the @code{format} attribute only applies to @code{d1}.
3814 An attribute specifier list may appear immediately before the comma,
3815 @code{=} or semicolon terminating the declaration of an identifier other
3816 than a function definition. Such attribute specifiers apply
3817 to the declared object or function. Where an
3818 assembler name for an object or function is specified (@pxref{Asm
3819 Labels}), the attribute must follow the @code{asm}
3822 An attribute specifier list may, in future, be permitted to appear after
3823 the declarator in a function definition (before any old-style parameter
3824 declarations or the function body).
3826 Attribute specifiers may be mixed with type qualifiers appearing inside
3827 the @code{[]} of a parameter array declarator, in the C99 construct by
3828 which such qualifiers are applied to the pointer to which the array is
3829 implicitly converted. Such attribute specifiers apply to the pointer,
3830 not to the array, but at present this is not implemented and they are
3833 An attribute specifier list may appear at the start of a nested
3834 declarator. At present, there are some limitations in this usage: the
3835 attributes correctly apply to the declarator, but for most individual
3836 attributes the semantics this implies are not implemented.
3837 When attribute specifiers follow the @code{*} of a pointer
3838 declarator, they may be mixed with any type qualifiers present.
3839 The following describes the formal semantics of this syntax. It will make the
3840 most sense if you are familiar with the formal specification of
3841 declarators in the ISO C standard.
3843 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
3844 D1}, where @code{T} contains declaration specifiers that specify a type
3845 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
3846 contains an identifier @var{ident}. The type specified for @var{ident}
3847 for derived declarators whose type does not include an attribute
3848 specifier is as in the ISO C standard.
3850 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
3851 and the declaration @code{T D} specifies the type
3852 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3853 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3854 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
3856 If @code{D1} has the form @code{*
3857 @var{type-qualifier-and-attribute-specifier-list} D}, and the
3858 declaration @code{T D} specifies the type
3859 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3860 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3861 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
3867 void (__attribute__((noreturn)) ****f) (void);
3871 specifies the type ``pointer to pointer to pointer to pointer to
3872 non-returning function returning @code{void}''. As another example,
3875 char *__attribute__((aligned(8))) *f;
3879 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
3880 Note again that this does not work with most attributes; for example,
3881 the usage of @samp{aligned} and @samp{noreturn} attributes given above
3882 is not yet supported.
3884 For compatibility with existing code written for compiler versions that
3885 did not implement attributes on nested declarators, some laxity is
3886 allowed in the placing of attributes. If an attribute that only applies
3887 to types is applied to a declaration, it will be treated as applying to
3888 the type of that declaration. If an attribute that only applies to
3889 declarations is applied to the type of a declaration, it will be treated
3890 as applying to that declaration; and, for compatibility with code
3891 placing the attributes immediately before the identifier declared, such
3892 an attribute applied to a function return type will be treated as
3893 applying to the function type, and such an attribute applied to an array
3894 element type will be treated as applying to the array type. If an
3895 attribute that only applies to function types is applied to a
3896 pointer-to-function type, it will be treated as applying to the pointer
3897 target type; if such an attribute is applied to a function return type
3898 that is not a pointer-to-function type, it will be treated as applying
3899 to the function type.
3901 @node Function Prototypes
3902 @section Prototypes and Old-Style Function Definitions
3903 @cindex function prototype declarations
3904 @cindex old-style function definitions
3905 @cindex promotion of formal parameters
3907 GNU C extends ISO C to allow a function prototype to override a later
3908 old-style non-prototype definition. Consider the following example:
3911 /* @r{Use prototypes unless the compiler is old-fashioned.} */
3918 /* @r{Prototype function declaration.} */
3919 int isroot P((uid_t));
3921 /* @r{Old-style function definition.} */
3923 isroot (x) /* @r{??? lossage here ???} */
3930 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
3931 not allow this example, because subword arguments in old-style
3932 non-prototype definitions are promoted. Therefore in this example the
3933 function definition's argument is really an @code{int}, which does not
3934 match the prototype argument type of @code{short}.
3936 This restriction of ISO C makes it hard to write code that is portable
3937 to traditional C compilers, because the programmer does not know
3938 whether the @code{uid_t} type is @code{short}, @code{int}, or
3939 @code{long}. Therefore, in cases like these GNU C allows a prototype
3940 to override a later old-style definition. More precisely, in GNU C, a
3941 function prototype argument type overrides the argument type specified
3942 by a later old-style definition if the former type is the same as the
3943 latter type before promotion. Thus in GNU C the above example is
3944 equivalent to the following:
3957 GNU C++ does not support old-style function definitions, so this
3958 extension is irrelevant.
3961 @section C++ Style Comments
3963 @cindex C++ comments
3964 @cindex comments, C++ style
3966 In GNU C, you may use C++ style comments, which start with @samp{//} and
3967 continue until the end of the line. Many other C implementations allow
3968 such comments, and they are included in the 1999 C standard. However,
3969 C++ style comments are not recognized if you specify an @option{-std}
3970 option specifying a version of ISO C before C99, or @option{-ansi}
3971 (equivalent to @option{-std=c90}).
3974 @section Dollar Signs in Identifier Names
3976 @cindex dollar signs in identifier names
3977 @cindex identifier names, dollar signs in
3979 In GNU C, you may normally use dollar signs in identifier names.
3980 This is because many traditional C implementations allow such identifiers.
3981 However, dollar signs in identifiers are not supported on a few target
3982 machines, typically because the target assembler does not allow them.
3984 @node Character Escapes
3985 @section The Character @key{ESC} in Constants
3987 You can use the sequence @samp{\e} in a string or character constant to
3988 stand for the ASCII character @key{ESC}.
3991 @section Inquiring on Alignment of Types or Variables
3993 @cindex type alignment
3994 @cindex variable alignment
3996 The keyword @code{__alignof__} allows you to inquire about how an object
3997 is aligned, or the minimum alignment usually required by a type. Its
3998 syntax is just like @code{sizeof}.
4000 For example, if the target machine requires a @code{double} value to be
4001 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
4002 This is true on many RISC machines. On more traditional machine
4003 designs, @code{__alignof__ (double)} is 4 or even 2.
4005 Some machines never actually require alignment; they allow reference to any
4006 data type even at an odd address. For these machines, @code{__alignof__}
4007 reports the smallest alignment that GCC will give the data type, usually as
4008 mandated by the target ABI.
4010 If the operand of @code{__alignof__} is an lvalue rather than a type,
4011 its value is the required alignment for its type, taking into account
4012 any minimum alignment specified with GCC's @code{__attribute__}
4013 extension (@pxref{Variable Attributes}). For example, after this
4017 struct foo @{ int x; char y; @} foo1;
4021 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
4022 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
4024 It is an error to ask for the alignment of an incomplete type.
4026 @node Variable Attributes
4027 @section Specifying Attributes of Variables
4028 @cindex attribute of variables
4029 @cindex variable attributes
4031 The keyword @code{__attribute__} allows you to specify special
4032 attributes of variables or structure fields. This keyword is followed
4033 by an attribute specification inside double parentheses. Some
4034 attributes are currently defined generically for variables.
4035 Other attributes are defined for variables on particular target
4036 systems. Other attributes are available for functions
4037 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
4038 Other front ends might define more attributes
4039 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
4041 You may also specify attributes with @samp{__} preceding and following
4042 each keyword. This allows you to use them in header files without
4043 being concerned about a possible macro of the same name. For example,
4044 you may use @code{__aligned__} instead of @code{aligned}.
4046 @xref{Attribute Syntax}, for details of the exact syntax for using
4050 @cindex @code{aligned} attribute
4051 @item aligned (@var{alignment})
4052 This attribute specifies a minimum alignment for the variable or
4053 structure field, measured in bytes. For example, the declaration:
4056 int x __attribute__ ((aligned (16))) = 0;
4060 causes the compiler to allocate the global variable @code{x} on a
4061 16-byte boundary. On a 68040, this could be used in conjunction with
4062 an @code{asm} expression to access the @code{move16} instruction which
4063 requires 16-byte aligned operands.
4065 You can also specify the alignment of structure fields. For example, to
4066 create a double-word aligned @code{int} pair, you could write:
4069 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
4073 This is an alternative to creating a union with a @code{double} member
4074 that forces the union to be double-word aligned.
4076 As in the preceding examples, you can explicitly specify the alignment
4077 (in bytes) that you wish the compiler to use for a given variable or
4078 structure field. Alternatively, you can leave out the alignment factor
4079 and just ask the compiler to align a variable or field to the
4080 default alignment for the target architecture you are compiling for.
4081 The default alignment is sufficient for all scalar types, but may not be
4082 enough for all vector types on a target which supports vector operations.
4083 The default alignment is fixed for a particular target ABI.
4085 Gcc also provides a target specific macro @code{__BIGGEST_ALIGNMENT__},
4086 which is the largest alignment ever used for any data type on the
4087 target machine you are compiling for. For example, you could write:
4090 short array[3] __attribute__ ((aligned (__BIGGEST_ALIGNMENT__)));
4093 The compiler automatically sets the alignment for the declared
4094 variable or field to @code{__BIGGEST_ALIGNMENT__}. Doing this can
4095 often make copy operations more efficient, because the compiler can
4096 use whatever instructions copy the biggest chunks of memory when
4097 performing copies to or from the variables or fields that you have
4098 aligned this way. Note that the value of @code{__BIGGEST_ALIGNMENT__}
4099 may change depending on command line options.
4101 When used on a struct, or struct member, the @code{aligned} attribute can
4102 only increase the alignment; in order to decrease it, the @code{packed}
4103 attribute must be specified as well. When used as part of a typedef, the
4104 @code{aligned} attribute can both increase and decrease alignment, and
4105 specifying the @code{packed} attribute will generate a warning.
4107 Note that the effectiveness of @code{aligned} attributes may be limited
4108 by inherent limitations in your linker. On many systems, the linker is
4109 only able to arrange for variables to be aligned up to a certain maximum
4110 alignment. (For some linkers, the maximum supported alignment may
4111 be very very small.) If your linker is only able to align variables
4112 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4113 in an @code{__attribute__} will still only provide you with 8 byte
4114 alignment. See your linker documentation for further information.
4116 The @code{aligned} attribute can also be used for functions
4117 (@pxref{Function Attributes}.)
4119 @item cleanup (@var{cleanup_function})
4120 @cindex @code{cleanup} attribute
4121 The @code{cleanup} attribute runs a function when the variable goes
4122 out of scope. This attribute can only be applied to auto function
4123 scope variables; it may not be applied to parameters or variables
4124 with static storage duration. The function must take one parameter,
4125 a pointer to a type compatible with the variable. The return value
4126 of the function (if any) is ignored.
4128 If @option{-fexceptions} is enabled, then @var{cleanup_function}
4129 will be run during the stack unwinding that happens during the
4130 processing of the exception. Note that the @code{cleanup} attribute
4131 does not allow the exception to be caught, only to perform an action.
4132 It is undefined what happens if @var{cleanup_function} does not
4137 @cindex @code{common} attribute
4138 @cindex @code{nocommon} attribute
4141 The @code{common} attribute requests GCC to place a variable in
4142 ``common'' storage. The @code{nocommon} attribute requests the
4143 opposite---to allocate space for it directly.
4145 These attributes override the default chosen by the
4146 @option{-fno-common} and @option{-fcommon} flags respectively.
4149 @itemx deprecated (@var{msg})
4150 @cindex @code{deprecated} attribute
4151 The @code{deprecated} attribute results in a warning if the variable
4152 is used anywhere in the source file. This is useful when identifying
4153 variables that are expected to be removed in a future version of a
4154 program. The warning also includes the location of the declaration
4155 of the deprecated variable, to enable users to easily find further
4156 information about why the variable is deprecated, or what they should
4157 do instead. Note that the warning only occurs for uses:
4160 extern int old_var __attribute__ ((deprecated));
4162 int new_fn () @{ return old_var; @}
4165 results in a warning on line 3 but not line 2. The optional msg
4166 argument, which must be a string, will be printed in the warning if
4169 The @code{deprecated} attribute can also be used for functions and
4170 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
4172 @item mode (@var{mode})
4173 @cindex @code{mode} attribute
4174 This attribute specifies the data type for the declaration---whichever
4175 type corresponds to the mode @var{mode}. This in effect lets you
4176 request an integer or floating point type according to its width.
4178 You may also specify a mode of @samp{byte} or @samp{__byte__} to
4179 indicate the mode corresponding to a one-byte integer, @samp{word} or
4180 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
4181 or @samp{__pointer__} for the mode used to represent pointers.
4184 @cindex @code{packed} attribute
4185 The @code{packed} attribute specifies that a variable or structure field
4186 should have the smallest possible alignment---one byte for a variable,
4187 and one bit for a field, unless you specify a larger value with the
4188 @code{aligned} attribute.
4190 Here is a structure in which the field @code{x} is packed, so that it
4191 immediately follows @code{a}:
4197 int x[2] __attribute__ ((packed));
4201 @emph{Note:} The 4.1, 4.2 and 4.3 series of GCC ignore the
4202 @code{packed} attribute on bit-fields of type @code{char}. This has
4203 been fixed in GCC 4.4 but the change can lead to differences in the
4204 structure layout. See the documentation of
4205 @option{-Wpacked-bitfield-compat} for more information.
4207 @item section ("@var{section-name}")
4208 @cindex @code{section} variable attribute
4209 Normally, the compiler places the objects it generates in sections like
4210 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
4211 or you need certain particular variables to appear in special sections,
4212 for example to map to special hardware. The @code{section}
4213 attribute specifies that a variable (or function) lives in a particular
4214 section. For example, this small program uses several specific section names:
4217 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
4218 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
4219 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
4220 int init_data __attribute__ ((section ("INITDATA")));
4224 /* @r{Initialize stack pointer} */
4225 init_sp (stack + sizeof (stack));
4227 /* @r{Initialize initialized data} */
4228 memcpy (&init_data, &data, &edata - &data);
4230 /* @r{Turn on the serial ports} */
4237 Use the @code{section} attribute with
4238 @emph{global} variables and not @emph{local} variables,
4239 as shown in the example.
4241 You may use the @code{section} attribute with initialized or
4242 uninitialized global variables but the linker requires
4243 each object be defined once, with the exception that uninitialized
4244 variables tentatively go in the @code{common} (or @code{bss}) section
4245 and can be multiply ``defined''. Using the @code{section} attribute
4246 will change what section the variable goes into and may cause the
4247 linker to issue an error if an uninitialized variable has multiple
4248 definitions. You can force a variable to be initialized with the
4249 @option{-fno-common} flag or the @code{nocommon} attribute.
4251 Some file formats do not support arbitrary sections so the @code{section}
4252 attribute is not available on all platforms.
4253 If you need to map the entire contents of a module to a particular
4254 section, consider using the facilities of the linker instead.
4257 @cindex @code{shared} variable attribute
4258 On Microsoft Windows, in addition to putting variable definitions in a named
4259 section, the section can also be shared among all running copies of an
4260 executable or DLL@. For example, this small program defines shared data
4261 by putting it in a named section @code{shared} and marking the section
4265 int foo __attribute__((section ("shared"), shared)) = 0;
4270 /* @r{Read and write foo. All running
4271 copies see the same value.} */
4277 You may only use the @code{shared} attribute along with @code{section}
4278 attribute with a fully initialized global definition because of the way
4279 linkers work. See @code{section} attribute for more information.
4281 The @code{shared} attribute is only available on Microsoft Windows@.
4283 @item tls_model ("@var{tls_model}")
4284 @cindex @code{tls_model} attribute
4285 The @code{tls_model} attribute sets thread-local storage model
4286 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
4287 overriding @option{-ftls-model=} command-line switch on a per-variable
4289 The @var{tls_model} argument should be one of @code{global-dynamic},
4290 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
4292 Not all targets support this attribute.
4295 This attribute, attached to a variable, means that the variable is meant
4296 to be possibly unused. GCC will not produce a warning for this
4300 This attribute, attached to a variable, means that the variable must be
4301 emitted even if it appears that the variable is not referenced.
4303 @item vector_size (@var{bytes})
4304 This attribute specifies the vector size for the variable, measured in
4305 bytes. For example, the declaration:
4308 int foo __attribute__ ((vector_size (16)));
4312 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
4313 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
4314 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
4316 This attribute is only applicable to integral and float scalars,
4317 although arrays, pointers, and function return values are allowed in
4318 conjunction with this construct.
4320 Aggregates with this attribute are invalid, even if they are of the same
4321 size as a corresponding scalar. For example, the declaration:
4324 struct S @{ int a; @};
4325 struct S __attribute__ ((vector_size (16))) foo;
4329 is invalid even if the size of the structure is the same as the size of
4333 The @code{selectany} attribute causes an initialized global variable to
4334 have link-once semantics. When multiple definitions of the variable are
4335 encountered by the linker, the first is selected and the remainder are
4336 discarded. Following usage by the Microsoft compiler, the linker is told
4337 @emph{not} to warn about size or content differences of the multiple
4340 Although the primary usage of this attribute is for POD types, the
4341 attribute can also be applied to global C++ objects that are initialized
4342 by a constructor. In this case, the static initialization and destruction
4343 code for the object is emitted in each translation defining the object,
4344 but the calls to the constructor and destructor are protected by a
4345 link-once guard variable.
4347 The @code{selectany} attribute is only available on Microsoft Windows
4348 targets. You can use @code{__declspec (selectany)} as a synonym for
4349 @code{__attribute__ ((selectany))} for compatibility with other
4353 The @code{weak} attribute is described in @ref{Function Attributes}.
4356 The @code{dllimport} attribute is described in @ref{Function Attributes}.
4359 The @code{dllexport} attribute is described in @ref{Function Attributes}.
4363 @subsection Blackfin Variable Attributes
4365 Three attributes are currently defined for the Blackfin.
4371 @cindex @code{l1_data} variable attribute
4372 @cindex @code{l1_data_A} variable attribute
4373 @cindex @code{l1_data_B} variable attribute
4374 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
4375 Variables with @code{l1_data} attribute will be put into the specific section
4376 named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into
4377 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
4378 attribute will be put into the specific section named @code{.l1.data.B}.
4381 @cindex @code{l2} variable attribute
4382 Use this attribute on the Blackfin to place the variable into L2 SRAM.
4383 Variables with @code{l2} attribute will be put into the specific section
4384 named @code{.l2.data}.
4387 @subsection M32R/D Variable Attributes
4389 One attribute is currently defined for the M32R/D@.
4392 @item model (@var{model-name})
4393 @cindex variable addressability on the M32R/D
4394 Use this attribute on the M32R/D to set the addressability of an object.
4395 The identifier @var{model-name} is one of @code{small}, @code{medium},
4396 or @code{large}, representing each of the code models.
4398 Small model objects live in the lower 16MB of memory (so that their
4399 addresses can be loaded with the @code{ld24} instruction).
4401 Medium and large model objects may live anywhere in the 32-bit address space
4402 (the compiler will generate @code{seth/add3} instructions to load their
4406 @anchor{MeP Variable Attributes}
4407 @subsection MeP Variable Attributes
4409 The MeP target has a number of addressing modes and busses. The
4410 @code{near} space spans the standard memory space's first 16 megabytes
4411 (24 bits). The @code{far} space spans the entire 32-bit memory space.
4412 The @code{based} space is a 128 byte region in the memory space which
4413 is addressed relative to the @code{$tp} register. The @code{tiny}
4414 space is a 65536 byte region relative to the @code{$gp} register. In
4415 addition to these memory regions, the MeP target has a separate 16-bit
4416 control bus which is specified with @code{cb} attributes.
4421 Any variable with the @code{based} attribute will be assigned to the
4422 @code{.based} section, and will be accessed with relative to the
4423 @code{$tp} register.
4426 Likewise, the @code{tiny} attribute assigned variables to the
4427 @code{.tiny} section, relative to the @code{$gp} register.
4430 Variables with the @code{near} attribute are assumed to have addresses
4431 that fit in a 24-bit addressing mode. This is the default for large
4432 variables (@code{-mtiny=4} is the default) but this attribute can
4433 override @code{-mtiny=} for small variables, or override @code{-ml}.
4436 Variables with the @code{far} attribute are addressed using a full
4437 32-bit address. Since this covers the entire memory space, this
4438 allows modules to make no assumptions about where variables might be
4442 @itemx io (@var{addr})
4443 Variables with the @code{io} attribute are used to address
4444 memory-mapped peripherals. If an address is specified, the variable
4445 is assigned that address, else it is not assigned an address (it is
4446 assumed some other module will assign an address). Example:
4449 int timer_count __attribute__((io(0x123)));
4453 @itemx cb (@var{addr})
4454 Variables with the @code{cb} attribute are used to access the control
4455 bus, using special instructions. @code{addr} indicates the control bus
4459 int cpu_clock __attribute__((cb(0x123)));
4464 @anchor{i386 Variable Attributes}
4465 @subsection i386 Variable Attributes
4467 Two attributes are currently defined for i386 configurations:
4468 @code{ms_struct} and @code{gcc_struct}
4473 @cindex @code{ms_struct} attribute
4474 @cindex @code{gcc_struct} attribute
4476 If @code{packed} is used on a structure, or if bit-fields are used
4477 it may be that the Microsoft ABI packs them differently
4478 than GCC would normally pack them. Particularly when moving packed
4479 data between functions compiled with GCC and the native Microsoft compiler
4480 (either via function call or as data in a file), it may be necessary to access
4483 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4484 compilers to match the native Microsoft compiler.
4486 The Microsoft structure layout algorithm is fairly simple with the exception
4487 of the bitfield packing:
4489 The padding and alignment of members of structures and whether a bit field
4490 can straddle a storage-unit boundary
4493 @item Structure members are stored sequentially in the order in which they are
4494 declared: the first member has the lowest memory address and the last member
4497 @item Every data object has an alignment-requirement. The alignment-requirement
4498 for all data except structures, unions, and arrays is either the size of the
4499 object or the current packing size (specified with either the aligned attribute
4500 or the pack pragma), whichever is less. For structures, unions, and arrays,
4501 the alignment-requirement is the largest alignment-requirement of its members.
4502 Every object is allocated an offset so that:
4504 offset % alignment-requirement == 0
4506 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
4507 unit if the integral types are the same size and if the next bit field fits
4508 into the current allocation unit without crossing the boundary imposed by the
4509 common alignment requirements of the bit fields.
4512 Handling of zero-length bitfields:
4514 MSVC interprets zero-length bitfields in the following ways:
4517 @item If a zero-length bitfield is inserted between two bitfields that would
4518 normally be coalesced, the bitfields will not be coalesced.
4525 unsigned long bf_1 : 12;
4527 unsigned long bf_2 : 12;
4531 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
4532 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
4534 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
4535 alignment of the zero-length bitfield is greater than the member that follows it,
4536 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
4556 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
4557 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
4558 bitfield will not affect the alignment of @code{bar} or, as a result, the size
4561 Taking this into account, it is important to note the following:
4564 @item If a zero-length bitfield follows a normal bitfield, the type of the
4565 zero-length bitfield may affect the alignment of the structure as whole. For
4566 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
4567 normal bitfield, and is of type short.
4569 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
4570 still affect the alignment of the structure:
4580 Here, @code{t4} will take up 4 bytes.
4583 @item Zero-length bitfields following non-bitfield members are ignored:
4594 Here, @code{t5} will take up 2 bytes.
4598 @subsection PowerPC Variable Attributes
4600 Three attributes currently are defined for PowerPC configurations:
4601 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4603 For full documentation of the struct attributes please see the
4604 documentation in @ref{i386 Variable Attributes}.
4606 For documentation of @code{altivec} attribute please see the
4607 documentation in @ref{PowerPC Type Attributes}.
4609 @subsection SPU Variable Attributes
4611 The SPU supports the @code{spu_vector} attribute for variables. For
4612 documentation of this attribute please see the documentation in
4613 @ref{SPU Type Attributes}.
4615 @subsection Xstormy16 Variable Attributes
4617 One attribute is currently defined for xstormy16 configurations:
4622 @cindex @code{below100} attribute
4624 If a variable has the @code{below100} attribute (@code{BELOW100} is
4625 allowed also), GCC will place the variable in the first 0x100 bytes of
4626 memory and use special opcodes to access it. Such variables will be
4627 placed in either the @code{.bss_below100} section or the
4628 @code{.data_below100} section.
4632 @subsection AVR Variable Attributes
4636 @cindex @code{progmem} variable attribute
4637 The @code{progmem} attribute is used on the AVR to place data in the Program
4638 Memory address space. The AVR is a Harvard Architecture processor and data
4639 normally resides in the Data Memory address space.
4642 @node Type Attributes
4643 @section Specifying Attributes of Types
4644 @cindex attribute of types
4645 @cindex type attributes
4647 The keyword @code{__attribute__} allows you to specify special
4648 attributes of @code{struct} and @code{union} types when you define
4649 such types. This keyword is followed by an attribute specification
4650 inside double parentheses. Seven attributes are currently defined for
4651 types: @code{aligned}, @code{packed}, @code{transparent_union},
4652 @code{unused}, @code{deprecated}, @code{visibility}, and
4653 @code{may_alias}. Other attributes are defined for functions
4654 (@pxref{Function Attributes}) and for variables (@pxref{Variable
4657 You may also specify any one of these attributes with @samp{__}
4658 preceding and following its keyword. This allows you to use these
4659 attributes in header files without being concerned about a possible
4660 macro of the same name. For example, you may use @code{__aligned__}
4661 instead of @code{aligned}.
4663 You may specify type attributes in an enum, struct or union type
4664 declaration or definition, or for other types in a @code{typedef}
4667 For an enum, struct or union type, you may specify attributes either
4668 between the enum, struct or union tag and the name of the type, or
4669 just past the closing curly brace of the @emph{definition}. The
4670 former syntax is preferred.
4672 @xref{Attribute Syntax}, for details of the exact syntax for using
4676 @cindex @code{aligned} attribute
4677 @item aligned (@var{alignment})
4678 This attribute specifies a minimum alignment (in bytes) for variables
4679 of the specified type. For example, the declarations:
4682 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
4683 typedef int more_aligned_int __attribute__ ((aligned (8)));
4687 force the compiler to insure (as far as it can) that each variable whose
4688 type is @code{struct S} or @code{more_aligned_int} will be allocated and
4689 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
4690 variables of type @code{struct S} aligned to 8-byte boundaries allows
4691 the compiler to use the @code{ldd} and @code{std} (doubleword load and
4692 store) instructions when copying one variable of type @code{struct S} to
4693 another, thus improving run-time efficiency.
4695 Note that the alignment of any given @code{struct} or @code{union} type
4696 is required by the ISO C standard to be at least a perfect multiple of
4697 the lowest common multiple of the alignments of all of the members of
4698 the @code{struct} or @code{union} in question. This means that you @emph{can}
4699 effectively adjust the alignment of a @code{struct} or @code{union}
4700 type by attaching an @code{aligned} attribute to any one of the members
4701 of such a type, but the notation illustrated in the example above is a
4702 more obvious, intuitive, and readable way to request the compiler to
4703 adjust the alignment of an entire @code{struct} or @code{union} type.
4705 As in the preceding example, you can explicitly specify the alignment
4706 (in bytes) that you wish the compiler to use for a given @code{struct}
4707 or @code{union} type. Alternatively, you can leave out the alignment factor
4708 and just ask the compiler to align a type to the maximum
4709 useful alignment for the target machine you are compiling for. For
4710 example, you could write:
4713 struct S @{ short f[3]; @} __attribute__ ((aligned));
4716 Whenever you leave out the alignment factor in an @code{aligned}
4717 attribute specification, the compiler automatically sets the alignment
4718 for the type to the largest alignment which is ever used for any data
4719 type on the target machine you are compiling for. Doing this can often
4720 make copy operations more efficient, because the compiler can use
4721 whatever instructions copy the biggest chunks of memory when performing
4722 copies to or from the variables which have types that you have aligned
4725 In the example above, if the size of each @code{short} is 2 bytes, then
4726 the size of the entire @code{struct S} type is 6 bytes. The smallest
4727 power of two which is greater than or equal to that is 8, so the
4728 compiler sets the alignment for the entire @code{struct S} type to 8
4731 Note that although you can ask the compiler to select a time-efficient
4732 alignment for a given type and then declare only individual stand-alone
4733 objects of that type, the compiler's ability to select a time-efficient
4734 alignment is primarily useful only when you plan to create arrays of
4735 variables having the relevant (efficiently aligned) type. If you
4736 declare or use arrays of variables of an efficiently-aligned type, then
4737 it is likely that your program will also be doing pointer arithmetic (or
4738 subscripting, which amounts to the same thing) on pointers to the
4739 relevant type, and the code that the compiler generates for these
4740 pointer arithmetic operations will often be more efficient for
4741 efficiently-aligned types than for other types.
4743 The @code{aligned} attribute can only increase the alignment; but you
4744 can decrease it by specifying @code{packed} as well. See below.
4746 Note that the effectiveness of @code{aligned} attributes may be limited
4747 by inherent limitations in your linker. On many systems, the linker is
4748 only able to arrange for variables to be aligned up to a certain maximum
4749 alignment. (For some linkers, the maximum supported alignment may
4750 be very very small.) If your linker is only able to align variables
4751 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4752 in an @code{__attribute__} will still only provide you with 8 byte
4753 alignment. See your linker documentation for further information.
4756 This attribute, attached to @code{struct} or @code{union} type
4757 definition, specifies that each member (other than zero-width bitfields)
4758 of the structure or union is placed to minimize the memory required. When
4759 attached to an @code{enum} definition, it indicates that the smallest
4760 integral type should be used.
4762 @opindex fshort-enums
4763 Specifying this attribute for @code{struct} and @code{union} types is
4764 equivalent to specifying the @code{packed} attribute on each of the
4765 structure or union members. Specifying the @option{-fshort-enums}
4766 flag on the line is equivalent to specifying the @code{packed}
4767 attribute on all @code{enum} definitions.
4769 In the following example @code{struct my_packed_struct}'s members are
4770 packed closely together, but the internal layout of its @code{s} member
4771 is not packed---to do that, @code{struct my_unpacked_struct} would need to
4775 struct my_unpacked_struct
4781 struct __attribute__ ((__packed__)) my_packed_struct
4785 struct my_unpacked_struct s;
4789 You may only specify this attribute on the definition of an @code{enum},
4790 @code{struct} or @code{union}, not on a @code{typedef} which does not
4791 also define the enumerated type, structure or union.
4793 @item transparent_union
4794 This attribute, attached to a @code{union} type definition, indicates
4795 that any function parameter having that union type causes calls to that
4796 function to be treated in a special way.
4798 First, the argument corresponding to a transparent union type can be of
4799 any type in the union; no cast is required. Also, if the union contains
4800 a pointer type, the corresponding argument can be a null pointer
4801 constant or a void pointer expression; and if the union contains a void
4802 pointer type, the corresponding argument can be any pointer expression.
4803 If the union member type is a pointer, qualifiers like @code{const} on
4804 the referenced type must be respected, just as with normal pointer
4807 Second, the argument is passed to the function using the calling
4808 conventions of the first member of the transparent union, not the calling
4809 conventions of the union itself. All members of the union must have the
4810 same machine representation; this is necessary for this argument passing
4813 Transparent unions are designed for library functions that have multiple
4814 interfaces for compatibility reasons. For example, suppose the
4815 @code{wait} function must accept either a value of type @code{int *} to
4816 comply with Posix, or a value of type @code{union wait *} to comply with
4817 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
4818 @code{wait} would accept both kinds of arguments, but it would also
4819 accept any other pointer type and this would make argument type checking
4820 less useful. Instead, @code{<sys/wait.h>} might define the interface
4824 typedef union __attribute__ ((__transparent_union__))
4828 @} wait_status_ptr_t;
4830 pid_t wait (wait_status_ptr_t);
4833 This interface allows either @code{int *} or @code{union wait *}
4834 arguments to be passed, using the @code{int *} calling convention.
4835 The program can call @code{wait} with arguments of either type:
4838 int w1 () @{ int w; return wait (&w); @}
4839 int w2 () @{ union wait w; return wait (&w); @}
4842 With this interface, @code{wait}'s implementation might look like this:
4845 pid_t wait (wait_status_ptr_t p)
4847 return waitpid (-1, p.__ip, 0);
4852 When attached to a type (including a @code{union} or a @code{struct}),
4853 this attribute means that variables of that type are meant to appear
4854 possibly unused. GCC will not produce a warning for any variables of
4855 that type, even if the variable appears to do nothing. This is often
4856 the case with lock or thread classes, which are usually defined and then
4857 not referenced, but contain constructors and destructors that have
4858 nontrivial bookkeeping functions.
4861 @itemx deprecated (@var{msg})
4862 The @code{deprecated} attribute results in a warning if the type
4863 is used anywhere in the source file. This is useful when identifying
4864 types that are expected to be removed in a future version of a program.
4865 If possible, the warning also includes the location of the declaration
4866 of the deprecated type, to enable users to easily find further
4867 information about why the type is deprecated, or what they should do
4868 instead. Note that the warnings only occur for uses and then only
4869 if the type is being applied to an identifier that itself is not being
4870 declared as deprecated.
4873 typedef int T1 __attribute__ ((deprecated));
4877 typedef T1 T3 __attribute__ ((deprecated));
4878 T3 z __attribute__ ((deprecated));
4881 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
4882 warning is issued for line 4 because T2 is not explicitly
4883 deprecated. Line 5 has no warning because T3 is explicitly
4884 deprecated. Similarly for line 6. The optional msg
4885 argument, which must be a string, will be printed in the warning if
4888 The @code{deprecated} attribute can also be used for functions and
4889 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
4892 Accesses through pointers to types with this attribute are not subject
4893 to type-based alias analysis, but are instead assumed to be able to alias
4894 any other type of objects. In the context of 6.5/7 an lvalue expression
4895 dereferencing such a pointer is treated like having a character type.
4896 See @option{-fstrict-aliasing} for more information on aliasing issues.
4897 This extension exists to support some vector APIs, in which pointers to
4898 one vector type are permitted to alias pointers to a different vector type.
4900 Note that an object of a type with this attribute does not have any
4906 typedef short __attribute__((__may_alias__)) short_a;
4912 short_a *b = (short_a *) &a;
4916 if (a == 0x12345678)
4923 If you replaced @code{short_a} with @code{short} in the variable
4924 declaration, the above program would abort when compiled with
4925 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
4926 above in recent GCC versions.
4929 In C++, attribute visibility (@pxref{Function Attributes}) can also be
4930 applied to class, struct, union and enum types. Unlike other type
4931 attributes, the attribute must appear between the initial keyword and
4932 the name of the type; it cannot appear after the body of the type.
4934 Note that the type visibility is applied to vague linkage entities
4935 associated with the class (vtable, typeinfo node, etc.). In
4936 particular, if a class is thrown as an exception in one shared object
4937 and caught in another, the class must have default visibility.
4938 Otherwise the two shared objects will be unable to use the same
4939 typeinfo node and exception handling will break.
4943 @subsection ARM Type Attributes
4945 On those ARM targets that support @code{dllimport} (such as Symbian
4946 OS), you can use the @code{notshared} attribute to indicate that the
4947 virtual table and other similar data for a class should not be
4948 exported from a DLL@. For example:
4951 class __declspec(notshared) C @{
4953 __declspec(dllimport) C();
4957 __declspec(dllexport)
4961 In this code, @code{C::C} is exported from the current DLL, but the
4962 virtual table for @code{C} is not exported. (You can use
4963 @code{__attribute__} instead of @code{__declspec} if you prefer, but
4964 most Symbian OS code uses @code{__declspec}.)
4966 @anchor{MeP Type Attributes}
4967 @subsection MeP Type Attributes
4969 Many of the MeP variable attributes may be applied to types as well.
4970 Specifically, the @code{based}, @code{tiny}, @code{near}, and
4971 @code{far} attributes may be applied to either. The @code{io} and
4972 @code{cb} attributes may not be applied to types.
4974 @anchor{i386 Type Attributes}
4975 @subsection i386 Type Attributes
4977 Two attributes are currently defined for i386 configurations:
4978 @code{ms_struct} and @code{gcc_struct}.
4984 @cindex @code{ms_struct}
4985 @cindex @code{gcc_struct}
4987 If @code{packed} is used on a structure, or if bit-fields are used
4988 it may be that the Microsoft ABI packs them differently
4989 than GCC would normally pack them. Particularly when moving packed
4990 data between functions compiled with GCC and the native Microsoft compiler
4991 (either via function call or as data in a file), it may be necessary to access
4994 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4995 compilers to match the native Microsoft compiler.
4998 To specify multiple attributes, separate them by commas within the
4999 double parentheses: for example, @samp{__attribute__ ((aligned (16),
5002 @anchor{PowerPC Type Attributes}
5003 @subsection PowerPC Type Attributes
5005 Three attributes currently are defined for PowerPC configurations:
5006 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
5008 For full documentation of the @code{ms_struct} and @code{gcc_struct}
5009 attributes please see the documentation in @ref{i386 Type Attributes}.
5011 The @code{altivec} attribute allows one to declare AltiVec vector data
5012 types supported by the AltiVec Programming Interface Manual. The
5013 attribute requires an argument to specify one of three vector types:
5014 @code{vector__}, @code{pixel__} (always followed by unsigned short),
5015 and @code{bool__} (always followed by unsigned).
5018 __attribute__((altivec(vector__)))
5019 __attribute__((altivec(pixel__))) unsigned short
5020 __attribute__((altivec(bool__))) unsigned
5023 These attributes mainly are intended to support the @code{__vector},
5024 @code{__pixel}, and @code{__bool} AltiVec keywords.
5026 @anchor{SPU Type Attributes}
5027 @subsection SPU Type Attributes
5029 The SPU supports the @code{spu_vector} attribute for types. This attribute
5030 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
5031 Language Extensions Specification. It is intended to support the
5032 @code{__vector} keyword.
5036 @section An Inline Function is As Fast As a Macro
5037 @cindex inline functions
5038 @cindex integrating function code
5040 @cindex macros, inline alternative
5042 By declaring a function inline, you can direct GCC to make
5043 calls to that function faster. One way GCC can achieve this is to
5044 integrate that function's code into the code for its callers. This
5045 makes execution faster by eliminating the function-call overhead; in
5046 addition, if any of the actual argument values are constant, their
5047 known values may permit simplifications at compile time so that not
5048 all of the inline function's code needs to be included. The effect on
5049 code size is less predictable; object code may be larger or smaller
5050 with function inlining, depending on the particular case. You can
5051 also direct GCC to try to integrate all ``simple enough'' functions
5052 into their callers with the option @option{-finline-functions}.
5054 GCC implements three different semantics of declaring a function
5055 inline. One is available with @option{-std=gnu89} or
5056 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
5057 on all inline declarations, another when
5058 @option{-std=c99}, @option{-std=c1x},
5059 @option{-std=gnu99} or @option{-std=gnu1x}
5060 (without @option{-fgnu89-inline}), and the third
5061 is used when compiling C++.
5063 To declare a function inline, use the @code{inline} keyword in its
5064 declaration, like this:
5074 If you are writing a header file to be included in ISO C90 programs, write
5075 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
5077 The three types of inlining behave similarly in two important cases:
5078 when the @code{inline} keyword is used on a @code{static} function,
5079 like the example above, and when a function is first declared without
5080 using the @code{inline} keyword and then is defined with
5081 @code{inline}, like this:
5084 extern int inc (int *a);
5092 In both of these common cases, the program behaves the same as if you
5093 had not used the @code{inline} keyword, except for its speed.
5095 @cindex inline functions, omission of
5096 @opindex fkeep-inline-functions
5097 When a function is both inline and @code{static}, if all calls to the
5098 function are integrated into the caller, and the function's address is
5099 never used, then the function's own assembler code is never referenced.
5100 In this case, GCC does not actually output assembler code for the
5101 function, unless you specify the option @option{-fkeep-inline-functions}.
5102 Some calls cannot be integrated for various reasons (in particular,
5103 calls that precede the function's definition cannot be integrated, and
5104 neither can recursive calls within the definition). If there is a
5105 nonintegrated call, then the function is compiled to assembler code as
5106 usual. The function must also be compiled as usual if the program
5107 refers to its address, because that can't be inlined.
5110 Note that certain usages in a function definition can make it unsuitable
5111 for inline substitution. Among these usages are: use of varargs, use of
5112 alloca, use of variable sized data types (@pxref{Variable Length}),
5113 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
5114 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
5115 will warn when a function marked @code{inline} could not be substituted,
5116 and will give the reason for the failure.
5118 @cindex automatic @code{inline} for C++ member fns
5119 @cindex @code{inline} automatic for C++ member fns
5120 @cindex member fns, automatically @code{inline}
5121 @cindex C++ member fns, automatically @code{inline}
5122 @opindex fno-default-inline
5123 As required by ISO C++, GCC considers member functions defined within
5124 the body of a class to be marked inline even if they are
5125 not explicitly declared with the @code{inline} keyword. You can
5126 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
5127 Options,,Options Controlling C++ Dialect}.
5129 GCC does not inline any functions when not optimizing unless you specify
5130 the @samp{always_inline} attribute for the function, like this:
5133 /* @r{Prototype.} */
5134 inline void foo (const char) __attribute__((always_inline));
5137 The remainder of this section is specific to GNU C90 inlining.
5139 @cindex non-static inline function
5140 When an inline function is not @code{static}, then the compiler must assume
5141 that there may be calls from other source files; since a global symbol can
5142 be defined only once in any program, the function must not be defined in
5143 the other source files, so the calls therein cannot be integrated.
5144 Therefore, a non-@code{static} inline function is always compiled on its
5145 own in the usual fashion.
5147 If you specify both @code{inline} and @code{extern} in the function
5148 definition, then the definition is used only for inlining. In no case
5149 is the function compiled on its own, not even if you refer to its
5150 address explicitly. Such an address becomes an external reference, as
5151 if you had only declared the function, and had not defined it.
5153 This combination of @code{inline} and @code{extern} has almost the
5154 effect of a macro. The way to use it is to put a function definition in
5155 a header file with these keywords, and put another copy of the
5156 definition (lacking @code{inline} and @code{extern}) in a library file.
5157 The definition in the header file will cause most calls to the function
5158 to be inlined. If any uses of the function remain, they will refer to
5159 the single copy in the library.
5162 @section When is a Volatile Object Accessed?
5163 @cindex accessing volatiles
5164 @cindex volatile read
5165 @cindex volatile write
5166 @cindex volatile access
5168 C has the concept of volatile objects. These are normally accessed by
5169 pointers and used for accessing hardware or inter-thread
5170 communication. The standard encourage compilers to refrain from
5171 optimizations concerning accesses to volatile objects, but leaves it
5172 implementation defined as to what constitutes a volatile access. The
5173 minimum requirement is that at a sequence point all previous accesses
5174 to volatile objects have stabilized and no subsequent accesses have
5175 occurred. Thus an implementation is free to reorder and combine
5176 volatile accesses which occur between sequence points, but cannot do
5177 so for accesses across a sequence point. The use of volatiles does
5178 not allow you to violate the restriction on updating objects multiple
5179 times between two sequence points.
5181 Accesses to non-volatile objects are not ordered with respect to
5182 volatile accesses. You cannot use a volatile object as a memory
5183 barrier to order a sequence of writes to non-volatile memory. For
5187 int *ptr = @var{something};
5189 *ptr = @var{something};
5193 Unless @var{*ptr} and @var{vobj} can be aliased, it is not guaranteed
5194 that the write to @var{*ptr} will have occurred by the time the update
5195 of @var{vobj} has happened. If you need this guarantee, you must use
5196 a stronger memory barrier such as:
5199 int *ptr = @var{something};
5201 *ptr = @var{something};
5202 asm volatile ("" : : : "memory");
5206 A scalar volatile object is read, when it is accessed in a void context:
5209 volatile int *src = @var{somevalue};
5213 Such expressions are rvalues, and GCC implements this as a
5214 read of the volatile object being pointed to.
5216 Assignments are also expressions and have an rvalue. However when
5217 assigning to a scalar volatile, the volatile object is not reread,
5218 regardless of whether the assignment expression's rvalue is used or
5219 not. If the assignment's rvalue is used, the value is that assigned
5220 to the volatile object. For instance, there is no read of @var{vobj}
5221 in all the following cases:
5226 vobj = @var{something};
5227 obj = vobj = @var{something};
5228 obj ? vobj = @var{onething} : vobj = @var{anotherthing};
5229 obj = (@var{something}, vobj = @var{anotherthing});
5232 If you need to read the volatile object after an assignment has
5233 occurred, you must use a separate expression with an intervening
5236 As bitfields are not individually addressable, volatile bitfields may
5237 be implicitly read when written to, or when adjacent bitfields are
5238 accessed. Bitfield operations may be optimized such that adjacent
5239 bitfields are only partially accessed, if they straddle a storage unit
5240 boundary. For these reasons it is unwise to use volatile bitfields to
5244 @section Assembler Instructions with C Expression Operands
5245 @cindex extended @code{asm}
5246 @cindex @code{asm} expressions
5247 @cindex assembler instructions
5250 In an assembler instruction using @code{asm}, you can specify the
5251 operands of the instruction using C expressions. This means you need not
5252 guess which registers or memory locations will contain the data you want
5255 You must specify an assembler instruction template much like what
5256 appears in a machine description, plus an operand constraint string for
5259 For example, here is how to use the 68881's @code{fsinx} instruction:
5262 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
5266 Here @code{angle} is the C expression for the input operand while
5267 @code{result} is that of the output operand. Each has @samp{"f"} as its
5268 operand constraint, saying that a floating point register is required.
5269 The @samp{=} in @samp{=f} indicates that the operand is an output; all
5270 output operands' constraints must use @samp{=}. The constraints use the
5271 same language used in the machine description (@pxref{Constraints}).
5273 Each operand is described by an operand-constraint string followed by
5274 the C expression in parentheses. A colon separates the assembler
5275 template from the first output operand and another separates the last
5276 output operand from the first input, if any. Commas separate the
5277 operands within each group. The total number of operands is currently
5278 limited to 30; this limitation may be lifted in some future version of
5281 If there are no output operands but there are input operands, you must
5282 place two consecutive colons surrounding the place where the output
5285 As of GCC version 3.1, it is also possible to specify input and output
5286 operands using symbolic names which can be referenced within the
5287 assembler code. These names are specified inside square brackets
5288 preceding the constraint string, and can be referenced inside the
5289 assembler code using @code{%[@var{name}]} instead of a percentage sign
5290 followed by the operand number. Using named operands the above example
5294 asm ("fsinx %[angle],%[output]"
5295 : [output] "=f" (result)
5296 : [angle] "f" (angle));
5300 Note that the symbolic operand names have no relation whatsoever to
5301 other C identifiers. You may use any name you like, even those of
5302 existing C symbols, but you must ensure that no two operands within the same
5303 assembler construct use the same symbolic name.
5305 Output operand expressions must be lvalues; the compiler can check this.
5306 The input operands need not be lvalues. The compiler cannot check
5307 whether the operands have data types that are reasonable for the
5308 instruction being executed. It does not parse the assembler instruction
5309 template and does not know what it means or even whether it is valid
5310 assembler input. The extended @code{asm} feature is most often used for
5311 machine instructions the compiler itself does not know exist. If
5312 the output expression cannot be directly addressed (for example, it is a
5313 bit-field), your constraint must allow a register. In that case, GCC
5314 will use the register as the output of the @code{asm}, and then store
5315 that register into the output.
5317 The ordinary output operands must be write-only; GCC will assume that
5318 the values in these operands before the instruction are dead and need
5319 not be generated. Extended asm supports input-output or read-write
5320 operands. Use the constraint character @samp{+} to indicate such an
5321 operand and list it with the output operands. You should only use
5322 read-write operands when the constraints for the operand (or the
5323 operand in which only some of the bits are to be changed) allow a
5326 You may, as an alternative, logically split its function into two
5327 separate operands, one input operand and one write-only output
5328 operand. The connection between them is expressed by constraints
5329 which say they need to be in the same location when the instruction
5330 executes. You can use the same C expression for both operands, or
5331 different expressions. For example, here we write the (fictitious)
5332 @samp{combine} instruction with @code{bar} as its read-only source
5333 operand and @code{foo} as its read-write destination:
5336 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
5340 The constraint @samp{"0"} for operand 1 says that it must occupy the
5341 same location as operand 0. A number in constraint is allowed only in
5342 an input operand and it must refer to an output operand.
5344 Only a number in the constraint can guarantee that one operand will be in
5345 the same place as another. The mere fact that @code{foo} is the value
5346 of both operands is not enough to guarantee that they will be in the
5347 same place in the generated assembler code. The following would not
5351 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
5354 Various optimizations or reloading could cause operands 0 and 1 to be in
5355 different registers; GCC knows no reason not to do so. For example, the
5356 compiler might find a copy of the value of @code{foo} in one register and
5357 use it for operand 1, but generate the output operand 0 in a different
5358 register (copying it afterward to @code{foo}'s own address). Of course,
5359 since the register for operand 1 is not even mentioned in the assembler
5360 code, the result will not work, but GCC can't tell that.
5362 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
5363 the operand number for a matching constraint. For example:
5366 asm ("cmoveq %1,%2,%[result]"
5367 : [result] "=r"(result)
5368 : "r" (test), "r"(new), "[result]"(old));
5371 Sometimes you need to make an @code{asm} operand be a specific register,
5372 but there's no matching constraint letter for that register @emph{by
5373 itself}. To force the operand into that register, use a local variable
5374 for the operand and specify the register in the variable declaration.
5375 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
5376 register constraint letter that matches the register:
5379 register int *p1 asm ("r0") = @dots{};
5380 register int *p2 asm ("r1") = @dots{};
5381 register int *result asm ("r0");
5382 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5385 @anchor{Example of asm with clobbered asm reg}
5386 In the above example, beware that a register that is call-clobbered by
5387 the target ABI will be overwritten by any function call in the
5388 assignment, including library calls for arithmetic operators.
5389 Also a register may be clobbered when generating some operations,
5390 like variable shift, memory copy or memory move on x86.
5391 Assuming it is a call-clobbered register, this may happen to @code{r0}
5392 above by the assignment to @code{p2}. If you have to use such a
5393 register, use temporary variables for expressions between the register
5398 register int *p1 asm ("r0") = @dots{};
5399 register int *p2 asm ("r1") = t1;
5400 register int *result asm ("r0");
5401 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5404 Some instructions clobber specific hard registers. To describe this,
5405 write a third colon after the input operands, followed by the names of
5406 the clobbered hard registers (given as strings). Here is a realistic
5407 example for the VAX:
5410 asm volatile ("movc3 %0,%1,%2"
5411 : /* @r{no outputs} */
5412 : "g" (from), "g" (to), "g" (count)
5413 : "r0", "r1", "r2", "r3", "r4", "r5");
5416 You may not write a clobber description in a way that overlaps with an
5417 input or output operand. For example, you may not have an operand
5418 describing a register class with one member if you mention that register
5419 in the clobber list. Variables declared to live in specific registers
5420 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
5421 have no part mentioned in the clobber description.
5422 There is no way for you to specify that an input
5423 operand is modified without also specifying it as an output
5424 operand. Note that if all the output operands you specify are for this
5425 purpose (and hence unused), you will then also need to specify
5426 @code{volatile} for the @code{asm} construct, as described below, to
5427 prevent GCC from deleting the @code{asm} statement as unused.
5429 If you refer to a particular hardware register from the assembler code,
5430 you will probably have to list the register after the third colon to
5431 tell the compiler the register's value is modified. In some assemblers,
5432 the register names begin with @samp{%}; to produce one @samp{%} in the
5433 assembler code, you must write @samp{%%} in the input.
5435 If your assembler instruction can alter the condition code register, add
5436 @samp{cc} to the list of clobbered registers. GCC on some machines
5437 represents the condition codes as a specific hardware register;
5438 @samp{cc} serves to name this register. On other machines, the
5439 condition code is handled differently, and specifying @samp{cc} has no
5440 effect. But it is valid no matter what the machine.
5442 If your assembler instructions access memory in an unpredictable
5443 fashion, add @samp{memory} to the list of clobbered registers. This
5444 will cause GCC to not keep memory values cached in registers across the
5445 assembler instruction and not optimize stores or loads to that memory.
5446 You will also want to add the @code{volatile} keyword if the memory
5447 affected is not listed in the inputs or outputs of the @code{asm}, as
5448 the @samp{memory} clobber does not count as a side-effect of the
5449 @code{asm}. If you know how large the accessed memory is, you can add
5450 it as input or output but if this is not known, you should add
5451 @samp{memory}. As an example, if you access ten bytes of a string, you
5452 can use a memory input like:
5455 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
5458 Note that in the following example the memory input is necessary,
5459 otherwise GCC might optimize the store to @code{x} away:
5466 asm ("magic stuff accessing an 'int' pointed to by '%1'"
5467 "=&d" (r) : "a" (y), "m" (*y));
5472 You can put multiple assembler instructions together in a single
5473 @code{asm} template, separated by the characters normally used in assembly
5474 code for the system. A combination that works in most places is a newline
5475 to break the line, plus a tab character to move to the instruction field
5476 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
5477 assembler allows semicolons as a line-breaking character. Note that some
5478 assembler dialects use semicolons to start a comment.
5479 The input operands are guaranteed not to use any of the clobbered
5480 registers, and neither will the output operands' addresses, so you can
5481 read and write the clobbered registers as many times as you like. Here
5482 is an example of multiple instructions in a template; it assumes the
5483 subroutine @code{_foo} accepts arguments in registers 9 and 10:
5486 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
5488 : "g" (from), "g" (to)
5492 Unless an output operand has the @samp{&} constraint modifier, GCC
5493 may allocate it in the same register as an unrelated input operand, on
5494 the assumption the inputs are consumed before the outputs are produced.
5495 This assumption may be false if the assembler code actually consists of
5496 more than one instruction. In such a case, use @samp{&} for each output
5497 operand that may not overlap an input. @xref{Modifiers}.
5499 If you want to test the condition code produced by an assembler
5500 instruction, you must include a branch and a label in the @code{asm}
5501 construct, as follows:
5504 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
5510 This assumes your assembler supports local labels, as the GNU assembler
5511 and most Unix assemblers do.
5513 Speaking of labels, jumps from one @code{asm} to another are not
5514 supported. The compiler's optimizers do not know about these jumps, and
5515 therefore they cannot take account of them when deciding how to
5516 optimize. @xref{Extended asm with goto}.
5518 @cindex macros containing @code{asm}
5519 Usually the most convenient way to use these @code{asm} instructions is to
5520 encapsulate them in macros that look like functions. For example,
5524 (@{ double __value, __arg = (x); \
5525 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
5530 Here the variable @code{__arg} is used to make sure that the instruction
5531 operates on a proper @code{double} value, and to accept only those
5532 arguments @code{x} which can convert automatically to a @code{double}.
5534 Another way to make sure the instruction operates on the correct data
5535 type is to use a cast in the @code{asm}. This is different from using a
5536 variable @code{__arg} in that it converts more different types. For
5537 example, if the desired type were @code{int}, casting the argument to
5538 @code{int} would accept a pointer with no complaint, while assigning the
5539 argument to an @code{int} variable named @code{__arg} would warn about
5540 using a pointer unless the caller explicitly casts it.
5542 If an @code{asm} has output operands, GCC assumes for optimization
5543 purposes the instruction has no side effects except to change the output
5544 operands. This does not mean instructions with a side effect cannot be
5545 used, but you must be careful, because the compiler may eliminate them
5546 if the output operands aren't used, or move them out of loops, or
5547 replace two with one if they constitute a common subexpression. Also,
5548 if your instruction does have a side effect on a variable that otherwise
5549 appears not to change, the old value of the variable may be reused later
5550 if it happens to be found in a register.
5552 You can prevent an @code{asm} instruction from being deleted
5553 by writing the keyword @code{volatile} after
5554 the @code{asm}. For example:
5557 #define get_and_set_priority(new) \
5559 asm volatile ("get_and_set_priority %0, %1" \
5560 : "=g" (__old) : "g" (new)); \
5565 The @code{volatile} keyword indicates that the instruction has
5566 important side-effects. GCC will not delete a volatile @code{asm} if
5567 it is reachable. (The instruction can still be deleted if GCC can
5568 prove that control-flow will never reach the location of the
5569 instruction.) Note that even a volatile @code{asm} instruction
5570 can be moved relative to other code, including across jump
5571 instructions. For example, on many targets there is a system
5572 register which can be set to control the rounding mode of
5573 floating point operations. You might try
5574 setting it with a volatile @code{asm}, like this PowerPC example:
5577 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
5582 This will not work reliably, as the compiler may move the addition back
5583 before the volatile @code{asm}. To make it work you need to add an
5584 artificial dependency to the @code{asm} referencing a variable in the code
5585 you don't want moved, for example:
5588 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
5592 Similarly, you can't expect a
5593 sequence of volatile @code{asm} instructions to remain perfectly
5594 consecutive. If you want consecutive output, use a single @code{asm}.
5595 Also, GCC will perform some optimizations across a volatile @code{asm}
5596 instruction; GCC does not ``forget everything'' when it encounters
5597 a volatile @code{asm} instruction the way some other compilers do.
5599 An @code{asm} instruction without any output operands will be treated
5600 identically to a volatile @code{asm} instruction.
5602 It is a natural idea to look for a way to give access to the condition
5603 code left by the assembler instruction. However, when we attempted to
5604 implement this, we found no way to make it work reliably. The problem
5605 is that output operands might need reloading, which would result in
5606 additional following ``store'' instructions. On most machines, these
5607 instructions would alter the condition code before there was time to
5608 test it. This problem doesn't arise for ordinary ``test'' and
5609 ``compare'' instructions because they don't have any output operands.
5611 For reasons similar to those described above, it is not possible to give
5612 an assembler instruction access to the condition code left by previous
5615 @anchor{Extended asm with goto}
5616 As of GCC version 4.5, @code{asm goto} may be used to have the assembly
5617 jump to one or more C labels. In this form, a fifth section after the
5618 clobber list contains a list of all C labels to which the assembly may jump.
5619 Each label operand is implicitly self-named. The @code{asm} is also assumed
5620 to fall through to the next statement.
5622 This form of @code{asm} is restricted to not have outputs. This is due
5623 to a internal restriction in the compiler that control transfer instructions
5624 cannot have outputs. This restriction on @code{asm goto} may be lifted
5625 in some future version of the compiler. In the mean time, @code{asm goto}
5626 may include a memory clobber, and so leave outputs in memory.
5632 asm goto ("frob %%r5, %1; jc %l[error]; mov (%2), %%r5"
5633 : : "r"(x), "r"(&y) : "r5", "memory" : error);
5640 In this (inefficient) example, the @code{frob} instruction sets the
5641 carry bit to indicate an error. The @code{jc} instruction detects
5642 this and branches to the @code{error} label. Finally, the output
5643 of the @code{frob} instruction (@code{%r5}) is stored into the memory
5644 for variable @code{y}, which is later read by the @code{return} statement.
5650 asm goto ("mfsr %%r1, 123; jmp %%r1;"
5651 ".pushsection doit_table;"
5652 ".long %l0, %l1, %l2, %l3;"
5654 : : : "r1" : label1, label2, label3, label4);
5655 __builtin_unreachable ();
5670 In this (also inefficient) example, the @code{mfsr} instruction reads
5671 an address from some out-of-band machine register, and the following
5672 @code{jmp} instruction branches to that address. The address read by
5673 the @code{mfsr} instruction is assumed to have been previously set via
5674 some application-specific mechanism to be one of the four values stored
5675 in the @code{doit_table} section. Finally, the @code{asm} is followed
5676 by a call to @code{__builtin_unreachable} to indicate that the @code{asm}
5677 does not in fact fall through.
5680 #define TRACE1(NUM) \
5682 asm goto ("0: nop;" \
5683 ".pushsection trace_table;" \
5686 : : : : trace#NUM); \
5687 if (0) @{ trace#NUM: trace(); @} \
5689 #define TRACE TRACE1(__COUNTER__)
5692 In this example (which in fact inspired the @code{asm goto} feature)
5693 we want on rare occasions to call the @code{trace} function; on other
5694 occasions we'd like to keep the overhead to the absolute minimum.
5695 The normal code path consists of a single @code{nop} instruction.
5696 However, we record the address of this @code{nop} together with the
5697 address of a label that calls the @code{trace} function. This allows
5698 the @code{nop} instruction to be patched at runtime to be an
5699 unconditional branch to the stored label. It is assumed that an
5700 optimizing compiler will move the labeled block out of line, to
5701 optimize the fall through path from the @code{asm}.
5703 If you are writing a header file that should be includable in ISO C
5704 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
5707 @subsection Size of an @code{asm}
5709 Some targets require that GCC track the size of each instruction used in
5710 order to generate correct code. Because the final length of an
5711 @code{asm} is only known by the assembler, GCC must make an estimate as
5712 to how big it will be. The estimate is formed by counting the number of
5713 statements in the pattern of the @code{asm} and multiplying that by the
5714 length of the longest instruction on that processor. Statements in the
5715 @code{asm} are identified by newline characters and whatever statement
5716 separator characters are supported by the assembler; on most processors
5717 this is the `@code{;}' character.
5719 Normally, GCC's estimate is perfectly adequate to ensure that correct
5720 code is generated, but it is possible to confuse the compiler if you use
5721 pseudo instructions or assembler macros that expand into multiple real
5722 instructions or if you use assembler directives that expand to more
5723 space in the object file than would be needed for a single instruction.
5724 If this happens then the assembler will produce a diagnostic saying that
5725 a label is unreachable.
5727 @subsection i386 floating point asm operands
5729 There are several rules on the usage of stack-like regs in
5730 asm_operands insns. These rules apply only to the operands that are
5735 Given a set of input regs that die in an asm_operands, it is
5736 necessary to know which are implicitly popped by the asm, and
5737 which must be explicitly popped by gcc.
5739 An input reg that is implicitly popped by the asm must be
5740 explicitly clobbered, unless it is constrained to match an
5744 For any input reg that is implicitly popped by an asm, it is
5745 necessary to know how to adjust the stack to compensate for the pop.
5746 If any non-popped input is closer to the top of the reg-stack than
5747 the implicitly popped reg, it would not be possible to know what the
5748 stack looked like---it's not clear how the rest of the stack ``slides
5751 All implicitly popped input regs must be closer to the top of
5752 the reg-stack than any input that is not implicitly popped.
5754 It is possible that if an input dies in an insn, reload might
5755 use the input reg for an output reload. Consider this example:
5758 asm ("foo" : "=t" (a) : "f" (b));
5761 This asm says that input B is not popped by the asm, and that
5762 the asm pushes a result onto the reg-stack, i.e., the stack is one
5763 deeper after the asm than it was before. But, it is possible that
5764 reload will think that it can use the same reg for both the input and
5765 the output, if input B dies in this insn.
5767 If any input operand uses the @code{f} constraint, all output reg
5768 constraints must use the @code{&} earlyclobber.
5770 The asm above would be written as
5773 asm ("foo" : "=&t" (a) : "f" (b));
5777 Some operands need to be in particular places on the stack. All
5778 output operands fall in this category---there is no other way to
5779 know which regs the outputs appear in unless the user indicates
5780 this in the constraints.
5782 Output operands must specifically indicate which reg an output
5783 appears in after an asm. @code{=f} is not allowed: the operand
5784 constraints must select a class with a single reg.
5787 Output operands may not be ``inserted'' between existing stack regs.
5788 Since no 387 opcode uses a read/write operand, all output operands
5789 are dead before the asm_operands, and are pushed by the asm_operands.
5790 It makes no sense to push anywhere but the top of the reg-stack.
5792 Output operands must start at the top of the reg-stack: output
5793 operands may not ``skip'' a reg.
5796 Some asm statements may need extra stack space for internal
5797 calculations. This can be guaranteed by clobbering stack registers
5798 unrelated to the inputs and outputs.
5802 Here are a couple of reasonable asms to want to write. This asm
5803 takes one input, which is internally popped, and produces two outputs.
5806 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
5809 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
5810 and replaces them with one output. The user must code the @code{st(1)}
5811 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
5814 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
5820 @section Controlling Names Used in Assembler Code
5821 @cindex assembler names for identifiers
5822 @cindex names used in assembler code
5823 @cindex identifiers, names in assembler code
5825 You can specify the name to be used in the assembler code for a C
5826 function or variable by writing the @code{asm} (or @code{__asm__})
5827 keyword after the declarator as follows:
5830 int foo asm ("myfoo") = 2;
5834 This specifies that the name to be used for the variable @code{foo} in
5835 the assembler code should be @samp{myfoo} rather than the usual
5838 On systems where an underscore is normally prepended to the name of a C
5839 function or variable, this feature allows you to define names for the
5840 linker that do not start with an underscore.
5842 It does not make sense to use this feature with a non-static local
5843 variable since such variables do not have assembler names. If you are
5844 trying to put the variable in a particular register, see @ref{Explicit
5845 Reg Vars}. GCC presently accepts such code with a warning, but will
5846 probably be changed to issue an error, rather than a warning, in the
5849 You cannot use @code{asm} in this way in a function @emph{definition}; but
5850 you can get the same effect by writing a declaration for the function
5851 before its definition and putting @code{asm} there, like this:
5854 extern func () asm ("FUNC");
5861 It is up to you to make sure that the assembler names you choose do not
5862 conflict with any other assembler symbols. Also, you must not use a
5863 register name; that would produce completely invalid assembler code. GCC
5864 does not as yet have the ability to store static variables in registers.
5865 Perhaps that will be added.
5867 @node Explicit Reg Vars
5868 @section Variables in Specified Registers
5869 @cindex explicit register variables
5870 @cindex variables in specified registers
5871 @cindex specified registers
5872 @cindex registers, global allocation
5874 GNU C allows you to put a few global variables into specified hardware
5875 registers. You can also specify the register in which an ordinary
5876 register variable should be allocated.
5880 Global register variables reserve registers throughout the program.
5881 This may be useful in programs such as programming language
5882 interpreters which have a couple of global variables that are accessed
5886 Local register variables in specific registers do not reserve the
5887 registers, except at the point where they are used as input or output
5888 operands in an @code{asm} statement and the @code{asm} statement itself is
5889 not deleted. The compiler's data flow analysis is capable of determining
5890 where the specified registers contain live values, and where they are
5891 available for other uses. Stores into local register variables may be deleted
5892 when they appear to be dead according to dataflow analysis. References
5893 to local register variables may be deleted or moved or simplified.
5895 These local variables are sometimes convenient for use with the extended
5896 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
5897 output of the assembler instruction directly into a particular register.
5898 (This will work provided the register you specify fits the constraints
5899 specified for that operand in the @code{asm}.)
5907 @node Global Reg Vars
5908 @subsection Defining Global Register Variables
5909 @cindex global register variables
5910 @cindex registers, global variables in
5912 You can define a global register variable in GNU C like this:
5915 register int *foo asm ("a5");
5919 Here @code{a5} is the name of the register which should be used. Choose a
5920 register which is normally saved and restored by function calls on your
5921 machine, so that library routines will not clobber it.
5923 Naturally the register name is cpu-dependent, so you would need to
5924 conditionalize your program according to cpu type. The register
5925 @code{a5} would be a good choice on a 68000 for a variable of pointer
5926 type. On machines with register windows, be sure to choose a ``global''
5927 register that is not affected magically by the function call mechanism.
5929 In addition, operating systems on one type of cpu may differ in how they
5930 name the registers; then you would need additional conditionals. For
5931 example, some 68000 operating systems call this register @code{%a5}.
5933 Eventually there may be a way of asking the compiler to choose a register
5934 automatically, but first we need to figure out how it should choose and
5935 how to enable you to guide the choice. No solution is evident.
5937 Defining a global register variable in a certain register reserves that
5938 register entirely for this use, at least within the current compilation.
5939 The register will not be allocated for any other purpose in the functions
5940 in the current compilation. The register will not be saved and restored by
5941 these functions. Stores into this register are never deleted even if they
5942 would appear to be dead, but references may be deleted or moved or
5945 It is not safe to access the global register variables from signal
5946 handlers, or from more than one thread of control, because the system
5947 library routines may temporarily use the register for other things (unless
5948 you recompile them specially for the task at hand).
5950 @cindex @code{qsort}, and global register variables
5951 It is not safe for one function that uses a global register variable to
5952 call another such function @code{foo} by way of a third function
5953 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
5954 different source file in which the variable wasn't declared). This is
5955 because @code{lose} might save the register and put some other value there.
5956 For example, you can't expect a global register variable to be available in
5957 the comparison-function that you pass to @code{qsort}, since @code{qsort}
5958 might have put something else in that register. (If you are prepared to
5959 recompile @code{qsort} with the same global register variable, you can
5960 solve this problem.)
5962 If you want to recompile @code{qsort} or other source files which do not
5963 actually use your global register variable, so that they will not use that
5964 register for any other purpose, then it suffices to specify the compiler
5965 option @option{-ffixed-@var{reg}}. You need not actually add a global
5966 register declaration to their source code.
5968 A function which can alter the value of a global register variable cannot
5969 safely be called from a function compiled without this variable, because it
5970 could clobber the value the caller expects to find there on return.
5971 Therefore, the function which is the entry point into the part of the
5972 program that uses the global register variable must explicitly save and
5973 restore the value which belongs to its caller.
5975 @cindex register variable after @code{longjmp}
5976 @cindex global register after @code{longjmp}
5977 @cindex value after @code{longjmp}
5980 On most machines, @code{longjmp} will restore to each global register
5981 variable the value it had at the time of the @code{setjmp}. On some
5982 machines, however, @code{longjmp} will not change the value of global
5983 register variables. To be portable, the function that called @code{setjmp}
5984 should make other arrangements to save the values of the global register
5985 variables, and to restore them in a @code{longjmp}. This way, the same
5986 thing will happen regardless of what @code{longjmp} does.
5988 All global register variable declarations must precede all function
5989 definitions. If such a declaration could appear after function
5990 definitions, the declaration would be too late to prevent the register from
5991 being used for other purposes in the preceding functions.
5993 Global register variables may not have initial values, because an
5994 executable file has no means to supply initial contents for a register.
5996 On the SPARC, there are reports that g3 @dots{} g7 are suitable
5997 registers, but certain library functions, such as @code{getwd}, as well
5998 as the subroutines for division and remainder, modify g3 and g4. g1 and
5999 g2 are local temporaries.
6001 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
6002 Of course, it will not do to use more than a few of those.
6004 @node Local Reg Vars
6005 @subsection Specifying Registers for Local Variables
6006 @cindex local variables, specifying registers
6007 @cindex specifying registers for local variables
6008 @cindex registers for local variables
6010 You can define a local register variable with a specified register
6014 register int *foo asm ("a5");
6018 Here @code{a5} is the name of the register which should be used. Note
6019 that this is the same syntax used for defining global register
6020 variables, but for a local variable it would appear within a function.
6022 Naturally the register name is cpu-dependent, but this is not a
6023 problem, since specific registers are most often useful with explicit
6024 assembler instructions (@pxref{Extended Asm}). Both of these things
6025 generally require that you conditionalize your program according to
6028 In addition, operating systems on one type of cpu may differ in how they
6029 name the registers; then you would need additional conditionals. For
6030 example, some 68000 operating systems call this register @code{%a5}.
6032 Defining such a register variable does not reserve the register; it
6033 remains available for other uses in places where flow control determines
6034 the variable's value is not live.
6036 This option does not guarantee that GCC will generate code that has
6037 this variable in the register you specify at all times. You may not
6038 code an explicit reference to this register in the @emph{assembler
6039 instruction template} part of an @code{asm} statement and assume it will
6040 always refer to this variable. However, using the variable as an
6041 @code{asm} @emph{operand} guarantees that the specified register is used
6044 Stores into local register variables may be deleted when they appear to be dead
6045 according to dataflow analysis. References to local register variables may
6046 be deleted or moved or simplified.
6048 As for global register variables, it's recommended that you choose a
6049 register which is normally saved and restored by function calls on
6050 your machine, so that library routines will not clobber it. A common
6051 pitfall is to initialize multiple call-clobbered registers with
6052 arbitrary expressions, where a function call or library call for an
6053 arithmetic operator will overwrite a register value from a previous
6054 assignment, for example @code{r0} below:
6056 register int *p1 asm ("r0") = @dots{};
6057 register int *p2 asm ("r1") = @dots{};
6059 In those cases, a solution is to use a temporary variable for
6060 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
6062 @node Alternate Keywords
6063 @section Alternate Keywords
6064 @cindex alternate keywords
6065 @cindex keywords, alternate
6067 @option{-ansi} and the various @option{-std} options disable certain
6068 keywords. This causes trouble when you want to use GNU C extensions, or
6069 a general-purpose header file that should be usable by all programs,
6070 including ISO C programs. The keywords @code{asm}, @code{typeof} and
6071 @code{inline} are not available in programs compiled with
6072 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
6073 program compiled with @option{-std=c99} or @option{-std=c1x}). The
6075 @code{restrict} is only available when @option{-std=gnu99} (which will
6076 eventually be the default) or @option{-std=c99} (or the equivalent
6077 @option{-std=iso9899:1999}), or an option for a later standard
6080 The way to solve these problems is to put @samp{__} at the beginning and
6081 end of each problematical keyword. For example, use @code{__asm__}
6082 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
6084 Other C compilers won't accept these alternative keywords; if you want to
6085 compile with another compiler, you can define the alternate keywords as
6086 macros to replace them with the customary keywords. It looks like this:
6094 @findex __extension__
6096 @option{-pedantic} and other options cause warnings for many GNU C extensions.
6098 prevent such warnings within one expression by writing
6099 @code{__extension__} before the expression. @code{__extension__} has no
6100 effect aside from this.
6102 @node Incomplete Enums
6103 @section Incomplete @code{enum} Types
6105 You can define an @code{enum} tag without specifying its possible values.
6106 This results in an incomplete type, much like what you get if you write
6107 @code{struct foo} without describing the elements. A later declaration
6108 which does specify the possible values completes the type.
6110 You can't allocate variables or storage using the type while it is
6111 incomplete. However, you can work with pointers to that type.
6113 This extension may not be very useful, but it makes the handling of
6114 @code{enum} more consistent with the way @code{struct} and @code{union}
6117 This extension is not supported by GNU C++.
6119 @node Function Names
6120 @section Function Names as Strings
6121 @cindex @code{__func__} identifier
6122 @cindex @code{__FUNCTION__} identifier
6123 @cindex @code{__PRETTY_FUNCTION__} identifier
6125 GCC provides three magic variables which hold the name of the current
6126 function, as a string. The first of these is @code{__func__}, which
6127 is part of the C99 standard:
6129 The identifier @code{__func__} is implicitly declared by the translator
6130 as if, immediately following the opening brace of each function
6131 definition, the declaration
6134 static const char __func__[] = "function-name";
6138 appeared, where function-name is the name of the lexically-enclosing
6139 function. This name is the unadorned name of the function.
6141 @code{__FUNCTION__} is another name for @code{__func__}. Older
6142 versions of GCC recognize only this name. However, it is not
6143 standardized. For maximum portability, we recommend you use
6144 @code{__func__}, but provide a fallback definition with the
6148 #if __STDC_VERSION__ < 199901L
6150 # define __func__ __FUNCTION__
6152 # define __func__ "<unknown>"
6157 In C, @code{__PRETTY_FUNCTION__} is yet another name for
6158 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
6159 the type signature of the function as well as its bare name. For
6160 example, this program:
6164 extern int printf (char *, ...);
6171 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
6172 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
6190 __PRETTY_FUNCTION__ = void a::sub(int)
6193 These identifiers are not preprocessor macros. In GCC 3.3 and
6194 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
6195 were treated as string literals; they could be used to initialize
6196 @code{char} arrays, and they could be concatenated with other string
6197 literals. GCC 3.4 and later treat them as variables, like
6198 @code{__func__}. In C++, @code{__FUNCTION__} and
6199 @code{__PRETTY_FUNCTION__} have always been variables.
6201 @node Return Address
6202 @section Getting the Return or Frame Address of a Function
6204 These functions may be used to get information about the callers of a
6207 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
6208 This function returns the return address of the current function, or of
6209 one of its callers. The @var{level} argument is number of frames to
6210 scan up the call stack. A value of @code{0} yields the return address
6211 of the current function, a value of @code{1} yields the return address
6212 of the caller of the current function, and so forth. When inlining
6213 the expected behavior is that the function will return the address of
6214 the function that will be returned to. To work around this behavior use
6215 the @code{noinline} function attribute.
6217 The @var{level} argument must be a constant integer.
6219 On some machines it may be impossible to determine the return address of
6220 any function other than the current one; in such cases, or when the top
6221 of the stack has been reached, this function will return @code{0} or a
6222 random value. In addition, @code{__builtin_frame_address} may be used
6223 to determine if the top of the stack has been reached.
6225 Additional post-processing of the returned value may be needed, see
6226 @code{__builtin_extract_return_address}.
6228 This function should only be used with a nonzero argument for debugging
6232 @deftypefn {Built-in Function} {void *} __builtin_extract_return_address (void *@var{addr})
6233 The address as returned by @code{__builtin_return_address} may have to be fed
6234 through this function to get the actual encoded address. For example, on the
6235 31-bit S/390 platform the highest bit has to be masked out, or on SPARC
6236 platforms an offset has to be added for the true next instruction to be
6239 If no fixup is needed, this function simply passes through @var{addr}.
6242 @deftypefn {Built-in Function} {void *} __builtin_frob_return_address (void *@var{addr})
6243 This function does the reverse of @code{__builtin_extract_return_address}.
6246 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
6247 This function is similar to @code{__builtin_return_address}, but it
6248 returns the address of the function frame rather than the return address
6249 of the function. Calling @code{__builtin_frame_address} with a value of
6250 @code{0} yields the frame address of the current function, a value of
6251 @code{1} yields the frame address of the caller of the current function,
6254 The frame is the area on the stack which holds local variables and saved
6255 registers. The frame address is normally the address of the first word
6256 pushed on to the stack by the function. However, the exact definition
6257 depends upon the processor and the calling convention. If the processor
6258 has a dedicated frame pointer register, and the function has a frame,
6259 then @code{__builtin_frame_address} will return the value of the frame
6262 On some machines it may be impossible to determine the frame address of
6263 any function other than the current one; in such cases, or when the top
6264 of the stack has been reached, this function will return @code{0} if
6265 the first frame pointer is properly initialized by the startup code.
6267 This function should only be used with a nonzero argument for debugging
6271 @node Vector Extensions
6272 @section Using vector instructions through built-in functions
6274 On some targets, the instruction set contains SIMD vector instructions that
6275 operate on multiple values contained in one large register at the same time.
6276 For example, on the i386 the MMX, 3DNow!@: and SSE extensions can be used
6279 The first step in using these extensions is to provide the necessary data
6280 types. This should be done using an appropriate @code{typedef}:
6283 typedef int v4si __attribute__ ((vector_size (16)));
6286 The @code{int} type specifies the base type, while the attribute specifies
6287 the vector size for the variable, measured in bytes. For example, the
6288 declaration above causes the compiler to set the mode for the @code{v4si}
6289 type to be 16 bytes wide and divided into @code{int} sized units. For
6290 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
6291 corresponding mode of @code{foo} will be @acronym{V4SI}.
6293 The @code{vector_size} attribute is only applicable to integral and
6294 float scalars, although arrays, pointers, and function return values
6295 are allowed in conjunction with this construct.
6297 All the basic integer types can be used as base types, both as signed
6298 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
6299 @code{long long}. In addition, @code{float} and @code{double} can be
6300 used to build floating-point vector types.
6302 Specifying a combination that is not valid for the current architecture
6303 will cause GCC to synthesize the instructions using a narrower mode.
6304 For example, if you specify a variable of type @code{V4SI} and your
6305 architecture does not allow for this specific SIMD type, GCC will
6306 produce code that uses 4 @code{SIs}.
6308 The types defined in this manner can be used with a subset of normal C
6309 operations. Currently, GCC will allow using the following operators
6310 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~, %}@.
6312 The operations behave like C++ @code{valarrays}. Addition is defined as
6313 the addition of the corresponding elements of the operands. For
6314 example, in the code below, each of the 4 elements in @var{a} will be
6315 added to the corresponding 4 elements in @var{b} and the resulting
6316 vector will be stored in @var{c}.
6319 typedef int v4si __attribute__ ((vector_size (16)));
6326 Subtraction, multiplication, division, and the logical operations
6327 operate in a similar manner. Likewise, the result of using the unary
6328 minus or complement operators on a vector type is a vector whose
6329 elements are the negative or complemented values of the corresponding
6330 elements in the operand.
6332 In C it is possible to use shifting operators @code{<<}, @code{>>} on
6333 integer-type vectors. The operation is defined as following: @code{@{a0,
6334 a1, @dots{}, an@} >> @{b0, b1, @dots{}, bn@} == @{a0 >> b0, a1 >> b1,
6335 @dots{}, an >> bn@}}@. Vector operands must have the same number of
6336 elements. Additionally second operands can be a scalar integer in which
6337 case the scalar is converted to the type used by the vector operand (with
6338 possible truncation) and each element of this new vector is the scalar's
6340 Consider the following code.
6343 typedef int v4si __attribute__ ((vector_size (16)));
6347 b = a >> 1; /* b = a >> @{1,1,1,1@}; */
6350 In C vectors can be subscripted as if the vector were an array with
6351 the same number of elements and base type. Out of bound accesses
6352 invoke undefined behavior at runtime. Warnings for out of bound
6353 accesses for vector subscription can be enabled with
6354 @option{-Warray-bounds}.
6356 You can declare variables and use them in function calls and returns, as
6357 well as in assignments and some casts. You can specify a vector type as
6358 a return type for a function. Vector types can also be used as function
6359 arguments. It is possible to cast from one vector type to another,
6360 provided they are of the same size (in fact, you can also cast vectors
6361 to and from other datatypes of the same size).
6363 You cannot operate between vectors of different lengths or different
6364 signedness without a cast.
6366 A port that supports hardware vector operations, usually provides a set
6367 of built-in functions that can be used to operate on vectors. For
6368 example, a function to add two vectors and multiply the result by a
6369 third could look like this:
6372 v4si f (v4si a, v4si b, v4si c)
6374 v4si tmp = __builtin_addv4si (a, b);
6375 return __builtin_mulv4si (tmp, c);
6382 @findex __builtin_offsetof
6384 GCC implements for both C and C++ a syntactic extension to implement
6385 the @code{offsetof} macro.
6389 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
6391 offsetof_member_designator:
6393 | offsetof_member_designator "." @code{identifier}
6394 | offsetof_member_designator "[" @code{expr} "]"
6397 This extension is sufficient such that
6400 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
6403 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
6404 may be dependent. In either case, @var{member} may consist of a single
6405 identifier, or a sequence of member accesses and array references.
6407 @node Atomic Builtins
6408 @section Built-in functions for atomic memory access
6410 The following builtins are intended to be compatible with those described
6411 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
6412 section 7.4. As such, they depart from the normal GCC practice of using
6413 the ``__builtin_'' prefix, and further that they are overloaded such that
6414 they work on multiple types.
6416 The definition given in the Intel documentation allows only for the use of
6417 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
6418 counterparts. GCC will allow any integral scalar or pointer type that is
6419 1, 2, 4 or 8 bytes in length.
6421 Not all operations are supported by all target processors. If a particular
6422 operation cannot be implemented on the target processor, a warning will be
6423 generated and a call an external function will be generated. The external
6424 function will carry the same name as the builtin, with an additional suffix
6425 @samp{_@var{n}} where @var{n} is the size of the data type.
6427 @c ??? Should we have a mechanism to suppress this warning? This is almost
6428 @c useful for implementing the operation under the control of an external
6431 In most cases, these builtins are considered a @dfn{full barrier}. That is,
6432 no memory operand will be moved across the operation, either forward or
6433 backward. Further, instructions will be issued as necessary to prevent the
6434 processor from speculating loads across the operation and from queuing stores
6435 after the operation.
6437 All of the routines are described in the Intel documentation to take
6438 ``an optional list of variables protected by the memory barrier''. It's
6439 not clear what is meant by that; it could mean that @emph{only} the
6440 following variables are protected, or it could mean that these variables
6441 should in addition be protected. At present GCC ignores this list and
6442 protects all variables which are globally accessible. If in the future
6443 we make some use of this list, an empty list will continue to mean all
6444 globally accessible variables.
6447 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
6448 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
6449 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
6450 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
6451 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
6452 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
6453 @findex __sync_fetch_and_add
6454 @findex __sync_fetch_and_sub
6455 @findex __sync_fetch_and_or
6456 @findex __sync_fetch_and_and
6457 @findex __sync_fetch_and_xor
6458 @findex __sync_fetch_and_nand
6459 These builtins perform the operation suggested by the name, and
6460 returns the value that had previously been in memory. That is,
6463 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
6464 @{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; @} // nand
6467 @emph{Note:} GCC 4.4 and later implement @code{__sync_fetch_and_nand}
6468 builtin as @code{*ptr = ~(tmp & value)} instead of @code{*ptr = ~tmp & value}.
6470 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
6471 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
6472 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
6473 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
6474 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
6475 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
6476 @findex __sync_add_and_fetch
6477 @findex __sync_sub_and_fetch
6478 @findex __sync_or_and_fetch
6479 @findex __sync_and_and_fetch
6480 @findex __sync_xor_and_fetch
6481 @findex __sync_nand_and_fetch
6482 These builtins perform the operation suggested by the name, and
6483 return the new value. That is,
6486 @{ *ptr @var{op}= value; return *ptr; @}
6487 @{ *ptr = ~(*ptr & value); return *ptr; @} // nand
6490 @emph{Note:} GCC 4.4 and later implement @code{__sync_nand_and_fetch}
6491 builtin as @code{*ptr = ~(*ptr & value)} instead of
6492 @code{*ptr = ~*ptr & value}.
6494 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6495 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6496 @findex __sync_bool_compare_and_swap
6497 @findex __sync_val_compare_and_swap
6498 These builtins perform an atomic compare and swap. That is, if the current
6499 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
6502 The ``bool'' version returns true if the comparison is successful and
6503 @var{newval} was written. The ``val'' version returns the contents
6504 of @code{*@var{ptr}} before the operation.
6506 @item __sync_synchronize (...)
6507 @findex __sync_synchronize
6508 This builtin issues a full memory barrier.
6510 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
6511 @findex __sync_lock_test_and_set
6512 This builtin, as described by Intel, is not a traditional test-and-set
6513 operation, but rather an atomic exchange operation. It writes @var{value}
6514 into @code{*@var{ptr}}, and returns the previous contents of
6517 Many targets have only minimal support for such locks, and do not support
6518 a full exchange operation. In this case, a target may support reduced
6519 functionality here by which the @emph{only} valid value to store is the
6520 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
6521 is implementation defined.
6523 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
6524 This means that references after the builtin cannot move to (or be
6525 speculated to) before the builtin, but previous memory stores may not
6526 be globally visible yet, and previous memory loads may not yet be
6529 @item void __sync_lock_release (@var{type} *ptr, ...)
6530 @findex __sync_lock_release
6531 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
6532 Normally this means writing the constant 0 to @code{*@var{ptr}}.
6534 This builtin is not a full barrier, but rather a @dfn{release barrier}.
6535 This means that all previous memory stores are globally visible, and all
6536 previous memory loads have been satisfied, but following memory reads
6537 are not prevented from being speculated to before the barrier.
6540 @node Object Size Checking
6541 @section Object Size Checking Builtins
6542 @findex __builtin_object_size
6543 @findex __builtin___memcpy_chk
6544 @findex __builtin___mempcpy_chk
6545 @findex __builtin___memmove_chk
6546 @findex __builtin___memset_chk
6547 @findex __builtin___strcpy_chk
6548 @findex __builtin___stpcpy_chk
6549 @findex __builtin___strncpy_chk
6550 @findex __builtin___strcat_chk
6551 @findex __builtin___strncat_chk
6552 @findex __builtin___sprintf_chk
6553 @findex __builtin___snprintf_chk
6554 @findex __builtin___vsprintf_chk
6555 @findex __builtin___vsnprintf_chk
6556 @findex __builtin___printf_chk
6557 @findex __builtin___vprintf_chk
6558 @findex __builtin___fprintf_chk
6559 @findex __builtin___vfprintf_chk
6561 GCC implements a limited buffer overflow protection mechanism
6562 that can prevent some buffer overflow attacks.
6564 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
6565 is a built-in construct that returns a constant number of bytes from
6566 @var{ptr} to the end of the object @var{ptr} pointer points to
6567 (if known at compile time). @code{__builtin_object_size} never evaluates
6568 its arguments for side-effects. If there are any side-effects in them, it
6569 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6570 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
6571 point to and all of them are known at compile time, the returned number
6572 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
6573 0 and minimum if nonzero. If it is not possible to determine which objects
6574 @var{ptr} points to at compile time, @code{__builtin_object_size} should
6575 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6576 for @var{type} 2 or 3.
6578 @var{type} is an integer constant from 0 to 3. If the least significant
6579 bit is clear, objects are whole variables, if it is set, a closest
6580 surrounding subobject is considered the object a pointer points to.
6581 The second bit determines if maximum or minimum of remaining bytes
6585 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
6586 char *p = &var.buf1[1], *q = &var.b;
6588 /* Here the object p points to is var. */
6589 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
6590 /* The subobject p points to is var.buf1. */
6591 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
6592 /* The object q points to is var. */
6593 assert (__builtin_object_size (q, 0)
6594 == (char *) (&var + 1) - (char *) &var.b);
6595 /* The subobject q points to is var.b. */
6596 assert (__builtin_object_size (q, 1) == sizeof (var.b));
6600 There are built-in functions added for many common string operation
6601 functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk}
6602 built-in is provided. This built-in has an additional last argument,
6603 which is the number of bytes remaining in object the @var{dest}
6604 argument points to or @code{(size_t) -1} if the size is not known.
6606 The built-in functions are optimized into the normal string functions
6607 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
6608 it is known at compile time that the destination object will not
6609 be overflown. If the compiler can determine at compile time the
6610 object will be always overflown, it issues a warning.
6612 The intended use can be e.g.
6616 #define bos0(dest) __builtin_object_size (dest, 0)
6617 #define memcpy(dest, src, n) \
6618 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
6622 /* It is unknown what object p points to, so this is optimized
6623 into plain memcpy - no checking is possible. */
6624 memcpy (p, "abcde", n);
6625 /* Destination is known and length too. It is known at compile
6626 time there will be no overflow. */
6627 memcpy (&buf[5], "abcde", 5);
6628 /* Destination is known, but the length is not known at compile time.
6629 This will result in __memcpy_chk call that can check for overflow
6631 memcpy (&buf[5], "abcde", n);
6632 /* Destination is known and it is known at compile time there will
6633 be overflow. There will be a warning and __memcpy_chk call that
6634 will abort the program at runtime. */
6635 memcpy (&buf[6], "abcde", 5);
6638 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
6639 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
6640 @code{strcat} and @code{strncat}.
6642 There are also checking built-in functions for formatted output functions.
6644 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
6645 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6646 const char *fmt, ...);
6647 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
6649 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6650 const char *fmt, va_list ap);
6653 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
6654 etc.@: functions and can contain implementation specific flags on what
6655 additional security measures the checking function might take, such as
6656 handling @code{%n} differently.
6658 The @var{os} argument is the object size @var{s} points to, like in the
6659 other built-in functions. There is a small difference in the behavior
6660 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
6661 optimized into the non-checking functions only if @var{flag} is 0, otherwise
6662 the checking function is called with @var{os} argument set to
6665 In addition to this, there are checking built-in functions
6666 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
6667 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
6668 These have just one additional argument, @var{flag}, right before
6669 format string @var{fmt}. If the compiler is able to optimize them to
6670 @code{fputc} etc.@: functions, it will, otherwise the checking function
6671 should be called and the @var{flag} argument passed to it.
6673 @node Other Builtins
6674 @section Other built-in functions provided by GCC
6675 @cindex built-in functions
6676 @findex __builtin_fpclassify
6677 @findex __builtin_isfinite
6678 @findex __builtin_isnormal
6679 @findex __builtin_isgreater
6680 @findex __builtin_isgreaterequal
6681 @findex __builtin_isinf_sign
6682 @findex __builtin_isless
6683 @findex __builtin_islessequal
6684 @findex __builtin_islessgreater
6685 @findex __builtin_isunordered
6686 @findex __builtin_powi
6687 @findex __builtin_powif
6688 @findex __builtin_powil
6846 @findex fprintf_unlocked
6848 @findex fputs_unlocked
6965 @findex printf_unlocked
6997 @findex significandf
6998 @findex significandl
7069 GCC provides a large number of built-in functions other than the ones
7070 mentioned above. Some of these are for internal use in the processing
7071 of exceptions or variable-length argument lists and will not be
7072 documented here because they may change from time to time; we do not
7073 recommend general use of these functions.
7075 The remaining functions are provided for optimization purposes.
7077 @opindex fno-builtin
7078 GCC includes built-in versions of many of the functions in the standard
7079 C library. The versions prefixed with @code{__builtin_} will always be
7080 treated as having the same meaning as the C library function even if you
7081 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
7082 Many of these functions are only optimized in certain cases; if they are
7083 not optimized in a particular case, a call to the library function will
7088 Outside strict ISO C mode (@option{-ansi}, @option{-std=c90},
7089 @option{-std=c99} or @option{-std=c1x}), the functions
7090 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
7091 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
7092 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
7093 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
7094 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
7095 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
7096 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
7097 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
7098 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
7099 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
7100 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
7101 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
7102 @code{signbitd64}, @code{signbitd128}, @code{significandf},
7103 @code{significandl}, @code{significand}, @code{sincosf},
7104 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
7105 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
7106 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
7107 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
7109 may be handled as built-in functions.
7110 All these functions have corresponding versions
7111 prefixed with @code{__builtin_}, which may be used even in strict C90
7114 The ISO C99 functions
7115 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
7116 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
7117 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
7118 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
7119 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
7120 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
7121 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
7122 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
7123 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
7124 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
7125 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
7126 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
7127 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
7128 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
7129 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
7130 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
7131 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
7132 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
7133 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
7134 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
7135 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
7136 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
7137 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
7138 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
7139 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
7140 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
7141 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
7142 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
7143 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
7144 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
7145 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
7146 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
7147 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
7148 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
7149 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
7150 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
7151 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
7152 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
7153 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
7154 are handled as built-in functions
7155 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
7157 There are also built-in versions of the ISO C99 functions
7158 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
7159 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
7160 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
7161 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
7162 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
7163 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
7164 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
7165 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
7166 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
7167 that are recognized in any mode since ISO C90 reserves these names for
7168 the purpose to which ISO C99 puts them. All these functions have
7169 corresponding versions prefixed with @code{__builtin_}.
7171 The ISO C94 functions
7172 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
7173 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
7174 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
7176 are handled as built-in functions
7177 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
7179 The ISO C90 functions
7180 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
7181 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
7182 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
7183 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
7184 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
7185 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
7186 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
7187 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
7188 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
7189 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
7190 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
7191 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
7192 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
7193 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
7194 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
7195 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
7196 are all recognized as built-in functions unless
7197 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
7198 is specified for an individual function). All of these functions have
7199 corresponding versions prefixed with @code{__builtin_}.
7201 GCC provides built-in versions of the ISO C99 floating point comparison
7202 macros that avoid raising exceptions for unordered operands. They have
7203 the same names as the standard macros ( @code{isgreater},
7204 @code{isgreaterequal}, @code{isless}, @code{islessequal},
7205 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
7206 prefixed. We intend for a library implementor to be able to simply
7207 @code{#define} each standard macro to its built-in equivalent.
7208 In the same fashion, GCC provides @code{fpclassify}, @code{isfinite},
7209 @code{isinf_sign} and @code{isnormal} built-ins used with
7210 @code{__builtin_} prefixed. The @code{isinf} and @code{isnan}
7211 builtins appear both with and without the @code{__builtin_} prefix.
7213 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
7215 You can use the built-in function @code{__builtin_types_compatible_p} to
7216 determine whether two types are the same.
7218 This built-in function returns 1 if the unqualified versions of the
7219 types @var{type1} and @var{type2} (which are types, not expressions) are
7220 compatible, 0 otherwise. The result of this built-in function can be
7221 used in integer constant expressions.
7223 This built-in function ignores top level qualifiers (e.g., @code{const},
7224 @code{volatile}). For example, @code{int} is equivalent to @code{const
7227 The type @code{int[]} and @code{int[5]} are compatible. On the other
7228 hand, @code{int} and @code{char *} are not compatible, even if the size
7229 of their types, on the particular architecture are the same. Also, the
7230 amount of pointer indirection is taken into account when determining
7231 similarity. Consequently, @code{short *} is not similar to
7232 @code{short **}. Furthermore, two types that are typedefed are
7233 considered compatible if their underlying types are compatible.
7235 An @code{enum} type is not considered to be compatible with another
7236 @code{enum} type even if both are compatible with the same integer
7237 type; this is what the C standard specifies.
7238 For example, @code{enum @{foo, bar@}} is not similar to
7239 @code{enum @{hot, dog@}}.
7241 You would typically use this function in code whose execution varies
7242 depending on the arguments' types. For example:
7247 typeof (x) tmp = (x); \
7248 if (__builtin_types_compatible_p (typeof (x), long double)) \
7249 tmp = foo_long_double (tmp); \
7250 else if (__builtin_types_compatible_p (typeof (x), double)) \
7251 tmp = foo_double (tmp); \
7252 else if (__builtin_types_compatible_p (typeof (x), float)) \
7253 tmp = foo_float (tmp); \
7260 @emph{Note:} This construct is only available for C@.
7264 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
7266 You can use the built-in function @code{__builtin_choose_expr} to
7267 evaluate code depending on the value of a constant expression. This
7268 built-in function returns @var{exp1} if @var{const_exp}, which is an
7269 integer constant expression, is nonzero. Otherwise it returns @var{exp2}.
7271 This built-in function is analogous to the @samp{? :} operator in C,
7272 except that the expression returned has its type unaltered by promotion
7273 rules. Also, the built-in function does not evaluate the expression
7274 that was not chosen. For example, if @var{const_exp} evaluates to true,
7275 @var{exp2} is not evaluated even if it has side-effects.
7277 This built-in function can return an lvalue if the chosen argument is an
7280 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
7281 type. Similarly, if @var{exp2} is returned, its return type is the same
7288 __builtin_choose_expr ( \
7289 __builtin_types_compatible_p (typeof (x), double), \
7291 __builtin_choose_expr ( \
7292 __builtin_types_compatible_p (typeof (x), float), \
7294 /* @r{The void expression results in a compile-time error} \
7295 @r{when assigning the result to something.} */ \
7299 @emph{Note:} This construct is only available for C@. Furthermore, the
7300 unused expression (@var{exp1} or @var{exp2} depending on the value of
7301 @var{const_exp}) may still generate syntax errors. This may change in
7306 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
7307 You can use the built-in function @code{__builtin_constant_p} to
7308 determine if a value is known to be constant at compile-time and hence
7309 that GCC can perform constant-folding on expressions involving that
7310 value. The argument of the function is the value to test. The function
7311 returns the integer 1 if the argument is known to be a compile-time
7312 constant and 0 if it is not known to be a compile-time constant. A
7313 return of 0 does not indicate that the value is @emph{not} a constant,
7314 but merely that GCC cannot prove it is a constant with the specified
7315 value of the @option{-O} option.
7317 You would typically use this function in an embedded application where
7318 memory was a critical resource. If you have some complex calculation,
7319 you may want it to be folded if it involves constants, but need to call
7320 a function if it does not. For example:
7323 #define Scale_Value(X) \
7324 (__builtin_constant_p (X) \
7325 ? ((X) * SCALE + OFFSET) : Scale (X))
7328 You may use this built-in function in either a macro or an inline
7329 function. However, if you use it in an inlined function and pass an
7330 argument of the function as the argument to the built-in, GCC will
7331 never return 1 when you call the inline function with a string constant
7332 or compound literal (@pxref{Compound Literals}) and will not return 1
7333 when you pass a constant numeric value to the inline function unless you
7334 specify the @option{-O} option.
7336 You may also use @code{__builtin_constant_p} in initializers for static
7337 data. For instance, you can write
7340 static const int table[] = @{
7341 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
7347 This is an acceptable initializer even if @var{EXPRESSION} is not a
7348 constant expression, including the case where
7349 @code{__builtin_constant_p} returns 1 because @var{EXPRESSION} can be
7350 folded to a constant but @var{EXPRESSION} contains operands that would
7351 not otherwise be permitted in a static initializer (for example,
7352 @code{0 && foo ()}). GCC must be more conservative about evaluating the
7353 built-in in this case, because it has no opportunity to perform
7356 Previous versions of GCC did not accept this built-in in data
7357 initializers. The earliest version where it is completely safe is
7361 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
7362 @opindex fprofile-arcs
7363 You may use @code{__builtin_expect} to provide the compiler with
7364 branch prediction information. In general, you should prefer to
7365 use actual profile feedback for this (@option{-fprofile-arcs}), as
7366 programmers are notoriously bad at predicting how their programs
7367 actually perform. However, there are applications in which this
7368 data is hard to collect.
7370 The return value is the value of @var{exp}, which should be an integral
7371 expression. The semantics of the built-in are that it is expected that
7372 @var{exp} == @var{c}. For example:
7375 if (__builtin_expect (x, 0))
7380 would indicate that we do not expect to call @code{foo}, since
7381 we expect @code{x} to be zero. Since you are limited to integral
7382 expressions for @var{exp}, you should use constructions such as
7385 if (__builtin_expect (ptr != NULL, 1))
7390 when testing pointer or floating-point values.
7393 @deftypefn {Built-in Function} void __builtin_trap (void)
7394 This function causes the program to exit abnormally. GCC implements
7395 this function by using a target-dependent mechanism (such as
7396 intentionally executing an illegal instruction) or by calling
7397 @code{abort}. The mechanism used may vary from release to release so
7398 you should not rely on any particular implementation.
7401 @deftypefn {Built-in Function} void __builtin_unreachable (void)
7402 If control flow reaches the point of the @code{__builtin_unreachable},
7403 the program is undefined. It is useful in situations where the
7404 compiler cannot deduce the unreachability of the code.
7406 One such case is immediately following an @code{asm} statement that
7407 will either never terminate, or one that transfers control elsewhere
7408 and never returns. In this example, without the
7409 @code{__builtin_unreachable}, GCC would issue a warning that control
7410 reaches the end of a non-void function. It would also generate code
7411 to return after the @code{asm}.
7414 int f (int c, int v)
7422 asm("jmp error_handler");
7423 __builtin_unreachable ();
7428 Because the @code{asm} statement unconditionally transfers control out
7429 of the function, control will never reach the end of the function
7430 body. The @code{__builtin_unreachable} is in fact unreachable and
7431 communicates this fact to the compiler.
7433 Another use for @code{__builtin_unreachable} is following a call a
7434 function that never returns but that is not declared
7435 @code{__attribute__((noreturn))}, as in this example:
7438 void function_that_never_returns (void);
7448 function_that_never_returns ();
7449 __builtin_unreachable ();
7456 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
7457 This function is used to flush the processor's instruction cache for
7458 the region of memory between @var{begin} inclusive and @var{end}
7459 exclusive. Some targets require that the instruction cache be
7460 flushed, after modifying memory containing code, in order to obtain
7461 deterministic behavior.
7463 If the target does not require instruction cache flushes,
7464 @code{__builtin___clear_cache} has no effect. Otherwise either
7465 instructions are emitted in-line to clear the instruction cache or a
7466 call to the @code{__clear_cache} function in libgcc is made.
7469 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
7470 This function is used to minimize cache-miss latency by moving data into
7471 a cache before it is accessed.
7472 You can insert calls to @code{__builtin_prefetch} into code for which
7473 you know addresses of data in memory that is likely to be accessed soon.
7474 If the target supports them, data prefetch instructions will be generated.
7475 If the prefetch is done early enough before the access then the data will
7476 be in the cache by the time it is accessed.
7478 The value of @var{addr} is the address of the memory to prefetch.
7479 There are two optional arguments, @var{rw} and @var{locality}.
7480 The value of @var{rw} is a compile-time constant one or zero; one
7481 means that the prefetch is preparing for a write to the memory address
7482 and zero, the default, means that the prefetch is preparing for a read.
7483 The value @var{locality} must be a compile-time constant integer between
7484 zero and three. A value of zero means that the data has no temporal
7485 locality, so it need not be left in the cache after the access. A value
7486 of three means that the data has a high degree of temporal locality and
7487 should be left in all levels of cache possible. Values of one and two
7488 mean, respectively, a low or moderate degree of temporal locality. The
7492 for (i = 0; i < n; i++)
7495 __builtin_prefetch (&a[i+j], 1, 1);
7496 __builtin_prefetch (&b[i+j], 0, 1);
7501 Data prefetch does not generate faults if @var{addr} is invalid, but
7502 the address expression itself must be valid. For example, a prefetch
7503 of @code{p->next} will not fault if @code{p->next} is not a valid
7504 address, but evaluation will fault if @code{p} is not a valid address.
7506 If the target does not support data prefetch, the address expression
7507 is evaluated if it includes side effects but no other code is generated
7508 and GCC does not issue a warning.
7511 @deftypefn {Built-in Function} double __builtin_huge_val (void)
7512 Returns a positive infinity, if supported by the floating-point format,
7513 else @code{DBL_MAX}. This function is suitable for implementing the
7514 ISO C macro @code{HUGE_VAL}.
7517 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
7518 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
7521 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
7522 Similar to @code{__builtin_huge_val}, except the return
7523 type is @code{long double}.
7526 @deftypefn {Built-in Function} int __builtin_fpclassify (int, int, int, int, int, ...)
7527 This built-in implements the C99 fpclassify functionality. The first
7528 five int arguments should be the target library's notion of the
7529 possible FP classes and are used for return values. They must be
7530 constant values and they must appear in this order: @code{FP_NAN},
7531 @code{FP_INFINITE}, @code{FP_NORMAL}, @code{FP_SUBNORMAL} and
7532 @code{FP_ZERO}. The ellipsis is for exactly one floating point value
7533 to classify. GCC treats the last argument as type-generic, which
7534 means it does not do default promotion from float to double.
7537 @deftypefn {Built-in Function} double __builtin_inf (void)
7538 Similar to @code{__builtin_huge_val}, except a warning is generated
7539 if the target floating-point format does not support infinities.
7542 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
7543 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
7546 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
7547 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
7550 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
7551 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
7554 @deftypefn {Built-in Function} float __builtin_inff (void)
7555 Similar to @code{__builtin_inf}, except the return type is @code{float}.
7556 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
7559 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
7560 Similar to @code{__builtin_inf}, except the return
7561 type is @code{long double}.
7564 @deftypefn {Built-in Function} int __builtin_isinf_sign (...)
7565 Similar to @code{isinf}, except the return value will be negative for
7566 an argument of @code{-Inf}. Note while the parameter list is an
7567 ellipsis, this function only accepts exactly one floating point
7568 argument. GCC treats this parameter as type-generic, which means it
7569 does not do default promotion from float to double.
7572 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
7573 This is an implementation of the ISO C99 function @code{nan}.
7575 Since ISO C99 defines this function in terms of @code{strtod}, which we
7576 do not implement, a description of the parsing is in order. The string
7577 is parsed as by @code{strtol}; that is, the base is recognized by
7578 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
7579 in the significand such that the least significant bit of the number
7580 is at the least significant bit of the significand. The number is
7581 truncated to fit the significand field provided. The significand is
7582 forced to be a quiet NaN@.
7584 This function, if given a string literal all of which would have been
7585 consumed by strtol, is evaluated early enough that it is considered a
7586 compile-time constant.
7589 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
7590 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
7593 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
7594 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
7597 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
7598 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
7601 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
7602 Similar to @code{__builtin_nan}, except the return type is @code{float}.
7605 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
7606 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
7609 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
7610 Similar to @code{__builtin_nan}, except the significand is forced
7611 to be a signaling NaN@. The @code{nans} function is proposed by
7612 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
7615 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
7616 Similar to @code{__builtin_nans}, except the return type is @code{float}.
7619 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
7620 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
7623 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
7624 Returns one plus the index of the least significant 1-bit of @var{x}, or
7625 if @var{x} is zero, returns zero.
7628 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
7629 Returns the number of leading 0-bits in @var{x}, starting at the most
7630 significant bit position. If @var{x} is 0, the result is undefined.
7633 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
7634 Returns the number of trailing 0-bits in @var{x}, starting at the least
7635 significant bit position. If @var{x} is 0, the result is undefined.
7638 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
7639 Returns the number of 1-bits in @var{x}.
7642 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
7643 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
7647 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
7648 Similar to @code{__builtin_ffs}, except the argument type is
7649 @code{unsigned long}.
7652 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
7653 Similar to @code{__builtin_clz}, except the argument type is
7654 @code{unsigned long}.
7657 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
7658 Similar to @code{__builtin_ctz}, except the argument type is
7659 @code{unsigned long}.
7662 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
7663 Similar to @code{__builtin_popcount}, except the argument type is
7664 @code{unsigned long}.
7667 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
7668 Similar to @code{__builtin_parity}, except the argument type is
7669 @code{unsigned long}.
7672 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
7673 Similar to @code{__builtin_ffs}, except the argument type is
7674 @code{unsigned long long}.
7677 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
7678 Similar to @code{__builtin_clz}, except the argument type is
7679 @code{unsigned long long}.
7682 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
7683 Similar to @code{__builtin_ctz}, except the argument type is
7684 @code{unsigned long long}.
7687 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
7688 Similar to @code{__builtin_popcount}, except the argument type is
7689 @code{unsigned long long}.
7692 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
7693 Similar to @code{__builtin_parity}, except the argument type is
7694 @code{unsigned long long}.
7697 @deftypefn {Built-in Function} double __builtin_powi (double, int)
7698 Returns the first argument raised to the power of the second. Unlike the
7699 @code{pow} function no guarantees about precision and rounding are made.
7702 @deftypefn {Built-in Function} float __builtin_powif (float, int)
7703 Similar to @code{__builtin_powi}, except the argument and return types
7707 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
7708 Similar to @code{__builtin_powi}, except the argument and return types
7709 are @code{long double}.
7712 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
7713 Returns @var{x} with the order of the bytes reversed; for example,
7714 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
7718 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
7719 Similar to @code{__builtin_bswap32}, except the argument and return types
7723 @node Target Builtins
7724 @section Built-in Functions Specific to Particular Target Machines
7726 On some target machines, GCC supports many built-in functions specific
7727 to those machines. Generally these generate calls to specific machine
7728 instructions, but allow the compiler to schedule those calls.
7731 * Alpha Built-in Functions::
7732 * ARM iWMMXt Built-in Functions::
7733 * ARM NEON Intrinsics::
7734 * Blackfin Built-in Functions::
7735 * FR-V Built-in Functions::
7736 * X86 Built-in Functions::
7737 * MIPS DSP Built-in Functions::
7738 * MIPS Paired-Single Support::
7739 * MIPS Loongson Built-in Functions::
7740 * Other MIPS Built-in Functions::
7741 * picoChip Built-in Functions::
7742 * PowerPC AltiVec/VSX Built-in Functions::
7743 * RX Built-in Functions::
7744 * SPARC VIS Built-in Functions::
7745 * SPU Built-in Functions::
7748 @node Alpha Built-in Functions
7749 @subsection Alpha Built-in Functions
7751 These built-in functions are available for the Alpha family of
7752 processors, depending on the command-line switches used.
7754 The following built-in functions are always available. They
7755 all generate the machine instruction that is part of the name.
7758 long __builtin_alpha_implver (void)
7759 long __builtin_alpha_rpcc (void)
7760 long __builtin_alpha_amask (long)
7761 long __builtin_alpha_cmpbge (long, long)
7762 long __builtin_alpha_extbl (long, long)
7763 long __builtin_alpha_extwl (long, long)
7764 long __builtin_alpha_extll (long, long)
7765 long __builtin_alpha_extql (long, long)
7766 long __builtin_alpha_extwh (long, long)
7767 long __builtin_alpha_extlh (long, long)
7768 long __builtin_alpha_extqh (long, long)
7769 long __builtin_alpha_insbl (long, long)
7770 long __builtin_alpha_inswl (long, long)
7771 long __builtin_alpha_insll (long, long)
7772 long __builtin_alpha_insql (long, long)
7773 long __builtin_alpha_inswh (long, long)
7774 long __builtin_alpha_inslh (long, long)
7775 long __builtin_alpha_insqh (long, long)
7776 long __builtin_alpha_mskbl (long, long)
7777 long __builtin_alpha_mskwl (long, long)
7778 long __builtin_alpha_mskll (long, long)
7779 long __builtin_alpha_mskql (long, long)
7780 long __builtin_alpha_mskwh (long, long)
7781 long __builtin_alpha_msklh (long, long)
7782 long __builtin_alpha_mskqh (long, long)
7783 long __builtin_alpha_umulh (long, long)
7784 long __builtin_alpha_zap (long, long)
7785 long __builtin_alpha_zapnot (long, long)
7788 The following built-in functions are always with @option{-mmax}
7789 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
7790 later. They all generate the machine instruction that is part
7794 long __builtin_alpha_pklb (long)
7795 long __builtin_alpha_pkwb (long)
7796 long __builtin_alpha_unpkbl (long)
7797 long __builtin_alpha_unpkbw (long)
7798 long __builtin_alpha_minub8 (long, long)
7799 long __builtin_alpha_minsb8 (long, long)
7800 long __builtin_alpha_minuw4 (long, long)
7801 long __builtin_alpha_minsw4 (long, long)
7802 long __builtin_alpha_maxub8 (long, long)
7803 long __builtin_alpha_maxsb8 (long, long)
7804 long __builtin_alpha_maxuw4 (long, long)
7805 long __builtin_alpha_maxsw4 (long, long)
7806 long __builtin_alpha_perr (long, long)
7809 The following built-in functions are always with @option{-mcix}
7810 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
7811 later. They all generate the machine instruction that is part
7815 long __builtin_alpha_cttz (long)
7816 long __builtin_alpha_ctlz (long)
7817 long __builtin_alpha_ctpop (long)
7820 The following builtins are available on systems that use the OSF/1
7821 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
7822 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
7823 @code{rdval} and @code{wrval}.
7826 void *__builtin_thread_pointer (void)
7827 void __builtin_set_thread_pointer (void *)
7830 @node ARM iWMMXt Built-in Functions
7831 @subsection ARM iWMMXt Built-in Functions
7833 These built-in functions are available for the ARM family of
7834 processors when the @option{-mcpu=iwmmxt} switch is used:
7837 typedef int v2si __attribute__ ((vector_size (8)));
7838 typedef short v4hi __attribute__ ((vector_size (8)));
7839 typedef char v8qi __attribute__ ((vector_size (8)));
7841 int __builtin_arm_getwcx (int)
7842 void __builtin_arm_setwcx (int, int)
7843 int __builtin_arm_textrmsb (v8qi, int)
7844 int __builtin_arm_textrmsh (v4hi, int)
7845 int __builtin_arm_textrmsw (v2si, int)
7846 int __builtin_arm_textrmub (v8qi, int)
7847 int __builtin_arm_textrmuh (v4hi, int)
7848 int __builtin_arm_textrmuw (v2si, int)
7849 v8qi __builtin_arm_tinsrb (v8qi, int)
7850 v4hi __builtin_arm_tinsrh (v4hi, int)
7851 v2si __builtin_arm_tinsrw (v2si, int)
7852 long long __builtin_arm_tmia (long long, int, int)
7853 long long __builtin_arm_tmiabb (long long, int, int)
7854 long long __builtin_arm_tmiabt (long long, int, int)
7855 long long __builtin_arm_tmiaph (long long, int, int)
7856 long long __builtin_arm_tmiatb (long long, int, int)
7857 long long __builtin_arm_tmiatt (long long, int, int)
7858 int __builtin_arm_tmovmskb (v8qi)
7859 int __builtin_arm_tmovmskh (v4hi)
7860 int __builtin_arm_tmovmskw (v2si)
7861 long long __builtin_arm_waccb (v8qi)
7862 long long __builtin_arm_wacch (v4hi)
7863 long long __builtin_arm_waccw (v2si)
7864 v8qi __builtin_arm_waddb (v8qi, v8qi)
7865 v8qi __builtin_arm_waddbss (v8qi, v8qi)
7866 v8qi __builtin_arm_waddbus (v8qi, v8qi)
7867 v4hi __builtin_arm_waddh (v4hi, v4hi)
7868 v4hi __builtin_arm_waddhss (v4hi, v4hi)
7869 v4hi __builtin_arm_waddhus (v4hi, v4hi)
7870 v2si __builtin_arm_waddw (v2si, v2si)
7871 v2si __builtin_arm_waddwss (v2si, v2si)
7872 v2si __builtin_arm_waddwus (v2si, v2si)
7873 v8qi __builtin_arm_walign (v8qi, v8qi, int)
7874 long long __builtin_arm_wand(long long, long long)
7875 long long __builtin_arm_wandn (long long, long long)
7876 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
7877 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
7878 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
7879 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
7880 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
7881 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
7882 v2si __builtin_arm_wcmpeqw (v2si, v2si)
7883 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
7884 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
7885 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
7886 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
7887 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
7888 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
7889 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
7890 long long __builtin_arm_wmacsz (v4hi, v4hi)
7891 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
7892 long long __builtin_arm_wmacuz (v4hi, v4hi)
7893 v4hi __builtin_arm_wmadds (v4hi, v4hi)
7894 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
7895 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
7896 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
7897 v2si __builtin_arm_wmaxsw (v2si, v2si)
7898 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
7899 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
7900 v2si __builtin_arm_wmaxuw (v2si, v2si)
7901 v8qi __builtin_arm_wminsb (v8qi, v8qi)
7902 v4hi __builtin_arm_wminsh (v4hi, v4hi)
7903 v2si __builtin_arm_wminsw (v2si, v2si)
7904 v8qi __builtin_arm_wminub (v8qi, v8qi)
7905 v4hi __builtin_arm_wminuh (v4hi, v4hi)
7906 v2si __builtin_arm_wminuw (v2si, v2si)
7907 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
7908 v4hi __builtin_arm_wmulul (v4hi, v4hi)
7909 v4hi __builtin_arm_wmulum (v4hi, v4hi)
7910 long long __builtin_arm_wor (long long, long long)
7911 v2si __builtin_arm_wpackdss (long long, long long)
7912 v2si __builtin_arm_wpackdus (long long, long long)
7913 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
7914 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
7915 v4hi __builtin_arm_wpackwss (v2si, v2si)
7916 v4hi __builtin_arm_wpackwus (v2si, v2si)
7917 long long __builtin_arm_wrord (long long, long long)
7918 long long __builtin_arm_wrordi (long long, int)
7919 v4hi __builtin_arm_wrorh (v4hi, long long)
7920 v4hi __builtin_arm_wrorhi (v4hi, int)
7921 v2si __builtin_arm_wrorw (v2si, long long)
7922 v2si __builtin_arm_wrorwi (v2si, int)
7923 v2si __builtin_arm_wsadb (v8qi, v8qi)
7924 v2si __builtin_arm_wsadbz (v8qi, v8qi)
7925 v2si __builtin_arm_wsadh (v4hi, v4hi)
7926 v2si __builtin_arm_wsadhz (v4hi, v4hi)
7927 v4hi __builtin_arm_wshufh (v4hi, int)
7928 long long __builtin_arm_wslld (long long, long long)
7929 long long __builtin_arm_wslldi (long long, int)
7930 v4hi __builtin_arm_wsllh (v4hi, long long)
7931 v4hi __builtin_arm_wsllhi (v4hi, int)
7932 v2si __builtin_arm_wsllw (v2si, long long)
7933 v2si __builtin_arm_wsllwi (v2si, int)
7934 long long __builtin_arm_wsrad (long long, long long)
7935 long long __builtin_arm_wsradi (long long, int)
7936 v4hi __builtin_arm_wsrah (v4hi, long long)
7937 v4hi __builtin_arm_wsrahi (v4hi, int)
7938 v2si __builtin_arm_wsraw (v2si, long long)
7939 v2si __builtin_arm_wsrawi (v2si, int)
7940 long long __builtin_arm_wsrld (long long, long long)
7941 long long __builtin_arm_wsrldi (long long, int)
7942 v4hi __builtin_arm_wsrlh (v4hi, long long)
7943 v4hi __builtin_arm_wsrlhi (v4hi, int)
7944 v2si __builtin_arm_wsrlw (v2si, long long)
7945 v2si __builtin_arm_wsrlwi (v2si, int)
7946 v8qi __builtin_arm_wsubb (v8qi, v8qi)
7947 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
7948 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
7949 v4hi __builtin_arm_wsubh (v4hi, v4hi)
7950 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
7951 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
7952 v2si __builtin_arm_wsubw (v2si, v2si)
7953 v2si __builtin_arm_wsubwss (v2si, v2si)
7954 v2si __builtin_arm_wsubwus (v2si, v2si)
7955 v4hi __builtin_arm_wunpckehsb (v8qi)
7956 v2si __builtin_arm_wunpckehsh (v4hi)
7957 long long __builtin_arm_wunpckehsw (v2si)
7958 v4hi __builtin_arm_wunpckehub (v8qi)
7959 v2si __builtin_arm_wunpckehuh (v4hi)
7960 long long __builtin_arm_wunpckehuw (v2si)
7961 v4hi __builtin_arm_wunpckelsb (v8qi)
7962 v2si __builtin_arm_wunpckelsh (v4hi)
7963 long long __builtin_arm_wunpckelsw (v2si)
7964 v4hi __builtin_arm_wunpckelub (v8qi)
7965 v2si __builtin_arm_wunpckeluh (v4hi)
7966 long long __builtin_arm_wunpckeluw (v2si)
7967 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
7968 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
7969 v2si __builtin_arm_wunpckihw (v2si, v2si)
7970 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
7971 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
7972 v2si __builtin_arm_wunpckilw (v2si, v2si)
7973 long long __builtin_arm_wxor (long long, long long)
7974 long long __builtin_arm_wzero ()
7977 @node ARM NEON Intrinsics
7978 @subsection ARM NEON Intrinsics
7980 These built-in intrinsics for the ARM Advanced SIMD extension are available
7981 when the @option{-mfpu=neon} switch is used:
7983 @include arm-neon-intrinsics.texi
7985 @node Blackfin Built-in Functions
7986 @subsection Blackfin Built-in Functions
7988 Currently, there are two Blackfin-specific built-in functions. These are
7989 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
7990 using inline assembly; by using these built-in functions the compiler can
7991 automatically add workarounds for hardware errata involving these
7992 instructions. These functions are named as follows:
7995 void __builtin_bfin_csync (void)
7996 void __builtin_bfin_ssync (void)
7999 @node FR-V Built-in Functions
8000 @subsection FR-V Built-in Functions
8002 GCC provides many FR-V-specific built-in functions. In general,
8003 these functions are intended to be compatible with those described
8004 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
8005 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
8006 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
8007 pointer rather than by value.
8009 Most of the functions are named after specific FR-V instructions.
8010 Such functions are said to be ``directly mapped'' and are summarized
8011 here in tabular form.
8015 * Directly-mapped Integer Functions::
8016 * Directly-mapped Media Functions::
8017 * Raw read/write Functions::
8018 * Other Built-in Functions::
8021 @node Argument Types
8022 @subsubsection Argument Types
8024 The arguments to the built-in functions can be divided into three groups:
8025 register numbers, compile-time constants and run-time values. In order
8026 to make this classification clear at a glance, the arguments and return
8027 values are given the following pseudo types:
8029 @multitable @columnfractions .20 .30 .15 .35
8030 @item Pseudo type @tab Real C type @tab Constant? @tab Description
8031 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
8032 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
8033 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
8034 @item @code{uw2} @tab @code{unsigned long long} @tab No
8035 @tab an unsigned doubleword
8036 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
8037 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
8038 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
8039 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
8042 These pseudo types are not defined by GCC, they are simply a notational
8043 convenience used in this manual.
8045 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
8046 and @code{sw2} are evaluated at run time. They correspond to
8047 register operands in the underlying FR-V instructions.
8049 @code{const} arguments represent immediate operands in the underlying
8050 FR-V instructions. They must be compile-time constants.
8052 @code{acc} arguments are evaluated at compile time and specify the number
8053 of an accumulator register. For example, an @code{acc} argument of 2
8054 will select the ACC2 register.
8056 @code{iacc} arguments are similar to @code{acc} arguments but specify the
8057 number of an IACC register. See @pxref{Other Built-in Functions}
8060 @node Directly-mapped Integer Functions
8061 @subsubsection Directly-mapped Integer Functions
8063 The functions listed below map directly to FR-V I-type instructions.
8065 @multitable @columnfractions .45 .32 .23
8066 @item Function prototype @tab Example usage @tab Assembly output
8067 @item @code{sw1 __ADDSS (sw1, sw1)}
8068 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
8069 @tab @code{ADDSS @var{a},@var{b},@var{c}}
8070 @item @code{sw1 __SCAN (sw1, sw1)}
8071 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
8072 @tab @code{SCAN @var{a},@var{b},@var{c}}
8073 @item @code{sw1 __SCUTSS (sw1)}
8074 @tab @code{@var{b} = __SCUTSS (@var{a})}
8075 @tab @code{SCUTSS @var{a},@var{b}}
8076 @item @code{sw1 __SLASS (sw1, sw1)}
8077 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
8078 @tab @code{SLASS @var{a},@var{b},@var{c}}
8079 @item @code{void __SMASS (sw1, sw1)}
8080 @tab @code{__SMASS (@var{a}, @var{b})}
8081 @tab @code{SMASS @var{a},@var{b}}
8082 @item @code{void __SMSSS (sw1, sw1)}
8083 @tab @code{__SMSSS (@var{a}, @var{b})}
8084 @tab @code{SMSSS @var{a},@var{b}}
8085 @item @code{void __SMU (sw1, sw1)}
8086 @tab @code{__SMU (@var{a}, @var{b})}
8087 @tab @code{SMU @var{a},@var{b}}
8088 @item @code{sw2 __SMUL (sw1, sw1)}
8089 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
8090 @tab @code{SMUL @var{a},@var{b},@var{c}}
8091 @item @code{sw1 __SUBSS (sw1, sw1)}
8092 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
8093 @tab @code{SUBSS @var{a},@var{b},@var{c}}
8094 @item @code{uw2 __UMUL (uw1, uw1)}
8095 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
8096 @tab @code{UMUL @var{a},@var{b},@var{c}}
8099 @node Directly-mapped Media Functions
8100 @subsubsection Directly-mapped Media Functions
8102 The functions listed below map directly to FR-V M-type instructions.
8104 @multitable @columnfractions .45 .32 .23
8105 @item Function prototype @tab Example usage @tab Assembly output
8106 @item @code{uw1 __MABSHS (sw1)}
8107 @tab @code{@var{b} = __MABSHS (@var{a})}
8108 @tab @code{MABSHS @var{a},@var{b}}
8109 @item @code{void __MADDACCS (acc, acc)}
8110 @tab @code{__MADDACCS (@var{b}, @var{a})}
8111 @tab @code{MADDACCS @var{a},@var{b}}
8112 @item @code{sw1 __MADDHSS (sw1, sw1)}
8113 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
8114 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
8115 @item @code{uw1 __MADDHUS (uw1, uw1)}
8116 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
8117 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
8118 @item @code{uw1 __MAND (uw1, uw1)}
8119 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
8120 @tab @code{MAND @var{a},@var{b},@var{c}}
8121 @item @code{void __MASACCS (acc, acc)}
8122 @tab @code{__MASACCS (@var{b}, @var{a})}
8123 @tab @code{MASACCS @var{a},@var{b}}
8124 @item @code{uw1 __MAVEH (uw1, uw1)}
8125 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
8126 @tab @code{MAVEH @var{a},@var{b},@var{c}}
8127 @item @code{uw2 __MBTOH (uw1)}
8128 @tab @code{@var{b} = __MBTOH (@var{a})}
8129 @tab @code{MBTOH @var{a},@var{b}}
8130 @item @code{void __MBTOHE (uw1 *, uw1)}
8131 @tab @code{__MBTOHE (&@var{b}, @var{a})}
8132 @tab @code{MBTOHE @var{a},@var{b}}
8133 @item @code{void __MCLRACC (acc)}
8134 @tab @code{__MCLRACC (@var{a})}
8135 @tab @code{MCLRACC @var{a}}
8136 @item @code{void __MCLRACCA (void)}
8137 @tab @code{__MCLRACCA ()}
8138 @tab @code{MCLRACCA}
8139 @item @code{uw1 __Mcop1 (uw1, uw1)}
8140 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
8141 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
8142 @item @code{uw1 __Mcop2 (uw1, uw1)}
8143 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
8144 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
8145 @item @code{uw1 __MCPLHI (uw2, const)}
8146 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
8147 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
8148 @item @code{uw1 __MCPLI (uw2, const)}
8149 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
8150 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
8151 @item @code{void __MCPXIS (acc, sw1, sw1)}
8152 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
8153 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
8154 @item @code{void __MCPXIU (acc, uw1, uw1)}
8155 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
8156 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
8157 @item @code{void __MCPXRS (acc, sw1, sw1)}
8158 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
8159 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
8160 @item @code{void __MCPXRU (acc, uw1, uw1)}
8161 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
8162 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
8163 @item @code{uw1 __MCUT (acc, uw1)}
8164 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
8165 @tab @code{MCUT @var{a},@var{b},@var{c}}
8166 @item @code{uw1 __MCUTSS (acc, sw1)}
8167 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
8168 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
8169 @item @code{void __MDADDACCS (acc, acc)}
8170 @tab @code{__MDADDACCS (@var{b}, @var{a})}
8171 @tab @code{MDADDACCS @var{a},@var{b}}
8172 @item @code{void __MDASACCS (acc, acc)}
8173 @tab @code{__MDASACCS (@var{b}, @var{a})}
8174 @tab @code{MDASACCS @var{a},@var{b}}
8175 @item @code{uw2 __MDCUTSSI (acc, const)}
8176 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
8177 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
8178 @item @code{uw2 __MDPACKH (uw2, uw2)}
8179 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
8180 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
8181 @item @code{uw2 __MDROTLI (uw2, const)}
8182 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
8183 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
8184 @item @code{void __MDSUBACCS (acc, acc)}
8185 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
8186 @tab @code{MDSUBACCS @var{a},@var{b}}
8187 @item @code{void __MDUNPACKH (uw1 *, uw2)}
8188 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
8189 @tab @code{MDUNPACKH @var{a},@var{b}}
8190 @item @code{uw2 __MEXPDHD (uw1, const)}
8191 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
8192 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
8193 @item @code{uw1 __MEXPDHW (uw1, const)}
8194 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
8195 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
8196 @item @code{uw1 __MHDSETH (uw1, const)}
8197 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
8198 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
8199 @item @code{sw1 __MHDSETS (const)}
8200 @tab @code{@var{b} = __MHDSETS (@var{a})}
8201 @tab @code{MHDSETS #@var{a},@var{b}}
8202 @item @code{uw1 __MHSETHIH (uw1, const)}
8203 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
8204 @tab @code{MHSETHIH #@var{a},@var{b}}
8205 @item @code{sw1 __MHSETHIS (sw1, const)}
8206 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
8207 @tab @code{MHSETHIS #@var{a},@var{b}}
8208 @item @code{uw1 __MHSETLOH (uw1, const)}
8209 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
8210 @tab @code{MHSETLOH #@var{a},@var{b}}
8211 @item @code{sw1 __MHSETLOS (sw1, const)}
8212 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
8213 @tab @code{MHSETLOS #@var{a},@var{b}}
8214 @item @code{uw1 __MHTOB (uw2)}
8215 @tab @code{@var{b} = __MHTOB (@var{a})}
8216 @tab @code{MHTOB @var{a},@var{b}}
8217 @item @code{void __MMACHS (acc, sw1, sw1)}
8218 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
8219 @tab @code{MMACHS @var{a},@var{b},@var{c}}
8220 @item @code{void __MMACHU (acc, uw1, uw1)}
8221 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
8222 @tab @code{MMACHU @var{a},@var{b},@var{c}}
8223 @item @code{void __MMRDHS (acc, sw1, sw1)}
8224 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
8225 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
8226 @item @code{void __MMRDHU (acc, uw1, uw1)}
8227 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
8228 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
8229 @item @code{void __MMULHS (acc, sw1, sw1)}
8230 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
8231 @tab @code{MMULHS @var{a},@var{b},@var{c}}
8232 @item @code{void __MMULHU (acc, uw1, uw1)}
8233 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
8234 @tab @code{MMULHU @var{a},@var{b},@var{c}}
8235 @item @code{void __MMULXHS (acc, sw1, sw1)}
8236 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
8237 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
8238 @item @code{void __MMULXHU (acc, uw1, uw1)}
8239 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
8240 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
8241 @item @code{uw1 __MNOT (uw1)}
8242 @tab @code{@var{b} = __MNOT (@var{a})}
8243 @tab @code{MNOT @var{a},@var{b}}
8244 @item @code{uw1 __MOR (uw1, uw1)}
8245 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
8246 @tab @code{MOR @var{a},@var{b},@var{c}}
8247 @item @code{uw1 __MPACKH (uh, uh)}
8248 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
8249 @tab @code{MPACKH @var{a},@var{b},@var{c}}
8250 @item @code{sw2 __MQADDHSS (sw2, sw2)}
8251 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
8252 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
8253 @item @code{uw2 __MQADDHUS (uw2, uw2)}
8254 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
8255 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
8256 @item @code{void __MQCPXIS (acc, sw2, sw2)}
8257 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
8258 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
8259 @item @code{void __MQCPXIU (acc, uw2, uw2)}
8260 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
8261 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
8262 @item @code{void __MQCPXRS (acc, sw2, sw2)}
8263 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
8264 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
8265 @item @code{void __MQCPXRU (acc, uw2, uw2)}
8266 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
8267 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
8268 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
8269 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
8270 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
8271 @item @code{sw2 __MQLMTHS (sw2, sw2)}
8272 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
8273 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
8274 @item @code{void __MQMACHS (acc, sw2, sw2)}
8275 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
8276 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
8277 @item @code{void __MQMACHU (acc, uw2, uw2)}
8278 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
8279 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
8280 @item @code{void __MQMACXHS (acc, sw2, sw2)}
8281 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
8282 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
8283 @item @code{void __MQMULHS (acc, sw2, sw2)}
8284 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
8285 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
8286 @item @code{void __MQMULHU (acc, uw2, uw2)}
8287 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
8288 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
8289 @item @code{void __MQMULXHS (acc, sw2, sw2)}
8290 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
8291 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
8292 @item @code{void __MQMULXHU (acc, uw2, uw2)}
8293 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
8294 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
8295 @item @code{sw2 __MQSATHS (sw2, sw2)}
8296 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
8297 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
8298 @item @code{uw2 __MQSLLHI (uw2, int)}
8299 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
8300 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
8301 @item @code{sw2 __MQSRAHI (sw2, int)}
8302 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
8303 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
8304 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
8305 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
8306 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
8307 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
8308 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
8309 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
8310 @item @code{void __MQXMACHS (acc, sw2, sw2)}
8311 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
8312 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
8313 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
8314 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
8315 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
8316 @item @code{uw1 __MRDACC (acc)}
8317 @tab @code{@var{b} = __MRDACC (@var{a})}
8318 @tab @code{MRDACC @var{a},@var{b}}
8319 @item @code{uw1 __MRDACCG (acc)}
8320 @tab @code{@var{b} = __MRDACCG (@var{a})}
8321 @tab @code{MRDACCG @var{a},@var{b}}
8322 @item @code{uw1 __MROTLI (uw1, const)}
8323 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
8324 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
8325 @item @code{uw1 __MROTRI (uw1, const)}
8326 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
8327 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
8328 @item @code{sw1 __MSATHS (sw1, sw1)}
8329 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
8330 @tab @code{MSATHS @var{a},@var{b},@var{c}}
8331 @item @code{uw1 __MSATHU (uw1, uw1)}
8332 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
8333 @tab @code{MSATHU @var{a},@var{b},@var{c}}
8334 @item @code{uw1 __MSLLHI (uw1, const)}
8335 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
8336 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
8337 @item @code{sw1 __MSRAHI (sw1, const)}
8338 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
8339 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
8340 @item @code{uw1 __MSRLHI (uw1, const)}
8341 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
8342 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
8343 @item @code{void __MSUBACCS (acc, acc)}
8344 @tab @code{__MSUBACCS (@var{b}, @var{a})}
8345 @tab @code{MSUBACCS @var{a},@var{b}}
8346 @item @code{sw1 __MSUBHSS (sw1, sw1)}
8347 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
8348 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
8349 @item @code{uw1 __MSUBHUS (uw1, uw1)}
8350 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
8351 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
8352 @item @code{void __MTRAP (void)}
8353 @tab @code{__MTRAP ()}
8355 @item @code{uw2 __MUNPACKH (uw1)}
8356 @tab @code{@var{b} = __MUNPACKH (@var{a})}
8357 @tab @code{MUNPACKH @var{a},@var{b}}
8358 @item @code{uw1 __MWCUT (uw2, uw1)}
8359 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
8360 @tab @code{MWCUT @var{a},@var{b},@var{c}}
8361 @item @code{void __MWTACC (acc, uw1)}
8362 @tab @code{__MWTACC (@var{b}, @var{a})}
8363 @tab @code{MWTACC @var{a},@var{b}}
8364 @item @code{void __MWTACCG (acc, uw1)}
8365 @tab @code{__MWTACCG (@var{b}, @var{a})}
8366 @tab @code{MWTACCG @var{a},@var{b}}
8367 @item @code{uw1 __MXOR (uw1, uw1)}
8368 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
8369 @tab @code{MXOR @var{a},@var{b},@var{c}}
8372 @node Raw read/write Functions
8373 @subsubsection Raw read/write Functions
8375 This sections describes built-in functions related to read and write
8376 instructions to access memory. These functions generate
8377 @code{membar} instructions to flush the I/O load and stores where
8378 appropriate, as described in Fujitsu's manual described above.
8382 @item unsigned char __builtin_read8 (void *@var{data})
8383 @item unsigned short __builtin_read16 (void *@var{data})
8384 @item unsigned long __builtin_read32 (void *@var{data})
8385 @item unsigned long long __builtin_read64 (void *@var{data})
8387 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
8388 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
8389 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
8390 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
8393 @node Other Built-in Functions
8394 @subsubsection Other Built-in Functions
8396 This section describes built-in functions that are not named after
8397 a specific FR-V instruction.
8400 @item sw2 __IACCreadll (iacc @var{reg})
8401 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
8402 for future expansion and must be 0.
8404 @item sw1 __IACCreadl (iacc @var{reg})
8405 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
8406 Other values of @var{reg} are rejected as invalid.
8408 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
8409 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
8410 is reserved for future expansion and must be 0.
8412 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
8413 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
8414 is 1. Other values of @var{reg} are rejected as invalid.
8416 @item void __data_prefetch0 (const void *@var{x})
8417 Use the @code{dcpl} instruction to load the contents of address @var{x}
8418 into the data cache.
8420 @item void __data_prefetch (const void *@var{x})
8421 Use the @code{nldub} instruction to load the contents of address @var{x}
8422 into the data cache. The instruction will be issued in slot I1@.
8425 @node X86 Built-in Functions
8426 @subsection X86 Built-in Functions
8428 These built-in functions are available for the i386 and x86-64 family
8429 of computers, depending on the command-line switches used.
8431 Note that, if you specify command-line switches such as @option{-msse},
8432 the compiler could use the extended instruction sets even if the built-ins
8433 are not used explicitly in the program. For this reason, applications
8434 which perform runtime CPU detection must compile separate files for each
8435 supported architecture, using the appropriate flags. In particular,
8436 the file containing the CPU detection code should be compiled without
8439 The following machine modes are available for use with MMX built-in functions
8440 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
8441 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
8442 vector of eight 8-bit integers. Some of the built-in functions operate on
8443 MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode.
8445 If 3DNow!@: extensions are enabled, @code{V2SF} is used as a mode for a vector
8446 of two 32-bit floating point values.
8448 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
8449 floating point values. Some instructions use a vector of four 32-bit
8450 integers, these use @code{V4SI}. Finally, some instructions operate on an
8451 entire vector register, interpreting it as a 128-bit integer, these use mode
8454 In 64-bit mode, the x86-64 family of processors uses additional built-in
8455 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
8456 floating point and @code{TC} 128-bit complex floating point values.
8458 The following floating point built-in functions are available in 64-bit
8459 mode. All of them implement the function that is part of the name.
8462 __float128 __builtin_fabsq (__float128)
8463 __float128 __builtin_copysignq (__float128, __float128)
8466 The following floating point built-in functions are made available in the
8470 @item __float128 __builtin_infq (void)
8471 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
8472 @findex __builtin_infq
8474 @item __float128 __builtin_huge_valq (void)
8475 Similar to @code{__builtin_huge_val}, except the return type is @code{__float128}.
8476 @findex __builtin_huge_valq
8479 The following built-in functions are made available by @option{-mmmx}.
8480 All of them generate the machine instruction that is part of the name.
8483 v8qi __builtin_ia32_paddb (v8qi, v8qi)
8484 v4hi __builtin_ia32_paddw (v4hi, v4hi)
8485 v2si __builtin_ia32_paddd (v2si, v2si)
8486 v8qi __builtin_ia32_psubb (v8qi, v8qi)
8487 v4hi __builtin_ia32_psubw (v4hi, v4hi)
8488 v2si __builtin_ia32_psubd (v2si, v2si)
8489 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
8490 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
8491 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
8492 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
8493 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
8494 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
8495 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
8496 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
8497 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
8498 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
8499 di __builtin_ia32_pand (di, di)
8500 di __builtin_ia32_pandn (di,di)
8501 di __builtin_ia32_por (di, di)
8502 di __builtin_ia32_pxor (di, di)
8503 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
8504 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
8505 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
8506 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
8507 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
8508 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
8509 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
8510 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
8511 v2si __builtin_ia32_punpckhdq (v2si, v2si)
8512 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
8513 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
8514 v2si __builtin_ia32_punpckldq (v2si, v2si)
8515 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
8516 v4hi __builtin_ia32_packssdw (v2si, v2si)
8517 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
8519 v4hi __builtin_ia32_psllw (v4hi, v4hi)
8520 v2si __builtin_ia32_pslld (v2si, v2si)
8521 v1di __builtin_ia32_psllq (v1di, v1di)
8522 v4hi __builtin_ia32_psrlw (v4hi, v4hi)
8523 v2si __builtin_ia32_psrld (v2si, v2si)
8524 v1di __builtin_ia32_psrlq (v1di, v1di)
8525 v4hi __builtin_ia32_psraw (v4hi, v4hi)
8526 v2si __builtin_ia32_psrad (v2si, v2si)
8527 v4hi __builtin_ia32_psllwi (v4hi, int)
8528 v2si __builtin_ia32_pslldi (v2si, int)
8529 v1di __builtin_ia32_psllqi (v1di, int)
8530 v4hi __builtin_ia32_psrlwi (v4hi, int)
8531 v2si __builtin_ia32_psrldi (v2si, int)
8532 v1di __builtin_ia32_psrlqi (v1di, int)
8533 v4hi __builtin_ia32_psrawi (v4hi, int)
8534 v2si __builtin_ia32_psradi (v2si, int)
8538 The following built-in functions are made available either with
8539 @option{-msse}, or with a combination of @option{-m3dnow} and
8540 @option{-march=athlon}. All of them generate the machine
8541 instruction that is part of the name.
8544 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
8545 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
8546 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
8547 v1di __builtin_ia32_psadbw (v8qi, v8qi)
8548 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
8549 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
8550 v8qi __builtin_ia32_pminub (v8qi, v8qi)
8551 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
8552 int __builtin_ia32_pextrw (v4hi, int)
8553 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
8554 int __builtin_ia32_pmovmskb (v8qi)
8555 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
8556 void __builtin_ia32_movntq (di *, di)
8557 void __builtin_ia32_sfence (void)
8560 The following built-in functions are available when @option{-msse} is used.
8561 All of them generate the machine instruction that is part of the name.
8564 int __builtin_ia32_comieq (v4sf, v4sf)
8565 int __builtin_ia32_comineq (v4sf, v4sf)
8566 int __builtin_ia32_comilt (v4sf, v4sf)
8567 int __builtin_ia32_comile (v4sf, v4sf)
8568 int __builtin_ia32_comigt (v4sf, v4sf)
8569 int __builtin_ia32_comige (v4sf, v4sf)
8570 int __builtin_ia32_ucomieq (v4sf, v4sf)
8571 int __builtin_ia32_ucomineq (v4sf, v4sf)
8572 int __builtin_ia32_ucomilt (v4sf, v4sf)
8573 int __builtin_ia32_ucomile (v4sf, v4sf)
8574 int __builtin_ia32_ucomigt (v4sf, v4sf)
8575 int __builtin_ia32_ucomige (v4sf, v4sf)
8576 v4sf __builtin_ia32_addps (v4sf, v4sf)
8577 v4sf __builtin_ia32_subps (v4sf, v4sf)
8578 v4sf __builtin_ia32_mulps (v4sf, v4sf)
8579 v4sf __builtin_ia32_divps (v4sf, v4sf)
8580 v4sf __builtin_ia32_addss (v4sf, v4sf)
8581 v4sf __builtin_ia32_subss (v4sf, v4sf)
8582 v4sf __builtin_ia32_mulss (v4sf, v4sf)
8583 v4sf __builtin_ia32_divss (v4sf, v4sf)
8584 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
8585 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
8586 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
8587 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
8588 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
8589 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
8590 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
8591 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
8592 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
8593 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
8594 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
8595 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
8596 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
8597 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
8598 v4si __builtin_ia32_cmpless (v4sf, v4sf)
8599 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
8600 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
8601 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
8602 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
8603 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
8604 v4sf __builtin_ia32_maxps (v4sf, v4sf)
8605 v4sf __builtin_ia32_maxss (v4sf, v4sf)
8606 v4sf __builtin_ia32_minps (v4sf, v4sf)
8607 v4sf __builtin_ia32_minss (v4sf, v4sf)
8608 v4sf __builtin_ia32_andps (v4sf, v4sf)
8609 v4sf __builtin_ia32_andnps (v4sf, v4sf)
8610 v4sf __builtin_ia32_orps (v4sf, v4sf)
8611 v4sf __builtin_ia32_xorps (v4sf, v4sf)
8612 v4sf __builtin_ia32_movss (v4sf, v4sf)
8613 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
8614 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
8615 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
8616 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
8617 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
8618 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
8619 v2si __builtin_ia32_cvtps2pi (v4sf)
8620 int __builtin_ia32_cvtss2si (v4sf)
8621 v2si __builtin_ia32_cvttps2pi (v4sf)
8622 int __builtin_ia32_cvttss2si (v4sf)
8623 v4sf __builtin_ia32_rcpps (v4sf)
8624 v4sf __builtin_ia32_rsqrtps (v4sf)
8625 v4sf __builtin_ia32_sqrtps (v4sf)
8626 v4sf __builtin_ia32_rcpss (v4sf)
8627 v4sf __builtin_ia32_rsqrtss (v4sf)
8628 v4sf __builtin_ia32_sqrtss (v4sf)
8629 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
8630 void __builtin_ia32_movntps (float *, v4sf)
8631 int __builtin_ia32_movmskps (v4sf)
8634 The following built-in functions are available when @option{-msse} is used.
8637 @item v4sf __builtin_ia32_loadaps (float *)
8638 Generates the @code{movaps} machine instruction as a load from memory.
8639 @item void __builtin_ia32_storeaps (float *, v4sf)
8640 Generates the @code{movaps} machine instruction as a store to memory.
8641 @item v4sf __builtin_ia32_loadups (float *)
8642 Generates the @code{movups} machine instruction as a load from memory.
8643 @item void __builtin_ia32_storeups (float *, v4sf)
8644 Generates the @code{movups} machine instruction as a store to memory.
8645 @item v4sf __builtin_ia32_loadsss (float *)
8646 Generates the @code{movss} machine instruction as a load from memory.
8647 @item void __builtin_ia32_storess (float *, v4sf)
8648 Generates the @code{movss} machine instruction as a store to memory.
8649 @item v4sf __builtin_ia32_loadhps (v4sf, const v2sf *)
8650 Generates the @code{movhps} machine instruction as a load from memory.
8651 @item v4sf __builtin_ia32_loadlps (v4sf, const v2sf *)
8652 Generates the @code{movlps} machine instruction as a load from memory
8653 @item void __builtin_ia32_storehps (v2sf *, v4sf)
8654 Generates the @code{movhps} machine instruction as a store to memory.
8655 @item void __builtin_ia32_storelps (v2sf *, v4sf)
8656 Generates the @code{movlps} machine instruction as a store to memory.
8659 The following built-in functions are available when @option{-msse2} is used.
8660 All of them generate the machine instruction that is part of the name.
8663 int __builtin_ia32_comisdeq (v2df, v2df)
8664 int __builtin_ia32_comisdlt (v2df, v2df)
8665 int __builtin_ia32_comisdle (v2df, v2df)
8666 int __builtin_ia32_comisdgt (v2df, v2df)
8667 int __builtin_ia32_comisdge (v2df, v2df)
8668 int __builtin_ia32_comisdneq (v2df, v2df)
8669 int __builtin_ia32_ucomisdeq (v2df, v2df)
8670 int __builtin_ia32_ucomisdlt (v2df, v2df)
8671 int __builtin_ia32_ucomisdle (v2df, v2df)
8672 int __builtin_ia32_ucomisdgt (v2df, v2df)
8673 int __builtin_ia32_ucomisdge (v2df, v2df)
8674 int __builtin_ia32_ucomisdneq (v2df, v2df)
8675 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
8676 v2df __builtin_ia32_cmpltpd (v2df, v2df)
8677 v2df __builtin_ia32_cmplepd (v2df, v2df)
8678 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
8679 v2df __builtin_ia32_cmpgepd (v2df, v2df)
8680 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
8681 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
8682 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
8683 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
8684 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
8685 v2df __builtin_ia32_cmpngepd (v2df, v2df)
8686 v2df __builtin_ia32_cmpordpd (v2df, v2df)
8687 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
8688 v2df __builtin_ia32_cmpltsd (v2df, v2df)
8689 v2df __builtin_ia32_cmplesd (v2df, v2df)
8690 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
8691 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
8692 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
8693 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
8694 v2df __builtin_ia32_cmpordsd (v2df, v2df)
8695 v2di __builtin_ia32_paddq (v2di, v2di)
8696 v2di __builtin_ia32_psubq (v2di, v2di)
8697 v2df __builtin_ia32_addpd (v2df, v2df)
8698 v2df __builtin_ia32_subpd (v2df, v2df)
8699 v2df __builtin_ia32_mulpd (v2df, v2df)
8700 v2df __builtin_ia32_divpd (v2df, v2df)
8701 v2df __builtin_ia32_addsd (v2df, v2df)
8702 v2df __builtin_ia32_subsd (v2df, v2df)
8703 v2df __builtin_ia32_mulsd (v2df, v2df)
8704 v2df __builtin_ia32_divsd (v2df, v2df)
8705 v2df __builtin_ia32_minpd (v2df, v2df)
8706 v2df __builtin_ia32_maxpd (v2df, v2df)
8707 v2df __builtin_ia32_minsd (v2df, v2df)
8708 v2df __builtin_ia32_maxsd (v2df, v2df)
8709 v2df __builtin_ia32_andpd (v2df, v2df)
8710 v2df __builtin_ia32_andnpd (v2df, v2df)
8711 v2df __builtin_ia32_orpd (v2df, v2df)
8712 v2df __builtin_ia32_xorpd (v2df, v2df)
8713 v2df __builtin_ia32_movsd (v2df, v2df)
8714 v2df __builtin_ia32_unpckhpd (v2df, v2df)
8715 v2df __builtin_ia32_unpcklpd (v2df, v2df)
8716 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
8717 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
8718 v4si __builtin_ia32_paddd128 (v4si, v4si)
8719 v2di __builtin_ia32_paddq128 (v2di, v2di)
8720 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
8721 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
8722 v4si __builtin_ia32_psubd128 (v4si, v4si)
8723 v2di __builtin_ia32_psubq128 (v2di, v2di)
8724 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
8725 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
8726 v2di __builtin_ia32_pand128 (v2di, v2di)
8727 v2di __builtin_ia32_pandn128 (v2di, v2di)
8728 v2di __builtin_ia32_por128 (v2di, v2di)
8729 v2di __builtin_ia32_pxor128 (v2di, v2di)
8730 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
8731 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
8732 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
8733 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
8734 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
8735 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
8736 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
8737 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
8738 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
8739 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
8740 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
8741 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
8742 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
8743 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
8744 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
8745 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
8746 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
8747 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
8748 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
8749 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
8750 v16qi __builtin_ia32_packsswb128 (v8hi, v8hi)
8751 v8hi __builtin_ia32_packssdw128 (v4si, v4si)
8752 v16qi __builtin_ia32_packuswb128 (v8hi, v8hi)
8753 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
8754 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
8755 v2df __builtin_ia32_loadupd (double *)
8756 void __builtin_ia32_storeupd (double *, v2df)
8757 v2df __builtin_ia32_loadhpd (v2df, double const *)
8758 v2df __builtin_ia32_loadlpd (v2df, double const *)
8759 int __builtin_ia32_movmskpd (v2df)
8760 int __builtin_ia32_pmovmskb128 (v16qi)
8761 void __builtin_ia32_movnti (int *, int)
8762 void __builtin_ia32_movntpd (double *, v2df)
8763 void __builtin_ia32_movntdq (v2df *, v2df)
8764 v4si __builtin_ia32_pshufd (v4si, int)
8765 v8hi __builtin_ia32_pshuflw (v8hi, int)
8766 v8hi __builtin_ia32_pshufhw (v8hi, int)
8767 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
8768 v2df __builtin_ia32_sqrtpd (v2df)
8769 v2df __builtin_ia32_sqrtsd (v2df)
8770 v2df __builtin_ia32_shufpd (v2df, v2df, int)
8771 v2df __builtin_ia32_cvtdq2pd (v4si)
8772 v4sf __builtin_ia32_cvtdq2ps (v4si)
8773 v4si __builtin_ia32_cvtpd2dq (v2df)
8774 v2si __builtin_ia32_cvtpd2pi (v2df)
8775 v4sf __builtin_ia32_cvtpd2ps (v2df)
8776 v4si __builtin_ia32_cvttpd2dq (v2df)
8777 v2si __builtin_ia32_cvttpd2pi (v2df)
8778 v2df __builtin_ia32_cvtpi2pd (v2si)
8779 int __builtin_ia32_cvtsd2si (v2df)
8780 int __builtin_ia32_cvttsd2si (v2df)
8781 long long __builtin_ia32_cvtsd2si64 (v2df)
8782 long long __builtin_ia32_cvttsd2si64 (v2df)
8783 v4si __builtin_ia32_cvtps2dq (v4sf)
8784 v2df __builtin_ia32_cvtps2pd (v4sf)
8785 v4si __builtin_ia32_cvttps2dq (v4sf)
8786 v2df __builtin_ia32_cvtsi2sd (v2df, int)
8787 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
8788 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
8789 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
8790 void __builtin_ia32_clflush (const void *)
8791 void __builtin_ia32_lfence (void)
8792 void __builtin_ia32_mfence (void)
8793 v16qi __builtin_ia32_loaddqu (const char *)
8794 void __builtin_ia32_storedqu (char *, v16qi)
8795 v1di __builtin_ia32_pmuludq (v2si, v2si)
8796 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
8797 v8hi __builtin_ia32_psllw128 (v8hi, v8hi)
8798 v4si __builtin_ia32_pslld128 (v4si, v4si)
8799 v2di __builtin_ia32_psllq128 (v2di, v2di)
8800 v8hi __builtin_ia32_psrlw128 (v8hi, v8hi)
8801 v4si __builtin_ia32_psrld128 (v4si, v4si)
8802 v2di __builtin_ia32_psrlq128 (v2di, v2di)
8803 v8hi __builtin_ia32_psraw128 (v8hi, v8hi)
8804 v4si __builtin_ia32_psrad128 (v4si, v4si)
8805 v2di __builtin_ia32_pslldqi128 (v2di, int)
8806 v8hi __builtin_ia32_psllwi128 (v8hi, int)
8807 v4si __builtin_ia32_pslldi128 (v4si, int)
8808 v2di __builtin_ia32_psllqi128 (v2di, int)
8809 v2di __builtin_ia32_psrldqi128 (v2di, int)
8810 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
8811 v4si __builtin_ia32_psrldi128 (v4si, int)
8812 v2di __builtin_ia32_psrlqi128 (v2di, int)
8813 v8hi __builtin_ia32_psrawi128 (v8hi, int)
8814 v4si __builtin_ia32_psradi128 (v4si, int)
8815 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
8816 v2di __builtin_ia32_movq128 (v2di)
8819 The following built-in functions are available when @option{-msse3} is used.
8820 All of them generate the machine instruction that is part of the name.
8823 v2df __builtin_ia32_addsubpd (v2df, v2df)
8824 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
8825 v2df __builtin_ia32_haddpd (v2df, v2df)
8826 v4sf __builtin_ia32_haddps (v4sf, v4sf)
8827 v2df __builtin_ia32_hsubpd (v2df, v2df)
8828 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
8829 v16qi __builtin_ia32_lddqu (char const *)
8830 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
8831 v2df __builtin_ia32_movddup (v2df)
8832 v4sf __builtin_ia32_movshdup (v4sf)
8833 v4sf __builtin_ia32_movsldup (v4sf)
8834 void __builtin_ia32_mwait (unsigned int, unsigned int)
8837 The following built-in functions are available when @option{-msse3} is used.
8840 @item v2df __builtin_ia32_loadddup (double const *)
8841 Generates the @code{movddup} machine instruction as a load from memory.
8844 The following built-in functions are available when @option{-mssse3} is used.
8845 All of them generate the machine instruction that is part of the name
8849 v2si __builtin_ia32_phaddd (v2si, v2si)
8850 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
8851 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
8852 v2si __builtin_ia32_phsubd (v2si, v2si)
8853 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
8854 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
8855 v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi)
8856 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
8857 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
8858 v8qi __builtin_ia32_psignb (v8qi, v8qi)
8859 v2si __builtin_ia32_psignd (v2si, v2si)
8860 v4hi __builtin_ia32_psignw (v4hi, v4hi)
8861 v1di __builtin_ia32_palignr (v1di, v1di, int)
8862 v8qi __builtin_ia32_pabsb (v8qi)
8863 v2si __builtin_ia32_pabsd (v2si)
8864 v4hi __builtin_ia32_pabsw (v4hi)
8867 The following built-in functions are available when @option{-mssse3} is used.
8868 All of them generate the machine instruction that is part of the name
8872 v4si __builtin_ia32_phaddd128 (v4si, v4si)
8873 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
8874 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
8875 v4si __builtin_ia32_phsubd128 (v4si, v4si)
8876 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
8877 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
8878 v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
8879 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
8880 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
8881 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
8882 v4si __builtin_ia32_psignd128 (v4si, v4si)
8883 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
8884 v2di __builtin_ia32_palignr128 (v2di, v2di, int)
8885 v16qi __builtin_ia32_pabsb128 (v16qi)
8886 v4si __builtin_ia32_pabsd128 (v4si)
8887 v8hi __builtin_ia32_pabsw128 (v8hi)
8890 The following built-in functions are available when @option{-msse4.1} is
8891 used. All of them generate the machine instruction that is part of the
8895 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
8896 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
8897 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
8898 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
8899 v2df __builtin_ia32_dppd (v2df, v2df, const int)
8900 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
8901 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
8902 v2di __builtin_ia32_movntdqa (v2di *);
8903 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
8904 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
8905 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
8906 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
8907 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
8908 v8hi __builtin_ia32_phminposuw128 (v8hi)
8909 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
8910 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
8911 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
8912 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
8913 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
8914 v4si __builtin_ia32_pminsd128 (v4si, v4si)
8915 v4si __builtin_ia32_pminud128 (v4si, v4si)
8916 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
8917 v4si __builtin_ia32_pmovsxbd128 (v16qi)
8918 v2di __builtin_ia32_pmovsxbq128 (v16qi)
8919 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
8920 v2di __builtin_ia32_pmovsxdq128 (v4si)
8921 v4si __builtin_ia32_pmovsxwd128 (v8hi)
8922 v2di __builtin_ia32_pmovsxwq128 (v8hi)
8923 v4si __builtin_ia32_pmovzxbd128 (v16qi)
8924 v2di __builtin_ia32_pmovzxbq128 (v16qi)
8925 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
8926 v2di __builtin_ia32_pmovzxdq128 (v4si)
8927 v4si __builtin_ia32_pmovzxwd128 (v8hi)
8928 v2di __builtin_ia32_pmovzxwq128 (v8hi)
8929 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
8930 v4si __builtin_ia32_pmulld128 (v4si, v4si)
8931 int __builtin_ia32_ptestc128 (v2di, v2di)
8932 int __builtin_ia32_ptestnzc128 (v2di, v2di)
8933 int __builtin_ia32_ptestz128 (v2di, v2di)
8934 v2df __builtin_ia32_roundpd (v2df, const int)
8935 v4sf __builtin_ia32_roundps (v4sf, const int)
8936 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
8937 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
8940 The following built-in functions are available when @option{-msse4.1} is
8944 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
8945 Generates the @code{insertps} machine instruction.
8946 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
8947 Generates the @code{pextrb} machine instruction.
8948 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
8949 Generates the @code{pinsrb} machine instruction.
8950 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
8951 Generates the @code{pinsrd} machine instruction.
8952 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
8953 Generates the @code{pinsrq} machine instruction in 64bit mode.
8956 The following built-in functions are changed to generate new SSE4.1
8957 instructions when @option{-msse4.1} is used.
8960 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
8961 Generates the @code{extractps} machine instruction.
8962 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
8963 Generates the @code{pextrd} machine instruction.
8964 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
8965 Generates the @code{pextrq} machine instruction in 64bit mode.
8968 The following built-in functions are available when @option{-msse4.2} is
8969 used. All of them generate the machine instruction that is part of the
8973 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
8974 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
8975 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
8976 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
8977 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
8978 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
8979 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
8980 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
8981 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
8982 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
8983 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
8984 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
8985 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
8986 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
8987 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
8990 The following built-in functions are available when @option{-msse4.2} is
8994 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
8995 Generates the @code{crc32b} machine instruction.
8996 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
8997 Generates the @code{crc32w} machine instruction.
8998 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
8999 Generates the @code{crc32l} machine instruction.
9000 @item unsigned long long __builtin_ia32_crc32di (unsigned long long, unsigned long long)
9001 Generates the @code{crc32q} machine instruction.
9004 The following built-in functions are changed to generate new SSE4.2
9005 instructions when @option{-msse4.2} is used.
9008 @item int __builtin_popcount (unsigned int)
9009 Generates the @code{popcntl} machine instruction.
9010 @item int __builtin_popcountl (unsigned long)
9011 Generates the @code{popcntl} or @code{popcntq} machine instruction,
9012 depending on the size of @code{unsigned long}.
9013 @item int __builtin_popcountll (unsigned long long)
9014 Generates the @code{popcntq} machine instruction.
9017 The following built-in functions are available when @option{-mavx} is
9018 used. All of them generate the machine instruction that is part of the
9022 v4df __builtin_ia32_addpd256 (v4df,v4df)
9023 v8sf __builtin_ia32_addps256 (v8sf,v8sf)
9024 v4df __builtin_ia32_addsubpd256 (v4df,v4df)
9025 v8sf __builtin_ia32_addsubps256 (v8sf,v8sf)
9026 v4df __builtin_ia32_andnpd256 (v4df,v4df)
9027 v8sf __builtin_ia32_andnps256 (v8sf,v8sf)
9028 v4df __builtin_ia32_andpd256 (v4df,v4df)
9029 v8sf __builtin_ia32_andps256 (v8sf,v8sf)
9030 v4df __builtin_ia32_blendpd256 (v4df,v4df,int)
9031 v8sf __builtin_ia32_blendps256 (v8sf,v8sf,int)
9032 v4df __builtin_ia32_blendvpd256 (v4df,v4df,v4df)
9033 v8sf __builtin_ia32_blendvps256 (v8sf,v8sf,v8sf)
9034 v2df __builtin_ia32_cmppd (v2df,v2df,int)
9035 v4df __builtin_ia32_cmppd256 (v4df,v4df,int)
9036 v4sf __builtin_ia32_cmpps (v4sf,v4sf,int)
9037 v8sf __builtin_ia32_cmpps256 (v8sf,v8sf,int)
9038 v2df __builtin_ia32_cmpsd (v2df,v2df,int)
9039 v4sf __builtin_ia32_cmpss (v4sf,v4sf,int)
9040 v4df __builtin_ia32_cvtdq2pd256 (v4si)
9041 v8sf __builtin_ia32_cvtdq2ps256 (v8si)
9042 v4si __builtin_ia32_cvtpd2dq256 (v4df)
9043 v4sf __builtin_ia32_cvtpd2ps256 (v4df)
9044 v8si __builtin_ia32_cvtps2dq256 (v8sf)
9045 v4df __builtin_ia32_cvtps2pd256 (v4sf)
9046 v4si __builtin_ia32_cvttpd2dq256 (v4df)
9047 v8si __builtin_ia32_cvttps2dq256 (v8sf)
9048 v4df __builtin_ia32_divpd256 (v4df,v4df)
9049 v8sf __builtin_ia32_divps256 (v8sf,v8sf)
9050 v8sf __builtin_ia32_dpps256 (v8sf,v8sf,int)
9051 v4df __builtin_ia32_haddpd256 (v4df,v4df)
9052 v8sf __builtin_ia32_haddps256 (v8sf,v8sf)
9053 v4df __builtin_ia32_hsubpd256 (v4df,v4df)
9054 v8sf __builtin_ia32_hsubps256 (v8sf,v8sf)
9055 v32qi __builtin_ia32_lddqu256 (pcchar)
9056 v32qi __builtin_ia32_loaddqu256 (pcchar)
9057 v4df __builtin_ia32_loadupd256 (pcdouble)
9058 v8sf __builtin_ia32_loadups256 (pcfloat)
9059 v2df __builtin_ia32_maskloadpd (pcv2df,v2df)
9060 v4df __builtin_ia32_maskloadpd256 (pcv4df,v4df)
9061 v4sf __builtin_ia32_maskloadps (pcv4sf,v4sf)
9062 v8sf __builtin_ia32_maskloadps256 (pcv8sf,v8sf)
9063 void __builtin_ia32_maskstorepd (pv2df,v2df,v2df)
9064 void __builtin_ia32_maskstorepd256 (pv4df,v4df,v4df)
9065 void __builtin_ia32_maskstoreps (pv4sf,v4sf,v4sf)
9066 void __builtin_ia32_maskstoreps256 (pv8sf,v8sf,v8sf)
9067 v4df __builtin_ia32_maxpd256 (v4df,v4df)
9068 v8sf __builtin_ia32_maxps256 (v8sf,v8sf)
9069 v4df __builtin_ia32_minpd256 (v4df,v4df)
9070 v8sf __builtin_ia32_minps256 (v8sf,v8sf)
9071 v4df __builtin_ia32_movddup256 (v4df)
9072 int __builtin_ia32_movmskpd256 (v4df)
9073 int __builtin_ia32_movmskps256 (v8sf)
9074 v8sf __builtin_ia32_movshdup256 (v8sf)
9075 v8sf __builtin_ia32_movsldup256 (v8sf)
9076 v4df __builtin_ia32_mulpd256 (v4df,v4df)
9077 v8sf __builtin_ia32_mulps256 (v8sf,v8sf)
9078 v4df __builtin_ia32_orpd256 (v4df,v4df)
9079 v8sf __builtin_ia32_orps256 (v8sf,v8sf)
9080 v2df __builtin_ia32_pd_pd256 (v4df)
9081 v4df __builtin_ia32_pd256_pd (v2df)
9082 v4sf __builtin_ia32_ps_ps256 (v8sf)
9083 v8sf __builtin_ia32_ps256_ps (v4sf)
9084 int __builtin_ia32_ptestc256 (v4di,v4di,ptest)
9085 int __builtin_ia32_ptestnzc256 (v4di,v4di,ptest)
9086 int __builtin_ia32_ptestz256 (v4di,v4di,ptest)
9087 v8sf __builtin_ia32_rcpps256 (v8sf)
9088 v4df __builtin_ia32_roundpd256 (v4df,int)
9089 v8sf __builtin_ia32_roundps256 (v8sf,int)
9090 v8sf __builtin_ia32_rsqrtps_nr256 (v8sf)
9091 v8sf __builtin_ia32_rsqrtps256 (v8sf)
9092 v4df __builtin_ia32_shufpd256 (v4df,v4df,int)
9093 v8sf __builtin_ia32_shufps256 (v8sf,v8sf,int)
9094 v4si __builtin_ia32_si_si256 (v8si)
9095 v8si __builtin_ia32_si256_si (v4si)
9096 v4df __builtin_ia32_sqrtpd256 (v4df)
9097 v8sf __builtin_ia32_sqrtps_nr256 (v8sf)
9098 v8sf __builtin_ia32_sqrtps256 (v8sf)
9099 void __builtin_ia32_storedqu256 (pchar,v32qi)
9100 void __builtin_ia32_storeupd256 (pdouble,v4df)
9101 void __builtin_ia32_storeups256 (pfloat,v8sf)
9102 v4df __builtin_ia32_subpd256 (v4df,v4df)
9103 v8sf __builtin_ia32_subps256 (v8sf,v8sf)
9104 v4df __builtin_ia32_unpckhpd256 (v4df,v4df)
9105 v8sf __builtin_ia32_unpckhps256 (v8sf,v8sf)
9106 v4df __builtin_ia32_unpcklpd256 (v4df,v4df)
9107 v8sf __builtin_ia32_unpcklps256 (v8sf,v8sf)
9108 v4df __builtin_ia32_vbroadcastf128_pd256 (pcv2df)
9109 v8sf __builtin_ia32_vbroadcastf128_ps256 (pcv4sf)
9110 v4df __builtin_ia32_vbroadcastsd256 (pcdouble)
9111 v4sf __builtin_ia32_vbroadcastss (pcfloat)
9112 v8sf __builtin_ia32_vbroadcastss256 (pcfloat)
9113 v2df __builtin_ia32_vextractf128_pd256 (v4df,int)
9114 v4sf __builtin_ia32_vextractf128_ps256 (v8sf,int)
9115 v4si __builtin_ia32_vextractf128_si256 (v8si,int)
9116 v4df __builtin_ia32_vinsertf128_pd256 (v4df,v2df,int)
9117 v8sf __builtin_ia32_vinsertf128_ps256 (v8sf,v4sf,int)
9118 v8si __builtin_ia32_vinsertf128_si256 (v8si,v4si,int)
9119 v4df __builtin_ia32_vperm2f128_pd256 (v4df,v4df,int)
9120 v8sf __builtin_ia32_vperm2f128_ps256 (v8sf,v8sf,int)
9121 v8si __builtin_ia32_vperm2f128_si256 (v8si,v8si,int)
9122 v2df __builtin_ia32_vpermil2pd (v2df,v2df,v2di,int)
9123 v4df __builtin_ia32_vpermil2pd256 (v4df,v4df,v4di,int)
9124 v4sf __builtin_ia32_vpermil2ps (v4sf,v4sf,v4si,int)
9125 v8sf __builtin_ia32_vpermil2ps256 (v8sf,v8sf,v8si,int)
9126 v2df __builtin_ia32_vpermilpd (v2df,int)
9127 v4df __builtin_ia32_vpermilpd256 (v4df,int)
9128 v4sf __builtin_ia32_vpermilps (v4sf,int)
9129 v8sf __builtin_ia32_vpermilps256 (v8sf,int)
9130 v2df __builtin_ia32_vpermilvarpd (v2df,v2di)
9131 v4df __builtin_ia32_vpermilvarpd256 (v4df,v4di)
9132 v4sf __builtin_ia32_vpermilvarps (v4sf,v4si)
9133 v8sf __builtin_ia32_vpermilvarps256 (v8sf,v8si)
9134 int __builtin_ia32_vtestcpd (v2df,v2df,ptest)
9135 int __builtin_ia32_vtestcpd256 (v4df,v4df,ptest)
9136 int __builtin_ia32_vtestcps (v4sf,v4sf,ptest)
9137 int __builtin_ia32_vtestcps256 (v8sf,v8sf,ptest)
9138 int __builtin_ia32_vtestnzcpd (v2df,v2df,ptest)
9139 int __builtin_ia32_vtestnzcpd256 (v4df,v4df,ptest)
9140 int __builtin_ia32_vtestnzcps (v4sf,v4sf,ptest)
9141 int __builtin_ia32_vtestnzcps256 (v8sf,v8sf,ptest)
9142 int __builtin_ia32_vtestzpd (v2df,v2df,ptest)
9143 int __builtin_ia32_vtestzpd256 (v4df,v4df,ptest)
9144 int __builtin_ia32_vtestzps (v4sf,v4sf,ptest)
9145 int __builtin_ia32_vtestzps256 (v8sf,v8sf,ptest)
9146 void __builtin_ia32_vzeroall (void)
9147 void __builtin_ia32_vzeroupper (void)
9148 v4df __builtin_ia32_xorpd256 (v4df,v4df)
9149 v8sf __builtin_ia32_xorps256 (v8sf,v8sf)
9152 The following built-in functions are available when @option{-maes} is
9153 used. All of them generate the machine instruction that is part of the
9157 v2di __builtin_ia32_aesenc128 (v2di, v2di)
9158 v2di __builtin_ia32_aesenclast128 (v2di, v2di)
9159 v2di __builtin_ia32_aesdec128 (v2di, v2di)
9160 v2di __builtin_ia32_aesdeclast128 (v2di, v2di)
9161 v2di __builtin_ia32_aeskeygenassist128 (v2di, const int)
9162 v2di __builtin_ia32_aesimc128 (v2di)
9165 The following built-in function is available when @option{-mpclmul} is
9169 @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int)
9170 Generates the @code{pclmulqdq} machine instruction.
9173 The following built-in function is available when @option{-mfsgsbase} is
9174 used. All of them generate the machine instruction that is part of the
9178 unsigned int __builtin_ia32_rdfsbase32 (void)
9179 unsigned long long __builtin_ia32_rdfsbase64 (void)
9180 unsigned int __builtin_ia32_rdgsbase32 (void)
9181 unsigned long long __builtin_ia32_rdgsbase64 (void)
9182 void _writefsbase_u32 (unsigned int)
9183 void _writefsbase_u64 (unsigned long long)
9184 void _writegsbase_u32 (unsigned int)
9185 void _writegsbase_u64 (unsigned long long)
9188 The following built-in function is available when @option{-mrdrnd} is
9189 used. All of them generate the machine instruction that is part of the
9193 unsigned short __builtin_ia32_rdrand16 (void)
9194 unsigned int __builtin_ia32_rdrand32 (void)
9195 unsigned long long __builtin_ia32_rdrand64 (void)
9198 The following built-in functions are available when @option{-msse4a} is used.
9199 All of them generate the machine instruction that is part of the name.
9202 void __builtin_ia32_movntsd (double *, v2df)
9203 void __builtin_ia32_movntss (float *, v4sf)
9204 v2di __builtin_ia32_extrq (v2di, v16qi)
9205 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
9206 v2di __builtin_ia32_insertq (v2di, v2di)
9207 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
9210 The following built-in functions are available when @option{-mxop} is used.
9212 v2df __builtin_ia32_vfrczpd (v2df)
9213 v4sf __builtin_ia32_vfrczps (v4sf)
9214 v2df __builtin_ia32_vfrczsd (v2df, v2df)
9215 v4sf __builtin_ia32_vfrczss (v4sf, v4sf)
9216 v4df __builtin_ia32_vfrczpd256 (v4df)
9217 v8sf __builtin_ia32_vfrczps256 (v8sf)
9218 v2di __builtin_ia32_vpcmov (v2di, v2di, v2di)
9219 v2di __builtin_ia32_vpcmov_v2di (v2di, v2di, v2di)
9220 v4si __builtin_ia32_vpcmov_v4si (v4si, v4si, v4si)
9221 v8hi __builtin_ia32_vpcmov_v8hi (v8hi, v8hi, v8hi)
9222 v16qi __builtin_ia32_vpcmov_v16qi (v16qi, v16qi, v16qi)
9223 v2df __builtin_ia32_vpcmov_v2df (v2df, v2df, v2df)
9224 v4sf __builtin_ia32_vpcmov_v4sf (v4sf, v4sf, v4sf)
9225 v4di __builtin_ia32_vpcmov_v4di256 (v4di, v4di, v4di)
9226 v8si __builtin_ia32_vpcmov_v8si256 (v8si, v8si, v8si)
9227 v16hi __builtin_ia32_vpcmov_v16hi256 (v16hi, v16hi, v16hi)
9228 v32qi __builtin_ia32_vpcmov_v32qi256 (v32qi, v32qi, v32qi)
9229 v4df __builtin_ia32_vpcmov_v4df256 (v4df, v4df, v4df)
9230 v8sf __builtin_ia32_vpcmov_v8sf256 (v8sf, v8sf, v8sf)
9231 v16qi __builtin_ia32_vpcomeqb (v16qi, v16qi)
9232 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
9233 v4si __builtin_ia32_vpcomeqd (v4si, v4si)
9234 v2di __builtin_ia32_vpcomeqq (v2di, v2di)
9235 v16qi __builtin_ia32_vpcomequb (v16qi, v16qi)
9236 v4si __builtin_ia32_vpcomequd (v4si, v4si)
9237 v2di __builtin_ia32_vpcomequq (v2di, v2di)
9238 v8hi __builtin_ia32_vpcomequw (v8hi, v8hi)
9239 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
9240 v16qi __builtin_ia32_vpcomfalseb (v16qi, v16qi)
9241 v4si __builtin_ia32_vpcomfalsed (v4si, v4si)
9242 v2di __builtin_ia32_vpcomfalseq (v2di, v2di)
9243 v16qi __builtin_ia32_vpcomfalseub (v16qi, v16qi)
9244 v4si __builtin_ia32_vpcomfalseud (v4si, v4si)
9245 v2di __builtin_ia32_vpcomfalseuq (v2di, v2di)
9246 v8hi __builtin_ia32_vpcomfalseuw (v8hi, v8hi)
9247 v8hi __builtin_ia32_vpcomfalsew (v8hi, v8hi)
9248 v16qi __builtin_ia32_vpcomgeb (v16qi, v16qi)
9249 v4si __builtin_ia32_vpcomged (v4si, v4si)
9250 v2di __builtin_ia32_vpcomgeq (v2di, v2di)
9251 v16qi __builtin_ia32_vpcomgeub (v16qi, v16qi)
9252 v4si __builtin_ia32_vpcomgeud (v4si, v4si)
9253 v2di __builtin_ia32_vpcomgeuq (v2di, v2di)
9254 v8hi __builtin_ia32_vpcomgeuw (v8hi, v8hi)
9255 v8hi __builtin_ia32_vpcomgew (v8hi, v8hi)
9256 v16qi __builtin_ia32_vpcomgtb (v16qi, v16qi)
9257 v4si __builtin_ia32_vpcomgtd (v4si, v4si)
9258 v2di __builtin_ia32_vpcomgtq (v2di, v2di)
9259 v16qi __builtin_ia32_vpcomgtub (v16qi, v16qi)
9260 v4si __builtin_ia32_vpcomgtud (v4si, v4si)
9261 v2di __builtin_ia32_vpcomgtuq (v2di, v2di)
9262 v8hi __builtin_ia32_vpcomgtuw (v8hi, v8hi)
9263 v8hi __builtin_ia32_vpcomgtw (v8hi, v8hi)
9264 v16qi __builtin_ia32_vpcomleb (v16qi, v16qi)
9265 v4si __builtin_ia32_vpcomled (v4si, v4si)
9266 v2di __builtin_ia32_vpcomleq (v2di, v2di)
9267 v16qi __builtin_ia32_vpcomleub (v16qi, v16qi)
9268 v4si __builtin_ia32_vpcomleud (v4si, v4si)
9269 v2di __builtin_ia32_vpcomleuq (v2di, v2di)
9270 v8hi __builtin_ia32_vpcomleuw (v8hi, v8hi)
9271 v8hi __builtin_ia32_vpcomlew (v8hi, v8hi)
9272 v16qi __builtin_ia32_vpcomltb (v16qi, v16qi)
9273 v4si __builtin_ia32_vpcomltd (v4si, v4si)
9274 v2di __builtin_ia32_vpcomltq (v2di, v2di)
9275 v16qi __builtin_ia32_vpcomltub (v16qi, v16qi)
9276 v4si __builtin_ia32_vpcomltud (v4si, v4si)
9277 v2di __builtin_ia32_vpcomltuq (v2di, v2di)
9278 v8hi __builtin_ia32_vpcomltuw (v8hi, v8hi)
9279 v8hi __builtin_ia32_vpcomltw (v8hi, v8hi)
9280 v16qi __builtin_ia32_vpcomneb (v16qi, v16qi)
9281 v4si __builtin_ia32_vpcomned (v4si, v4si)
9282 v2di __builtin_ia32_vpcomneq (v2di, v2di)
9283 v16qi __builtin_ia32_vpcomneub (v16qi, v16qi)
9284 v4si __builtin_ia32_vpcomneud (v4si, v4si)
9285 v2di __builtin_ia32_vpcomneuq (v2di, v2di)
9286 v8hi __builtin_ia32_vpcomneuw (v8hi, v8hi)
9287 v8hi __builtin_ia32_vpcomnew (v8hi, v8hi)
9288 v16qi __builtin_ia32_vpcomtrueb (v16qi, v16qi)
9289 v4si __builtin_ia32_vpcomtrued (v4si, v4si)
9290 v2di __builtin_ia32_vpcomtrueq (v2di, v2di)
9291 v16qi __builtin_ia32_vpcomtrueub (v16qi, v16qi)
9292 v4si __builtin_ia32_vpcomtrueud (v4si, v4si)
9293 v2di __builtin_ia32_vpcomtrueuq (v2di, v2di)
9294 v8hi __builtin_ia32_vpcomtrueuw (v8hi, v8hi)
9295 v8hi __builtin_ia32_vpcomtruew (v8hi, v8hi)
9296 v4si __builtin_ia32_vphaddbd (v16qi)
9297 v2di __builtin_ia32_vphaddbq (v16qi)
9298 v8hi __builtin_ia32_vphaddbw (v16qi)
9299 v2di __builtin_ia32_vphadddq (v4si)
9300 v4si __builtin_ia32_vphaddubd (v16qi)
9301 v2di __builtin_ia32_vphaddubq (v16qi)
9302 v8hi __builtin_ia32_vphaddubw (v16qi)
9303 v2di __builtin_ia32_vphaddudq (v4si)
9304 v4si __builtin_ia32_vphadduwd (v8hi)
9305 v2di __builtin_ia32_vphadduwq (v8hi)
9306 v4si __builtin_ia32_vphaddwd (v8hi)
9307 v2di __builtin_ia32_vphaddwq (v8hi)
9308 v8hi __builtin_ia32_vphsubbw (v16qi)
9309 v2di __builtin_ia32_vphsubdq (v4si)
9310 v4si __builtin_ia32_vphsubwd (v8hi)
9311 v4si __builtin_ia32_vpmacsdd (v4si, v4si, v4si)
9312 v2di __builtin_ia32_vpmacsdqh (v4si, v4si, v2di)
9313 v2di __builtin_ia32_vpmacsdql (v4si, v4si, v2di)
9314 v4si __builtin_ia32_vpmacssdd (v4si, v4si, v4si)
9315 v2di __builtin_ia32_vpmacssdqh (v4si, v4si, v2di)
9316 v2di __builtin_ia32_vpmacssdql (v4si, v4si, v2di)
9317 v4si __builtin_ia32_vpmacsswd (v8hi, v8hi, v4si)
9318 v8hi __builtin_ia32_vpmacssww (v8hi, v8hi, v8hi)
9319 v4si __builtin_ia32_vpmacswd (v8hi, v8hi, v4si)
9320 v8hi __builtin_ia32_vpmacsww (v8hi, v8hi, v8hi)
9321 v4si __builtin_ia32_vpmadcsswd (v8hi, v8hi, v4si)
9322 v4si __builtin_ia32_vpmadcswd (v8hi, v8hi, v4si)
9323 v16qi __builtin_ia32_vpperm (v16qi, v16qi, v16qi)
9324 v16qi __builtin_ia32_vprotb (v16qi, v16qi)
9325 v4si __builtin_ia32_vprotd (v4si, v4si)
9326 v2di __builtin_ia32_vprotq (v2di, v2di)
9327 v8hi __builtin_ia32_vprotw (v8hi, v8hi)
9328 v16qi __builtin_ia32_vpshab (v16qi, v16qi)
9329 v4si __builtin_ia32_vpshad (v4si, v4si)
9330 v2di __builtin_ia32_vpshaq (v2di, v2di)
9331 v8hi __builtin_ia32_vpshaw (v8hi, v8hi)
9332 v16qi __builtin_ia32_vpshlb (v16qi, v16qi)
9333 v4si __builtin_ia32_vpshld (v4si, v4si)
9334 v2di __builtin_ia32_vpshlq (v2di, v2di)
9335 v8hi __builtin_ia32_vpshlw (v8hi, v8hi)
9338 The following built-in functions are available when @option{-mfma4} is used.
9339 All of them generate the machine instruction that is part of the name
9343 v2df __builtin_ia32_fmaddpd (v2df, v2df, v2df)
9344 v4sf __builtin_ia32_fmaddps (v4sf, v4sf, v4sf)
9345 v2df __builtin_ia32_fmaddsd (v2df, v2df, v2df)
9346 v4sf __builtin_ia32_fmaddss (v4sf, v4sf, v4sf)
9347 v2df __builtin_ia32_fmsubpd (v2df, v2df, v2df)
9348 v4sf __builtin_ia32_fmsubps (v4sf, v4sf, v4sf)
9349 v2df __builtin_ia32_fmsubsd (v2df, v2df, v2df)
9350 v4sf __builtin_ia32_fmsubss (v4sf, v4sf, v4sf)
9351 v2df __builtin_ia32_fnmaddpd (v2df, v2df, v2df)
9352 v4sf __builtin_ia32_fnmaddps (v4sf, v4sf, v4sf)
9353 v2df __builtin_ia32_fnmaddsd (v2df, v2df, v2df)
9354 v4sf __builtin_ia32_fnmaddss (v4sf, v4sf, v4sf)
9355 v2df __builtin_ia32_fnmsubpd (v2df, v2df, v2df)
9356 v4sf __builtin_ia32_fnmsubps (v4sf, v4sf, v4sf)
9357 v2df __builtin_ia32_fnmsubsd (v2df, v2df, v2df)
9358 v4sf __builtin_ia32_fnmsubss (v4sf, v4sf, v4sf)
9359 v2df __builtin_ia32_fmaddsubpd (v2df, v2df, v2df)
9360 v4sf __builtin_ia32_fmaddsubps (v4sf, v4sf, v4sf)
9361 v2df __builtin_ia32_fmsubaddpd (v2df, v2df, v2df)
9362 v4sf __builtin_ia32_fmsubaddps (v4sf, v4sf, v4sf)
9363 v4df __builtin_ia32_fmaddpd256 (v4df, v4df, v4df)
9364 v8sf __builtin_ia32_fmaddps256 (v8sf, v8sf, v8sf)
9365 v4df __builtin_ia32_fmsubpd256 (v4df, v4df, v4df)
9366 v8sf __builtin_ia32_fmsubps256 (v8sf, v8sf, v8sf)
9367 v4df __builtin_ia32_fnmaddpd256 (v4df, v4df, v4df)
9368 v8sf __builtin_ia32_fnmaddps256 (v8sf, v8sf, v8sf)
9369 v4df __builtin_ia32_fnmsubpd256 (v4df, v4df, v4df)
9370 v8sf __builtin_ia32_fnmsubps256 (v8sf, v8sf, v8sf)
9371 v4df __builtin_ia32_fmaddsubpd256 (v4df, v4df, v4df)
9372 v8sf __builtin_ia32_fmaddsubps256 (v8sf, v8sf, v8sf)
9373 v4df __builtin_ia32_fmsubaddpd256 (v4df, v4df, v4df)
9374 v8sf __builtin_ia32_fmsubaddps256 (v8sf, v8sf, v8sf)
9378 The following built-in functions are available when @option{-mlwp} is used.
9381 void __builtin_ia32_llwpcb16 (void *);
9382 void __builtin_ia32_llwpcb32 (void *);
9383 void __builtin_ia32_llwpcb64 (void *);
9384 void * __builtin_ia32_llwpcb16 (void);
9385 void * __builtin_ia32_llwpcb32 (void);
9386 void * __builtin_ia32_llwpcb64 (void);
9387 void __builtin_ia32_lwpval16 (unsigned short, unsigned int, unsigned short)
9388 void __builtin_ia32_lwpval32 (unsigned int, unsigned int, unsigned int)
9389 void __builtin_ia32_lwpval64 (unsigned __int64, unsigned int, unsigned int)
9390 unsigned char __builtin_ia32_lwpins16 (unsigned short, unsigned int, unsigned short)
9391 unsigned char __builtin_ia32_lwpins32 (unsigned int, unsigned int, unsigned int)
9392 unsigned char __builtin_ia32_lwpins64 (unsigned __int64, unsigned int, unsigned int)
9395 The following built-in functions are available when @option{-mbmi} is used.
9396 All of them generate the machine instruction that is part of the name.
9398 unsigned int __builtin_ia32_bextr_u32(unsigned int, unsigned int);
9399 unsigned long long __builtin_ia32_bextr_u64 (unsigned long long, unsigned long long);
9400 unsigned short __builtin_ia32_lzcnt_16(unsigned short);
9401 unsigned int __builtin_ia32_lzcnt_u32(unsigned int);
9402 unsigned long long __builtin_ia32_lzcnt_u64 (unsigned long long);
9405 The following built-in functions are available when @option{-m3dnow} is used.
9406 All of them generate the machine instruction that is part of the name.
9409 void __builtin_ia32_femms (void)
9410 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
9411 v2si __builtin_ia32_pf2id (v2sf)
9412 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
9413 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
9414 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
9415 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
9416 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
9417 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
9418 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
9419 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
9420 v2sf __builtin_ia32_pfrcp (v2sf)
9421 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
9422 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
9423 v2sf __builtin_ia32_pfrsqrt (v2sf)
9424 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
9425 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
9426 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
9427 v2sf __builtin_ia32_pi2fd (v2si)
9428 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
9431 The following built-in functions are available when both @option{-m3dnow}
9432 and @option{-march=athlon} are used. All of them generate the machine
9433 instruction that is part of the name.
9436 v2si __builtin_ia32_pf2iw (v2sf)
9437 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
9438 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
9439 v2sf __builtin_ia32_pi2fw (v2si)
9440 v2sf __builtin_ia32_pswapdsf (v2sf)
9441 v2si __builtin_ia32_pswapdsi (v2si)
9444 @node MIPS DSP Built-in Functions
9445 @subsection MIPS DSP Built-in Functions
9447 The MIPS DSP Application-Specific Extension (ASE) includes new
9448 instructions that are designed to improve the performance of DSP and
9449 media applications. It provides instructions that operate on packed
9450 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
9452 GCC supports MIPS DSP operations using both the generic
9453 vector extensions (@pxref{Vector Extensions}) and a collection of
9454 MIPS-specific built-in functions. Both kinds of support are
9455 enabled by the @option{-mdsp} command-line option.
9457 Revision 2 of the ASE was introduced in the second half of 2006.
9458 This revision adds extra instructions to the original ASE, but is
9459 otherwise backwards-compatible with it. You can select revision 2
9460 using the command-line option @option{-mdspr2}; this option implies
9463 The SCOUNT and POS bits of the DSP control register are global. The
9464 WRDSP, EXTPDP, EXTPDPV and MTHLIP instructions modify the SCOUNT and
9465 POS bits. During optimization, the compiler will not delete these
9466 instructions and it will not delete calls to functions containing
9469 At present, GCC only provides support for operations on 32-bit
9470 vectors. The vector type associated with 8-bit integer data is
9471 usually called @code{v4i8}, the vector type associated with Q7
9472 is usually called @code{v4q7}, the vector type associated with 16-bit
9473 integer data is usually called @code{v2i16}, and the vector type
9474 associated with Q15 is usually called @code{v2q15}. They can be
9475 defined in C as follows:
9478 typedef signed char v4i8 __attribute__ ((vector_size(4)));
9479 typedef signed char v4q7 __attribute__ ((vector_size(4)));
9480 typedef short v2i16 __attribute__ ((vector_size(4)));
9481 typedef short v2q15 __attribute__ ((vector_size(4)));
9484 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
9485 initialized in the same way as aggregates. For example:
9488 v4i8 a = @{1, 2, 3, 4@};
9490 b = (v4i8) @{5, 6, 7, 8@};
9492 v2q15 c = @{0x0fcb, 0x3a75@};
9494 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
9497 @emph{Note:} The CPU's endianness determines the order in which values
9498 are packed. On little-endian targets, the first value is the least
9499 significant and the last value is the most significant. The opposite
9500 order applies to big-endian targets. For example, the code above will
9501 set the lowest byte of @code{a} to @code{1} on little-endian targets
9502 and @code{4} on big-endian targets.
9504 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
9505 representation. As shown in this example, the integer representation
9506 of a Q7 value can be obtained by multiplying the fractional value by
9507 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
9508 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
9511 The table below lists the @code{v4i8} and @code{v2q15} operations for which
9512 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
9513 and @code{c} and @code{d} are @code{v2q15} values.
9515 @multitable @columnfractions .50 .50
9516 @item C code @tab MIPS instruction
9517 @item @code{a + b} @tab @code{addu.qb}
9518 @item @code{c + d} @tab @code{addq.ph}
9519 @item @code{a - b} @tab @code{subu.qb}
9520 @item @code{c - d} @tab @code{subq.ph}
9523 The table below lists the @code{v2i16} operation for which
9524 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
9525 @code{v2i16} values.
9527 @multitable @columnfractions .50 .50
9528 @item C code @tab MIPS instruction
9529 @item @code{e * f} @tab @code{mul.ph}
9532 It is easier to describe the DSP built-in functions if we first define
9533 the following types:
9538 typedef unsigned int ui32;
9539 typedef long long a64;
9542 @code{q31} and @code{i32} are actually the same as @code{int}, but we
9543 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
9544 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
9545 @code{long long}, but we use @code{a64} to indicate values that will
9546 be placed in one of the four DSP accumulators (@code{$ac0},
9547 @code{$ac1}, @code{$ac2} or @code{$ac3}).
9549 Also, some built-in functions prefer or require immediate numbers as
9550 parameters, because the corresponding DSP instructions accept both immediate
9551 numbers and register operands, or accept immediate numbers only. The
9552 immediate parameters are listed as follows.
9561 imm_n32_31: -32 to 31.
9562 imm_n512_511: -512 to 511.
9565 The following built-in functions map directly to a particular MIPS DSP
9566 instruction. Please refer to the architecture specification
9567 for details on what each instruction does.
9570 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
9571 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
9572 q31 __builtin_mips_addq_s_w (q31, q31)
9573 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
9574 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
9575 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
9576 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
9577 q31 __builtin_mips_subq_s_w (q31, q31)
9578 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
9579 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
9580 i32 __builtin_mips_addsc (i32, i32)
9581 i32 __builtin_mips_addwc (i32, i32)
9582 i32 __builtin_mips_modsub (i32, i32)
9583 i32 __builtin_mips_raddu_w_qb (v4i8)
9584 v2q15 __builtin_mips_absq_s_ph (v2q15)
9585 q31 __builtin_mips_absq_s_w (q31)
9586 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
9587 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
9588 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
9589 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
9590 q31 __builtin_mips_preceq_w_phl (v2q15)
9591 q31 __builtin_mips_preceq_w_phr (v2q15)
9592 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
9593 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
9594 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
9595 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
9596 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
9597 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
9598 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
9599 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
9600 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
9601 v4i8 __builtin_mips_shll_qb (v4i8, i32)
9602 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
9603 v2q15 __builtin_mips_shll_ph (v2q15, i32)
9604 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
9605 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
9606 q31 __builtin_mips_shll_s_w (q31, imm0_31)
9607 q31 __builtin_mips_shll_s_w (q31, i32)
9608 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
9609 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
9610 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
9611 v2q15 __builtin_mips_shra_ph (v2q15, i32)
9612 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
9613 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
9614 q31 __builtin_mips_shra_r_w (q31, imm0_31)
9615 q31 __builtin_mips_shra_r_w (q31, i32)
9616 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
9617 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
9618 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
9619 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
9620 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
9621 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
9622 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
9623 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
9624 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
9625 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
9626 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
9627 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
9628 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
9629 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
9630 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
9631 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
9632 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
9633 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
9634 i32 __builtin_mips_bitrev (i32)
9635 i32 __builtin_mips_insv (i32, i32)
9636 v4i8 __builtin_mips_repl_qb (imm0_255)
9637 v4i8 __builtin_mips_repl_qb (i32)
9638 v2q15 __builtin_mips_repl_ph (imm_n512_511)
9639 v2q15 __builtin_mips_repl_ph (i32)
9640 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
9641 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
9642 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
9643 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
9644 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
9645 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
9646 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
9647 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
9648 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
9649 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
9650 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
9651 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
9652 i32 __builtin_mips_extr_w (a64, imm0_31)
9653 i32 __builtin_mips_extr_w (a64, i32)
9654 i32 __builtin_mips_extr_r_w (a64, imm0_31)
9655 i32 __builtin_mips_extr_s_h (a64, i32)
9656 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
9657 i32 __builtin_mips_extr_rs_w (a64, i32)
9658 i32 __builtin_mips_extr_s_h (a64, imm0_31)
9659 i32 __builtin_mips_extr_r_w (a64, i32)
9660 i32 __builtin_mips_extp (a64, imm0_31)
9661 i32 __builtin_mips_extp (a64, i32)
9662 i32 __builtin_mips_extpdp (a64, imm0_31)
9663 i32 __builtin_mips_extpdp (a64, i32)
9664 a64 __builtin_mips_shilo (a64, imm_n32_31)
9665 a64 __builtin_mips_shilo (a64, i32)
9666 a64 __builtin_mips_mthlip (a64, i32)
9667 void __builtin_mips_wrdsp (i32, imm0_63)
9668 i32 __builtin_mips_rddsp (imm0_63)
9669 i32 __builtin_mips_lbux (void *, i32)
9670 i32 __builtin_mips_lhx (void *, i32)
9671 i32 __builtin_mips_lwx (void *, i32)
9672 i32 __builtin_mips_bposge32 (void)
9673 a64 __builtin_mips_madd (a64, i32, i32);
9674 a64 __builtin_mips_maddu (a64, ui32, ui32);
9675 a64 __builtin_mips_msub (a64, i32, i32);
9676 a64 __builtin_mips_msubu (a64, ui32, ui32);
9677 a64 __builtin_mips_mult (i32, i32);
9678 a64 __builtin_mips_multu (ui32, ui32);
9681 The following built-in functions map directly to a particular MIPS DSP REV 2
9682 instruction. Please refer to the architecture specification
9683 for details on what each instruction does.
9686 v4q7 __builtin_mips_absq_s_qb (v4q7);
9687 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
9688 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
9689 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
9690 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
9691 i32 __builtin_mips_append (i32, i32, imm0_31);
9692 i32 __builtin_mips_balign (i32, i32, imm0_3);
9693 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
9694 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
9695 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
9696 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
9697 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
9698 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
9699 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
9700 q31 __builtin_mips_mulq_rs_w (q31, q31);
9701 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
9702 q31 __builtin_mips_mulq_s_w (q31, q31);
9703 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
9704 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
9705 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
9706 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
9707 i32 __builtin_mips_prepend (i32, i32, imm0_31);
9708 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
9709 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
9710 v4i8 __builtin_mips_shra_qb (v4i8, i32);
9711 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
9712 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
9713 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
9714 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
9715 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
9716 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
9717 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
9718 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
9719 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
9720 q31 __builtin_mips_addqh_w (q31, q31);
9721 q31 __builtin_mips_addqh_r_w (q31, q31);
9722 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
9723 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
9724 q31 __builtin_mips_subqh_w (q31, q31);
9725 q31 __builtin_mips_subqh_r_w (q31, q31);
9726 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
9727 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
9728 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
9729 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
9730 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
9731 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
9735 @node MIPS Paired-Single Support
9736 @subsection MIPS Paired-Single Support
9738 The MIPS64 architecture includes a number of instructions that
9739 operate on pairs of single-precision floating-point values.
9740 Each pair is packed into a 64-bit floating-point register,
9741 with one element being designated the ``upper half'' and
9742 the other being designated the ``lower half''.
9744 GCC supports paired-single operations using both the generic
9745 vector extensions (@pxref{Vector Extensions}) and a collection of
9746 MIPS-specific built-in functions. Both kinds of support are
9747 enabled by the @option{-mpaired-single} command-line option.
9749 The vector type associated with paired-single values is usually
9750 called @code{v2sf}. It can be defined in C as follows:
9753 typedef float v2sf __attribute__ ((vector_size (8)));
9756 @code{v2sf} values are initialized in the same way as aggregates.
9760 v2sf a = @{1.5, 9.1@};
9763 b = (v2sf) @{e, f@};
9766 @emph{Note:} The CPU's endianness determines which value is stored in
9767 the upper half of a register and which value is stored in the lower half.
9768 On little-endian targets, the first value is the lower one and the second
9769 value is the upper one. The opposite order applies to big-endian targets.
9770 For example, the code above will set the lower half of @code{a} to
9771 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
9773 @node MIPS Loongson Built-in Functions
9774 @subsection MIPS Loongson Built-in Functions
9776 GCC provides intrinsics to access the SIMD instructions provided by the
9777 ST Microelectronics Loongson-2E and -2F processors. These intrinsics,
9778 available after inclusion of the @code{loongson.h} header file,
9779 operate on the following 64-bit vector types:
9782 @item @code{uint8x8_t}, a vector of eight unsigned 8-bit integers;
9783 @item @code{uint16x4_t}, a vector of four unsigned 16-bit integers;
9784 @item @code{uint32x2_t}, a vector of two unsigned 32-bit integers;
9785 @item @code{int8x8_t}, a vector of eight signed 8-bit integers;
9786 @item @code{int16x4_t}, a vector of four signed 16-bit integers;
9787 @item @code{int32x2_t}, a vector of two signed 32-bit integers.
9790 The intrinsics provided are listed below; each is named after the
9791 machine instruction to which it corresponds, with suffixes added as
9792 appropriate to distinguish intrinsics that expand to the same machine
9793 instruction yet have different argument types. Refer to the architecture
9794 documentation for a description of the functionality of each
9798 int16x4_t packsswh (int32x2_t s, int32x2_t t);
9799 int8x8_t packsshb (int16x4_t s, int16x4_t t);
9800 uint8x8_t packushb (uint16x4_t s, uint16x4_t t);
9801 uint32x2_t paddw_u (uint32x2_t s, uint32x2_t t);
9802 uint16x4_t paddh_u (uint16x4_t s, uint16x4_t t);
9803 uint8x8_t paddb_u (uint8x8_t s, uint8x8_t t);
9804 int32x2_t paddw_s (int32x2_t s, int32x2_t t);
9805 int16x4_t paddh_s (int16x4_t s, int16x4_t t);
9806 int8x8_t paddb_s (int8x8_t s, int8x8_t t);
9807 uint64_t paddd_u (uint64_t s, uint64_t t);
9808 int64_t paddd_s (int64_t s, int64_t t);
9809 int16x4_t paddsh (int16x4_t s, int16x4_t t);
9810 int8x8_t paddsb (int8x8_t s, int8x8_t t);
9811 uint16x4_t paddush (uint16x4_t s, uint16x4_t t);
9812 uint8x8_t paddusb (uint8x8_t s, uint8x8_t t);
9813 uint64_t pandn_ud (uint64_t s, uint64_t t);
9814 uint32x2_t pandn_uw (uint32x2_t s, uint32x2_t t);
9815 uint16x4_t pandn_uh (uint16x4_t s, uint16x4_t t);
9816 uint8x8_t pandn_ub (uint8x8_t s, uint8x8_t t);
9817 int64_t pandn_sd (int64_t s, int64_t t);
9818 int32x2_t pandn_sw (int32x2_t s, int32x2_t t);
9819 int16x4_t pandn_sh (int16x4_t s, int16x4_t t);
9820 int8x8_t pandn_sb (int8x8_t s, int8x8_t t);
9821 uint16x4_t pavgh (uint16x4_t s, uint16x4_t t);
9822 uint8x8_t pavgb (uint8x8_t s, uint8x8_t t);
9823 uint32x2_t pcmpeqw_u (uint32x2_t s, uint32x2_t t);
9824 uint16x4_t pcmpeqh_u (uint16x4_t s, uint16x4_t t);
9825 uint8x8_t pcmpeqb_u (uint8x8_t s, uint8x8_t t);
9826 int32x2_t pcmpeqw_s (int32x2_t s, int32x2_t t);
9827 int16x4_t pcmpeqh_s (int16x4_t s, int16x4_t t);
9828 int8x8_t pcmpeqb_s (int8x8_t s, int8x8_t t);
9829 uint32x2_t pcmpgtw_u (uint32x2_t s, uint32x2_t t);
9830 uint16x4_t pcmpgth_u (uint16x4_t s, uint16x4_t t);
9831 uint8x8_t pcmpgtb_u (uint8x8_t s, uint8x8_t t);
9832 int32x2_t pcmpgtw_s (int32x2_t s, int32x2_t t);
9833 int16x4_t pcmpgth_s (int16x4_t s, int16x4_t t);
9834 int8x8_t pcmpgtb_s (int8x8_t s, int8x8_t t);
9835 uint16x4_t pextrh_u (uint16x4_t s, int field);
9836 int16x4_t pextrh_s (int16x4_t s, int field);
9837 uint16x4_t pinsrh_0_u (uint16x4_t s, uint16x4_t t);
9838 uint16x4_t pinsrh_1_u (uint16x4_t s, uint16x4_t t);
9839 uint16x4_t pinsrh_2_u (uint16x4_t s, uint16x4_t t);
9840 uint16x4_t pinsrh_3_u (uint16x4_t s, uint16x4_t t);
9841 int16x4_t pinsrh_0_s (int16x4_t s, int16x4_t t);
9842 int16x4_t pinsrh_1_s (int16x4_t s, int16x4_t t);
9843 int16x4_t pinsrh_2_s (int16x4_t s, int16x4_t t);
9844 int16x4_t pinsrh_3_s (int16x4_t s, int16x4_t t);
9845 int32x2_t pmaddhw (int16x4_t s, int16x4_t t);
9846 int16x4_t pmaxsh (int16x4_t s, int16x4_t t);
9847 uint8x8_t pmaxub (uint8x8_t s, uint8x8_t t);
9848 int16x4_t pminsh (int16x4_t s, int16x4_t t);
9849 uint8x8_t pminub (uint8x8_t s, uint8x8_t t);
9850 uint8x8_t pmovmskb_u (uint8x8_t s);
9851 int8x8_t pmovmskb_s (int8x8_t s);
9852 uint16x4_t pmulhuh (uint16x4_t s, uint16x4_t t);
9853 int16x4_t pmulhh (int16x4_t s, int16x4_t t);
9854 int16x4_t pmullh (int16x4_t s, int16x4_t t);
9855 int64_t pmuluw (uint32x2_t s, uint32x2_t t);
9856 uint8x8_t pasubub (uint8x8_t s, uint8x8_t t);
9857 uint16x4_t biadd (uint8x8_t s);
9858 uint16x4_t psadbh (uint8x8_t s, uint8x8_t t);
9859 uint16x4_t pshufh_u (uint16x4_t dest, uint16x4_t s, uint8_t order);
9860 int16x4_t pshufh_s (int16x4_t dest, int16x4_t s, uint8_t order);
9861 uint16x4_t psllh_u (uint16x4_t s, uint8_t amount);
9862 int16x4_t psllh_s (int16x4_t s, uint8_t amount);
9863 uint32x2_t psllw_u (uint32x2_t s, uint8_t amount);
9864 int32x2_t psllw_s (int32x2_t s, uint8_t amount);
9865 uint16x4_t psrlh_u (uint16x4_t s, uint8_t amount);
9866 int16x4_t psrlh_s (int16x4_t s, uint8_t amount);
9867 uint32x2_t psrlw_u (uint32x2_t s, uint8_t amount);
9868 int32x2_t psrlw_s (int32x2_t s, uint8_t amount);
9869 uint16x4_t psrah_u (uint16x4_t s, uint8_t amount);
9870 int16x4_t psrah_s (int16x4_t s, uint8_t amount);
9871 uint32x2_t psraw_u (uint32x2_t s, uint8_t amount);
9872 int32x2_t psraw_s (int32x2_t s, uint8_t amount);
9873 uint32x2_t psubw_u (uint32x2_t s, uint32x2_t t);
9874 uint16x4_t psubh_u (uint16x4_t s, uint16x4_t t);
9875 uint8x8_t psubb_u (uint8x8_t s, uint8x8_t t);
9876 int32x2_t psubw_s (int32x2_t s, int32x2_t t);
9877 int16x4_t psubh_s (int16x4_t s, int16x4_t t);
9878 int8x8_t psubb_s (int8x8_t s, int8x8_t t);
9879 uint64_t psubd_u (uint64_t s, uint64_t t);
9880 int64_t psubd_s (int64_t s, int64_t t);
9881 int16x4_t psubsh (int16x4_t s, int16x4_t t);
9882 int8x8_t psubsb (int8x8_t s, int8x8_t t);
9883 uint16x4_t psubush (uint16x4_t s, uint16x4_t t);
9884 uint8x8_t psubusb (uint8x8_t s, uint8x8_t t);
9885 uint32x2_t punpckhwd_u (uint32x2_t s, uint32x2_t t);
9886 uint16x4_t punpckhhw_u (uint16x4_t s, uint16x4_t t);
9887 uint8x8_t punpckhbh_u (uint8x8_t s, uint8x8_t t);
9888 int32x2_t punpckhwd_s (int32x2_t s, int32x2_t t);
9889 int16x4_t punpckhhw_s (int16x4_t s, int16x4_t t);
9890 int8x8_t punpckhbh_s (int8x8_t s, int8x8_t t);
9891 uint32x2_t punpcklwd_u (uint32x2_t s, uint32x2_t t);
9892 uint16x4_t punpcklhw_u (uint16x4_t s, uint16x4_t t);
9893 uint8x8_t punpcklbh_u (uint8x8_t s, uint8x8_t t);
9894 int32x2_t punpcklwd_s (int32x2_t s, int32x2_t t);
9895 int16x4_t punpcklhw_s (int16x4_t s, int16x4_t t);
9896 int8x8_t punpcklbh_s (int8x8_t s, int8x8_t t);
9900 * Paired-Single Arithmetic::
9901 * Paired-Single Built-in Functions::
9902 * MIPS-3D Built-in Functions::
9905 @node Paired-Single Arithmetic
9906 @subsubsection Paired-Single Arithmetic
9908 The table below lists the @code{v2sf} operations for which hardware
9909 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
9910 values and @code{x} is an integral value.
9912 @multitable @columnfractions .50 .50
9913 @item C code @tab MIPS instruction
9914 @item @code{a + b} @tab @code{add.ps}
9915 @item @code{a - b} @tab @code{sub.ps}
9916 @item @code{-a} @tab @code{neg.ps}
9917 @item @code{a * b} @tab @code{mul.ps}
9918 @item @code{a * b + c} @tab @code{madd.ps}
9919 @item @code{a * b - c} @tab @code{msub.ps}
9920 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
9921 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
9922 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
9925 Note that the multiply-accumulate instructions can be disabled
9926 using the command-line option @code{-mno-fused-madd}.
9928 @node Paired-Single Built-in Functions
9929 @subsubsection Paired-Single Built-in Functions
9931 The following paired-single functions map directly to a particular
9932 MIPS instruction. Please refer to the architecture specification
9933 for details on what each instruction does.
9936 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
9937 Pair lower lower (@code{pll.ps}).
9939 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
9940 Pair upper lower (@code{pul.ps}).
9942 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
9943 Pair lower upper (@code{plu.ps}).
9945 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
9946 Pair upper upper (@code{puu.ps}).
9948 @item v2sf __builtin_mips_cvt_ps_s (float, float)
9949 Convert pair to paired single (@code{cvt.ps.s}).
9951 @item float __builtin_mips_cvt_s_pl (v2sf)
9952 Convert pair lower to single (@code{cvt.s.pl}).
9954 @item float __builtin_mips_cvt_s_pu (v2sf)
9955 Convert pair upper to single (@code{cvt.s.pu}).
9957 @item v2sf __builtin_mips_abs_ps (v2sf)
9958 Absolute value (@code{abs.ps}).
9960 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
9961 Align variable (@code{alnv.ps}).
9963 @emph{Note:} The value of the third parameter must be 0 or 4
9964 modulo 8, otherwise the result will be unpredictable. Please read the
9965 instruction description for details.
9968 The following multi-instruction functions are also available.
9969 In each case, @var{cond} can be any of the 16 floating-point conditions:
9970 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
9971 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
9972 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
9975 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9976 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9977 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
9978 @code{movt.ps}/@code{movf.ps}).
9980 The @code{movt} functions return the value @var{x} computed by:
9983 c.@var{cond}.ps @var{cc},@var{a},@var{b}
9984 mov.ps @var{x},@var{c}
9985 movt.ps @var{x},@var{d},@var{cc}
9988 The @code{movf} functions are similar but use @code{movf.ps} instead
9991 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9992 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9993 Comparison of two paired-single values (@code{c.@var{cond}.ps},
9994 @code{bc1t}/@code{bc1f}).
9996 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
9997 and return either the upper or lower half of the result. For example:
10001 if (__builtin_mips_upper_c_eq_ps (a, b))
10002 upper_halves_are_equal ();
10004 upper_halves_are_unequal ();
10006 if (__builtin_mips_lower_c_eq_ps (a, b))
10007 lower_halves_are_equal ();
10009 lower_halves_are_unequal ();
10013 @node MIPS-3D Built-in Functions
10014 @subsubsection MIPS-3D Built-in Functions
10016 The MIPS-3D Application-Specific Extension (ASE) includes additional
10017 paired-single instructions that are designed to improve the performance
10018 of 3D graphics operations. Support for these instructions is controlled
10019 by the @option{-mips3d} command-line option.
10021 The functions listed below map directly to a particular MIPS-3D
10022 instruction. Please refer to the architecture specification for
10023 more details on what each instruction does.
10026 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
10027 Reduction add (@code{addr.ps}).
10029 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
10030 Reduction multiply (@code{mulr.ps}).
10032 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
10033 Convert paired single to paired word (@code{cvt.pw.ps}).
10035 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
10036 Convert paired word to paired single (@code{cvt.ps.pw}).
10038 @item float __builtin_mips_recip1_s (float)
10039 @itemx double __builtin_mips_recip1_d (double)
10040 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
10041 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
10043 @item float __builtin_mips_recip2_s (float, float)
10044 @itemx double __builtin_mips_recip2_d (double, double)
10045 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
10046 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
10048 @item float __builtin_mips_rsqrt1_s (float)
10049 @itemx double __builtin_mips_rsqrt1_d (double)
10050 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
10051 Reduced precision reciprocal square root (sequence step 1)
10052 (@code{rsqrt1.@var{fmt}}).
10054 @item float __builtin_mips_rsqrt2_s (float, float)
10055 @itemx double __builtin_mips_rsqrt2_d (double, double)
10056 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
10057 Reduced precision reciprocal square root (sequence step 2)
10058 (@code{rsqrt2.@var{fmt}}).
10061 The following multi-instruction functions are also available.
10062 In each case, @var{cond} can be any of the 16 floating-point conditions:
10063 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
10064 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
10065 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
10068 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
10069 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
10070 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
10071 @code{bc1t}/@code{bc1f}).
10073 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
10074 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
10079 if (__builtin_mips_cabs_eq_s (a, b))
10085 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10086 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10087 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
10088 @code{bc1t}/@code{bc1f}).
10090 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
10091 and return either the upper or lower half of the result. For example:
10095 if (__builtin_mips_upper_cabs_eq_ps (a, b))
10096 upper_halves_are_equal ();
10098 upper_halves_are_unequal ();
10100 if (__builtin_mips_lower_cabs_eq_ps (a, b))
10101 lower_halves_are_equal ();
10103 lower_halves_are_unequal ();
10106 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10107 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10108 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
10109 @code{movt.ps}/@code{movf.ps}).
10111 The @code{movt} functions return the value @var{x} computed by:
10114 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
10115 mov.ps @var{x},@var{c}
10116 movt.ps @var{x},@var{d},@var{cc}
10119 The @code{movf} functions are similar but use @code{movf.ps} instead
10122 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10123 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10124 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10125 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10126 Comparison of two paired-single values
10127 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
10128 @code{bc1any2t}/@code{bc1any2f}).
10130 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
10131 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
10132 result is true and the @code{all} forms return true if both results are true.
10137 if (__builtin_mips_any_c_eq_ps (a, b))
10142 if (__builtin_mips_all_c_eq_ps (a, b))
10148 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10149 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10150 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10151 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10152 Comparison of four paired-single values
10153 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
10154 @code{bc1any4t}/@code{bc1any4f}).
10156 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
10157 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
10158 The @code{any} forms return true if any of the four results are true
10159 and the @code{all} forms return true if all four results are true.
10164 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
10169 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
10176 @node picoChip Built-in Functions
10177 @subsection picoChip Built-in Functions
10179 GCC provides an interface to selected machine instructions from the
10180 picoChip instruction set.
10183 @item int __builtin_sbc (int @var{value})
10184 Sign bit count. Return the number of consecutive bits in @var{value}
10185 which have the same value as the sign-bit. The result is the number of
10186 leading sign bits minus one, giving the number of redundant sign bits in
10189 @item int __builtin_byteswap (int @var{value})
10190 Byte swap. Return the result of swapping the upper and lower bytes of
10193 @item int __builtin_brev (int @var{value})
10194 Bit reversal. Return the result of reversing the bits in
10195 @var{value}. Bit 15 is swapped with bit 0, bit 14 is swapped with bit 1,
10198 @item int __builtin_adds (int @var{x}, int @var{y})
10199 Saturating addition. Return the result of adding @var{x} and @var{y},
10200 storing the value 32767 if the result overflows.
10202 @item int __builtin_subs (int @var{x}, int @var{y})
10203 Saturating subtraction. Return the result of subtracting @var{y} from
10204 @var{x}, storing the value @minus{}32768 if the result overflows.
10206 @item void __builtin_halt (void)
10207 Halt. The processor will stop execution. This built-in is useful for
10208 implementing assertions.
10212 @node Other MIPS Built-in Functions
10213 @subsection Other MIPS Built-in Functions
10215 GCC provides other MIPS-specific built-in functions:
10218 @item void __builtin_mips_cache (int @var{op}, const volatile void *@var{addr})
10219 Insert a @samp{cache} instruction with operands @var{op} and @var{addr}.
10220 GCC defines the preprocessor macro @code{___GCC_HAVE_BUILTIN_MIPS_CACHE}
10221 when this function is available.
10224 @node PowerPC AltiVec/VSX Built-in Functions
10225 @subsection PowerPC AltiVec Built-in Functions
10227 GCC provides an interface for the PowerPC family of processors to access
10228 the AltiVec operations described in Motorola's AltiVec Programming
10229 Interface Manual. The interface is made available by including
10230 @code{<altivec.h>} and using @option{-maltivec} and
10231 @option{-mabi=altivec}. The interface supports the following vector
10235 vector unsigned char
10239 vector unsigned short
10240 vector signed short
10244 vector unsigned int
10250 If @option{-mvsx} is used the following additional vector types are
10254 vector unsigned long
10259 The long types are only implemented for 64-bit code generation, and
10260 the long type is only used in the floating point/integer conversion
10263 GCC's implementation of the high-level language interface available from
10264 C and C++ code differs from Motorola's documentation in several ways.
10269 A vector constant is a list of constant expressions within curly braces.
10272 A vector initializer requires no cast if the vector constant is of the
10273 same type as the variable it is initializing.
10276 If @code{signed} or @code{unsigned} is omitted, the signedness of the
10277 vector type is the default signedness of the base type. The default
10278 varies depending on the operating system, so a portable program should
10279 always specify the signedness.
10282 Compiling with @option{-maltivec} adds keywords @code{__vector},
10283 @code{vector}, @code{__pixel}, @code{pixel}, @code{__bool} and
10284 @code{bool}. When compiling ISO C, the context-sensitive substitution
10285 of the keywords @code{vector}, @code{pixel} and @code{bool} is
10286 disabled. To use them, you must include @code{<altivec.h>} instead.
10289 GCC allows using a @code{typedef} name as the type specifier for a
10293 For C, overloaded functions are implemented with macros so the following
10297 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
10300 Since @code{vec_add} is a macro, the vector constant in the example
10301 is treated as four separate arguments. Wrap the entire argument in
10302 parentheses for this to work.
10305 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
10306 Internally, GCC uses built-in functions to achieve the functionality in
10307 the aforementioned header file, but they are not supported and are
10308 subject to change without notice.
10310 The following interfaces are supported for the generic and specific
10311 AltiVec operations and the AltiVec predicates. In cases where there
10312 is a direct mapping between generic and specific operations, only the
10313 generic names are shown here, although the specific operations can also
10316 Arguments that are documented as @code{const int} require literal
10317 integral values within the range required for that operation.
10320 vector signed char vec_abs (vector signed char);
10321 vector signed short vec_abs (vector signed short);
10322 vector signed int vec_abs (vector signed int);
10323 vector float vec_abs (vector float);
10325 vector signed char vec_abss (vector signed char);
10326 vector signed short vec_abss (vector signed short);
10327 vector signed int vec_abss (vector signed int);
10329 vector signed char vec_add (vector bool char, vector signed char);
10330 vector signed char vec_add (vector signed char, vector bool char);
10331 vector signed char vec_add (vector signed char, vector signed char);
10332 vector unsigned char vec_add (vector bool char, vector unsigned char);
10333 vector unsigned char vec_add (vector unsigned char, vector bool char);
10334 vector unsigned char vec_add (vector unsigned char,
10335 vector unsigned char);
10336 vector signed short vec_add (vector bool short, vector signed short);
10337 vector signed short vec_add (vector signed short, vector bool short);
10338 vector signed short vec_add (vector signed short, vector signed short);
10339 vector unsigned short vec_add (vector bool short,
10340 vector unsigned short);
10341 vector unsigned short vec_add (vector unsigned short,
10342 vector bool short);
10343 vector unsigned short vec_add (vector unsigned short,
10344 vector unsigned short);
10345 vector signed int vec_add (vector bool int, vector signed int);
10346 vector signed int vec_add (vector signed int, vector bool int);
10347 vector signed int vec_add (vector signed int, vector signed int);
10348 vector unsigned int vec_add (vector bool int, vector unsigned int);
10349 vector unsigned int vec_add (vector unsigned int, vector bool int);
10350 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
10351 vector float vec_add (vector float, vector float);
10353 vector float vec_vaddfp (vector float, vector float);
10355 vector signed int vec_vadduwm (vector bool int, vector signed int);
10356 vector signed int vec_vadduwm (vector signed int, vector bool int);
10357 vector signed int vec_vadduwm (vector signed int, vector signed int);
10358 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
10359 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
10360 vector unsigned int vec_vadduwm (vector unsigned int,
10361 vector unsigned int);
10363 vector signed short vec_vadduhm (vector bool short,
10364 vector signed short);
10365 vector signed short vec_vadduhm (vector signed short,
10366 vector bool short);
10367 vector signed short vec_vadduhm (vector signed short,
10368 vector signed short);
10369 vector unsigned short vec_vadduhm (vector bool short,
10370 vector unsigned short);
10371 vector unsigned short vec_vadduhm (vector unsigned short,
10372 vector bool short);
10373 vector unsigned short vec_vadduhm (vector unsigned short,
10374 vector unsigned short);
10376 vector signed char vec_vaddubm (vector bool char, vector signed char);
10377 vector signed char vec_vaddubm (vector signed char, vector bool char);
10378 vector signed char vec_vaddubm (vector signed char, vector signed char);
10379 vector unsigned char vec_vaddubm (vector bool char,
10380 vector unsigned char);
10381 vector unsigned char vec_vaddubm (vector unsigned char,
10383 vector unsigned char vec_vaddubm (vector unsigned char,
10384 vector unsigned char);
10386 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
10388 vector unsigned char vec_adds (vector bool char, vector unsigned char);
10389 vector unsigned char vec_adds (vector unsigned char, vector bool char);
10390 vector unsigned char vec_adds (vector unsigned char,
10391 vector unsigned char);
10392 vector signed char vec_adds (vector bool char, vector signed char);
10393 vector signed char vec_adds (vector signed char, vector bool char);
10394 vector signed char vec_adds (vector signed char, vector signed char);
10395 vector unsigned short vec_adds (vector bool short,
10396 vector unsigned short);
10397 vector unsigned short vec_adds (vector unsigned short,
10398 vector bool short);
10399 vector unsigned short vec_adds (vector unsigned short,
10400 vector unsigned short);
10401 vector signed short vec_adds (vector bool short, vector signed short);
10402 vector signed short vec_adds (vector signed short, vector bool short);
10403 vector signed short vec_adds (vector signed short, vector signed short);
10404 vector unsigned int vec_adds (vector bool int, vector unsigned int);
10405 vector unsigned int vec_adds (vector unsigned int, vector bool int);
10406 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
10407 vector signed int vec_adds (vector bool int, vector signed int);
10408 vector signed int vec_adds (vector signed int, vector bool int);
10409 vector signed int vec_adds (vector signed int, vector signed int);
10411 vector signed int vec_vaddsws (vector bool int, vector signed int);
10412 vector signed int vec_vaddsws (vector signed int, vector bool int);
10413 vector signed int vec_vaddsws (vector signed int, vector signed int);
10415 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
10416 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
10417 vector unsigned int vec_vadduws (vector unsigned int,
10418 vector unsigned int);
10420 vector signed short vec_vaddshs (vector bool short,
10421 vector signed short);
10422 vector signed short vec_vaddshs (vector signed short,
10423 vector bool short);
10424 vector signed short vec_vaddshs (vector signed short,
10425 vector signed short);
10427 vector unsigned short vec_vadduhs (vector bool short,
10428 vector unsigned short);
10429 vector unsigned short vec_vadduhs (vector unsigned short,
10430 vector bool short);
10431 vector unsigned short vec_vadduhs (vector unsigned short,
10432 vector unsigned short);
10434 vector signed char vec_vaddsbs (vector bool char, vector signed char);
10435 vector signed char vec_vaddsbs (vector signed char, vector bool char);
10436 vector signed char vec_vaddsbs (vector signed char, vector signed char);
10438 vector unsigned char vec_vaddubs (vector bool char,
10439 vector unsigned char);
10440 vector unsigned char vec_vaddubs (vector unsigned char,
10442 vector unsigned char vec_vaddubs (vector unsigned char,
10443 vector unsigned char);
10445 vector float vec_and (vector float, vector float);
10446 vector float vec_and (vector float, vector bool int);
10447 vector float vec_and (vector bool int, vector float);
10448 vector bool int vec_and (vector bool int, vector bool int);
10449 vector signed int vec_and (vector bool int, vector signed int);
10450 vector signed int vec_and (vector signed int, vector bool int);
10451 vector signed int vec_and (vector signed int, vector signed int);
10452 vector unsigned int vec_and (vector bool int, vector unsigned int);
10453 vector unsigned int vec_and (vector unsigned int, vector bool int);
10454 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
10455 vector bool short vec_and (vector bool short, vector bool short);
10456 vector signed short vec_and (vector bool short, vector signed short);
10457 vector signed short vec_and (vector signed short, vector bool short);
10458 vector signed short vec_and (vector signed short, vector signed short);
10459 vector unsigned short vec_and (vector bool short,
10460 vector unsigned short);
10461 vector unsigned short vec_and (vector unsigned short,
10462 vector bool short);
10463 vector unsigned short vec_and (vector unsigned short,
10464 vector unsigned short);
10465 vector signed char vec_and (vector bool char, vector signed char);
10466 vector bool char vec_and (vector bool char, vector bool char);
10467 vector signed char vec_and (vector signed char, vector bool char);
10468 vector signed char vec_and (vector signed char, vector signed char);
10469 vector unsigned char vec_and (vector bool char, vector unsigned char);
10470 vector unsigned char vec_and (vector unsigned char, vector bool char);
10471 vector unsigned char vec_and (vector unsigned char,
10472 vector unsigned char);
10474 vector float vec_andc (vector float, vector float);
10475 vector float vec_andc (vector float, vector bool int);
10476 vector float vec_andc (vector bool int, vector float);
10477 vector bool int vec_andc (vector bool int, vector bool int);
10478 vector signed int vec_andc (vector bool int, vector signed int);
10479 vector signed int vec_andc (vector signed int, vector bool int);
10480 vector signed int vec_andc (vector signed int, vector signed int);
10481 vector unsigned int vec_andc (vector bool int, vector unsigned int);
10482 vector unsigned int vec_andc (vector unsigned int, vector bool int);
10483 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
10484 vector bool short vec_andc (vector bool short, vector bool short);
10485 vector signed short vec_andc (vector bool short, vector signed short);
10486 vector signed short vec_andc (vector signed short, vector bool short);
10487 vector signed short vec_andc (vector signed short, vector signed short);
10488 vector unsigned short vec_andc (vector bool short,
10489 vector unsigned short);
10490 vector unsigned short vec_andc (vector unsigned short,
10491 vector bool short);
10492 vector unsigned short vec_andc (vector unsigned short,
10493 vector unsigned short);
10494 vector signed char vec_andc (vector bool char, vector signed char);
10495 vector bool char vec_andc (vector bool char, vector bool char);
10496 vector signed char vec_andc (vector signed char, vector bool char);
10497 vector signed char vec_andc (vector signed char, vector signed char);
10498 vector unsigned char vec_andc (vector bool char, vector unsigned char);
10499 vector unsigned char vec_andc (vector unsigned char, vector bool char);
10500 vector unsigned char vec_andc (vector unsigned char,
10501 vector unsigned char);
10503 vector unsigned char vec_avg (vector unsigned char,
10504 vector unsigned char);
10505 vector signed char vec_avg (vector signed char, vector signed char);
10506 vector unsigned short vec_avg (vector unsigned short,
10507 vector unsigned short);
10508 vector signed short vec_avg (vector signed short, vector signed short);
10509 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
10510 vector signed int vec_avg (vector signed int, vector signed int);
10512 vector signed int vec_vavgsw (vector signed int, vector signed int);
10514 vector unsigned int vec_vavguw (vector unsigned int,
10515 vector unsigned int);
10517 vector signed short vec_vavgsh (vector signed short,
10518 vector signed short);
10520 vector unsigned short vec_vavguh (vector unsigned short,
10521 vector unsigned short);
10523 vector signed char vec_vavgsb (vector signed char, vector signed char);
10525 vector unsigned char vec_vavgub (vector unsigned char,
10526 vector unsigned char);
10528 vector float vec_copysign (vector float);
10530 vector float vec_ceil (vector float);
10532 vector signed int vec_cmpb (vector float, vector float);
10534 vector bool char vec_cmpeq (vector signed char, vector signed char);
10535 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
10536 vector bool short vec_cmpeq (vector signed short, vector signed short);
10537 vector bool short vec_cmpeq (vector unsigned short,
10538 vector unsigned short);
10539 vector bool int vec_cmpeq (vector signed int, vector signed int);
10540 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
10541 vector bool int vec_cmpeq (vector float, vector float);
10543 vector bool int vec_vcmpeqfp (vector float, vector float);
10545 vector bool int vec_vcmpequw (vector signed int, vector signed int);
10546 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
10548 vector bool short vec_vcmpequh (vector signed short,
10549 vector signed short);
10550 vector bool short vec_vcmpequh (vector unsigned short,
10551 vector unsigned short);
10553 vector bool char vec_vcmpequb (vector signed char, vector signed char);
10554 vector bool char vec_vcmpequb (vector unsigned char,
10555 vector unsigned char);
10557 vector bool int vec_cmpge (vector float, vector float);
10559 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
10560 vector bool char vec_cmpgt (vector signed char, vector signed char);
10561 vector bool short vec_cmpgt (vector unsigned short,
10562 vector unsigned short);
10563 vector bool short vec_cmpgt (vector signed short, vector signed short);
10564 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
10565 vector bool int vec_cmpgt (vector signed int, vector signed int);
10566 vector bool int vec_cmpgt (vector float, vector float);
10568 vector bool int vec_vcmpgtfp (vector float, vector float);
10570 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
10572 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
10574 vector bool short vec_vcmpgtsh (vector signed short,
10575 vector signed short);
10577 vector bool short vec_vcmpgtuh (vector unsigned short,
10578 vector unsigned short);
10580 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
10582 vector bool char vec_vcmpgtub (vector unsigned char,
10583 vector unsigned char);
10585 vector bool int vec_cmple (vector float, vector float);
10587 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
10588 vector bool char vec_cmplt (vector signed char, vector signed char);
10589 vector bool short vec_cmplt (vector unsigned short,
10590 vector unsigned short);
10591 vector bool short vec_cmplt (vector signed short, vector signed short);
10592 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
10593 vector bool int vec_cmplt (vector signed int, vector signed int);
10594 vector bool int vec_cmplt (vector float, vector float);
10596 vector float vec_ctf (vector unsigned int, const int);
10597 vector float vec_ctf (vector signed int, const int);
10599 vector float vec_vcfsx (vector signed int, const int);
10601 vector float vec_vcfux (vector unsigned int, const int);
10603 vector signed int vec_cts (vector float, const int);
10605 vector unsigned int vec_ctu (vector float, const int);
10607 void vec_dss (const int);
10609 void vec_dssall (void);
10611 void vec_dst (const vector unsigned char *, int, const int);
10612 void vec_dst (const vector signed char *, int, const int);
10613 void vec_dst (const vector bool char *, int, const int);
10614 void vec_dst (const vector unsigned short *, int, const int);
10615 void vec_dst (const vector signed short *, int, const int);
10616 void vec_dst (const vector bool short *, int, const int);
10617 void vec_dst (const vector pixel *, int, const int);
10618 void vec_dst (const vector unsigned int *, int, const int);
10619 void vec_dst (const vector signed int *, int, const int);
10620 void vec_dst (const vector bool int *, int, const int);
10621 void vec_dst (const vector float *, int, const int);
10622 void vec_dst (const unsigned char *, int, const int);
10623 void vec_dst (const signed char *, int, const int);
10624 void vec_dst (const unsigned short *, int, const int);
10625 void vec_dst (const short *, int, const int);
10626 void vec_dst (const unsigned int *, int, const int);
10627 void vec_dst (const int *, int, const int);
10628 void vec_dst (const unsigned long *, int, const int);
10629 void vec_dst (const long *, int, const int);
10630 void vec_dst (const float *, int, const int);
10632 void vec_dstst (const vector unsigned char *, int, const int);
10633 void vec_dstst (const vector signed char *, int, const int);
10634 void vec_dstst (const vector bool char *, int, const int);
10635 void vec_dstst (const vector unsigned short *, int, const int);
10636 void vec_dstst (const vector signed short *, int, const int);
10637 void vec_dstst (const vector bool short *, int, const int);
10638 void vec_dstst (const vector pixel *, int, const int);
10639 void vec_dstst (const vector unsigned int *, int, const int);
10640 void vec_dstst (const vector signed int *, int, const int);
10641 void vec_dstst (const vector bool int *, int, const int);
10642 void vec_dstst (const vector float *, int, const int);
10643 void vec_dstst (const unsigned char *, int, const int);
10644 void vec_dstst (const signed char *, int, const int);
10645 void vec_dstst (const unsigned short *, int, const int);
10646 void vec_dstst (const short *, int, const int);
10647 void vec_dstst (const unsigned int *, int, const int);
10648 void vec_dstst (const int *, int, const int);
10649 void vec_dstst (const unsigned long *, int, const int);
10650 void vec_dstst (const long *, int, const int);
10651 void vec_dstst (const float *, int, const int);
10653 void vec_dststt (const vector unsigned char *, int, const int);
10654 void vec_dststt (const vector signed char *, int, const int);
10655 void vec_dststt (const vector bool char *, int, const int);
10656 void vec_dststt (const vector unsigned short *, int, const int);
10657 void vec_dststt (const vector signed short *, int, const int);
10658 void vec_dststt (const vector bool short *, int, const int);
10659 void vec_dststt (const vector pixel *, int, const int);
10660 void vec_dststt (const vector unsigned int *, int, const int);
10661 void vec_dststt (const vector signed int *, int, const int);
10662 void vec_dststt (const vector bool int *, int, const int);
10663 void vec_dststt (const vector float *, int, const int);
10664 void vec_dststt (const unsigned char *, int, const int);
10665 void vec_dststt (const signed char *, int, const int);
10666 void vec_dststt (const unsigned short *, int, const int);
10667 void vec_dststt (const short *, int, const int);
10668 void vec_dststt (const unsigned int *, int, const int);
10669 void vec_dststt (const int *, int, const int);
10670 void vec_dststt (const unsigned long *, int, const int);
10671 void vec_dststt (const long *, int, const int);
10672 void vec_dststt (const float *, int, const int);
10674 void vec_dstt (const vector unsigned char *, int, const int);
10675 void vec_dstt (const vector signed char *, int, const int);
10676 void vec_dstt (const vector bool char *, int, const int);
10677 void vec_dstt (const vector unsigned short *, int, const int);
10678 void vec_dstt (const vector signed short *, int, const int);
10679 void vec_dstt (const vector bool short *, int, const int);
10680 void vec_dstt (const vector pixel *, int, const int);
10681 void vec_dstt (const vector unsigned int *, int, const int);
10682 void vec_dstt (const vector signed int *, int, const int);
10683 void vec_dstt (const vector bool int *, int, const int);
10684 void vec_dstt (const vector float *, int, const int);
10685 void vec_dstt (const unsigned char *, int, const int);
10686 void vec_dstt (const signed char *, int, const int);
10687 void vec_dstt (const unsigned short *, int, const int);
10688 void vec_dstt (const short *, int, const int);
10689 void vec_dstt (const unsigned int *, int, const int);
10690 void vec_dstt (const int *, int, const int);
10691 void vec_dstt (const unsigned long *, int, const int);
10692 void vec_dstt (const long *, int, const int);
10693 void vec_dstt (const float *, int, const int);
10695 vector float vec_expte (vector float);
10697 vector float vec_floor (vector float);
10699 vector float vec_ld (int, const vector float *);
10700 vector float vec_ld (int, const float *);
10701 vector bool int vec_ld (int, const vector bool int *);
10702 vector signed int vec_ld (int, const vector signed int *);
10703 vector signed int vec_ld (int, const int *);
10704 vector signed int vec_ld (int, const long *);
10705 vector unsigned int vec_ld (int, const vector unsigned int *);
10706 vector unsigned int vec_ld (int, const unsigned int *);
10707 vector unsigned int vec_ld (int, const unsigned long *);
10708 vector bool short vec_ld (int, const vector bool short *);
10709 vector pixel vec_ld (int, const vector pixel *);
10710 vector signed short vec_ld (int, const vector signed short *);
10711 vector signed short vec_ld (int, const short *);
10712 vector unsigned short vec_ld (int, const vector unsigned short *);
10713 vector unsigned short vec_ld (int, const unsigned short *);
10714 vector bool char vec_ld (int, const vector bool char *);
10715 vector signed char vec_ld (int, const vector signed char *);
10716 vector signed char vec_ld (int, const signed char *);
10717 vector unsigned char vec_ld (int, const vector unsigned char *);
10718 vector unsigned char vec_ld (int, const unsigned char *);
10720 vector signed char vec_lde (int, const signed char *);
10721 vector unsigned char vec_lde (int, const unsigned char *);
10722 vector signed short vec_lde (int, const short *);
10723 vector unsigned short vec_lde (int, const unsigned short *);
10724 vector float vec_lde (int, const float *);
10725 vector signed int vec_lde (int, const int *);
10726 vector unsigned int vec_lde (int, const unsigned int *);
10727 vector signed int vec_lde (int, const long *);
10728 vector unsigned int vec_lde (int, const unsigned long *);
10730 vector float vec_lvewx (int, float *);
10731 vector signed int vec_lvewx (int, int *);
10732 vector unsigned int vec_lvewx (int, unsigned int *);
10733 vector signed int vec_lvewx (int, long *);
10734 vector unsigned int vec_lvewx (int, unsigned long *);
10736 vector signed short vec_lvehx (int, short *);
10737 vector unsigned short vec_lvehx (int, unsigned short *);
10739 vector signed char vec_lvebx (int, char *);
10740 vector unsigned char vec_lvebx (int, unsigned char *);
10742 vector float vec_ldl (int, const vector float *);
10743 vector float vec_ldl (int, const float *);
10744 vector bool int vec_ldl (int, const vector bool int *);
10745 vector signed int vec_ldl (int, const vector signed int *);
10746 vector signed int vec_ldl (int, const int *);
10747 vector signed int vec_ldl (int, const long *);
10748 vector unsigned int vec_ldl (int, const vector unsigned int *);
10749 vector unsigned int vec_ldl (int, const unsigned int *);
10750 vector unsigned int vec_ldl (int, const unsigned long *);
10751 vector bool short vec_ldl (int, const vector bool short *);
10752 vector pixel vec_ldl (int, const vector pixel *);
10753 vector signed short vec_ldl (int, const vector signed short *);
10754 vector signed short vec_ldl (int, const short *);
10755 vector unsigned short vec_ldl (int, const vector unsigned short *);
10756 vector unsigned short vec_ldl (int, const unsigned short *);
10757 vector bool char vec_ldl (int, const vector bool char *);
10758 vector signed char vec_ldl (int, const vector signed char *);
10759 vector signed char vec_ldl (int, const signed char *);
10760 vector unsigned char vec_ldl (int, const vector unsigned char *);
10761 vector unsigned char vec_ldl (int, const unsigned char *);
10763 vector float vec_loge (vector float);
10765 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
10766 vector unsigned char vec_lvsl (int, const volatile signed char *);
10767 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
10768 vector unsigned char vec_lvsl (int, const volatile short *);
10769 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
10770 vector unsigned char vec_lvsl (int, const volatile int *);
10771 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
10772 vector unsigned char vec_lvsl (int, const volatile long *);
10773 vector unsigned char vec_lvsl (int, const volatile float *);
10775 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
10776 vector unsigned char vec_lvsr (int, const volatile signed char *);
10777 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
10778 vector unsigned char vec_lvsr (int, const volatile short *);
10779 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
10780 vector unsigned char vec_lvsr (int, const volatile int *);
10781 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
10782 vector unsigned char vec_lvsr (int, const volatile long *);
10783 vector unsigned char vec_lvsr (int, const volatile float *);
10785 vector float vec_madd (vector float, vector float, vector float);
10787 vector signed short vec_madds (vector signed short,
10788 vector signed short,
10789 vector signed short);
10791 vector unsigned char vec_max (vector bool char, vector unsigned char);
10792 vector unsigned char vec_max (vector unsigned char, vector bool char);
10793 vector unsigned char vec_max (vector unsigned char,
10794 vector unsigned char);
10795 vector signed char vec_max (vector bool char, vector signed char);
10796 vector signed char vec_max (vector signed char, vector bool char);
10797 vector signed char vec_max (vector signed char, vector signed char);
10798 vector unsigned short vec_max (vector bool short,
10799 vector unsigned short);
10800 vector unsigned short vec_max (vector unsigned short,
10801 vector bool short);
10802 vector unsigned short vec_max (vector unsigned short,
10803 vector unsigned short);
10804 vector signed short vec_max (vector bool short, vector signed short);
10805 vector signed short vec_max (vector signed short, vector bool short);
10806 vector signed short vec_max (vector signed short, vector signed short);
10807 vector unsigned int vec_max (vector bool int, vector unsigned int);
10808 vector unsigned int vec_max (vector unsigned int, vector bool int);
10809 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
10810 vector signed int vec_max (vector bool int, vector signed int);
10811 vector signed int vec_max (vector signed int, vector bool int);
10812 vector signed int vec_max (vector signed int, vector signed int);
10813 vector float vec_max (vector float, vector float);
10815 vector float vec_vmaxfp (vector float, vector float);
10817 vector signed int vec_vmaxsw (vector bool int, vector signed int);
10818 vector signed int vec_vmaxsw (vector signed int, vector bool int);
10819 vector signed int vec_vmaxsw (vector signed int, vector signed int);
10821 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
10822 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
10823 vector unsigned int vec_vmaxuw (vector unsigned int,
10824 vector unsigned int);
10826 vector signed short vec_vmaxsh (vector bool short, vector signed short);
10827 vector signed short vec_vmaxsh (vector signed short, vector bool short);
10828 vector signed short vec_vmaxsh (vector signed short,
10829 vector signed short);
10831 vector unsigned short vec_vmaxuh (vector bool short,
10832 vector unsigned short);
10833 vector unsigned short vec_vmaxuh (vector unsigned short,
10834 vector bool short);
10835 vector unsigned short vec_vmaxuh (vector unsigned short,
10836 vector unsigned short);
10838 vector signed char vec_vmaxsb (vector bool char, vector signed char);
10839 vector signed char vec_vmaxsb (vector signed char, vector bool char);
10840 vector signed char vec_vmaxsb (vector signed char, vector signed char);
10842 vector unsigned char vec_vmaxub (vector bool char,
10843 vector unsigned char);
10844 vector unsigned char vec_vmaxub (vector unsigned char,
10846 vector unsigned char vec_vmaxub (vector unsigned char,
10847 vector unsigned char);
10849 vector bool char vec_mergeh (vector bool char, vector bool char);
10850 vector signed char vec_mergeh (vector signed char, vector signed char);
10851 vector unsigned char vec_mergeh (vector unsigned char,
10852 vector unsigned char);
10853 vector bool short vec_mergeh (vector bool short, vector bool short);
10854 vector pixel vec_mergeh (vector pixel, vector pixel);
10855 vector signed short vec_mergeh (vector signed short,
10856 vector signed short);
10857 vector unsigned short vec_mergeh (vector unsigned short,
10858 vector unsigned short);
10859 vector float vec_mergeh (vector float, vector float);
10860 vector bool int vec_mergeh (vector bool int, vector bool int);
10861 vector signed int vec_mergeh (vector signed int, vector signed int);
10862 vector unsigned int vec_mergeh (vector unsigned int,
10863 vector unsigned int);
10865 vector float vec_vmrghw (vector float, vector float);
10866 vector bool int vec_vmrghw (vector bool int, vector bool int);
10867 vector signed int vec_vmrghw (vector signed int, vector signed int);
10868 vector unsigned int vec_vmrghw (vector unsigned int,
10869 vector unsigned int);
10871 vector bool short vec_vmrghh (vector bool short, vector bool short);
10872 vector signed short vec_vmrghh (vector signed short,
10873 vector signed short);
10874 vector unsigned short vec_vmrghh (vector unsigned short,
10875 vector unsigned short);
10876 vector pixel vec_vmrghh (vector pixel, vector pixel);
10878 vector bool char vec_vmrghb (vector bool char, vector bool char);
10879 vector signed char vec_vmrghb (vector signed char, vector signed char);
10880 vector unsigned char vec_vmrghb (vector unsigned char,
10881 vector unsigned char);
10883 vector bool char vec_mergel (vector bool char, vector bool char);
10884 vector signed char vec_mergel (vector signed char, vector signed char);
10885 vector unsigned char vec_mergel (vector unsigned char,
10886 vector unsigned char);
10887 vector bool short vec_mergel (vector bool short, vector bool short);
10888 vector pixel vec_mergel (vector pixel, vector pixel);
10889 vector signed short vec_mergel (vector signed short,
10890 vector signed short);
10891 vector unsigned short vec_mergel (vector unsigned short,
10892 vector unsigned short);
10893 vector float vec_mergel (vector float, vector float);
10894 vector bool int vec_mergel (vector bool int, vector bool int);
10895 vector signed int vec_mergel (vector signed int, vector signed int);
10896 vector unsigned int vec_mergel (vector unsigned int,
10897 vector unsigned int);
10899 vector float vec_vmrglw (vector float, vector float);
10900 vector signed int vec_vmrglw (vector signed int, vector signed int);
10901 vector unsigned int vec_vmrglw (vector unsigned int,
10902 vector unsigned int);
10903 vector bool int vec_vmrglw (vector bool int, vector bool int);
10905 vector bool short vec_vmrglh (vector bool short, vector bool short);
10906 vector signed short vec_vmrglh (vector signed short,
10907 vector signed short);
10908 vector unsigned short vec_vmrglh (vector unsigned short,
10909 vector unsigned short);
10910 vector pixel vec_vmrglh (vector pixel, vector pixel);
10912 vector bool char vec_vmrglb (vector bool char, vector bool char);
10913 vector signed char vec_vmrglb (vector signed char, vector signed char);
10914 vector unsigned char vec_vmrglb (vector unsigned char,
10915 vector unsigned char);
10917 vector unsigned short vec_mfvscr (void);
10919 vector unsigned char vec_min (vector bool char, vector unsigned char);
10920 vector unsigned char vec_min (vector unsigned char, vector bool char);
10921 vector unsigned char vec_min (vector unsigned char,
10922 vector unsigned char);
10923 vector signed char vec_min (vector bool char, vector signed char);
10924 vector signed char vec_min (vector signed char, vector bool char);
10925 vector signed char vec_min (vector signed char, vector signed char);
10926 vector unsigned short vec_min (vector bool short,
10927 vector unsigned short);
10928 vector unsigned short vec_min (vector unsigned short,
10929 vector bool short);
10930 vector unsigned short vec_min (vector unsigned short,
10931 vector unsigned short);
10932 vector signed short vec_min (vector bool short, vector signed short);
10933 vector signed short vec_min (vector signed short, vector bool short);
10934 vector signed short vec_min (vector signed short, vector signed short);
10935 vector unsigned int vec_min (vector bool int, vector unsigned int);
10936 vector unsigned int vec_min (vector unsigned int, vector bool int);
10937 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
10938 vector signed int vec_min (vector bool int, vector signed int);
10939 vector signed int vec_min (vector signed int, vector bool int);
10940 vector signed int vec_min (vector signed int, vector signed int);
10941 vector float vec_min (vector float, vector float);
10943 vector float vec_vminfp (vector float, vector float);
10945 vector signed int vec_vminsw (vector bool int, vector signed int);
10946 vector signed int vec_vminsw (vector signed int, vector bool int);
10947 vector signed int vec_vminsw (vector signed int, vector signed int);
10949 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
10950 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
10951 vector unsigned int vec_vminuw (vector unsigned int,
10952 vector unsigned int);
10954 vector signed short vec_vminsh (vector bool short, vector signed short);
10955 vector signed short vec_vminsh (vector signed short, vector bool short);
10956 vector signed short vec_vminsh (vector signed short,
10957 vector signed short);
10959 vector unsigned short vec_vminuh (vector bool short,
10960 vector unsigned short);
10961 vector unsigned short vec_vminuh (vector unsigned short,
10962 vector bool short);
10963 vector unsigned short vec_vminuh (vector unsigned short,
10964 vector unsigned short);
10966 vector signed char vec_vminsb (vector bool char, vector signed char);
10967 vector signed char vec_vminsb (vector signed char, vector bool char);
10968 vector signed char vec_vminsb (vector signed char, vector signed char);
10970 vector unsigned char vec_vminub (vector bool char,
10971 vector unsigned char);
10972 vector unsigned char vec_vminub (vector unsigned char,
10974 vector unsigned char vec_vminub (vector unsigned char,
10975 vector unsigned char);
10977 vector signed short vec_mladd (vector signed short,
10978 vector signed short,
10979 vector signed short);
10980 vector signed short vec_mladd (vector signed short,
10981 vector unsigned short,
10982 vector unsigned short);
10983 vector signed short vec_mladd (vector unsigned short,
10984 vector signed short,
10985 vector signed short);
10986 vector unsigned short vec_mladd (vector unsigned short,
10987 vector unsigned short,
10988 vector unsigned short);
10990 vector signed short vec_mradds (vector signed short,
10991 vector signed short,
10992 vector signed short);
10994 vector unsigned int vec_msum (vector unsigned char,
10995 vector unsigned char,
10996 vector unsigned int);
10997 vector signed int vec_msum (vector signed char,
10998 vector unsigned char,
10999 vector signed int);
11000 vector unsigned int vec_msum (vector unsigned short,
11001 vector unsigned short,
11002 vector unsigned int);
11003 vector signed int vec_msum (vector signed short,
11004 vector signed short,
11005 vector signed int);
11007 vector signed int vec_vmsumshm (vector signed short,
11008 vector signed short,
11009 vector signed int);
11011 vector unsigned int vec_vmsumuhm (vector unsigned short,
11012 vector unsigned short,
11013 vector unsigned int);
11015 vector signed int vec_vmsummbm (vector signed char,
11016 vector unsigned char,
11017 vector signed int);
11019 vector unsigned int vec_vmsumubm (vector unsigned char,
11020 vector unsigned char,
11021 vector unsigned int);
11023 vector unsigned int vec_msums (vector unsigned short,
11024 vector unsigned short,
11025 vector unsigned int);
11026 vector signed int vec_msums (vector signed short,
11027 vector signed short,
11028 vector signed int);
11030 vector signed int vec_vmsumshs (vector signed short,
11031 vector signed short,
11032 vector signed int);
11034 vector unsigned int vec_vmsumuhs (vector unsigned short,
11035 vector unsigned short,
11036 vector unsigned int);
11038 void vec_mtvscr (vector signed int);
11039 void vec_mtvscr (vector unsigned int);
11040 void vec_mtvscr (vector bool int);
11041 void vec_mtvscr (vector signed short);
11042 void vec_mtvscr (vector unsigned short);
11043 void vec_mtvscr (vector bool short);
11044 void vec_mtvscr (vector pixel);
11045 void vec_mtvscr (vector signed char);
11046 void vec_mtvscr (vector unsigned char);
11047 void vec_mtvscr (vector bool char);
11049 vector unsigned short vec_mule (vector unsigned char,
11050 vector unsigned char);
11051 vector signed short vec_mule (vector signed char,
11052 vector signed char);
11053 vector unsigned int vec_mule (vector unsigned short,
11054 vector unsigned short);
11055 vector signed int vec_mule (vector signed short, vector signed short);
11057 vector signed int vec_vmulesh (vector signed short,
11058 vector signed short);
11060 vector unsigned int vec_vmuleuh (vector unsigned short,
11061 vector unsigned short);
11063 vector signed short vec_vmulesb (vector signed char,
11064 vector signed char);
11066 vector unsigned short vec_vmuleub (vector unsigned char,
11067 vector unsigned char);
11069 vector unsigned short vec_mulo (vector unsigned char,
11070 vector unsigned char);
11071 vector signed short vec_mulo (vector signed char, vector signed char);
11072 vector unsigned int vec_mulo (vector unsigned short,
11073 vector unsigned short);
11074 vector signed int vec_mulo (vector signed short, vector signed short);
11076 vector signed int vec_vmulosh (vector signed short,
11077 vector signed short);
11079 vector unsigned int vec_vmulouh (vector unsigned short,
11080 vector unsigned short);
11082 vector signed short vec_vmulosb (vector signed char,
11083 vector signed char);
11085 vector unsigned short vec_vmuloub (vector unsigned char,
11086 vector unsigned char);
11088 vector float vec_nmsub (vector float, vector float, vector float);
11090 vector float vec_nor (vector float, vector float);
11091 vector signed int vec_nor (vector signed int, vector signed int);
11092 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
11093 vector bool int vec_nor (vector bool int, vector bool int);
11094 vector signed short vec_nor (vector signed short, vector signed short);
11095 vector unsigned short vec_nor (vector unsigned short,
11096 vector unsigned short);
11097 vector bool short vec_nor (vector bool short, vector bool short);
11098 vector signed char vec_nor (vector signed char, vector signed char);
11099 vector unsigned char vec_nor (vector unsigned char,
11100 vector unsigned char);
11101 vector bool char vec_nor (vector bool char, vector bool char);
11103 vector float vec_or (vector float, vector float);
11104 vector float vec_or (vector float, vector bool int);
11105 vector float vec_or (vector bool int, vector float);
11106 vector bool int vec_or (vector bool int, vector bool int);
11107 vector signed int vec_or (vector bool int, vector signed int);
11108 vector signed int vec_or (vector signed int, vector bool int);
11109 vector signed int vec_or (vector signed int, vector signed int);
11110 vector unsigned int vec_or (vector bool int, vector unsigned int);
11111 vector unsigned int vec_or (vector unsigned int, vector bool int);
11112 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
11113 vector bool short vec_or (vector bool short, vector bool short);
11114 vector signed short vec_or (vector bool short, vector signed short);
11115 vector signed short vec_or (vector signed short, vector bool short);
11116 vector signed short vec_or (vector signed short, vector signed short);
11117 vector unsigned short vec_or (vector bool short, vector unsigned short);
11118 vector unsigned short vec_or (vector unsigned short, vector bool short);
11119 vector unsigned short vec_or (vector unsigned short,
11120 vector unsigned short);
11121 vector signed char vec_or (vector bool char, vector signed char);
11122 vector bool char vec_or (vector bool char, vector bool char);
11123 vector signed char vec_or (vector signed char, vector bool char);
11124 vector signed char vec_or (vector signed char, vector signed char);
11125 vector unsigned char vec_or (vector bool char, vector unsigned char);
11126 vector unsigned char vec_or (vector unsigned char, vector bool char);
11127 vector unsigned char vec_or (vector unsigned char,
11128 vector unsigned char);
11130 vector signed char vec_pack (vector signed short, vector signed short);
11131 vector unsigned char vec_pack (vector unsigned short,
11132 vector unsigned short);
11133 vector bool char vec_pack (vector bool short, vector bool short);
11134 vector signed short vec_pack (vector signed int, vector signed int);
11135 vector unsigned short vec_pack (vector unsigned int,
11136 vector unsigned int);
11137 vector bool short vec_pack (vector bool int, vector bool int);
11139 vector bool short vec_vpkuwum (vector bool int, vector bool int);
11140 vector signed short vec_vpkuwum (vector signed int, vector signed int);
11141 vector unsigned short vec_vpkuwum (vector unsigned int,
11142 vector unsigned int);
11144 vector bool char vec_vpkuhum (vector bool short, vector bool short);
11145 vector signed char vec_vpkuhum (vector signed short,
11146 vector signed short);
11147 vector unsigned char vec_vpkuhum (vector unsigned short,
11148 vector unsigned short);
11150 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
11152 vector unsigned char vec_packs (vector unsigned short,
11153 vector unsigned short);
11154 vector signed char vec_packs (vector signed short, vector signed short);
11155 vector unsigned short vec_packs (vector unsigned int,
11156 vector unsigned int);
11157 vector signed short vec_packs (vector signed int, vector signed int);
11159 vector signed short vec_vpkswss (vector signed int, vector signed int);
11161 vector unsigned short vec_vpkuwus (vector unsigned int,
11162 vector unsigned int);
11164 vector signed char vec_vpkshss (vector signed short,
11165 vector signed short);
11167 vector unsigned char vec_vpkuhus (vector unsigned short,
11168 vector unsigned short);
11170 vector unsigned char vec_packsu (vector unsigned short,
11171 vector unsigned short);
11172 vector unsigned char vec_packsu (vector signed short,
11173 vector signed short);
11174 vector unsigned short vec_packsu (vector unsigned int,
11175 vector unsigned int);
11176 vector unsigned short vec_packsu (vector signed int, vector signed int);
11178 vector unsigned short vec_vpkswus (vector signed int,
11179 vector signed int);
11181 vector unsigned char vec_vpkshus (vector signed short,
11182 vector signed short);
11184 vector float vec_perm (vector float,
11186 vector unsigned char);
11187 vector signed int vec_perm (vector signed int,
11189 vector unsigned char);
11190 vector unsigned int vec_perm (vector unsigned int,
11191 vector unsigned int,
11192 vector unsigned char);
11193 vector bool int vec_perm (vector bool int,
11195 vector unsigned char);
11196 vector signed short vec_perm (vector signed short,
11197 vector signed short,
11198 vector unsigned char);
11199 vector unsigned short vec_perm (vector unsigned short,
11200 vector unsigned short,
11201 vector unsigned char);
11202 vector bool short vec_perm (vector bool short,
11204 vector unsigned char);
11205 vector pixel vec_perm (vector pixel,
11207 vector unsigned char);
11208 vector signed char vec_perm (vector signed char,
11209 vector signed char,
11210 vector unsigned char);
11211 vector unsigned char vec_perm (vector unsigned char,
11212 vector unsigned char,
11213 vector unsigned char);
11214 vector bool char vec_perm (vector bool char,
11216 vector unsigned char);
11218 vector float vec_re (vector float);
11220 vector signed char vec_rl (vector signed char,
11221 vector unsigned char);
11222 vector unsigned char vec_rl (vector unsigned char,
11223 vector unsigned char);
11224 vector signed short vec_rl (vector signed short, vector unsigned short);
11225 vector unsigned short vec_rl (vector unsigned short,
11226 vector unsigned short);
11227 vector signed int vec_rl (vector signed int, vector unsigned int);
11228 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
11230 vector signed int vec_vrlw (vector signed int, vector unsigned int);
11231 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
11233 vector signed short vec_vrlh (vector signed short,
11234 vector unsigned short);
11235 vector unsigned short vec_vrlh (vector unsigned short,
11236 vector unsigned short);
11238 vector signed char vec_vrlb (vector signed char, vector unsigned char);
11239 vector unsigned char vec_vrlb (vector unsigned char,
11240 vector unsigned char);
11242 vector float vec_round (vector float);
11244 vector float vec_recip (vector float, vector float);
11246 vector float vec_rsqrt (vector float);
11248 vector float vec_rsqrte (vector float);
11250 vector float vec_sel (vector float, vector float, vector bool int);
11251 vector float vec_sel (vector float, vector float, vector unsigned int);
11252 vector signed int vec_sel (vector signed int,
11255 vector signed int vec_sel (vector signed int,
11257 vector unsigned int);
11258 vector unsigned int vec_sel (vector unsigned int,
11259 vector unsigned int,
11261 vector unsigned int vec_sel (vector unsigned int,
11262 vector unsigned int,
11263 vector unsigned int);
11264 vector bool int vec_sel (vector bool int,
11267 vector bool int vec_sel (vector bool int,
11269 vector unsigned int);
11270 vector signed short vec_sel (vector signed short,
11271 vector signed short,
11272 vector bool short);
11273 vector signed short vec_sel (vector signed short,
11274 vector signed short,
11275 vector unsigned short);
11276 vector unsigned short vec_sel (vector unsigned short,
11277 vector unsigned short,
11278 vector bool short);
11279 vector unsigned short vec_sel (vector unsigned short,
11280 vector unsigned short,
11281 vector unsigned short);
11282 vector bool short vec_sel (vector bool short,
11284 vector bool short);
11285 vector bool short vec_sel (vector bool short,
11287 vector unsigned short);
11288 vector signed char vec_sel (vector signed char,
11289 vector signed char,
11291 vector signed char vec_sel (vector signed char,
11292 vector signed char,
11293 vector unsigned char);
11294 vector unsigned char vec_sel (vector unsigned char,
11295 vector unsigned char,
11297 vector unsigned char vec_sel (vector unsigned char,
11298 vector unsigned char,
11299 vector unsigned char);
11300 vector bool char vec_sel (vector bool char,
11303 vector bool char vec_sel (vector bool char,
11305 vector unsigned char);
11307 vector signed char vec_sl (vector signed char,
11308 vector unsigned char);
11309 vector unsigned char vec_sl (vector unsigned char,
11310 vector unsigned char);
11311 vector signed short vec_sl (vector signed short, vector unsigned short);
11312 vector unsigned short vec_sl (vector unsigned short,
11313 vector unsigned short);
11314 vector signed int vec_sl (vector signed int, vector unsigned int);
11315 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
11317 vector signed int vec_vslw (vector signed int, vector unsigned int);
11318 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
11320 vector signed short vec_vslh (vector signed short,
11321 vector unsigned short);
11322 vector unsigned short vec_vslh (vector unsigned short,
11323 vector unsigned short);
11325 vector signed char vec_vslb (vector signed char, vector unsigned char);
11326 vector unsigned char vec_vslb (vector unsigned char,
11327 vector unsigned char);
11329 vector float vec_sld (vector float, vector float, const int);
11330 vector signed int vec_sld (vector signed int,
11333 vector unsigned int vec_sld (vector unsigned int,
11334 vector unsigned int,
11336 vector bool int vec_sld (vector bool int,
11339 vector signed short vec_sld (vector signed short,
11340 vector signed short,
11342 vector unsigned short vec_sld (vector unsigned short,
11343 vector unsigned short,
11345 vector bool short vec_sld (vector bool short,
11348 vector pixel vec_sld (vector pixel,
11351 vector signed char vec_sld (vector signed char,
11352 vector signed char,
11354 vector unsigned char vec_sld (vector unsigned char,
11355 vector unsigned char,
11357 vector bool char vec_sld (vector bool char,
11361 vector signed int vec_sll (vector signed int,
11362 vector unsigned int);
11363 vector signed int vec_sll (vector signed int,
11364 vector unsigned short);
11365 vector signed int vec_sll (vector signed int,
11366 vector unsigned char);
11367 vector unsigned int vec_sll (vector unsigned int,
11368 vector unsigned int);
11369 vector unsigned int vec_sll (vector unsigned int,
11370 vector unsigned short);
11371 vector unsigned int vec_sll (vector unsigned int,
11372 vector unsigned char);
11373 vector bool int vec_sll (vector bool int,
11374 vector unsigned int);
11375 vector bool int vec_sll (vector bool int,
11376 vector unsigned short);
11377 vector bool int vec_sll (vector bool int,
11378 vector unsigned char);
11379 vector signed short vec_sll (vector signed short,
11380 vector unsigned int);
11381 vector signed short vec_sll (vector signed short,
11382 vector unsigned short);
11383 vector signed short vec_sll (vector signed short,
11384 vector unsigned char);
11385 vector unsigned short vec_sll (vector unsigned short,
11386 vector unsigned int);
11387 vector unsigned short vec_sll (vector unsigned short,
11388 vector unsigned short);
11389 vector unsigned short vec_sll (vector unsigned short,
11390 vector unsigned char);
11391 vector bool short vec_sll (vector bool short, vector unsigned int);
11392 vector bool short vec_sll (vector bool short, vector unsigned short);
11393 vector bool short vec_sll (vector bool short, vector unsigned char);
11394 vector pixel vec_sll (vector pixel, vector unsigned int);
11395 vector pixel vec_sll (vector pixel, vector unsigned short);
11396 vector pixel vec_sll (vector pixel, vector unsigned char);
11397 vector signed char vec_sll (vector signed char, vector unsigned int);
11398 vector signed char vec_sll (vector signed char, vector unsigned short);
11399 vector signed char vec_sll (vector signed char, vector unsigned char);
11400 vector unsigned char vec_sll (vector unsigned char,
11401 vector unsigned int);
11402 vector unsigned char vec_sll (vector unsigned char,
11403 vector unsigned short);
11404 vector unsigned char vec_sll (vector unsigned char,
11405 vector unsigned char);
11406 vector bool char vec_sll (vector bool char, vector unsigned int);
11407 vector bool char vec_sll (vector bool char, vector unsigned short);
11408 vector bool char vec_sll (vector bool char, vector unsigned char);
11410 vector float vec_slo (vector float, vector signed char);
11411 vector float vec_slo (vector float, vector unsigned char);
11412 vector signed int vec_slo (vector signed int, vector signed char);
11413 vector signed int vec_slo (vector signed int, vector unsigned char);
11414 vector unsigned int vec_slo (vector unsigned int, vector signed char);
11415 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
11416 vector signed short vec_slo (vector signed short, vector signed char);
11417 vector signed short vec_slo (vector signed short, vector unsigned char);
11418 vector unsigned short vec_slo (vector unsigned short,
11419 vector signed char);
11420 vector unsigned short vec_slo (vector unsigned short,
11421 vector unsigned char);
11422 vector pixel vec_slo (vector pixel, vector signed char);
11423 vector pixel vec_slo (vector pixel, vector unsigned char);
11424 vector signed char vec_slo (vector signed char, vector signed char);
11425 vector signed char vec_slo (vector signed char, vector unsigned char);
11426 vector unsigned char vec_slo (vector unsigned char, vector signed char);
11427 vector unsigned char vec_slo (vector unsigned char,
11428 vector unsigned char);
11430 vector signed char vec_splat (vector signed char, const int);
11431 vector unsigned char vec_splat (vector unsigned char, const int);
11432 vector bool char vec_splat (vector bool char, const int);
11433 vector signed short vec_splat (vector signed short, const int);
11434 vector unsigned short vec_splat (vector unsigned short, const int);
11435 vector bool short vec_splat (vector bool short, const int);
11436 vector pixel vec_splat (vector pixel, const int);
11437 vector float vec_splat (vector float, const int);
11438 vector signed int vec_splat (vector signed int, const int);
11439 vector unsigned int vec_splat (vector unsigned int, const int);
11440 vector bool int vec_splat (vector bool int, const int);
11442 vector float vec_vspltw (vector float, const int);
11443 vector signed int vec_vspltw (vector signed int, const int);
11444 vector unsigned int vec_vspltw (vector unsigned int, const int);
11445 vector bool int vec_vspltw (vector bool int, const int);
11447 vector bool short vec_vsplth (vector bool short, const int);
11448 vector signed short vec_vsplth (vector signed short, const int);
11449 vector unsigned short vec_vsplth (vector unsigned short, const int);
11450 vector pixel vec_vsplth (vector pixel, const int);
11452 vector signed char vec_vspltb (vector signed char, const int);
11453 vector unsigned char vec_vspltb (vector unsigned char, const int);
11454 vector bool char vec_vspltb (vector bool char, const int);
11456 vector signed char vec_splat_s8 (const int);
11458 vector signed short vec_splat_s16 (const int);
11460 vector signed int vec_splat_s32 (const int);
11462 vector unsigned char vec_splat_u8 (const int);
11464 vector unsigned short vec_splat_u16 (const int);
11466 vector unsigned int vec_splat_u32 (const int);
11468 vector signed char vec_sr (vector signed char, vector unsigned char);
11469 vector unsigned char vec_sr (vector unsigned char,
11470 vector unsigned char);
11471 vector signed short vec_sr (vector signed short,
11472 vector unsigned short);
11473 vector unsigned short vec_sr (vector unsigned short,
11474 vector unsigned short);
11475 vector signed int vec_sr (vector signed int, vector unsigned int);
11476 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
11478 vector signed int vec_vsrw (vector signed int, vector unsigned int);
11479 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
11481 vector signed short vec_vsrh (vector signed short,
11482 vector unsigned short);
11483 vector unsigned short vec_vsrh (vector unsigned short,
11484 vector unsigned short);
11486 vector signed char vec_vsrb (vector signed char, vector unsigned char);
11487 vector unsigned char vec_vsrb (vector unsigned char,
11488 vector unsigned char);
11490 vector signed char vec_sra (vector signed char, vector unsigned char);
11491 vector unsigned char vec_sra (vector unsigned char,
11492 vector unsigned char);
11493 vector signed short vec_sra (vector signed short,
11494 vector unsigned short);
11495 vector unsigned short vec_sra (vector unsigned short,
11496 vector unsigned short);
11497 vector signed int vec_sra (vector signed int, vector unsigned int);
11498 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
11500 vector signed int vec_vsraw (vector signed int, vector unsigned int);
11501 vector unsigned int vec_vsraw (vector unsigned int,
11502 vector unsigned int);
11504 vector signed short vec_vsrah (vector signed short,
11505 vector unsigned short);
11506 vector unsigned short vec_vsrah (vector unsigned short,
11507 vector unsigned short);
11509 vector signed char vec_vsrab (vector signed char, vector unsigned char);
11510 vector unsigned char vec_vsrab (vector unsigned char,
11511 vector unsigned char);
11513 vector signed int vec_srl (vector signed int, vector unsigned int);
11514 vector signed int vec_srl (vector signed int, vector unsigned short);
11515 vector signed int vec_srl (vector signed int, vector unsigned char);
11516 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
11517 vector unsigned int vec_srl (vector unsigned int,
11518 vector unsigned short);
11519 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
11520 vector bool int vec_srl (vector bool int, vector unsigned int);
11521 vector bool int vec_srl (vector bool int, vector unsigned short);
11522 vector bool int vec_srl (vector bool int, vector unsigned char);
11523 vector signed short vec_srl (vector signed short, vector unsigned int);
11524 vector signed short vec_srl (vector signed short,
11525 vector unsigned short);
11526 vector signed short vec_srl (vector signed short, vector unsigned char);
11527 vector unsigned short vec_srl (vector unsigned short,
11528 vector unsigned int);
11529 vector unsigned short vec_srl (vector unsigned short,
11530 vector unsigned short);
11531 vector unsigned short vec_srl (vector unsigned short,
11532 vector unsigned char);
11533 vector bool short vec_srl (vector bool short, vector unsigned int);
11534 vector bool short vec_srl (vector bool short, vector unsigned short);
11535 vector bool short vec_srl (vector bool short, vector unsigned char);
11536 vector pixel vec_srl (vector pixel, vector unsigned int);
11537 vector pixel vec_srl (vector pixel, vector unsigned short);
11538 vector pixel vec_srl (vector pixel, vector unsigned char);
11539 vector signed char vec_srl (vector signed char, vector unsigned int);
11540 vector signed char vec_srl (vector signed char, vector unsigned short);
11541 vector signed char vec_srl (vector signed char, vector unsigned char);
11542 vector unsigned char vec_srl (vector unsigned char,
11543 vector unsigned int);
11544 vector unsigned char vec_srl (vector unsigned char,
11545 vector unsigned short);
11546 vector unsigned char vec_srl (vector unsigned char,
11547 vector unsigned char);
11548 vector bool char vec_srl (vector bool char, vector unsigned int);
11549 vector bool char vec_srl (vector bool char, vector unsigned short);
11550 vector bool char vec_srl (vector bool char, vector unsigned char);
11552 vector float vec_sro (vector float, vector signed char);
11553 vector float vec_sro (vector float, vector unsigned char);
11554 vector signed int vec_sro (vector signed int, vector signed char);
11555 vector signed int vec_sro (vector signed int, vector unsigned char);
11556 vector unsigned int vec_sro (vector unsigned int, vector signed char);
11557 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
11558 vector signed short vec_sro (vector signed short, vector signed char);
11559 vector signed short vec_sro (vector signed short, vector unsigned char);
11560 vector unsigned short vec_sro (vector unsigned short,
11561 vector signed char);
11562 vector unsigned short vec_sro (vector unsigned short,
11563 vector unsigned char);
11564 vector pixel vec_sro (vector pixel, vector signed char);
11565 vector pixel vec_sro (vector pixel, vector unsigned char);
11566 vector signed char vec_sro (vector signed char, vector signed char);
11567 vector signed char vec_sro (vector signed char, vector unsigned char);
11568 vector unsigned char vec_sro (vector unsigned char, vector signed char);
11569 vector unsigned char vec_sro (vector unsigned char,
11570 vector unsigned char);
11572 void vec_st (vector float, int, vector float *);
11573 void vec_st (vector float, int, float *);
11574 void vec_st (vector signed int, int, vector signed int *);
11575 void vec_st (vector signed int, int, int *);
11576 void vec_st (vector unsigned int, int, vector unsigned int *);
11577 void vec_st (vector unsigned int, int, unsigned int *);
11578 void vec_st (vector bool int, int, vector bool int *);
11579 void vec_st (vector bool int, int, unsigned int *);
11580 void vec_st (vector bool int, int, int *);
11581 void vec_st (vector signed short, int, vector signed short *);
11582 void vec_st (vector signed short, int, short *);
11583 void vec_st (vector unsigned short, int, vector unsigned short *);
11584 void vec_st (vector unsigned short, int, unsigned short *);
11585 void vec_st (vector bool short, int, vector bool short *);
11586 void vec_st (vector bool short, int, unsigned short *);
11587 void vec_st (vector pixel, int, vector pixel *);
11588 void vec_st (vector pixel, int, unsigned short *);
11589 void vec_st (vector pixel, int, short *);
11590 void vec_st (vector bool short, int, short *);
11591 void vec_st (vector signed char, int, vector signed char *);
11592 void vec_st (vector signed char, int, signed char *);
11593 void vec_st (vector unsigned char, int, vector unsigned char *);
11594 void vec_st (vector unsigned char, int, unsigned char *);
11595 void vec_st (vector bool char, int, vector bool char *);
11596 void vec_st (vector bool char, int, unsigned char *);
11597 void vec_st (vector bool char, int, signed char *);
11599 void vec_ste (vector signed char, int, signed char *);
11600 void vec_ste (vector unsigned char, int, unsigned char *);
11601 void vec_ste (vector bool char, int, signed char *);
11602 void vec_ste (vector bool char, int, unsigned char *);
11603 void vec_ste (vector signed short, int, short *);
11604 void vec_ste (vector unsigned short, int, unsigned short *);
11605 void vec_ste (vector bool short, int, short *);
11606 void vec_ste (vector bool short, int, unsigned short *);
11607 void vec_ste (vector pixel, int, short *);
11608 void vec_ste (vector pixel, int, unsigned short *);
11609 void vec_ste (vector float, int, float *);
11610 void vec_ste (vector signed int, int, int *);
11611 void vec_ste (vector unsigned int, int, unsigned int *);
11612 void vec_ste (vector bool int, int, int *);
11613 void vec_ste (vector bool int, int, unsigned int *);
11615 void vec_stvewx (vector float, int, float *);
11616 void vec_stvewx (vector signed int, int, int *);
11617 void vec_stvewx (vector unsigned int, int, unsigned int *);
11618 void vec_stvewx (vector bool int, int, int *);
11619 void vec_stvewx (vector bool int, int, unsigned int *);
11621 void vec_stvehx (vector signed short, int, short *);
11622 void vec_stvehx (vector unsigned short, int, unsigned short *);
11623 void vec_stvehx (vector bool short, int, short *);
11624 void vec_stvehx (vector bool short, int, unsigned short *);
11625 void vec_stvehx (vector pixel, int, short *);
11626 void vec_stvehx (vector pixel, int, unsigned short *);
11628 void vec_stvebx (vector signed char, int, signed char *);
11629 void vec_stvebx (vector unsigned char, int, unsigned char *);
11630 void vec_stvebx (vector bool char, int, signed char *);
11631 void vec_stvebx (vector bool char, int, unsigned char *);
11633 void vec_stl (vector float, int, vector float *);
11634 void vec_stl (vector float, int, float *);
11635 void vec_stl (vector signed int, int, vector signed int *);
11636 void vec_stl (vector signed int, int, int *);
11637 void vec_stl (vector unsigned int, int, vector unsigned int *);
11638 void vec_stl (vector unsigned int, int, unsigned int *);
11639 void vec_stl (vector bool int, int, vector bool int *);
11640 void vec_stl (vector bool int, int, unsigned int *);
11641 void vec_stl (vector bool int, int, int *);
11642 void vec_stl (vector signed short, int, vector signed short *);
11643 void vec_stl (vector signed short, int, short *);
11644 void vec_stl (vector unsigned short, int, vector unsigned short *);
11645 void vec_stl (vector unsigned short, int, unsigned short *);
11646 void vec_stl (vector bool short, int, vector bool short *);
11647 void vec_stl (vector bool short, int, unsigned short *);
11648 void vec_stl (vector bool short, int, short *);
11649 void vec_stl (vector pixel, int, vector pixel *);
11650 void vec_stl (vector pixel, int, unsigned short *);
11651 void vec_stl (vector pixel, int, short *);
11652 void vec_stl (vector signed char, int, vector signed char *);
11653 void vec_stl (vector signed char, int, signed char *);
11654 void vec_stl (vector unsigned char, int, vector unsigned char *);
11655 void vec_stl (vector unsigned char, int, unsigned char *);
11656 void vec_stl (vector bool char, int, vector bool char *);
11657 void vec_stl (vector bool char, int, unsigned char *);
11658 void vec_stl (vector bool char, int, signed char *);
11660 vector signed char vec_sub (vector bool char, vector signed char);
11661 vector signed char vec_sub (vector signed char, vector bool char);
11662 vector signed char vec_sub (vector signed char, vector signed char);
11663 vector unsigned char vec_sub (vector bool char, vector unsigned char);
11664 vector unsigned char vec_sub (vector unsigned char, vector bool char);
11665 vector unsigned char vec_sub (vector unsigned char,
11666 vector unsigned char);
11667 vector signed short vec_sub (vector bool short, vector signed short);
11668 vector signed short vec_sub (vector signed short, vector bool short);
11669 vector signed short vec_sub (vector signed short, vector signed short);
11670 vector unsigned short vec_sub (vector bool short,
11671 vector unsigned short);
11672 vector unsigned short vec_sub (vector unsigned short,
11673 vector bool short);
11674 vector unsigned short vec_sub (vector unsigned short,
11675 vector unsigned short);
11676 vector signed int vec_sub (vector bool int, vector signed int);
11677 vector signed int vec_sub (vector signed int, vector bool int);
11678 vector signed int vec_sub (vector signed int, vector signed int);
11679 vector unsigned int vec_sub (vector bool int, vector unsigned int);
11680 vector unsigned int vec_sub (vector unsigned int, vector bool int);
11681 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
11682 vector float vec_sub (vector float, vector float);
11684 vector float vec_vsubfp (vector float, vector float);
11686 vector signed int vec_vsubuwm (vector bool int, vector signed int);
11687 vector signed int vec_vsubuwm (vector signed int, vector bool int);
11688 vector signed int vec_vsubuwm (vector signed int, vector signed int);
11689 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
11690 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
11691 vector unsigned int vec_vsubuwm (vector unsigned int,
11692 vector unsigned int);
11694 vector signed short vec_vsubuhm (vector bool short,
11695 vector signed short);
11696 vector signed short vec_vsubuhm (vector signed short,
11697 vector bool short);
11698 vector signed short vec_vsubuhm (vector signed short,
11699 vector signed short);
11700 vector unsigned short vec_vsubuhm (vector bool short,
11701 vector unsigned short);
11702 vector unsigned short vec_vsubuhm (vector unsigned short,
11703 vector bool short);
11704 vector unsigned short vec_vsubuhm (vector unsigned short,
11705 vector unsigned short);
11707 vector signed char vec_vsububm (vector bool char, vector signed char);
11708 vector signed char vec_vsububm (vector signed char, vector bool char);
11709 vector signed char vec_vsububm (vector signed char, vector signed char);
11710 vector unsigned char vec_vsububm (vector bool char,
11711 vector unsigned char);
11712 vector unsigned char vec_vsububm (vector unsigned char,
11714 vector unsigned char vec_vsububm (vector unsigned char,
11715 vector unsigned char);
11717 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
11719 vector unsigned char vec_subs (vector bool char, vector unsigned char);
11720 vector unsigned char vec_subs (vector unsigned char, vector bool char);
11721 vector unsigned char vec_subs (vector unsigned char,
11722 vector unsigned char);
11723 vector signed char vec_subs (vector bool char, vector signed char);
11724 vector signed char vec_subs (vector signed char, vector bool char);
11725 vector signed char vec_subs (vector signed char, vector signed char);
11726 vector unsigned short vec_subs (vector bool short,
11727 vector unsigned short);
11728 vector unsigned short vec_subs (vector unsigned short,
11729 vector bool short);
11730 vector unsigned short vec_subs (vector unsigned short,
11731 vector unsigned short);
11732 vector signed short vec_subs (vector bool short, vector signed short);
11733 vector signed short vec_subs (vector signed short, vector bool short);
11734 vector signed short vec_subs (vector signed short, vector signed short);
11735 vector unsigned int vec_subs (vector bool int, vector unsigned int);
11736 vector unsigned int vec_subs (vector unsigned int, vector bool int);
11737 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
11738 vector signed int vec_subs (vector bool int, vector signed int);
11739 vector signed int vec_subs (vector signed int, vector bool int);
11740 vector signed int vec_subs (vector signed int, vector signed int);
11742 vector signed int vec_vsubsws (vector bool int, vector signed int);
11743 vector signed int vec_vsubsws (vector signed int, vector bool int);
11744 vector signed int vec_vsubsws (vector signed int, vector signed int);
11746 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
11747 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
11748 vector unsigned int vec_vsubuws (vector unsigned int,
11749 vector unsigned int);
11751 vector signed short vec_vsubshs (vector bool short,
11752 vector signed short);
11753 vector signed short vec_vsubshs (vector signed short,
11754 vector bool short);
11755 vector signed short vec_vsubshs (vector signed short,
11756 vector signed short);
11758 vector unsigned short vec_vsubuhs (vector bool short,
11759 vector unsigned short);
11760 vector unsigned short vec_vsubuhs (vector unsigned short,
11761 vector bool short);
11762 vector unsigned short vec_vsubuhs (vector unsigned short,
11763 vector unsigned short);
11765 vector signed char vec_vsubsbs (vector bool char, vector signed char);
11766 vector signed char vec_vsubsbs (vector signed char, vector bool char);
11767 vector signed char vec_vsubsbs (vector signed char, vector signed char);
11769 vector unsigned char vec_vsububs (vector bool char,
11770 vector unsigned char);
11771 vector unsigned char vec_vsububs (vector unsigned char,
11773 vector unsigned char vec_vsububs (vector unsigned char,
11774 vector unsigned char);
11776 vector unsigned int vec_sum4s (vector unsigned char,
11777 vector unsigned int);
11778 vector signed int vec_sum4s (vector signed char, vector signed int);
11779 vector signed int vec_sum4s (vector signed short, vector signed int);
11781 vector signed int vec_vsum4shs (vector signed short, vector signed int);
11783 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
11785 vector unsigned int vec_vsum4ubs (vector unsigned char,
11786 vector unsigned int);
11788 vector signed int vec_sum2s (vector signed int, vector signed int);
11790 vector signed int vec_sums (vector signed int, vector signed int);
11792 vector float vec_trunc (vector float);
11794 vector signed short vec_unpackh (vector signed char);
11795 vector bool short vec_unpackh (vector bool char);
11796 vector signed int vec_unpackh (vector signed short);
11797 vector bool int vec_unpackh (vector bool short);
11798 vector unsigned int vec_unpackh (vector pixel);
11800 vector bool int vec_vupkhsh (vector bool short);
11801 vector signed int vec_vupkhsh (vector signed short);
11803 vector unsigned int vec_vupkhpx (vector pixel);
11805 vector bool short vec_vupkhsb (vector bool char);
11806 vector signed short vec_vupkhsb (vector signed char);
11808 vector signed short vec_unpackl (vector signed char);
11809 vector bool short vec_unpackl (vector bool char);
11810 vector unsigned int vec_unpackl (vector pixel);
11811 vector signed int vec_unpackl (vector signed short);
11812 vector bool int vec_unpackl (vector bool short);
11814 vector unsigned int vec_vupklpx (vector pixel);
11816 vector bool int vec_vupklsh (vector bool short);
11817 vector signed int vec_vupklsh (vector signed short);
11819 vector bool short vec_vupklsb (vector bool char);
11820 vector signed short vec_vupklsb (vector signed char);
11822 vector float vec_xor (vector float, vector float);
11823 vector float vec_xor (vector float, vector bool int);
11824 vector float vec_xor (vector bool int, vector float);
11825 vector bool int vec_xor (vector bool int, vector bool int);
11826 vector signed int vec_xor (vector bool int, vector signed int);
11827 vector signed int vec_xor (vector signed int, vector bool int);
11828 vector signed int vec_xor (vector signed int, vector signed int);
11829 vector unsigned int vec_xor (vector bool int, vector unsigned int);
11830 vector unsigned int vec_xor (vector unsigned int, vector bool int);
11831 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
11832 vector bool short vec_xor (vector bool short, vector bool short);
11833 vector signed short vec_xor (vector bool short, vector signed short);
11834 vector signed short vec_xor (vector signed short, vector bool short);
11835 vector signed short vec_xor (vector signed short, vector signed short);
11836 vector unsigned short vec_xor (vector bool short,
11837 vector unsigned short);
11838 vector unsigned short vec_xor (vector unsigned short,
11839 vector bool short);
11840 vector unsigned short vec_xor (vector unsigned short,
11841 vector unsigned short);
11842 vector signed char vec_xor (vector bool char, vector signed char);
11843 vector bool char vec_xor (vector bool char, vector bool char);
11844 vector signed char vec_xor (vector signed char, vector bool char);
11845 vector signed char vec_xor (vector signed char, vector signed char);
11846 vector unsigned char vec_xor (vector bool char, vector unsigned char);
11847 vector unsigned char vec_xor (vector unsigned char, vector bool char);
11848 vector unsigned char vec_xor (vector unsigned char,
11849 vector unsigned char);
11851 int vec_all_eq (vector signed char, vector bool char);
11852 int vec_all_eq (vector signed char, vector signed char);
11853 int vec_all_eq (vector unsigned char, vector bool char);
11854 int vec_all_eq (vector unsigned char, vector unsigned char);
11855 int vec_all_eq (vector bool char, vector bool char);
11856 int vec_all_eq (vector bool char, vector unsigned char);
11857 int vec_all_eq (vector bool char, vector signed char);
11858 int vec_all_eq (vector signed short, vector bool short);
11859 int vec_all_eq (vector signed short, vector signed short);
11860 int vec_all_eq (vector unsigned short, vector bool short);
11861 int vec_all_eq (vector unsigned short, vector unsigned short);
11862 int vec_all_eq (vector bool short, vector bool short);
11863 int vec_all_eq (vector bool short, vector unsigned short);
11864 int vec_all_eq (vector bool short, vector signed short);
11865 int vec_all_eq (vector pixel, vector pixel);
11866 int vec_all_eq (vector signed int, vector bool int);
11867 int vec_all_eq (vector signed int, vector signed int);
11868 int vec_all_eq (vector unsigned int, vector bool int);
11869 int vec_all_eq (vector unsigned int, vector unsigned int);
11870 int vec_all_eq (vector bool int, vector bool int);
11871 int vec_all_eq (vector bool int, vector unsigned int);
11872 int vec_all_eq (vector bool int, vector signed int);
11873 int vec_all_eq (vector float, vector float);
11875 int vec_all_ge (vector bool char, vector unsigned char);
11876 int vec_all_ge (vector unsigned char, vector bool char);
11877 int vec_all_ge (vector unsigned char, vector unsigned char);
11878 int vec_all_ge (vector bool char, vector signed char);
11879 int vec_all_ge (vector signed char, vector bool char);
11880 int vec_all_ge (vector signed char, vector signed char);
11881 int vec_all_ge (vector bool short, vector unsigned short);
11882 int vec_all_ge (vector unsigned short, vector bool short);
11883 int vec_all_ge (vector unsigned short, vector unsigned short);
11884 int vec_all_ge (vector signed short, vector signed short);
11885 int vec_all_ge (vector bool short, vector signed short);
11886 int vec_all_ge (vector signed short, vector bool short);
11887 int vec_all_ge (vector bool int, vector unsigned int);
11888 int vec_all_ge (vector unsigned int, vector bool int);
11889 int vec_all_ge (vector unsigned int, vector unsigned int);
11890 int vec_all_ge (vector bool int, vector signed int);
11891 int vec_all_ge (vector signed int, vector bool int);
11892 int vec_all_ge (vector signed int, vector signed int);
11893 int vec_all_ge (vector float, vector float);
11895 int vec_all_gt (vector bool char, vector unsigned char);
11896 int vec_all_gt (vector unsigned char, vector bool char);
11897 int vec_all_gt (vector unsigned char, vector unsigned char);
11898 int vec_all_gt (vector bool char, vector signed char);
11899 int vec_all_gt (vector signed char, vector bool char);
11900 int vec_all_gt (vector signed char, vector signed char);
11901 int vec_all_gt (vector bool short, vector unsigned short);
11902 int vec_all_gt (vector unsigned short, vector bool short);
11903 int vec_all_gt (vector unsigned short, vector unsigned short);
11904 int vec_all_gt (vector bool short, vector signed short);
11905 int vec_all_gt (vector signed short, vector bool short);
11906 int vec_all_gt (vector signed short, vector signed short);
11907 int vec_all_gt (vector bool int, vector unsigned int);
11908 int vec_all_gt (vector unsigned int, vector bool int);
11909 int vec_all_gt (vector unsigned int, vector unsigned int);
11910 int vec_all_gt (vector bool int, vector signed int);
11911 int vec_all_gt (vector signed int, vector bool int);
11912 int vec_all_gt (vector signed int, vector signed int);
11913 int vec_all_gt (vector float, vector float);
11915 int vec_all_in (vector float, vector float);
11917 int vec_all_le (vector bool char, vector unsigned char);
11918 int vec_all_le (vector unsigned char, vector bool char);
11919 int vec_all_le (vector unsigned char, vector unsigned char);
11920 int vec_all_le (vector bool char, vector signed char);
11921 int vec_all_le (vector signed char, vector bool char);
11922 int vec_all_le (vector signed char, vector signed char);
11923 int vec_all_le (vector bool short, vector unsigned short);
11924 int vec_all_le (vector unsigned short, vector bool short);
11925 int vec_all_le (vector unsigned short, vector unsigned short);
11926 int vec_all_le (vector bool short, vector signed short);
11927 int vec_all_le (vector signed short, vector bool short);
11928 int vec_all_le (vector signed short, vector signed short);
11929 int vec_all_le (vector bool int, vector unsigned int);
11930 int vec_all_le (vector unsigned int, vector bool int);
11931 int vec_all_le (vector unsigned int, vector unsigned int);
11932 int vec_all_le (vector bool int, vector signed int);
11933 int vec_all_le (vector signed int, vector bool int);
11934 int vec_all_le (vector signed int, vector signed int);
11935 int vec_all_le (vector float, vector float);
11937 int vec_all_lt (vector bool char, vector unsigned char);
11938 int vec_all_lt (vector unsigned char, vector bool char);
11939 int vec_all_lt (vector unsigned char, vector unsigned char);
11940 int vec_all_lt (vector bool char, vector signed char);
11941 int vec_all_lt (vector signed char, vector bool char);
11942 int vec_all_lt (vector signed char, vector signed char);
11943 int vec_all_lt (vector bool short, vector unsigned short);
11944 int vec_all_lt (vector unsigned short, vector bool short);
11945 int vec_all_lt (vector unsigned short, vector unsigned short);
11946 int vec_all_lt (vector bool short, vector signed short);
11947 int vec_all_lt (vector signed short, vector bool short);
11948 int vec_all_lt (vector signed short, vector signed short);
11949 int vec_all_lt (vector bool int, vector unsigned int);
11950 int vec_all_lt (vector unsigned int, vector bool int);
11951 int vec_all_lt (vector unsigned int, vector unsigned int);
11952 int vec_all_lt (vector bool int, vector signed int);
11953 int vec_all_lt (vector signed int, vector bool int);
11954 int vec_all_lt (vector signed int, vector signed int);
11955 int vec_all_lt (vector float, vector float);
11957 int vec_all_nan (vector float);
11959 int vec_all_ne (vector signed char, vector bool char);
11960 int vec_all_ne (vector signed char, vector signed char);
11961 int vec_all_ne (vector unsigned char, vector bool char);
11962 int vec_all_ne (vector unsigned char, vector unsigned char);
11963 int vec_all_ne (vector bool char, vector bool char);
11964 int vec_all_ne (vector bool char, vector unsigned char);
11965 int vec_all_ne (vector bool char, vector signed char);
11966 int vec_all_ne (vector signed short, vector bool short);
11967 int vec_all_ne (vector signed short, vector signed short);
11968 int vec_all_ne (vector unsigned short, vector bool short);
11969 int vec_all_ne (vector unsigned short, vector unsigned short);
11970 int vec_all_ne (vector bool short, vector bool short);
11971 int vec_all_ne (vector bool short, vector unsigned short);
11972 int vec_all_ne (vector bool short, vector signed short);
11973 int vec_all_ne (vector pixel, vector pixel);
11974 int vec_all_ne (vector signed int, vector bool int);
11975 int vec_all_ne (vector signed int, vector signed int);
11976 int vec_all_ne (vector unsigned int, vector bool int);
11977 int vec_all_ne (vector unsigned int, vector unsigned int);
11978 int vec_all_ne (vector bool int, vector bool int);
11979 int vec_all_ne (vector bool int, vector unsigned int);
11980 int vec_all_ne (vector bool int, vector signed int);
11981 int vec_all_ne (vector float, vector float);
11983 int vec_all_nge (vector float, vector float);
11985 int vec_all_ngt (vector float, vector float);
11987 int vec_all_nle (vector float, vector float);
11989 int vec_all_nlt (vector float, vector float);
11991 int vec_all_numeric (vector float);
11993 int vec_any_eq (vector signed char, vector bool char);
11994 int vec_any_eq (vector signed char, vector signed char);
11995 int vec_any_eq (vector unsigned char, vector bool char);
11996 int vec_any_eq (vector unsigned char, vector unsigned char);
11997 int vec_any_eq (vector bool char, vector bool char);
11998 int vec_any_eq (vector bool char, vector unsigned char);
11999 int vec_any_eq (vector bool char, vector signed char);
12000 int vec_any_eq (vector signed short, vector bool short);
12001 int vec_any_eq (vector signed short, vector signed short);
12002 int vec_any_eq (vector unsigned short, vector bool short);
12003 int vec_any_eq (vector unsigned short, vector unsigned short);
12004 int vec_any_eq (vector bool short, vector bool short);
12005 int vec_any_eq (vector bool short, vector unsigned short);
12006 int vec_any_eq (vector bool short, vector signed short);
12007 int vec_any_eq (vector pixel, vector pixel);
12008 int vec_any_eq (vector signed int, vector bool int);
12009 int vec_any_eq (vector signed int, vector signed int);
12010 int vec_any_eq (vector unsigned int, vector bool int);
12011 int vec_any_eq (vector unsigned int, vector unsigned int);
12012 int vec_any_eq (vector bool int, vector bool int);
12013 int vec_any_eq (vector bool int, vector unsigned int);
12014 int vec_any_eq (vector bool int, vector signed int);
12015 int vec_any_eq (vector float, vector float);
12017 int vec_any_ge (vector signed char, vector bool char);
12018 int vec_any_ge (vector unsigned char, vector bool char);
12019 int vec_any_ge (vector unsigned char, vector unsigned char);
12020 int vec_any_ge (vector signed char, vector signed char);
12021 int vec_any_ge (vector bool char, vector unsigned char);
12022 int vec_any_ge (vector bool char, vector signed char);
12023 int vec_any_ge (vector unsigned short, vector bool short);
12024 int vec_any_ge (vector unsigned short, vector unsigned short);
12025 int vec_any_ge (vector signed short, vector signed short);
12026 int vec_any_ge (vector signed short, vector bool short);
12027 int vec_any_ge (vector bool short, vector unsigned short);
12028 int vec_any_ge (vector bool short, vector signed short);
12029 int vec_any_ge (vector signed int, vector bool int);
12030 int vec_any_ge (vector unsigned int, vector bool int);
12031 int vec_any_ge (vector unsigned int, vector unsigned int);
12032 int vec_any_ge (vector signed int, vector signed int);
12033 int vec_any_ge (vector bool int, vector unsigned int);
12034 int vec_any_ge (vector bool int, vector signed int);
12035 int vec_any_ge (vector float, vector float);
12037 int vec_any_gt (vector bool char, vector unsigned char);
12038 int vec_any_gt (vector unsigned char, vector bool char);
12039 int vec_any_gt (vector unsigned char, vector unsigned char);
12040 int vec_any_gt (vector bool char, vector signed char);
12041 int vec_any_gt (vector signed char, vector bool char);
12042 int vec_any_gt (vector signed char, vector signed char);
12043 int vec_any_gt (vector bool short, vector unsigned short);
12044 int vec_any_gt (vector unsigned short, vector bool short);
12045 int vec_any_gt (vector unsigned short, vector unsigned short);
12046 int vec_any_gt (vector bool short, vector signed short);
12047 int vec_any_gt (vector signed short, vector bool short);
12048 int vec_any_gt (vector signed short, vector signed short);
12049 int vec_any_gt (vector bool int, vector unsigned int);
12050 int vec_any_gt (vector unsigned int, vector bool int);
12051 int vec_any_gt (vector unsigned int, vector unsigned int);
12052 int vec_any_gt (vector bool int, vector signed int);
12053 int vec_any_gt (vector signed int, vector bool int);
12054 int vec_any_gt (vector signed int, vector signed int);
12055 int vec_any_gt (vector float, vector float);
12057 int vec_any_le (vector bool char, vector unsigned char);
12058 int vec_any_le (vector unsigned char, vector bool char);
12059 int vec_any_le (vector unsigned char, vector unsigned char);
12060 int vec_any_le (vector bool char, vector signed char);
12061 int vec_any_le (vector signed char, vector bool char);
12062 int vec_any_le (vector signed char, vector signed char);
12063 int vec_any_le (vector bool short, vector unsigned short);
12064 int vec_any_le (vector unsigned short, vector bool short);
12065 int vec_any_le (vector unsigned short, vector unsigned short);
12066 int vec_any_le (vector bool short, vector signed short);
12067 int vec_any_le (vector signed short, vector bool short);
12068 int vec_any_le (vector signed short, vector signed short);
12069 int vec_any_le (vector bool int, vector unsigned int);
12070 int vec_any_le (vector unsigned int, vector bool int);
12071 int vec_any_le (vector unsigned int, vector unsigned int);
12072 int vec_any_le (vector bool int, vector signed int);
12073 int vec_any_le (vector signed int, vector bool int);
12074 int vec_any_le (vector signed int, vector signed int);
12075 int vec_any_le (vector float, vector float);
12077 int vec_any_lt (vector bool char, vector unsigned char);
12078 int vec_any_lt (vector unsigned char, vector bool char);
12079 int vec_any_lt (vector unsigned char, vector unsigned char);
12080 int vec_any_lt (vector bool char, vector signed char);
12081 int vec_any_lt (vector signed char, vector bool char);
12082 int vec_any_lt (vector signed char, vector signed char);
12083 int vec_any_lt (vector bool short, vector unsigned short);
12084 int vec_any_lt (vector unsigned short, vector bool short);
12085 int vec_any_lt (vector unsigned short, vector unsigned short);
12086 int vec_any_lt (vector bool short, vector signed short);
12087 int vec_any_lt (vector signed short, vector bool short);
12088 int vec_any_lt (vector signed short, vector signed short);
12089 int vec_any_lt (vector bool int, vector unsigned int);
12090 int vec_any_lt (vector unsigned int, vector bool int);
12091 int vec_any_lt (vector unsigned int, vector unsigned int);
12092 int vec_any_lt (vector bool int, vector signed int);
12093 int vec_any_lt (vector signed int, vector bool int);
12094 int vec_any_lt (vector signed int, vector signed int);
12095 int vec_any_lt (vector float, vector float);
12097 int vec_any_nan (vector float);
12099 int vec_any_ne (vector signed char, vector bool char);
12100 int vec_any_ne (vector signed char, vector signed char);
12101 int vec_any_ne (vector unsigned char, vector bool char);
12102 int vec_any_ne (vector unsigned char, vector unsigned char);
12103 int vec_any_ne (vector bool char, vector bool char);
12104 int vec_any_ne (vector bool char, vector unsigned char);
12105 int vec_any_ne (vector bool char, vector signed char);
12106 int vec_any_ne (vector signed short, vector bool short);
12107 int vec_any_ne (vector signed short, vector signed short);
12108 int vec_any_ne (vector unsigned short, vector bool short);
12109 int vec_any_ne (vector unsigned short, vector unsigned short);
12110 int vec_any_ne (vector bool short, vector bool short);
12111 int vec_any_ne (vector bool short, vector unsigned short);
12112 int vec_any_ne (vector bool short, vector signed short);
12113 int vec_any_ne (vector pixel, vector pixel);
12114 int vec_any_ne (vector signed int, vector bool int);
12115 int vec_any_ne (vector signed int, vector signed int);
12116 int vec_any_ne (vector unsigned int, vector bool int);
12117 int vec_any_ne (vector unsigned int, vector unsigned int);
12118 int vec_any_ne (vector bool int, vector bool int);
12119 int vec_any_ne (vector bool int, vector unsigned int);
12120 int vec_any_ne (vector bool int, vector signed int);
12121 int vec_any_ne (vector float, vector float);
12123 int vec_any_nge (vector float, vector float);
12125 int vec_any_ngt (vector float, vector float);
12127 int vec_any_nle (vector float, vector float);
12129 int vec_any_nlt (vector float, vector float);
12131 int vec_any_numeric (vector float);
12133 int vec_any_out (vector float, vector float);
12136 If the vector/scalar (VSX) instruction set is available, the following
12137 additional functions are available:
12140 vector double vec_abs (vector double);
12141 vector double vec_add (vector double, vector double);
12142 vector double vec_and (vector double, vector double);
12143 vector double vec_and (vector double, vector bool long);
12144 vector double vec_and (vector bool long, vector double);
12145 vector double vec_andc (vector double, vector double);
12146 vector double vec_andc (vector double, vector bool long);
12147 vector double vec_andc (vector bool long, vector double);
12148 vector double vec_ceil (vector double);
12149 vector bool long vec_cmpeq (vector double, vector double);
12150 vector bool long vec_cmpge (vector double, vector double);
12151 vector bool long vec_cmpgt (vector double, vector double);
12152 vector bool long vec_cmple (vector double, vector double);
12153 vector bool long vec_cmplt (vector double, vector double);
12154 vector float vec_div (vector float, vector float);
12155 vector double vec_div (vector double, vector double);
12156 vector double vec_floor (vector double);
12157 vector double vec_madd (vector double, vector double, vector double);
12158 vector double vec_max (vector double, vector double);
12159 vector double vec_min (vector double, vector double);
12160 vector float vec_msub (vector float, vector float, vector float);
12161 vector double vec_msub (vector double, vector double, vector double);
12162 vector float vec_mul (vector float, vector float);
12163 vector double vec_mul (vector double, vector double);
12164 vector float vec_nearbyint (vector float);
12165 vector double vec_nearbyint (vector double);
12166 vector float vec_nmadd (vector float, vector float, vector float);
12167 vector double vec_nmadd (vector double, vector double, vector double);
12168 vector double vec_nmsub (vector double, vector double, vector double);
12169 vector double vec_nor (vector double, vector double);
12170 vector double vec_or (vector double, vector double);
12171 vector double vec_or (vector double, vector bool long);
12172 vector double vec_or (vector bool long, vector double);
12173 vector double vec_perm (vector double,
12175 vector unsigned char);
12176 vector double vec_rint (vector double);
12177 vector double vec_recip (vector double, vector double);
12178 vector double vec_rsqrt (vector double);
12179 vector double vec_rsqrte (vector double);
12180 vector double vec_sel (vector double, vector double, vector bool long);
12181 vector double vec_sel (vector double, vector double, vector unsigned long);
12182 vector double vec_sub (vector double, vector double);
12183 vector float vec_sqrt (vector float);
12184 vector double vec_sqrt (vector double);
12185 vector double vec_trunc (vector double);
12186 vector double vec_xor (vector double, vector double);
12187 vector double vec_xor (vector double, vector bool long);
12188 vector double vec_xor (vector bool long, vector double);
12189 int vec_all_eq (vector double, vector double);
12190 int vec_all_ge (vector double, vector double);
12191 int vec_all_gt (vector double, vector double);
12192 int vec_all_le (vector double, vector double);
12193 int vec_all_lt (vector double, vector double);
12194 int vec_all_nan (vector double);
12195 int vec_all_ne (vector double, vector double);
12196 int vec_all_nge (vector double, vector double);
12197 int vec_all_ngt (vector double, vector double);
12198 int vec_all_nle (vector double, vector double);
12199 int vec_all_nlt (vector double, vector double);
12200 int vec_all_numeric (vector double);
12201 int vec_any_eq (vector double, vector double);
12202 int vec_any_ge (vector double, vector double);
12203 int vec_any_gt (vector double, vector double);
12204 int vec_any_le (vector double, vector double);
12205 int vec_any_lt (vector double, vector double);
12206 int vec_any_nan (vector double);
12207 int vec_any_ne (vector double, vector double);
12208 int vec_any_nge (vector double, vector double);
12209 int vec_any_ngt (vector double, vector double);
12210 int vec_any_nle (vector double, vector double);
12211 int vec_any_nlt (vector double, vector double);
12212 int vec_any_numeric (vector double);
12215 GCC provides a few other builtins on Powerpc to access certain instructions:
12217 float __builtin_recipdivf (float, float);
12218 float __builtin_rsqrtf (float);
12219 double __builtin_recipdiv (double, double);
12220 double __builtin_rsqrt (double);
12221 long __builtin_bpermd (long, long);
12222 int __builtin_bswap16 (int);
12225 The @code{vec_rsqrt}, @code{__builtin_rsqrt}, and
12226 @code{__builtin_rsqrtf} functions generate multiple instructions to
12227 implement the reciprocal sqrt functionality using reciprocal sqrt
12228 estimate instructions.
12230 The @code{__builtin_recipdiv}, and @code{__builtin_recipdivf}
12231 functions generate multiple instructions to implement division using
12232 the reciprocal estimate instructions.
12234 @node RX Built-in Functions
12235 @subsection RX Built-in Functions
12236 GCC supports some of the RX instructions which cannot be expressed in
12237 the C programming language via the use of built-in functions. The
12238 following functions are supported:
12240 @deftypefn {Built-in Function} void __builtin_rx_brk (void)
12241 Generates the @code{brk} machine instruction.
12244 @deftypefn {Built-in Function} void __builtin_rx_clrpsw (int)
12245 Generates the @code{clrpsw} machine instruction to clear the specified
12246 bit in the processor status word.
12249 @deftypefn {Built-in Function} void __builtin_rx_int (int)
12250 Generates the @code{int} machine instruction to generate an interrupt
12251 with the specified value.
12254 @deftypefn {Built-in Function} void __builtin_rx_machi (int, int)
12255 Generates the @code{machi} machine instruction to add the result of
12256 multiplying the top 16-bits of the two arguments into the
12260 @deftypefn {Built-in Function} void __builtin_rx_maclo (int, int)
12261 Generates the @code{maclo} machine instruction to add the result of
12262 multiplying the bottom 16-bits of the two arguments into the
12266 @deftypefn {Built-in Function} void __builtin_rx_mulhi (int, int)
12267 Generates the @code{mulhi} machine instruction to place the result of
12268 multiplying the top 16-bits of the two arguments into the
12272 @deftypefn {Built-in Function} void __builtin_rx_mullo (int, int)
12273 Generates the @code{mullo} machine instruction to place the result of
12274 multiplying the bottom 16-bits of the two arguments into the
12278 @deftypefn {Built-in Function} int __builtin_rx_mvfachi (void)
12279 Generates the @code{mvfachi} machine instruction to read the top
12280 32-bits of the accumulator.
12283 @deftypefn {Built-in Function} int __builtin_rx_mvfacmi (void)
12284 Generates the @code{mvfacmi} machine instruction to read the middle
12285 32-bits of the accumulator.
12288 @deftypefn {Built-in Function} int __builtin_rx_mvfc (int)
12289 Generates the @code{mvfc} machine instruction which reads the control
12290 register specified in its argument and returns its value.
12293 @deftypefn {Built-in Function} void __builtin_rx_mvtachi (int)
12294 Generates the @code{mvtachi} machine instruction to set the top
12295 32-bits of the accumulator.
12298 @deftypefn {Built-in Function} void __builtin_rx_mvtaclo (int)
12299 Generates the @code{mvtaclo} machine instruction to set the bottom
12300 32-bits of the accumulator.
12303 @deftypefn {Built-in Function} void __builtin_rx_mvtc (int reg, int val)
12304 Generates the @code{mvtc} machine instruction which sets control
12305 register number @code{reg} to @code{val}.
12308 @deftypefn {Built-in Function} void __builtin_rx_mvtipl (int)
12309 Generates the @code{mvtipl} machine instruction set the interrupt
12313 @deftypefn {Built-in Function} void __builtin_rx_racw (int)
12314 Generates the @code{racw} machine instruction to round the accumulator
12315 according to the specified mode.
12318 @deftypefn {Built-in Function} int __builtin_rx_revw (int)
12319 Generates the @code{revw} machine instruction which swaps the bytes in
12320 the argument so that bits 0--7 now occupy bits 8--15 and vice versa,
12321 and also bits 16--23 occupy bits 24--31 and vice versa.
12324 @deftypefn {Built-in Function} void __builtin_rx_rmpa (void)
12325 Generates the @code{rmpa} machine instruction which initiates a
12326 repeated multiply and accumulate sequence.
12329 @deftypefn {Built-in Function} void __builtin_rx_round (float)
12330 Generates the @code{round} machine instruction which returns the
12331 floating point argument rounded according to the current rounding mode
12332 set in the floating point status word register.
12335 @deftypefn {Built-in Function} int __builtin_rx_sat (int)
12336 Generates the @code{sat} machine instruction which returns the
12337 saturated value of the argument.
12340 @deftypefn {Built-in Function} void __builtin_rx_setpsw (int)
12341 Generates the @code{setpsw} machine instruction to set the specified
12342 bit in the processor status word.
12345 @deftypefn {Built-in Function} void __builtin_rx_wait (void)
12346 Generates the @code{wait} machine instruction.
12349 @node SPARC VIS Built-in Functions
12350 @subsection SPARC VIS Built-in Functions
12352 GCC supports SIMD operations on the SPARC using both the generic vector
12353 extensions (@pxref{Vector Extensions}) as well as built-in functions for
12354 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
12355 switch, the VIS extension is exposed as the following built-in functions:
12358 typedef int v2si __attribute__ ((vector_size (8)));
12359 typedef short v4hi __attribute__ ((vector_size (8)));
12360 typedef short v2hi __attribute__ ((vector_size (4)));
12361 typedef char v8qi __attribute__ ((vector_size (8)));
12362 typedef char v4qi __attribute__ ((vector_size (4)));
12364 void * __builtin_vis_alignaddr (void *, long);
12365 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
12366 v2si __builtin_vis_faligndatav2si (v2si, v2si);
12367 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
12368 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
12370 v4hi __builtin_vis_fexpand (v4qi);
12372 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
12373 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
12374 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
12375 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
12376 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
12377 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
12378 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
12380 v4qi __builtin_vis_fpack16 (v4hi);
12381 v8qi __builtin_vis_fpack32 (v2si, v2si);
12382 v2hi __builtin_vis_fpackfix (v2si);
12383 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
12385 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
12388 @node SPU Built-in Functions
12389 @subsection SPU Built-in Functions
12391 GCC provides extensions for the SPU processor as described in the
12392 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
12393 found at @uref{http://cell.scei.co.jp/} or
12394 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
12395 implementation differs in several ways.
12400 The optional extension of specifying vector constants in parentheses is
12404 A vector initializer requires no cast if the vector constant is of the
12405 same type as the variable it is initializing.
12408 If @code{signed} or @code{unsigned} is omitted, the signedness of the
12409 vector type is the default signedness of the base type. The default
12410 varies depending on the operating system, so a portable program should
12411 always specify the signedness.
12414 By default, the keyword @code{__vector} is added. The macro
12415 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
12419 GCC allows using a @code{typedef} name as the type specifier for a
12423 For C, overloaded functions are implemented with macros so the following
12427 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
12430 Since @code{spu_add} is a macro, the vector constant in the example
12431 is treated as four separate arguments. Wrap the entire argument in
12432 parentheses for this to work.
12435 The extended version of @code{__builtin_expect} is not supported.
12439 @emph{Note:} Only the interface described in the aforementioned
12440 specification is supported. Internally, GCC uses built-in functions to
12441 implement the required functionality, but these are not supported and
12442 are subject to change without notice.
12444 @node Target Format Checks
12445 @section Format Checks Specific to Particular Target Machines
12447 For some target machines, GCC supports additional options to the
12449 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
12452 * Solaris Format Checks::
12453 * Darwin Format Checks::
12456 @node Solaris Format Checks
12457 @subsection Solaris Format Checks
12459 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
12460 check. @code{cmn_err} accepts a subset of the standard @code{printf}
12461 conversions, and the two-argument @code{%b} conversion for displaying
12462 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
12464 @node Darwin Format Checks
12465 @subsection Darwin Format Checks
12467 Darwin targets support the @code{CFString} (or @code{__CFString__}) in the format
12468 attribute context. Declarations made with such attribution will be parsed for correct syntax
12469 and format argument types. However, parsing of the format string itself is currently undefined
12470 and will not be carried out by this version of the compiler.
12472 Additionally, @code{CFStringRefs} (defined by the @code{CoreFoundation} headers) may
12473 also be used as format arguments. Note that the relevant headers are only likely to be
12474 available on Darwin (OSX) installations. On such installations, the XCode and system
12475 documentation provide descriptions of @code{CFString}, @code{CFStringRefs} and
12476 associated functions.
12479 @section Pragmas Accepted by GCC
12481 @cindex @code{#pragma}
12483 GCC supports several types of pragmas, primarily in order to compile
12484 code originally written for other compilers. Note that in general
12485 we do not recommend the use of pragmas; @xref{Function Attributes},
12486 for further explanation.
12492 * RS/6000 and PowerPC Pragmas::
12494 * Solaris Pragmas::
12495 * Symbol-Renaming Pragmas::
12496 * Structure-Packing Pragmas::
12498 * Diagnostic Pragmas::
12499 * Visibility Pragmas::
12500 * Push/Pop Macro Pragmas::
12501 * Function Specific Option Pragmas::
12505 @subsection ARM Pragmas
12507 The ARM target defines pragmas for controlling the default addition of
12508 @code{long_call} and @code{short_call} attributes to functions.
12509 @xref{Function Attributes}, for information about the effects of these
12514 @cindex pragma, long_calls
12515 Set all subsequent functions to have the @code{long_call} attribute.
12517 @item no_long_calls
12518 @cindex pragma, no_long_calls
12519 Set all subsequent functions to have the @code{short_call} attribute.
12521 @item long_calls_off
12522 @cindex pragma, long_calls_off
12523 Do not affect the @code{long_call} or @code{short_call} attributes of
12524 subsequent functions.
12528 @subsection M32C Pragmas
12531 @item GCC memregs @var{number}
12532 @cindex pragma, memregs
12533 Overrides the command-line option @code{-memregs=} for the current
12534 file. Use with care! This pragma must be before any function in the
12535 file, and mixing different memregs values in different objects may
12536 make them incompatible. This pragma is useful when a
12537 performance-critical function uses a memreg for temporary values,
12538 as it may allow you to reduce the number of memregs used.
12540 @item ADDRESS @var{name} @var{address}
12541 @cindex pragma, address
12542 For any declared symbols matching @var{name}, this does three things
12543 to that symbol: it forces the symbol to be located at the given
12544 address (a number), it forces the symbol to be volatile, and it
12545 changes the symbol's scope to be static. This pragma exists for
12546 compatibility with other compilers, but note that the common
12547 @code{1234H} numeric syntax is not supported (use @code{0x1234}
12551 #pragma ADDRESS port3 0x103
12558 @subsection MeP Pragmas
12562 @item custom io_volatile (on|off)
12563 @cindex pragma, custom io_volatile
12564 Overrides the command line option @code{-mio-volatile} for the current
12565 file. Note that for compatibility with future GCC releases, this
12566 option should only be used once before any @code{io} variables in each
12569 @item GCC coprocessor available @var{registers}
12570 @cindex pragma, coprocessor available
12571 Specifies which coprocessor registers are available to the register
12572 allocator. @var{registers} may be a single register, register range
12573 separated by ellipses, or comma-separated list of those. Example:
12576 #pragma GCC coprocessor available $c0...$c10, $c28
12579 @item GCC coprocessor call_saved @var{registers}
12580 @cindex pragma, coprocessor call_saved
12581 Specifies which coprocessor registers are to be saved and restored by
12582 any function using them. @var{registers} may be a single register,
12583 register range separated by ellipses, or comma-separated list of
12587 #pragma GCC coprocessor call_saved $c4...$c6, $c31
12590 @item GCC coprocessor subclass '(A|B|C|D)' = @var{registers}
12591 @cindex pragma, coprocessor subclass
12592 Creates and defines a register class. These register classes can be
12593 used by inline @code{asm} constructs. @var{registers} may be a single
12594 register, register range separated by ellipses, or comma-separated
12595 list of those. Example:
12598 #pragma GCC coprocessor subclass 'B' = $c2, $c4, $c6
12600 asm ("cpfoo %0" : "=B" (x));
12603 @item GCC disinterrupt @var{name} , @var{name} @dots{}
12604 @cindex pragma, disinterrupt
12605 For the named functions, the compiler adds code to disable interrupts
12606 for the duration of those functions. Any functions so named, which
12607 are not encountered in the source, cause a warning that the pragma was
12608 not used. Examples:
12611 #pragma disinterrupt foo
12612 #pragma disinterrupt bar, grill
12613 int foo () @{ @dots{} @}
12616 @item GCC call @var{name} , @var{name} @dots{}
12617 @cindex pragma, call
12618 For the named functions, the compiler always uses a register-indirect
12619 call model when calling the named functions. Examples:
12628 @node RS/6000 and PowerPC Pragmas
12629 @subsection RS/6000 and PowerPC Pragmas
12631 The RS/6000 and PowerPC targets define one pragma for controlling
12632 whether or not the @code{longcall} attribute is added to function
12633 declarations by default. This pragma overrides the @option{-mlongcall}
12634 option, but not the @code{longcall} and @code{shortcall} attributes.
12635 @xref{RS/6000 and PowerPC Options}, for more information about when long
12636 calls are and are not necessary.
12640 @cindex pragma, longcall
12641 Apply the @code{longcall} attribute to all subsequent function
12645 Do not apply the @code{longcall} attribute to subsequent function
12649 @c Describe h8300 pragmas here.
12650 @c Describe sh pragmas here.
12651 @c Describe v850 pragmas here.
12653 @node Darwin Pragmas
12654 @subsection Darwin Pragmas
12656 The following pragmas are available for all architectures running the
12657 Darwin operating system. These are useful for compatibility with other
12661 @item mark @var{tokens}@dots{}
12662 @cindex pragma, mark
12663 This pragma is accepted, but has no effect.
12665 @item options align=@var{alignment}
12666 @cindex pragma, options align
12667 This pragma sets the alignment of fields in structures. The values of
12668 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
12669 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
12670 properly; to restore the previous setting, use @code{reset} for the
12673 @item segment @var{tokens}@dots{}
12674 @cindex pragma, segment
12675 This pragma is accepted, but has no effect.
12677 @item unused (@var{var} [, @var{var}]@dots{})
12678 @cindex pragma, unused
12679 This pragma declares variables to be possibly unused. GCC will not
12680 produce warnings for the listed variables. The effect is similar to
12681 that of the @code{unused} attribute, except that this pragma may appear
12682 anywhere within the variables' scopes.
12685 @node Solaris Pragmas
12686 @subsection Solaris Pragmas
12688 The Solaris target supports @code{#pragma redefine_extname}
12689 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
12690 @code{#pragma} directives for compatibility with the system compiler.
12693 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
12694 @cindex pragma, align
12696 Increase the minimum alignment of each @var{variable} to @var{alignment}.
12697 This is the same as GCC's @code{aligned} attribute @pxref{Variable
12698 Attributes}). Macro expansion occurs on the arguments to this pragma
12699 when compiling C and Objective-C@. It does not currently occur when
12700 compiling C++, but this is a bug which may be fixed in a future
12703 @item fini (@var{function} [, @var{function}]...)
12704 @cindex pragma, fini
12706 This pragma causes each listed @var{function} to be called after
12707 main, or during shared module unloading, by adding a call to the
12708 @code{.fini} section.
12710 @item init (@var{function} [, @var{function}]...)
12711 @cindex pragma, init
12713 This pragma causes each listed @var{function} to be called during
12714 initialization (before @code{main}) or during shared module loading, by
12715 adding a call to the @code{.init} section.
12719 @node Symbol-Renaming Pragmas
12720 @subsection Symbol-Renaming Pragmas
12722 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
12723 supports two @code{#pragma} directives which change the name used in
12724 assembly for a given declaration. @code{#pragma extern_prefix} is only
12725 available on platforms whose system headers need it. To get this effect
12726 on all platforms supported by GCC, use the asm labels extension (@pxref{Asm
12730 @item redefine_extname @var{oldname} @var{newname}
12731 @cindex pragma, redefine_extname
12733 This pragma gives the C function @var{oldname} the assembly symbol
12734 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
12735 will be defined if this pragma is available (currently on all platforms).
12737 @item extern_prefix @var{string}
12738 @cindex pragma, extern_prefix
12740 This pragma causes all subsequent external function and variable
12741 declarations to have @var{string} prepended to their assembly symbols.
12742 This effect may be terminated with another @code{extern_prefix} pragma
12743 whose argument is an empty string. The preprocessor macro
12744 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
12745 available (currently only on Tru64 UNIX)@.
12748 These pragmas and the asm labels extension interact in a complicated
12749 manner. Here are some corner cases you may want to be aware of.
12752 @item Both pragmas silently apply only to declarations with external
12753 linkage. Asm labels do not have this restriction.
12755 @item In C++, both pragmas silently apply only to declarations with
12756 ``C'' linkage. Again, asm labels do not have this restriction.
12758 @item If any of the three ways of changing the assembly name of a
12759 declaration is applied to a declaration whose assembly name has
12760 already been determined (either by a previous use of one of these
12761 features, or because the compiler needed the assembly name in order to
12762 generate code), and the new name is different, a warning issues and
12763 the name does not change.
12765 @item The @var{oldname} used by @code{#pragma redefine_extname} is
12766 always the C-language name.
12768 @item If @code{#pragma extern_prefix} is in effect, and a declaration
12769 occurs with an asm label attached, the prefix is silently ignored for
12772 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
12773 apply to the same declaration, whichever triggered first wins, and a
12774 warning issues if they contradict each other. (We would like to have
12775 @code{#pragma redefine_extname} always win, for consistency with asm
12776 labels, but if @code{#pragma extern_prefix} triggers first we have no
12777 way of knowing that that happened.)
12780 @node Structure-Packing Pragmas
12781 @subsection Structure-Packing Pragmas
12783 For compatibility with Microsoft Windows compilers, GCC supports a
12784 set of @code{#pragma} directives which change the maximum alignment of
12785 members of structures (other than zero-width bitfields), unions, and
12786 classes subsequently defined. The @var{n} value below always is required
12787 to be a small power of two and specifies the new alignment in bytes.
12790 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
12791 @item @code{#pragma pack()} sets the alignment to the one that was in
12792 effect when compilation started (see also command-line option
12793 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
12794 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
12795 setting on an internal stack and then optionally sets the new alignment.
12796 @item @code{#pragma pack(pop)} restores the alignment setting to the one
12797 saved at the top of the internal stack (and removes that stack entry).
12798 Note that @code{#pragma pack([@var{n}])} does not influence this internal
12799 stack; thus it is possible to have @code{#pragma pack(push)} followed by
12800 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
12801 @code{#pragma pack(pop)}.
12804 Some targets, e.g.@: i386 and powerpc, support the @code{ms_struct}
12805 @code{#pragma} which lays out a structure as the documented
12806 @code{__attribute__ ((ms_struct))}.
12808 @item @code{#pragma ms_struct on} turns on the layout for structures
12810 @item @code{#pragma ms_struct off} turns off the layout for structures
12812 @item @code{#pragma ms_struct reset} goes back to the default layout.
12816 @subsection Weak Pragmas
12818 For compatibility with SVR4, GCC supports a set of @code{#pragma}
12819 directives for declaring symbols to be weak, and defining weak
12823 @item #pragma weak @var{symbol}
12824 @cindex pragma, weak
12825 This pragma declares @var{symbol} to be weak, as if the declaration
12826 had the attribute of the same name. The pragma may appear before
12827 or after the declaration of @var{symbol}, but must appear before
12828 either its first use or its definition. It is not an error for
12829 @var{symbol} to never be defined at all.
12831 @item #pragma weak @var{symbol1} = @var{symbol2}
12832 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
12833 It is an error if @var{symbol2} is not defined in the current
12837 @node Diagnostic Pragmas
12838 @subsection Diagnostic Pragmas
12840 GCC allows the user to selectively enable or disable certain types of
12841 diagnostics, and change the kind of the diagnostic. For example, a
12842 project's policy might require that all sources compile with
12843 @option{-Werror} but certain files might have exceptions allowing
12844 specific types of warnings. Or, a project might selectively enable
12845 diagnostics and treat them as errors depending on which preprocessor
12846 macros are defined.
12849 @item #pragma GCC diagnostic @var{kind} @var{option}
12850 @cindex pragma, diagnostic
12852 Modifies the disposition of a diagnostic. Note that not all
12853 diagnostics are modifiable; at the moment only warnings (normally
12854 controlled by @samp{-W@dots{}}) can be controlled, and not all of them.
12855 Use @option{-fdiagnostics-show-option} to determine which diagnostics
12856 are controllable and which option controls them.
12858 @var{kind} is @samp{error} to treat this diagnostic as an error,
12859 @samp{warning} to treat it like a warning (even if @option{-Werror} is
12860 in effect), or @samp{ignored} if the diagnostic is to be ignored.
12861 @var{option} is a double quoted string which matches the command-line
12865 #pragma GCC diagnostic warning "-Wformat"
12866 #pragma GCC diagnostic error "-Wformat"
12867 #pragma GCC diagnostic ignored "-Wformat"
12870 Note that these pragmas override any command-line options. GCC keeps
12871 track of the location of each pragma, and issues diagnostics according
12872 to the state as of that point in the source file. Thus, pragmas occurring
12873 after a line do not affect diagnostics caused by that line.
12875 @item #pragma GCC diagnostic push
12876 @itemx #pragma GCC diagnostic pop
12878 Causes GCC to remember the state of the diagnostics as of each
12879 @code{push}, and restore to that point at each @code{pop}. If a
12880 @code{pop} has no matching @code{push}, the command line options are
12884 #pragma GCC diagnostic error "-Wuninitialized"
12885 foo(a); /* error is given for this one */
12886 #pragma GCC diagnostic push
12887 #pragma GCC diagnostic ignored "-Wuninitialized"
12888 foo(b); /* no diagnostic for this one */
12889 #pragma GCC diagnostic pop
12890 foo(c); /* error is given for this one */
12891 #pragma GCC diagnostic pop
12892 foo(d); /* depends on command line options */
12897 GCC also offers a simple mechanism for printing messages during
12901 @item #pragma message @var{string}
12902 @cindex pragma, diagnostic
12904 Prints @var{string} as a compiler message on compilation. The message
12905 is informational only, and is neither a compilation warning nor an error.
12908 #pragma message "Compiling " __FILE__ "..."
12911 @var{string} may be parenthesized, and is printed with location
12912 information. For example,
12915 #define DO_PRAGMA(x) _Pragma (#x)
12916 #define TODO(x) DO_PRAGMA(message ("TODO - " #x))
12918 TODO(Remember to fix this)
12921 prints @samp{/tmp/file.c:4: note: #pragma message:
12922 TODO - Remember to fix this}.
12926 @node Visibility Pragmas
12927 @subsection Visibility Pragmas
12930 @item #pragma GCC visibility push(@var{visibility})
12931 @itemx #pragma GCC visibility pop
12932 @cindex pragma, visibility
12934 This pragma allows the user to set the visibility for multiple
12935 declarations without having to give each a visibility attribute
12936 @xref{Function Attributes}, for more information about visibility and
12937 the attribute syntax.
12939 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
12940 declarations. Class members and template specializations are not
12941 affected; if you want to override the visibility for a particular
12942 member or instantiation, you must use an attribute.
12947 @node Push/Pop Macro Pragmas
12948 @subsection Push/Pop Macro Pragmas
12950 For compatibility with Microsoft Windows compilers, GCC supports
12951 @samp{#pragma push_macro(@var{"macro_name"})}
12952 and @samp{#pragma pop_macro(@var{"macro_name"})}.
12955 @item #pragma push_macro(@var{"macro_name"})
12956 @cindex pragma, push_macro
12957 This pragma saves the value of the macro named as @var{macro_name} to
12958 the top of the stack for this macro.
12960 @item #pragma pop_macro(@var{"macro_name"})
12961 @cindex pragma, pop_macro
12962 This pragma sets the value of the macro named as @var{macro_name} to
12963 the value on top of the stack for this macro. If the stack for
12964 @var{macro_name} is empty, the value of the macro remains unchanged.
12971 #pragma push_macro("X")
12974 #pragma pop_macro("X")
12978 In this example, the definition of X as 1 is saved by @code{#pragma
12979 push_macro} and restored by @code{#pragma pop_macro}.
12981 @node Function Specific Option Pragmas
12982 @subsection Function Specific Option Pragmas
12985 @item #pragma GCC target (@var{"string"}...)
12986 @cindex pragma GCC target
12988 This pragma allows you to set target specific options for functions
12989 defined later in the source file. One or more strings can be
12990 specified. Each function that is defined after this point will be as
12991 if @code{attribute((target("STRING")))} was specified for that
12992 function. The parenthesis around the options is optional.
12993 @xref{Function Attributes}, for more information about the
12994 @code{target} attribute and the attribute syntax.
12996 The @samp{#pragma GCC target} pragma is not implemented in GCC
12997 versions earlier than 4.4, and is currently only implemented for the
12998 386 and x86_64 backends.
13002 @item #pragma GCC optimize (@var{"string"}...)
13003 @cindex pragma GCC optimize
13005 This pragma allows you to set global optimization options for functions
13006 defined later in the source file. One or more strings can be
13007 specified. Each function that is defined after this point will be as
13008 if @code{attribute((optimize("STRING")))} was specified for that
13009 function. The parenthesis around the options is optional.
13010 @xref{Function Attributes}, for more information about the
13011 @code{optimize} attribute and the attribute syntax.
13013 The @samp{#pragma GCC optimize} pragma is not implemented in GCC
13014 versions earlier than 4.4.
13018 @item #pragma GCC push_options
13019 @itemx #pragma GCC pop_options
13020 @cindex pragma GCC push_options
13021 @cindex pragma GCC pop_options
13023 These pragmas maintain a stack of the current target and optimization
13024 options. It is intended for include files where you temporarily want
13025 to switch to using a different @samp{#pragma GCC target} or
13026 @samp{#pragma GCC optimize} and then to pop back to the previous
13029 The @samp{#pragma GCC push_options} and @samp{#pragma GCC pop_options}
13030 pragmas are not implemented in GCC versions earlier than 4.4.
13034 @item #pragma GCC reset_options
13035 @cindex pragma GCC reset_options
13037 This pragma clears the current @code{#pragma GCC target} and
13038 @code{#pragma GCC optimize} to use the default switches as specified
13039 on the command line.
13041 The @samp{#pragma GCC reset_options} pragma is not implemented in GCC
13042 versions earlier than 4.4.
13045 @node Unnamed Fields
13046 @section Unnamed struct/union fields within structs/unions
13047 @cindex @code{struct}
13048 @cindex @code{union}
13050 As permitted by ISO C1X and for compatibility with other compilers,
13051 GCC allows you to define
13052 a structure or union that contains, as fields, structures and unions
13053 without names. For example:
13066 In this example, the user would be able to access members of the unnamed
13067 union with code like @samp{foo.b}. Note that only unnamed structs and
13068 unions are allowed, you may not have, for example, an unnamed
13071 You must never create such structures that cause ambiguous field definitions.
13072 For example, this structure:
13083 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
13084 The compiler gives errors for such constructs.
13086 @opindex fms-extensions
13087 Unless @option{-fms-extensions} is used, the unnamed field must be a
13088 structure or union definition without a tag (for example, @samp{struct
13089 @{ int a; @};}), or a @code{typedef} name for such a structure or
13090 union. If @option{-fms-extensions} is used, the field may
13091 also be a definition with a tag such as @samp{struct foo @{ int a;
13092 @};}, a reference to a previously defined structure or union such as
13093 @samp{struct foo;}, or a reference to a @code{typedef} name for a
13094 previously defined structure or union type with a tag.
13096 @opindex fplan9-extensions
13097 The option @option{-fplan9-extensions} enables
13098 @option{-fms-extensions} as well as two other extensions. First, a
13099 pointer to a structure is automatically converted to a pointer to an
13100 anonymous field for assignments and function calls. For example:
13103 struct s1 @{ int a; @};
13104 struct s2 @{ struct s1; @};
13105 extern void f1 (struct s1 *);
13106 void f2 (struct s2 *p) @{ f1 (p); @}
13109 In the call to @code{f1} inside @code{f2}, the pointer @code{p} is
13110 converted into a pointer to the anonymous field.
13112 Second, when the type of an anonymous field is a @code{typedef} for a
13113 @code{struct} or @code{union}, code may refer to the field using the
13114 name of the @code{typedef}.
13117 typedef struct @{ int a; @} s1;
13118 struct s2 @{ s1; @};
13119 s1 f1 (struct s2 *p) @{ return p->s1; @}
13122 These usages are only permitted when they are not ambiguous.
13125 @section Thread-Local Storage
13126 @cindex Thread-Local Storage
13127 @cindex @acronym{TLS}
13128 @cindex @code{__thread}
13130 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
13131 are allocated such that there is one instance of the variable per extant
13132 thread. The run-time model GCC uses to implement this originates
13133 in the IA-64 processor-specific ABI, but has since been migrated
13134 to other processors as well. It requires significant support from
13135 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
13136 system libraries (@file{libc.so} and @file{libpthread.so}), so it
13137 is not available everywhere.
13139 At the user level, the extension is visible with a new storage
13140 class keyword: @code{__thread}. For example:
13144 extern __thread struct state s;
13145 static __thread char *p;
13148 The @code{__thread} specifier may be used alone, with the @code{extern}
13149 or @code{static} specifiers, but with no other storage class specifier.
13150 When used with @code{extern} or @code{static}, @code{__thread} must appear
13151 immediately after the other storage class specifier.
13153 The @code{__thread} specifier may be applied to any global, file-scoped
13154 static, function-scoped static, or static data member of a class. It may
13155 not be applied to block-scoped automatic or non-static data member.
13157 When the address-of operator is applied to a thread-local variable, it is
13158 evaluated at run-time and returns the address of the current thread's
13159 instance of that variable. An address so obtained may be used by any
13160 thread. When a thread terminates, any pointers to thread-local variables
13161 in that thread become invalid.
13163 No static initialization may refer to the address of a thread-local variable.
13165 In C++, if an initializer is present for a thread-local variable, it must
13166 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
13169 See @uref{http://people.redhat.com/drepper/tls.pdf,
13170 ELF Handling For Thread-Local Storage} for a detailed explanation of
13171 the four thread-local storage addressing models, and how the run-time
13172 is expected to function.
13175 * C99 Thread-Local Edits::
13176 * C++98 Thread-Local Edits::
13179 @node C99 Thread-Local Edits
13180 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
13182 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
13183 that document the exact semantics of the language extension.
13187 @cite{5.1.2 Execution environments}
13189 Add new text after paragraph 1
13192 Within either execution environment, a @dfn{thread} is a flow of
13193 control within a program. It is implementation defined whether
13194 or not there may be more than one thread associated with a program.
13195 It is implementation defined how threads beyond the first are
13196 created, the name and type of the function called at thread
13197 startup, and how threads may be terminated. However, objects
13198 with thread storage duration shall be initialized before thread
13203 @cite{6.2.4 Storage durations of objects}
13205 Add new text before paragraph 3
13208 An object whose identifier is declared with the storage-class
13209 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
13210 Its lifetime is the entire execution of the thread, and its
13211 stored value is initialized only once, prior to thread startup.
13215 @cite{6.4.1 Keywords}
13217 Add @code{__thread}.
13220 @cite{6.7.1 Storage-class specifiers}
13222 Add @code{__thread} to the list of storage class specifiers in
13225 Change paragraph 2 to
13228 With the exception of @code{__thread}, at most one storage-class
13229 specifier may be given [@dots{}]. The @code{__thread} specifier may
13230 be used alone, or immediately following @code{extern} or
13234 Add new text after paragraph 6
13237 The declaration of an identifier for a variable that has
13238 block scope that specifies @code{__thread} shall also
13239 specify either @code{extern} or @code{static}.
13241 The @code{__thread} specifier shall be used only with
13246 @node C++98 Thread-Local Edits
13247 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
13249 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
13250 that document the exact semantics of the language extension.
13254 @b{[intro.execution]}
13256 New text after paragraph 4
13259 A @dfn{thread} is a flow of control within the abstract machine.
13260 It is implementation defined whether or not there may be more than
13264 New text after paragraph 7
13267 It is unspecified whether additional action must be taken to
13268 ensure when and whether side effects are visible to other threads.
13274 Add @code{__thread}.
13277 @b{[basic.start.main]}
13279 Add after paragraph 5
13282 The thread that begins execution at the @code{main} function is called
13283 the @dfn{main thread}. It is implementation defined how functions
13284 beginning threads other than the main thread are designated or typed.
13285 A function so designated, as well as the @code{main} function, is called
13286 a @dfn{thread startup function}. It is implementation defined what
13287 happens if a thread startup function returns. It is implementation
13288 defined what happens to other threads when any thread calls @code{exit}.
13292 @b{[basic.start.init]}
13294 Add after paragraph 4
13297 The storage for an object of thread storage duration shall be
13298 statically initialized before the first statement of the thread startup
13299 function. An object of thread storage duration shall not require
13300 dynamic initialization.
13304 @b{[basic.start.term]}
13306 Add after paragraph 3
13309 The type of an object with thread storage duration shall not have a
13310 non-trivial destructor, nor shall it be an array type whose elements
13311 (directly or indirectly) have non-trivial destructors.
13317 Add ``thread storage duration'' to the list in paragraph 1.
13322 Thread, static, and automatic storage durations are associated with
13323 objects introduced by declarations [@dots{}].
13326 Add @code{__thread} to the list of specifiers in paragraph 3.
13329 @b{[basic.stc.thread]}
13331 New section before @b{[basic.stc.static]}
13334 The keyword @code{__thread} applied to a non-local object gives the
13335 object thread storage duration.
13337 A local variable or class data member declared both @code{static}
13338 and @code{__thread} gives the variable or member thread storage
13343 @b{[basic.stc.static]}
13348 All objects which have neither thread storage duration, dynamic
13349 storage duration nor are local [@dots{}].
13355 Add @code{__thread} to the list in paragraph 1.
13360 With the exception of @code{__thread}, at most one
13361 @var{storage-class-specifier} shall appear in a given
13362 @var{decl-specifier-seq}. The @code{__thread} specifier may
13363 be used alone, or immediately following the @code{extern} or
13364 @code{static} specifiers. [@dots{}]
13367 Add after paragraph 5
13370 The @code{__thread} specifier can be applied only to the names of objects
13371 and to anonymous unions.
13377 Add after paragraph 6
13380 Non-@code{static} members shall not be @code{__thread}.
13384 @node Binary constants
13385 @section Binary constants using the @samp{0b} prefix
13386 @cindex Binary constants using the @samp{0b} prefix
13388 Integer constants can be written as binary constants, consisting of a
13389 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
13390 @samp{0B}. This is particularly useful in environments that operate a
13391 lot on the bit-level (like microcontrollers).
13393 The following statements are identical:
13402 The type of these constants follows the same rules as for octal or
13403 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
13406 @node C++ Extensions
13407 @chapter Extensions to the C++ Language
13408 @cindex extensions, C++ language
13409 @cindex C++ language extensions
13411 The GNU compiler provides these extensions to the C++ language (and you
13412 can also use most of the C language extensions in your C++ programs). If you
13413 want to write code that checks whether these features are available, you can
13414 test for the GNU compiler the same way as for C programs: check for a
13415 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
13416 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
13417 Predefined Macros,cpp,The GNU C Preprocessor}).
13420 * C++ Volatiles:: What constitutes an access to a volatile object.
13421 * Restricted Pointers:: C99 restricted pointers and references.
13422 * Vague Linkage:: Where G++ puts inlines, vtables and such.
13423 * C++ Interface:: You can use a single C++ header file for both
13424 declarations and definitions.
13425 * Template Instantiation:: Methods for ensuring that exactly one copy of
13426 each needed template instantiation is emitted.
13427 * Bound member functions:: You can extract a function pointer to the
13428 method denoted by a @samp{->*} or @samp{.*} expression.
13429 * C++ Attributes:: Variable, function, and type attributes for C++ only.
13430 * Namespace Association:: Strong using-directives for namespace association.
13431 * Type Traits:: Compiler support for type traits
13432 * Java Exceptions:: Tweaking exception handling to work with Java.
13433 * Deprecated Features:: Things will disappear from g++.
13434 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
13437 @node C++ Volatiles
13438 @section When is a Volatile C++ Object Accessed?
13439 @cindex accessing volatiles
13440 @cindex volatile read
13441 @cindex volatile write
13442 @cindex volatile access
13444 The C++ standard differs from the C standard in its treatment of
13445 volatile objects. It fails to specify what constitutes a volatile
13446 access, except to say that C++ should behave in a similar manner to C
13447 with respect to volatiles, where possible. However, the different
13448 lvalueness of expressions between C and C++ complicate the behaviour.
13449 G++ behaves the same as GCC for volatile access, @xref{C
13450 Extensions,,Volatiles}, for a description of GCC's behaviour.
13452 The C and C++ language specifications differ when an object is
13453 accessed in a void context:
13456 volatile int *src = @var{somevalue};
13460 The C++ standard specifies that such expressions do not undergo lvalue
13461 to rvalue conversion, and that the type of the dereferenced object may
13462 be incomplete. The C++ standard does not specify explicitly that it
13463 is lvalue to rvalue conversion which is responsible for causing an
13464 access. There is reason to believe that it is, because otherwise
13465 certain simple expressions become undefined. However, because it
13466 would surprise most programmers, G++ treats dereferencing a pointer to
13467 volatile object of complete type as GCC would do for an equivalent
13468 type in C@. When the object has incomplete type, G++ issues a
13469 warning; if you wish to force an error, you must force a conversion to
13470 rvalue with, for instance, a static cast.
13472 When using a reference to volatile, G++ does not treat equivalent
13473 expressions as accesses to volatiles, but instead issues a warning that
13474 no volatile is accessed. The rationale for this is that otherwise it
13475 becomes difficult to determine where volatile access occur, and not
13476 possible to ignore the return value from functions returning volatile
13477 references. Again, if you wish to force a read, cast the reference to
13480 G++ implements the same behaviour as GCC does when assigning to a
13481 volatile object -- there is no reread of the assigned-to object, the
13482 assigned rvalue is reused. Note that in C++ assignment expressions
13483 are lvalues, and if used as an lvalue, the volatile object will be
13484 referred to. For instance, @var{vref} will refer to @var{vobj}, as
13485 expected, in the following example:
13489 volatile int &vref = vobj = @var{something};
13492 @node Restricted Pointers
13493 @section Restricting Pointer Aliasing
13494 @cindex restricted pointers
13495 @cindex restricted references
13496 @cindex restricted this pointer
13498 As with the C front end, G++ understands the C99 feature of restricted pointers,
13499 specified with the @code{__restrict__}, or @code{__restrict} type
13500 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
13501 language flag, @code{restrict} is not a keyword in C++.
13503 In addition to allowing restricted pointers, you can specify restricted
13504 references, which indicate that the reference is not aliased in the local
13508 void fn (int *__restrict__ rptr, int &__restrict__ rref)
13515 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
13516 @var{rref} refers to a (different) unaliased integer.
13518 You may also specify whether a member function's @var{this} pointer is
13519 unaliased by using @code{__restrict__} as a member function qualifier.
13522 void T::fn () __restrict__
13529 Within the body of @code{T::fn}, @var{this} will have the effective
13530 definition @code{T *__restrict__ const this}. Notice that the
13531 interpretation of a @code{__restrict__} member function qualifier is
13532 different to that of @code{const} or @code{volatile} qualifier, in that it
13533 is applied to the pointer rather than the object. This is consistent with
13534 other compilers which implement restricted pointers.
13536 As with all outermost parameter qualifiers, @code{__restrict__} is
13537 ignored in function definition matching. This means you only need to
13538 specify @code{__restrict__} in a function definition, rather than
13539 in a function prototype as well.
13541 @node Vague Linkage
13542 @section Vague Linkage
13543 @cindex vague linkage
13545 There are several constructs in C++ which require space in the object
13546 file but are not clearly tied to a single translation unit. We say that
13547 these constructs have ``vague linkage''. Typically such constructs are
13548 emitted wherever they are needed, though sometimes we can be more
13552 @item Inline Functions
13553 Inline functions are typically defined in a header file which can be
13554 included in many different compilations. Hopefully they can usually be
13555 inlined, but sometimes an out-of-line copy is necessary, if the address
13556 of the function is taken or if inlining fails. In general, we emit an
13557 out-of-line copy in all translation units where one is needed. As an
13558 exception, we only emit inline virtual functions with the vtable, since
13559 it will always require a copy.
13561 Local static variables and string constants used in an inline function
13562 are also considered to have vague linkage, since they must be shared
13563 between all inlined and out-of-line instances of the function.
13567 C++ virtual functions are implemented in most compilers using a lookup
13568 table, known as a vtable. The vtable contains pointers to the virtual
13569 functions provided by a class, and each object of the class contains a
13570 pointer to its vtable (or vtables, in some multiple-inheritance
13571 situations). If the class declares any non-inline, non-pure virtual
13572 functions, the first one is chosen as the ``key method'' for the class,
13573 and the vtable is only emitted in the translation unit where the key
13576 @emph{Note:} If the chosen key method is later defined as inline, the
13577 vtable will still be emitted in every translation unit which defines it.
13578 Make sure that any inline virtuals are declared inline in the class
13579 body, even if they are not defined there.
13581 @item @code{type_info} objects
13582 @cindex @code{type_info}
13584 C++ requires information about types to be written out in order to
13585 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
13586 For polymorphic classes (classes with virtual functions), the @samp{type_info}
13587 object is written out along with the vtable so that @samp{dynamic_cast}
13588 can determine the dynamic type of a class object at runtime. For all
13589 other types, we write out the @samp{type_info} object when it is used: when
13590 applying @samp{typeid} to an expression, throwing an object, or
13591 referring to a type in a catch clause or exception specification.
13593 @item Template Instantiations
13594 Most everything in this section also applies to template instantiations,
13595 but there are other options as well.
13596 @xref{Template Instantiation,,Where's the Template?}.
13600 When used with GNU ld version 2.8 or later on an ELF system such as
13601 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
13602 these constructs will be discarded at link time. This is known as
13605 On targets that don't support COMDAT, but do support weak symbols, GCC
13606 will use them. This way one copy will override all the others, but
13607 the unused copies will still take up space in the executable.
13609 For targets which do not support either COMDAT or weak symbols,
13610 most entities with vague linkage will be emitted as local symbols to
13611 avoid duplicate definition errors from the linker. This will not happen
13612 for local statics in inlines, however, as having multiple copies will
13613 almost certainly break things.
13615 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
13616 another way to control placement of these constructs.
13618 @node C++ Interface
13619 @section #pragma interface and implementation
13621 @cindex interface and implementation headers, C++
13622 @cindex C++ interface and implementation headers
13623 @cindex pragmas, interface and implementation
13625 @code{#pragma interface} and @code{#pragma implementation} provide the
13626 user with a way of explicitly directing the compiler to emit entities
13627 with vague linkage (and debugging information) in a particular
13630 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
13631 most cases, because of COMDAT support and the ``key method'' heuristic
13632 mentioned in @ref{Vague Linkage}. Using them can actually cause your
13633 program to grow due to unnecessary out-of-line copies of inline
13634 functions. Currently (3.4) the only benefit of these
13635 @code{#pragma}s is reduced duplication of debugging information, and
13636 that should be addressed soon on DWARF 2 targets with the use of
13640 @item #pragma interface
13641 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
13642 @kindex #pragma interface
13643 Use this directive in @emph{header files} that define object classes, to save
13644 space in most of the object files that use those classes. Normally,
13645 local copies of certain information (backup copies of inline member
13646 functions, debugging information, and the internal tables that implement
13647 virtual functions) must be kept in each object file that includes class
13648 definitions. You can use this pragma to avoid such duplication. When a
13649 header file containing @samp{#pragma interface} is included in a
13650 compilation, this auxiliary information will not be generated (unless
13651 the main input source file itself uses @samp{#pragma implementation}).
13652 Instead, the object files will contain references to be resolved at link
13655 The second form of this directive is useful for the case where you have
13656 multiple headers with the same name in different directories. If you
13657 use this form, you must specify the same string to @samp{#pragma
13660 @item #pragma implementation
13661 @itemx #pragma implementation "@var{objects}.h"
13662 @kindex #pragma implementation
13663 Use this pragma in a @emph{main input file}, when you want full output from
13664 included header files to be generated (and made globally visible). The
13665 included header file, in turn, should use @samp{#pragma interface}.
13666 Backup copies of inline member functions, debugging information, and the
13667 internal tables used to implement virtual functions are all generated in
13668 implementation files.
13670 @cindex implied @code{#pragma implementation}
13671 @cindex @code{#pragma implementation}, implied
13672 @cindex naming convention, implementation headers
13673 If you use @samp{#pragma implementation} with no argument, it applies to
13674 an include file with the same basename@footnote{A file's @dfn{basename}
13675 was the name stripped of all leading path information and of trailing
13676 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
13677 file. For example, in @file{allclass.cc}, giving just
13678 @samp{#pragma implementation}
13679 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
13681 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
13682 an implementation file whenever you would include it from
13683 @file{allclass.cc} even if you never specified @samp{#pragma
13684 implementation}. This was deemed to be more trouble than it was worth,
13685 however, and disabled.
13687 Use the string argument if you want a single implementation file to
13688 include code from multiple header files. (You must also use
13689 @samp{#include} to include the header file; @samp{#pragma
13690 implementation} only specifies how to use the file---it doesn't actually
13693 There is no way to split up the contents of a single header file into
13694 multiple implementation files.
13697 @cindex inlining and C++ pragmas
13698 @cindex C++ pragmas, effect on inlining
13699 @cindex pragmas in C++, effect on inlining
13700 @samp{#pragma implementation} and @samp{#pragma interface} also have an
13701 effect on function inlining.
13703 If you define a class in a header file marked with @samp{#pragma
13704 interface}, the effect on an inline function defined in that class is
13705 similar to an explicit @code{extern} declaration---the compiler emits
13706 no code at all to define an independent version of the function. Its
13707 definition is used only for inlining with its callers.
13709 @opindex fno-implement-inlines
13710 Conversely, when you include the same header file in a main source file
13711 that declares it as @samp{#pragma implementation}, the compiler emits
13712 code for the function itself; this defines a version of the function
13713 that can be found via pointers (or by callers compiled without
13714 inlining). If all calls to the function can be inlined, you can avoid
13715 emitting the function by compiling with @option{-fno-implement-inlines}.
13716 If any calls were not inlined, you will get linker errors.
13718 @node Template Instantiation
13719 @section Where's the Template?
13720 @cindex template instantiation
13722 C++ templates are the first language feature to require more
13723 intelligence from the environment than one usually finds on a UNIX
13724 system. Somehow the compiler and linker have to make sure that each
13725 template instance occurs exactly once in the executable if it is needed,
13726 and not at all otherwise. There are two basic approaches to this
13727 problem, which are referred to as the Borland model and the Cfront model.
13730 @item Borland model
13731 Borland C++ solved the template instantiation problem by adding the code
13732 equivalent of common blocks to their linker; the compiler emits template
13733 instances in each translation unit that uses them, and the linker
13734 collapses them together. The advantage of this model is that the linker
13735 only has to consider the object files themselves; there is no external
13736 complexity to worry about. This disadvantage is that compilation time
13737 is increased because the template code is being compiled repeatedly.
13738 Code written for this model tends to include definitions of all
13739 templates in the header file, since they must be seen to be
13743 The AT&T C++ translator, Cfront, solved the template instantiation
13744 problem by creating the notion of a template repository, an
13745 automatically maintained place where template instances are stored. A
13746 more modern version of the repository works as follows: As individual
13747 object files are built, the compiler places any template definitions and
13748 instantiations encountered in the repository. At link time, the link
13749 wrapper adds in the objects in the repository and compiles any needed
13750 instances that were not previously emitted. The advantages of this
13751 model are more optimal compilation speed and the ability to use the
13752 system linker; to implement the Borland model a compiler vendor also
13753 needs to replace the linker. The disadvantages are vastly increased
13754 complexity, and thus potential for error; for some code this can be
13755 just as transparent, but in practice it can been very difficult to build
13756 multiple programs in one directory and one program in multiple
13757 directories. Code written for this model tends to separate definitions
13758 of non-inline member templates into a separate file, which should be
13759 compiled separately.
13762 When used with GNU ld version 2.8 or later on an ELF system such as
13763 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
13764 Borland model. On other systems, G++ implements neither automatic
13767 A future version of G++ will support a hybrid model whereby the compiler
13768 will emit any instantiations for which the template definition is
13769 included in the compile, and store template definitions and
13770 instantiation context information into the object file for the rest.
13771 The link wrapper will extract that information as necessary and invoke
13772 the compiler to produce the remaining instantiations. The linker will
13773 then combine duplicate instantiations.
13775 In the mean time, you have the following options for dealing with
13776 template instantiations:
13781 Compile your template-using code with @option{-frepo}. The compiler will
13782 generate files with the extension @samp{.rpo} listing all of the
13783 template instantiations used in the corresponding object files which
13784 could be instantiated there; the link wrapper, @samp{collect2}, will
13785 then update the @samp{.rpo} files to tell the compiler where to place
13786 those instantiations and rebuild any affected object files. The
13787 link-time overhead is negligible after the first pass, as the compiler
13788 will continue to place the instantiations in the same files.
13790 This is your best option for application code written for the Borland
13791 model, as it will just work. Code written for the Cfront model will
13792 need to be modified so that the template definitions are available at
13793 one or more points of instantiation; usually this is as simple as adding
13794 @code{#include <tmethods.cc>} to the end of each template header.
13796 For library code, if you want the library to provide all of the template
13797 instantiations it needs, just try to link all of its object files
13798 together; the link will fail, but cause the instantiations to be
13799 generated as a side effect. Be warned, however, that this may cause
13800 conflicts if multiple libraries try to provide the same instantiations.
13801 For greater control, use explicit instantiation as described in the next
13805 @opindex fno-implicit-templates
13806 Compile your code with @option{-fno-implicit-templates} to disable the
13807 implicit generation of template instances, and explicitly instantiate
13808 all the ones you use. This approach requires more knowledge of exactly
13809 which instances you need than do the others, but it's less
13810 mysterious and allows greater control. You can scatter the explicit
13811 instantiations throughout your program, perhaps putting them in the
13812 translation units where the instances are used or the translation units
13813 that define the templates themselves; you can put all of the explicit
13814 instantiations you need into one big file; or you can create small files
13821 template class Foo<int>;
13822 template ostream& operator <<
13823 (ostream&, const Foo<int>&);
13826 for each of the instances you need, and create a template instantiation
13827 library from those.
13829 If you are using Cfront-model code, you can probably get away with not
13830 using @option{-fno-implicit-templates} when compiling files that don't
13831 @samp{#include} the member template definitions.
13833 If you use one big file to do the instantiations, you may want to
13834 compile it without @option{-fno-implicit-templates} so you get all of the
13835 instances required by your explicit instantiations (but not by any
13836 other files) without having to specify them as well.
13838 G++ has extended the template instantiation syntax given in the ISO
13839 standard to allow forward declaration of explicit instantiations
13840 (with @code{extern}), instantiation of the compiler support data for a
13841 template class (i.e.@: the vtable) without instantiating any of its
13842 members (with @code{inline}), and instantiation of only the static data
13843 members of a template class, without the support data or member
13844 functions (with (@code{static}):
13847 extern template int max (int, int);
13848 inline template class Foo<int>;
13849 static template class Foo<int>;
13853 Do nothing. Pretend G++ does implement automatic instantiation
13854 management. Code written for the Borland model will work fine, but
13855 each translation unit will contain instances of each of the templates it
13856 uses. In a large program, this can lead to an unacceptable amount of code
13860 @node Bound member functions
13861 @section Extracting the function pointer from a bound pointer to member function
13863 @cindex pointer to member function
13864 @cindex bound pointer to member function
13866 In C++, pointer to member functions (PMFs) are implemented using a wide
13867 pointer of sorts to handle all the possible call mechanisms; the PMF
13868 needs to store information about how to adjust the @samp{this} pointer,
13869 and if the function pointed to is virtual, where to find the vtable, and
13870 where in the vtable to look for the member function. If you are using
13871 PMFs in an inner loop, you should really reconsider that decision. If
13872 that is not an option, you can extract the pointer to the function that
13873 would be called for a given object/PMF pair and call it directly inside
13874 the inner loop, to save a bit of time.
13876 Note that you will still be paying the penalty for the call through a
13877 function pointer; on most modern architectures, such a call defeats the
13878 branch prediction features of the CPU@. This is also true of normal
13879 virtual function calls.
13881 The syntax for this extension is
13885 extern int (A::*fp)();
13886 typedef int (*fptr)(A *);
13888 fptr p = (fptr)(a.*fp);
13891 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
13892 no object is needed to obtain the address of the function. They can be
13893 converted to function pointers directly:
13896 fptr p1 = (fptr)(&A::foo);
13899 @opindex Wno-pmf-conversions
13900 You must specify @option{-Wno-pmf-conversions} to use this extension.
13902 @node C++ Attributes
13903 @section C++-Specific Variable, Function, and Type Attributes
13905 Some attributes only make sense for C++ programs.
13908 @item init_priority (@var{priority})
13909 @cindex @code{init_priority} attribute
13912 In Standard C++, objects defined at namespace scope are guaranteed to be
13913 initialized in an order in strict accordance with that of their definitions
13914 @emph{in a given translation unit}. No guarantee is made for initializations
13915 across translation units. However, GNU C++ allows users to control the
13916 order of initialization of objects defined at namespace scope with the
13917 @code{init_priority} attribute by specifying a relative @var{priority},
13918 a constant integral expression currently bounded between 101 and 65535
13919 inclusive. Lower numbers indicate a higher priority.
13921 In the following example, @code{A} would normally be created before
13922 @code{B}, but the @code{init_priority} attribute has reversed that order:
13925 Some_Class A __attribute__ ((init_priority (2000)));
13926 Some_Class B __attribute__ ((init_priority (543)));
13930 Note that the particular values of @var{priority} do not matter; only their
13933 @item java_interface
13934 @cindex @code{java_interface} attribute
13936 This type attribute informs C++ that the class is a Java interface. It may
13937 only be applied to classes declared within an @code{extern "Java"} block.
13938 Calls to methods declared in this interface will be dispatched using GCJ's
13939 interface table mechanism, instead of regular virtual table dispatch.
13943 See also @ref{Namespace Association}.
13945 @node Namespace Association
13946 @section Namespace Association
13948 @strong{Caution:} The semantics of this extension are not fully
13949 defined. Users should refrain from using this extension as its
13950 semantics may change subtly over time. It is possible that this
13951 extension will be removed in future versions of G++.
13953 A using-directive with @code{__attribute ((strong))} is stronger
13954 than a normal using-directive in two ways:
13958 Templates from the used namespace can be specialized and explicitly
13959 instantiated as though they were members of the using namespace.
13962 The using namespace is considered an associated namespace of all
13963 templates in the used namespace for purposes of argument-dependent
13967 The used namespace must be nested within the using namespace so that
13968 normal unqualified lookup works properly.
13970 This is useful for composing a namespace transparently from
13971 implementation namespaces. For example:
13976 template <class T> struct A @{ @};
13978 using namespace debug __attribute ((__strong__));
13979 template <> struct A<int> @{ @}; // @r{ok to specialize}
13981 template <class T> void f (A<T>);
13986 f (std::A<float>()); // @r{lookup finds} std::f
13992 @section Type Traits
13994 The C++ front-end implements syntactic extensions that allow to
13995 determine at compile time various characteristics of a type (or of a
13999 @item __has_nothrow_assign (type)
14000 If @code{type} is const qualified or is a reference type then the trait is
14001 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
14002 is true, else if @code{type} is a cv class or union type with copy assignment
14003 operators that are known not to throw an exception then the trait is true,
14004 else it is false. Requires: @code{type} shall be a complete type, an array
14005 type of unknown bound, or is a @code{void} type.
14007 @item __has_nothrow_copy (type)
14008 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
14009 @code{type} is a cv class or union type with copy constructors that
14010 are known not to throw an exception then the trait is true, else it is false.
14011 Requires: @code{type} shall be a complete type, an array type of
14012 unknown bound, or is a @code{void} type.
14014 @item __has_nothrow_constructor (type)
14015 If @code{__has_trivial_constructor (type)} is true then the trait is
14016 true, else if @code{type} is a cv class or union type (or array
14017 thereof) with a default constructor that is known not to throw an
14018 exception then the trait is true, else it is false. Requires:
14019 @code{type} shall be a complete type, an array type of unknown bound,
14020 or is a @code{void} type.
14022 @item __has_trivial_assign (type)
14023 If @code{type} is const qualified or is a reference type then the trait is
14024 false. Otherwise if @code{__is_pod (type)} is true then the trait is
14025 true, else if @code{type} is a cv class or union type with a trivial
14026 copy assignment ([class.copy]) then the trait is true, else it is
14027 false. Requires: @code{type} shall be a complete type, an array type
14028 of unknown bound, or is a @code{void} type.
14030 @item __has_trivial_copy (type)
14031 If @code{__is_pod (type)} is true or @code{type} is a reference type
14032 then the trait is true, else if @code{type} is a cv class or union type
14033 with a trivial copy constructor ([class.copy]) then the trait
14034 is true, else it is false. Requires: @code{type} shall be a complete
14035 type, an array type of unknown bound, or is a @code{void} type.
14037 @item __has_trivial_constructor (type)
14038 If @code{__is_pod (type)} is true then the trait is true, else if
14039 @code{type} is a cv class or union type (or array thereof) with a
14040 trivial default constructor ([class.ctor]) then the trait is true,
14041 else it is false. Requires: @code{type} shall be a complete type, an
14042 array type of unknown bound, or is a @code{void} type.
14044 @item __has_trivial_destructor (type)
14045 If @code{__is_pod (type)} is true or @code{type} is a reference type then
14046 the trait is true, else if @code{type} is a cv class or union type (or
14047 array thereof) with a trivial destructor ([class.dtor]) then the trait
14048 is true, else it is false. Requires: @code{type} shall be a complete
14049 type, an array type of unknown bound, or is a @code{void} type.
14051 @item __has_virtual_destructor (type)
14052 If @code{type} is a class type with a virtual destructor
14053 ([class.dtor]) then the trait is true, else it is false. Requires:
14054 @code{type} shall be a complete type, an array type of unknown bound,
14055 or is a @code{void} type.
14057 @item __is_abstract (type)
14058 If @code{type} is an abstract class ([class.abstract]) then the trait
14059 is true, else it is false. Requires: @code{type} shall be a complete
14060 type, an array type of unknown bound, or is a @code{void} type.
14062 @item __is_base_of (base_type, derived_type)
14063 If @code{base_type} is a base class of @code{derived_type}
14064 ([class.derived]) then the trait is true, otherwise it is false.
14065 Top-level cv qualifications of @code{base_type} and
14066 @code{derived_type} are ignored. For the purposes of this trait, a
14067 class type is considered is own base. Requires: if @code{__is_class
14068 (base_type)} and @code{__is_class (derived_type)} are true and
14069 @code{base_type} and @code{derived_type} are not the same type
14070 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
14071 type. Diagnostic is produced if this requirement is not met.
14073 @item __is_class (type)
14074 If @code{type} is a cv class type, and not a union type
14075 ([basic.compound]) the trait is true, else it is false.
14077 @item __is_empty (type)
14078 If @code{__is_class (type)} is false then the trait is false.
14079 Otherwise @code{type} is considered empty if and only if: @code{type}
14080 has no non-static data members, or all non-static data members, if
14081 any, are bit-fields of length 0, and @code{type} has no virtual
14082 members, and @code{type} has no virtual base classes, and @code{type}
14083 has no base classes @code{base_type} for which
14084 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
14085 be a complete type, an array type of unknown bound, or is a
14088 @item __is_enum (type)
14089 If @code{type} is a cv enumeration type ([basic.compound]) the trait is
14090 true, else it is false.
14092 @item __is_pod (type)
14093 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
14094 else it is false. Requires: @code{type} shall be a complete type,
14095 an array type of unknown bound, or is a @code{void} type.
14097 @item __is_polymorphic (type)
14098 If @code{type} is a polymorphic class ([class.virtual]) then the trait
14099 is true, else it is false. Requires: @code{type} shall be a complete
14100 type, an array type of unknown bound, or is a @code{void} type.
14102 @item __is_union (type)
14103 If @code{type} is a cv union type ([basic.compound]) the trait is
14104 true, else it is false.
14108 @node Java Exceptions
14109 @section Java Exceptions
14111 The Java language uses a slightly different exception handling model
14112 from C++. Normally, GNU C++ will automatically detect when you are
14113 writing C++ code that uses Java exceptions, and handle them
14114 appropriately. However, if C++ code only needs to execute destructors
14115 when Java exceptions are thrown through it, GCC will guess incorrectly.
14116 Sample problematic code is:
14119 struct S @{ ~S(); @};
14120 extern void bar(); // @r{is written in Java, and may throw exceptions}
14129 The usual effect of an incorrect guess is a link failure, complaining of
14130 a missing routine called @samp{__gxx_personality_v0}.
14132 You can inform the compiler that Java exceptions are to be used in a
14133 translation unit, irrespective of what it might think, by writing
14134 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
14135 @samp{#pragma} must appear before any functions that throw or catch
14136 exceptions, or run destructors when exceptions are thrown through them.
14138 You cannot mix Java and C++ exceptions in the same translation unit. It
14139 is believed to be safe to throw a C++ exception from one file through
14140 another file compiled for the Java exception model, or vice versa, but
14141 there may be bugs in this area.
14143 @node Deprecated Features
14144 @section Deprecated Features
14146 In the past, the GNU C++ compiler was extended to experiment with new
14147 features, at a time when the C++ language was still evolving. Now that
14148 the C++ standard is complete, some of those features are superseded by
14149 superior alternatives. Using the old features might cause a warning in
14150 some cases that the feature will be dropped in the future. In other
14151 cases, the feature might be gone already.
14153 While the list below is not exhaustive, it documents some of the options
14154 that are now deprecated:
14157 @item -fexternal-templates
14158 @itemx -falt-external-templates
14159 These are two of the many ways for G++ to implement template
14160 instantiation. @xref{Template Instantiation}. The C++ standard clearly
14161 defines how template definitions have to be organized across
14162 implementation units. G++ has an implicit instantiation mechanism that
14163 should work just fine for standard-conforming code.
14165 @item -fstrict-prototype
14166 @itemx -fno-strict-prototype
14167 Previously it was possible to use an empty prototype parameter list to
14168 indicate an unspecified number of parameters (like C), rather than no
14169 parameters, as C++ demands. This feature has been removed, except where
14170 it is required for backwards compatibility. @xref{Backwards Compatibility}.
14173 G++ allows a virtual function returning @samp{void *} to be overridden
14174 by one returning a different pointer type. This extension to the
14175 covariant return type rules is now deprecated and will be removed from a
14178 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
14179 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
14180 and are now removed from G++. Code using these operators should be
14181 modified to use @code{std::min} and @code{std::max} instead.
14183 The named return value extension has been deprecated, and is now
14186 The use of initializer lists with new expressions has been deprecated,
14187 and is now removed from G++.
14189 Floating and complex non-type template parameters have been deprecated,
14190 and are now removed from G++.
14192 The implicit typename extension has been deprecated and is now
14195 The use of default arguments in function pointers, function typedefs
14196 and other places where they are not permitted by the standard is
14197 deprecated and will be removed from a future version of G++.
14199 G++ allows floating-point literals to appear in integral constant expressions,
14200 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
14201 This extension is deprecated and will be removed from a future version.
14203 G++ allows static data members of const floating-point type to be declared
14204 with an initializer in a class definition. The standard only allows
14205 initializers for static members of const integral types and const
14206 enumeration types so this extension has been deprecated and will be removed
14207 from a future version.
14209 @node Backwards Compatibility
14210 @section Backwards Compatibility
14211 @cindex Backwards Compatibility
14212 @cindex ARM [Annotated C++ Reference Manual]
14214 Now that there is a definitive ISO standard C++, G++ has a specification
14215 to adhere to. The C++ language evolved over time, and features that
14216 used to be acceptable in previous drafts of the standard, such as the ARM
14217 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
14218 compilation of C++ written to such drafts, G++ contains some backwards
14219 compatibilities. @emph{All such backwards compatibility features are
14220 liable to disappear in future versions of G++.} They should be considered
14221 deprecated. @xref{Deprecated Features}.
14225 If a variable is declared at for scope, it used to remain in scope until
14226 the end of the scope which contained the for statement (rather than just
14227 within the for scope). G++ retains this, but issues a warning, if such a
14228 variable is accessed outside the for scope.
14230 @item Implicit C language
14231 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
14232 scope to set the language. On such systems, all header files are
14233 implicitly scoped inside a C language scope. Also, an empty prototype
14234 @code{()} will be treated as an unspecified number of arguments, rather
14235 than no arguments, as C++ demands.