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 callee_pop_aggregate_return (@var{number})
2827 @cindex @code{callee_pop_aggregate_return} attribute
2829 On 32-bit i?86-*-* targets, you can control by those attribute for
2830 aggregate return in memory, if the caller is responsible to pop the hidden
2831 pointer together with the rest of the arguments - @var{number} equal to
2832 zero -, or if the callee is responsible to pop hidden pointer - @var{number}
2835 For i?86-netware, the caller pops the stack for the hidden arguments pointing
2836 to aggregate return value. This differs from the default i386 ABI which assumes
2837 that the callee pops the stack for hidden pointer.
2839 @item ms_hook_prologue
2840 @cindex @code{ms_hook_prologue} attribute
2842 On 32 bit i[34567]86-*-* targets and 64 bit x86_64-*-* targets, you can use
2843 this function attribute to make gcc generate the "hot-patching" function
2844 prologue used in Win32 API functions in Microsoft Windows XP Service Pack 2
2848 @cindex function without a prologue/epilogue code
2849 Use this attribute on the ARM, AVR, MCORE, RX and SPU ports to indicate that
2850 the specified function does not need prologue/epilogue sequences generated by
2851 the compiler. It is up to the programmer to provide these sequences. The
2852 only statements that can be safely included in naked functions are
2853 @code{asm} statements that do not have operands. All other statements,
2854 including declarations of local variables, @code{if} statements, and so
2855 forth, should be avoided. Naked functions should be used to implement the
2856 body of an assembly function, while allowing the compiler to construct
2857 the requisite function declaration for the assembler.
2860 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2861 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2862 use the normal calling convention based on @code{jsr} and @code{rts}.
2863 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2866 On MeP targets this attribute causes the compiler to assume the called
2867 function is close enough to use the normal calling convention,
2868 overriding the @code{-mtf} command line option.
2871 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2872 Use this attribute together with @code{interrupt_handler},
2873 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2874 entry code should enable nested interrupts or exceptions.
2877 @cindex NMI handler functions on the Blackfin processor
2878 Use this attribute on the Blackfin to indicate that the specified function
2879 is an NMI handler. The compiler will generate function entry and
2880 exit sequences suitable for use in an NMI handler when this
2881 attribute is present.
2883 @item no_instrument_function
2884 @cindex @code{no_instrument_function} function attribute
2885 @opindex finstrument-functions
2886 If @option{-finstrument-functions} is given, profiling function calls will
2887 be generated at entry and exit of most user-compiled functions.
2888 Functions with this attribute will not be so instrumented.
2890 @item no_split_stack
2891 @cindex @code{no_split_stack} function attribute
2892 @opindex fsplit-stack
2893 If @option{-fsplit-stack} is given, functions will have a small
2894 prologue which decides whether to split the stack. Functions with the
2895 @code{no_split_stack} attribute will not have that prologue, and thus
2896 may run with only a small amount of stack space available.
2899 @cindex @code{noinline} function attribute
2900 This function attribute prevents a function from being considered for
2902 @c Don't enumerate the optimizations by name here; we try to be
2903 @c future-compatible with this mechanism.
2904 If the function does not have side-effects, there are optimizations
2905 other than inlining that causes function calls to be optimized away,
2906 although the function call is live. To keep such calls from being
2911 (@pxref{Extended Asm}) in the called function, to serve as a special
2915 @cindex @code{noclone} function attribute
2916 This function attribute prevents a function from being considered for
2917 cloning - a mechanism which produces specialized copies of functions
2918 and which is (currently) performed by interprocedural constant
2921 @item nonnull (@var{arg-index}, @dots{})
2922 @cindex @code{nonnull} function attribute
2923 The @code{nonnull} attribute specifies that some function parameters should
2924 be non-null pointers. For instance, the declaration:
2928 my_memcpy (void *dest, const void *src, size_t len)
2929 __attribute__((nonnull (1, 2)));
2933 causes the compiler to check that, in calls to @code{my_memcpy},
2934 arguments @var{dest} and @var{src} are non-null. If the compiler
2935 determines that a null pointer is passed in an argument slot marked
2936 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2937 is issued. The compiler may also choose to make optimizations based
2938 on the knowledge that certain function arguments will not be null.
2940 If no argument index list is given to the @code{nonnull} attribute,
2941 all pointer arguments are marked as non-null. To illustrate, the
2942 following declaration is equivalent to the previous example:
2946 my_memcpy (void *dest, const void *src, size_t len)
2947 __attribute__((nonnull));
2951 @cindex @code{noreturn} function attribute
2952 A few standard library functions, such as @code{abort} and @code{exit},
2953 cannot return. GCC knows this automatically. Some programs define
2954 their own functions that never return. You can declare them
2955 @code{noreturn} to tell the compiler this fact. For example,
2959 void fatal () __attribute__ ((noreturn));
2962 fatal (/* @r{@dots{}} */)
2964 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2970 The @code{noreturn} keyword tells the compiler to assume that
2971 @code{fatal} cannot return. It can then optimize without regard to what
2972 would happen if @code{fatal} ever did return. This makes slightly
2973 better code. More importantly, it helps avoid spurious warnings of
2974 uninitialized variables.
2976 The @code{noreturn} keyword does not affect the exceptional path when that
2977 applies: a @code{noreturn}-marked function may still return to the caller
2978 by throwing an exception or calling @code{longjmp}.
2980 Do not assume that registers saved by the calling function are
2981 restored before calling the @code{noreturn} function.
2983 It does not make sense for a @code{noreturn} function to have a return
2984 type other than @code{void}.
2986 The attribute @code{noreturn} is not implemented in GCC versions
2987 earlier than 2.5. An alternative way to declare that a function does
2988 not return, which works in the current version and in some older
2989 versions, is as follows:
2992 typedef void voidfn ();
2994 volatile voidfn fatal;
2997 This approach does not work in GNU C++.
3000 @cindex @code{nothrow} function attribute
3001 The @code{nothrow} attribute is used to inform the compiler that a
3002 function cannot throw an exception. For example, most functions in
3003 the standard C library can be guaranteed not to throw an exception
3004 with the notable exceptions of @code{qsort} and @code{bsearch} that
3005 take function pointer arguments. The @code{nothrow} attribute is not
3006 implemented in GCC versions earlier than 3.3.
3009 @cindex @code{optimize} function attribute
3010 The @code{optimize} attribute is used to specify that a function is to
3011 be compiled with different optimization options than specified on the
3012 command line. Arguments can either be numbers or strings. Numbers
3013 are assumed to be an optimization level. Strings that begin with
3014 @code{O} are assumed to be an optimization option, while other options
3015 are assumed to be used with a @code{-f} prefix. You can also use the
3016 @samp{#pragma GCC optimize} pragma to set the optimization options
3017 that affect more than one function.
3018 @xref{Function Specific Option Pragmas}, for details about the
3019 @samp{#pragma GCC optimize} pragma.
3021 This can be used for instance to have frequently executed functions
3022 compiled with more aggressive optimization options that produce faster
3023 and larger code, while other functions can be called with less
3027 @cindex @code{pcs} function attribute
3029 The @code{pcs} attribute can be used to control the calling convention
3030 used for a function on ARM. The attribute takes an argument that specifies
3031 the calling convention to use.
3033 When compiling using the AAPCS ABI (or a variant of that) then valid
3034 values for the argument are @code{"aapcs"} and @code{"aapcs-vfp"}. In
3035 order to use a variant other than @code{"aapcs"} then the compiler must
3036 be permitted to use the appropriate co-processor registers (i.e., the
3037 VFP registers must be available in order to use @code{"aapcs-vfp"}).
3041 /* Argument passed in r0, and result returned in r0+r1. */
3042 double f2d (float) __attribute__((pcs("aapcs")));
3045 Variadic functions always use the @code{"aapcs"} calling convention and
3046 the compiler will reject attempts to specify an alternative.
3049 @cindex @code{pure} function attribute
3050 Many functions have no effects except the return value and their
3051 return value depends only on the parameters and/or global variables.
3052 Such a function can be subject
3053 to common subexpression elimination and loop optimization just as an
3054 arithmetic operator would be. These functions should be declared
3055 with the attribute @code{pure}. For example,
3058 int square (int) __attribute__ ((pure));
3062 says that the hypothetical function @code{square} is safe to call
3063 fewer times than the program says.
3065 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
3066 Interesting non-pure functions are functions with infinite loops or those
3067 depending on volatile memory or other system resource, that may change between
3068 two consecutive calls (such as @code{feof} in a multithreading environment).
3070 The attribute @code{pure} is not implemented in GCC versions earlier
3074 @cindex @code{hot} function attribute
3075 The @code{hot} attribute is used to inform the compiler that a function is a
3076 hot spot of the compiled program. The function is optimized more aggressively
3077 and on many target it is placed into special subsection of the text section so
3078 all hot functions appears close together improving locality.
3080 When profile feedback is available, via @option{-fprofile-use}, hot functions
3081 are automatically detected and this attribute is ignored.
3083 The @code{hot} attribute is not implemented in GCC versions earlier
3087 @cindex @code{cold} function attribute
3088 The @code{cold} attribute is used to inform the compiler that a function is
3089 unlikely executed. The function is optimized for size rather than speed and on
3090 many targets it is placed into special subsection of the text section so all
3091 cold functions appears close together improving code locality of non-cold parts
3092 of program. The paths leading to call of cold functions within code are marked
3093 as unlikely by the branch prediction mechanism. It is thus useful to mark
3094 functions used to handle unlikely conditions, such as @code{perror}, as cold to
3095 improve optimization of hot functions that do call marked functions in rare
3098 When profile feedback is available, via @option{-fprofile-use}, hot functions
3099 are automatically detected and this attribute is ignored.
3101 The @code{cold} attribute is not implemented in GCC versions earlier than 4.3.
3103 @item regparm (@var{number})
3104 @cindex @code{regparm} attribute
3105 @cindex functions that are passed arguments in registers on the 386
3106 On the Intel 386, the @code{regparm} attribute causes the compiler to
3107 pass arguments number one to @var{number} if they are of integral type
3108 in registers EAX, EDX, and ECX instead of on the stack. Functions that
3109 take a variable number of arguments will continue to be passed all of their
3110 arguments on the stack.
3112 Beware that on some ELF systems this attribute is unsuitable for
3113 global functions in shared libraries with lazy binding (which is the
3114 default). Lazy binding will send the first call via resolving code in
3115 the loader, which might assume EAX, EDX and ECX can be clobbered, as
3116 per the standard calling conventions. Solaris 8 is affected by this.
3117 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
3118 safe since the loaders there save EAX, EDX and ECX. (Lazy binding can be
3119 disabled with the linker or the loader if desired, to avoid the
3123 @cindex @code{sseregparm} attribute
3124 On the Intel 386 with SSE support, the @code{sseregparm} attribute
3125 causes the compiler to pass up to 3 floating point arguments in
3126 SSE registers instead of on the stack. Functions that take a
3127 variable number of arguments will continue to pass all of their
3128 floating point arguments on the stack.
3130 @item force_align_arg_pointer
3131 @cindex @code{force_align_arg_pointer} attribute
3132 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
3133 applied to individual function definitions, generating an alternate
3134 prologue and epilogue that realigns the runtime stack if necessary.
3135 This supports mixing legacy codes that run with a 4-byte aligned stack
3136 with modern codes that keep a 16-byte stack for SSE compatibility.
3139 @cindex @code{resbank} attribute
3140 On the SH2A target, this attribute enables the high-speed register
3141 saving and restoration using a register bank for @code{interrupt_handler}
3142 routines. Saving to the bank is performed automatically after the CPU
3143 accepts an interrupt that uses a register bank.
3145 The nineteen 32-bit registers comprising general register R0 to R14,
3146 control register GBR, and system registers MACH, MACL, and PR and the
3147 vector table address offset are saved into a register bank. Register
3148 banks are stacked in first-in last-out (FILO) sequence. Restoration
3149 from the bank is executed by issuing a RESBANK instruction.
3152 @cindex @code{returns_twice} attribute
3153 The @code{returns_twice} attribute tells the compiler that a function may
3154 return more than one time. The compiler will ensure that all registers
3155 are dead before calling such a function and will emit a warning about
3156 the variables that may be clobbered after the second return from the
3157 function. Examples of such functions are @code{setjmp} and @code{vfork}.
3158 The @code{longjmp}-like counterpart of such function, if any, might need
3159 to be marked with the @code{noreturn} attribute.
3162 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
3163 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
3164 all registers except the stack pointer should be saved in the prologue
3165 regardless of whether they are used or not.
3167 @item save_volatiles
3168 @cindex save volatile registers on the MicroBlaze
3169 Use this attribute on the MicroBlaze to indicate that the function is
3170 an interrupt handler. All volatile registers (in addition to non-volatile
3171 registers) will be saved in the function prologue. If the function is a leaf
3172 function, only volatiles used by the function are saved. A normal function
3173 return is generated instead of a return from interrupt.
3175 @item section ("@var{section-name}")
3176 @cindex @code{section} function attribute
3177 Normally, the compiler places the code it generates in the @code{text} section.
3178 Sometimes, however, you need additional sections, or you need certain
3179 particular functions to appear in special sections. The @code{section}
3180 attribute specifies that a function lives in a particular section.
3181 For example, the declaration:
3184 extern void foobar (void) __attribute__ ((section ("bar")));
3188 puts the function @code{foobar} in the @code{bar} section.
3190 Some file formats do not support arbitrary sections so the @code{section}
3191 attribute is not available on all platforms.
3192 If you need to map the entire contents of a module to a particular
3193 section, consider using the facilities of the linker instead.
3196 @cindex @code{sentinel} function attribute
3197 This function attribute ensures that a parameter in a function call is
3198 an explicit @code{NULL}. The attribute is only valid on variadic
3199 functions. By default, the sentinel is located at position zero, the
3200 last parameter of the function call. If an optional integer position
3201 argument P is supplied to the attribute, the sentinel must be located at
3202 position P counting backwards from the end of the argument list.
3205 __attribute__ ((sentinel))
3207 __attribute__ ((sentinel(0)))
3210 The attribute is automatically set with a position of 0 for the built-in
3211 functions @code{execl} and @code{execlp}. The built-in function
3212 @code{execle} has the attribute set with a position of 1.
3214 A valid @code{NULL} in this context is defined as zero with any pointer
3215 type. If your system defines the @code{NULL} macro with an integer type
3216 then you need to add an explicit cast. GCC replaces @code{stddef.h}
3217 with a copy that redefines NULL appropriately.
3219 The warnings for missing or incorrect sentinels are enabled with
3223 See long_call/short_call.
3226 See longcall/shortcall.
3229 @cindex signal handler functions on the AVR processors
3230 Use this attribute on the AVR to indicate that the specified
3231 function is a signal handler. The compiler will generate function
3232 entry and exit sequences suitable for use in a signal handler when this
3233 attribute is present. Interrupts will be disabled inside the function.
3236 Use this attribute on the SH to indicate an @code{interrupt_handler}
3237 function should switch to an alternate stack. It expects a string
3238 argument that names a global variable holding the address of the
3243 void f () __attribute__ ((interrupt_handler,
3244 sp_switch ("alt_stack")));
3248 @cindex functions that pop the argument stack on the 386
3249 On the Intel 386, the @code{stdcall} attribute causes the compiler to
3250 assume that the called function will pop off the stack space used to
3251 pass arguments, unless it takes a variable number of arguments.
3253 @item syscall_linkage
3254 @cindex @code{syscall_linkage} attribute
3255 This attribute is used to modify the IA64 calling convention by marking
3256 all input registers as live at all function exits. This makes it possible
3257 to restart a system call after an interrupt without having to save/restore
3258 the input registers. This also prevents kernel data from leaking into
3262 @cindex @code{target} function attribute
3263 The @code{target} attribute is used to specify that a function is to
3264 be compiled with different target options than specified on the
3265 command line. This can be used for instance to have functions
3266 compiled with a different ISA (instruction set architecture) than the
3267 default. You can also use the @samp{#pragma GCC target} pragma to set
3268 more than one function to be compiled with specific target options.
3269 @xref{Function Specific Option Pragmas}, for details about the
3270 @samp{#pragma GCC target} pragma.
3272 For instance on a 386, you could compile one function with
3273 @code{target("sse4.1,arch=core2")} and another with
3274 @code{target("sse4a,arch=amdfam10")} that would be equivalent to
3275 compiling the first function with @option{-msse4.1} and
3276 @option{-march=core2} options, and the second function with
3277 @option{-msse4a} and @option{-march=amdfam10} options. It is up to the
3278 user to make sure that a function is only invoked on a machine that
3279 supports the particular ISA it was compiled for (for example by using
3280 @code{cpuid} on 386 to determine what feature bits and architecture
3284 int core2_func (void) __attribute__ ((__target__ ("arch=core2")));
3285 int sse3_func (void) __attribute__ ((__target__ ("sse3")));
3289 @item i386 target attributes
3290 On the 386, the following options are allowed:
3295 @cindex @code{target("abm")} attribute
3296 Enable/disable the generation of the advanced bit instructions.
3300 @cindex @code{target("aes")} attribute
3301 Enable/disable the generation of the AES instructions.
3305 @cindex @code{target("mmx")} attribute
3306 Enable/disable the generation of the MMX instructions.
3310 @cindex @code{target("pclmul")} attribute
3311 Enable/disable the generation of the PCLMUL instructions.
3315 @cindex @code{target("popcnt")} attribute
3316 Enable/disable the generation of the POPCNT instruction.
3320 @cindex @code{target("sse")} attribute
3321 Enable/disable the generation of the SSE instructions.
3325 @cindex @code{target("sse2")} attribute
3326 Enable/disable the generation of the SSE2 instructions.
3330 @cindex @code{target("sse3")} attribute
3331 Enable/disable the generation of the SSE3 instructions.
3335 @cindex @code{target("sse4")} attribute
3336 Enable/disable the generation of the SSE4 instructions (both SSE4.1
3341 @cindex @code{target("sse4.1")} attribute
3342 Enable/disable the generation of the sse4.1 instructions.
3346 @cindex @code{target("sse4.2")} attribute
3347 Enable/disable the generation of the sse4.2 instructions.
3351 @cindex @code{target("sse4a")} attribute
3352 Enable/disable the generation of the SSE4A instructions.
3356 @cindex @code{target("fma4")} attribute
3357 Enable/disable the generation of the FMA4 instructions.
3361 @cindex @code{target("xop")} attribute
3362 Enable/disable the generation of the XOP instructions.
3366 @cindex @code{target("lwp")} attribute
3367 Enable/disable the generation of the LWP instructions.
3371 @cindex @code{target("ssse3")} attribute
3372 Enable/disable the generation of the SSSE3 instructions.
3376 @cindex @code{target("cld")} attribute
3377 Enable/disable the generation of the CLD before string moves.
3379 @item fancy-math-387
3380 @itemx no-fancy-math-387
3381 @cindex @code{target("fancy-math-387")} attribute
3382 Enable/disable the generation of the @code{sin}, @code{cos}, and
3383 @code{sqrt} instructions on the 387 floating point unit.
3386 @itemx no-fused-madd
3387 @cindex @code{target("fused-madd")} attribute
3388 Enable/disable the generation of the fused multiply/add instructions.
3392 @cindex @code{target("ieee-fp")} attribute
3393 Enable/disable the generation of floating point that depends on IEEE arithmetic.
3395 @item inline-all-stringops
3396 @itemx no-inline-all-stringops
3397 @cindex @code{target("inline-all-stringops")} attribute
3398 Enable/disable inlining of string operations.
3400 @item inline-stringops-dynamically
3401 @itemx no-inline-stringops-dynamically
3402 @cindex @code{target("inline-stringops-dynamically")} attribute
3403 Enable/disable the generation of the inline code to do small string
3404 operations and calling the library routines for large operations.
3406 @item align-stringops
3407 @itemx no-align-stringops
3408 @cindex @code{target("align-stringops")} attribute
3409 Do/do not align destination of inlined string operations.
3413 @cindex @code{target("recip")} attribute
3414 Enable/disable the generation of RCPSS, RCPPS, RSQRTSS and RSQRTPS
3415 instructions followed an additional Newton-Raphson step instead of
3416 doing a floating point division.
3418 @item arch=@var{ARCH}
3419 @cindex @code{target("arch=@var{ARCH}")} attribute
3420 Specify the architecture to generate code for in compiling the function.
3422 @item tune=@var{TUNE}
3423 @cindex @code{target("tune=@var{TUNE}")} attribute
3424 Specify the architecture to tune for in compiling the function.
3426 @item fpmath=@var{FPMATH}
3427 @cindex @code{target("fpmath=@var{FPMATH}")} attribute
3428 Specify which floating point unit to use. The
3429 @code{target("fpmath=sse,387")} option must be specified as
3430 @code{target("fpmath=sse+387")} because the comma would separate
3433 @item PowerPC target attributes
3434 On the PowerPC, the following options are allowed:
3439 @cindex @code{target("altivec")} attribute
3440 Generate code that uses (does not use) AltiVec instructions. In
3441 32-bit code, you cannot enable Altivec instructions unless
3442 @option{-mabi=altivec} was used on the command line.
3446 @cindex @code{target("cmpb")} attribute
3447 Generate code that uses (does not use) the compare bytes instruction
3448 implemented on the POWER6 processor and other processors that support
3449 the PowerPC V2.05 architecture.
3453 @cindex @code{target("dlmzb")} attribute
3454 Generate code that uses (does not use) the string-search @samp{dlmzb}
3455 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
3456 generated by default when targetting those processors.
3460 @cindex @code{target("fprnd")} attribute
3461 Generate code that uses (does not use) the FP round to integer
3462 instructions implemented on the POWER5+ processor and other processors
3463 that support the PowerPC V2.03 architecture.
3467 @cindex @code{target("hard-dfp")} attribute
3468 Generate code that uses (does not use) the decimal floating point
3469 instructions implemented on some POWER processors.
3473 @cindex @code{target("isel")} attribute
3474 Generate code that uses (does not use) ISEL instruction.
3478 @cindex @code{target("mfcrf")} attribute
3479 Generate code that uses (does not use) the move from condition
3480 register field instruction implemented on the POWER4 processor and
3481 other processors that support the PowerPC V2.01 architecture.
3485 @cindex @code{target("mfpgpr")} attribute
3486 Generate code that uses (does not use) the FP move to/from general
3487 purpose register instructions implemented on the POWER6X processor and
3488 other processors that support the extended PowerPC V2.05 architecture.
3492 @cindex @code{target("mulhw")} attribute
3493 Generate code that uses (does not use) the half-word multiply and
3494 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
3495 These instructions are generated by default when targetting those
3500 @cindex @code{target("multiple")} attribute
3501 Generate code that uses (does not use) the load multiple word
3502 instructions and the store multiple word instructions.
3506 @cindex @code{target("update")} attribute
3507 Generate code that uses (does not use) the load or store instructions
3508 that update the base register to the address of the calculated memory
3513 @cindex @code{target("popcntb")} attribute
3514 Generate code that uses (does not use) the popcount and double
3515 precision FP reciprocal estimate instruction implemented on the POWER5
3516 processor and other processors that support the PowerPC V2.02
3521 @cindex @code{target("popcntd")} attribute
3522 Generate code that uses (does not use) the popcount instruction
3523 implemented on the POWER7 processor and other processors that support
3524 the PowerPC V2.06 architecture.
3526 @item powerpc-gfxopt
3527 @itemx no-powerpc-gfxopt
3528 @cindex @code{target("powerpc-gfxopt")} attribute
3529 Generate code that uses (does not use) the optional PowerPC
3530 architecture instructions in the Graphics group, including
3531 floating-point select.
3534 @itemx no-powerpc-gpopt
3535 @cindex @code{target("powerpc-gpopt")} attribute
3536 Generate code that uses (does not use) the optional PowerPC
3537 architecture instructions in the General Purpose group, including
3538 floating-point square root.
3540 @item recip-precision
3541 @itemx no-recip-precision
3542 @cindex @code{target("recip-precision")} attribute
3543 Assume (do not assume) that the reciprocal estimate instructions
3544 provide higher precision estimates than is mandated by the powerpc
3549 @cindex @code{target("string")} attribute
3550 Generate code that uses (does not use) the load string instructions
3551 and the store string word instructions to save multiple registers and
3552 do small block moves.
3556 @cindex @code{target("vsx")} attribute
3557 Generate code that uses (does not use) vector/scalar (VSX)
3558 instructions, and also enable the use of built-in functions that allow
3559 more direct access to the VSX instruction set. In 32-bit code, you
3560 cannot enable VSX or Altivec instructions unless
3561 @option{-mabi=altivec} was used on the command line.
3565 @cindex @code{target("friz")} attribute
3566 Generate (do not generate) the @code{friz} instruction when the
3567 @option{-funsafe-math-optimizations} option is used to optimize
3568 rounding a floating point value to 64-bit integer and back to floating
3569 point. The @code{friz} instruction does not return the same value if
3570 the floating point number is too large to fit in an integer.
3572 @item avoid-indexed-addresses
3573 @itemx no-avoid-indexed-addresses
3574 @cindex @code{target("avoid-indexed-addresses")} attribute
3575 Generate code that tries to avoid (not avoid) the use of indexed load
3576 or store instructions.
3580 @cindex @code{target("paired")} attribute
3581 Generate code that uses (does not use) the generation of PAIRED simd
3586 @cindex @code{target("longcall")} attribute
3587 Generate code that assumes (does not assume) that all calls are far
3588 away so that a longer more expensive calling sequence is required.
3591 @cindex @code{target("cpu=@var{CPU}")} attribute
3592 Specify the architecture to generate code for in compiling the
3593 function. If you select @code{"target("cpu=power7)"} attribute when
3594 generating 32-bit code, VSX and Altivec instructions are not generated
3595 unless you use the @option{-mabi=altivec} option on the command line.
3597 @item tune=@var{TUNE}
3598 @cindex @code{target("tune=@var{TUNE}")} attribute
3599 Specify the architecture to tune for in compiling the function. If
3600 you do not specify the @code{target("tune=@var{TUNE}")} attribute and
3601 you do specifiy the @code{target("cpu=@var{CPU}")} attribute,
3602 compilation will tune for the @var{CPU} architecture, and not the
3603 default tuning specified on the command line.
3608 On the 386/x86_64 and PowerPC backends, you can use either multiple
3609 strings to specify multiple options, or you can separate the option
3610 with a comma (@code{,}).
3612 On the 386/x86_64 and PowerPC backends, the inliner will not inline a
3613 function that has different target options than the caller, unless the
3614 callee has a subset of the target options of the caller. For example
3615 a function declared with @code{target("sse3")} can inline a function
3616 with @code{target("sse2")}, since @code{-msse3} implies @code{-msse2}.
3618 The @code{target} attribute is not implemented in GCC versions earlier
3619 than 4.4 for the i386/x86_64 and 4.6 for the PowerPC backends. It is
3620 not currently implemented for other backends.
3623 @cindex tiny data section on the H8/300H and H8S
3624 Use this attribute on the H8/300H and H8S to indicate that the specified
3625 variable should be placed into the tiny data section.
3626 The compiler will generate more efficient code for loads and stores
3627 on data in the tiny data section. Note the tiny data area is limited to
3628 slightly under 32kbytes of data.
3631 Use this attribute on the SH for an @code{interrupt_handler} to return using
3632 @code{trapa} instead of @code{rte}. This attribute expects an integer
3633 argument specifying the trap number to be used.
3636 @cindex @code{unused} attribute.
3637 This attribute, attached to a function, means that the function is meant
3638 to be possibly unused. GCC will not produce a warning for this
3642 @cindex @code{used} attribute.
3643 This attribute, attached to a function, means that code must be emitted
3644 for the function even if it appears that the function is not referenced.
3645 This is useful, for example, when the function is referenced only in
3649 @cindex @code{version_id} attribute
3650 This IA64 HP-UX attribute, attached to a global variable or function, renames a
3651 symbol to contain a version string, thus allowing for function level
3652 versioning. HP-UX system header files may use version level functioning
3653 for some system calls.
3656 extern int foo () __attribute__((version_id ("20040821")));
3659 Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}.
3661 @item visibility ("@var{visibility_type}")
3662 @cindex @code{visibility} attribute
3663 This attribute affects the linkage of the declaration to which it is attached.
3664 There are four supported @var{visibility_type} values: default,
3665 hidden, protected or internal visibility.
3668 void __attribute__ ((visibility ("protected")))
3669 f () @{ /* @r{Do something.} */; @}
3670 int i __attribute__ ((visibility ("hidden")));
3673 The possible values of @var{visibility_type} correspond to the
3674 visibility settings in the ELF gABI.
3677 @c keep this list of visibilities in alphabetical order.
3680 Default visibility is the normal case for the object file format.
3681 This value is available for the visibility attribute to override other
3682 options that may change the assumed visibility of entities.
3684 On ELF, default visibility means that the declaration is visible to other
3685 modules and, in shared libraries, means that the declared entity may be
3688 On Darwin, default visibility means that the declaration is visible to
3691 Default visibility corresponds to ``external linkage'' in the language.
3694 Hidden visibility indicates that the entity declared will have a new
3695 form of linkage, which we'll call ``hidden linkage''. Two
3696 declarations of an object with hidden linkage refer to the same object
3697 if they are in the same shared object.
3700 Internal visibility is like hidden visibility, but with additional
3701 processor specific semantics. Unless otherwise specified by the
3702 psABI, GCC defines internal visibility to mean that a function is
3703 @emph{never} called from another module. Compare this with hidden
3704 functions which, while they cannot be referenced directly by other
3705 modules, can be referenced indirectly via function pointers. By
3706 indicating that a function cannot be called from outside the module,
3707 GCC may for instance omit the load of a PIC register since it is known
3708 that the calling function loaded the correct value.
3711 Protected visibility is like default visibility except that it
3712 indicates that references within the defining module will bind to the
3713 definition in that module. That is, the declared entity cannot be
3714 overridden by another module.
3718 All visibilities are supported on many, but not all, ELF targets
3719 (supported when the assembler supports the @samp{.visibility}
3720 pseudo-op). Default visibility is supported everywhere. Hidden
3721 visibility is supported on Darwin targets.
3723 The visibility attribute should be applied only to declarations which
3724 would otherwise have external linkage. The attribute should be applied
3725 consistently, so that the same entity should not be declared with
3726 different settings of the attribute.
3728 In C++, the visibility attribute applies to types as well as functions
3729 and objects, because in C++ types have linkage. A class must not have
3730 greater visibility than its non-static data member types and bases,
3731 and class members default to the visibility of their class. Also, a
3732 declaration without explicit visibility is limited to the visibility
3735 In C++, you can mark member functions and static member variables of a
3736 class with the visibility attribute. This is useful if you know a
3737 particular method or static member variable should only be used from
3738 one shared object; then you can mark it hidden while the rest of the
3739 class has default visibility. Care must be taken to avoid breaking
3740 the One Definition Rule; for example, it is usually not useful to mark
3741 an inline method as hidden without marking the whole class as hidden.
3743 A C++ namespace declaration can also have the visibility attribute.
3744 This attribute applies only to the particular namespace body, not to
3745 other definitions of the same namespace; it is equivalent to using
3746 @samp{#pragma GCC visibility} before and after the namespace
3747 definition (@pxref{Visibility Pragmas}).
3749 In C++, if a template argument has limited visibility, this
3750 restriction is implicitly propagated to the template instantiation.
3751 Otherwise, template instantiations and specializations default to the
3752 visibility of their template.
3754 If both the template and enclosing class have explicit visibility, the
3755 visibility from the template is used.
3758 @cindex @code{vliw} attribute
3759 On MeP, the @code{vliw} attribute tells the compiler to emit
3760 instructions in VLIW mode instead of core mode. Note that this
3761 attribute is not allowed unless a VLIW coprocessor has been configured
3762 and enabled through command line options.
3764 @item warn_unused_result
3765 @cindex @code{warn_unused_result} attribute
3766 The @code{warn_unused_result} attribute causes a warning to be emitted
3767 if a caller of the function with this attribute does not use its
3768 return value. This is useful for functions where not checking
3769 the result is either a security problem or always a bug, such as
3773 int fn () __attribute__ ((warn_unused_result));
3776 if (fn () < 0) return -1;
3782 results in warning on line 5.
3785 @cindex @code{weak} attribute
3786 The @code{weak} attribute causes the declaration to be emitted as a weak
3787 symbol rather than a global. This is primarily useful in defining
3788 library functions which can be overridden in user code, though it can
3789 also be used with non-function declarations. Weak symbols are supported
3790 for ELF targets, and also for a.out targets when using the GNU assembler
3794 @itemx weakref ("@var{target}")
3795 @cindex @code{weakref} attribute
3796 The @code{weakref} attribute marks a declaration as a weak reference.
3797 Without arguments, it should be accompanied by an @code{alias} attribute
3798 naming the target symbol. Optionally, the @var{target} may be given as
3799 an argument to @code{weakref} itself. In either case, @code{weakref}
3800 implicitly marks the declaration as @code{weak}. Without a
3801 @var{target}, given as an argument to @code{weakref} or to @code{alias},
3802 @code{weakref} is equivalent to @code{weak}.
3805 static int x() __attribute__ ((weakref ("y")));
3806 /* is equivalent to... */
3807 static int x() __attribute__ ((weak, weakref, alias ("y")));
3809 static int x() __attribute__ ((weakref));
3810 static int x() __attribute__ ((alias ("y")));
3813 A weak reference is an alias that does not by itself require a
3814 definition to be given for the target symbol. If the target symbol is
3815 only referenced through weak references, then it becomes a @code{weak}
3816 undefined symbol. If it is directly referenced, however, then such
3817 strong references prevail, and a definition will be required for the
3818 symbol, not necessarily in the same translation unit.
3820 The effect is equivalent to moving all references to the alias to a
3821 separate translation unit, renaming the alias to the aliased symbol,
3822 declaring it as weak, compiling the two separate translation units and
3823 performing a reloadable link on them.
3825 At present, a declaration to which @code{weakref} is attached can
3826 only be @code{static}.
3830 You can specify multiple attributes in a declaration by separating them
3831 by commas within the double parentheses or by immediately following an
3832 attribute declaration with another attribute declaration.
3834 @cindex @code{#pragma}, reason for not using
3835 @cindex pragma, reason for not using
3836 Some people object to the @code{__attribute__} feature, suggesting that
3837 ISO C's @code{#pragma} should be used instead. At the time
3838 @code{__attribute__} was designed, there were two reasons for not doing
3843 It is impossible to generate @code{#pragma} commands from a macro.
3846 There is no telling what the same @code{#pragma} might mean in another
3850 These two reasons applied to almost any application that might have been
3851 proposed for @code{#pragma}. It was basically a mistake to use
3852 @code{#pragma} for @emph{anything}.
3854 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
3855 to be generated from macros. In addition, a @code{#pragma GCC}
3856 namespace is now in use for GCC-specific pragmas. However, it has been
3857 found convenient to use @code{__attribute__} to achieve a natural
3858 attachment of attributes to their corresponding declarations, whereas
3859 @code{#pragma GCC} is of use for constructs that do not naturally form
3860 part of the grammar. @xref{Other Directives,,Miscellaneous
3861 Preprocessing Directives, cpp, The GNU C Preprocessor}.
3863 @node Attribute Syntax
3864 @section Attribute Syntax
3865 @cindex attribute syntax
3867 This section describes the syntax with which @code{__attribute__} may be
3868 used, and the constructs to which attribute specifiers bind, for the C
3869 language. Some details may vary for C++ and Objective-C@. Because of
3870 infelicities in the grammar for attributes, some forms described here
3871 may not be successfully parsed in all cases.
3873 There are some problems with the semantics of attributes in C++. For
3874 example, there are no manglings for attributes, although they may affect
3875 code generation, so problems may arise when attributed types are used in
3876 conjunction with templates or overloading. Similarly, @code{typeid}
3877 does not distinguish between types with different attributes. Support
3878 for attributes in C++ may be restricted in future to attributes on
3879 declarations only, but not on nested declarators.
3881 @xref{Function Attributes}, for details of the semantics of attributes
3882 applying to functions. @xref{Variable Attributes}, for details of the
3883 semantics of attributes applying to variables. @xref{Type Attributes},
3884 for details of the semantics of attributes applying to structure, union
3885 and enumerated types.
3887 An @dfn{attribute specifier} is of the form
3888 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
3889 is a possibly empty comma-separated sequence of @dfn{attributes}, where
3890 each attribute is one of the following:
3894 Empty. Empty attributes are ignored.
3897 A word (which may be an identifier such as @code{unused}, or a reserved
3898 word such as @code{const}).
3901 A word, followed by, in parentheses, parameters for the attribute.
3902 These parameters take one of the following forms:
3906 An identifier. For example, @code{mode} attributes use this form.
3909 An identifier followed by a comma and a non-empty comma-separated list
3910 of expressions. For example, @code{format} attributes use this form.
3913 A possibly empty comma-separated list of expressions. For example,
3914 @code{format_arg} attributes use this form with the list being a single
3915 integer constant expression, and @code{alias} attributes use this form
3916 with the list being a single string constant.
3920 An @dfn{attribute specifier list} is a sequence of one or more attribute
3921 specifiers, not separated by any other tokens.
3923 In GNU C, an attribute specifier list may appear after the colon following a
3924 label, other than a @code{case} or @code{default} label. The only
3925 attribute it makes sense to use after a label is @code{unused}. This
3926 feature is intended for code generated by programs which contains labels
3927 that may be unused but which is compiled with @option{-Wall}. It would
3928 not normally be appropriate to use in it human-written code, though it
3929 could be useful in cases where the code that jumps to the label is
3930 contained within an @code{#ifdef} conditional. GNU C++ only permits
3931 attributes on labels if the attribute specifier is immediately
3932 followed by a semicolon (i.e., the label applies to an empty
3933 statement). If the semicolon is missing, C++ label attributes are
3934 ambiguous, as it is permissible for a declaration, which could begin
3935 with an attribute list, to be labelled in C++. Declarations cannot be
3936 labelled in C90 or C99, so the ambiguity does not arise there.
3938 An attribute specifier list may appear as part of a @code{struct},
3939 @code{union} or @code{enum} specifier. It may go either immediately
3940 after the @code{struct}, @code{union} or @code{enum} keyword, or after
3941 the closing brace. The former syntax is preferred.
3942 Where attribute specifiers follow the closing brace, they are considered
3943 to relate to the structure, union or enumerated type defined, not to any
3944 enclosing declaration the type specifier appears in, and the type
3945 defined is not complete until after the attribute specifiers.
3946 @c Otherwise, there would be the following problems: a shift/reduce
3947 @c conflict between attributes binding the struct/union/enum and
3948 @c binding to the list of specifiers/qualifiers; and "aligned"
3949 @c attributes could use sizeof for the structure, but the size could be
3950 @c changed later by "packed" attributes.
3952 Otherwise, an attribute specifier appears as part of a declaration,
3953 counting declarations of unnamed parameters and type names, and relates
3954 to that declaration (which may be nested in another declaration, for
3955 example in the case of a parameter declaration), or to a particular declarator
3956 within a declaration. Where an
3957 attribute specifier is applied to a parameter declared as a function or
3958 an array, it should apply to the function or array rather than the
3959 pointer to which the parameter is implicitly converted, but this is not
3960 yet correctly implemented.
3962 Any list of specifiers and qualifiers at the start of a declaration may
3963 contain attribute specifiers, whether or not such a list may in that
3964 context contain storage class specifiers. (Some attributes, however,
3965 are essentially in the nature of storage class specifiers, and only make
3966 sense where storage class specifiers may be used; for example,
3967 @code{section}.) There is one necessary limitation to this syntax: the
3968 first old-style parameter declaration in a function definition cannot
3969 begin with an attribute specifier, because such an attribute applies to
3970 the function instead by syntax described below (which, however, is not
3971 yet implemented in this case). In some other cases, attribute
3972 specifiers are permitted by this grammar but not yet supported by the
3973 compiler. All attribute specifiers in this place relate to the
3974 declaration as a whole. In the obsolescent usage where a type of
3975 @code{int} is implied by the absence of type specifiers, such a list of
3976 specifiers and qualifiers may be an attribute specifier list with no
3977 other specifiers or qualifiers.
3979 At present, the first parameter in a function prototype must have some
3980 type specifier which is not an attribute specifier; this resolves an
3981 ambiguity in the interpretation of @code{void f(int
3982 (__attribute__((foo)) x))}, but is subject to change. At present, if
3983 the parentheses of a function declarator contain only attributes then
3984 those attributes are ignored, rather than yielding an error or warning
3985 or implying a single parameter of type int, but this is subject to
3988 An attribute specifier list may appear immediately before a declarator
3989 (other than the first) in a comma-separated list of declarators in a
3990 declaration of more than one identifier using a single list of
3991 specifiers and qualifiers. Such attribute specifiers apply
3992 only to the identifier before whose declarator they appear. For
3996 __attribute__((noreturn)) void d0 (void),
3997 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
4002 the @code{noreturn} attribute applies to all the functions
4003 declared; the @code{format} attribute only applies to @code{d1}.
4005 An attribute specifier list may appear immediately before the comma,
4006 @code{=} or semicolon terminating the declaration of an identifier other
4007 than a function definition. Such attribute specifiers apply
4008 to the declared object or function. Where an
4009 assembler name for an object or function is specified (@pxref{Asm
4010 Labels}), the attribute must follow the @code{asm}
4013 An attribute specifier list may, in future, be permitted to appear after
4014 the declarator in a function definition (before any old-style parameter
4015 declarations or the function body).
4017 Attribute specifiers may be mixed with type qualifiers appearing inside
4018 the @code{[]} of a parameter array declarator, in the C99 construct by
4019 which such qualifiers are applied to the pointer to which the array is
4020 implicitly converted. Such attribute specifiers apply to the pointer,
4021 not to the array, but at present this is not implemented and they are
4024 An attribute specifier list may appear at the start of a nested
4025 declarator. At present, there are some limitations in this usage: the
4026 attributes correctly apply to the declarator, but for most individual
4027 attributes the semantics this implies are not implemented.
4028 When attribute specifiers follow the @code{*} of a pointer
4029 declarator, they may be mixed with any type qualifiers present.
4030 The following describes the formal semantics of this syntax. It will make the
4031 most sense if you are familiar with the formal specification of
4032 declarators in the ISO C standard.
4034 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
4035 D1}, where @code{T} contains declaration specifiers that specify a type
4036 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
4037 contains an identifier @var{ident}. The type specified for @var{ident}
4038 for derived declarators whose type does not include an attribute
4039 specifier is as in the ISO C standard.
4041 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
4042 and the declaration @code{T D} specifies the type
4043 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
4044 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
4045 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
4047 If @code{D1} has the form @code{*
4048 @var{type-qualifier-and-attribute-specifier-list} D}, and the
4049 declaration @code{T D} specifies the type
4050 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
4051 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
4052 @var{type-qualifier-and-attribute-specifier-list} pointer to @var{Type}'' for
4058 void (__attribute__((noreturn)) ****f) (void);
4062 specifies the type ``pointer to pointer to pointer to pointer to
4063 non-returning function returning @code{void}''. As another example,
4066 char *__attribute__((aligned(8))) *f;
4070 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
4071 Note again that this does not work with most attributes; for example,
4072 the usage of @samp{aligned} and @samp{noreturn} attributes given above
4073 is not yet supported.
4075 For compatibility with existing code written for compiler versions that
4076 did not implement attributes on nested declarators, some laxity is
4077 allowed in the placing of attributes. If an attribute that only applies
4078 to types is applied to a declaration, it will be treated as applying to
4079 the type of that declaration. If an attribute that only applies to
4080 declarations is applied to the type of a declaration, it will be treated
4081 as applying to that declaration; and, for compatibility with code
4082 placing the attributes immediately before the identifier declared, such
4083 an attribute applied to a function return type will be treated as
4084 applying to the function type, and such an attribute applied to an array
4085 element type will be treated as applying to the array type. If an
4086 attribute that only applies to function types is applied to a
4087 pointer-to-function type, it will be treated as applying to the pointer
4088 target type; if such an attribute is applied to a function return type
4089 that is not a pointer-to-function type, it will be treated as applying
4090 to the function type.
4092 @node Function Prototypes
4093 @section Prototypes and Old-Style Function Definitions
4094 @cindex function prototype declarations
4095 @cindex old-style function definitions
4096 @cindex promotion of formal parameters
4098 GNU C extends ISO C to allow a function prototype to override a later
4099 old-style non-prototype definition. Consider the following example:
4102 /* @r{Use prototypes unless the compiler is old-fashioned.} */
4109 /* @r{Prototype function declaration.} */
4110 int isroot P((uid_t));
4112 /* @r{Old-style function definition.} */
4114 isroot (x) /* @r{??? lossage here ???} */
4121 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
4122 not allow this example, because subword arguments in old-style
4123 non-prototype definitions are promoted. Therefore in this example the
4124 function definition's argument is really an @code{int}, which does not
4125 match the prototype argument type of @code{short}.
4127 This restriction of ISO C makes it hard to write code that is portable
4128 to traditional C compilers, because the programmer does not know
4129 whether the @code{uid_t} type is @code{short}, @code{int}, or
4130 @code{long}. Therefore, in cases like these GNU C allows a prototype
4131 to override a later old-style definition. More precisely, in GNU C, a
4132 function prototype argument type overrides the argument type specified
4133 by a later old-style definition if the former type is the same as the
4134 latter type before promotion. Thus in GNU C the above example is
4135 equivalent to the following:
4148 GNU C++ does not support old-style function definitions, so this
4149 extension is irrelevant.
4152 @section C++ Style Comments
4154 @cindex C++ comments
4155 @cindex comments, C++ style
4157 In GNU C, you may use C++ style comments, which start with @samp{//} and
4158 continue until the end of the line. Many other C implementations allow
4159 such comments, and they are included in the 1999 C standard. However,
4160 C++ style comments are not recognized if you specify an @option{-std}
4161 option specifying a version of ISO C before C99, or @option{-ansi}
4162 (equivalent to @option{-std=c90}).
4165 @section Dollar Signs in Identifier Names
4167 @cindex dollar signs in identifier names
4168 @cindex identifier names, dollar signs in
4170 In GNU C, you may normally use dollar signs in identifier names.
4171 This is because many traditional C implementations allow such identifiers.
4172 However, dollar signs in identifiers are not supported on a few target
4173 machines, typically because the target assembler does not allow them.
4175 @node Character Escapes
4176 @section The Character @key{ESC} in Constants
4178 You can use the sequence @samp{\e} in a string or character constant to
4179 stand for the ASCII character @key{ESC}.
4182 @section Inquiring on Alignment of Types or Variables
4184 @cindex type alignment
4185 @cindex variable alignment
4187 The keyword @code{__alignof__} allows you to inquire about how an object
4188 is aligned, or the minimum alignment usually required by a type. Its
4189 syntax is just like @code{sizeof}.
4191 For example, if the target machine requires a @code{double} value to be
4192 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
4193 This is true on many RISC machines. On more traditional machine
4194 designs, @code{__alignof__ (double)} is 4 or even 2.
4196 Some machines never actually require alignment; they allow reference to any
4197 data type even at an odd address. For these machines, @code{__alignof__}
4198 reports the smallest alignment that GCC will give the data type, usually as
4199 mandated by the target ABI.
4201 If the operand of @code{__alignof__} is an lvalue rather than a type,
4202 its value is the required alignment for its type, taking into account
4203 any minimum alignment specified with GCC's @code{__attribute__}
4204 extension (@pxref{Variable Attributes}). For example, after this
4208 struct foo @{ int x; char y; @} foo1;
4212 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
4213 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
4215 It is an error to ask for the alignment of an incomplete type.
4217 @node Variable Attributes
4218 @section Specifying Attributes of Variables
4219 @cindex attribute of variables
4220 @cindex variable attributes
4222 The keyword @code{__attribute__} allows you to specify special
4223 attributes of variables or structure fields. This keyword is followed
4224 by an attribute specification inside double parentheses. Some
4225 attributes are currently defined generically for variables.
4226 Other attributes are defined for variables on particular target
4227 systems. Other attributes are available for functions
4228 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
4229 Other front ends might define more attributes
4230 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
4232 You may also specify attributes with @samp{__} preceding and following
4233 each keyword. This allows you to use them in header files without
4234 being concerned about a possible macro of the same name. For example,
4235 you may use @code{__aligned__} instead of @code{aligned}.
4237 @xref{Attribute Syntax}, for details of the exact syntax for using
4241 @cindex @code{aligned} attribute
4242 @item aligned (@var{alignment})
4243 This attribute specifies a minimum alignment for the variable or
4244 structure field, measured in bytes. For example, the declaration:
4247 int x __attribute__ ((aligned (16))) = 0;
4251 causes the compiler to allocate the global variable @code{x} on a
4252 16-byte boundary. On a 68040, this could be used in conjunction with
4253 an @code{asm} expression to access the @code{move16} instruction which
4254 requires 16-byte aligned operands.
4256 You can also specify the alignment of structure fields. For example, to
4257 create a double-word aligned @code{int} pair, you could write:
4260 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
4264 This is an alternative to creating a union with a @code{double} member
4265 that forces the union to be double-word aligned.
4267 As in the preceding examples, you can explicitly specify the alignment
4268 (in bytes) that you wish the compiler to use for a given variable or
4269 structure field. Alternatively, you can leave out the alignment factor
4270 and just ask the compiler to align a variable or field to the
4271 default alignment for the target architecture you are compiling for.
4272 The default alignment is sufficient for all scalar types, but may not be
4273 enough for all vector types on a target which supports vector operations.
4274 The default alignment is fixed for a particular target ABI.
4276 Gcc also provides a target specific macro @code{__BIGGEST_ALIGNMENT__},
4277 which is the largest alignment ever used for any data type on the
4278 target machine you are compiling for. For example, you could write:
4281 short array[3] __attribute__ ((aligned (__BIGGEST_ALIGNMENT__)));
4284 The compiler automatically sets the alignment for the declared
4285 variable or field to @code{__BIGGEST_ALIGNMENT__}. Doing this can
4286 often make copy operations more efficient, because the compiler can
4287 use whatever instructions copy the biggest chunks of memory when
4288 performing copies to or from the variables or fields that you have
4289 aligned this way. Note that the value of @code{__BIGGEST_ALIGNMENT__}
4290 may change depending on command line options.
4292 When used on a struct, or struct member, the @code{aligned} attribute can
4293 only increase the alignment; in order to decrease it, the @code{packed}
4294 attribute must be specified as well. When used as part of a typedef, the
4295 @code{aligned} attribute can both increase and decrease alignment, and
4296 specifying the @code{packed} attribute will generate a warning.
4298 Note that the effectiveness of @code{aligned} attributes may be limited
4299 by inherent limitations in your linker. On many systems, the linker is
4300 only able to arrange for variables to be aligned up to a certain maximum
4301 alignment. (For some linkers, the maximum supported alignment may
4302 be very very small.) If your linker is only able to align variables
4303 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4304 in an @code{__attribute__} will still only provide you with 8 byte
4305 alignment. See your linker documentation for further information.
4307 The @code{aligned} attribute can also be used for functions
4308 (@pxref{Function Attributes}.)
4310 @item cleanup (@var{cleanup_function})
4311 @cindex @code{cleanup} attribute
4312 The @code{cleanup} attribute runs a function when the variable goes
4313 out of scope. This attribute can only be applied to auto function
4314 scope variables; it may not be applied to parameters or variables
4315 with static storage duration. The function must take one parameter,
4316 a pointer to a type compatible with the variable. The return value
4317 of the function (if any) is ignored.
4319 If @option{-fexceptions} is enabled, then @var{cleanup_function}
4320 will be run during the stack unwinding that happens during the
4321 processing of the exception. Note that the @code{cleanup} attribute
4322 does not allow the exception to be caught, only to perform an action.
4323 It is undefined what happens if @var{cleanup_function} does not
4328 @cindex @code{common} attribute
4329 @cindex @code{nocommon} attribute
4332 The @code{common} attribute requests GCC to place a variable in
4333 ``common'' storage. The @code{nocommon} attribute requests the
4334 opposite---to allocate space for it directly.
4336 These attributes override the default chosen by the
4337 @option{-fno-common} and @option{-fcommon} flags respectively.
4340 @itemx deprecated (@var{msg})
4341 @cindex @code{deprecated} attribute
4342 The @code{deprecated} attribute results in a warning if the variable
4343 is used anywhere in the source file. This is useful when identifying
4344 variables that are expected to be removed in a future version of a
4345 program. The warning also includes the location of the declaration
4346 of the deprecated variable, to enable users to easily find further
4347 information about why the variable is deprecated, or what they should
4348 do instead. Note that the warning only occurs for uses:
4351 extern int old_var __attribute__ ((deprecated));
4353 int new_fn () @{ return old_var; @}
4356 results in a warning on line 3 but not line 2. The optional msg
4357 argument, which must be a string, will be printed in the warning if
4360 The @code{deprecated} attribute can also be used for functions and
4361 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
4363 @item mode (@var{mode})
4364 @cindex @code{mode} attribute
4365 This attribute specifies the data type for the declaration---whichever
4366 type corresponds to the mode @var{mode}. This in effect lets you
4367 request an integer or floating point type according to its width.
4369 You may also specify a mode of @samp{byte} or @samp{__byte__} to
4370 indicate the mode corresponding to a one-byte integer, @samp{word} or
4371 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
4372 or @samp{__pointer__} for the mode used to represent pointers.
4375 @cindex @code{packed} attribute
4376 The @code{packed} attribute specifies that a variable or structure field
4377 should have the smallest possible alignment---one byte for a variable,
4378 and one bit for a field, unless you specify a larger value with the
4379 @code{aligned} attribute.
4381 Here is a structure in which the field @code{x} is packed, so that it
4382 immediately follows @code{a}:
4388 int x[2] __attribute__ ((packed));
4392 @emph{Note:} The 4.1, 4.2 and 4.3 series of GCC ignore the
4393 @code{packed} attribute on bit-fields of type @code{char}. This has
4394 been fixed in GCC 4.4 but the change can lead to differences in the
4395 structure layout. See the documentation of
4396 @option{-Wpacked-bitfield-compat} for more information.
4398 @item section ("@var{section-name}")
4399 @cindex @code{section} variable attribute
4400 Normally, the compiler places the objects it generates in sections like
4401 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
4402 or you need certain particular variables to appear in special sections,
4403 for example to map to special hardware. The @code{section}
4404 attribute specifies that a variable (or function) lives in a particular
4405 section. For example, this small program uses several specific section names:
4408 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
4409 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
4410 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
4411 int init_data __attribute__ ((section ("INITDATA")));
4415 /* @r{Initialize stack pointer} */
4416 init_sp (stack + sizeof (stack));
4418 /* @r{Initialize initialized data} */
4419 memcpy (&init_data, &data, &edata - &data);
4421 /* @r{Turn on the serial ports} */
4428 Use the @code{section} attribute with
4429 @emph{global} variables and not @emph{local} variables,
4430 as shown in the example.
4432 You may use the @code{section} attribute with initialized or
4433 uninitialized global variables but the linker requires
4434 each object be defined once, with the exception that uninitialized
4435 variables tentatively go in the @code{common} (or @code{bss}) section
4436 and can be multiply ``defined''. Using the @code{section} attribute
4437 will change what section the variable goes into and may cause the
4438 linker to issue an error if an uninitialized variable has multiple
4439 definitions. You can force a variable to be initialized with the
4440 @option{-fno-common} flag or the @code{nocommon} attribute.
4442 Some file formats do not support arbitrary sections so the @code{section}
4443 attribute is not available on all platforms.
4444 If you need to map the entire contents of a module to a particular
4445 section, consider using the facilities of the linker instead.
4448 @cindex @code{shared} variable attribute
4449 On Microsoft Windows, in addition to putting variable definitions in a named
4450 section, the section can also be shared among all running copies of an
4451 executable or DLL@. For example, this small program defines shared data
4452 by putting it in a named section @code{shared} and marking the section
4456 int foo __attribute__((section ("shared"), shared)) = 0;
4461 /* @r{Read and write foo. All running
4462 copies see the same value.} */
4468 You may only use the @code{shared} attribute along with @code{section}
4469 attribute with a fully initialized global definition because of the way
4470 linkers work. See @code{section} attribute for more information.
4472 The @code{shared} attribute is only available on Microsoft Windows@.
4474 @item tls_model ("@var{tls_model}")
4475 @cindex @code{tls_model} attribute
4476 The @code{tls_model} attribute sets thread-local storage model
4477 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
4478 overriding @option{-ftls-model=} command-line switch on a per-variable
4480 The @var{tls_model} argument should be one of @code{global-dynamic},
4481 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
4483 Not all targets support this attribute.
4486 This attribute, attached to a variable, means that the variable is meant
4487 to be possibly unused. GCC will not produce a warning for this
4491 This attribute, attached to a variable, means that the variable must be
4492 emitted even if it appears that the variable is not referenced.
4494 @item vector_size (@var{bytes})
4495 This attribute specifies the vector size for the variable, measured in
4496 bytes. For example, the declaration:
4499 int foo __attribute__ ((vector_size (16)));
4503 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
4504 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
4505 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
4507 This attribute is only applicable to integral and float scalars,
4508 although arrays, pointers, and function return values are allowed in
4509 conjunction with this construct.
4511 Aggregates with this attribute are invalid, even if they are of the same
4512 size as a corresponding scalar. For example, the declaration:
4515 struct S @{ int a; @};
4516 struct S __attribute__ ((vector_size (16))) foo;
4520 is invalid even if the size of the structure is the same as the size of
4524 The @code{selectany} attribute causes an initialized global variable to
4525 have link-once semantics. When multiple definitions of the variable are
4526 encountered by the linker, the first is selected and the remainder are
4527 discarded. Following usage by the Microsoft compiler, the linker is told
4528 @emph{not} to warn about size or content differences of the multiple
4531 Although the primary usage of this attribute is for POD types, the
4532 attribute can also be applied to global C++ objects that are initialized
4533 by a constructor. In this case, the static initialization and destruction
4534 code for the object is emitted in each translation defining the object,
4535 but the calls to the constructor and destructor are protected by a
4536 link-once guard variable.
4538 The @code{selectany} attribute is only available on Microsoft Windows
4539 targets. You can use @code{__declspec (selectany)} as a synonym for
4540 @code{__attribute__ ((selectany))} for compatibility with other
4544 The @code{weak} attribute is described in @ref{Function Attributes}.
4547 The @code{dllimport} attribute is described in @ref{Function Attributes}.
4550 The @code{dllexport} attribute is described in @ref{Function Attributes}.
4554 @subsection Blackfin Variable Attributes
4556 Three attributes are currently defined for the Blackfin.
4562 @cindex @code{l1_data} variable attribute
4563 @cindex @code{l1_data_A} variable attribute
4564 @cindex @code{l1_data_B} variable attribute
4565 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
4566 Variables with @code{l1_data} attribute will be put into the specific section
4567 named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into
4568 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
4569 attribute will be put into the specific section named @code{.l1.data.B}.
4572 @cindex @code{l2} variable attribute
4573 Use this attribute on the Blackfin to place the variable into L2 SRAM.
4574 Variables with @code{l2} attribute will be put into the specific section
4575 named @code{.l2.data}.
4578 @subsection M32R/D Variable Attributes
4580 One attribute is currently defined for the M32R/D@.
4583 @item model (@var{model-name})
4584 @cindex variable addressability on the M32R/D
4585 Use this attribute on the M32R/D to set the addressability of an object.
4586 The identifier @var{model-name} is one of @code{small}, @code{medium},
4587 or @code{large}, representing each of the code models.
4589 Small model objects live in the lower 16MB of memory (so that their
4590 addresses can be loaded with the @code{ld24} instruction).
4592 Medium and large model objects may live anywhere in the 32-bit address space
4593 (the compiler will generate @code{seth/add3} instructions to load their
4597 @anchor{MeP Variable Attributes}
4598 @subsection MeP Variable Attributes
4600 The MeP target has a number of addressing modes and busses. The
4601 @code{near} space spans the standard memory space's first 16 megabytes
4602 (24 bits). The @code{far} space spans the entire 32-bit memory space.
4603 The @code{based} space is a 128 byte region in the memory space which
4604 is addressed relative to the @code{$tp} register. The @code{tiny}
4605 space is a 65536 byte region relative to the @code{$gp} register. In
4606 addition to these memory regions, the MeP target has a separate 16-bit
4607 control bus which is specified with @code{cb} attributes.
4612 Any variable with the @code{based} attribute will be assigned to the
4613 @code{.based} section, and will be accessed with relative to the
4614 @code{$tp} register.
4617 Likewise, the @code{tiny} attribute assigned variables to the
4618 @code{.tiny} section, relative to the @code{$gp} register.
4621 Variables with the @code{near} attribute are assumed to have addresses
4622 that fit in a 24-bit addressing mode. This is the default for large
4623 variables (@code{-mtiny=4} is the default) but this attribute can
4624 override @code{-mtiny=} for small variables, or override @code{-ml}.
4627 Variables with the @code{far} attribute are addressed using a full
4628 32-bit address. Since this covers the entire memory space, this
4629 allows modules to make no assumptions about where variables might be
4633 @itemx io (@var{addr})
4634 Variables with the @code{io} attribute are used to address
4635 memory-mapped peripherals. If an address is specified, the variable
4636 is assigned that address, else it is not assigned an address (it is
4637 assumed some other module will assign an address). Example:
4640 int timer_count __attribute__((io(0x123)));
4644 @itemx cb (@var{addr})
4645 Variables with the @code{cb} attribute are used to access the control
4646 bus, using special instructions. @code{addr} indicates the control bus
4650 int cpu_clock __attribute__((cb(0x123)));
4655 @anchor{i386 Variable Attributes}
4656 @subsection i386 Variable Attributes
4658 Two attributes are currently defined for i386 configurations:
4659 @code{ms_struct} and @code{gcc_struct}
4664 @cindex @code{ms_struct} attribute
4665 @cindex @code{gcc_struct} attribute
4667 If @code{packed} is used on a structure, or if bit-fields are used
4668 it may be that the Microsoft ABI packs them differently
4669 than GCC would normally pack them. Particularly when moving packed
4670 data between functions compiled with GCC and the native Microsoft compiler
4671 (either via function call or as data in a file), it may be necessary to access
4674 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4675 compilers to match the native Microsoft compiler.
4677 The Microsoft structure layout algorithm is fairly simple with the exception
4678 of the bitfield packing:
4680 The padding and alignment of members of structures and whether a bit field
4681 can straddle a storage-unit boundary
4684 @item Structure members are stored sequentially in the order in which they are
4685 declared: the first member has the lowest memory address and the last member
4688 @item Every data object has an alignment-requirement. The alignment-requirement
4689 for all data except structures, unions, and arrays is either the size of the
4690 object or the current packing size (specified with either the aligned attribute
4691 or the pack pragma), whichever is less. For structures, unions, and arrays,
4692 the alignment-requirement is the largest alignment-requirement of its members.
4693 Every object is allocated an offset so that:
4695 offset % alignment-requirement == 0
4697 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
4698 unit if the integral types are the same size and if the next bit field fits
4699 into the current allocation unit without crossing the boundary imposed by the
4700 common alignment requirements of the bit fields.
4703 Handling of zero-length bitfields:
4705 MSVC interprets zero-length bitfields in the following ways:
4708 @item If a zero-length bitfield is inserted between two bitfields that would
4709 normally be coalesced, the bitfields will not be coalesced.
4716 unsigned long bf_1 : 12;
4718 unsigned long bf_2 : 12;
4722 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
4723 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
4725 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
4726 alignment of the zero-length bitfield is greater than the member that follows it,
4727 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
4747 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
4748 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
4749 bitfield will not affect the alignment of @code{bar} or, as a result, the size
4752 Taking this into account, it is important to note the following:
4755 @item If a zero-length bitfield follows a normal bitfield, the type of the
4756 zero-length bitfield may affect the alignment of the structure as whole. For
4757 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
4758 normal bitfield, and is of type short.
4760 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
4761 still affect the alignment of the structure:
4771 Here, @code{t4} will take up 4 bytes.
4774 @item Zero-length bitfields following non-bitfield members are ignored:
4785 Here, @code{t5} will take up 2 bytes.
4789 @subsection PowerPC Variable Attributes
4791 Three attributes currently are defined for PowerPC configurations:
4792 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4794 For full documentation of the struct attributes please see the
4795 documentation in @ref{i386 Variable Attributes}.
4797 For documentation of @code{altivec} attribute please see the
4798 documentation in @ref{PowerPC Type Attributes}.
4800 @subsection SPU Variable Attributes
4802 The SPU supports the @code{spu_vector} attribute for variables. For
4803 documentation of this attribute please see the documentation in
4804 @ref{SPU Type Attributes}.
4806 @subsection Xstormy16 Variable Attributes
4808 One attribute is currently defined for xstormy16 configurations:
4813 @cindex @code{below100} attribute
4815 If a variable has the @code{below100} attribute (@code{BELOW100} is
4816 allowed also), GCC will place the variable in the first 0x100 bytes of
4817 memory and use special opcodes to access it. Such variables will be
4818 placed in either the @code{.bss_below100} section or the
4819 @code{.data_below100} section.
4823 @subsection AVR Variable Attributes
4827 @cindex @code{progmem} variable attribute
4828 The @code{progmem} attribute is used on the AVR to place data in the Program
4829 Memory address space. The AVR is a Harvard Architecture processor and data
4830 normally resides in the Data Memory address space.
4833 @node Type Attributes
4834 @section Specifying Attributes of Types
4835 @cindex attribute of types
4836 @cindex type attributes
4838 The keyword @code{__attribute__} allows you to specify special
4839 attributes of @code{struct} and @code{union} types when you define
4840 such types. This keyword is followed by an attribute specification
4841 inside double parentheses. Seven attributes are currently defined for
4842 types: @code{aligned}, @code{packed}, @code{transparent_union},
4843 @code{unused}, @code{deprecated}, @code{visibility}, and
4844 @code{may_alias}. Other attributes are defined for functions
4845 (@pxref{Function Attributes}) and for variables (@pxref{Variable
4848 You may also specify any one of these attributes with @samp{__}
4849 preceding and following its keyword. This allows you to use these
4850 attributes in header files without being concerned about a possible
4851 macro of the same name. For example, you may use @code{__aligned__}
4852 instead of @code{aligned}.
4854 You may specify type attributes in an enum, struct or union type
4855 declaration or definition, or for other types in a @code{typedef}
4858 For an enum, struct or union type, you may specify attributes either
4859 between the enum, struct or union tag and the name of the type, or
4860 just past the closing curly brace of the @emph{definition}. The
4861 former syntax is preferred.
4863 @xref{Attribute Syntax}, for details of the exact syntax for using
4867 @cindex @code{aligned} attribute
4868 @item aligned (@var{alignment})
4869 This attribute specifies a minimum alignment (in bytes) for variables
4870 of the specified type. For example, the declarations:
4873 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
4874 typedef int more_aligned_int __attribute__ ((aligned (8)));
4878 force the compiler to insure (as far as it can) that each variable whose
4879 type is @code{struct S} or @code{more_aligned_int} will be allocated and
4880 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
4881 variables of type @code{struct S} aligned to 8-byte boundaries allows
4882 the compiler to use the @code{ldd} and @code{std} (doubleword load and
4883 store) instructions when copying one variable of type @code{struct S} to
4884 another, thus improving run-time efficiency.
4886 Note that the alignment of any given @code{struct} or @code{union} type
4887 is required by the ISO C standard to be at least a perfect multiple of
4888 the lowest common multiple of the alignments of all of the members of
4889 the @code{struct} or @code{union} in question. This means that you @emph{can}
4890 effectively adjust the alignment of a @code{struct} or @code{union}
4891 type by attaching an @code{aligned} attribute to any one of the members
4892 of such a type, but the notation illustrated in the example above is a
4893 more obvious, intuitive, and readable way to request the compiler to
4894 adjust the alignment of an entire @code{struct} or @code{union} type.
4896 As in the preceding example, you can explicitly specify the alignment
4897 (in bytes) that you wish the compiler to use for a given @code{struct}
4898 or @code{union} type. Alternatively, you can leave out the alignment factor
4899 and just ask the compiler to align a type to the maximum
4900 useful alignment for the target machine you are compiling for. For
4901 example, you could write:
4904 struct S @{ short f[3]; @} __attribute__ ((aligned));
4907 Whenever you leave out the alignment factor in an @code{aligned}
4908 attribute specification, the compiler automatically sets the alignment
4909 for the type to the largest alignment which is ever used for any data
4910 type on the target machine you are compiling for. Doing this can often
4911 make copy operations more efficient, because the compiler can use
4912 whatever instructions copy the biggest chunks of memory when performing
4913 copies to or from the variables which have types that you have aligned
4916 In the example above, if the size of each @code{short} is 2 bytes, then
4917 the size of the entire @code{struct S} type is 6 bytes. The smallest
4918 power of two which is greater than or equal to that is 8, so the
4919 compiler sets the alignment for the entire @code{struct S} type to 8
4922 Note that although you can ask the compiler to select a time-efficient
4923 alignment for a given type and then declare only individual stand-alone
4924 objects of that type, the compiler's ability to select a time-efficient
4925 alignment is primarily useful only when you plan to create arrays of
4926 variables having the relevant (efficiently aligned) type. If you
4927 declare or use arrays of variables of an efficiently-aligned type, then
4928 it is likely that your program will also be doing pointer arithmetic (or
4929 subscripting, which amounts to the same thing) on pointers to the
4930 relevant type, and the code that the compiler generates for these
4931 pointer arithmetic operations will often be more efficient for
4932 efficiently-aligned types than for other types.
4934 The @code{aligned} attribute can only increase the alignment; but you
4935 can decrease it by specifying @code{packed} as well. See below.
4937 Note that the effectiveness of @code{aligned} attributes may be limited
4938 by inherent limitations in your linker. On many systems, the linker is
4939 only able to arrange for variables to be aligned up to a certain maximum
4940 alignment. (For some linkers, the maximum supported alignment may
4941 be very very small.) If your linker is only able to align variables
4942 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4943 in an @code{__attribute__} will still only provide you with 8 byte
4944 alignment. See your linker documentation for further information.
4947 This attribute, attached to @code{struct} or @code{union} type
4948 definition, specifies that each member (other than zero-width bitfields)
4949 of the structure or union is placed to minimize the memory required. When
4950 attached to an @code{enum} definition, it indicates that the smallest
4951 integral type should be used.
4953 @opindex fshort-enums
4954 Specifying this attribute for @code{struct} and @code{union} types is
4955 equivalent to specifying the @code{packed} attribute on each of the
4956 structure or union members. Specifying the @option{-fshort-enums}
4957 flag on the line is equivalent to specifying the @code{packed}
4958 attribute on all @code{enum} definitions.
4960 In the following example @code{struct my_packed_struct}'s members are
4961 packed closely together, but the internal layout of its @code{s} member
4962 is not packed---to do that, @code{struct my_unpacked_struct} would need to
4966 struct my_unpacked_struct
4972 struct __attribute__ ((__packed__)) my_packed_struct
4976 struct my_unpacked_struct s;
4980 You may only specify this attribute on the definition of an @code{enum},
4981 @code{struct} or @code{union}, not on a @code{typedef} which does not
4982 also define the enumerated type, structure or union.
4984 @item transparent_union
4985 This attribute, attached to a @code{union} type definition, indicates
4986 that any function parameter having that union type causes calls to that
4987 function to be treated in a special way.
4989 First, the argument corresponding to a transparent union type can be of
4990 any type in the union; no cast is required. Also, if the union contains
4991 a pointer type, the corresponding argument can be a null pointer
4992 constant or a void pointer expression; and if the union contains a void
4993 pointer type, the corresponding argument can be any pointer expression.
4994 If the union member type is a pointer, qualifiers like @code{const} on
4995 the referenced type must be respected, just as with normal pointer
4998 Second, the argument is passed to the function using the calling
4999 conventions of the first member of the transparent union, not the calling
5000 conventions of the union itself. All members of the union must have the
5001 same machine representation; this is necessary for this argument passing
5004 Transparent unions are designed for library functions that have multiple
5005 interfaces for compatibility reasons. For example, suppose the
5006 @code{wait} function must accept either a value of type @code{int *} to
5007 comply with Posix, or a value of type @code{union wait *} to comply with
5008 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
5009 @code{wait} would accept both kinds of arguments, but it would also
5010 accept any other pointer type and this would make argument type checking
5011 less useful. Instead, @code{<sys/wait.h>} might define the interface
5015 typedef union __attribute__ ((__transparent_union__))
5019 @} wait_status_ptr_t;
5021 pid_t wait (wait_status_ptr_t);
5024 This interface allows either @code{int *} or @code{union wait *}
5025 arguments to be passed, using the @code{int *} calling convention.
5026 The program can call @code{wait} with arguments of either type:
5029 int w1 () @{ int w; return wait (&w); @}
5030 int w2 () @{ union wait w; return wait (&w); @}
5033 With this interface, @code{wait}'s implementation might look like this:
5036 pid_t wait (wait_status_ptr_t p)
5038 return waitpid (-1, p.__ip, 0);
5043 When attached to a type (including a @code{union} or a @code{struct}),
5044 this attribute means that variables of that type are meant to appear
5045 possibly unused. GCC will not produce a warning for any variables of
5046 that type, even if the variable appears to do nothing. This is often
5047 the case with lock or thread classes, which are usually defined and then
5048 not referenced, but contain constructors and destructors that have
5049 nontrivial bookkeeping functions.
5052 @itemx deprecated (@var{msg})
5053 The @code{deprecated} attribute results in a warning if the type
5054 is used anywhere in the source file. This is useful when identifying
5055 types that are expected to be removed in a future version of a program.
5056 If possible, the warning also includes the location of the declaration
5057 of the deprecated type, to enable users to easily find further
5058 information about why the type is deprecated, or what they should do
5059 instead. Note that the warnings only occur for uses and then only
5060 if the type is being applied to an identifier that itself is not being
5061 declared as deprecated.
5064 typedef int T1 __attribute__ ((deprecated));
5068 typedef T1 T3 __attribute__ ((deprecated));
5069 T3 z __attribute__ ((deprecated));
5072 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
5073 warning is issued for line 4 because T2 is not explicitly
5074 deprecated. Line 5 has no warning because T3 is explicitly
5075 deprecated. Similarly for line 6. The optional msg
5076 argument, which must be a string, will be printed in the warning if
5079 The @code{deprecated} attribute can also be used for functions and
5080 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
5083 Accesses through pointers to types with this attribute are not subject
5084 to type-based alias analysis, but are instead assumed to be able to alias
5085 any other type of objects. In the context of 6.5/7 an lvalue expression
5086 dereferencing such a pointer is treated like having a character type.
5087 See @option{-fstrict-aliasing} for more information on aliasing issues.
5088 This extension exists to support some vector APIs, in which pointers to
5089 one vector type are permitted to alias pointers to a different vector type.
5091 Note that an object of a type with this attribute does not have any
5097 typedef short __attribute__((__may_alias__)) short_a;
5103 short_a *b = (short_a *) &a;
5107 if (a == 0x12345678)
5114 If you replaced @code{short_a} with @code{short} in the variable
5115 declaration, the above program would abort when compiled with
5116 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
5117 above in recent GCC versions.
5120 In C++, attribute visibility (@pxref{Function Attributes}) can also be
5121 applied to class, struct, union and enum types. Unlike other type
5122 attributes, the attribute must appear between the initial keyword and
5123 the name of the type; it cannot appear after the body of the type.
5125 Note that the type visibility is applied to vague linkage entities
5126 associated with the class (vtable, typeinfo node, etc.). In
5127 particular, if a class is thrown as an exception in one shared object
5128 and caught in another, the class must have default visibility.
5129 Otherwise the two shared objects will be unable to use the same
5130 typeinfo node and exception handling will break.
5134 @subsection ARM Type Attributes
5136 On those ARM targets that support @code{dllimport} (such as Symbian
5137 OS), you can use the @code{notshared} attribute to indicate that the
5138 virtual table and other similar data for a class should not be
5139 exported from a DLL@. For example:
5142 class __declspec(notshared) C @{
5144 __declspec(dllimport) C();
5148 __declspec(dllexport)
5152 In this code, @code{C::C} is exported from the current DLL, but the
5153 virtual table for @code{C} is not exported. (You can use
5154 @code{__attribute__} instead of @code{__declspec} if you prefer, but
5155 most Symbian OS code uses @code{__declspec}.)
5157 @anchor{MeP Type Attributes}
5158 @subsection MeP Type Attributes
5160 Many of the MeP variable attributes may be applied to types as well.
5161 Specifically, the @code{based}, @code{tiny}, @code{near}, and
5162 @code{far} attributes may be applied to either. The @code{io} and
5163 @code{cb} attributes may not be applied to types.
5165 @anchor{i386 Type Attributes}
5166 @subsection i386 Type Attributes
5168 Two attributes are currently defined for i386 configurations:
5169 @code{ms_struct} and @code{gcc_struct}.
5175 @cindex @code{ms_struct}
5176 @cindex @code{gcc_struct}
5178 If @code{packed} is used on a structure, or if bit-fields are used
5179 it may be that the Microsoft ABI packs them differently
5180 than GCC would normally pack them. Particularly when moving packed
5181 data between functions compiled with GCC and the native Microsoft compiler
5182 (either via function call or as data in a file), it may be necessary to access
5185 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
5186 compilers to match the native Microsoft compiler.
5189 To specify multiple attributes, separate them by commas within the
5190 double parentheses: for example, @samp{__attribute__ ((aligned (16),
5193 @anchor{PowerPC Type Attributes}
5194 @subsection PowerPC Type Attributes
5196 Three attributes currently are defined for PowerPC configurations:
5197 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
5199 For full documentation of the @code{ms_struct} and @code{gcc_struct}
5200 attributes please see the documentation in @ref{i386 Type Attributes}.
5202 The @code{altivec} attribute allows one to declare AltiVec vector data
5203 types supported by the AltiVec Programming Interface Manual. The
5204 attribute requires an argument to specify one of three vector types:
5205 @code{vector__}, @code{pixel__} (always followed by unsigned short),
5206 and @code{bool__} (always followed by unsigned).
5209 __attribute__((altivec(vector__)))
5210 __attribute__((altivec(pixel__))) unsigned short
5211 __attribute__((altivec(bool__))) unsigned
5214 These attributes mainly are intended to support the @code{__vector},
5215 @code{__pixel}, and @code{__bool} AltiVec keywords.
5217 @anchor{SPU Type Attributes}
5218 @subsection SPU Type Attributes
5220 The SPU supports the @code{spu_vector} attribute for types. This attribute
5221 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
5222 Language Extensions Specification. It is intended to support the
5223 @code{__vector} keyword.
5227 @section An Inline Function is As Fast As a Macro
5228 @cindex inline functions
5229 @cindex integrating function code
5231 @cindex macros, inline alternative
5233 By declaring a function inline, you can direct GCC to make
5234 calls to that function faster. One way GCC can achieve this is to
5235 integrate that function's code into the code for its callers. This
5236 makes execution faster by eliminating the function-call overhead; in
5237 addition, if any of the actual argument values are constant, their
5238 known values may permit simplifications at compile time so that not
5239 all of the inline function's code needs to be included. The effect on
5240 code size is less predictable; object code may be larger or smaller
5241 with function inlining, depending on the particular case. You can
5242 also direct GCC to try to integrate all ``simple enough'' functions
5243 into their callers with the option @option{-finline-functions}.
5245 GCC implements three different semantics of declaring a function
5246 inline. One is available with @option{-std=gnu89} or
5247 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
5248 on all inline declarations, another when
5249 @option{-std=c99}, @option{-std=c1x},
5250 @option{-std=gnu99} or @option{-std=gnu1x}
5251 (without @option{-fgnu89-inline}), and the third
5252 is used when compiling C++.
5254 To declare a function inline, use the @code{inline} keyword in its
5255 declaration, like this:
5265 If you are writing a header file to be included in ISO C90 programs, write
5266 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
5268 The three types of inlining behave similarly in two important cases:
5269 when the @code{inline} keyword is used on a @code{static} function,
5270 like the example above, and when a function is first declared without
5271 using the @code{inline} keyword and then is defined with
5272 @code{inline}, like this:
5275 extern int inc (int *a);
5283 In both of these common cases, the program behaves the same as if you
5284 had not used the @code{inline} keyword, except for its speed.
5286 @cindex inline functions, omission of
5287 @opindex fkeep-inline-functions
5288 When a function is both inline and @code{static}, if all calls to the
5289 function are integrated into the caller, and the function's address is
5290 never used, then the function's own assembler code is never referenced.
5291 In this case, GCC does not actually output assembler code for the
5292 function, unless you specify the option @option{-fkeep-inline-functions}.
5293 Some calls cannot be integrated for various reasons (in particular,
5294 calls that precede the function's definition cannot be integrated, and
5295 neither can recursive calls within the definition). If there is a
5296 nonintegrated call, then the function is compiled to assembler code as
5297 usual. The function must also be compiled as usual if the program
5298 refers to its address, because that can't be inlined.
5301 Note that certain usages in a function definition can make it unsuitable
5302 for inline substitution. Among these usages are: use of varargs, use of
5303 alloca, use of variable sized data types (@pxref{Variable Length}),
5304 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
5305 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
5306 will warn when a function marked @code{inline} could not be substituted,
5307 and will give the reason for the failure.
5309 @cindex automatic @code{inline} for C++ member fns
5310 @cindex @code{inline} automatic for C++ member fns
5311 @cindex member fns, automatically @code{inline}
5312 @cindex C++ member fns, automatically @code{inline}
5313 @opindex fno-default-inline
5314 As required by ISO C++, GCC considers member functions defined within
5315 the body of a class to be marked inline even if they are
5316 not explicitly declared with the @code{inline} keyword. You can
5317 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
5318 Options,,Options Controlling C++ Dialect}.
5320 GCC does not inline any functions when not optimizing unless you specify
5321 the @samp{always_inline} attribute for the function, like this:
5324 /* @r{Prototype.} */
5325 inline void foo (const char) __attribute__((always_inline));
5328 The remainder of this section is specific to GNU C90 inlining.
5330 @cindex non-static inline function
5331 When an inline function is not @code{static}, then the compiler must assume
5332 that there may be calls from other source files; since a global symbol can
5333 be defined only once in any program, the function must not be defined in
5334 the other source files, so the calls therein cannot be integrated.
5335 Therefore, a non-@code{static} inline function is always compiled on its
5336 own in the usual fashion.
5338 If you specify both @code{inline} and @code{extern} in the function
5339 definition, then the definition is used only for inlining. In no case
5340 is the function compiled on its own, not even if you refer to its
5341 address explicitly. Such an address becomes an external reference, as
5342 if you had only declared the function, and had not defined it.
5344 This combination of @code{inline} and @code{extern} has almost the
5345 effect of a macro. The way to use it is to put a function definition in
5346 a header file with these keywords, and put another copy of the
5347 definition (lacking @code{inline} and @code{extern}) in a library file.
5348 The definition in the header file will cause most calls to the function
5349 to be inlined. If any uses of the function remain, they will refer to
5350 the single copy in the library.
5353 @section When is a Volatile Object Accessed?
5354 @cindex accessing volatiles
5355 @cindex volatile read
5356 @cindex volatile write
5357 @cindex volatile access
5359 C has the concept of volatile objects. These are normally accessed by
5360 pointers and used for accessing hardware or inter-thread
5361 communication. The standard encourage compilers to refrain from
5362 optimizations concerning accesses to volatile objects, but leaves it
5363 implementation defined as to what constitutes a volatile access. The
5364 minimum requirement is that at a sequence point all previous accesses
5365 to volatile objects have stabilized and no subsequent accesses have
5366 occurred. Thus an implementation is free to reorder and combine
5367 volatile accesses which occur between sequence points, but cannot do
5368 so for accesses across a sequence point. The use of volatiles does
5369 not allow you to violate the restriction on updating objects multiple
5370 times between two sequence points.
5372 Accesses to non-volatile objects are not ordered with respect to
5373 volatile accesses. You cannot use a volatile object as a memory
5374 barrier to order a sequence of writes to non-volatile memory. For
5378 int *ptr = @var{something};
5380 *ptr = @var{something};
5384 Unless @var{*ptr} and @var{vobj} can be aliased, it is not guaranteed
5385 that the write to @var{*ptr} will have occurred by the time the update
5386 of @var{vobj} has happened. If you need this guarantee, you must use
5387 a stronger memory barrier such as:
5390 int *ptr = @var{something};
5392 *ptr = @var{something};
5393 asm volatile ("" : : : "memory");
5397 A scalar volatile object is read, when it is accessed in a void context:
5400 volatile int *src = @var{somevalue};
5404 Such expressions are rvalues, and GCC implements this as a
5405 read of the volatile object being pointed to.
5407 Assignments are also expressions and have an rvalue. However when
5408 assigning to a scalar volatile, the volatile object is not reread,
5409 regardless of whether the assignment expression's rvalue is used or
5410 not. If the assignment's rvalue is used, the value is that assigned
5411 to the volatile object. For instance, there is no read of @var{vobj}
5412 in all the following cases:
5417 vobj = @var{something};
5418 obj = vobj = @var{something};
5419 obj ? vobj = @var{onething} : vobj = @var{anotherthing};
5420 obj = (@var{something}, vobj = @var{anotherthing});
5423 If you need to read the volatile object after an assignment has
5424 occurred, you must use a separate expression with an intervening
5427 As bitfields are not individually addressable, volatile bitfields may
5428 be implicitly read when written to, or when adjacent bitfields are
5429 accessed. Bitfield operations may be optimized such that adjacent
5430 bitfields are only partially accessed, if they straddle a storage unit
5431 boundary. For these reasons it is unwise to use volatile bitfields to
5435 @section Assembler Instructions with C Expression Operands
5436 @cindex extended @code{asm}
5437 @cindex @code{asm} expressions
5438 @cindex assembler instructions
5441 In an assembler instruction using @code{asm}, you can specify the
5442 operands of the instruction using C expressions. This means you need not
5443 guess which registers or memory locations will contain the data you want
5446 You must specify an assembler instruction template much like what
5447 appears in a machine description, plus an operand constraint string for
5450 For example, here is how to use the 68881's @code{fsinx} instruction:
5453 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
5457 Here @code{angle} is the C expression for the input operand while
5458 @code{result} is that of the output operand. Each has @samp{"f"} as its
5459 operand constraint, saying that a floating point register is required.
5460 The @samp{=} in @samp{=f} indicates that the operand is an output; all
5461 output operands' constraints must use @samp{=}. The constraints use the
5462 same language used in the machine description (@pxref{Constraints}).
5464 Each operand is described by an operand-constraint string followed by
5465 the C expression in parentheses. A colon separates the assembler
5466 template from the first output operand and another separates the last
5467 output operand from the first input, if any. Commas separate the
5468 operands within each group. The total number of operands is currently
5469 limited to 30; this limitation may be lifted in some future version of
5472 If there are no output operands but there are input operands, you must
5473 place two consecutive colons surrounding the place where the output
5476 As of GCC version 3.1, it is also possible to specify input and output
5477 operands using symbolic names which can be referenced within the
5478 assembler code. These names are specified inside square brackets
5479 preceding the constraint string, and can be referenced inside the
5480 assembler code using @code{%[@var{name}]} instead of a percentage sign
5481 followed by the operand number. Using named operands the above example
5485 asm ("fsinx %[angle],%[output]"
5486 : [output] "=f" (result)
5487 : [angle] "f" (angle));
5491 Note that the symbolic operand names have no relation whatsoever to
5492 other C identifiers. You may use any name you like, even those of
5493 existing C symbols, but you must ensure that no two operands within the same
5494 assembler construct use the same symbolic name.
5496 Output operand expressions must be lvalues; the compiler can check this.
5497 The input operands need not be lvalues. The compiler cannot check
5498 whether the operands have data types that are reasonable for the
5499 instruction being executed. It does not parse the assembler instruction
5500 template and does not know what it means or even whether it is valid
5501 assembler input. The extended @code{asm} feature is most often used for
5502 machine instructions the compiler itself does not know exist. If
5503 the output expression cannot be directly addressed (for example, it is a
5504 bit-field), your constraint must allow a register. In that case, GCC
5505 will use the register as the output of the @code{asm}, and then store
5506 that register into the output.
5508 The ordinary output operands must be write-only; GCC will assume that
5509 the values in these operands before the instruction are dead and need
5510 not be generated. Extended asm supports input-output or read-write
5511 operands. Use the constraint character @samp{+} to indicate such an
5512 operand and list it with the output operands. You should only use
5513 read-write operands when the constraints for the operand (or the
5514 operand in which only some of the bits are to be changed) allow a
5517 You may, as an alternative, logically split its function into two
5518 separate operands, one input operand and one write-only output
5519 operand. The connection between them is expressed by constraints
5520 which say they need to be in the same location when the instruction
5521 executes. You can use the same C expression for both operands, or
5522 different expressions. For example, here we write the (fictitious)
5523 @samp{combine} instruction with @code{bar} as its read-only source
5524 operand and @code{foo} as its read-write destination:
5527 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
5531 The constraint @samp{"0"} for operand 1 says that it must occupy the
5532 same location as operand 0. A number in constraint is allowed only in
5533 an input operand and it must refer to an output operand.
5535 Only a number in the constraint can guarantee that one operand will be in
5536 the same place as another. The mere fact that @code{foo} is the value
5537 of both operands is not enough to guarantee that they will be in the
5538 same place in the generated assembler code. The following would not
5542 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
5545 Various optimizations or reloading could cause operands 0 and 1 to be in
5546 different registers; GCC knows no reason not to do so. For example, the
5547 compiler might find a copy of the value of @code{foo} in one register and
5548 use it for operand 1, but generate the output operand 0 in a different
5549 register (copying it afterward to @code{foo}'s own address). Of course,
5550 since the register for operand 1 is not even mentioned in the assembler
5551 code, the result will not work, but GCC can't tell that.
5553 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
5554 the operand number for a matching constraint. For example:
5557 asm ("cmoveq %1,%2,%[result]"
5558 : [result] "=r"(result)
5559 : "r" (test), "r"(new), "[result]"(old));
5562 Sometimes you need to make an @code{asm} operand be a specific register,
5563 but there's no matching constraint letter for that register @emph{by
5564 itself}. To force the operand into that register, use a local variable
5565 for the operand and specify the register in the variable declaration.
5566 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
5567 register constraint letter that matches the register:
5570 register int *p1 asm ("r0") = @dots{};
5571 register int *p2 asm ("r1") = @dots{};
5572 register int *result asm ("r0");
5573 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5576 @anchor{Example of asm with clobbered asm reg}
5577 In the above example, beware that a register that is call-clobbered by
5578 the target ABI will be overwritten by any function call in the
5579 assignment, including library calls for arithmetic operators.
5580 Also a register may be clobbered when generating some operations,
5581 like variable shift, memory copy or memory move on x86.
5582 Assuming it is a call-clobbered register, this may happen to @code{r0}
5583 above by the assignment to @code{p2}. If you have to use such a
5584 register, use temporary variables for expressions between the register
5589 register int *p1 asm ("r0") = @dots{};
5590 register int *p2 asm ("r1") = t1;
5591 register int *result asm ("r0");
5592 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5595 Some instructions clobber specific hard registers. To describe this,
5596 write a third colon after the input operands, followed by the names of
5597 the clobbered hard registers (given as strings). Here is a realistic
5598 example for the VAX:
5601 asm volatile ("movc3 %0,%1,%2"
5602 : /* @r{no outputs} */
5603 : "g" (from), "g" (to), "g" (count)
5604 : "r0", "r1", "r2", "r3", "r4", "r5");
5607 You may not write a clobber description in a way that overlaps with an
5608 input or output operand. For example, you may not have an operand
5609 describing a register class with one member if you mention that register
5610 in the clobber list. Variables declared to live in specific registers
5611 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
5612 have no part mentioned in the clobber description.
5613 There is no way for you to specify that an input
5614 operand is modified without also specifying it as an output
5615 operand. Note that if all the output operands you specify are for this
5616 purpose (and hence unused), you will then also need to specify
5617 @code{volatile} for the @code{asm} construct, as described below, to
5618 prevent GCC from deleting the @code{asm} statement as unused.
5620 If you refer to a particular hardware register from the assembler code,
5621 you will probably have to list the register after the third colon to
5622 tell the compiler the register's value is modified. In some assemblers,
5623 the register names begin with @samp{%}; to produce one @samp{%} in the
5624 assembler code, you must write @samp{%%} in the input.
5626 If your assembler instruction can alter the condition code register, add
5627 @samp{cc} to the list of clobbered registers. GCC on some machines
5628 represents the condition codes as a specific hardware register;
5629 @samp{cc} serves to name this register. On other machines, the
5630 condition code is handled differently, and specifying @samp{cc} has no
5631 effect. But it is valid no matter what the machine.
5633 If your assembler instructions access memory in an unpredictable
5634 fashion, add @samp{memory} to the list of clobbered registers. This
5635 will cause GCC to not keep memory values cached in registers across the
5636 assembler instruction and not optimize stores or loads to that memory.
5637 You will also want to add the @code{volatile} keyword if the memory
5638 affected is not listed in the inputs or outputs of the @code{asm}, as
5639 the @samp{memory} clobber does not count as a side-effect of the
5640 @code{asm}. If you know how large the accessed memory is, you can add
5641 it as input or output but if this is not known, you should add
5642 @samp{memory}. As an example, if you access ten bytes of a string, you
5643 can use a memory input like:
5646 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
5649 Note that in the following example the memory input is necessary,
5650 otherwise GCC might optimize the store to @code{x} away:
5657 asm ("magic stuff accessing an 'int' pointed to by '%1'"
5658 "=&d" (r) : "a" (y), "m" (*y));
5663 You can put multiple assembler instructions together in a single
5664 @code{asm} template, separated by the characters normally used in assembly
5665 code for the system. A combination that works in most places is a newline
5666 to break the line, plus a tab character to move to the instruction field
5667 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
5668 assembler allows semicolons as a line-breaking character. Note that some
5669 assembler dialects use semicolons to start a comment.
5670 The input operands are guaranteed not to use any of the clobbered
5671 registers, and neither will the output operands' addresses, so you can
5672 read and write the clobbered registers as many times as you like. Here
5673 is an example of multiple instructions in a template; it assumes the
5674 subroutine @code{_foo} accepts arguments in registers 9 and 10:
5677 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
5679 : "g" (from), "g" (to)
5683 Unless an output operand has the @samp{&} constraint modifier, GCC
5684 may allocate it in the same register as an unrelated input operand, on
5685 the assumption the inputs are consumed before the outputs are produced.
5686 This assumption may be false if the assembler code actually consists of
5687 more than one instruction. In such a case, use @samp{&} for each output
5688 operand that may not overlap an input. @xref{Modifiers}.
5690 If you want to test the condition code produced by an assembler
5691 instruction, you must include a branch and a label in the @code{asm}
5692 construct, as follows:
5695 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
5701 This assumes your assembler supports local labels, as the GNU assembler
5702 and most Unix assemblers do.
5704 Speaking of labels, jumps from one @code{asm} to another are not
5705 supported. The compiler's optimizers do not know about these jumps, and
5706 therefore they cannot take account of them when deciding how to
5707 optimize. @xref{Extended asm with goto}.
5709 @cindex macros containing @code{asm}
5710 Usually the most convenient way to use these @code{asm} instructions is to
5711 encapsulate them in macros that look like functions. For example,
5715 (@{ double __value, __arg = (x); \
5716 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
5721 Here the variable @code{__arg} is used to make sure that the instruction
5722 operates on a proper @code{double} value, and to accept only those
5723 arguments @code{x} which can convert automatically to a @code{double}.
5725 Another way to make sure the instruction operates on the correct data
5726 type is to use a cast in the @code{asm}. This is different from using a
5727 variable @code{__arg} in that it converts more different types. For
5728 example, if the desired type were @code{int}, casting the argument to
5729 @code{int} would accept a pointer with no complaint, while assigning the
5730 argument to an @code{int} variable named @code{__arg} would warn about
5731 using a pointer unless the caller explicitly casts it.
5733 If an @code{asm} has output operands, GCC assumes for optimization
5734 purposes the instruction has no side effects except to change the output
5735 operands. This does not mean instructions with a side effect cannot be
5736 used, but you must be careful, because the compiler may eliminate them
5737 if the output operands aren't used, or move them out of loops, or
5738 replace two with one if they constitute a common subexpression. Also,
5739 if your instruction does have a side effect on a variable that otherwise
5740 appears not to change, the old value of the variable may be reused later
5741 if it happens to be found in a register.
5743 You can prevent an @code{asm} instruction from being deleted
5744 by writing the keyword @code{volatile} after
5745 the @code{asm}. For example:
5748 #define get_and_set_priority(new) \
5750 asm volatile ("get_and_set_priority %0, %1" \
5751 : "=g" (__old) : "g" (new)); \
5756 The @code{volatile} keyword indicates that the instruction has
5757 important side-effects. GCC will not delete a volatile @code{asm} if
5758 it is reachable. (The instruction can still be deleted if GCC can
5759 prove that control-flow will never reach the location of the
5760 instruction.) Note that even a volatile @code{asm} instruction
5761 can be moved relative to other code, including across jump
5762 instructions. For example, on many targets there is a system
5763 register which can be set to control the rounding mode of
5764 floating point operations. You might try
5765 setting it with a volatile @code{asm}, like this PowerPC example:
5768 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
5773 This will not work reliably, as the compiler may move the addition back
5774 before the volatile @code{asm}. To make it work you need to add an
5775 artificial dependency to the @code{asm} referencing a variable in the code
5776 you don't want moved, for example:
5779 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
5783 Similarly, you can't expect a
5784 sequence of volatile @code{asm} instructions to remain perfectly
5785 consecutive. If you want consecutive output, use a single @code{asm}.
5786 Also, GCC will perform some optimizations across a volatile @code{asm}
5787 instruction; GCC does not ``forget everything'' when it encounters
5788 a volatile @code{asm} instruction the way some other compilers do.
5790 An @code{asm} instruction without any output operands will be treated
5791 identically to a volatile @code{asm} instruction.
5793 It is a natural idea to look for a way to give access to the condition
5794 code left by the assembler instruction. However, when we attempted to
5795 implement this, we found no way to make it work reliably. The problem
5796 is that output operands might need reloading, which would result in
5797 additional following ``store'' instructions. On most machines, these
5798 instructions would alter the condition code before there was time to
5799 test it. This problem doesn't arise for ordinary ``test'' and
5800 ``compare'' instructions because they don't have any output operands.
5802 For reasons similar to those described above, it is not possible to give
5803 an assembler instruction access to the condition code left by previous
5806 @anchor{Extended asm with goto}
5807 As of GCC version 4.5, @code{asm goto} may be used to have the assembly
5808 jump to one or more C labels. In this form, a fifth section after the
5809 clobber list contains a list of all C labels to which the assembly may jump.
5810 Each label operand is implicitly self-named. The @code{asm} is also assumed
5811 to fall through to the next statement.
5813 This form of @code{asm} is restricted to not have outputs. This is due
5814 to a internal restriction in the compiler that control transfer instructions
5815 cannot have outputs. This restriction on @code{asm goto} may be lifted
5816 in some future version of the compiler. In the mean time, @code{asm goto}
5817 may include a memory clobber, and so leave outputs in memory.
5823 asm goto ("frob %%r5, %1; jc %l[error]; mov (%2), %%r5"
5824 : : "r"(x), "r"(&y) : "r5", "memory" : error);
5831 In this (inefficient) example, the @code{frob} instruction sets the
5832 carry bit to indicate an error. The @code{jc} instruction detects
5833 this and branches to the @code{error} label. Finally, the output
5834 of the @code{frob} instruction (@code{%r5}) is stored into the memory
5835 for variable @code{y}, which is later read by the @code{return} statement.
5841 asm goto ("mfsr %%r1, 123; jmp %%r1;"
5842 ".pushsection doit_table;"
5843 ".long %l0, %l1, %l2, %l3;"
5845 : : : "r1" : label1, label2, label3, label4);
5846 __builtin_unreachable ();
5861 In this (also inefficient) example, the @code{mfsr} instruction reads
5862 an address from some out-of-band machine register, and the following
5863 @code{jmp} instruction branches to that address. The address read by
5864 the @code{mfsr} instruction is assumed to have been previously set via
5865 some application-specific mechanism to be one of the four values stored
5866 in the @code{doit_table} section. Finally, the @code{asm} is followed
5867 by a call to @code{__builtin_unreachable} to indicate that the @code{asm}
5868 does not in fact fall through.
5871 #define TRACE1(NUM) \
5873 asm goto ("0: nop;" \
5874 ".pushsection trace_table;" \
5877 : : : : trace#NUM); \
5878 if (0) @{ trace#NUM: trace(); @} \
5880 #define TRACE TRACE1(__COUNTER__)
5883 In this example (which in fact inspired the @code{asm goto} feature)
5884 we want on rare occasions to call the @code{trace} function; on other
5885 occasions we'd like to keep the overhead to the absolute minimum.
5886 The normal code path consists of a single @code{nop} instruction.
5887 However, we record the address of this @code{nop} together with the
5888 address of a label that calls the @code{trace} function. This allows
5889 the @code{nop} instruction to be patched at runtime to be an
5890 unconditional branch to the stored label. It is assumed that an
5891 optimizing compiler will move the labeled block out of line, to
5892 optimize the fall through path from the @code{asm}.
5894 If you are writing a header file that should be includable in ISO C
5895 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
5898 @subsection Size of an @code{asm}
5900 Some targets require that GCC track the size of each instruction used in
5901 order to generate correct code. Because the final length of an
5902 @code{asm} is only known by the assembler, GCC must make an estimate as
5903 to how big it will be. The estimate is formed by counting the number of
5904 statements in the pattern of the @code{asm} and multiplying that by the
5905 length of the longest instruction on that processor. Statements in the
5906 @code{asm} are identified by newline characters and whatever statement
5907 separator characters are supported by the assembler; on most processors
5908 this is the `@code{;}' character.
5910 Normally, GCC's estimate is perfectly adequate to ensure that correct
5911 code is generated, but it is possible to confuse the compiler if you use
5912 pseudo instructions or assembler macros that expand into multiple real
5913 instructions or if you use assembler directives that expand to more
5914 space in the object file than would be needed for a single instruction.
5915 If this happens then the assembler will produce a diagnostic saying that
5916 a label is unreachable.
5918 @subsection i386 floating point asm operands
5920 There are several rules on the usage of stack-like regs in
5921 asm_operands insns. These rules apply only to the operands that are
5926 Given a set of input regs that die in an asm_operands, it is
5927 necessary to know which are implicitly popped by the asm, and
5928 which must be explicitly popped by gcc.
5930 An input reg that is implicitly popped by the asm must be
5931 explicitly clobbered, unless it is constrained to match an
5935 For any input reg that is implicitly popped by an asm, it is
5936 necessary to know how to adjust the stack to compensate for the pop.
5937 If any non-popped input is closer to the top of the reg-stack than
5938 the implicitly popped reg, it would not be possible to know what the
5939 stack looked like---it's not clear how the rest of the stack ``slides
5942 All implicitly popped input regs must be closer to the top of
5943 the reg-stack than any input that is not implicitly popped.
5945 It is possible that if an input dies in an insn, reload might
5946 use the input reg for an output reload. Consider this example:
5949 asm ("foo" : "=t" (a) : "f" (b));
5952 This asm says that input B is not popped by the asm, and that
5953 the asm pushes a result onto the reg-stack, i.e., the stack is one
5954 deeper after the asm than it was before. But, it is possible that
5955 reload will think that it can use the same reg for both the input and
5956 the output, if input B dies in this insn.
5958 If any input operand uses the @code{f} constraint, all output reg
5959 constraints must use the @code{&} earlyclobber.
5961 The asm above would be written as
5964 asm ("foo" : "=&t" (a) : "f" (b));
5968 Some operands need to be in particular places on the stack. All
5969 output operands fall in this category---there is no other way to
5970 know which regs the outputs appear in unless the user indicates
5971 this in the constraints.
5973 Output operands must specifically indicate which reg an output
5974 appears in after an asm. @code{=f} is not allowed: the operand
5975 constraints must select a class with a single reg.
5978 Output operands may not be ``inserted'' between existing stack regs.
5979 Since no 387 opcode uses a read/write operand, all output operands
5980 are dead before the asm_operands, and are pushed by the asm_operands.
5981 It makes no sense to push anywhere but the top of the reg-stack.
5983 Output operands must start at the top of the reg-stack: output
5984 operands may not ``skip'' a reg.
5987 Some asm statements may need extra stack space for internal
5988 calculations. This can be guaranteed by clobbering stack registers
5989 unrelated to the inputs and outputs.
5993 Here are a couple of reasonable asms to want to write. This asm
5994 takes one input, which is internally popped, and produces two outputs.
5997 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
6000 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
6001 and replaces them with one output. The user must code the @code{st(1)}
6002 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
6005 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
6011 @section Controlling Names Used in Assembler Code
6012 @cindex assembler names for identifiers
6013 @cindex names used in assembler code
6014 @cindex identifiers, names in assembler code
6016 You can specify the name to be used in the assembler code for a C
6017 function or variable by writing the @code{asm} (or @code{__asm__})
6018 keyword after the declarator as follows:
6021 int foo asm ("myfoo") = 2;
6025 This specifies that the name to be used for the variable @code{foo} in
6026 the assembler code should be @samp{myfoo} rather than the usual
6029 On systems where an underscore is normally prepended to the name of a C
6030 function or variable, this feature allows you to define names for the
6031 linker that do not start with an underscore.
6033 It does not make sense to use this feature with a non-static local
6034 variable since such variables do not have assembler names. If you are
6035 trying to put the variable in a particular register, see @ref{Explicit
6036 Reg Vars}. GCC presently accepts such code with a warning, but will
6037 probably be changed to issue an error, rather than a warning, in the
6040 You cannot use @code{asm} in this way in a function @emph{definition}; but
6041 you can get the same effect by writing a declaration for the function
6042 before its definition and putting @code{asm} there, like this:
6045 extern func () asm ("FUNC");
6052 It is up to you to make sure that the assembler names you choose do not
6053 conflict with any other assembler symbols. Also, you must not use a
6054 register name; that would produce completely invalid assembler code. GCC
6055 does not as yet have the ability to store static variables in registers.
6056 Perhaps that will be added.
6058 @node Explicit Reg Vars
6059 @section Variables in Specified Registers
6060 @cindex explicit register variables
6061 @cindex variables in specified registers
6062 @cindex specified registers
6063 @cindex registers, global allocation
6065 GNU C allows you to put a few global variables into specified hardware
6066 registers. You can also specify the register in which an ordinary
6067 register variable should be allocated.
6071 Global register variables reserve registers throughout the program.
6072 This may be useful in programs such as programming language
6073 interpreters which have a couple of global variables that are accessed
6077 Local register variables in specific registers do not reserve the
6078 registers, except at the point where they are used as input or output
6079 operands in an @code{asm} statement and the @code{asm} statement itself is
6080 not deleted. The compiler's data flow analysis is capable of determining
6081 where the specified registers contain live values, and where they are
6082 available for other uses. Stores into local register variables may be deleted
6083 when they appear to be dead according to dataflow analysis. References
6084 to local register variables may be deleted or moved or simplified.
6086 These local variables are sometimes convenient for use with the extended
6087 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
6088 output of the assembler instruction directly into a particular register.
6089 (This will work provided the register you specify fits the constraints
6090 specified for that operand in the @code{asm}.)
6098 @node Global Reg Vars
6099 @subsection Defining Global Register Variables
6100 @cindex global register variables
6101 @cindex registers, global variables in
6103 You can define a global register variable in GNU C like this:
6106 register int *foo asm ("a5");
6110 Here @code{a5} is the name of the register which should be used. Choose a
6111 register which is normally saved and restored by function calls on your
6112 machine, so that library routines will not clobber it.
6114 Naturally the register name is cpu-dependent, so you would need to
6115 conditionalize your program according to cpu type. The register
6116 @code{a5} would be a good choice on a 68000 for a variable of pointer
6117 type. On machines with register windows, be sure to choose a ``global''
6118 register that is not affected magically by the function call mechanism.
6120 In addition, operating systems on one type of cpu may differ in how they
6121 name the registers; then you would need additional conditionals. For
6122 example, some 68000 operating systems call this register @code{%a5}.
6124 Eventually there may be a way of asking the compiler to choose a register
6125 automatically, but first we need to figure out how it should choose and
6126 how to enable you to guide the choice. No solution is evident.
6128 Defining a global register variable in a certain register reserves that
6129 register entirely for this use, at least within the current compilation.
6130 The register will not be allocated for any other purpose in the functions
6131 in the current compilation. The register will not be saved and restored by
6132 these functions. Stores into this register are never deleted even if they
6133 would appear to be dead, but references may be deleted or moved or
6136 It is not safe to access the global register variables from signal
6137 handlers, or from more than one thread of control, because the system
6138 library routines may temporarily use the register for other things (unless
6139 you recompile them specially for the task at hand).
6141 @cindex @code{qsort}, and global register variables
6142 It is not safe for one function that uses a global register variable to
6143 call another such function @code{foo} by way of a third function
6144 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
6145 different source file in which the variable wasn't declared). This is
6146 because @code{lose} might save the register and put some other value there.
6147 For example, you can't expect a global register variable to be available in
6148 the comparison-function that you pass to @code{qsort}, since @code{qsort}
6149 might have put something else in that register. (If you are prepared to
6150 recompile @code{qsort} with the same global register variable, you can
6151 solve this problem.)
6153 If you want to recompile @code{qsort} or other source files which do not
6154 actually use your global register variable, so that they will not use that
6155 register for any other purpose, then it suffices to specify the compiler
6156 option @option{-ffixed-@var{reg}}. You need not actually add a global
6157 register declaration to their source code.
6159 A function which can alter the value of a global register variable cannot
6160 safely be called from a function compiled without this variable, because it
6161 could clobber the value the caller expects to find there on return.
6162 Therefore, the function which is the entry point into the part of the
6163 program that uses the global register variable must explicitly save and
6164 restore the value which belongs to its caller.
6166 @cindex register variable after @code{longjmp}
6167 @cindex global register after @code{longjmp}
6168 @cindex value after @code{longjmp}
6171 On most machines, @code{longjmp} will restore to each global register
6172 variable the value it had at the time of the @code{setjmp}. On some
6173 machines, however, @code{longjmp} will not change the value of global
6174 register variables. To be portable, the function that called @code{setjmp}
6175 should make other arrangements to save the values of the global register
6176 variables, and to restore them in a @code{longjmp}. This way, the same
6177 thing will happen regardless of what @code{longjmp} does.
6179 All global register variable declarations must precede all function
6180 definitions. If such a declaration could appear after function
6181 definitions, the declaration would be too late to prevent the register from
6182 being used for other purposes in the preceding functions.
6184 Global register variables may not have initial values, because an
6185 executable file has no means to supply initial contents for a register.
6187 On the SPARC, there are reports that g3 @dots{} g7 are suitable
6188 registers, but certain library functions, such as @code{getwd}, as well
6189 as the subroutines for division and remainder, modify g3 and g4. g1 and
6190 g2 are local temporaries.
6192 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
6193 Of course, it will not do to use more than a few of those.
6195 @node Local Reg Vars
6196 @subsection Specifying Registers for Local Variables
6197 @cindex local variables, specifying registers
6198 @cindex specifying registers for local variables
6199 @cindex registers for local variables
6201 You can define a local register variable with a specified register
6205 register int *foo asm ("a5");
6209 Here @code{a5} is the name of the register which should be used. Note
6210 that this is the same syntax used for defining global register
6211 variables, but for a local variable it would appear within a function.
6213 Naturally the register name is cpu-dependent, but this is not a
6214 problem, since specific registers are most often useful with explicit
6215 assembler instructions (@pxref{Extended Asm}). Both of these things
6216 generally require that you conditionalize your program according to
6219 In addition, operating systems on one type of cpu may differ in how they
6220 name the registers; then you would need additional conditionals. For
6221 example, some 68000 operating systems call this register @code{%a5}.
6223 Defining such a register variable does not reserve the register; it
6224 remains available for other uses in places where flow control determines
6225 the variable's value is not live.
6227 This option does not guarantee that GCC will generate code that has
6228 this variable in the register you specify at all times. You may not
6229 code an explicit reference to this register in the @emph{assembler
6230 instruction template} part of an @code{asm} statement and assume it will
6231 always refer to this variable. However, using the variable as an
6232 @code{asm} @emph{operand} guarantees that the specified register is used
6235 Stores into local register variables may be deleted when they appear to be dead
6236 according to dataflow analysis. References to local register variables may
6237 be deleted or moved or simplified.
6239 As for global register variables, it's recommended that you choose a
6240 register which is normally saved and restored by function calls on
6241 your machine, so that library routines will not clobber it. A common
6242 pitfall is to initialize multiple call-clobbered registers with
6243 arbitrary expressions, where a function call or library call for an
6244 arithmetic operator will overwrite a register value from a previous
6245 assignment, for example @code{r0} below:
6247 register int *p1 asm ("r0") = @dots{};
6248 register int *p2 asm ("r1") = @dots{};
6250 In those cases, a solution is to use a temporary variable for
6251 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
6253 @node Alternate Keywords
6254 @section Alternate Keywords
6255 @cindex alternate keywords
6256 @cindex keywords, alternate
6258 @option{-ansi} and the various @option{-std} options disable certain
6259 keywords. This causes trouble when you want to use GNU C extensions, or
6260 a general-purpose header file that should be usable by all programs,
6261 including ISO C programs. The keywords @code{asm}, @code{typeof} and
6262 @code{inline} are not available in programs compiled with
6263 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
6264 program compiled with @option{-std=c99} or @option{-std=c1x}). The
6266 @code{restrict} is only available when @option{-std=gnu99} (which will
6267 eventually be the default) or @option{-std=c99} (or the equivalent
6268 @option{-std=iso9899:1999}), or an option for a later standard
6271 The way to solve these problems is to put @samp{__} at the beginning and
6272 end of each problematical keyword. For example, use @code{__asm__}
6273 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
6275 Other C compilers won't accept these alternative keywords; if you want to
6276 compile with another compiler, you can define the alternate keywords as
6277 macros to replace them with the customary keywords. It looks like this:
6285 @findex __extension__
6287 @option{-pedantic} and other options cause warnings for many GNU C extensions.
6289 prevent such warnings within one expression by writing
6290 @code{__extension__} before the expression. @code{__extension__} has no
6291 effect aside from this.
6293 @node Incomplete Enums
6294 @section Incomplete @code{enum} Types
6296 You can define an @code{enum} tag without specifying its possible values.
6297 This results in an incomplete type, much like what you get if you write
6298 @code{struct foo} without describing the elements. A later declaration
6299 which does specify the possible values completes the type.
6301 You can't allocate variables or storage using the type while it is
6302 incomplete. However, you can work with pointers to that type.
6304 This extension may not be very useful, but it makes the handling of
6305 @code{enum} more consistent with the way @code{struct} and @code{union}
6308 This extension is not supported by GNU C++.
6310 @node Function Names
6311 @section Function Names as Strings
6312 @cindex @code{__func__} identifier
6313 @cindex @code{__FUNCTION__} identifier
6314 @cindex @code{__PRETTY_FUNCTION__} identifier
6316 GCC provides three magic variables which hold the name of the current
6317 function, as a string. The first of these is @code{__func__}, which
6318 is part of the C99 standard:
6320 The identifier @code{__func__} is implicitly declared by the translator
6321 as if, immediately following the opening brace of each function
6322 definition, the declaration
6325 static const char __func__[] = "function-name";
6329 appeared, where function-name is the name of the lexically-enclosing
6330 function. This name is the unadorned name of the function.
6332 @code{__FUNCTION__} is another name for @code{__func__}. Older
6333 versions of GCC recognize only this name. However, it is not
6334 standardized. For maximum portability, we recommend you use
6335 @code{__func__}, but provide a fallback definition with the
6339 #if __STDC_VERSION__ < 199901L
6341 # define __func__ __FUNCTION__
6343 # define __func__ "<unknown>"
6348 In C, @code{__PRETTY_FUNCTION__} is yet another name for
6349 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
6350 the type signature of the function as well as its bare name. For
6351 example, this program:
6355 extern int printf (char *, ...);
6362 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
6363 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
6381 __PRETTY_FUNCTION__ = void a::sub(int)
6384 These identifiers are not preprocessor macros. In GCC 3.3 and
6385 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
6386 were treated as string literals; they could be used to initialize
6387 @code{char} arrays, and they could be concatenated with other string
6388 literals. GCC 3.4 and later treat them as variables, like
6389 @code{__func__}. In C++, @code{__FUNCTION__} and
6390 @code{__PRETTY_FUNCTION__} have always been variables.
6392 @node Return Address
6393 @section Getting the Return or Frame Address of a Function
6395 These functions may be used to get information about the callers of a
6398 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
6399 This function returns the return address of the current function, or of
6400 one of its callers. The @var{level} argument is number of frames to
6401 scan up the call stack. A value of @code{0} yields the return address
6402 of the current function, a value of @code{1} yields the return address
6403 of the caller of the current function, and so forth. When inlining
6404 the expected behavior is that the function will return the address of
6405 the function that will be returned to. To work around this behavior use
6406 the @code{noinline} function attribute.
6408 The @var{level} argument must be a constant integer.
6410 On some machines it may be impossible to determine the return address of
6411 any function other than the current one; in such cases, or when the top
6412 of the stack has been reached, this function will return @code{0} or a
6413 random value. In addition, @code{__builtin_frame_address} may be used
6414 to determine if the top of the stack has been reached.
6416 Additional post-processing of the returned value may be needed, see
6417 @code{__builtin_extract_return_address}.
6419 This function should only be used with a nonzero argument for debugging
6423 @deftypefn {Built-in Function} {void *} __builtin_extract_return_address (void *@var{addr})
6424 The address as returned by @code{__builtin_return_address} may have to be fed
6425 through this function to get the actual encoded address. For example, on the
6426 31-bit S/390 platform the highest bit has to be masked out, or on SPARC
6427 platforms an offset has to be added for the true next instruction to be
6430 If no fixup is needed, this function simply passes through @var{addr}.
6433 @deftypefn {Built-in Function} {void *} __builtin_frob_return_address (void *@var{addr})
6434 This function does the reverse of @code{__builtin_extract_return_address}.
6437 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
6438 This function is similar to @code{__builtin_return_address}, but it
6439 returns the address of the function frame rather than the return address
6440 of the function. Calling @code{__builtin_frame_address} with a value of
6441 @code{0} yields the frame address of the current function, a value of
6442 @code{1} yields the frame address of the caller of the current function,
6445 The frame is the area on the stack which holds local variables and saved
6446 registers. The frame address is normally the address of the first word
6447 pushed on to the stack by the function. However, the exact definition
6448 depends upon the processor and the calling convention. If the processor
6449 has a dedicated frame pointer register, and the function has a frame,
6450 then @code{__builtin_frame_address} will return the value of the frame
6453 On some machines it may be impossible to determine the frame address of
6454 any function other than the current one; in such cases, or when the top
6455 of the stack has been reached, this function will return @code{0} if
6456 the first frame pointer is properly initialized by the startup code.
6458 This function should only be used with a nonzero argument for debugging
6462 @node Vector Extensions
6463 @section Using vector instructions through built-in functions
6465 On some targets, the instruction set contains SIMD vector instructions that
6466 operate on multiple values contained in one large register at the same time.
6467 For example, on the i386 the MMX, 3DNow!@: and SSE extensions can be used
6470 The first step in using these extensions is to provide the necessary data
6471 types. This should be done using an appropriate @code{typedef}:
6474 typedef int v4si __attribute__ ((vector_size (16)));
6477 The @code{int} type specifies the base type, while the attribute specifies
6478 the vector size for the variable, measured in bytes. For example, the
6479 declaration above causes the compiler to set the mode for the @code{v4si}
6480 type to be 16 bytes wide and divided into @code{int} sized units. For
6481 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
6482 corresponding mode of @code{foo} will be @acronym{V4SI}.
6484 The @code{vector_size} attribute is only applicable to integral and
6485 float scalars, although arrays, pointers, and function return values
6486 are allowed in conjunction with this construct.
6488 All the basic integer types can be used as base types, both as signed
6489 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
6490 @code{long long}. In addition, @code{float} and @code{double} can be
6491 used to build floating-point vector types.
6493 Specifying a combination that is not valid for the current architecture
6494 will cause GCC to synthesize the instructions using a narrower mode.
6495 For example, if you specify a variable of type @code{V4SI} and your
6496 architecture does not allow for this specific SIMD type, GCC will
6497 produce code that uses 4 @code{SIs}.
6499 The types defined in this manner can be used with a subset of normal C
6500 operations. Currently, GCC will allow using the following operators
6501 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~, %}@.
6503 The operations behave like C++ @code{valarrays}. Addition is defined as
6504 the addition of the corresponding elements of the operands. For
6505 example, in the code below, each of the 4 elements in @var{a} will be
6506 added to the corresponding 4 elements in @var{b} and the resulting
6507 vector will be stored in @var{c}.
6510 typedef int v4si __attribute__ ((vector_size (16)));
6517 Subtraction, multiplication, division, and the logical operations
6518 operate in a similar manner. Likewise, the result of using the unary
6519 minus or complement operators on a vector type is a vector whose
6520 elements are the negative or complemented values of the corresponding
6521 elements in the operand.
6523 In C it is possible to use shifting operators @code{<<}, @code{>>} on
6524 integer-type vectors. The operation is defined as following: @code{@{a0,
6525 a1, @dots{}, an@} >> @{b0, b1, @dots{}, bn@} == @{a0 >> b0, a1 >> b1,
6526 @dots{}, an >> bn@}}@. Vector operands must have the same number of
6527 elements. Additionally second operands can be a scalar integer in which
6528 case the scalar is converted to the type used by the vector operand (with
6529 possible truncation) and each element of this new vector is the scalar's
6531 Consider the following code.
6534 typedef int v4si __attribute__ ((vector_size (16)));
6538 b = a >> 1; /* b = a >> @{1,1,1,1@}; */
6541 In C vectors can be subscripted as if the vector were an array with
6542 the same number of elements and base type. Out of bound accesses
6543 invoke undefined behavior at runtime. Warnings for out of bound
6544 accesses for vector subscription can be enabled with
6545 @option{-Warray-bounds}.
6547 You can declare variables and use them in function calls and returns, as
6548 well as in assignments and some casts. You can specify a vector type as
6549 a return type for a function. Vector types can also be used as function
6550 arguments. It is possible to cast from one vector type to another,
6551 provided they are of the same size (in fact, you can also cast vectors
6552 to and from other datatypes of the same size).
6554 You cannot operate between vectors of different lengths or different
6555 signedness without a cast.
6557 A port that supports hardware vector operations, usually provides a set
6558 of built-in functions that can be used to operate on vectors. For
6559 example, a function to add two vectors and multiply the result by a
6560 third could look like this:
6563 v4si f (v4si a, v4si b, v4si c)
6565 v4si tmp = __builtin_addv4si (a, b);
6566 return __builtin_mulv4si (tmp, c);
6573 @findex __builtin_offsetof
6575 GCC implements for both C and C++ a syntactic extension to implement
6576 the @code{offsetof} macro.
6580 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
6582 offsetof_member_designator:
6584 | offsetof_member_designator "." @code{identifier}
6585 | offsetof_member_designator "[" @code{expr} "]"
6588 This extension is sufficient such that
6591 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
6594 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
6595 may be dependent. In either case, @var{member} may consist of a single
6596 identifier, or a sequence of member accesses and array references.
6598 @node Atomic Builtins
6599 @section Built-in functions for atomic memory access
6601 The following builtins are intended to be compatible with those described
6602 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
6603 section 7.4. As such, they depart from the normal GCC practice of using
6604 the ``__builtin_'' prefix, and further that they are overloaded such that
6605 they work on multiple types.
6607 The definition given in the Intel documentation allows only for the use of
6608 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
6609 counterparts. GCC will allow any integral scalar or pointer type that is
6610 1, 2, 4 or 8 bytes in length.
6612 Not all operations are supported by all target processors. If a particular
6613 operation cannot be implemented on the target processor, a warning will be
6614 generated and a call an external function will be generated. The external
6615 function will carry the same name as the builtin, with an additional suffix
6616 @samp{_@var{n}} where @var{n} is the size of the data type.
6618 @c ??? Should we have a mechanism to suppress this warning? This is almost
6619 @c useful for implementing the operation under the control of an external
6622 In most cases, these builtins are considered a @dfn{full barrier}. That is,
6623 no memory operand will be moved across the operation, either forward or
6624 backward. Further, instructions will be issued as necessary to prevent the
6625 processor from speculating loads across the operation and from queuing stores
6626 after the operation.
6628 All of the routines are described in the Intel documentation to take
6629 ``an optional list of variables protected by the memory barrier''. It's
6630 not clear what is meant by that; it could mean that @emph{only} the
6631 following variables are protected, or it could mean that these variables
6632 should in addition be protected. At present GCC ignores this list and
6633 protects all variables which are globally accessible. If in the future
6634 we make some use of this list, an empty list will continue to mean all
6635 globally accessible variables.
6638 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
6639 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
6640 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
6641 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
6642 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
6643 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
6644 @findex __sync_fetch_and_add
6645 @findex __sync_fetch_and_sub
6646 @findex __sync_fetch_and_or
6647 @findex __sync_fetch_and_and
6648 @findex __sync_fetch_and_xor
6649 @findex __sync_fetch_and_nand
6650 These builtins perform the operation suggested by the name, and
6651 returns the value that had previously been in memory. That is,
6654 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
6655 @{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; @} // nand
6658 @emph{Note:} GCC 4.4 and later implement @code{__sync_fetch_and_nand}
6659 builtin as @code{*ptr = ~(tmp & value)} instead of @code{*ptr = ~tmp & value}.
6661 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
6662 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
6663 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
6664 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
6665 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
6666 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
6667 @findex __sync_add_and_fetch
6668 @findex __sync_sub_and_fetch
6669 @findex __sync_or_and_fetch
6670 @findex __sync_and_and_fetch
6671 @findex __sync_xor_and_fetch
6672 @findex __sync_nand_and_fetch
6673 These builtins perform the operation suggested by the name, and
6674 return the new value. That is,
6677 @{ *ptr @var{op}= value; return *ptr; @}
6678 @{ *ptr = ~(*ptr & value); return *ptr; @} // nand
6681 @emph{Note:} GCC 4.4 and later implement @code{__sync_nand_and_fetch}
6682 builtin as @code{*ptr = ~(*ptr & value)} instead of
6683 @code{*ptr = ~*ptr & value}.
6685 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6686 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6687 @findex __sync_bool_compare_and_swap
6688 @findex __sync_val_compare_and_swap
6689 These builtins perform an atomic compare and swap. That is, if the current
6690 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
6693 The ``bool'' version returns true if the comparison is successful and
6694 @var{newval} was written. The ``val'' version returns the contents
6695 of @code{*@var{ptr}} before the operation.
6697 @item __sync_synchronize (...)
6698 @findex __sync_synchronize
6699 This builtin issues a full memory barrier.
6701 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
6702 @findex __sync_lock_test_and_set
6703 This builtin, as described by Intel, is not a traditional test-and-set
6704 operation, but rather an atomic exchange operation. It writes @var{value}
6705 into @code{*@var{ptr}}, and returns the previous contents of
6708 Many targets have only minimal support for such locks, and do not support
6709 a full exchange operation. In this case, a target may support reduced
6710 functionality here by which the @emph{only} valid value to store is the
6711 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
6712 is implementation defined.
6714 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
6715 This means that references after the builtin cannot move to (or be
6716 speculated to) before the builtin, but previous memory stores may not
6717 be globally visible yet, and previous memory loads may not yet be
6720 @item void __sync_lock_release (@var{type} *ptr, ...)
6721 @findex __sync_lock_release
6722 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
6723 Normally this means writing the constant 0 to @code{*@var{ptr}}.
6725 This builtin is not a full barrier, but rather a @dfn{release barrier}.
6726 This means that all previous memory stores are globally visible, and all
6727 previous memory loads have been satisfied, but following memory reads
6728 are not prevented from being speculated to before the barrier.
6731 @node Object Size Checking
6732 @section Object Size Checking Builtins
6733 @findex __builtin_object_size
6734 @findex __builtin___memcpy_chk
6735 @findex __builtin___mempcpy_chk
6736 @findex __builtin___memmove_chk
6737 @findex __builtin___memset_chk
6738 @findex __builtin___strcpy_chk
6739 @findex __builtin___stpcpy_chk
6740 @findex __builtin___strncpy_chk
6741 @findex __builtin___strcat_chk
6742 @findex __builtin___strncat_chk
6743 @findex __builtin___sprintf_chk
6744 @findex __builtin___snprintf_chk
6745 @findex __builtin___vsprintf_chk
6746 @findex __builtin___vsnprintf_chk
6747 @findex __builtin___printf_chk
6748 @findex __builtin___vprintf_chk
6749 @findex __builtin___fprintf_chk
6750 @findex __builtin___vfprintf_chk
6752 GCC implements a limited buffer overflow protection mechanism
6753 that can prevent some buffer overflow attacks.
6755 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
6756 is a built-in construct that returns a constant number of bytes from
6757 @var{ptr} to the end of the object @var{ptr} pointer points to
6758 (if known at compile time). @code{__builtin_object_size} never evaluates
6759 its arguments for side-effects. If there are any side-effects in them, it
6760 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6761 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
6762 point to and all of them are known at compile time, the returned number
6763 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
6764 0 and minimum if nonzero. If it is not possible to determine which objects
6765 @var{ptr} points to at compile time, @code{__builtin_object_size} should
6766 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6767 for @var{type} 2 or 3.
6769 @var{type} is an integer constant from 0 to 3. If the least significant
6770 bit is clear, objects are whole variables, if it is set, a closest
6771 surrounding subobject is considered the object a pointer points to.
6772 The second bit determines if maximum or minimum of remaining bytes
6776 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
6777 char *p = &var.buf1[1], *q = &var.b;
6779 /* Here the object p points to is var. */
6780 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
6781 /* The subobject p points to is var.buf1. */
6782 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
6783 /* The object q points to is var. */
6784 assert (__builtin_object_size (q, 0)
6785 == (char *) (&var + 1) - (char *) &var.b);
6786 /* The subobject q points to is var.b. */
6787 assert (__builtin_object_size (q, 1) == sizeof (var.b));
6791 There are built-in functions added for many common string operation
6792 functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk}
6793 built-in is provided. This built-in has an additional last argument,
6794 which is the number of bytes remaining in object the @var{dest}
6795 argument points to or @code{(size_t) -1} if the size is not known.
6797 The built-in functions are optimized into the normal string functions
6798 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
6799 it is known at compile time that the destination object will not
6800 be overflown. If the compiler can determine at compile time the
6801 object will be always overflown, it issues a warning.
6803 The intended use can be e.g.
6807 #define bos0(dest) __builtin_object_size (dest, 0)
6808 #define memcpy(dest, src, n) \
6809 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
6813 /* It is unknown what object p points to, so this is optimized
6814 into plain memcpy - no checking is possible. */
6815 memcpy (p, "abcde", n);
6816 /* Destination is known and length too. It is known at compile
6817 time there will be no overflow. */
6818 memcpy (&buf[5], "abcde", 5);
6819 /* Destination is known, but the length is not known at compile time.
6820 This will result in __memcpy_chk call that can check for overflow
6822 memcpy (&buf[5], "abcde", n);
6823 /* Destination is known and it is known at compile time there will
6824 be overflow. There will be a warning and __memcpy_chk call that
6825 will abort the program at runtime. */
6826 memcpy (&buf[6], "abcde", 5);
6829 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
6830 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
6831 @code{strcat} and @code{strncat}.
6833 There are also checking built-in functions for formatted output functions.
6835 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
6836 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6837 const char *fmt, ...);
6838 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
6840 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6841 const char *fmt, va_list ap);
6844 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
6845 etc.@: functions and can contain implementation specific flags on what
6846 additional security measures the checking function might take, such as
6847 handling @code{%n} differently.
6849 The @var{os} argument is the object size @var{s} points to, like in the
6850 other built-in functions. There is a small difference in the behavior
6851 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
6852 optimized into the non-checking functions only if @var{flag} is 0, otherwise
6853 the checking function is called with @var{os} argument set to
6856 In addition to this, there are checking built-in functions
6857 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
6858 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
6859 These have just one additional argument, @var{flag}, right before
6860 format string @var{fmt}. If the compiler is able to optimize them to
6861 @code{fputc} etc.@: functions, it will, otherwise the checking function
6862 should be called and the @var{flag} argument passed to it.
6864 @node Other Builtins
6865 @section Other built-in functions provided by GCC
6866 @cindex built-in functions
6867 @findex __builtin_fpclassify
6868 @findex __builtin_isfinite
6869 @findex __builtin_isnormal
6870 @findex __builtin_isgreater
6871 @findex __builtin_isgreaterequal
6872 @findex __builtin_isinf_sign
6873 @findex __builtin_isless
6874 @findex __builtin_islessequal
6875 @findex __builtin_islessgreater
6876 @findex __builtin_isunordered
6877 @findex __builtin_powi
6878 @findex __builtin_powif
6879 @findex __builtin_powil
7037 @findex fprintf_unlocked
7039 @findex fputs_unlocked
7156 @findex printf_unlocked
7188 @findex significandf
7189 @findex significandl
7260 GCC provides a large number of built-in functions other than the ones
7261 mentioned above. Some of these are for internal use in the processing
7262 of exceptions or variable-length argument lists and will not be
7263 documented here because they may change from time to time; we do not
7264 recommend general use of these functions.
7266 The remaining functions are provided for optimization purposes.
7268 @opindex fno-builtin
7269 GCC includes built-in versions of many of the functions in the standard
7270 C library. The versions prefixed with @code{__builtin_} will always be
7271 treated as having the same meaning as the C library function even if you
7272 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
7273 Many of these functions are only optimized in certain cases; if they are
7274 not optimized in a particular case, a call to the library function will
7279 Outside strict ISO C mode (@option{-ansi}, @option{-std=c90},
7280 @option{-std=c99} or @option{-std=c1x}), the functions
7281 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
7282 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
7283 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
7284 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
7285 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
7286 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
7287 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
7288 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
7289 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
7290 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
7291 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
7292 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
7293 @code{signbitd64}, @code{signbitd128}, @code{significandf},
7294 @code{significandl}, @code{significand}, @code{sincosf},
7295 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
7296 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
7297 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
7298 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
7300 may be handled as built-in functions.
7301 All these functions have corresponding versions
7302 prefixed with @code{__builtin_}, which may be used even in strict C90
7305 The ISO C99 functions
7306 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
7307 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
7308 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
7309 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
7310 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
7311 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
7312 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
7313 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
7314 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
7315 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
7316 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
7317 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
7318 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
7319 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
7320 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
7321 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
7322 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
7323 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
7324 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
7325 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
7326 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
7327 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
7328 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
7329 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
7330 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
7331 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
7332 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
7333 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
7334 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
7335 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
7336 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
7337 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
7338 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
7339 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
7340 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
7341 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
7342 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
7343 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
7344 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
7345 are handled as built-in functions
7346 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
7348 There are also built-in versions of the ISO C99 functions
7349 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
7350 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
7351 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
7352 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
7353 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
7354 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
7355 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
7356 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
7357 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
7358 that are recognized in any mode since ISO C90 reserves these names for
7359 the purpose to which ISO C99 puts them. All these functions have
7360 corresponding versions prefixed with @code{__builtin_}.
7362 The ISO C94 functions
7363 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
7364 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
7365 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
7367 are handled as built-in functions
7368 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
7370 The ISO C90 functions
7371 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
7372 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
7373 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
7374 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
7375 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
7376 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
7377 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
7378 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
7379 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
7380 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
7381 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
7382 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
7383 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
7384 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
7385 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
7386 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
7387 are all recognized as built-in functions unless
7388 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
7389 is specified for an individual function). All of these functions have
7390 corresponding versions prefixed with @code{__builtin_}.
7392 GCC provides built-in versions of the ISO C99 floating point comparison
7393 macros that avoid raising exceptions for unordered operands. They have
7394 the same names as the standard macros ( @code{isgreater},
7395 @code{isgreaterequal}, @code{isless}, @code{islessequal},
7396 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
7397 prefixed. We intend for a library implementor to be able to simply
7398 @code{#define} each standard macro to its built-in equivalent.
7399 In the same fashion, GCC provides @code{fpclassify}, @code{isfinite},
7400 @code{isinf_sign} and @code{isnormal} built-ins used with
7401 @code{__builtin_} prefixed. The @code{isinf} and @code{isnan}
7402 builtins appear both with and without the @code{__builtin_} prefix.
7404 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
7406 You can use the built-in function @code{__builtin_types_compatible_p} to
7407 determine whether two types are the same.
7409 This built-in function returns 1 if the unqualified versions of the
7410 types @var{type1} and @var{type2} (which are types, not expressions) are
7411 compatible, 0 otherwise. The result of this built-in function can be
7412 used in integer constant expressions.
7414 This built-in function ignores top level qualifiers (e.g., @code{const},
7415 @code{volatile}). For example, @code{int} is equivalent to @code{const
7418 The type @code{int[]} and @code{int[5]} are compatible. On the other
7419 hand, @code{int} and @code{char *} are not compatible, even if the size
7420 of their types, on the particular architecture are the same. Also, the
7421 amount of pointer indirection is taken into account when determining
7422 similarity. Consequently, @code{short *} is not similar to
7423 @code{short **}. Furthermore, two types that are typedefed are
7424 considered compatible if their underlying types are compatible.
7426 An @code{enum} type is not considered to be compatible with another
7427 @code{enum} type even if both are compatible with the same integer
7428 type; this is what the C standard specifies.
7429 For example, @code{enum @{foo, bar@}} is not similar to
7430 @code{enum @{hot, dog@}}.
7432 You would typically use this function in code whose execution varies
7433 depending on the arguments' types. For example:
7438 typeof (x) tmp = (x); \
7439 if (__builtin_types_compatible_p (typeof (x), long double)) \
7440 tmp = foo_long_double (tmp); \
7441 else if (__builtin_types_compatible_p (typeof (x), double)) \
7442 tmp = foo_double (tmp); \
7443 else if (__builtin_types_compatible_p (typeof (x), float)) \
7444 tmp = foo_float (tmp); \
7451 @emph{Note:} This construct is only available for C@.
7455 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
7457 You can use the built-in function @code{__builtin_choose_expr} to
7458 evaluate code depending on the value of a constant expression. This
7459 built-in function returns @var{exp1} if @var{const_exp}, which is an
7460 integer constant expression, is nonzero. Otherwise it returns @var{exp2}.
7462 This built-in function is analogous to the @samp{? :} operator in C,
7463 except that the expression returned has its type unaltered by promotion
7464 rules. Also, the built-in function does not evaluate the expression
7465 that was not chosen. For example, if @var{const_exp} evaluates to true,
7466 @var{exp2} is not evaluated even if it has side-effects.
7468 This built-in function can return an lvalue if the chosen argument is an
7471 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
7472 type. Similarly, if @var{exp2} is returned, its return type is the same
7479 __builtin_choose_expr ( \
7480 __builtin_types_compatible_p (typeof (x), double), \
7482 __builtin_choose_expr ( \
7483 __builtin_types_compatible_p (typeof (x), float), \
7485 /* @r{The void expression results in a compile-time error} \
7486 @r{when assigning the result to something.} */ \
7490 @emph{Note:} This construct is only available for C@. Furthermore, the
7491 unused expression (@var{exp1} or @var{exp2} depending on the value of
7492 @var{const_exp}) may still generate syntax errors. This may change in
7497 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
7498 You can use the built-in function @code{__builtin_constant_p} to
7499 determine if a value is known to be constant at compile-time and hence
7500 that GCC can perform constant-folding on expressions involving that
7501 value. The argument of the function is the value to test. The function
7502 returns the integer 1 if the argument is known to be a compile-time
7503 constant and 0 if it is not known to be a compile-time constant. A
7504 return of 0 does not indicate that the value is @emph{not} a constant,
7505 but merely that GCC cannot prove it is a constant with the specified
7506 value of the @option{-O} option.
7508 You would typically use this function in an embedded application where
7509 memory was a critical resource. If you have some complex calculation,
7510 you may want it to be folded if it involves constants, but need to call
7511 a function if it does not. For example:
7514 #define Scale_Value(X) \
7515 (__builtin_constant_p (X) \
7516 ? ((X) * SCALE + OFFSET) : Scale (X))
7519 You may use this built-in function in either a macro or an inline
7520 function. However, if you use it in an inlined function and pass an
7521 argument of the function as the argument to the built-in, GCC will
7522 never return 1 when you call the inline function with a string constant
7523 or compound literal (@pxref{Compound Literals}) and will not return 1
7524 when you pass a constant numeric value to the inline function unless you
7525 specify the @option{-O} option.
7527 You may also use @code{__builtin_constant_p} in initializers for static
7528 data. For instance, you can write
7531 static const int table[] = @{
7532 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
7538 This is an acceptable initializer even if @var{EXPRESSION} is not a
7539 constant expression, including the case where
7540 @code{__builtin_constant_p} returns 1 because @var{EXPRESSION} can be
7541 folded to a constant but @var{EXPRESSION} contains operands that would
7542 not otherwise be permitted in a static initializer (for example,
7543 @code{0 && foo ()}). GCC must be more conservative about evaluating the
7544 built-in in this case, because it has no opportunity to perform
7547 Previous versions of GCC did not accept this built-in in data
7548 initializers. The earliest version where it is completely safe is
7552 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
7553 @opindex fprofile-arcs
7554 You may use @code{__builtin_expect} to provide the compiler with
7555 branch prediction information. In general, you should prefer to
7556 use actual profile feedback for this (@option{-fprofile-arcs}), as
7557 programmers are notoriously bad at predicting how their programs
7558 actually perform. However, there are applications in which this
7559 data is hard to collect.
7561 The return value is the value of @var{exp}, which should be an integral
7562 expression. The semantics of the built-in are that it is expected that
7563 @var{exp} == @var{c}. For example:
7566 if (__builtin_expect (x, 0))
7571 would indicate that we do not expect to call @code{foo}, since
7572 we expect @code{x} to be zero. Since you are limited to integral
7573 expressions for @var{exp}, you should use constructions such as
7576 if (__builtin_expect (ptr != NULL, 1))
7581 when testing pointer or floating-point values.
7584 @deftypefn {Built-in Function} void __builtin_trap (void)
7585 This function causes the program to exit abnormally. GCC implements
7586 this function by using a target-dependent mechanism (such as
7587 intentionally executing an illegal instruction) or by calling
7588 @code{abort}. The mechanism used may vary from release to release so
7589 you should not rely on any particular implementation.
7592 @deftypefn {Built-in Function} void __builtin_unreachable (void)
7593 If control flow reaches the point of the @code{__builtin_unreachable},
7594 the program is undefined. It is useful in situations where the
7595 compiler cannot deduce the unreachability of the code.
7597 One such case is immediately following an @code{asm} statement that
7598 will either never terminate, or one that transfers control elsewhere
7599 and never returns. In this example, without the
7600 @code{__builtin_unreachable}, GCC would issue a warning that control
7601 reaches the end of a non-void function. It would also generate code
7602 to return after the @code{asm}.
7605 int f (int c, int v)
7613 asm("jmp error_handler");
7614 __builtin_unreachable ();
7619 Because the @code{asm} statement unconditionally transfers control out
7620 of the function, control will never reach the end of the function
7621 body. The @code{__builtin_unreachable} is in fact unreachable and
7622 communicates this fact to the compiler.
7624 Another use for @code{__builtin_unreachable} is following a call a
7625 function that never returns but that is not declared
7626 @code{__attribute__((noreturn))}, as in this example:
7629 void function_that_never_returns (void);
7639 function_that_never_returns ();
7640 __builtin_unreachable ();
7647 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
7648 This function is used to flush the processor's instruction cache for
7649 the region of memory between @var{begin} inclusive and @var{end}
7650 exclusive. Some targets require that the instruction cache be
7651 flushed, after modifying memory containing code, in order to obtain
7652 deterministic behavior.
7654 If the target does not require instruction cache flushes,
7655 @code{__builtin___clear_cache} has no effect. Otherwise either
7656 instructions are emitted in-line to clear the instruction cache or a
7657 call to the @code{__clear_cache} function in libgcc is made.
7660 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
7661 This function is used to minimize cache-miss latency by moving data into
7662 a cache before it is accessed.
7663 You can insert calls to @code{__builtin_prefetch} into code for which
7664 you know addresses of data in memory that is likely to be accessed soon.
7665 If the target supports them, data prefetch instructions will be generated.
7666 If the prefetch is done early enough before the access then the data will
7667 be in the cache by the time it is accessed.
7669 The value of @var{addr} is the address of the memory to prefetch.
7670 There are two optional arguments, @var{rw} and @var{locality}.
7671 The value of @var{rw} is a compile-time constant one or zero; one
7672 means that the prefetch is preparing for a write to the memory address
7673 and zero, the default, means that the prefetch is preparing for a read.
7674 The value @var{locality} must be a compile-time constant integer between
7675 zero and three. A value of zero means that the data has no temporal
7676 locality, so it need not be left in the cache after the access. A value
7677 of three means that the data has a high degree of temporal locality and
7678 should be left in all levels of cache possible. Values of one and two
7679 mean, respectively, a low or moderate degree of temporal locality. The
7683 for (i = 0; i < n; i++)
7686 __builtin_prefetch (&a[i+j], 1, 1);
7687 __builtin_prefetch (&b[i+j], 0, 1);
7692 Data prefetch does not generate faults if @var{addr} is invalid, but
7693 the address expression itself must be valid. For example, a prefetch
7694 of @code{p->next} will not fault if @code{p->next} is not a valid
7695 address, but evaluation will fault if @code{p} is not a valid address.
7697 If the target does not support data prefetch, the address expression
7698 is evaluated if it includes side effects but no other code is generated
7699 and GCC does not issue a warning.
7702 @deftypefn {Built-in Function} double __builtin_huge_val (void)
7703 Returns a positive infinity, if supported by the floating-point format,
7704 else @code{DBL_MAX}. This function is suitable for implementing the
7705 ISO C macro @code{HUGE_VAL}.
7708 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
7709 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
7712 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
7713 Similar to @code{__builtin_huge_val}, except the return
7714 type is @code{long double}.
7717 @deftypefn {Built-in Function} int __builtin_fpclassify (int, int, int, int, int, ...)
7718 This built-in implements the C99 fpclassify functionality. The first
7719 five int arguments should be the target library's notion of the
7720 possible FP classes and are used for return values. They must be
7721 constant values and they must appear in this order: @code{FP_NAN},
7722 @code{FP_INFINITE}, @code{FP_NORMAL}, @code{FP_SUBNORMAL} and
7723 @code{FP_ZERO}. The ellipsis is for exactly one floating point value
7724 to classify. GCC treats the last argument as type-generic, which
7725 means it does not do default promotion from float to double.
7728 @deftypefn {Built-in Function} double __builtin_inf (void)
7729 Similar to @code{__builtin_huge_val}, except a warning is generated
7730 if the target floating-point format does not support infinities.
7733 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
7734 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
7737 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
7738 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
7741 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
7742 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
7745 @deftypefn {Built-in Function} float __builtin_inff (void)
7746 Similar to @code{__builtin_inf}, except the return type is @code{float}.
7747 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
7750 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
7751 Similar to @code{__builtin_inf}, except the return
7752 type is @code{long double}.
7755 @deftypefn {Built-in Function} int __builtin_isinf_sign (...)
7756 Similar to @code{isinf}, except the return value will be negative for
7757 an argument of @code{-Inf}. Note while the parameter list is an
7758 ellipsis, this function only accepts exactly one floating point
7759 argument. GCC treats this parameter as type-generic, which means it
7760 does not do default promotion from float to double.
7763 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
7764 This is an implementation of the ISO C99 function @code{nan}.
7766 Since ISO C99 defines this function in terms of @code{strtod}, which we
7767 do not implement, a description of the parsing is in order. The string
7768 is parsed as by @code{strtol}; that is, the base is recognized by
7769 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
7770 in the significand such that the least significant bit of the number
7771 is at the least significant bit of the significand. The number is
7772 truncated to fit the significand field provided. The significand is
7773 forced to be a quiet NaN@.
7775 This function, if given a string literal all of which would have been
7776 consumed by strtol, is evaluated early enough that it is considered a
7777 compile-time constant.
7780 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
7781 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
7784 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
7785 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
7788 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
7789 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
7792 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
7793 Similar to @code{__builtin_nan}, except the return type is @code{float}.
7796 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
7797 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
7800 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
7801 Similar to @code{__builtin_nan}, except the significand is forced
7802 to be a signaling NaN@. The @code{nans} function is proposed by
7803 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
7806 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
7807 Similar to @code{__builtin_nans}, except the return type is @code{float}.
7810 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
7811 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
7814 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
7815 Returns one plus the index of the least significant 1-bit of @var{x}, or
7816 if @var{x} is zero, returns zero.
7819 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
7820 Returns the number of leading 0-bits in @var{x}, starting at the most
7821 significant bit position. If @var{x} is 0, the result is undefined.
7824 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
7825 Returns the number of trailing 0-bits in @var{x}, starting at the least
7826 significant bit position. If @var{x} is 0, the result is undefined.
7829 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
7830 Returns the number of 1-bits in @var{x}.
7833 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
7834 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
7838 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
7839 Similar to @code{__builtin_ffs}, except the argument type is
7840 @code{unsigned long}.
7843 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
7844 Similar to @code{__builtin_clz}, except the argument type is
7845 @code{unsigned long}.
7848 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
7849 Similar to @code{__builtin_ctz}, except the argument type is
7850 @code{unsigned long}.
7853 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
7854 Similar to @code{__builtin_popcount}, except the argument type is
7855 @code{unsigned long}.
7858 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
7859 Similar to @code{__builtin_parity}, except the argument type is
7860 @code{unsigned long}.
7863 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
7864 Similar to @code{__builtin_ffs}, except the argument type is
7865 @code{unsigned long long}.
7868 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
7869 Similar to @code{__builtin_clz}, except the argument type is
7870 @code{unsigned long long}.
7873 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
7874 Similar to @code{__builtin_ctz}, except the argument type is
7875 @code{unsigned long long}.
7878 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
7879 Similar to @code{__builtin_popcount}, except the argument type is
7880 @code{unsigned long long}.
7883 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
7884 Similar to @code{__builtin_parity}, except the argument type is
7885 @code{unsigned long long}.
7888 @deftypefn {Built-in Function} double __builtin_powi (double, int)
7889 Returns the first argument raised to the power of the second. Unlike the
7890 @code{pow} function no guarantees about precision and rounding are made.
7893 @deftypefn {Built-in Function} float __builtin_powif (float, int)
7894 Similar to @code{__builtin_powi}, except the argument and return types
7898 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
7899 Similar to @code{__builtin_powi}, except the argument and return types
7900 are @code{long double}.
7903 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
7904 Returns @var{x} with the order of the bytes reversed; for example,
7905 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
7909 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
7910 Similar to @code{__builtin_bswap32}, except the argument and return types
7914 @node Target Builtins
7915 @section Built-in Functions Specific to Particular Target Machines
7917 On some target machines, GCC supports many built-in functions specific
7918 to those machines. Generally these generate calls to specific machine
7919 instructions, but allow the compiler to schedule those calls.
7922 * Alpha Built-in Functions::
7923 * ARM iWMMXt Built-in Functions::
7924 * ARM NEON Intrinsics::
7925 * Blackfin Built-in Functions::
7926 * FR-V Built-in Functions::
7927 * X86 Built-in Functions::
7928 * MIPS DSP Built-in Functions::
7929 * MIPS Paired-Single Support::
7930 * MIPS Loongson Built-in Functions::
7931 * Other MIPS Built-in Functions::
7932 * picoChip Built-in Functions::
7933 * PowerPC AltiVec/VSX Built-in Functions::
7934 * RX Built-in Functions::
7935 * SPARC VIS Built-in Functions::
7936 * SPU Built-in Functions::
7939 @node Alpha Built-in Functions
7940 @subsection Alpha Built-in Functions
7942 These built-in functions are available for the Alpha family of
7943 processors, depending on the command-line switches used.
7945 The following built-in functions are always available. They
7946 all generate the machine instruction that is part of the name.
7949 long __builtin_alpha_implver (void)
7950 long __builtin_alpha_rpcc (void)
7951 long __builtin_alpha_amask (long)
7952 long __builtin_alpha_cmpbge (long, long)
7953 long __builtin_alpha_extbl (long, long)
7954 long __builtin_alpha_extwl (long, long)
7955 long __builtin_alpha_extll (long, long)
7956 long __builtin_alpha_extql (long, long)
7957 long __builtin_alpha_extwh (long, long)
7958 long __builtin_alpha_extlh (long, long)
7959 long __builtin_alpha_extqh (long, long)
7960 long __builtin_alpha_insbl (long, long)
7961 long __builtin_alpha_inswl (long, long)
7962 long __builtin_alpha_insll (long, long)
7963 long __builtin_alpha_insql (long, long)
7964 long __builtin_alpha_inswh (long, long)
7965 long __builtin_alpha_inslh (long, long)
7966 long __builtin_alpha_insqh (long, long)
7967 long __builtin_alpha_mskbl (long, long)
7968 long __builtin_alpha_mskwl (long, long)
7969 long __builtin_alpha_mskll (long, long)
7970 long __builtin_alpha_mskql (long, long)
7971 long __builtin_alpha_mskwh (long, long)
7972 long __builtin_alpha_msklh (long, long)
7973 long __builtin_alpha_mskqh (long, long)
7974 long __builtin_alpha_umulh (long, long)
7975 long __builtin_alpha_zap (long, long)
7976 long __builtin_alpha_zapnot (long, long)
7979 The following built-in functions are always with @option{-mmax}
7980 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
7981 later. They all generate the machine instruction that is part
7985 long __builtin_alpha_pklb (long)
7986 long __builtin_alpha_pkwb (long)
7987 long __builtin_alpha_unpkbl (long)
7988 long __builtin_alpha_unpkbw (long)
7989 long __builtin_alpha_minub8 (long, long)
7990 long __builtin_alpha_minsb8 (long, long)
7991 long __builtin_alpha_minuw4 (long, long)
7992 long __builtin_alpha_minsw4 (long, long)
7993 long __builtin_alpha_maxub8 (long, long)
7994 long __builtin_alpha_maxsb8 (long, long)
7995 long __builtin_alpha_maxuw4 (long, long)
7996 long __builtin_alpha_maxsw4 (long, long)
7997 long __builtin_alpha_perr (long, long)
8000 The following built-in functions are always with @option{-mcix}
8001 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
8002 later. They all generate the machine instruction that is part
8006 long __builtin_alpha_cttz (long)
8007 long __builtin_alpha_ctlz (long)
8008 long __builtin_alpha_ctpop (long)
8011 The following builtins are available on systems that use the OSF/1
8012 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
8013 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
8014 @code{rdval} and @code{wrval}.
8017 void *__builtin_thread_pointer (void)
8018 void __builtin_set_thread_pointer (void *)
8021 @node ARM iWMMXt Built-in Functions
8022 @subsection ARM iWMMXt Built-in Functions
8024 These built-in functions are available for the ARM family of
8025 processors when the @option{-mcpu=iwmmxt} switch is used:
8028 typedef int v2si __attribute__ ((vector_size (8)));
8029 typedef short v4hi __attribute__ ((vector_size (8)));
8030 typedef char v8qi __attribute__ ((vector_size (8)));
8032 int __builtin_arm_getwcx (int)
8033 void __builtin_arm_setwcx (int, int)
8034 int __builtin_arm_textrmsb (v8qi, int)
8035 int __builtin_arm_textrmsh (v4hi, int)
8036 int __builtin_arm_textrmsw (v2si, int)
8037 int __builtin_arm_textrmub (v8qi, int)
8038 int __builtin_arm_textrmuh (v4hi, int)
8039 int __builtin_arm_textrmuw (v2si, int)
8040 v8qi __builtin_arm_tinsrb (v8qi, int)
8041 v4hi __builtin_arm_tinsrh (v4hi, int)
8042 v2si __builtin_arm_tinsrw (v2si, int)
8043 long long __builtin_arm_tmia (long long, int, int)
8044 long long __builtin_arm_tmiabb (long long, int, int)
8045 long long __builtin_arm_tmiabt (long long, int, int)
8046 long long __builtin_arm_tmiaph (long long, int, int)
8047 long long __builtin_arm_tmiatb (long long, int, int)
8048 long long __builtin_arm_tmiatt (long long, int, int)
8049 int __builtin_arm_tmovmskb (v8qi)
8050 int __builtin_arm_tmovmskh (v4hi)
8051 int __builtin_arm_tmovmskw (v2si)
8052 long long __builtin_arm_waccb (v8qi)
8053 long long __builtin_arm_wacch (v4hi)
8054 long long __builtin_arm_waccw (v2si)
8055 v8qi __builtin_arm_waddb (v8qi, v8qi)
8056 v8qi __builtin_arm_waddbss (v8qi, v8qi)
8057 v8qi __builtin_arm_waddbus (v8qi, v8qi)
8058 v4hi __builtin_arm_waddh (v4hi, v4hi)
8059 v4hi __builtin_arm_waddhss (v4hi, v4hi)
8060 v4hi __builtin_arm_waddhus (v4hi, v4hi)
8061 v2si __builtin_arm_waddw (v2si, v2si)
8062 v2si __builtin_arm_waddwss (v2si, v2si)
8063 v2si __builtin_arm_waddwus (v2si, v2si)
8064 v8qi __builtin_arm_walign (v8qi, v8qi, int)
8065 long long __builtin_arm_wand(long long, long long)
8066 long long __builtin_arm_wandn (long long, long long)
8067 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
8068 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
8069 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
8070 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
8071 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
8072 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
8073 v2si __builtin_arm_wcmpeqw (v2si, v2si)
8074 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
8075 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
8076 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
8077 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
8078 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
8079 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
8080 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
8081 long long __builtin_arm_wmacsz (v4hi, v4hi)
8082 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
8083 long long __builtin_arm_wmacuz (v4hi, v4hi)
8084 v4hi __builtin_arm_wmadds (v4hi, v4hi)
8085 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
8086 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
8087 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
8088 v2si __builtin_arm_wmaxsw (v2si, v2si)
8089 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
8090 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
8091 v2si __builtin_arm_wmaxuw (v2si, v2si)
8092 v8qi __builtin_arm_wminsb (v8qi, v8qi)
8093 v4hi __builtin_arm_wminsh (v4hi, v4hi)
8094 v2si __builtin_arm_wminsw (v2si, v2si)
8095 v8qi __builtin_arm_wminub (v8qi, v8qi)
8096 v4hi __builtin_arm_wminuh (v4hi, v4hi)
8097 v2si __builtin_arm_wminuw (v2si, v2si)
8098 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
8099 v4hi __builtin_arm_wmulul (v4hi, v4hi)
8100 v4hi __builtin_arm_wmulum (v4hi, v4hi)
8101 long long __builtin_arm_wor (long long, long long)
8102 v2si __builtin_arm_wpackdss (long long, long long)
8103 v2si __builtin_arm_wpackdus (long long, long long)
8104 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
8105 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
8106 v4hi __builtin_arm_wpackwss (v2si, v2si)
8107 v4hi __builtin_arm_wpackwus (v2si, v2si)
8108 long long __builtin_arm_wrord (long long, long long)
8109 long long __builtin_arm_wrordi (long long, int)
8110 v4hi __builtin_arm_wrorh (v4hi, long long)
8111 v4hi __builtin_arm_wrorhi (v4hi, int)
8112 v2si __builtin_arm_wrorw (v2si, long long)
8113 v2si __builtin_arm_wrorwi (v2si, int)
8114 v2si __builtin_arm_wsadb (v8qi, v8qi)
8115 v2si __builtin_arm_wsadbz (v8qi, v8qi)
8116 v2si __builtin_arm_wsadh (v4hi, v4hi)
8117 v2si __builtin_arm_wsadhz (v4hi, v4hi)
8118 v4hi __builtin_arm_wshufh (v4hi, int)
8119 long long __builtin_arm_wslld (long long, long long)
8120 long long __builtin_arm_wslldi (long long, int)
8121 v4hi __builtin_arm_wsllh (v4hi, long long)
8122 v4hi __builtin_arm_wsllhi (v4hi, int)
8123 v2si __builtin_arm_wsllw (v2si, long long)
8124 v2si __builtin_arm_wsllwi (v2si, int)
8125 long long __builtin_arm_wsrad (long long, long long)
8126 long long __builtin_arm_wsradi (long long, int)
8127 v4hi __builtin_arm_wsrah (v4hi, long long)
8128 v4hi __builtin_arm_wsrahi (v4hi, int)
8129 v2si __builtin_arm_wsraw (v2si, long long)
8130 v2si __builtin_arm_wsrawi (v2si, int)
8131 long long __builtin_arm_wsrld (long long, long long)
8132 long long __builtin_arm_wsrldi (long long, int)
8133 v4hi __builtin_arm_wsrlh (v4hi, long long)
8134 v4hi __builtin_arm_wsrlhi (v4hi, int)
8135 v2si __builtin_arm_wsrlw (v2si, long long)
8136 v2si __builtin_arm_wsrlwi (v2si, int)
8137 v8qi __builtin_arm_wsubb (v8qi, v8qi)
8138 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
8139 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
8140 v4hi __builtin_arm_wsubh (v4hi, v4hi)
8141 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
8142 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
8143 v2si __builtin_arm_wsubw (v2si, v2si)
8144 v2si __builtin_arm_wsubwss (v2si, v2si)
8145 v2si __builtin_arm_wsubwus (v2si, v2si)
8146 v4hi __builtin_arm_wunpckehsb (v8qi)
8147 v2si __builtin_arm_wunpckehsh (v4hi)
8148 long long __builtin_arm_wunpckehsw (v2si)
8149 v4hi __builtin_arm_wunpckehub (v8qi)
8150 v2si __builtin_arm_wunpckehuh (v4hi)
8151 long long __builtin_arm_wunpckehuw (v2si)
8152 v4hi __builtin_arm_wunpckelsb (v8qi)
8153 v2si __builtin_arm_wunpckelsh (v4hi)
8154 long long __builtin_arm_wunpckelsw (v2si)
8155 v4hi __builtin_arm_wunpckelub (v8qi)
8156 v2si __builtin_arm_wunpckeluh (v4hi)
8157 long long __builtin_arm_wunpckeluw (v2si)
8158 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
8159 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
8160 v2si __builtin_arm_wunpckihw (v2si, v2si)
8161 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
8162 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
8163 v2si __builtin_arm_wunpckilw (v2si, v2si)
8164 long long __builtin_arm_wxor (long long, long long)
8165 long long __builtin_arm_wzero ()
8168 @node ARM NEON Intrinsics
8169 @subsection ARM NEON Intrinsics
8171 These built-in intrinsics for the ARM Advanced SIMD extension are available
8172 when the @option{-mfpu=neon} switch is used:
8174 @include arm-neon-intrinsics.texi
8176 @node Blackfin Built-in Functions
8177 @subsection Blackfin Built-in Functions
8179 Currently, there are two Blackfin-specific built-in functions. These are
8180 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
8181 using inline assembly; by using these built-in functions the compiler can
8182 automatically add workarounds for hardware errata involving these
8183 instructions. These functions are named as follows:
8186 void __builtin_bfin_csync (void)
8187 void __builtin_bfin_ssync (void)
8190 @node FR-V Built-in Functions
8191 @subsection FR-V Built-in Functions
8193 GCC provides many FR-V-specific built-in functions. In general,
8194 these functions are intended to be compatible with those described
8195 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
8196 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
8197 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
8198 pointer rather than by value.
8200 Most of the functions are named after specific FR-V instructions.
8201 Such functions are said to be ``directly mapped'' and are summarized
8202 here in tabular form.
8206 * Directly-mapped Integer Functions::
8207 * Directly-mapped Media Functions::
8208 * Raw read/write Functions::
8209 * Other Built-in Functions::
8212 @node Argument Types
8213 @subsubsection Argument Types
8215 The arguments to the built-in functions can be divided into three groups:
8216 register numbers, compile-time constants and run-time values. In order
8217 to make this classification clear at a glance, the arguments and return
8218 values are given the following pseudo types:
8220 @multitable @columnfractions .20 .30 .15 .35
8221 @item Pseudo type @tab Real C type @tab Constant? @tab Description
8222 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
8223 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
8224 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
8225 @item @code{uw2} @tab @code{unsigned long long} @tab No
8226 @tab an unsigned doubleword
8227 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
8228 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
8229 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
8230 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
8233 These pseudo types are not defined by GCC, they are simply a notational
8234 convenience used in this manual.
8236 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
8237 and @code{sw2} are evaluated at run time. They correspond to
8238 register operands in the underlying FR-V instructions.
8240 @code{const} arguments represent immediate operands in the underlying
8241 FR-V instructions. They must be compile-time constants.
8243 @code{acc} arguments are evaluated at compile time and specify the number
8244 of an accumulator register. For example, an @code{acc} argument of 2
8245 will select the ACC2 register.
8247 @code{iacc} arguments are similar to @code{acc} arguments but specify the
8248 number of an IACC register. See @pxref{Other Built-in Functions}
8251 @node Directly-mapped Integer Functions
8252 @subsubsection Directly-mapped Integer Functions
8254 The functions listed below map directly to FR-V I-type instructions.
8256 @multitable @columnfractions .45 .32 .23
8257 @item Function prototype @tab Example usage @tab Assembly output
8258 @item @code{sw1 __ADDSS (sw1, sw1)}
8259 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
8260 @tab @code{ADDSS @var{a},@var{b},@var{c}}
8261 @item @code{sw1 __SCAN (sw1, sw1)}
8262 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
8263 @tab @code{SCAN @var{a},@var{b},@var{c}}
8264 @item @code{sw1 __SCUTSS (sw1)}
8265 @tab @code{@var{b} = __SCUTSS (@var{a})}
8266 @tab @code{SCUTSS @var{a},@var{b}}
8267 @item @code{sw1 __SLASS (sw1, sw1)}
8268 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
8269 @tab @code{SLASS @var{a},@var{b},@var{c}}
8270 @item @code{void __SMASS (sw1, sw1)}
8271 @tab @code{__SMASS (@var{a}, @var{b})}
8272 @tab @code{SMASS @var{a},@var{b}}
8273 @item @code{void __SMSSS (sw1, sw1)}
8274 @tab @code{__SMSSS (@var{a}, @var{b})}
8275 @tab @code{SMSSS @var{a},@var{b}}
8276 @item @code{void __SMU (sw1, sw1)}
8277 @tab @code{__SMU (@var{a}, @var{b})}
8278 @tab @code{SMU @var{a},@var{b}}
8279 @item @code{sw2 __SMUL (sw1, sw1)}
8280 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
8281 @tab @code{SMUL @var{a},@var{b},@var{c}}
8282 @item @code{sw1 __SUBSS (sw1, sw1)}
8283 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
8284 @tab @code{SUBSS @var{a},@var{b},@var{c}}
8285 @item @code{uw2 __UMUL (uw1, uw1)}
8286 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
8287 @tab @code{UMUL @var{a},@var{b},@var{c}}
8290 @node Directly-mapped Media Functions
8291 @subsubsection Directly-mapped Media Functions
8293 The functions listed below map directly to FR-V M-type instructions.
8295 @multitable @columnfractions .45 .32 .23
8296 @item Function prototype @tab Example usage @tab Assembly output
8297 @item @code{uw1 __MABSHS (sw1)}
8298 @tab @code{@var{b} = __MABSHS (@var{a})}
8299 @tab @code{MABSHS @var{a},@var{b}}
8300 @item @code{void __MADDACCS (acc, acc)}
8301 @tab @code{__MADDACCS (@var{b}, @var{a})}
8302 @tab @code{MADDACCS @var{a},@var{b}}
8303 @item @code{sw1 __MADDHSS (sw1, sw1)}
8304 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
8305 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
8306 @item @code{uw1 __MADDHUS (uw1, uw1)}
8307 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
8308 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
8309 @item @code{uw1 __MAND (uw1, uw1)}
8310 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
8311 @tab @code{MAND @var{a},@var{b},@var{c}}
8312 @item @code{void __MASACCS (acc, acc)}
8313 @tab @code{__MASACCS (@var{b}, @var{a})}
8314 @tab @code{MASACCS @var{a},@var{b}}
8315 @item @code{uw1 __MAVEH (uw1, uw1)}
8316 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
8317 @tab @code{MAVEH @var{a},@var{b},@var{c}}
8318 @item @code{uw2 __MBTOH (uw1)}
8319 @tab @code{@var{b} = __MBTOH (@var{a})}
8320 @tab @code{MBTOH @var{a},@var{b}}
8321 @item @code{void __MBTOHE (uw1 *, uw1)}
8322 @tab @code{__MBTOHE (&@var{b}, @var{a})}
8323 @tab @code{MBTOHE @var{a},@var{b}}
8324 @item @code{void __MCLRACC (acc)}
8325 @tab @code{__MCLRACC (@var{a})}
8326 @tab @code{MCLRACC @var{a}}
8327 @item @code{void __MCLRACCA (void)}
8328 @tab @code{__MCLRACCA ()}
8329 @tab @code{MCLRACCA}
8330 @item @code{uw1 __Mcop1 (uw1, uw1)}
8331 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
8332 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
8333 @item @code{uw1 __Mcop2 (uw1, uw1)}
8334 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
8335 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
8336 @item @code{uw1 __MCPLHI (uw2, const)}
8337 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
8338 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
8339 @item @code{uw1 __MCPLI (uw2, const)}
8340 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
8341 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
8342 @item @code{void __MCPXIS (acc, sw1, sw1)}
8343 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
8344 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
8345 @item @code{void __MCPXIU (acc, uw1, uw1)}
8346 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
8347 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
8348 @item @code{void __MCPXRS (acc, sw1, sw1)}
8349 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
8350 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
8351 @item @code{void __MCPXRU (acc, uw1, uw1)}
8352 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
8353 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
8354 @item @code{uw1 __MCUT (acc, uw1)}
8355 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
8356 @tab @code{MCUT @var{a},@var{b},@var{c}}
8357 @item @code{uw1 __MCUTSS (acc, sw1)}
8358 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
8359 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
8360 @item @code{void __MDADDACCS (acc, acc)}
8361 @tab @code{__MDADDACCS (@var{b}, @var{a})}
8362 @tab @code{MDADDACCS @var{a},@var{b}}
8363 @item @code{void __MDASACCS (acc, acc)}
8364 @tab @code{__MDASACCS (@var{b}, @var{a})}
8365 @tab @code{MDASACCS @var{a},@var{b}}
8366 @item @code{uw2 __MDCUTSSI (acc, const)}
8367 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
8368 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
8369 @item @code{uw2 __MDPACKH (uw2, uw2)}
8370 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
8371 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
8372 @item @code{uw2 __MDROTLI (uw2, const)}
8373 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
8374 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
8375 @item @code{void __MDSUBACCS (acc, acc)}
8376 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
8377 @tab @code{MDSUBACCS @var{a},@var{b}}
8378 @item @code{void __MDUNPACKH (uw1 *, uw2)}
8379 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
8380 @tab @code{MDUNPACKH @var{a},@var{b}}
8381 @item @code{uw2 __MEXPDHD (uw1, const)}
8382 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
8383 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
8384 @item @code{uw1 __MEXPDHW (uw1, const)}
8385 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
8386 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
8387 @item @code{uw1 __MHDSETH (uw1, const)}
8388 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
8389 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
8390 @item @code{sw1 __MHDSETS (const)}
8391 @tab @code{@var{b} = __MHDSETS (@var{a})}
8392 @tab @code{MHDSETS #@var{a},@var{b}}
8393 @item @code{uw1 __MHSETHIH (uw1, const)}
8394 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
8395 @tab @code{MHSETHIH #@var{a},@var{b}}
8396 @item @code{sw1 __MHSETHIS (sw1, const)}
8397 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
8398 @tab @code{MHSETHIS #@var{a},@var{b}}
8399 @item @code{uw1 __MHSETLOH (uw1, const)}
8400 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
8401 @tab @code{MHSETLOH #@var{a},@var{b}}
8402 @item @code{sw1 __MHSETLOS (sw1, const)}
8403 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
8404 @tab @code{MHSETLOS #@var{a},@var{b}}
8405 @item @code{uw1 __MHTOB (uw2)}
8406 @tab @code{@var{b} = __MHTOB (@var{a})}
8407 @tab @code{MHTOB @var{a},@var{b}}
8408 @item @code{void __MMACHS (acc, sw1, sw1)}
8409 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
8410 @tab @code{MMACHS @var{a},@var{b},@var{c}}
8411 @item @code{void __MMACHU (acc, uw1, uw1)}
8412 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
8413 @tab @code{MMACHU @var{a},@var{b},@var{c}}
8414 @item @code{void __MMRDHS (acc, sw1, sw1)}
8415 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
8416 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
8417 @item @code{void __MMRDHU (acc, uw1, uw1)}
8418 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
8419 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
8420 @item @code{void __MMULHS (acc, sw1, sw1)}
8421 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
8422 @tab @code{MMULHS @var{a},@var{b},@var{c}}
8423 @item @code{void __MMULHU (acc, uw1, uw1)}
8424 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
8425 @tab @code{MMULHU @var{a},@var{b},@var{c}}
8426 @item @code{void __MMULXHS (acc, sw1, sw1)}
8427 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
8428 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
8429 @item @code{void __MMULXHU (acc, uw1, uw1)}
8430 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
8431 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
8432 @item @code{uw1 __MNOT (uw1)}
8433 @tab @code{@var{b} = __MNOT (@var{a})}
8434 @tab @code{MNOT @var{a},@var{b}}
8435 @item @code{uw1 __MOR (uw1, uw1)}
8436 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
8437 @tab @code{MOR @var{a},@var{b},@var{c}}
8438 @item @code{uw1 __MPACKH (uh, uh)}
8439 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
8440 @tab @code{MPACKH @var{a},@var{b},@var{c}}
8441 @item @code{sw2 __MQADDHSS (sw2, sw2)}
8442 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
8443 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
8444 @item @code{uw2 __MQADDHUS (uw2, uw2)}
8445 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
8446 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
8447 @item @code{void __MQCPXIS (acc, sw2, sw2)}
8448 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
8449 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
8450 @item @code{void __MQCPXIU (acc, uw2, uw2)}
8451 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
8452 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
8453 @item @code{void __MQCPXRS (acc, sw2, sw2)}
8454 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
8455 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
8456 @item @code{void __MQCPXRU (acc, uw2, uw2)}
8457 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
8458 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
8459 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
8460 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
8461 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
8462 @item @code{sw2 __MQLMTHS (sw2, sw2)}
8463 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
8464 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
8465 @item @code{void __MQMACHS (acc, sw2, sw2)}
8466 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
8467 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
8468 @item @code{void __MQMACHU (acc, uw2, uw2)}
8469 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
8470 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
8471 @item @code{void __MQMACXHS (acc, sw2, sw2)}
8472 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
8473 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
8474 @item @code{void __MQMULHS (acc, sw2, sw2)}
8475 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
8476 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
8477 @item @code{void __MQMULHU (acc, uw2, uw2)}
8478 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
8479 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
8480 @item @code{void __MQMULXHS (acc, sw2, sw2)}
8481 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
8482 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
8483 @item @code{void __MQMULXHU (acc, uw2, uw2)}
8484 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
8485 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
8486 @item @code{sw2 __MQSATHS (sw2, sw2)}
8487 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
8488 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
8489 @item @code{uw2 __MQSLLHI (uw2, int)}
8490 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
8491 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
8492 @item @code{sw2 __MQSRAHI (sw2, int)}
8493 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
8494 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
8495 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
8496 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
8497 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
8498 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
8499 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
8500 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
8501 @item @code{void __MQXMACHS (acc, sw2, sw2)}
8502 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
8503 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
8504 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
8505 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
8506 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
8507 @item @code{uw1 __MRDACC (acc)}
8508 @tab @code{@var{b} = __MRDACC (@var{a})}
8509 @tab @code{MRDACC @var{a},@var{b}}
8510 @item @code{uw1 __MRDACCG (acc)}
8511 @tab @code{@var{b} = __MRDACCG (@var{a})}
8512 @tab @code{MRDACCG @var{a},@var{b}}
8513 @item @code{uw1 __MROTLI (uw1, const)}
8514 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
8515 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
8516 @item @code{uw1 __MROTRI (uw1, const)}
8517 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
8518 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
8519 @item @code{sw1 __MSATHS (sw1, sw1)}
8520 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
8521 @tab @code{MSATHS @var{a},@var{b},@var{c}}
8522 @item @code{uw1 __MSATHU (uw1, uw1)}
8523 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
8524 @tab @code{MSATHU @var{a},@var{b},@var{c}}
8525 @item @code{uw1 __MSLLHI (uw1, const)}
8526 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
8527 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
8528 @item @code{sw1 __MSRAHI (sw1, const)}
8529 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
8530 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
8531 @item @code{uw1 __MSRLHI (uw1, const)}
8532 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
8533 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
8534 @item @code{void __MSUBACCS (acc, acc)}
8535 @tab @code{__MSUBACCS (@var{b}, @var{a})}
8536 @tab @code{MSUBACCS @var{a},@var{b}}
8537 @item @code{sw1 __MSUBHSS (sw1, sw1)}
8538 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
8539 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
8540 @item @code{uw1 __MSUBHUS (uw1, uw1)}
8541 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
8542 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
8543 @item @code{void __MTRAP (void)}
8544 @tab @code{__MTRAP ()}
8546 @item @code{uw2 __MUNPACKH (uw1)}
8547 @tab @code{@var{b} = __MUNPACKH (@var{a})}
8548 @tab @code{MUNPACKH @var{a},@var{b}}
8549 @item @code{uw1 __MWCUT (uw2, uw1)}
8550 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
8551 @tab @code{MWCUT @var{a},@var{b},@var{c}}
8552 @item @code{void __MWTACC (acc, uw1)}
8553 @tab @code{__MWTACC (@var{b}, @var{a})}
8554 @tab @code{MWTACC @var{a},@var{b}}
8555 @item @code{void __MWTACCG (acc, uw1)}
8556 @tab @code{__MWTACCG (@var{b}, @var{a})}
8557 @tab @code{MWTACCG @var{a},@var{b}}
8558 @item @code{uw1 __MXOR (uw1, uw1)}
8559 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
8560 @tab @code{MXOR @var{a},@var{b},@var{c}}
8563 @node Raw read/write Functions
8564 @subsubsection Raw read/write Functions
8566 This sections describes built-in functions related to read and write
8567 instructions to access memory. These functions generate
8568 @code{membar} instructions to flush the I/O load and stores where
8569 appropriate, as described in Fujitsu's manual described above.
8573 @item unsigned char __builtin_read8 (void *@var{data})
8574 @item unsigned short __builtin_read16 (void *@var{data})
8575 @item unsigned long __builtin_read32 (void *@var{data})
8576 @item unsigned long long __builtin_read64 (void *@var{data})
8578 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
8579 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
8580 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
8581 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
8584 @node Other Built-in Functions
8585 @subsubsection Other Built-in Functions
8587 This section describes built-in functions that are not named after
8588 a specific FR-V instruction.
8591 @item sw2 __IACCreadll (iacc @var{reg})
8592 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
8593 for future expansion and must be 0.
8595 @item sw1 __IACCreadl (iacc @var{reg})
8596 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
8597 Other values of @var{reg} are rejected as invalid.
8599 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
8600 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
8601 is reserved for future expansion and must be 0.
8603 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
8604 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
8605 is 1. Other values of @var{reg} are rejected as invalid.
8607 @item void __data_prefetch0 (const void *@var{x})
8608 Use the @code{dcpl} instruction to load the contents of address @var{x}
8609 into the data cache.
8611 @item void __data_prefetch (const void *@var{x})
8612 Use the @code{nldub} instruction to load the contents of address @var{x}
8613 into the data cache. The instruction will be issued in slot I1@.
8616 @node X86 Built-in Functions
8617 @subsection X86 Built-in Functions
8619 These built-in functions are available for the i386 and x86-64 family
8620 of computers, depending on the command-line switches used.
8622 Note that, if you specify command-line switches such as @option{-msse},
8623 the compiler could use the extended instruction sets even if the built-ins
8624 are not used explicitly in the program. For this reason, applications
8625 which perform runtime CPU detection must compile separate files for each
8626 supported architecture, using the appropriate flags. In particular,
8627 the file containing the CPU detection code should be compiled without
8630 The following machine modes are available for use with MMX built-in functions
8631 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
8632 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
8633 vector of eight 8-bit integers. Some of the built-in functions operate on
8634 MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode.
8636 If 3DNow!@: extensions are enabled, @code{V2SF} is used as a mode for a vector
8637 of two 32-bit floating point values.
8639 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
8640 floating point values. Some instructions use a vector of four 32-bit
8641 integers, these use @code{V4SI}. Finally, some instructions operate on an
8642 entire vector register, interpreting it as a 128-bit integer, these use mode
8645 In 64-bit mode, the x86-64 family of processors uses additional built-in
8646 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
8647 floating point and @code{TC} 128-bit complex floating point values.
8649 The following floating point built-in functions are available in 64-bit
8650 mode. All of them implement the function that is part of the name.
8653 __float128 __builtin_fabsq (__float128)
8654 __float128 __builtin_copysignq (__float128, __float128)
8657 The following floating point built-in functions are made available in the
8661 @item __float128 __builtin_infq (void)
8662 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
8663 @findex __builtin_infq
8665 @item __float128 __builtin_huge_valq (void)
8666 Similar to @code{__builtin_huge_val}, except the return type is @code{__float128}.
8667 @findex __builtin_huge_valq
8670 The following built-in functions are made available by @option{-mmmx}.
8671 All of them generate the machine instruction that is part of the name.
8674 v8qi __builtin_ia32_paddb (v8qi, v8qi)
8675 v4hi __builtin_ia32_paddw (v4hi, v4hi)
8676 v2si __builtin_ia32_paddd (v2si, v2si)
8677 v8qi __builtin_ia32_psubb (v8qi, v8qi)
8678 v4hi __builtin_ia32_psubw (v4hi, v4hi)
8679 v2si __builtin_ia32_psubd (v2si, v2si)
8680 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
8681 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
8682 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
8683 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
8684 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
8685 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
8686 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
8687 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
8688 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
8689 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
8690 di __builtin_ia32_pand (di, di)
8691 di __builtin_ia32_pandn (di,di)
8692 di __builtin_ia32_por (di, di)
8693 di __builtin_ia32_pxor (di, di)
8694 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
8695 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
8696 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
8697 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
8698 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
8699 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
8700 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
8701 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
8702 v2si __builtin_ia32_punpckhdq (v2si, v2si)
8703 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
8704 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
8705 v2si __builtin_ia32_punpckldq (v2si, v2si)
8706 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
8707 v4hi __builtin_ia32_packssdw (v2si, v2si)
8708 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
8710 v4hi __builtin_ia32_psllw (v4hi, v4hi)
8711 v2si __builtin_ia32_pslld (v2si, v2si)
8712 v1di __builtin_ia32_psllq (v1di, v1di)
8713 v4hi __builtin_ia32_psrlw (v4hi, v4hi)
8714 v2si __builtin_ia32_psrld (v2si, v2si)
8715 v1di __builtin_ia32_psrlq (v1di, v1di)
8716 v4hi __builtin_ia32_psraw (v4hi, v4hi)
8717 v2si __builtin_ia32_psrad (v2si, v2si)
8718 v4hi __builtin_ia32_psllwi (v4hi, int)
8719 v2si __builtin_ia32_pslldi (v2si, int)
8720 v1di __builtin_ia32_psllqi (v1di, int)
8721 v4hi __builtin_ia32_psrlwi (v4hi, int)
8722 v2si __builtin_ia32_psrldi (v2si, int)
8723 v1di __builtin_ia32_psrlqi (v1di, int)
8724 v4hi __builtin_ia32_psrawi (v4hi, int)
8725 v2si __builtin_ia32_psradi (v2si, int)
8729 The following built-in functions are made available either with
8730 @option{-msse}, or with a combination of @option{-m3dnow} and
8731 @option{-march=athlon}. All of them generate the machine
8732 instruction that is part of the name.
8735 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
8736 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
8737 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
8738 v1di __builtin_ia32_psadbw (v8qi, v8qi)
8739 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
8740 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
8741 v8qi __builtin_ia32_pminub (v8qi, v8qi)
8742 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
8743 int __builtin_ia32_pextrw (v4hi, int)
8744 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
8745 int __builtin_ia32_pmovmskb (v8qi)
8746 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
8747 void __builtin_ia32_movntq (di *, di)
8748 void __builtin_ia32_sfence (void)
8751 The following built-in functions are available when @option{-msse} is used.
8752 All of them generate the machine instruction that is part of the name.
8755 int __builtin_ia32_comieq (v4sf, v4sf)
8756 int __builtin_ia32_comineq (v4sf, v4sf)
8757 int __builtin_ia32_comilt (v4sf, v4sf)
8758 int __builtin_ia32_comile (v4sf, v4sf)
8759 int __builtin_ia32_comigt (v4sf, v4sf)
8760 int __builtin_ia32_comige (v4sf, v4sf)
8761 int __builtin_ia32_ucomieq (v4sf, v4sf)
8762 int __builtin_ia32_ucomineq (v4sf, v4sf)
8763 int __builtin_ia32_ucomilt (v4sf, v4sf)
8764 int __builtin_ia32_ucomile (v4sf, v4sf)
8765 int __builtin_ia32_ucomigt (v4sf, v4sf)
8766 int __builtin_ia32_ucomige (v4sf, v4sf)
8767 v4sf __builtin_ia32_addps (v4sf, v4sf)
8768 v4sf __builtin_ia32_subps (v4sf, v4sf)
8769 v4sf __builtin_ia32_mulps (v4sf, v4sf)
8770 v4sf __builtin_ia32_divps (v4sf, v4sf)
8771 v4sf __builtin_ia32_addss (v4sf, v4sf)
8772 v4sf __builtin_ia32_subss (v4sf, v4sf)
8773 v4sf __builtin_ia32_mulss (v4sf, v4sf)
8774 v4sf __builtin_ia32_divss (v4sf, v4sf)
8775 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
8776 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
8777 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
8778 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
8779 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
8780 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
8781 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
8782 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
8783 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
8784 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
8785 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
8786 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
8787 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
8788 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
8789 v4si __builtin_ia32_cmpless (v4sf, v4sf)
8790 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
8791 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
8792 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
8793 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
8794 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
8795 v4sf __builtin_ia32_maxps (v4sf, v4sf)
8796 v4sf __builtin_ia32_maxss (v4sf, v4sf)
8797 v4sf __builtin_ia32_minps (v4sf, v4sf)
8798 v4sf __builtin_ia32_minss (v4sf, v4sf)
8799 v4sf __builtin_ia32_andps (v4sf, v4sf)
8800 v4sf __builtin_ia32_andnps (v4sf, v4sf)
8801 v4sf __builtin_ia32_orps (v4sf, v4sf)
8802 v4sf __builtin_ia32_xorps (v4sf, v4sf)
8803 v4sf __builtin_ia32_movss (v4sf, v4sf)
8804 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
8805 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
8806 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
8807 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
8808 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
8809 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
8810 v2si __builtin_ia32_cvtps2pi (v4sf)
8811 int __builtin_ia32_cvtss2si (v4sf)
8812 v2si __builtin_ia32_cvttps2pi (v4sf)
8813 int __builtin_ia32_cvttss2si (v4sf)
8814 v4sf __builtin_ia32_rcpps (v4sf)
8815 v4sf __builtin_ia32_rsqrtps (v4sf)
8816 v4sf __builtin_ia32_sqrtps (v4sf)
8817 v4sf __builtin_ia32_rcpss (v4sf)
8818 v4sf __builtin_ia32_rsqrtss (v4sf)
8819 v4sf __builtin_ia32_sqrtss (v4sf)
8820 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
8821 void __builtin_ia32_movntps (float *, v4sf)
8822 int __builtin_ia32_movmskps (v4sf)
8825 The following built-in functions are available when @option{-msse} is used.
8828 @item v4sf __builtin_ia32_loadaps (float *)
8829 Generates the @code{movaps} machine instruction as a load from memory.
8830 @item void __builtin_ia32_storeaps (float *, v4sf)
8831 Generates the @code{movaps} machine instruction as a store to memory.
8832 @item v4sf __builtin_ia32_loadups (float *)
8833 Generates the @code{movups} machine instruction as a load from memory.
8834 @item void __builtin_ia32_storeups (float *, v4sf)
8835 Generates the @code{movups} machine instruction as a store to memory.
8836 @item v4sf __builtin_ia32_loadsss (float *)
8837 Generates the @code{movss} machine instruction as a load from memory.
8838 @item void __builtin_ia32_storess (float *, v4sf)
8839 Generates the @code{movss} machine instruction as a store to memory.
8840 @item v4sf __builtin_ia32_loadhps (v4sf, const v2sf *)
8841 Generates the @code{movhps} machine instruction as a load from memory.
8842 @item v4sf __builtin_ia32_loadlps (v4sf, const v2sf *)
8843 Generates the @code{movlps} machine instruction as a load from memory
8844 @item void __builtin_ia32_storehps (v2sf *, v4sf)
8845 Generates the @code{movhps} machine instruction as a store to memory.
8846 @item void __builtin_ia32_storelps (v2sf *, v4sf)
8847 Generates the @code{movlps} machine instruction as a store to memory.
8850 The following built-in functions are available when @option{-msse2} is used.
8851 All of them generate the machine instruction that is part of the name.
8854 int __builtin_ia32_comisdeq (v2df, v2df)
8855 int __builtin_ia32_comisdlt (v2df, v2df)
8856 int __builtin_ia32_comisdle (v2df, v2df)
8857 int __builtin_ia32_comisdgt (v2df, v2df)
8858 int __builtin_ia32_comisdge (v2df, v2df)
8859 int __builtin_ia32_comisdneq (v2df, v2df)
8860 int __builtin_ia32_ucomisdeq (v2df, v2df)
8861 int __builtin_ia32_ucomisdlt (v2df, v2df)
8862 int __builtin_ia32_ucomisdle (v2df, v2df)
8863 int __builtin_ia32_ucomisdgt (v2df, v2df)
8864 int __builtin_ia32_ucomisdge (v2df, v2df)
8865 int __builtin_ia32_ucomisdneq (v2df, v2df)
8866 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
8867 v2df __builtin_ia32_cmpltpd (v2df, v2df)
8868 v2df __builtin_ia32_cmplepd (v2df, v2df)
8869 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
8870 v2df __builtin_ia32_cmpgepd (v2df, v2df)
8871 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
8872 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
8873 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
8874 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
8875 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
8876 v2df __builtin_ia32_cmpngepd (v2df, v2df)
8877 v2df __builtin_ia32_cmpordpd (v2df, v2df)
8878 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
8879 v2df __builtin_ia32_cmpltsd (v2df, v2df)
8880 v2df __builtin_ia32_cmplesd (v2df, v2df)
8881 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
8882 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
8883 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
8884 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
8885 v2df __builtin_ia32_cmpordsd (v2df, v2df)
8886 v2di __builtin_ia32_paddq (v2di, v2di)
8887 v2di __builtin_ia32_psubq (v2di, v2di)
8888 v2df __builtin_ia32_addpd (v2df, v2df)
8889 v2df __builtin_ia32_subpd (v2df, v2df)
8890 v2df __builtin_ia32_mulpd (v2df, v2df)
8891 v2df __builtin_ia32_divpd (v2df, v2df)
8892 v2df __builtin_ia32_addsd (v2df, v2df)
8893 v2df __builtin_ia32_subsd (v2df, v2df)
8894 v2df __builtin_ia32_mulsd (v2df, v2df)
8895 v2df __builtin_ia32_divsd (v2df, v2df)
8896 v2df __builtin_ia32_minpd (v2df, v2df)
8897 v2df __builtin_ia32_maxpd (v2df, v2df)
8898 v2df __builtin_ia32_minsd (v2df, v2df)
8899 v2df __builtin_ia32_maxsd (v2df, v2df)
8900 v2df __builtin_ia32_andpd (v2df, v2df)
8901 v2df __builtin_ia32_andnpd (v2df, v2df)
8902 v2df __builtin_ia32_orpd (v2df, v2df)
8903 v2df __builtin_ia32_xorpd (v2df, v2df)
8904 v2df __builtin_ia32_movsd (v2df, v2df)
8905 v2df __builtin_ia32_unpckhpd (v2df, v2df)
8906 v2df __builtin_ia32_unpcklpd (v2df, v2df)
8907 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
8908 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
8909 v4si __builtin_ia32_paddd128 (v4si, v4si)
8910 v2di __builtin_ia32_paddq128 (v2di, v2di)
8911 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
8912 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
8913 v4si __builtin_ia32_psubd128 (v4si, v4si)
8914 v2di __builtin_ia32_psubq128 (v2di, v2di)
8915 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
8916 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
8917 v2di __builtin_ia32_pand128 (v2di, v2di)
8918 v2di __builtin_ia32_pandn128 (v2di, v2di)
8919 v2di __builtin_ia32_por128 (v2di, v2di)
8920 v2di __builtin_ia32_pxor128 (v2di, v2di)
8921 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
8922 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
8923 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
8924 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
8925 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
8926 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
8927 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
8928 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
8929 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
8930 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
8931 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
8932 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
8933 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
8934 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
8935 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
8936 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
8937 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
8938 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
8939 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
8940 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
8941 v16qi __builtin_ia32_packsswb128 (v8hi, v8hi)
8942 v8hi __builtin_ia32_packssdw128 (v4si, v4si)
8943 v16qi __builtin_ia32_packuswb128 (v8hi, v8hi)
8944 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
8945 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
8946 v2df __builtin_ia32_loadupd (double *)
8947 void __builtin_ia32_storeupd (double *, v2df)
8948 v2df __builtin_ia32_loadhpd (v2df, double const *)
8949 v2df __builtin_ia32_loadlpd (v2df, double const *)
8950 int __builtin_ia32_movmskpd (v2df)
8951 int __builtin_ia32_pmovmskb128 (v16qi)
8952 void __builtin_ia32_movnti (int *, int)
8953 void __builtin_ia32_movntpd (double *, v2df)
8954 void __builtin_ia32_movntdq (v2df *, v2df)
8955 v4si __builtin_ia32_pshufd (v4si, int)
8956 v8hi __builtin_ia32_pshuflw (v8hi, int)
8957 v8hi __builtin_ia32_pshufhw (v8hi, int)
8958 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
8959 v2df __builtin_ia32_sqrtpd (v2df)
8960 v2df __builtin_ia32_sqrtsd (v2df)
8961 v2df __builtin_ia32_shufpd (v2df, v2df, int)
8962 v2df __builtin_ia32_cvtdq2pd (v4si)
8963 v4sf __builtin_ia32_cvtdq2ps (v4si)
8964 v4si __builtin_ia32_cvtpd2dq (v2df)
8965 v2si __builtin_ia32_cvtpd2pi (v2df)
8966 v4sf __builtin_ia32_cvtpd2ps (v2df)
8967 v4si __builtin_ia32_cvttpd2dq (v2df)
8968 v2si __builtin_ia32_cvttpd2pi (v2df)
8969 v2df __builtin_ia32_cvtpi2pd (v2si)
8970 int __builtin_ia32_cvtsd2si (v2df)
8971 int __builtin_ia32_cvttsd2si (v2df)
8972 long long __builtin_ia32_cvtsd2si64 (v2df)
8973 long long __builtin_ia32_cvttsd2si64 (v2df)
8974 v4si __builtin_ia32_cvtps2dq (v4sf)
8975 v2df __builtin_ia32_cvtps2pd (v4sf)
8976 v4si __builtin_ia32_cvttps2dq (v4sf)
8977 v2df __builtin_ia32_cvtsi2sd (v2df, int)
8978 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
8979 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
8980 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
8981 void __builtin_ia32_clflush (const void *)
8982 void __builtin_ia32_lfence (void)
8983 void __builtin_ia32_mfence (void)
8984 v16qi __builtin_ia32_loaddqu (const char *)
8985 void __builtin_ia32_storedqu (char *, v16qi)
8986 v1di __builtin_ia32_pmuludq (v2si, v2si)
8987 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
8988 v8hi __builtin_ia32_psllw128 (v8hi, v8hi)
8989 v4si __builtin_ia32_pslld128 (v4si, v4si)
8990 v2di __builtin_ia32_psllq128 (v2di, v2di)
8991 v8hi __builtin_ia32_psrlw128 (v8hi, v8hi)
8992 v4si __builtin_ia32_psrld128 (v4si, v4si)
8993 v2di __builtin_ia32_psrlq128 (v2di, v2di)
8994 v8hi __builtin_ia32_psraw128 (v8hi, v8hi)
8995 v4si __builtin_ia32_psrad128 (v4si, v4si)
8996 v2di __builtin_ia32_pslldqi128 (v2di, int)
8997 v8hi __builtin_ia32_psllwi128 (v8hi, int)
8998 v4si __builtin_ia32_pslldi128 (v4si, int)
8999 v2di __builtin_ia32_psllqi128 (v2di, int)
9000 v2di __builtin_ia32_psrldqi128 (v2di, int)
9001 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
9002 v4si __builtin_ia32_psrldi128 (v4si, int)
9003 v2di __builtin_ia32_psrlqi128 (v2di, int)
9004 v8hi __builtin_ia32_psrawi128 (v8hi, int)
9005 v4si __builtin_ia32_psradi128 (v4si, int)
9006 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
9007 v2di __builtin_ia32_movq128 (v2di)
9010 The following built-in functions are available when @option{-msse3} is used.
9011 All of them generate the machine instruction that is part of the name.
9014 v2df __builtin_ia32_addsubpd (v2df, v2df)
9015 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
9016 v2df __builtin_ia32_haddpd (v2df, v2df)
9017 v4sf __builtin_ia32_haddps (v4sf, v4sf)
9018 v2df __builtin_ia32_hsubpd (v2df, v2df)
9019 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
9020 v16qi __builtin_ia32_lddqu (char const *)
9021 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
9022 v2df __builtin_ia32_movddup (v2df)
9023 v4sf __builtin_ia32_movshdup (v4sf)
9024 v4sf __builtin_ia32_movsldup (v4sf)
9025 void __builtin_ia32_mwait (unsigned int, unsigned int)
9028 The following built-in functions are available when @option{-msse3} is used.
9031 @item v2df __builtin_ia32_loadddup (double const *)
9032 Generates the @code{movddup} machine instruction as a load from memory.
9035 The following built-in functions are available when @option{-mssse3} is used.
9036 All of them generate the machine instruction that is part of the name
9040 v2si __builtin_ia32_phaddd (v2si, v2si)
9041 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
9042 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
9043 v2si __builtin_ia32_phsubd (v2si, v2si)
9044 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
9045 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
9046 v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi)
9047 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
9048 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
9049 v8qi __builtin_ia32_psignb (v8qi, v8qi)
9050 v2si __builtin_ia32_psignd (v2si, v2si)
9051 v4hi __builtin_ia32_psignw (v4hi, v4hi)
9052 v1di __builtin_ia32_palignr (v1di, v1di, int)
9053 v8qi __builtin_ia32_pabsb (v8qi)
9054 v2si __builtin_ia32_pabsd (v2si)
9055 v4hi __builtin_ia32_pabsw (v4hi)
9058 The following built-in functions are available when @option{-mssse3} is used.
9059 All of them generate the machine instruction that is part of the name
9063 v4si __builtin_ia32_phaddd128 (v4si, v4si)
9064 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
9065 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
9066 v4si __builtin_ia32_phsubd128 (v4si, v4si)
9067 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
9068 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
9069 v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
9070 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
9071 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
9072 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
9073 v4si __builtin_ia32_psignd128 (v4si, v4si)
9074 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
9075 v2di __builtin_ia32_palignr128 (v2di, v2di, int)
9076 v16qi __builtin_ia32_pabsb128 (v16qi)
9077 v4si __builtin_ia32_pabsd128 (v4si)
9078 v8hi __builtin_ia32_pabsw128 (v8hi)
9081 The following built-in functions are available when @option{-msse4.1} is
9082 used. All of them generate the machine instruction that is part of the
9086 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
9087 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
9088 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
9089 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
9090 v2df __builtin_ia32_dppd (v2df, v2df, const int)
9091 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
9092 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
9093 v2di __builtin_ia32_movntdqa (v2di *);
9094 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
9095 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
9096 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
9097 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
9098 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
9099 v8hi __builtin_ia32_phminposuw128 (v8hi)
9100 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
9101 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
9102 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
9103 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
9104 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
9105 v4si __builtin_ia32_pminsd128 (v4si, v4si)
9106 v4si __builtin_ia32_pminud128 (v4si, v4si)
9107 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
9108 v4si __builtin_ia32_pmovsxbd128 (v16qi)
9109 v2di __builtin_ia32_pmovsxbq128 (v16qi)
9110 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
9111 v2di __builtin_ia32_pmovsxdq128 (v4si)
9112 v4si __builtin_ia32_pmovsxwd128 (v8hi)
9113 v2di __builtin_ia32_pmovsxwq128 (v8hi)
9114 v4si __builtin_ia32_pmovzxbd128 (v16qi)
9115 v2di __builtin_ia32_pmovzxbq128 (v16qi)
9116 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
9117 v2di __builtin_ia32_pmovzxdq128 (v4si)
9118 v4si __builtin_ia32_pmovzxwd128 (v8hi)
9119 v2di __builtin_ia32_pmovzxwq128 (v8hi)
9120 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
9121 v4si __builtin_ia32_pmulld128 (v4si, v4si)
9122 int __builtin_ia32_ptestc128 (v2di, v2di)
9123 int __builtin_ia32_ptestnzc128 (v2di, v2di)
9124 int __builtin_ia32_ptestz128 (v2di, v2di)
9125 v2df __builtin_ia32_roundpd (v2df, const int)
9126 v4sf __builtin_ia32_roundps (v4sf, const int)
9127 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
9128 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
9131 The following built-in functions are available when @option{-msse4.1} is
9135 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
9136 Generates the @code{insertps} machine instruction.
9137 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
9138 Generates the @code{pextrb} machine instruction.
9139 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
9140 Generates the @code{pinsrb} machine instruction.
9141 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
9142 Generates the @code{pinsrd} machine instruction.
9143 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
9144 Generates the @code{pinsrq} machine instruction in 64bit mode.
9147 The following built-in functions are changed to generate new SSE4.1
9148 instructions when @option{-msse4.1} is used.
9151 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
9152 Generates the @code{extractps} machine instruction.
9153 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
9154 Generates the @code{pextrd} machine instruction.
9155 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
9156 Generates the @code{pextrq} machine instruction in 64bit mode.
9159 The following built-in functions are available when @option{-msse4.2} is
9160 used. All of them generate the machine instruction that is part of the
9164 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
9165 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
9166 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
9167 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
9168 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
9169 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
9170 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
9171 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
9172 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
9173 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
9174 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
9175 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
9176 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
9177 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
9178 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
9181 The following built-in functions are available when @option{-msse4.2} is
9185 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
9186 Generates the @code{crc32b} machine instruction.
9187 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
9188 Generates the @code{crc32w} machine instruction.
9189 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
9190 Generates the @code{crc32l} machine instruction.
9191 @item unsigned long long __builtin_ia32_crc32di (unsigned long long, unsigned long long)
9192 Generates the @code{crc32q} machine instruction.
9195 The following built-in functions are changed to generate new SSE4.2
9196 instructions when @option{-msse4.2} is used.
9199 @item int __builtin_popcount (unsigned int)
9200 Generates the @code{popcntl} machine instruction.
9201 @item int __builtin_popcountl (unsigned long)
9202 Generates the @code{popcntl} or @code{popcntq} machine instruction,
9203 depending on the size of @code{unsigned long}.
9204 @item int __builtin_popcountll (unsigned long long)
9205 Generates the @code{popcntq} machine instruction.
9208 The following built-in functions are available when @option{-mavx} is
9209 used. All of them generate the machine instruction that is part of the
9213 v4df __builtin_ia32_addpd256 (v4df,v4df)
9214 v8sf __builtin_ia32_addps256 (v8sf,v8sf)
9215 v4df __builtin_ia32_addsubpd256 (v4df,v4df)
9216 v8sf __builtin_ia32_addsubps256 (v8sf,v8sf)
9217 v4df __builtin_ia32_andnpd256 (v4df,v4df)
9218 v8sf __builtin_ia32_andnps256 (v8sf,v8sf)
9219 v4df __builtin_ia32_andpd256 (v4df,v4df)
9220 v8sf __builtin_ia32_andps256 (v8sf,v8sf)
9221 v4df __builtin_ia32_blendpd256 (v4df,v4df,int)
9222 v8sf __builtin_ia32_blendps256 (v8sf,v8sf,int)
9223 v4df __builtin_ia32_blendvpd256 (v4df,v4df,v4df)
9224 v8sf __builtin_ia32_blendvps256 (v8sf,v8sf,v8sf)
9225 v2df __builtin_ia32_cmppd (v2df,v2df,int)
9226 v4df __builtin_ia32_cmppd256 (v4df,v4df,int)
9227 v4sf __builtin_ia32_cmpps (v4sf,v4sf,int)
9228 v8sf __builtin_ia32_cmpps256 (v8sf,v8sf,int)
9229 v2df __builtin_ia32_cmpsd (v2df,v2df,int)
9230 v4sf __builtin_ia32_cmpss (v4sf,v4sf,int)
9231 v4df __builtin_ia32_cvtdq2pd256 (v4si)
9232 v8sf __builtin_ia32_cvtdq2ps256 (v8si)
9233 v4si __builtin_ia32_cvtpd2dq256 (v4df)
9234 v4sf __builtin_ia32_cvtpd2ps256 (v4df)
9235 v8si __builtin_ia32_cvtps2dq256 (v8sf)
9236 v4df __builtin_ia32_cvtps2pd256 (v4sf)
9237 v4si __builtin_ia32_cvttpd2dq256 (v4df)
9238 v8si __builtin_ia32_cvttps2dq256 (v8sf)
9239 v4df __builtin_ia32_divpd256 (v4df,v4df)
9240 v8sf __builtin_ia32_divps256 (v8sf,v8sf)
9241 v8sf __builtin_ia32_dpps256 (v8sf,v8sf,int)
9242 v4df __builtin_ia32_haddpd256 (v4df,v4df)
9243 v8sf __builtin_ia32_haddps256 (v8sf,v8sf)
9244 v4df __builtin_ia32_hsubpd256 (v4df,v4df)
9245 v8sf __builtin_ia32_hsubps256 (v8sf,v8sf)
9246 v32qi __builtin_ia32_lddqu256 (pcchar)
9247 v32qi __builtin_ia32_loaddqu256 (pcchar)
9248 v4df __builtin_ia32_loadupd256 (pcdouble)
9249 v8sf __builtin_ia32_loadups256 (pcfloat)
9250 v2df __builtin_ia32_maskloadpd (pcv2df,v2df)
9251 v4df __builtin_ia32_maskloadpd256 (pcv4df,v4df)
9252 v4sf __builtin_ia32_maskloadps (pcv4sf,v4sf)
9253 v8sf __builtin_ia32_maskloadps256 (pcv8sf,v8sf)
9254 void __builtin_ia32_maskstorepd (pv2df,v2df,v2df)
9255 void __builtin_ia32_maskstorepd256 (pv4df,v4df,v4df)
9256 void __builtin_ia32_maskstoreps (pv4sf,v4sf,v4sf)
9257 void __builtin_ia32_maskstoreps256 (pv8sf,v8sf,v8sf)
9258 v4df __builtin_ia32_maxpd256 (v4df,v4df)
9259 v8sf __builtin_ia32_maxps256 (v8sf,v8sf)
9260 v4df __builtin_ia32_minpd256 (v4df,v4df)
9261 v8sf __builtin_ia32_minps256 (v8sf,v8sf)
9262 v4df __builtin_ia32_movddup256 (v4df)
9263 int __builtin_ia32_movmskpd256 (v4df)
9264 int __builtin_ia32_movmskps256 (v8sf)
9265 v8sf __builtin_ia32_movshdup256 (v8sf)
9266 v8sf __builtin_ia32_movsldup256 (v8sf)
9267 v4df __builtin_ia32_mulpd256 (v4df,v4df)
9268 v8sf __builtin_ia32_mulps256 (v8sf,v8sf)
9269 v4df __builtin_ia32_orpd256 (v4df,v4df)
9270 v8sf __builtin_ia32_orps256 (v8sf,v8sf)
9271 v2df __builtin_ia32_pd_pd256 (v4df)
9272 v4df __builtin_ia32_pd256_pd (v2df)
9273 v4sf __builtin_ia32_ps_ps256 (v8sf)
9274 v8sf __builtin_ia32_ps256_ps (v4sf)
9275 int __builtin_ia32_ptestc256 (v4di,v4di,ptest)
9276 int __builtin_ia32_ptestnzc256 (v4di,v4di,ptest)
9277 int __builtin_ia32_ptestz256 (v4di,v4di,ptest)
9278 v8sf __builtin_ia32_rcpps256 (v8sf)
9279 v4df __builtin_ia32_roundpd256 (v4df,int)
9280 v8sf __builtin_ia32_roundps256 (v8sf,int)
9281 v8sf __builtin_ia32_rsqrtps_nr256 (v8sf)
9282 v8sf __builtin_ia32_rsqrtps256 (v8sf)
9283 v4df __builtin_ia32_shufpd256 (v4df,v4df,int)
9284 v8sf __builtin_ia32_shufps256 (v8sf,v8sf,int)
9285 v4si __builtin_ia32_si_si256 (v8si)
9286 v8si __builtin_ia32_si256_si (v4si)
9287 v4df __builtin_ia32_sqrtpd256 (v4df)
9288 v8sf __builtin_ia32_sqrtps_nr256 (v8sf)
9289 v8sf __builtin_ia32_sqrtps256 (v8sf)
9290 void __builtin_ia32_storedqu256 (pchar,v32qi)
9291 void __builtin_ia32_storeupd256 (pdouble,v4df)
9292 void __builtin_ia32_storeups256 (pfloat,v8sf)
9293 v4df __builtin_ia32_subpd256 (v4df,v4df)
9294 v8sf __builtin_ia32_subps256 (v8sf,v8sf)
9295 v4df __builtin_ia32_unpckhpd256 (v4df,v4df)
9296 v8sf __builtin_ia32_unpckhps256 (v8sf,v8sf)
9297 v4df __builtin_ia32_unpcklpd256 (v4df,v4df)
9298 v8sf __builtin_ia32_unpcklps256 (v8sf,v8sf)
9299 v4df __builtin_ia32_vbroadcastf128_pd256 (pcv2df)
9300 v8sf __builtin_ia32_vbroadcastf128_ps256 (pcv4sf)
9301 v4df __builtin_ia32_vbroadcastsd256 (pcdouble)
9302 v4sf __builtin_ia32_vbroadcastss (pcfloat)
9303 v8sf __builtin_ia32_vbroadcastss256 (pcfloat)
9304 v2df __builtin_ia32_vextractf128_pd256 (v4df,int)
9305 v4sf __builtin_ia32_vextractf128_ps256 (v8sf,int)
9306 v4si __builtin_ia32_vextractf128_si256 (v8si,int)
9307 v4df __builtin_ia32_vinsertf128_pd256 (v4df,v2df,int)
9308 v8sf __builtin_ia32_vinsertf128_ps256 (v8sf,v4sf,int)
9309 v8si __builtin_ia32_vinsertf128_si256 (v8si,v4si,int)
9310 v4df __builtin_ia32_vperm2f128_pd256 (v4df,v4df,int)
9311 v8sf __builtin_ia32_vperm2f128_ps256 (v8sf,v8sf,int)
9312 v8si __builtin_ia32_vperm2f128_si256 (v8si,v8si,int)
9313 v2df __builtin_ia32_vpermil2pd (v2df,v2df,v2di,int)
9314 v4df __builtin_ia32_vpermil2pd256 (v4df,v4df,v4di,int)
9315 v4sf __builtin_ia32_vpermil2ps (v4sf,v4sf,v4si,int)
9316 v8sf __builtin_ia32_vpermil2ps256 (v8sf,v8sf,v8si,int)
9317 v2df __builtin_ia32_vpermilpd (v2df,int)
9318 v4df __builtin_ia32_vpermilpd256 (v4df,int)
9319 v4sf __builtin_ia32_vpermilps (v4sf,int)
9320 v8sf __builtin_ia32_vpermilps256 (v8sf,int)
9321 v2df __builtin_ia32_vpermilvarpd (v2df,v2di)
9322 v4df __builtin_ia32_vpermilvarpd256 (v4df,v4di)
9323 v4sf __builtin_ia32_vpermilvarps (v4sf,v4si)
9324 v8sf __builtin_ia32_vpermilvarps256 (v8sf,v8si)
9325 int __builtin_ia32_vtestcpd (v2df,v2df,ptest)
9326 int __builtin_ia32_vtestcpd256 (v4df,v4df,ptest)
9327 int __builtin_ia32_vtestcps (v4sf,v4sf,ptest)
9328 int __builtin_ia32_vtestcps256 (v8sf,v8sf,ptest)
9329 int __builtin_ia32_vtestnzcpd (v2df,v2df,ptest)
9330 int __builtin_ia32_vtestnzcpd256 (v4df,v4df,ptest)
9331 int __builtin_ia32_vtestnzcps (v4sf,v4sf,ptest)
9332 int __builtin_ia32_vtestnzcps256 (v8sf,v8sf,ptest)
9333 int __builtin_ia32_vtestzpd (v2df,v2df,ptest)
9334 int __builtin_ia32_vtestzpd256 (v4df,v4df,ptest)
9335 int __builtin_ia32_vtestzps (v4sf,v4sf,ptest)
9336 int __builtin_ia32_vtestzps256 (v8sf,v8sf,ptest)
9337 void __builtin_ia32_vzeroall (void)
9338 void __builtin_ia32_vzeroupper (void)
9339 v4df __builtin_ia32_xorpd256 (v4df,v4df)
9340 v8sf __builtin_ia32_xorps256 (v8sf,v8sf)
9343 The following built-in functions are available when @option{-maes} is
9344 used. All of them generate the machine instruction that is part of the
9348 v2di __builtin_ia32_aesenc128 (v2di, v2di)
9349 v2di __builtin_ia32_aesenclast128 (v2di, v2di)
9350 v2di __builtin_ia32_aesdec128 (v2di, v2di)
9351 v2di __builtin_ia32_aesdeclast128 (v2di, v2di)
9352 v2di __builtin_ia32_aeskeygenassist128 (v2di, const int)
9353 v2di __builtin_ia32_aesimc128 (v2di)
9356 The following built-in function is available when @option{-mpclmul} is
9360 @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int)
9361 Generates the @code{pclmulqdq} machine instruction.
9364 The following built-in function is available when @option{-mfsgsbase} is
9365 used. All of them generate the machine instruction that is part of the
9369 unsigned int __builtin_ia32_rdfsbase32 (void)
9370 unsigned long long __builtin_ia32_rdfsbase64 (void)
9371 unsigned int __builtin_ia32_rdgsbase32 (void)
9372 unsigned long long __builtin_ia32_rdgsbase64 (void)
9373 void _writefsbase_u32 (unsigned int)
9374 void _writefsbase_u64 (unsigned long long)
9375 void _writegsbase_u32 (unsigned int)
9376 void _writegsbase_u64 (unsigned long long)
9379 The following built-in function is available when @option{-mrdrnd} is
9380 used. All of them generate the machine instruction that is part of the
9384 unsigned short __builtin_ia32_rdrand16 (void)
9385 unsigned int __builtin_ia32_rdrand32 (void)
9386 unsigned long long __builtin_ia32_rdrand64 (void)
9389 The following built-in functions are available when @option{-msse4a} is used.
9390 All of them generate the machine instruction that is part of the name.
9393 void __builtin_ia32_movntsd (double *, v2df)
9394 void __builtin_ia32_movntss (float *, v4sf)
9395 v2di __builtin_ia32_extrq (v2di, v16qi)
9396 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
9397 v2di __builtin_ia32_insertq (v2di, v2di)
9398 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
9401 The following built-in functions are available when @option{-mxop} is used.
9403 v2df __builtin_ia32_vfrczpd (v2df)
9404 v4sf __builtin_ia32_vfrczps (v4sf)
9405 v2df __builtin_ia32_vfrczsd (v2df, v2df)
9406 v4sf __builtin_ia32_vfrczss (v4sf, v4sf)
9407 v4df __builtin_ia32_vfrczpd256 (v4df)
9408 v8sf __builtin_ia32_vfrczps256 (v8sf)
9409 v2di __builtin_ia32_vpcmov (v2di, v2di, v2di)
9410 v2di __builtin_ia32_vpcmov_v2di (v2di, v2di, v2di)
9411 v4si __builtin_ia32_vpcmov_v4si (v4si, v4si, v4si)
9412 v8hi __builtin_ia32_vpcmov_v8hi (v8hi, v8hi, v8hi)
9413 v16qi __builtin_ia32_vpcmov_v16qi (v16qi, v16qi, v16qi)
9414 v2df __builtin_ia32_vpcmov_v2df (v2df, v2df, v2df)
9415 v4sf __builtin_ia32_vpcmov_v4sf (v4sf, v4sf, v4sf)
9416 v4di __builtin_ia32_vpcmov_v4di256 (v4di, v4di, v4di)
9417 v8si __builtin_ia32_vpcmov_v8si256 (v8si, v8si, v8si)
9418 v16hi __builtin_ia32_vpcmov_v16hi256 (v16hi, v16hi, v16hi)
9419 v32qi __builtin_ia32_vpcmov_v32qi256 (v32qi, v32qi, v32qi)
9420 v4df __builtin_ia32_vpcmov_v4df256 (v4df, v4df, v4df)
9421 v8sf __builtin_ia32_vpcmov_v8sf256 (v8sf, v8sf, v8sf)
9422 v16qi __builtin_ia32_vpcomeqb (v16qi, v16qi)
9423 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
9424 v4si __builtin_ia32_vpcomeqd (v4si, v4si)
9425 v2di __builtin_ia32_vpcomeqq (v2di, v2di)
9426 v16qi __builtin_ia32_vpcomequb (v16qi, v16qi)
9427 v4si __builtin_ia32_vpcomequd (v4si, v4si)
9428 v2di __builtin_ia32_vpcomequq (v2di, v2di)
9429 v8hi __builtin_ia32_vpcomequw (v8hi, v8hi)
9430 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
9431 v16qi __builtin_ia32_vpcomfalseb (v16qi, v16qi)
9432 v4si __builtin_ia32_vpcomfalsed (v4si, v4si)
9433 v2di __builtin_ia32_vpcomfalseq (v2di, v2di)
9434 v16qi __builtin_ia32_vpcomfalseub (v16qi, v16qi)
9435 v4si __builtin_ia32_vpcomfalseud (v4si, v4si)
9436 v2di __builtin_ia32_vpcomfalseuq (v2di, v2di)
9437 v8hi __builtin_ia32_vpcomfalseuw (v8hi, v8hi)
9438 v8hi __builtin_ia32_vpcomfalsew (v8hi, v8hi)
9439 v16qi __builtin_ia32_vpcomgeb (v16qi, v16qi)
9440 v4si __builtin_ia32_vpcomged (v4si, v4si)
9441 v2di __builtin_ia32_vpcomgeq (v2di, v2di)
9442 v16qi __builtin_ia32_vpcomgeub (v16qi, v16qi)
9443 v4si __builtin_ia32_vpcomgeud (v4si, v4si)
9444 v2di __builtin_ia32_vpcomgeuq (v2di, v2di)
9445 v8hi __builtin_ia32_vpcomgeuw (v8hi, v8hi)
9446 v8hi __builtin_ia32_vpcomgew (v8hi, v8hi)
9447 v16qi __builtin_ia32_vpcomgtb (v16qi, v16qi)
9448 v4si __builtin_ia32_vpcomgtd (v4si, v4si)
9449 v2di __builtin_ia32_vpcomgtq (v2di, v2di)
9450 v16qi __builtin_ia32_vpcomgtub (v16qi, v16qi)
9451 v4si __builtin_ia32_vpcomgtud (v4si, v4si)
9452 v2di __builtin_ia32_vpcomgtuq (v2di, v2di)
9453 v8hi __builtin_ia32_vpcomgtuw (v8hi, v8hi)
9454 v8hi __builtin_ia32_vpcomgtw (v8hi, v8hi)
9455 v16qi __builtin_ia32_vpcomleb (v16qi, v16qi)
9456 v4si __builtin_ia32_vpcomled (v4si, v4si)
9457 v2di __builtin_ia32_vpcomleq (v2di, v2di)
9458 v16qi __builtin_ia32_vpcomleub (v16qi, v16qi)
9459 v4si __builtin_ia32_vpcomleud (v4si, v4si)
9460 v2di __builtin_ia32_vpcomleuq (v2di, v2di)
9461 v8hi __builtin_ia32_vpcomleuw (v8hi, v8hi)
9462 v8hi __builtin_ia32_vpcomlew (v8hi, v8hi)
9463 v16qi __builtin_ia32_vpcomltb (v16qi, v16qi)
9464 v4si __builtin_ia32_vpcomltd (v4si, v4si)
9465 v2di __builtin_ia32_vpcomltq (v2di, v2di)
9466 v16qi __builtin_ia32_vpcomltub (v16qi, v16qi)
9467 v4si __builtin_ia32_vpcomltud (v4si, v4si)
9468 v2di __builtin_ia32_vpcomltuq (v2di, v2di)
9469 v8hi __builtin_ia32_vpcomltuw (v8hi, v8hi)
9470 v8hi __builtin_ia32_vpcomltw (v8hi, v8hi)
9471 v16qi __builtin_ia32_vpcomneb (v16qi, v16qi)
9472 v4si __builtin_ia32_vpcomned (v4si, v4si)
9473 v2di __builtin_ia32_vpcomneq (v2di, v2di)
9474 v16qi __builtin_ia32_vpcomneub (v16qi, v16qi)
9475 v4si __builtin_ia32_vpcomneud (v4si, v4si)
9476 v2di __builtin_ia32_vpcomneuq (v2di, v2di)
9477 v8hi __builtin_ia32_vpcomneuw (v8hi, v8hi)
9478 v8hi __builtin_ia32_vpcomnew (v8hi, v8hi)
9479 v16qi __builtin_ia32_vpcomtrueb (v16qi, v16qi)
9480 v4si __builtin_ia32_vpcomtrued (v4si, v4si)
9481 v2di __builtin_ia32_vpcomtrueq (v2di, v2di)
9482 v16qi __builtin_ia32_vpcomtrueub (v16qi, v16qi)
9483 v4si __builtin_ia32_vpcomtrueud (v4si, v4si)
9484 v2di __builtin_ia32_vpcomtrueuq (v2di, v2di)
9485 v8hi __builtin_ia32_vpcomtrueuw (v8hi, v8hi)
9486 v8hi __builtin_ia32_vpcomtruew (v8hi, v8hi)
9487 v4si __builtin_ia32_vphaddbd (v16qi)
9488 v2di __builtin_ia32_vphaddbq (v16qi)
9489 v8hi __builtin_ia32_vphaddbw (v16qi)
9490 v2di __builtin_ia32_vphadddq (v4si)
9491 v4si __builtin_ia32_vphaddubd (v16qi)
9492 v2di __builtin_ia32_vphaddubq (v16qi)
9493 v8hi __builtin_ia32_vphaddubw (v16qi)
9494 v2di __builtin_ia32_vphaddudq (v4si)
9495 v4si __builtin_ia32_vphadduwd (v8hi)
9496 v2di __builtin_ia32_vphadduwq (v8hi)
9497 v4si __builtin_ia32_vphaddwd (v8hi)
9498 v2di __builtin_ia32_vphaddwq (v8hi)
9499 v8hi __builtin_ia32_vphsubbw (v16qi)
9500 v2di __builtin_ia32_vphsubdq (v4si)
9501 v4si __builtin_ia32_vphsubwd (v8hi)
9502 v4si __builtin_ia32_vpmacsdd (v4si, v4si, v4si)
9503 v2di __builtin_ia32_vpmacsdqh (v4si, v4si, v2di)
9504 v2di __builtin_ia32_vpmacsdql (v4si, v4si, v2di)
9505 v4si __builtin_ia32_vpmacssdd (v4si, v4si, v4si)
9506 v2di __builtin_ia32_vpmacssdqh (v4si, v4si, v2di)
9507 v2di __builtin_ia32_vpmacssdql (v4si, v4si, v2di)
9508 v4si __builtin_ia32_vpmacsswd (v8hi, v8hi, v4si)
9509 v8hi __builtin_ia32_vpmacssww (v8hi, v8hi, v8hi)
9510 v4si __builtin_ia32_vpmacswd (v8hi, v8hi, v4si)
9511 v8hi __builtin_ia32_vpmacsww (v8hi, v8hi, v8hi)
9512 v4si __builtin_ia32_vpmadcsswd (v8hi, v8hi, v4si)
9513 v4si __builtin_ia32_vpmadcswd (v8hi, v8hi, v4si)
9514 v16qi __builtin_ia32_vpperm (v16qi, v16qi, v16qi)
9515 v16qi __builtin_ia32_vprotb (v16qi, v16qi)
9516 v4si __builtin_ia32_vprotd (v4si, v4si)
9517 v2di __builtin_ia32_vprotq (v2di, v2di)
9518 v8hi __builtin_ia32_vprotw (v8hi, v8hi)
9519 v16qi __builtin_ia32_vpshab (v16qi, v16qi)
9520 v4si __builtin_ia32_vpshad (v4si, v4si)
9521 v2di __builtin_ia32_vpshaq (v2di, v2di)
9522 v8hi __builtin_ia32_vpshaw (v8hi, v8hi)
9523 v16qi __builtin_ia32_vpshlb (v16qi, v16qi)
9524 v4si __builtin_ia32_vpshld (v4si, v4si)
9525 v2di __builtin_ia32_vpshlq (v2di, v2di)
9526 v8hi __builtin_ia32_vpshlw (v8hi, v8hi)
9529 The following built-in functions are available when @option{-mfma4} is used.
9530 All of them generate the machine instruction that is part of the name
9534 v2df __builtin_ia32_fmaddpd (v2df, v2df, v2df)
9535 v4sf __builtin_ia32_fmaddps (v4sf, v4sf, v4sf)
9536 v2df __builtin_ia32_fmaddsd (v2df, v2df, v2df)
9537 v4sf __builtin_ia32_fmaddss (v4sf, v4sf, v4sf)
9538 v2df __builtin_ia32_fmsubpd (v2df, v2df, v2df)
9539 v4sf __builtin_ia32_fmsubps (v4sf, v4sf, v4sf)
9540 v2df __builtin_ia32_fmsubsd (v2df, v2df, v2df)
9541 v4sf __builtin_ia32_fmsubss (v4sf, v4sf, v4sf)
9542 v2df __builtin_ia32_fnmaddpd (v2df, v2df, v2df)
9543 v4sf __builtin_ia32_fnmaddps (v4sf, v4sf, v4sf)
9544 v2df __builtin_ia32_fnmaddsd (v2df, v2df, v2df)
9545 v4sf __builtin_ia32_fnmaddss (v4sf, v4sf, v4sf)
9546 v2df __builtin_ia32_fnmsubpd (v2df, v2df, v2df)
9547 v4sf __builtin_ia32_fnmsubps (v4sf, v4sf, v4sf)
9548 v2df __builtin_ia32_fnmsubsd (v2df, v2df, v2df)
9549 v4sf __builtin_ia32_fnmsubss (v4sf, v4sf, v4sf)
9550 v2df __builtin_ia32_fmaddsubpd (v2df, v2df, v2df)
9551 v4sf __builtin_ia32_fmaddsubps (v4sf, v4sf, v4sf)
9552 v2df __builtin_ia32_fmsubaddpd (v2df, v2df, v2df)
9553 v4sf __builtin_ia32_fmsubaddps (v4sf, v4sf, v4sf)
9554 v4df __builtin_ia32_fmaddpd256 (v4df, v4df, v4df)
9555 v8sf __builtin_ia32_fmaddps256 (v8sf, v8sf, v8sf)
9556 v4df __builtin_ia32_fmsubpd256 (v4df, v4df, v4df)
9557 v8sf __builtin_ia32_fmsubps256 (v8sf, v8sf, v8sf)
9558 v4df __builtin_ia32_fnmaddpd256 (v4df, v4df, v4df)
9559 v8sf __builtin_ia32_fnmaddps256 (v8sf, v8sf, v8sf)
9560 v4df __builtin_ia32_fnmsubpd256 (v4df, v4df, v4df)
9561 v8sf __builtin_ia32_fnmsubps256 (v8sf, v8sf, v8sf)
9562 v4df __builtin_ia32_fmaddsubpd256 (v4df, v4df, v4df)
9563 v8sf __builtin_ia32_fmaddsubps256 (v8sf, v8sf, v8sf)
9564 v4df __builtin_ia32_fmsubaddpd256 (v4df, v4df, v4df)
9565 v8sf __builtin_ia32_fmsubaddps256 (v8sf, v8sf, v8sf)
9569 The following built-in functions are available when @option{-mlwp} is used.
9572 void __builtin_ia32_llwpcb16 (void *);
9573 void __builtin_ia32_llwpcb32 (void *);
9574 void __builtin_ia32_llwpcb64 (void *);
9575 void * __builtin_ia32_llwpcb16 (void);
9576 void * __builtin_ia32_llwpcb32 (void);
9577 void * __builtin_ia32_llwpcb64 (void);
9578 void __builtin_ia32_lwpval16 (unsigned short, unsigned int, unsigned short)
9579 void __builtin_ia32_lwpval32 (unsigned int, unsigned int, unsigned int)
9580 void __builtin_ia32_lwpval64 (unsigned __int64, unsigned int, unsigned int)
9581 unsigned char __builtin_ia32_lwpins16 (unsigned short, unsigned int, unsigned short)
9582 unsigned char __builtin_ia32_lwpins32 (unsigned int, unsigned int, unsigned int)
9583 unsigned char __builtin_ia32_lwpins64 (unsigned __int64, unsigned int, unsigned int)
9586 The following built-in functions are available when @option{-mbmi} is used.
9587 All of them generate the machine instruction that is part of the name.
9589 unsigned int __builtin_ia32_bextr_u32(unsigned int, unsigned int);
9590 unsigned long long __builtin_ia32_bextr_u64 (unsigned long long, unsigned long long);
9591 unsigned short __builtin_ia32_lzcnt_16(unsigned short);
9592 unsigned int __builtin_ia32_lzcnt_u32(unsigned int);
9593 unsigned long long __builtin_ia32_lzcnt_u64 (unsigned long long);
9596 The following built-in functions are available when @option{-mtbm} is used.
9597 Both of them generate the immediate form of the bextr machine instruction.
9599 unsigned int __builtin_ia32_bextri_u32 (unsigned int, const unsigned int);
9600 unsigned long long __builtin_ia32_bextri_u64 (unsigned long long, const unsigned long long);
9604 The following built-in functions are available when @option{-m3dnow} is used.
9605 All of them generate the machine instruction that is part of the name.
9608 void __builtin_ia32_femms (void)
9609 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
9610 v2si __builtin_ia32_pf2id (v2sf)
9611 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
9612 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
9613 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
9614 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
9615 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
9616 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
9617 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
9618 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
9619 v2sf __builtin_ia32_pfrcp (v2sf)
9620 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
9621 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
9622 v2sf __builtin_ia32_pfrsqrt (v2sf)
9623 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
9624 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
9625 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
9626 v2sf __builtin_ia32_pi2fd (v2si)
9627 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
9630 The following built-in functions are available when both @option{-m3dnow}
9631 and @option{-march=athlon} are used. All of them generate the machine
9632 instruction that is part of the name.
9635 v2si __builtin_ia32_pf2iw (v2sf)
9636 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
9637 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
9638 v2sf __builtin_ia32_pi2fw (v2si)
9639 v2sf __builtin_ia32_pswapdsf (v2sf)
9640 v2si __builtin_ia32_pswapdsi (v2si)
9643 @node MIPS DSP Built-in Functions
9644 @subsection MIPS DSP Built-in Functions
9646 The MIPS DSP Application-Specific Extension (ASE) includes new
9647 instructions that are designed to improve the performance of DSP and
9648 media applications. It provides instructions that operate on packed
9649 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
9651 GCC supports MIPS DSP operations using both the generic
9652 vector extensions (@pxref{Vector Extensions}) and a collection of
9653 MIPS-specific built-in functions. Both kinds of support are
9654 enabled by the @option{-mdsp} command-line option.
9656 Revision 2 of the ASE was introduced in the second half of 2006.
9657 This revision adds extra instructions to the original ASE, but is
9658 otherwise backwards-compatible with it. You can select revision 2
9659 using the command-line option @option{-mdspr2}; this option implies
9662 The SCOUNT and POS bits of the DSP control register are global. The
9663 WRDSP, EXTPDP, EXTPDPV and MTHLIP instructions modify the SCOUNT and
9664 POS bits. During optimization, the compiler will not delete these
9665 instructions and it will not delete calls to functions containing
9668 At present, GCC only provides support for operations on 32-bit
9669 vectors. The vector type associated with 8-bit integer data is
9670 usually called @code{v4i8}, the vector type associated with Q7
9671 is usually called @code{v4q7}, the vector type associated with 16-bit
9672 integer data is usually called @code{v2i16}, and the vector type
9673 associated with Q15 is usually called @code{v2q15}. They can be
9674 defined in C as follows:
9677 typedef signed char v4i8 __attribute__ ((vector_size(4)));
9678 typedef signed char v4q7 __attribute__ ((vector_size(4)));
9679 typedef short v2i16 __attribute__ ((vector_size(4)));
9680 typedef short v2q15 __attribute__ ((vector_size(4)));
9683 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
9684 initialized in the same way as aggregates. For example:
9687 v4i8 a = @{1, 2, 3, 4@};
9689 b = (v4i8) @{5, 6, 7, 8@};
9691 v2q15 c = @{0x0fcb, 0x3a75@};
9693 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
9696 @emph{Note:} The CPU's endianness determines the order in which values
9697 are packed. On little-endian targets, the first value is the least
9698 significant and the last value is the most significant. The opposite
9699 order applies to big-endian targets. For example, the code above will
9700 set the lowest byte of @code{a} to @code{1} on little-endian targets
9701 and @code{4} on big-endian targets.
9703 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
9704 representation. As shown in this example, the integer representation
9705 of a Q7 value can be obtained by multiplying the fractional value by
9706 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
9707 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
9710 The table below lists the @code{v4i8} and @code{v2q15} operations for which
9711 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
9712 and @code{c} and @code{d} are @code{v2q15} values.
9714 @multitable @columnfractions .50 .50
9715 @item C code @tab MIPS instruction
9716 @item @code{a + b} @tab @code{addu.qb}
9717 @item @code{c + d} @tab @code{addq.ph}
9718 @item @code{a - b} @tab @code{subu.qb}
9719 @item @code{c - d} @tab @code{subq.ph}
9722 The table below lists the @code{v2i16} operation for which
9723 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
9724 @code{v2i16} values.
9726 @multitable @columnfractions .50 .50
9727 @item C code @tab MIPS instruction
9728 @item @code{e * f} @tab @code{mul.ph}
9731 It is easier to describe the DSP built-in functions if we first define
9732 the following types:
9737 typedef unsigned int ui32;
9738 typedef long long a64;
9741 @code{q31} and @code{i32} are actually the same as @code{int}, but we
9742 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
9743 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
9744 @code{long long}, but we use @code{a64} to indicate values that will
9745 be placed in one of the four DSP accumulators (@code{$ac0},
9746 @code{$ac1}, @code{$ac2} or @code{$ac3}).
9748 Also, some built-in functions prefer or require immediate numbers as
9749 parameters, because the corresponding DSP instructions accept both immediate
9750 numbers and register operands, or accept immediate numbers only. The
9751 immediate parameters are listed as follows.
9760 imm_n32_31: -32 to 31.
9761 imm_n512_511: -512 to 511.
9764 The following built-in functions map directly to a particular MIPS DSP
9765 instruction. Please refer to the architecture specification
9766 for details on what each instruction does.
9769 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
9770 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
9771 q31 __builtin_mips_addq_s_w (q31, q31)
9772 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
9773 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
9774 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
9775 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
9776 q31 __builtin_mips_subq_s_w (q31, q31)
9777 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
9778 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
9779 i32 __builtin_mips_addsc (i32, i32)
9780 i32 __builtin_mips_addwc (i32, i32)
9781 i32 __builtin_mips_modsub (i32, i32)
9782 i32 __builtin_mips_raddu_w_qb (v4i8)
9783 v2q15 __builtin_mips_absq_s_ph (v2q15)
9784 q31 __builtin_mips_absq_s_w (q31)
9785 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
9786 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
9787 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
9788 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
9789 q31 __builtin_mips_preceq_w_phl (v2q15)
9790 q31 __builtin_mips_preceq_w_phr (v2q15)
9791 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
9792 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
9793 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
9794 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
9795 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
9796 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
9797 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
9798 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
9799 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
9800 v4i8 __builtin_mips_shll_qb (v4i8, i32)
9801 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
9802 v2q15 __builtin_mips_shll_ph (v2q15, i32)
9803 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
9804 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
9805 q31 __builtin_mips_shll_s_w (q31, imm0_31)
9806 q31 __builtin_mips_shll_s_w (q31, i32)
9807 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
9808 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
9809 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
9810 v2q15 __builtin_mips_shra_ph (v2q15, i32)
9811 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
9812 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
9813 q31 __builtin_mips_shra_r_w (q31, imm0_31)
9814 q31 __builtin_mips_shra_r_w (q31, i32)
9815 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
9816 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
9817 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
9818 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
9819 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
9820 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
9821 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
9822 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
9823 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
9824 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
9825 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
9826 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
9827 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
9828 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
9829 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
9830 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
9831 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
9832 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
9833 i32 __builtin_mips_bitrev (i32)
9834 i32 __builtin_mips_insv (i32, i32)
9835 v4i8 __builtin_mips_repl_qb (imm0_255)
9836 v4i8 __builtin_mips_repl_qb (i32)
9837 v2q15 __builtin_mips_repl_ph (imm_n512_511)
9838 v2q15 __builtin_mips_repl_ph (i32)
9839 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
9840 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
9841 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
9842 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
9843 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
9844 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
9845 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
9846 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
9847 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
9848 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
9849 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
9850 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
9851 i32 __builtin_mips_extr_w (a64, imm0_31)
9852 i32 __builtin_mips_extr_w (a64, i32)
9853 i32 __builtin_mips_extr_r_w (a64, imm0_31)
9854 i32 __builtin_mips_extr_s_h (a64, i32)
9855 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
9856 i32 __builtin_mips_extr_rs_w (a64, i32)
9857 i32 __builtin_mips_extr_s_h (a64, imm0_31)
9858 i32 __builtin_mips_extr_r_w (a64, i32)
9859 i32 __builtin_mips_extp (a64, imm0_31)
9860 i32 __builtin_mips_extp (a64, i32)
9861 i32 __builtin_mips_extpdp (a64, imm0_31)
9862 i32 __builtin_mips_extpdp (a64, i32)
9863 a64 __builtin_mips_shilo (a64, imm_n32_31)
9864 a64 __builtin_mips_shilo (a64, i32)
9865 a64 __builtin_mips_mthlip (a64, i32)
9866 void __builtin_mips_wrdsp (i32, imm0_63)
9867 i32 __builtin_mips_rddsp (imm0_63)
9868 i32 __builtin_mips_lbux (void *, i32)
9869 i32 __builtin_mips_lhx (void *, i32)
9870 i32 __builtin_mips_lwx (void *, i32)
9871 i32 __builtin_mips_bposge32 (void)
9872 a64 __builtin_mips_madd (a64, i32, i32);
9873 a64 __builtin_mips_maddu (a64, ui32, ui32);
9874 a64 __builtin_mips_msub (a64, i32, i32);
9875 a64 __builtin_mips_msubu (a64, ui32, ui32);
9876 a64 __builtin_mips_mult (i32, i32);
9877 a64 __builtin_mips_multu (ui32, ui32);
9880 The following built-in functions map directly to a particular MIPS DSP REV 2
9881 instruction. Please refer to the architecture specification
9882 for details on what each instruction does.
9885 v4q7 __builtin_mips_absq_s_qb (v4q7);
9886 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
9887 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
9888 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
9889 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
9890 i32 __builtin_mips_append (i32, i32, imm0_31);
9891 i32 __builtin_mips_balign (i32, i32, imm0_3);
9892 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
9893 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
9894 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
9895 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
9896 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
9897 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
9898 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
9899 q31 __builtin_mips_mulq_rs_w (q31, q31);
9900 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
9901 q31 __builtin_mips_mulq_s_w (q31, q31);
9902 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
9903 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
9904 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
9905 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
9906 i32 __builtin_mips_prepend (i32, i32, imm0_31);
9907 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
9908 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
9909 v4i8 __builtin_mips_shra_qb (v4i8, i32);
9910 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
9911 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
9912 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
9913 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
9914 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
9915 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
9916 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
9917 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
9918 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
9919 q31 __builtin_mips_addqh_w (q31, q31);
9920 q31 __builtin_mips_addqh_r_w (q31, q31);
9921 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
9922 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
9923 q31 __builtin_mips_subqh_w (q31, q31);
9924 q31 __builtin_mips_subqh_r_w (q31, q31);
9925 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
9926 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
9927 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
9928 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
9929 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
9930 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
9934 @node MIPS Paired-Single Support
9935 @subsection MIPS Paired-Single Support
9937 The MIPS64 architecture includes a number of instructions that
9938 operate on pairs of single-precision floating-point values.
9939 Each pair is packed into a 64-bit floating-point register,
9940 with one element being designated the ``upper half'' and
9941 the other being designated the ``lower half''.
9943 GCC supports paired-single operations using both the generic
9944 vector extensions (@pxref{Vector Extensions}) and a collection of
9945 MIPS-specific built-in functions. Both kinds of support are
9946 enabled by the @option{-mpaired-single} command-line option.
9948 The vector type associated with paired-single values is usually
9949 called @code{v2sf}. It can be defined in C as follows:
9952 typedef float v2sf __attribute__ ((vector_size (8)));
9955 @code{v2sf} values are initialized in the same way as aggregates.
9959 v2sf a = @{1.5, 9.1@};
9962 b = (v2sf) @{e, f@};
9965 @emph{Note:} The CPU's endianness determines which value is stored in
9966 the upper half of a register and which value is stored in the lower half.
9967 On little-endian targets, the first value is the lower one and the second
9968 value is the upper one. The opposite order applies to big-endian targets.
9969 For example, the code above will set the lower half of @code{a} to
9970 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
9972 @node MIPS Loongson Built-in Functions
9973 @subsection MIPS Loongson Built-in Functions
9975 GCC provides intrinsics to access the SIMD instructions provided by the
9976 ST Microelectronics Loongson-2E and -2F processors. These intrinsics,
9977 available after inclusion of the @code{loongson.h} header file,
9978 operate on the following 64-bit vector types:
9981 @item @code{uint8x8_t}, a vector of eight unsigned 8-bit integers;
9982 @item @code{uint16x4_t}, a vector of four unsigned 16-bit integers;
9983 @item @code{uint32x2_t}, a vector of two unsigned 32-bit integers;
9984 @item @code{int8x8_t}, a vector of eight signed 8-bit integers;
9985 @item @code{int16x4_t}, a vector of four signed 16-bit integers;
9986 @item @code{int32x2_t}, a vector of two signed 32-bit integers.
9989 The intrinsics provided are listed below; each is named after the
9990 machine instruction to which it corresponds, with suffixes added as
9991 appropriate to distinguish intrinsics that expand to the same machine
9992 instruction yet have different argument types. Refer to the architecture
9993 documentation for a description of the functionality of each
9997 int16x4_t packsswh (int32x2_t s, int32x2_t t);
9998 int8x8_t packsshb (int16x4_t s, int16x4_t t);
9999 uint8x8_t packushb (uint16x4_t s, uint16x4_t t);
10000 uint32x2_t paddw_u (uint32x2_t s, uint32x2_t t);
10001 uint16x4_t paddh_u (uint16x4_t s, uint16x4_t t);
10002 uint8x8_t paddb_u (uint8x8_t s, uint8x8_t t);
10003 int32x2_t paddw_s (int32x2_t s, int32x2_t t);
10004 int16x4_t paddh_s (int16x4_t s, int16x4_t t);
10005 int8x8_t paddb_s (int8x8_t s, int8x8_t t);
10006 uint64_t paddd_u (uint64_t s, uint64_t t);
10007 int64_t paddd_s (int64_t s, int64_t t);
10008 int16x4_t paddsh (int16x4_t s, int16x4_t t);
10009 int8x8_t paddsb (int8x8_t s, int8x8_t t);
10010 uint16x4_t paddush (uint16x4_t s, uint16x4_t t);
10011 uint8x8_t paddusb (uint8x8_t s, uint8x8_t t);
10012 uint64_t pandn_ud (uint64_t s, uint64_t t);
10013 uint32x2_t pandn_uw (uint32x2_t s, uint32x2_t t);
10014 uint16x4_t pandn_uh (uint16x4_t s, uint16x4_t t);
10015 uint8x8_t pandn_ub (uint8x8_t s, uint8x8_t t);
10016 int64_t pandn_sd (int64_t s, int64_t t);
10017 int32x2_t pandn_sw (int32x2_t s, int32x2_t t);
10018 int16x4_t pandn_sh (int16x4_t s, int16x4_t t);
10019 int8x8_t pandn_sb (int8x8_t s, int8x8_t t);
10020 uint16x4_t pavgh (uint16x4_t s, uint16x4_t t);
10021 uint8x8_t pavgb (uint8x8_t s, uint8x8_t t);
10022 uint32x2_t pcmpeqw_u (uint32x2_t s, uint32x2_t t);
10023 uint16x4_t pcmpeqh_u (uint16x4_t s, uint16x4_t t);
10024 uint8x8_t pcmpeqb_u (uint8x8_t s, uint8x8_t t);
10025 int32x2_t pcmpeqw_s (int32x2_t s, int32x2_t t);
10026 int16x4_t pcmpeqh_s (int16x4_t s, int16x4_t t);
10027 int8x8_t pcmpeqb_s (int8x8_t s, int8x8_t t);
10028 uint32x2_t pcmpgtw_u (uint32x2_t s, uint32x2_t t);
10029 uint16x4_t pcmpgth_u (uint16x4_t s, uint16x4_t t);
10030 uint8x8_t pcmpgtb_u (uint8x8_t s, uint8x8_t t);
10031 int32x2_t pcmpgtw_s (int32x2_t s, int32x2_t t);
10032 int16x4_t pcmpgth_s (int16x4_t s, int16x4_t t);
10033 int8x8_t pcmpgtb_s (int8x8_t s, int8x8_t t);
10034 uint16x4_t pextrh_u (uint16x4_t s, int field);
10035 int16x4_t pextrh_s (int16x4_t s, int field);
10036 uint16x4_t pinsrh_0_u (uint16x4_t s, uint16x4_t t);
10037 uint16x4_t pinsrh_1_u (uint16x4_t s, uint16x4_t t);
10038 uint16x4_t pinsrh_2_u (uint16x4_t s, uint16x4_t t);
10039 uint16x4_t pinsrh_3_u (uint16x4_t s, uint16x4_t t);
10040 int16x4_t pinsrh_0_s (int16x4_t s, int16x4_t t);
10041 int16x4_t pinsrh_1_s (int16x4_t s, int16x4_t t);
10042 int16x4_t pinsrh_2_s (int16x4_t s, int16x4_t t);
10043 int16x4_t pinsrh_3_s (int16x4_t s, int16x4_t t);
10044 int32x2_t pmaddhw (int16x4_t s, int16x4_t t);
10045 int16x4_t pmaxsh (int16x4_t s, int16x4_t t);
10046 uint8x8_t pmaxub (uint8x8_t s, uint8x8_t t);
10047 int16x4_t pminsh (int16x4_t s, int16x4_t t);
10048 uint8x8_t pminub (uint8x8_t s, uint8x8_t t);
10049 uint8x8_t pmovmskb_u (uint8x8_t s);
10050 int8x8_t pmovmskb_s (int8x8_t s);
10051 uint16x4_t pmulhuh (uint16x4_t s, uint16x4_t t);
10052 int16x4_t pmulhh (int16x4_t s, int16x4_t t);
10053 int16x4_t pmullh (int16x4_t s, int16x4_t t);
10054 int64_t pmuluw (uint32x2_t s, uint32x2_t t);
10055 uint8x8_t pasubub (uint8x8_t s, uint8x8_t t);
10056 uint16x4_t biadd (uint8x8_t s);
10057 uint16x4_t psadbh (uint8x8_t s, uint8x8_t t);
10058 uint16x4_t pshufh_u (uint16x4_t dest, uint16x4_t s, uint8_t order);
10059 int16x4_t pshufh_s (int16x4_t dest, int16x4_t s, uint8_t order);
10060 uint16x4_t psllh_u (uint16x4_t s, uint8_t amount);
10061 int16x4_t psllh_s (int16x4_t s, uint8_t amount);
10062 uint32x2_t psllw_u (uint32x2_t s, uint8_t amount);
10063 int32x2_t psllw_s (int32x2_t s, uint8_t amount);
10064 uint16x4_t psrlh_u (uint16x4_t s, uint8_t amount);
10065 int16x4_t psrlh_s (int16x4_t s, uint8_t amount);
10066 uint32x2_t psrlw_u (uint32x2_t s, uint8_t amount);
10067 int32x2_t psrlw_s (int32x2_t s, uint8_t amount);
10068 uint16x4_t psrah_u (uint16x4_t s, uint8_t amount);
10069 int16x4_t psrah_s (int16x4_t s, uint8_t amount);
10070 uint32x2_t psraw_u (uint32x2_t s, uint8_t amount);
10071 int32x2_t psraw_s (int32x2_t s, uint8_t amount);
10072 uint32x2_t psubw_u (uint32x2_t s, uint32x2_t t);
10073 uint16x4_t psubh_u (uint16x4_t s, uint16x4_t t);
10074 uint8x8_t psubb_u (uint8x8_t s, uint8x8_t t);
10075 int32x2_t psubw_s (int32x2_t s, int32x2_t t);
10076 int16x4_t psubh_s (int16x4_t s, int16x4_t t);
10077 int8x8_t psubb_s (int8x8_t s, int8x8_t t);
10078 uint64_t psubd_u (uint64_t s, uint64_t t);
10079 int64_t psubd_s (int64_t s, int64_t t);
10080 int16x4_t psubsh (int16x4_t s, int16x4_t t);
10081 int8x8_t psubsb (int8x8_t s, int8x8_t t);
10082 uint16x4_t psubush (uint16x4_t s, uint16x4_t t);
10083 uint8x8_t psubusb (uint8x8_t s, uint8x8_t t);
10084 uint32x2_t punpckhwd_u (uint32x2_t s, uint32x2_t t);
10085 uint16x4_t punpckhhw_u (uint16x4_t s, uint16x4_t t);
10086 uint8x8_t punpckhbh_u (uint8x8_t s, uint8x8_t t);
10087 int32x2_t punpckhwd_s (int32x2_t s, int32x2_t t);
10088 int16x4_t punpckhhw_s (int16x4_t s, int16x4_t t);
10089 int8x8_t punpckhbh_s (int8x8_t s, int8x8_t t);
10090 uint32x2_t punpcklwd_u (uint32x2_t s, uint32x2_t t);
10091 uint16x4_t punpcklhw_u (uint16x4_t s, uint16x4_t t);
10092 uint8x8_t punpcklbh_u (uint8x8_t s, uint8x8_t t);
10093 int32x2_t punpcklwd_s (int32x2_t s, int32x2_t t);
10094 int16x4_t punpcklhw_s (int16x4_t s, int16x4_t t);
10095 int8x8_t punpcklbh_s (int8x8_t s, int8x8_t t);
10099 * Paired-Single Arithmetic::
10100 * Paired-Single Built-in Functions::
10101 * MIPS-3D Built-in Functions::
10104 @node Paired-Single Arithmetic
10105 @subsubsection Paired-Single Arithmetic
10107 The table below lists the @code{v2sf} operations for which hardware
10108 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
10109 values and @code{x} is an integral value.
10111 @multitable @columnfractions .50 .50
10112 @item C code @tab MIPS instruction
10113 @item @code{a + b} @tab @code{add.ps}
10114 @item @code{a - b} @tab @code{sub.ps}
10115 @item @code{-a} @tab @code{neg.ps}
10116 @item @code{a * b} @tab @code{mul.ps}
10117 @item @code{a * b + c} @tab @code{madd.ps}
10118 @item @code{a * b - c} @tab @code{msub.ps}
10119 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
10120 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
10121 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
10124 Note that the multiply-accumulate instructions can be disabled
10125 using the command-line option @code{-mno-fused-madd}.
10127 @node Paired-Single Built-in Functions
10128 @subsubsection Paired-Single Built-in Functions
10130 The following paired-single functions map directly to a particular
10131 MIPS instruction. Please refer to the architecture specification
10132 for details on what each instruction does.
10135 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
10136 Pair lower lower (@code{pll.ps}).
10138 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
10139 Pair upper lower (@code{pul.ps}).
10141 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
10142 Pair lower upper (@code{plu.ps}).
10144 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
10145 Pair upper upper (@code{puu.ps}).
10147 @item v2sf __builtin_mips_cvt_ps_s (float, float)
10148 Convert pair to paired single (@code{cvt.ps.s}).
10150 @item float __builtin_mips_cvt_s_pl (v2sf)
10151 Convert pair lower to single (@code{cvt.s.pl}).
10153 @item float __builtin_mips_cvt_s_pu (v2sf)
10154 Convert pair upper to single (@code{cvt.s.pu}).
10156 @item v2sf __builtin_mips_abs_ps (v2sf)
10157 Absolute value (@code{abs.ps}).
10159 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
10160 Align variable (@code{alnv.ps}).
10162 @emph{Note:} The value of the third parameter must be 0 or 4
10163 modulo 8, otherwise the result will be unpredictable. Please read the
10164 instruction description for details.
10167 The following multi-instruction functions are also available.
10168 In each case, @var{cond} can be any of the 16 floating-point conditions:
10169 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
10170 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
10171 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
10174 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10175 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10176 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
10177 @code{movt.ps}/@code{movf.ps}).
10179 The @code{movt} functions return the value @var{x} computed by:
10182 c.@var{cond}.ps @var{cc},@var{a},@var{b}
10183 mov.ps @var{x},@var{c}
10184 movt.ps @var{x},@var{d},@var{cc}
10187 The @code{movf} functions are similar but use @code{movf.ps} instead
10190 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10191 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10192 Comparison of two paired-single values (@code{c.@var{cond}.ps},
10193 @code{bc1t}/@code{bc1f}).
10195 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
10196 and return either the upper or lower half of the result. For example:
10200 if (__builtin_mips_upper_c_eq_ps (a, b))
10201 upper_halves_are_equal ();
10203 upper_halves_are_unequal ();
10205 if (__builtin_mips_lower_c_eq_ps (a, b))
10206 lower_halves_are_equal ();
10208 lower_halves_are_unequal ();
10212 @node MIPS-3D Built-in Functions
10213 @subsubsection MIPS-3D Built-in Functions
10215 The MIPS-3D Application-Specific Extension (ASE) includes additional
10216 paired-single instructions that are designed to improve the performance
10217 of 3D graphics operations. Support for these instructions is controlled
10218 by the @option{-mips3d} command-line option.
10220 The functions listed below map directly to a particular MIPS-3D
10221 instruction. Please refer to the architecture specification for
10222 more details on what each instruction does.
10225 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
10226 Reduction add (@code{addr.ps}).
10228 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
10229 Reduction multiply (@code{mulr.ps}).
10231 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
10232 Convert paired single to paired word (@code{cvt.pw.ps}).
10234 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
10235 Convert paired word to paired single (@code{cvt.ps.pw}).
10237 @item float __builtin_mips_recip1_s (float)
10238 @itemx double __builtin_mips_recip1_d (double)
10239 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
10240 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
10242 @item float __builtin_mips_recip2_s (float, float)
10243 @itemx double __builtin_mips_recip2_d (double, double)
10244 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
10245 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
10247 @item float __builtin_mips_rsqrt1_s (float)
10248 @itemx double __builtin_mips_rsqrt1_d (double)
10249 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
10250 Reduced precision reciprocal square root (sequence step 1)
10251 (@code{rsqrt1.@var{fmt}}).
10253 @item float __builtin_mips_rsqrt2_s (float, float)
10254 @itemx double __builtin_mips_rsqrt2_d (double, double)
10255 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
10256 Reduced precision reciprocal square root (sequence step 2)
10257 (@code{rsqrt2.@var{fmt}}).
10260 The following multi-instruction functions are also available.
10261 In each case, @var{cond} can be any of the 16 floating-point conditions:
10262 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
10263 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
10264 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
10267 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
10268 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
10269 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
10270 @code{bc1t}/@code{bc1f}).
10272 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
10273 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
10278 if (__builtin_mips_cabs_eq_s (a, b))
10284 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10285 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10286 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
10287 @code{bc1t}/@code{bc1f}).
10289 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
10290 and return either the upper or lower half of the result. For example:
10294 if (__builtin_mips_upper_cabs_eq_ps (a, b))
10295 upper_halves_are_equal ();
10297 upper_halves_are_unequal ();
10299 if (__builtin_mips_lower_cabs_eq_ps (a, b))
10300 lower_halves_are_equal ();
10302 lower_halves_are_unequal ();
10305 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10306 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10307 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
10308 @code{movt.ps}/@code{movf.ps}).
10310 The @code{movt} functions return the value @var{x} computed by:
10313 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
10314 mov.ps @var{x},@var{c}
10315 movt.ps @var{x},@var{d},@var{cc}
10318 The @code{movf} functions are similar but use @code{movf.ps} instead
10321 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10322 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10323 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10324 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10325 Comparison of two paired-single values
10326 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
10327 @code{bc1any2t}/@code{bc1any2f}).
10329 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
10330 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
10331 result is true and the @code{all} forms return true if both results are true.
10336 if (__builtin_mips_any_c_eq_ps (a, b))
10341 if (__builtin_mips_all_c_eq_ps (a, b))
10347 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10348 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10349 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10350 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10351 Comparison of four paired-single values
10352 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
10353 @code{bc1any4t}/@code{bc1any4f}).
10355 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
10356 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
10357 The @code{any} forms return true if any of the four results are true
10358 and the @code{all} forms return true if all four results are true.
10363 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
10368 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
10375 @node picoChip Built-in Functions
10376 @subsection picoChip Built-in Functions
10378 GCC provides an interface to selected machine instructions from the
10379 picoChip instruction set.
10382 @item int __builtin_sbc (int @var{value})
10383 Sign bit count. Return the number of consecutive bits in @var{value}
10384 which have the same value as the sign-bit. The result is the number of
10385 leading sign bits minus one, giving the number of redundant sign bits in
10388 @item int __builtin_byteswap (int @var{value})
10389 Byte swap. Return the result of swapping the upper and lower bytes of
10392 @item int __builtin_brev (int @var{value})
10393 Bit reversal. Return the result of reversing the bits in
10394 @var{value}. Bit 15 is swapped with bit 0, bit 14 is swapped with bit 1,
10397 @item int __builtin_adds (int @var{x}, int @var{y})
10398 Saturating addition. Return the result of adding @var{x} and @var{y},
10399 storing the value 32767 if the result overflows.
10401 @item int __builtin_subs (int @var{x}, int @var{y})
10402 Saturating subtraction. Return the result of subtracting @var{y} from
10403 @var{x}, storing the value @minus{}32768 if the result overflows.
10405 @item void __builtin_halt (void)
10406 Halt. The processor will stop execution. This built-in is useful for
10407 implementing assertions.
10411 @node Other MIPS Built-in Functions
10412 @subsection Other MIPS Built-in Functions
10414 GCC provides other MIPS-specific built-in functions:
10417 @item void __builtin_mips_cache (int @var{op}, const volatile void *@var{addr})
10418 Insert a @samp{cache} instruction with operands @var{op} and @var{addr}.
10419 GCC defines the preprocessor macro @code{___GCC_HAVE_BUILTIN_MIPS_CACHE}
10420 when this function is available.
10423 @node PowerPC AltiVec/VSX Built-in Functions
10424 @subsection PowerPC AltiVec Built-in Functions
10426 GCC provides an interface for the PowerPC family of processors to access
10427 the AltiVec operations described in Motorola's AltiVec Programming
10428 Interface Manual. The interface is made available by including
10429 @code{<altivec.h>} and using @option{-maltivec} and
10430 @option{-mabi=altivec}. The interface supports the following vector
10434 vector unsigned char
10438 vector unsigned short
10439 vector signed short
10443 vector unsigned int
10449 If @option{-mvsx} is used the following additional vector types are
10453 vector unsigned long
10458 The long types are only implemented for 64-bit code generation, and
10459 the long type is only used in the floating point/integer conversion
10462 GCC's implementation of the high-level language interface available from
10463 C and C++ code differs from Motorola's documentation in several ways.
10468 A vector constant is a list of constant expressions within curly braces.
10471 A vector initializer requires no cast if the vector constant is of the
10472 same type as the variable it is initializing.
10475 If @code{signed} or @code{unsigned} is omitted, the signedness of the
10476 vector type is the default signedness of the base type. The default
10477 varies depending on the operating system, so a portable program should
10478 always specify the signedness.
10481 Compiling with @option{-maltivec} adds keywords @code{__vector},
10482 @code{vector}, @code{__pixel}, @code{pixel}, @code{__bool} and
10483 @code{bool}. When compiling ISO C, the context-sensitive substitution
10484 of the keywords @code{vector}, @code{pixel} and @code{bool} is
10485 disabled. To use them, you must include @code{<altivec.h>} instead.
10488 GCC allows using a @code{typedef} name as the type specifier for a
10492 For C, overloaded functions are implemented with macros so the following
10496 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
10499 Since @code{vec_add} is a macro, the vector constant in the example
10500 is treated as four separate arguments. Wrap the entire argument in
10501 parentheses for this to work.
10504 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
10505 Internally, GCC uses built-in functions to achieve the functionality in
10506 the aforementioned header file, but they are not supported and are
10507 subject to change without notice.
10509 The following interfaces are supported for the generic and specific
10510 AltiVec operations and the AltiVec predicates. In cases where there
10511 is a direct mapping between generic and specific operations, only the
10512 generic names are shown here, although the specific operations can also
10515 Arguments that are documented as @code{const int} require literal
10516 integral values within the range required for that operation.
10519 vector signed char vec_abs (vector signed char);
10520 vector signed short vec_abs (vector signed short);
10521 vector signed int vec_abs (vector signed int);
10522 vector float vec_abs (vector float);
10524 vector signed char vec_abss (vector signed char);
10525 vector signed short vec_abss (vector signed short);
10526 vector signed int vec_abss (vector signed int);
10528 vector signed char vec_add (vector bool char, vector signed char);
10529 vector signed char vec_add (vector signed char, vector bool char);
10530 vector signed char vec_add (vector signed char, vector signed char);
10531 vector unsigned char vec_add (vector bool char, vector unsigned char);
10532 vector unsigned char vec_add (vector unsigned char, vector bool char);
10533 vector unsigned char vec_add (vector unsigned char,
10534 vector unsigned char);
10535 vector signed short vec_add (vector bool short, vector signed short);
10536 vector signed short vec_add (vector signed short, vector bool short);
10537 vector signed short vec_add (vector signed short, vector signed short);
10538 vector unsigned short vec_add (vector bool short,
10539 vector unsigned short);
10540 vector unsigned short vec_add (vector unsigned short,
10541 vector bool short);
10542 vector unsigned short vec_add (vector unsigned short,
10543 vector unsigned short);
10544 vector signed int vec_add (vector bool int, vector signed int);
10545 vector signed int vec_add (vector signed int, vector bool int);
10546 vector signed int vec_add (vector signed int, vector signed int);
10547 vector unsigned int vec_add (vector bool int, vector unsigned int);
10548 vector unsigned int vec_add (vector unsigned int, vector bool int);
10549 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
10550 vector float vec_add (vector float, vector float);
10552 vector float vec_vaddfp (vector float, vector float);
10554 vector signed int vec_vadduwm (vector bool int, vector signed int);
10555 vector signed int vec_vadduwm (vector signed int, vector bool int);
10556 vector signed int vec_vadduwm (vector signed int, vector signed int);
10557 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
10558 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
10559 vector unsigned int vec_vadduwm (vector unsigned int,
10560 vector unsigned int);
10562 vector signed short vec_vadduhm (vector bool short,
10563 vector signed short);
10564 vector signed short vec_vadduhm (vector signed short,
10565 vector bool short);
10566 vector signed short vec_vadduhm (vector signed short,
10567 vector signed short);
10568 vector unsigned short vec_vadduhm (vector bool short,
10569 vector unsigned short);
10570 vector unsigned short vec_vadduhm (vector unsigned short,
10571 vector bool short);
10572 vector unsigned short vec_vadduhm (vector unsigned short,
10573 vector unsigned short);
10575 vector signed char vec_vaddubm (vector bool char, vector signed char);
10576 vector signed char vec_vaddubm (vector signed char, vector bool char);
10577 vector signed char vec_vaddubm (vector signed char, vector signed char);
10578 vector unsigned char vec_vaddubm (vector bool char,
10579 vector unsigned char);
10580 vector unsigned char vec_vaddubm (vector unsigned char,
10582 vector unsigned char vec_vaddubm (vector unsigned char,
10583 vector unsigned char);
10585 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
10587 vector unsigned char vec_adds (vector bool char, vector unsigned char);
10588 vector unsigned char vec_adds (vector unsigned char, vector bool char);
10589 vector unsigned char vec_adds (vector unsigned char,
10590 vector unsigned char);
10591 vector signed char vec_adds (vector bool char, vector signed char);
10592 vector signed char vec_adds (vector signed char, vector bool char);
10593 vector signed char vec_adds (vector signed char, vector signed char);
10594 vector unsigned short vec_adds (vector bool short,
10595 vector unsigned short);
10596 vector unsigned short vec_adds (vector unsigned short,
10597 vector bool short);
10598 vector unsigned short vec_adds (vector unsigned short,
10599 vector unsigned short);
10600 vector signed short vec_adds (vector bool short, vector signed short);
10601 vector signed short vec_adds (vector signed short, vector bool short);
10602 vector signed short vec_adds (vector signed short, vector signed short);
10603 vector unsigned int vec_adds (vector bool int, vector unsigned int);
10604 vector unsigned int vec_adds (vector unsigned int, vector bool int);
10605 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
10606 vector signed int vec_adds (vector bool int, vector signed int);
10607 vector signed int vec_adds (vector signed int, vector bool int);
10608 vector signed int vec_adds (vector signed int, vector signed int);
10610 vector signed int vec_vaddsws (vector bool int, vector signed int);
10611 vector signed int vec_vaddsws (vector signed int, vector bool int);
10612 vector signed int vec_vaddsws (vector signed int, vector signed int);
10614 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
10615 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
10616 vector unsigned int vec_vadduws (vector unsigned int,
10617 vector unsigned int);
10619 vector signed short vec_vaddshs (vector bool short,
10620 vector signed short);
10621 vector signed short vec_vaddshs (vector signed short,
10622 vector bool short);
10623 vector signed short vec_vaddshs (vector signed short,
10624 vector signed short);
10626 vector unsigned short vec_vadduhs (vector bool short,
10627 vector unsigned short);
10628 vector unsigned short vec_vadduhs (vector unsigned short,
10629 vector bool short);
10630 vector unsigned short vec_vadduhs (vector unsigned short,
10631 vector unsigned short);
10633 vector signed char vec_vaddsbs (vector bool char, vector signed char);
10634 vector signed char vec_vaddsbs (vector signed char, vector bool char);
10635 vector signed char vec_vaddsbs (vector signed char, vector signed char);
10637 vector unsigned char vec_vaddubs (vector bool char,
10638 vector unsigned char);
10639 vector unsigned char vec_vaddubs (vector unsigned char,
10641 vector unsigned char vec_vaddubs (vector unsigned char,
10642 vector unsigned char);
10644 vector float vec_and (vector float, vector float);
10645 vector float vec_and (vector float, vector bool int);
10646 vector float vec_and (vector bool int, vector float);
10647 vector bool int vec_and (vector bool int, vector bool int);
10648 vector signed int vec_and (vector bool int, vector signed int);
10649 vector signed int vec_and (vector signed int, vector bool int);
10650 vector signed int vec_and (vector signed int, vector signed int);
10651 vector unsigned int vec_and (vector bool int, vector unsigned int);
10652 vector unsigned int vec_and (vector unsigned int, vector bool int);
10653 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
10654 vector bool short vec_and (vector bool short, vector bool short);
10655 vector signed short vec_and (vector bool short, vector signed short);
10656 vector signed short vec_and (vector signed short, vector bool short);
10657 vector signed short vec_and (vector signed short, vector signed short);
10658 vector unsigned short vec_and (vector bool short,
10659 vector unsigned short);
10660 vector unsigned short vec_and (vector unsigned short,
10661 vector bool short);
10662 vector unsigned short vec_and (vector unsigned short,
10663 vector unsigned short);
10664 vector signed char vec_and (vector bool char, vector signed char);
10665 vector bool char vec_and (vector bool char, vector bool char);
10666 vector signed char vec_and (vector signed char, vector bool char);
10667 vector signed char vec_and (vector signed char, vector signed char);
10668 vector unsigned char vec_and (vector bool char, vector unsigned char);
10669 vector unsigned char vec_and (vector unsigned char, vector bool char);
10670 vector unsigned char vec_and (vector unsigned char,
10671 vector unsigned char);
10673 vector float vec_andc (vector float, vector float);
10674 vector float vec_andc (vector float, vector bool int);
10675 vector float vec_andc (vector bool int, vector float);
10676 vector bool int vec_andc (vector bool int, vector bool int);
10677 vector signed int vec_andc (vector bool int, vector signed int);
10678 vector signed int vec_andc (vector signed int, vector bool int);
10679 vector signed int vec_andc (vector signed int, vector signed int);
10680 vector unsigned int vec_andc (vector bool int, vector unsigned int);
10681 vector unsigned int vec_andc (vector unsigned int, vector bool int);
10682 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
10683 vector bool short vec_andc (vector bool short, vector bool short);
10684 vector signed short vec_andc (vector bool short, vector signed short);
10685 vector signed short vec_andc (vector signed short, vector bool short);
10686 vector signed short vec_andc (vector signed short, vector signed short);
10687 vector unsigned short vec_andc (vector bool short,
10688 vector unsigned short);
10689 vector unsigned short vec_andc (vector unsigned short,
10690 vector bool short);
10691 vector unsigned short vec_andc (vector unsigned short,
10692 vector unsigned short);
10693 vector signed char vec_andc (vector bool char, vector signed char);
10694 vector bool char vec_andc (vector bool char, vector bool char);
10695 vector signed char vec_andc (vector signed char, vector bool char);
10696 vector signed char vec_andc (vector signed char, vector signed char);
10697 vector unsigned char vec_andc (vector bool char, vector unsigned char);
10698 vector unsigned char vec_andc (vector unsigned char, vector bool char);
10699 vector unsigned char vec_andc (vector unsigned char,
10700 vector unsigned char);
10702 vector unsigned char vec_avg (vector unsigned char,
10703 vector unsigned char);
10704 vector signed char vec_avg (vector signed char, vector signed char);
10705 vector unsigned short vec_avg (vector unsigned short,
10706 vector unsigned short);
10707 vector signed short vec_avg (vector signed short, vector signed short);
10708 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
10709 vector signed int vec_avg (vector signed int, vector signed int);
10711 vector signed int vec_vavgsw (vector signed int, vector signed int);
10713 vector unsigned int vec_vavguw (vector unsigned int,
10714 vector unsigned int);
10716 vector signed short vec_vavgsh (vector signed short,
10717 vector signed short);
10719 vector unsigned short vec_vavguh (vector unsigned short,
10720 vector unsigned short);
10722 vector signed char vec_vavgsb (vector signed char, vector signed char);
10724 vector unsigned char vec_vavgub (vector unsigned char,
10725 vector unsigned char);
10727 vector float vec_copysign (vector float);
10729 vector float vec_ceil (vector float);
10731 vector signed int vec_cmpb (vector float, vector float);
10733 vector bool char vec_cmpeq (vector signed char, vector signed char);
10734 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
10735 vector bool short vec_cmpeq (vector signed short, vector signed short);
10736 vector bool short vec_cmpeq (vector unsigned short,
10737 vector unsigned short);
10738 vector bool int vec_cmpeq (vector signed int, vector signed int);
10739 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
10740 vector bool int vec_cmpeq (vector float, vector float);
10742 vector bool int vec_vcmpeqfp (vector float, vector float);
10744 vector bool int vec_vcmpequw (vector signed int, vector signed int);
10745 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
10747 vector bool short vec_vcmpequh (vector signed short,
10748 vector signed short);
10749 vector bool short vec_vcmpequh (vector unsigned short,
10750 vector unsigned short);
10752 vector bool char vec_vcmpequb (vector signed char, vector signed char);
10753 vector bool char vec_vcmpequb (vector unsigned char,
10754 vector unsigned char);
10756 vector bool int vec_cmpge (vector float, vector float);
10758 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
10759 vector bool char vec_cmpgt (vector signed char, vector signed char);
10760 vector bool short vec_cmpgt (vector unsigned short,
10761 vector unsigned short);
10762 vector bool short vec_cmpgt (vector signed short, vector signed short);
10763 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
10764 vector bool int vec_cmpgt (vector signed int, vector signed int);
10765 vector bool int vec_cmpgt (vector float, vector float);
10767 vector bool int vec_vcmpgtfp (vector float, vector float);
10769 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
10771 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
10773 vector bool short vec_vcmpgtsh (vector signed short,
10774 vector signed short);
10776 vector bool short vec_vcmpgtuh (vector unsigned short,
10777 vector unsigned short);
10779 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
10781 vector bool char vec_vcmpgtub (vector unsigned char,
10782 vector unsigned char);
10784 vector bool int vec_cmple (vector float, vector float);
10786 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
10787 vector bool char vec_cmplt (vector signed char, vector signed char);
10788 vector bool short vec_cmplt (vector unsigned short,
10789 vector unsigned short);
10790 vector bool short vec_cmplt (vector signed short, vector signed short);
10791 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
10792 vector bool int vec_cmplt (vector signed int, vector signed int);
10793 vector bool int vec_cmplt (vector float, vector float);
10795 vector float vec_ctf (vector unsigned int, const int);
10796 vector float vec_ctf (vector signed int, const int);
10798 vector float vec_vcfsx (vector signed int, const int);
10800 vector float vec_vcfux (vector unsigned int, const int);
10802 vector signed int vec_cts (vector float, const int);
10804 vector unsigned int vec_ctu (vector float, const int);
10806 void vec_dss (const int);
10808 void vec_dssall (void);
10810 void vec_dst (const vector unsigned char *, int, const int);
10811 void vec_dst (const vector signed char *, int, const int);
10812 void vec_dst (const vector bool char *, int, const int);
10813 void vec_dst (const vector unsigned short *, int, const int);
10814 void vec_dst (const vector signed short *, int, const int);
10815 void vec_dst (const vector bool short *, int, const int);
10816 void vec_dst (const vector pixel *, int, const int);
10817 void vec_dst (const vector unsigned int *, int, const int);
10818 void vec_dst (const vector signed int *, int, const int);
10819 void vec_dst (const vector bool int *, int, const int);
10820 void vec_dst (const vector float *, int, const int);
10821 void vec_dst (const unsigned char *, int, const int);
10822 void vec_dst (const signed char *, int, const int);
10823 void vec_dst (const unsigned short *, int, const int);
10824 void vec_dst (const short *, int, const int);
10825 void vec_dst (const unsigned int *, int, const int);
10826 void vec_dst (const int *, int, const int);
10827 void vec_dst (const unsigned long *, int, const int);
10828 void vec_dst (const long *, int, const int);
10829 void vec_dst (const float *, int, const int);
10831 void vec_dstst (const vector unsigned char *, int, const int);
10832 void vec_dstst (const vector signed char *, int, const int);
10833 void vec_dstst (const vector bool char *, int, const int);
10834 void vec_dstst (const vector unsigned short *, int, const int);
10835 void vec_dstst (const vector signed short *, int, const int);
10836 void vec_dstst (const vector bool short *, int, const int);
10837 void vec_dstst (const vector pixel *, int, const int);
10838 void vec_dstst (const vector unsigned int *, int, const int);
10839 void vec_dstst (const vector signed int *, int, const int);
10840 void vec_dstst (const vector bool int *, int, const int);
10841 void vec_dstst (const vector float *, int, const int);
10842 void vec_dstst (const unsigned char *, int, const int);
10843 void vec_dstst (const signed char *, int, const int);
10844 void vec_dstst (const unsigned short *, int, const int);
10845 void vec_dstst (const short *, int, const int);
10846 void vec_dstst (const unsigned int *, int, const int);
10847 void vec_dstst (const int *, int, const int);
10848 void vec_dstst (const unsigned long *, int, const int);
10849 void vec_dstst (const long *, int, const int);
10850 void vec_dstst (const float *, int, const int);
10852 void vec_dststt (const vector unsigned char *, int, const int);
10853 void vec_dststt (const vector signed char *, int, const int);
10854 void vec_dststt (const vector bool char *, int, const int);
10855 void vec_dststt (const vector unsigned short *, int, const int);
10856 void vec_dststt (const vector signed short *, int, const int);
10857 void vec_dststt (const vector bool short *, int, const int);
10858 void vec_dststt (const vector pixel *, int, const int);
10859 void vec_dststt (const vector unsigned int *, int, const int);
10860 void vec_dststt (const vector signed int *, int, const int);
10861 void vec_dststt (const vector bool int *, int, const int);
10862 void vec_dststt (const vector float *, int, const int);
10863 void vec_dststt (const unsigned char *, int, const int);
10864 void vec_dststt (const signed char *, int, const int);
10865 void vec_dststt (const unsigned short *, int, const int);
10866 void vec_dststt (const short *, int, const int);
10867 void vec_dststt (const unsigned int *, int, const int);
10868 void vec_dststt (const int *, int, const int);
10869 void vec_dststt (const unsigned long *, int, const int);
10870 void vec_dststt (const long *, int, const int);
10871 void vec_dststt (const float *, int, const int);
10873 void vec_dstt (const vector unsigned char *, int, const int);
10874 void vec_dstt (const vector signed char *, int, const int);
10875 void vec_dstt (const vector bool char *, int, const int);
10876 void vec_dstt (const vector unsigned short *, int, const int);
10877 void vec_dstt (const vector signed short *, int, const int);
10878 void vec_dstt (const vector bool short *, int, const int);
10879 void vec_dstt (const vector pixel *, int, const int);
10880 void vec_dstt (const vector unsigned int *, int, const int);
10881 void vec_dstt (const vector signed int *, int, const int);
10882 void vec_dstt (const vector bool int *, int, const int);
10883 void vec_dstt (const vector float *, int, const int);
10884 void vec_dstt (const unsigned char *, int, const int);
10885 void vec_dstt (const signed char *, int, const int);
10886 void vec_dstt (const unsigned short *, int, const int);
10887 void vec_dstt (const short *, int, const int);
10888 void vec_dstt (const unsigned int *, int, const int);
10889 void vec_dstt (const int *, int, const int);
10890 void vec_dstt (const unsigned long *, int, const int);
10891 void vec_dstt (const long *, int, const int);
10892 void vec_dstt (const float *, int, const int);
10894 vector float vec_expte (vector float);
10896 vector float vec_floor (vector float);
10898 vector float vec_ld (int, const vector float *);
10899 vector float vec_ld (int, const float *);
10900 vector bool int vec_ld (int, const vector bool int *);
10901 vector signed int vec_ld (int, const vector signed int *);
10902 vector signed int vec_ld (int, const int *);
10903 vector signed int vec_ld (int, const long *);
10904 vector unsigned int vec_ld (int, const vector unsigned int *);
10905 vector unsigned int vec_ld (int, const unsigned int *);
10906 vector unsigned int vec_ld (int, const unsigned long *);
10907 vector bool short vec_ld (int, const vector bool short *);
10908 vector pixel vec_ld (int, const vector pixel *);
10909 vector signed short vec_ld (int, const vector signed short *);
10910 vector signed short vec_ld (int, const short *);
10911 vector unsigned short vec_ld (int, const vector unsigned short *);
10912 vector unsigned short vec_ld (int, const unsigned short *);
10913 vector bool char vec_ld (int, const vector bool char *);
10914 vector signed char vec_ld (int, const vector signed char *);
10915 vector signed char vec_ld (int, const signed char *);
10916 vector unsigned char vec_ld (int, const vector unsigned char *);
10917 vector unsigned char vec_ld (int, const unsigned char *);
10919 vector signed char vec_lde (int, const signed char *);
10920 vector unsigned char vec_lde (int, const unsigned char *);
10921 vector signed short vec_lde (int, const short *);
10922 vector unsigned short vec_lde (int, const unsigned short *);
10923 vector float vec_lde (int, const float *);
10924 vector signed int vec_lde (int, const int *);
10925 vector unsigned int vec_lde (int, const unsigned int *);
10926 vector signed int vec_lde (int, const long *);
10927 vector unsigned int vec_lde (int, const unsigned long *);
10929 vector float vec_lvewx (int, float *);
10930 vector signed int vec_lvewx (int, int *);
10931 vector unsigned int vec_lvewx (int, unsigned int *);
10932 vector signed int vec_lvewx (int, long *);
10933 vector unsigned int vec_lvewx (int, unsigned long *);
10935 vector signed short vec_lvehx (int, short *);
10936 vector unsigned short vec_lvehx (int, unsigned short *);
10938 vector signed char vec_lvebx (int, char *);
10939 vector unsigned char vec_lvebx (int, unsigned char *);
10941 vector float vec_ldl (int, const vector float *);
10942 vector float vec_ldl (int, const float *);
10943 vector bool int vec_ldl (int, const vector bool int *);
10944 vector signed int vec_ldl (int, const vector signed int *);
10945 vector signed int vec_ldl (int, const int *);
10946 vector signed int vec_ldl (int, const long *);
10947 vector unsigned int vec_ldl (int, const vector unsigned int *);
10948 vector unsigned int vec_ldl (int, const unsigned int *);
10949 vector unsigned int vec_ldl (int, const unsigned long *);
10950 vector bool short vec_ldl (int, const vector bool short *);
10951 vector pixel vec_ldl (int, const vector pixel *);
10952 vector signed short vec_ldl (int, const vector signed short *);
10953 vector signed short vec_ldl (int, const short *);
10954 vector unsigned short vec_ldl (int, const vector unsigned short *);
10955 vector unsigned short vec_ldl (int, const unsigned short *);
10956 vector bool char vec_ldl (int, const vector bool char *);
10957 vector signed char vec_ldl (int, const vector signed char *);
10958 vector signed char vec_ldl (int, const signed char *);
10959 vector unsigned char vec_ldl (int, const vector unsigned char *);
10960 vector unsigned char vec_ldl (int, const unsigned char *);
10962 vector float vec_loge (vector float);
10964 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
10965 vector unsigned char vec_lvsl (int, const volatile signed char *);
10966 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
10967 vector unsigned char vec_lvsl (int, const volatile short *);
10968 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
10969 vector unsigned char vec_lvsl (int, const volatile int *);
10970 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
10971 vector unsigned char vec_lvsl (int, const volatile long *);
10972 vector unsigned char vec_lvsl (int, const volatile float *);
10974 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
10975 vector unsigned char vec_lvsr (int, const volatile signed char *);
10976 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
10977 vector unsigned char vec_lvsr (int, const volatile short *);
10978 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
10979 vector unsigned char vec_lvsr (int, const volatile int *);
10980 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
10981 vector unsigned char vec_lvsr (int, const volatile long *);
10982 vector unsigned char vec_lvsr (int, const volatile float *);
10984 vector float vec_madd (vector float, vector float, vector float);
10986 vector signed short vec_madds (vector signed short,
10987 vector signed short,
10988 vector signed short);
10990 vector unsigned char vec_max (vector bool char, vector unsigned char);
10991 vector unsigned char vec_max (vector unsigned char, vector bool char);
10992 vector unsigned char vec_max (vector unsigned char,
10993 vector unsigned char);
10994 vector signed char vec_max (vector bool char, vector signed char);
10995 vector signed char vec_max (vector signed char, vector bool char);
10996 vector signed char vec_max (vector signed char, vector signed char);
10997 vector unsigned short vec_max (vector bool short,
10998 vector unsigned short);
10999 vector unsigned short vec_max (vector unsigned short,
11000 vector bool short);
11001 vector unsigned short vec_max (vector unsigned short,
11002 vector unsigned short);
11003 vector signed short vec_max (vector bool short, vector signed short);
11004 vector signed short vec_max (vector signed short, vector bool short);
11005 vector signed short vec_max (vector signed short, vector signed short);
11006 vector unsigned int vec_max (vector bool int, vector unsigned int);
11007 vector unsigned int vec_max (vector unsigned int, vector bool int);
11008 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
11009 vector signed int vec_max (vector bool int, vector signed int);
11010 vector signed int vec_max (vector signed int, vector bool int);
11011 vector signed int vec_max (vector signed int, vector signed int);
11012 vector float vec_max (vector float, vector float);
11014 vector float vec_vmaxfp (vector float, vector float);
11016 vector signed int vec_vmaxsw (vector bool int, vector signed int);
11017 vector signed int vec_vmaxsw (vector signed int, vector bool int);
11018 vector signed int vec_vmaxsw (vector signed int, vector signed int);
11020 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
11021 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
11022 vector unsigned int vec_vmaxuw (vector unsigned int,
11023 vector unsigned int);
11025 vector signed short vec_vmaxsh (vector bool short, vector signed short);
11026 vector signed short vec_vmaxsh (vector signed short, vector bool short);
11027 vector signed short vec_vmaxsh (vector signed short,
11028 vector signed short);
11030 vector unsigned short vec_vmaxuh (vector bool short,
11031 vector unsigned short);
11032 vector unsigned short vec_vmaxuh (vector unsigned short,
11033 vector bool short);
11034 vector unsigned short vec_vmaxuh (vector unsigned short,
11035 vector unsigned short);
11037 vector signed char vec_vmaxsb (vector bool char, vector signed char);
11038 vector signed char vec_vmaxsb (vector signed char, vector bool char);
11039 vector signed char vec_vmaxsb (vector signed char, vector signed char);
11041 vector unsigned char vec_vmaxub (vector bool char,
11042 vector unsigned char);
11043 vector unsigned char vec_vmaxub (vector unsigned char,
11045 vector unsigned char vec_vmaxub (vector unsigned char,
11046 vector unsigned char);
11048 vector bool char vec_mergeh (vector bool char, vector bool char);
11049 vector signed char vec_mergeh (vector signed char, vector signed char);
11050 vector unsigned char vec_mergeh (vector unsigned char,
11051 vector unsigned char);
11052 vector bool short vec_mergeh (vector bool short, vector bool short);
11053 vector pixel vec_mergeh (vector pixel, vector pixel);
11054 vector signed short vec_mergeh (vector signed short,
11055 vector signed short);
11056 vector unsigned short vec_mergeh (vector unsigned short,
11057 vector unsigned short);
11058 vector float vec_mergeh (vector float, vector float);
11059 vector bool int vec_mergeh (vector bool int, vector bool int);
11060 vector signed int vec_mergeh (vector signed int, vector signed int);
11061 vector unsigned int vec_mergeh (vector unsigned int,
11062 vector unsigned int);
11064 vector float vec_vmrghw (vector float, vector float);
11065 vector bool int vec_vmrghw (vector bool int, vector bool int);
11066 vector signed int vec_vmrghw (vector signed int, vector signed int);
11067 vector unsigned int vec_vmrghw (vector unsigned int,
11068 vector unsigned int);
11070 vector bool short vec_vmrghh (vector bool short, vector bool short);
11071 vector signed short vec_vmrghh (vector signed short,
11072 vector signed short);
11073 vector unsigned short vec_vmrghh (vector unsigned short,
11074 vector unsigned short);
11075 vector pixel vec_vmrghh (vector pixel, vector pixel);
11077 vector bool char vec_vmrghb (vector bool char, vector bool char);
11078 vector signed char vec_vmrghb (vector signed char, vector signed char);
11079 vector unsigned char vec_vmrghb (vector unsigned char,
11080 vector unsigned char);
11082 vector bool char vec_mergel (vector bool char, vector bool char);
11083 vector signed char vec_mergel (vector signed char, vector signed char);
11084 vector unsigned char vec_mergel (vector unsigned char,
11085 vector unsigned char);
11086 vector bool short vec_mergel (vector bool short, vector bool short);
11087 vector pixel vec_mergel (vector pixel, vector pixel);
11088 vector signed short vec_mergel (vector signed short,
11089 vector signed short);
11090 vector unsigned short vec_mergel (vector unsigned short,
11091 vector unsigned short);
11092 vector float vec_mergel (vector float, vector float);
11093 vector bool int vec_mergel (vector bool int, vector bool int);
11094 vector signed int vec_mergel (vector signed int, vector signed int);
11095 vector unsigned int vec_mergel (vector unsigned int,
11096 vector unsigned int);
11098 vector float vec_vmrglw (vector float, vector float);
11099 vector signed int vec_vmrglw (vector signed int, vector signed int);
11100 vector unsigned int vec_vmrglw (vector unsigned int,
11101 vector unsigned int);
11102 vector bool int vec_vmrglw (vector bool int, vector bool int);
11104 vector bool short vec_vmrglh (vector bool short, vector bool short);
11105 vector signed short vec_vmrglh (vector signed short,
11106 vector signed short);
11107 vector unsigned short vec_vmrglh (vector unsigned short,
11108 vector unsigned short);
11109 vector pixel vec_vmrglh (vector pixel, vector pixel);
11111 vector bool char vec_vmrglb (vector bool char, vector bool char);
11112 vector signed char vec_vmrglb (vector signed char, vector signed char);
11113 vector unsigned char vec_vmrglb (vector unsigned char,
11114 vector unsigned char);
11116 vector unsigned short vec_mfvscr (void);
11118 vector unsigned char vec_min (vector bool char, vector unsigned char);
11119 vector unsigned char vec_min (vector unsigned char, vector bool char);
11120 vector unsigned char vec_min (vector unsigned char,
11121 vector unsigned char);
11122 vector signed char vec_min (vector bool char, vector signed char);
11123 vector signed char vec_min (vector signed char, vector bool char);
11124 vector signed char vec_min (vector signed char, vector signed char);
11125 vector unsigned short vec_min (vector bool short,
11126 vector unsigned short);
11127 vector unsigned short vec_min (vector unsigned short,
11128 vector bool short);
11129 vector unsigned short vec_min (vector unsigned short,
11130 vector unsigned short);
11131 vector signed short vec_min (vector bool short, vector signed short);
11132 vector signed short vec_min (vector signed short, vector bool short);
11133 vector signed short vec_min (vector signed short, vector signed short);
11134 vector unsigned int vec_min (vector bool int, vector unsigned int);
11135 vector unsigned int vec_min (vector unsigned int, vector bool int);
11136 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
11137 vector signed int vec_min (vector bool int, vector signed int);
11138 vector signed int vec_min (vector signed int, vector bool int);
11139 vector signed int vec_min (vector signed int, vector signed int);
11140 vector float vec_min (vector float, vector float);
11142 vector float vec_vminfp (vector float, vector float);
11144 vector signed int vec_vminsw (vector bool int, vector signed int);
11145 vector signed int vec_vminsw (vector signed int, vector bool int);
11146 vector signed int vec_vminsw (vector signed int, vector signed int);
11148 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
11149 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
11150 vector unsigned int vec_vminuw (vector unsigned int,
11151 vector unsigned int);
11153 vector signed short vec_vminsh (vector bool short, vector signed short);
11154 vector signed short vec_vminsh (vector signed short, vector bool short);
11155 vector signed short vec_vminsh (vector signed short,
11156 vector signed short);
11158 vector unsigned short vec_vminuh (vector bool short,
11159 vector unsigned short);
11160 vector unsigned short vec_vminuh (vector unsigned short,
11161 vector bool short);
11162 vector unsigned short vec_vminuh (vector unsigned short,
11163 vector unsigned short);
11165 vector signed char vec_vminsb (vector bool char, vector signed char);
11166 vector signed char vec_vminsb (vector signed char, vector bool char);
11167 vector signed char vec_vminsb (vector signed char, vector signed char);
11169 vector unsigned char vec_vminub (vector bool char,
11170 vector unsigned char);
11171 vector unsigned char vec_vminub (vector unsigned char,
11173 vector unsigned char vec_vminub (vector unsigned char,
11174 vector unsigned char);
11176 vector signed short vec_mladd (vector signed short,
11177 vector signed short,
11178 vector signed short);
11179 vector signed short vec_mladd (vector signed short,
11180 vector unsigned short,
11181 vector unsigned short);
11182 vector signed short vec_mladd (vector unsigned short,
11183 vector signed short,
11184 vector signed short);
11185 vector unsigned short vec_mladd (vector unsigned short,
11186 vector unsigned short,
11187 vector unsigned short);
11189 vector signed short vec_mradds (vector signed short,
11190 vector signed short,
11191 vector signed short);
11193 vector unsigned int vec_msum (vector unsigned char,
11194 vector unsigned char,
11195 vector unsigned int);
11196 vector signed int vec_msum (vector signed char,
11197 vector unsigned char,
11198 vector signed int);
11199 vector unsigned int vec_msum (vector unsigned short,
11200 vector unsigned short,
11201 vector unsigned int);
11202 vector signed int vec_msum (vector signed short,
11203 vector signed short,
11204 vector signed int);
11206 vector signed int vec_vmsumshm (vector signed short,
11207 vector signed short,
11208 vector signed int);
11210 vector unsigned int vec_vmsumuhm (vector unsigned short,
11211 vector unsigned short,
11212 vector unsigned int);
11214 vector signed int vec_vmsummbm (vector signed char,
11215 vector unsigned char,
11216 vector signed int);
11218 vector unsigned int vec_vmsumubm (vector unsigned char,
11219 vector unsigned char,
11220 vector unsigned int);
11222 vector unsigned int vec_msums (vector unsigned short,
11223 vector unsigned short,
11224 vector unsigned int);
11225 vector signed int vec_msums (vector signed short,
11226 vector signed short,
11227 vector signed int);
11229 vector signed int vec_vmsumshs (vector signed short,
11230 vector signed short,
11231 vector signed int);
11233 vector unsigned int vec_vmsumuhs (vector unsigned short,
11234 vector unsigned short,
11235 vector unsigned int);
11237 void vec_mtvscr (vector signed int);
11238 void vec_mtvscr (vector unsigned int);
11239 void vec_mtvscr (vector bool int);
11240 void vec_mtvscr (vector signed short);
11241 void vec_mtvscr (vector unsigned short);
11242 void vec_mtvscr (vector bool short);
11243 void vec_mtvscr (vector pixel);
11244 void vec_mtvscr (vector signed char);
11245 void vec_mtvscr (vector unsigned char);
11246 void vec_mtvscr (vector bool char);
11248 vector unsigned short vec_mule (vector unsigned char,
11249 vector unsigned char);
11250 vector signed short vec_mule (vector signed char,
11251 vector signed char);
11252 vector unsigned int vec_mule (vector unsigned short,
11253 vector unsigned short);
11254 vector signed int vec_mule (vector signed short, vector signed short);
11256 vector signed int vec_vmulesh (vector signed short,
11257 vector signed short);
11259 vector unsigned int vec_vmuleuh (vector unsigned short,
11260 vector unsigned short);
11262 vector signed short vec_vmulesb (vector signed char,
11263 vector signed char);
11265 vector unsigned short vec_vmuleub (vector unsigned char,
11266 vector unsigned char);
11268 vector unsigned short vec_mulo (vector unsigned char,
11269 vector unsigned char);
11270 vector signed short vec_mulo (vector signed char, vector signed char);
11271 vector unsigned int vec_mulo (vector unsigned short,
11272 vector unsigned short);
11273 vector signed int vec_mulo (vector signed short, vector signed short);
11275 vector signed int vec_vmulosh (vector signed short,
11276 vector signed short);
11278 vector unsigned int vec_vmulouh (vector unsigned short,
11279 vector unsigned short);
11281 vector signed short vec_vmulosb (vector signed char,
11282 vector signed char);
11284 vector unsigned short vec_vmuloub (vector unsigned char,
11285 vector unsigned char);
11287 vector float vec_nmsub (vector float, vector float, vector float);
11289 vector float vec_nor (vector float, vector float);
11290 vector signed int vec_nor (vector signed int, vector signed int);
11291 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
11292 vector bool int vec_nor (vector bool int, vector bool int);
11293 vector signed short vec_nor (vector signed short, vector signed short);
11294 vector unsigned short vec_nor (vector unsigned short,
11295 vector unsigned short);
11296 vector bool short vec_nor (vector bool short, vector bool short);
11297 vector signed char vec_nor (vector signed char, vector signed char);
11298 vector unsigned char vec_nor (vector unsigned char,
11299 vector unsigned char);
11300 vector bool char vec_nor (vector bool char, vector bool char);
11302 vector float vec_or (vector float, vector float);
11303 vector float vec_or (vector float, vector bool int);
11304 vector float vec_or (vector bool int, vector float);
11305 vector bool int vec_or (vector bool int, vector bool int);
11306 vector signed int vec_or (vector bool int, vector signed int);
11307 vector signed int vec_or (vector signed int, vector bool int);
11308 vector signed int vec_or (vector signed int, vector signed int);
11309 vector unsigned int vec_or (vector bool int, vector unsigned int);
11310 vector unsigned int vec_or (vector unsigned int, vector bool int);
11311 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
11312 vector bool short vec_or (vector bool short, vector bool short);
11313 vector signed short vec_or (vector bool short, vector signed short);
11314 vector signed short vec_or (vector signed short, vector bool short);
11315 vector signed short vec_or (vector signed short, vector signed short);
11316 vector unsigned short vec_or (vector bool short, vector unsigned short);
11317 vector unsigned short vec_or (vector unsigned short, vector bool short);
11318 vector unsigned short vec_or (vector unsigned short,
11319 vector unsigned short);
11320 vector signed char vec_or (vector bool char, vector signed char);
11321 vector bool char vec_or (vector bool char, vector bool char);
11322 vector signed char vec_or (vector signed char, vector bool char);
11323 vector signed char vec_or (vector signed char, vector signed char);
11324 vector unsigned char vec_or (vector bool char, vector unsigned char);
11325 vector unsigned char vec_or (vector unsigned char, vector bool char);
11326 vector unsigned char vec_or (vector unsigned char,
11327 vector unsigned char);
11329 vector signed char vec_pack (vector signed short, vector signed short);
11330 vector unsigned char vec_pack (vector unsigned short,
11331 vector unsigned short);
11332 vector bool char vec_pack (vector bool short, vector bool short);
11333 vector signed short vec_pack (vector signed int, vector signed int);
11334 vector unsigned short vec_pack (vector unsigned int,
11335 vector unsigned int);
11336 vector bool short vec_pack (vector bool int, vector bool int);
11338 vector bool short vec_vpkuwum (vector bool int, vector bool int);
11339 vector signed short vec_vpkuwum (vector signed int, vector signed int);
11340 vector unsigned short vec_vpkuwum (vector unsigned int,
11341 vector unsigned int);
11343 vector bool char vec_vpkuhum (vector bool short, vector bool short);
11344 vector signed char vec_vpkuhum (vector signed short,
11345 vector signed short);
11346 vector unsigned char vec_vpkuhum (vector unsigned short,
11347 vector unsigned short);
11349 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
11351 vector unsigned char vec_packs (vector unsigned short,
11352 vector unsigned short);
11353 vector signed char vec_packs (vector signed short, vector signed short);
11354 vector unsigned short vec_packs (vector unsigned int,
11355 vector unsigned int);
11356 vector signed short vec_packs (vector signed int, vector signed int);
11358 vector signed short vec_vpkswss (vector signed int, vector signed int);
11360 vector unsigned short vec_vpkuwus (vector unsigned int,
11361 vector unsigned int);
11363 vector signed char vec_vpkshss (vector signed short,
11364 vector signed short);
11366 vector unsigned char vec_vpkuhus (vector unsigned short,
11367 vector unsigned short);
11369 vector unsigned char vec_packsu (vector unsigned short,
11370 vector unsigned short);
11371 vector unsigned char vec_packsu (vector signed short,
11372 vector signed short);
11373 vector unsigned short vec_packsu (vector unsigned int,
11374 vector unsigned int);
11375 vector unsigned short vec_packsu (vector signed int, vector signed int);
11377 vector unsigned short vec_vpkswus (vector signed int,
11378 vector signed int);
11380 vector unsigned char vec_vpkshus (vector signed short,
11381 vector signed short);
11383 vector float vec_perm (vector float,
11385 vector unsigned char);
11386 vector signed int vec_perm (vector signed int,
11388 vector unsigned char);
11389 vector unsigned int vec_perm (vector unsigned int,
11390 vector unsigned int,
11391 vector unsigned char);
11392 vector bool int vec_perm (vector bool int,
11394 vector unsigned char);
11395 vector signed short vec_perm (vector signed short,
11396 vector signed short,
11397 vector unsigned char);
11398 vector unsigned short vec_perm (vector unsigned short,
11399 vector unsigned short,
11400 vector unsigned char);
11401 vector bool short vec_perm (vector bool short,
11403 vector unsigned char);
11404 vector pixel vec_perm (vector pixel,
11406 vector unsigned char);
11407 vector signed char vec_perm (vector signed char,
11408 vector signed char,
11409 vector unsigned char);
11410 vector unsigned char vec_perm (vector unsigned char,
11411 vector unsigned char,
11412 vector unsigned char);
11413 vector bool char vec_perm (vector bool char,
11415 vector unsigned char);
11417 vector float vec_re (vector float);
11419 vector signed char vec_rl (vector signed char,
11420 vector unsigned char);
11421 vector unsigned char vec_rl (vector unsigned char,
11422 vector unsigned char);
11423 vector signed short vec_rl (vector signed short, vector unsigned short);
11424 vector unsigned short vec_rl (vector unsigned short,
11425 vector unsigned short);
11426 vector signed int vec_rl (vector signed int, vector unsigned int);
11427 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
11429 vector signed int vec_vrlw (vector signed int, vector unsigned int);
11430 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
11432 vector signed short vec_vrlh (vector signed short,
11433 vector unsigned short);
11434 vector unsigned short vec_vrlh (vector unsigned short,
11435 vector unsigned short);
11437 vector signed char vec_vrlb (vector signed char, vector unsigned char);
11438 vector unsigned char vec_vrlb (vector unsigned char,
11439 vector unsigned char);
11441 vector float vec_round (vector float);
11443 vector float vec_recip (vector float, vector float);
11445 vector float vec_rsqrt (vector float);
11447 vector float vec_rsqrte (vector float);
11449 vector float vec_sel (vector float, vector float, vector bool int);
11450 vector float vec_sel (vector float, vector float, vector unsigned int);
11451 vector signed int vec_sel (vector signed int,
11454 vector signed int vec_sel (vector signed int,
11456 vector unsigned int);
11457 vector unsigned int vec_sel (vector unsigned int,
11458 vector unsigned int,
11460 vector unsigned int vec_sel (vector unsigned int,
11461 vector unsigned int,
11462 vector unsigned int);
11463 vector bool int vec_sel (vector bool int,
11466 vector bool int vec_sel (vector bool int,
11468 vector unsigned int);
11469 vector signed short vec_sel (vector signed short,
11470 vector signed short,
11471 vector bool short);
11472 vector signed short vec_sel (vector signed short,
11473 vector signed short,
11474 vector unsigned short);
11475 vector unsigned short vec_sel (vector unsigned short,
11476 vector unsigned short,
11477 vector bool short);
11478 vector unsigned short vec_sel (vector unsigned short,
11479 vector unsigned short,
11480 vector unsigned short);
11481 vector bool short vec_sel (vector bool short,
11483 vector bool short);
11484 vector bool short vec_sel (vector bool short,
11486 vector unsigned short);
11487 vector signed char vec_sel (vector signed char,
11488 vector signed char,
11490 vector signed char vec_sel (vector signed char,
11491 vector signed char,
11492 vector unsigned char);
11493 vector unsigned char vec_sel (vector unsigned char,
11494 vector unsigned char,
11496 vector unsigned char vec_sel (vector unsigned char,
11497 vector unsigned char,
11498 vector unsigned char);
11499 vector bool char vec_sel (vector bool char,
11502 vector bool char vec_sel (vector bool char,
11504 vector unsigned char);
11506 vector signed char vec_sl (vector signed char,
11507 vector unsigned char);
11508 vector unsigned char vec_sl (vector unsigned char,
11509 vector unsigned char);
11510 vector signed short vec_sl (vector signed short, vector unsigned short);
11511 vector unsigned short vec_sl (vector unsigned short,
11512 vector unsigned short);
11513 vector signed int vec_sl (vector signed int, vector unsigned int);
11514 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
11516 vector signed int vec_vslw (vector signed int, vector unsigned int);
11517 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
11519 vector signed short vec_vslh (vector signed short,
11520 vector unsigned short);
11521 vector unsigned short vec_vslh (vector unsigned short,
11522 vector unsigned short);
11524 vector signed char vec_vslb (vector signed char, vector unsigned char);
11525 vector unsigned char vec_vslb (vector unsigned char,
11526 vector unsigned char);
11528 vector float vec_sld (vector float, vector float, const int);
11529 vector signed int vec_sld (vector signed int,
11532 vector unsigned int vec_sld (vector unsigned int,
11533 vector unsigned int,
11535 vector bool int vec_sld (vector bool int,
11538 vector signed short vec_sld (vector signed short,
11539 vector signed short,
11541 vector unsigned short vec_sld (vector unsigned short,
11542 vector unsigned short,
11544 vector bool short vec_sld (vector bool short,
11547 vector pixel vec_sld (vector pixel,
11550 vector signed char vec_sld (vector signed char,
11551 vector signed char,
11553 vector unsigned char vec_sld (vector unsigned char,
11554 vector unsigned char,
11556 vector bool char vec_sld (vector bool char,
11560 vector signed int vec_sll (vector signed int,
11561 vector unsigned int);
11562 vector signed int vec_sll (vector signed int,
11563 vector unsigned short);
11564 vector signed int vec_sll (vector signed int,
11565 vector unsigned char);
11566 vector unsigned int vec_sll (vector unsigned int,
11567 vector unsigned int);
11568 vector unsigned int vec_sll (vector unsigned int,
11569 vector unsigned short);
11570 vector unsigned int vec_sll (vector unsigned int,
11571 vector unsigned char);
11572 vector bool int vec_sll (vector bool int,
11573 vector unsigned int);
11574 vector bool int vec_sll (vector bool int,
11575 vector unsigned short);
11576 vector bool int vec_sll (vector bool int,
11577 vector unsigned char);
11578 vector signed short vec_sll (vector signed short,
11579 vector unsigned int);
11580 vector signed short vec_sll (vector signed short,
11581 vector unsigned short);
11582 vector signed short vec_sll (vector signed short,
11583 vector unsigned char);
11584 vector unsigned short vec_sll (vector unsigned short,
11585 vector unsigned int);
11586 vector unsigned short vec_sll (vector unsigned short,
11587 vector unsigned short);
11588 vector unsigned short vec_sll (vector unsigned short,
11589 vector unsigned char);
11590 vector bool short vec_sll (vector bool short, vector unsigned int);
11591 vector bool short vec_sll (vector bool short, vector unsigned short);
11592 vector bool short vec_sll (vector bool short, vector unsigned char);
11593 vector pixel vec_sll (vector pixel, vector unsigned int);
11594 vector pixel vec_sll (vector pixel, vector unsigned short);
11595 vector pixel vec_sll (vector pixel, vector unsigned char);
11596 vector signed char vec_sll (vector signed char, vector unsigned int);
11597 vector signed char vec_sll (vector signed char, vector unsigned short);
11598 vector signed char vec_sll (vector signed char, vector unsigned char);
11599 vector unsigned char vec_sll (vector unsigned char,
11600 vector unsigned int);
11601 vector unsigned char vec_sll (vector unsigned char,
11602 vector unsigned short);
11603 vector unsigned char vec_sll (vector unsigned char,
11604 vector unsigned char);
11605 vector bool char vec_sll (vector bool char, vector unsigned int);
11606 vector bool char vec_sll (vector bool char, vector unsigned short);
11607 vector bool char vec_sll (vector bool char, vector unsigned char);
11609 vector float vec_slo (vector float, vector signed char);
11610 vector float vec_slo (vector float, vector unsigned char);
11611 vector signed int vec_slo (vector signed int, vector signed char);
11612 vector signed int vec_slo (vector signed int, vector unsigned char);
11613 vector unsigned int vec_slo (vector unsigned int, vector signed char);
11614 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
11615 vector signed short vec_slo (vector signed short, vector signed char);
11616 vector signed short vec_slo (vector signed short, vector unsigned char);
11617 vector unsigned short vec_slo (vector unsigned short,
11618 vector signed char);
11619 vector unsigned short vec_slo (vector unsigned short,
11620 vector unsigned char);
11621 vector pixel vec_slo (vector pixel, vector signed char);
11622 vector pixel vec_slo (vector pixel, vector unsigned char);
11623 vector signed char vec_slo (vector signed char, vector signed char);
11624 vector signed char vec_slo (vector signed char, vector unsigned char);
11625 vector unsigned char vec_slo (vector unsigned char, vector signed char);
11626 vector unsigned char vec_slo (vector unsigned char,
11627 vector unsigned char);
11629 vector signed char vec_splat (vector signed char, const int);
11630 vector unsigned char vec_splat (vector unsigned char, const int);
11631 vector bool char vec_splat (vector bool char, const int);
11632 vector signed short vec_splat (vector signed short, const int);
11633 vector unsigned short vec_splat (vector unsigned short, const int);
11634 vector bool short vec_splat (vector bool short, const int);
11635 vector pixel vec_splat (vector pixel, const int);
11636 vector float vec_splat (vector float, const int);
11637 vector signed int vec_splat (vector signed int, const int);
11638 vector unsigned int vec_splat (vector unsigned int, const int);
11639 vector bool int vec_splat (vector bool int, const int);
11641 vector float vec_vspltw (vector float, const int);
11642 vector signed int vec_vspltw (vector signed int, const int);
11643 vector unsigned int vec_vspltw (vector unsigned int, const int);
11644 vector bool int vec_vspltw (vector bool int, const int);
11646 vector bool short vec_vsplth (vector bool short, const int);
11647 vector signed short vec_vsplth (vector signed short, const int);
11648 vector unsigned short vec_vsplth (vector unsigned short, const int);
11649 vector pixel vec_vsplth (vector pixel, const int);
11651 vector signed char vec_vspltb (vector signed char, const int);
11652 vector unsigned char vec_vspltb (vector unsigned char, const int);
11653 vector bool char vec_vspltb (vector bool char, const int);
11655 vector signed char vec_splat_s8 (const int);
11657 vector signed short vec_splat_s16 (const int);
11659 vector signed int vec_splat_s32 (const int);
11661 vector unsigned char vec_splat_u8 (const int);
11663 vector unsigned short vec_splat_u16 (const int);
11665 vector unsigned int vec_splat_u32 (const int);
11667 vector signed char vec_sr (vector signed char, vector unsigned char);
11668 vector unsigned char vec_sr (vector unsigned char,
11669 vector unsigned char);
11670 vector signed short vec_sr (vector signed short,
11671 vector unsigned short);
11672 vector unsigned short vec_sr (vector unsigned short,
11673 vector unsigned short);
11674 vector signed int vec_sr (vector signed int, vector unsigned int);
11675 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
11677 vector signed int vec_vsrw (vector signed int, vector unsigned int);
11678 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
11680 vector signed short vec_vsrh (vector signed short,
11681 vector unsigned short);
11682 vector unsigned short vec_vsrh (vector unsigned short,
11683 vector unsigned short);
11685 vector signed char vec_vsrb (vector signed char, vector unsigned char);
11686 vector unsigned char vec_vsrb (vector unsigned char,
11687 vector unsigned char);
11689 vector signed char vec_sra (vector signed char, vector unsigned char);
11690 vector unsigned char vec_sra (vector unsigned char,
11691 vector unsigned char);
11692 vector signed short vec_sra (vector signed short,
11693 vector unsigned short);
11694 vector unsigned short vec_sra (vector unsigned short,
11695 vector unsigned short);
11696 vector signed int vec_sra (vector signed int, vector unsigned int);
11697 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
11699 vector signed int vec_vsraw (vector signed int, vector unsigned int);
11700 vector unsigned int vec_vsraw (vector unsigned int,
11701 vector unsigned int);
11703 vector signed short vec_vsrah (vector signed short,
11704 vector unsigned short);
11705 vector unsigned short vec_vsrah (vector unsigned short,
11706 vector unsigned short);
11708 vector signed char vec_vsrab (vector signed char, vector unsigned char);
11709 vector unsigned char vec_vsrab (vector unsigned char,
11710 vector unsigned char);
11712 vector signed int vec_srl (vector signed int, vector unsigned int);
11713 vector signed int vec_srl (vector signed int, vector unsigned short);
11714 vector signed int vec_srl (vector signed int, vector unsigned char);
11715 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
11716 vector unsigned int vec_srl (vector unsigned int,
11717 vector unsigned short);
11718 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
11719 vector bool int vec_srl (vector bool int, vector unsigned int);
11720 vector bool int vec_srl (vector bool int, vector unsigned short);
11721 vector bool int vec_srl (vector bool int, vector unsigned char);
11722 vector signed short vec_srl (vector signed short, vector unsigned int);
11723 vector signed short vec_srl (vector signed short,
11724 vector unsigned short);
11725 vector signed short vec_srl (vector signed short, vector unsigned char);
11726 vector unsigned short vec_srl (vector unsigned short,
11727 vector unsigned int);
11728 vector unsigned short vec_srl (vector unsigned short,
11729 vector unsigned short);
11730 vector unsigned short vec_srl (vector unsigned short,
11731 vector unsigned char);
11732 vector bool short vec_srl (vector bool short, vector unsigned int);
11733 vector bool short vec_srl (vector bool short, vector unsigned short);
11734 vector bool short vec_srl (vector bool short, vector unsigned char);
11735 vector pixel vec_srl (vector pixel, vector unsigned int);
11736 vector pixel vec_srl (vector pixel, vector unsigned short);
11737 vector pixel vec_srl (vector pixel, vector unsigned char);
11738 vector signed char vec_srl (vector signed char, vector unsigned int);
11739 vector signed char vec_srl (vector signed char, vector unsigned short);
11740 vector signed char vec_srl (vector signed char, vector unsigned char);
11741 vector unsigned char vec_srl (vector unsigned char,
11742 vector unsigned int);
11743 vector unsigned char vec_srl (vector unsigned char,
11744 vector unsigned short);
11745 vector unsigned char vec_srl (vector unsigned char,
11746 vector unsigned char);
11747 vector bool char vec_srl (vector bool char, vector unsigned int);
11748 vector bool char vec_srl (vector bool char, vector unsigned short);
11749 vector bool char vec_srl (vector bool char, vector unsigned char);
11751 vector float vec_sro (vector float, vector signed char);
11752 vector float vec_sro (vector float, vector unsigned char);
11753 vector signed int vec_sro (vector signed int, vector signed char);
11754 vector signed int vec_sro (vector signed int, vector unsigned char);
11755 vector unsigned int vec_sro (vector unsigned int, vector signed char);
11756 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
11757 vector signed short vec_sro (vector signed short, vector signed char);
11758 vector signed short vec_sro (vector signed short, vector unsigned char);
11759 vector unsigned short vec_sro (vector unsigned short,
11760 vector signed char);
11761 vector unsigned short vec_sro (vector unsigned short,
11762 vector unsigned char);
11763 vector pixel vec_sro (vector pixel, vector signed char);
11764 vector pixel vec_sro (vector pixel, vector unsigned char);
11765 vector signed char vec_sro (vector signed char, vector signed char);
11766 vector signed char vec_sro (vector signed char, vector unsigned char);
11767 vector unsigned char vec_sro (vector unsigned char, vector signed char);
11768 vector unsigned char vec_sro (vector unsigned char,
11769 vector unsigned char);
11771 void vec_st (vector float, int, vector float *);
11772 void vec_st (vector float, int, float *);
11773 void vec_st (vector signed int, int, vector signed int *);
11774 void vec_st (vector signed int, int, int *);
11775 void vec_st (vector unsigned int, int, vector unsigned int *);
11776 void vec_st (vector unsigned int, int, unsigned int *);
11777 void vec_st (vector bool int, int, vector bool int *);
11778 void vec_st (vector bool int, int, unsigned int *);
11779 void vec_st (vector bool int, int, int *);
11780 void vec_st (vector signed short, int, vector signed short *);
11781 void vec_st (vector signed short, int, short *);
11782 void vec_st (vector unsigned short, int, vector unsigned short *);
11783 void vec_st (vector unsigned short, int, unsigned short *);
11784 void vec_st (vector bool short, int, vector bool short *);
11785 void vec_st (vector bool short, int, unsigned short *);
11786 void vec_st (vector pixel, int, vector pixel *);
11787 void vec_st (vector pixel, int, unsigned short *);
11788 void vec_st (vector pixel, int, short *);
11789 void vec_st (vector bool short, int, short *);
11790 void vec_st (vector signed char, int, vector signed char *);
11791 void vec_st (vector signed char, int, signed char *);
11792 void vec_st (vector unsigned char, int, vector unsigned char *);
11793 void vec_st (vector unsigned char, int, unsigned char *);
11794 void vec_st (vector bool char, int, vector bool char *);
11795 void vec_st (vector bool char, int, unsigned char *);
11796 void vec_st (vector bool char, int, signed char *);
11798 void vec_ste (vector signed char, int, signed char *);
11799 void vec_ste (vector unsigned char, int, unsigned char *);
11800 void vec_ste (vector bool char, int, signed char *);
11801 void vec_ste (vector bool char, int, unsigned char *);
11802 void vec_ste (vector signed short, int, short *);
11803 void vec_ste (vector unsigned short, int, unsigned short *);
11804 void vec_ste (vector bool short, int, short *);
11805 void vec_ste (vector bool short, int, unsigned short *);
11806 void vec_ste (vector pixel, int, short *);
11807 void vec_ste (vector pixel, int, unsigned short *);
11808 void vec_ste (vector float, int, float *);
11809 void vec_ste (vector signed int, int, int *);
11810 void vec_ste (vector unsigned int, int, unsigned int *);
11811 void vec_ste (vector bool int, int, int *);
11812 void vec_ste (vector bool int, int, unsigned int *);
11814 void vec_stvewx (vector float, int, float *);
11815 void vec_stvewx (vector signed int, int, int *);
11816 void vec_stvewx (vector unsigned int, int, unsigned int *);
11817 void vec_stvewx (vector bool int, int, int *);
11818 void vec_stvewx (vector bool int, int, unsigned int *);
11820 void vec_stvehx (vector signed short, int, short *);
11821 void vec_stvehx (vector unsigned short, int, unsigned short *);
11822 void vec_stvehx (vector bool short, int, short *);
11823 void vec_stvehx (vector bool short, int, unsigned short *);
11824 void vec_stvehx (vector pixel, int, short *);
11825 void vec_stvehx (vector pixel, int, unsigned short *);
11827 void vec_stvebx (vector signed char, int, signed char *);
11828 void vec_stvebx (vector unsigned char, int, unsigned char *);
11829 void vec_stvebx (vector bool char, int, signed char *);
11830 void vec_stvebx (vector bool char, int, unsigned char *);
11832 void vec_stl (vector float, int, vector float *);
11833 void vec_stl (vector float, int, float *);
11834 void vec_stl (vector signed int, int, vector signed int *);
11835 void vec_stl (vector signed int, int, int *);
11836 void vec_stl (vector unsigned int, int, vector unsigned int *);
11837 void vec_stl (vector unsigned int, int, unsigned int *);
11838 void vec_stl (vector bool int, int, vector bool int *);
11839 void vec_stl (vector bool int, int, unsigned int *);
11840 void vec_stl (vector bool int, int, int *);
11841 void vec_stl (vector signed short, int, vector signed short *);
11842 void vec_stl (vector signed short, int, short *);
11843 void vec_stl (vector unsigned short, int, vector unsigned short *);
11844 void vec_stl (vector unsigned short, int, unsigned short *);
11845 void vec_stl (vector bool short, int, vector bool short *);
11846 void vec_stl (vector bool short, int, unsigned short *);
11847 void vec_stl (vector bool short, int, short *);
11848 void vec_stl (vector pixel, int, vector pixel *);
11849 void vec_stl (vector pixel, int, unsigned short *);
11850 void vec_stl (vector pixel, int, short *);
11851 void vec_stl (vector signed char, int, vector signed char *);
11852 void vec_stl (vector signed char, int, signed char *);
11853 void vec_stl (vector unsigned char, int, vector unsigned char *);
11854 void vec_stl (vector unsigned char, int, unsigned char *);
11855 void vec_stl (vector bool char, int, vector bool char *);
11856 void vec_stl (vector bool char, int, unsigned char *);
11857 void vec_stl (vector bool char, int, signed char *);
11859 vector signed char vec_sub (vector bool char, vector signed char);
11860 vector signed char vec_sub (vector signed char, vector bool char);
11861 vector signed char vec_sub (vector signed char, vector signed char);
11862 vector unsigned char vec_sub (vector bool char, vector unsigned char);
11863 vector unsigned char vec_sub (vector unsigned char, vector bool char);
11864 vector unsigned char vec_sub (vector unsigned char,
11865 vector unsigned char);
11866 vector signed short vec_sub (vector bool short, vector signed short);
11867 vector signed short vec_sub (vector signed short, vector bool short);
11868 vector signed short vec_sub (vector signed short, vector signed short);
11869 vector unsigned short vec_sub (vector bool short,
11870 vector unsigned short);
11871 vector unsigned short vec_sub (vector unsigned short,
11872 vector bool short);
11873 vector unsigned short vec_sub (vector unsigned short,
11874 vector unsigned short);
11875 vector signed int vec_sub (vector bool int, vector signed int);
11876 vector signed int vec_sub (vector signed int, vector bool int);
11877 vector signed int vec_sub (vector signed int, vector signed int);
11878 vector unsigned int vec_sub (vector bool int, vector unsigned int);
11879 vector unsigned int vec_sub (vector unsigned int, vector bool int);
11880 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
11881 vector float vec_sub (vector float, vector float);
11883 vector float vec_vsubfp (vector float, vector float);
11885 vector signed int vec_vsubuwm (vector bool int, vector signed int);
11886 vector signed int vec_vsubuwm (vector signed int, vector bool int);
11887 vector signed int vec_vsubuwm (vector signed int, vector signed int);
11888 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
11889 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
11890 vector unsigned int vec_vsubuwm (vector unsigned int,
11891 vector unsigned int);
11893 vector signed short vec_vsubuhm (vector bool short,
11894 vector signed short);
11895 vector signed short vec_vsubuhm (vector signed short,
11896 vector bool short);
11897 vector signed short vec_vsubuhm (vector signed short,
11898 vector signed short);
11899 vector unsigned short vec_vsubuhm (vector bool short,
11900 vector unsigned short);
11901 vector unsigned short vec_vsubuhm (vector unsigned short,
11902 vector bool short);
11903 vector unsigned short vec_vsubuhm (vector unsigned short,
11904 vector unsigned short);
11906 vector signed char vec_vsububm (vector bool char, vector signed char);
11907 vector signed char vec_vsububm (vector signed char, vector bool char);
11908 vector signed char vec_vsububm (vector signed char, vector signed char);
11909 vector unsigned char vec_vsububm (vector bool char,
11910 vector unsigned char);
11911 vector unsigned char vec_vsububm (vector unsigned char,
11913 vector unsigned char vec_vsububm (vector unsigned char,
11914 vector unsigned char);
11916 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
11918 vector unsigned char vec_subs (vector bool char, vector unsigned char);
11919 vector unsigned char vec_subs (vector unsigned char, vector bool char);
11920 vector unsigned char vec_subs (vector unsigned char,
11921 vector unsigned char);
11922 vector signed char vec_subs (vector bool char, vector signed char);
11923 vector signed char vec_subs (vector signed char, vector bool char);
11924 vector signed char vec_subs (vector signed char, vector signed char);
11925 vector unsigned short vec_subs (vector bool short,
11926 vector unsigned short);
11927 vector unsigned short vec_subs (vector unsigned short,
11928 vector bool short);
11929 vector unsigned short vec_subs (vector unsigned short,
11930 vector unsigned short);
11931 vector signed short vec_subs (vector bool short, vector signed short);
11932 vector signed short vec_subs (vector signed short, vector bool short);
11933 vector signed short vec_subs (vector signed short, vector signed short);
11934 vector unsigned int vec_subs (vector bool int, vector unsigned int);
11935 vector unsigned int vec_subs (vector unsigned int, vector bool int);
11936 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
11937 vector signed int vec_subs (vector bool int, vector signed int);
11938 vector signed int vec_subs (vector signed int, vector bool int);
11939 vector signed int vec_subs (vector signed int, vector signed int);
11941 vector signed int vec_vsubsws (vector bool int, vector signed int);
11942 vector signed int vec_vsubsws (vector signed int, vector bool int);
11943 vector signed int vec_vsubsws (vector signed int, vector signed int);
11945 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
11946 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
11947 vector unsigned int vec_vsubuws (vector unsigned int,
11948 vector unsigned int);
11950 vector signed short vec_vsubshs (vector bool short,
11951 vector signed short);
11952 vector signed short vec_vsubshs (vector signed short,
11953 vector bool short);
11954 vector signed short vec_vsubshs (vector signed short,
11955 vector signed short);
11957 vector unsigned short vec_vsubuhs (vector bool short,
11958 vector unsigned short);
11959 vector unsigned short vec_vsubuhs (vector unsigned short,
11960 vector bool short);
11961 vector unsigned short vec_vsubuhs (vector unsigned short,
11962 vector unsigned short);
11964 vector signed char vec_vsubsbs (vector bool char, vector signed char);
11965 vector signed char vec_vsubsbs (vector signed char, vector bool char);
11966 vector signed char vec_vsubsbs (vector signed char, vector signed char);
11968 vector unsigned char vec_vsububs (vector bool char,
11969 vector unsigned char);
11970 vector unsigned char vec_vsububs (vector unsigned char,
11972 vector unsigned char vec_vsububs (vector unsigned char,
11973 vector unsigned char);
11975 vector unsigned int vec_sum4s (vector unsigned char,
11976 vector unsigned int);
11977 vector signed int vec_sum4s (vector signed char, vector signed int);
11978 vector signed int vec_sum4s (vector signed short, vector signed int);
11980 vector signed int vec_vsum4shs (vector signed short, vector signed int);
11982 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
11984 vector unsigned int vec_vsum4ubs (vector unsigned char,
11985 vector unsigned int);
11987 vector signed int vec_sum2s (vector signed int, vector signed int);
11989 vector signed int vec_sums (vector signed int, vector signed int);
11991 vector float vec_trunc (vector float);
11993 vector signed short vec_unpackh (vector signed char);
11994 vector bool short vec_unpackh (vector bool char);
11995 vector signed int vec_unpackh (vector signed short);
11996 vector bool int vec_unpackh (vector bool short);
11997 vector unsigned int vec_unpackh (vector pixel);
11999 vector bool int vec_vupkhsh (vector bool short);
12000 vector signed int vec_vupkhsh (vector signed short);
12002 vector unsigned int vec_vupkhpx (vector pixel);
12004 vector bool short vec_vupkhsb (vector bool char);
12005 vector signed short vec_vupkhsb (vector signed char);
12007 vector signed short vec_unpackl (vector signed char);
12008 vector bool short vec_unpackl (vector bool char);
12009 vector unsigned int vec_unpackl (vector pixel);
12010 vector signed int vec_unpackl (vector signed short);
12011 vector bool int vec_unpackl (vector bool short);
12013 vector unsigned int vec_vupklpx (vector pixel);
12015 vector bool int vec_vupklsh (vector bool short);
12016 vector signed int vec_vupklsh (vector signed short);
12018 vector bool short vec_vupklsb (vector bool char);
12019 vector signed short vec_vupklsb (vector signed char);
12021 vector float vec_xor (vector float, vector float);
12022 vector float vec_xor (vector float, vector bool int);
12023 vector float vec_xor (vector bool int, vector float);
12024 vector bool int vec_xor (vector bool int, vector bool int);
12025 vector signed int vec_xor (vector bool int, vector signed int);
12026 vector signed int vec_xor (vector signed int, vector bool int);
12027 vector signed int vec_xor (vector signed int, vector signed int);
12028 vector unsigned int vec_xor (vector bool int, vector unsigned int);
12029 vector unsigned int vec_xor (vector unsigned int, vector bool int);
12030 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
12031 vector bool short vec_xor (vector bool short, vector bool short);
12032 vector signed short vec_xor (vector bool short, vector signed short);
12033 vector signed short vec_xor (vector signed short, vector bool short);
12034 vector signed short vec_xor (vector signed short, vector signed short);
12035 vector unsigned short vec_xor (vector bool short,
12036 vector unsigned short);
12037 vector unsigned short vec_xor (vector unsigned short,
12038 vector bool short);
12039 vector unsigned short vec_xor (vector unsigned short,
12040 vector unsigned short);
12041 vector signed char vec_xor (vector bool char, vector signed char);
12042 vector bool char vec_xor (vector bool char, vector bool char);
12043 vector signed char vec_xor (vector signed char, vector bool char);
12044 vector signed char vec_xor (vector signed char, vector signed char);
12045 vector unsigned char vec_xor (vector bool char, vector unsigned char);
12046 vector unsigned char vec_xor (vector unsigned char, vector bool char);
12047 vector unsigned char vec_xor (vector unsigned char,
12048 vector unsigned char);
12050 int vec_all_eq (vector signed char, vector bool char);
12051 int vec_all_eq (vector signed char, vector signed char);
12052 int vec_all_eq (vector unsigned char, vector bool char);
12053 int vec_all_eq (vector unsigned char, vector unsigned char);
12054 int vec_all_eq (vector bool char, vector bool char);
12055 int vec_all_eq (vector bool char, vector unsigned char);
12056 int vec_all_eq (vector bool char, vector signed char);
12057 int vec_all_eq (vector signed short, vector bool short);
12058 int vec_all_eq (vector signed short, vector signed short);
12059 int vec_all_eq (vector unsigned short, vector bool short);
12060 int vec_all_eq (vector unsigned short, vector unsigned short);
12061 int vec_all_eq (vector bool short, vector bool short);
12062 int vec_all_eq (vector bool short, vector unsigned short);
12063 int vec_all_eq (vector bool short, vector signed short);
12064 int vec_all_eq (vector pixel, vector pixel);
12065 int vec_all_eq (vector signed int, vector bool int);
12066 int vec_all_eq (vector signed int, vector signed int);
12067 int vec_all_eq (vector unsigned int, vector bool int);
12068 int vec_all_eq (vector unsigned int, vector unsigned int);
12069 int vec_all_eq (vector bool int, vector bool int);
12070 int vec_all_eq (vector bool int, vector unsigned int);
12071 int vec_all_eq (vector bool int, vector signed int);
12072 int vec_all_eq (vector float, vector float);
12074 int vec_all_ge (vector bool char, vector unsigned char);
12075 int vec_all_ge (vector unsigned char, vector bool char);
12076 int vec_all_ge (vector unsigned char, vector unsigned char);
12077 int vec_all_ge (vector bool char, vector signed char);
12078 int vec_all_ge (vector signed char, vector bool char);
12079 int vec_all_ge (vector signed char, vector signed char);
12080 int vec_all_ge (vector bool short, vector unsigned short);
12081 int vec_all_ge (vector unsigned short, vector bool short);
12082 int vec_all_ge (vector unsigned short, vector unsigned short);
12083 int vec_all_ge (vector signed short, vector signed short);
12084 int vec_all_ge (vector bool short, vector signed short);
12085 int vec_all_ge (vector signed short, vector bool short);
12086 int vec_all_ge (vector bool int, vector unsigned int);
12087 int vec_all_ge (vector unsigned int, vector bool int);
12088 int vec_all_ge (vector unsigned int, vector unsigned int);
12089 int vec_all_ge (vector bool int, vector signed int);
12090 int vec_all_ge (vector signed int, vector bool int);
12091 int vec_all_ge (vector signed int, vector signed int);
12092 int vec_all_ge (vector float, vector float);
12094 int vec_all_gt (vector bool char, vector unsigned char);
12095 int vec_all_gt (vector unsigned char, vector bool char);
12096 int vec_all_gt (vector unsigned char, vector unsigned char);
12097 int vec_all_gt (vector bool char, vector signed char);
12098 int vec_all_gt (vector signed char, vector bool char);
12099 int vec_all_gt (vector signed char, vector signed char);
12100 int vec_all_gt (vector bool short, vector unsigned short);
12101 int vec_all_gt (vector unsigned short, vector bool short);
12102 int vec_all_gt (vector unsigned short, vector unsigned short);
12103 int vec_all_gt (vector bool short, vector signed short);
12104 int vec_all_gt (vector signed short, vector bool short);
12105 int vec_all_gt (vector signed short, vector signed short);
12106 int vec_all_gt (vector bool int, vector unsigned int);
12107 int vec_all_gt (vector unsigned int, vector bool int);
12108 int vec_all_gt (vector unsigned int, vector unsigned int);
12109 int vec_all_gt (vector bool int, vector signed int);
12110 int vec_all_gt (vector signed int, vector bool int);
12111 int vec_all_gt (vector signed int, vector signed int);
12112 int vec_all_gt (vector float, vector float);
12114 int vec_all_in (vector float, vector float);
12116 int vec_all_le (vector bool char, vector unsigned char);
12117 int vec_all_le (vector unsigned char, vector bool char);
12118 int vec_all_le (vector unsigned char, vector unsigned char);
12119 int vec_all_le (vector bool char, vector signed char);
12120 int vec_all_le (vector signed char, vector bool char);
12121 int vec_all_le (vector signed char, vector signed char);
12122 int vec_all_le (vector bool short, vector unsigned short);
12123 int vec_all_le (vector unsigned short, vector bool short);
12124 int vec_all_le (vector unsigned short, vector unsigned short);
12125 int vec_all_le (vector bool short, vector signed short);
12126 int vec_all_le (vector signed short, vector bool short);
12127 int vec_all_le (vector signed short, vector signed short);
12128 int vec_all_le (vector bool int, vector unsigned int);
12129 int vec_all_le (vector unsigned int, vector bool int);
12130 int vec_all_le (vector unsigned int, vector unsigned int);
12131 int vec_all_le (vector bool int, vector signed int);
12132 int vec_all_le (vector signed int, vector bool int);
12133 int vec_all_le (vector signed int, vector signed int);
12134 int vec_all_le (vector float, vector float);
12136 int vec_all_lt (vector bool char, vector unsigned char);
12137 int vec_all_lt (vector unsigned char, vector bool char);
12138 int vec_all_lt (vector unsigned char, vector unsigned char);
12139 int vec_all_lt (vector bool char, vector signed char);
12140 int vec_all_lt (vector signed char, vector bool char);
12141 int vec_all_lt (vector signed char, vector signed char);
12142 int vec_all_lt (vector bool short, vector unsigned short);
12143 int vec_all_lt (vector unsigned short, vector bool short);
12144 int vec_all_lt (vector unsigned short, vector unsigned short);
12145 int vec_all_lt (vector bool short, vector signed short);
12146 int vec_all_lt (vector signed short, vector bool short);
12147 int vec_all_lt (vector signed short, vector signed short);
12148 int vec_all_lt (vector bool int, vector unsigned int);
12149 int vec_all_lt (vector unsigned int, vector bool int);
12150 int vec_all_lt (vector unsigned int, vector unsigned int);
12151 int vec_all_lt (vector bool int, vector signed int);
12152 int vec_all_lt (vector signed int, vector bool int);
12153 int vec_all_lt (vector signed int, vector signed int);
12154 int vec_all_lt (vector float, vector float);
12156 int vec_all_nan (vector float);
12158 int vec_all_ne (vector signed char, vector bool char);
12159 int vec_all_ne (vector signed char, vector signed char);
12160 int vec_all_ne (vector unsigned char, vector bool char);
12161 int vec_all_ne (vector unsigned char, vector unsigned char);
12162 int vec_all_ne (vector bool char, vector bool char);
12163 int vec_all_ne (vector bool char, vector unsigned char);
12164 int vec_all_ne (vector bool char, vector signed char);
12165 int vec_all_ne (vector signed short, vector bool short);
12166 int vec_all_ne (vector signed short, vector signed short);
12167 int vec_all_ne (vector unsigned short, vector bool short);
12168 int vec_all_ne (vector unsigned short, vector unsigned short);
12169 int vec_all_ne (vector bool short, vector bool short);
12170 int vec_all_ne (vector bool short, vector unsigned short);
12171 int vec_all_ne (vector bool short, vector signed short);
12172 int vec_all_ne (vector pixel, vector pixel);
12173 int vec_all_ne (vector signed int, vector bool int);
12174 int vec_all_ne (vector signed int, vector signed int);
12175 int vec_all_ne (vector unsigned int, vector bool int);
12176 int vec_all_ne (vector unsigned int, vector unsigned int);
12177 int vec_all_ne (vector bool int, vector bool int);
12178 int vec_all_ne (vector bool int, vector unsigned int);
12179 int vec_all_ne (vector bool int, vector signed int);
12180 int vec_all_ne (vector float, vector float);
12182 int vec_all_nge (vector float, vector float);
12184 int vec_all_ngt (vector float, vector float);
12186 int vec_all_nle (vector float, vector float);
12188 int vec_all_nlt (vector float, vector float);
12190 int vec_all_numeric (vector float);
12192 int vec_any_eq (vector signed char, vector bool char);
12193 int vec_any_eq (vector signed char, vector signed char);
12194 int vec_any_eq (vector unsigned char, vector bool char);
12195 int vec_any_eq (vector unsigned char, vector unsigned char);
12196 int vec_any_eq (vector bool char, vector bool char);
12197 int vec_any_eq (vector bool char, vector unsigned char);
12198 int vec_any_eq (vector bool char, vector signed char);
12199 int vec_any_eq (vector signed short, vector bool short);
12200 int vec_any_eq (vector signed short, vector signed short);
12201 int vec_any_eq (vector unsigned short, vector bool short);
12202 int vec_any_eq (vector unsigned short, vector unsigned short);
12203 int vec_any_eq (vector bool short, vector bool short);
12204 int vec_any_eq (vector bool short, vector unsigned short);
12205 int vec_any_eq (vector bool short, vector signed short);
12206 int vec_any_eq (vector pixel, vector pixel);
12207 int vec_any_eq (vector signed int, vector bool int);
12208 int vec_any_eq (vector signed int, vector signed int);
12209 int vec_any_eq (vector unsigned int, vector bool int);
12210 int vec_any_eq (vector unsigned int, vector unsigned int);
12211 int vec_any_eq (vector bool int, vector bool int);
12212 int vec_any_eq (vector bool int, vector unsigned int);
12213 int vec_any_eq (vector bool int, vector signed int);
12214 int vec_any_eq (vector float, vector float);
12216 int vec_any_ge (vector signed char, vector bool char);
12217 int vec_any_ge (vector unsigned char, vector bool char);
12218 int vec_any_ge (vector unsigned char, vector unsigned char);
12219 int vec_any_ge (vector signed char, vector signed char);
12220 int vec_any_ge (vector bool char, vector unsigned char);
12221 int vec_any_ge (vector bool char, vector signed char);
12222 int vec_any_ge (vector unsigned short, vector bool short);
12223 int vec_any_ge (vector unsigned short, vector unsigned short);
12224 int vec_any_ge (vector signed short, vector signed short);
12225 int vec_any_ge (vector signed short, vector bool short);
12226 int vec_any_ge (vector bool short, vector unsigned short);
12227 int vec_any_ge (vector bool short, vector signed short);
12228 int vec_any_ge (vector signed int, vector bool int);
12229 int vec_any_ge (vector unsigned int, vector bool int);
12230 int vec_any_ge (vector unsigned int, vector unsigned int);
12231 int vec_any_ge (vector signed int, vector signed int);
12232 int vec_any_ge (vector bool int, vector unsigned int);
12233 int vec_any_ge (vector bool int, vector signed int);
12234 int vec_any_ge (vector float, vector float);
12236 int vec_any_gt (vector bool char, vector unsigned char);
12237 int vec_any_gt (vector unsigned char, vector bool char);
12238 int vec_any_gt (vector unsigned char, vector unsigned char);
12239 int vec_any_gt (vector bool char, vector signed char);
12240 int vec_any_gt (vector signed char, vector bool char);
12241 int vec_any_gt (vector signed char, vector signed char);
12242 int vec_any_gt (vector bool short, vector unsigned short);
12243 int vec_any_gt (vector unsigned short, vector bool short);
12244 int vec_any_gt (vector unsigned short, vector unsigned short);
12245 int vec_any_gt (vector bool short, vector signed short);
12246 int vec_any_gt (vector signed short, vector bool short);
12247 int vec_any_gt (vector signed short, vector signed short);
12248 int vec_any_gt (vector bool int, vector unsigned int);
12249 int vec_any_gt (vector unsigned int, vector bool int);
12250 int vec_any_gt (vector unsigned int, vector unsigned int);
12251 int vec_any_gt (vector bool int, vector signed int);
12252 int vec_any_gt (vector signed int, vector bool int);
12253 int vec_any_gt (vector signed int, vector signed int);
12254 int vec_any_gt (vector float, vector float);
12256 int vec_any_le (vector bool char, vector unsigned char);
12257 int vec_any_le (vector unsigned char, vector bool char);
12258 int vec_any_le (vector unsigned char, vector unsigned char);
12259 int vec_any_le (vector bool char, vector signed char);
12260 int vec_any_le (vector signed char, vector bool char);
12261 int vec_any_le (vector signed char, vector signed char);
12262 int vec_any_le (vector bool short, vector unsigned short);
12263 int vec_any_le (vector unsigned short, vector bool short);
12264 int vec_any_le (vector unsigned short, vector unsigned short);
12265 int vec_any_le (vector bool short, vector signed short);
12266 int vec_any_le (vector signed short, vector bool short);
12267 int vec_any_le (vector signed short, vector signed short);
12268 int vec_any_le (vector bool int, vector unsigned int);
12269 int vec_any_le (vector unsigned int, vector bool int);
12270 int vec_any_le (vector unsigned int, vector unsigned int);
12271 int vec_any_le (vector bool int, vector signed int);
12272 int vec_any_le (vector signed int, vector bool int);
12273 int vec_any_le (vector signed int, vector signed int);
12274 int vec_any_le (vector float, vector float);
12276 int vec_any_lt (vector bool char, vector unsigned char);
12277 int vec_any_lt (vector unsigned char, vector bool char);
12278 int vec_any_lt (vector unsigned char, vector unsigned char);
12279 int vec_any_lt (vector bool char, vector signed char);
12280 int vec_any_lt (vector signed char, vector bool char);
12281 int vec_any_lt (vector signed char, vector signed char);
12282 int vec_any_lt (vector bool short, vector unsigned short);
12283 int vec_any_lt (vector unsigned short, vector bool short);
12284 int vec_any_lt (vector unsigned short, vector unsigned short);
12285 int vec_any_lt (vector bool short, vector signed short);
12286 int vec_any_lt (vector signed short, vector bool short);
12287 int vec_any_lt (vector signed short, vector signed short);
12288 int vec_any_lt (vector bool int, vector unsigned int);
12289 int vec_any_lt (vector unsigned int, vector bool int);
12290 int vec_any_lt (vector unsigned int, vector unsigned int);
12291 int vec_any_lt (vector bool int, vector signed int);
12292 int vec_any_lt (vector signed int, vector bool int);
12293 int vec_any_lt (vector signed int, vector signed int);
12294 int vec_any_lt (vector float, vector float);
12296 int vec_any_nan (vector float);
12298 int vec_any_ne (vector signed char, vector bool char);
12299 int vec_any_ne (vector signed char, vector signed char);
12300 int vec_any_ne (vector unsigned char, vector bool char);
12301 int vec_any_ne (vector unsigned char, vector unsigned char);
12302 int vec_any_ne (vector bool char, vector bool char);
12303 int vec_any_ne (vector bool char, vector unsigned char);
12304 int vec_any_ne (vector bool char, vector signed char);
12305 int vec_any_ne (vector signed short, vector bool short);
12306 int vec_any_ne (vector signed short, vector signed short);
12307 int vec_any_ne (vector unsigned short, vector bool short);
12308 int vec_any_ne (vector unsigned short, vector unsigned short);
12309 int vec_any_ne (vector bool short, vector bool short);
12310 int vec_any_ne (vector bool short, vector unsigned short);
12311 int vec_any_ne (vector bool short, vector signed short);
12312 int vec_any_ne (vector pixel, vector pixel);
12313 int vec_any_ne (vector signed int, vector bool int);
12314 int vec_any_ne (vector signed int, vector signed int);
12315 int vec_any_ne (vector unsigned int, vector bool int);
12316 int vec_any_ne (vector unsigned int, vector unsigned int);
12317 int vec_any_ne (vector bool int, vector bool int);
12318 int vec_any_ne (vector bool int, vector unsigned int);
12319 int vec_any_ne (vector bool int, vector signed int);
12320 int vec_any_ne (vector float, vector float);
12322 int vec_any_nge (vector float, vector float);
12324 int vec_any_ngt (vector float, vector float);
12326 int vec_any_nle (vector float, vector float);
12328 int vec_any_nlt (vector float, vector float);
12330 int vec_any_numeric (vector float);
12332 int vec_any_out (vector float, vector float);
12335 If the vector/scalar (VSX) instruction set is available, the following
12336 additional functions are available:
12339 vector double vec_abs (vector double);
12340 vector double vec_add (vector double, vector double);
12341 vector double vec_and (vector double, vector double);
12342 vector double vec_and (vector double, vector bool long);
12343 vector double vec_and (vector bool long, vector double);
12344 vector double vec_andc (vector double, vector double);
12345 vector double vec_andc (vector double, vector bool long);
12346 vector double vec_andc (vector bool long, vector double);
12347 vector double vec_ceil (vector double);
12348 vector bool long vec_cmpeq (vector double, vector double);
12349 vector bool long vec_cmpge (vector double, vector double);
12350 vector bool long vec_cmpgt (vector double, vector double);
12351 vector bool long vec_cmple (vector double, vector double);
12352 vector bool long vec_cmplt (vector double, vector double);
12353 vector float vec_div (vector float, vector float);
12354 vector double vec_div (vector double, vector double);
12355 vector double vec_floor (vector double);
12356 vector double vec_madd (vector double, vector double, vector double);
12357 vector double vec_max (vector double, vector double);
12358 vector double vec_min (vector double, vector double);
12359 vector float vec_msub (vector float, vector float, vector float);
12360 vector double vec_msub (vector double, vector double, vector double);
12361 vector float vec_mul (vector float, vector float);
12362 vector double vec_mul (vector double, vector double);
12363 vector float vec_nearbyint (vector float);
12364 vector double vec_nearbyint (vector double);
12365 vector float vec_nmadd (vector float, vector float, vector float);
12366 vector double vec_nmadd (vector double, vector double, vector double);
12367 vector double vec_nmsub (vector double, vector double, vector double);
12368 vector double vec_nor (vector double, vector double);
12369 vector double vec_or (vector double, vector double);
12370 vector double vec_or (vector double, vector bool long);
12371 vector double vec_or (vector bool long, vector double);
12372 vector double vec_perm (vector double,
12374 vector unsigned char);
12375 vector double vec_rint (vector double);
12376 vector double vec_recip (vector double, vector double);
12377 vector double vec_rsqrt (vector double);
12378 vector double vec_rsqrte (vector double);
12379 vector double vec_sel (vector double, vector double, vector bool long);
12380 vector double vec_sel (vector double, vector double, vector unsigned long);
12381 vector double vec_sub (vector double, vector double);
12382 vector float vec_sqrt (vector float);
12383 vector double vec_sqrt (vector double);
12384 vector double vec_trunc (vector double);
12385 vector double vec_xor (vector double, vector double);
12386 vector double vec_xor (vector double, vector bool long);
12387 vector double vec_xor (vector bool long, vector double);
12388 int vec_all_eq (vector double, vector double);
12389 int vec_all_ge (vector double, vector double);
12390 int vec_all_gt (vector double, vector double);
12391 int vec_all_le (vector double, vector double);
12392 int vec_all_lt (vector double, vector double);
12393 int vec_all_nan (vector double);
12394 int vec_all_ne (vector double, vector double);
12395 int vec_all_nge (vector double, vector double);
12396 int vec_all_ngt (vector double, vector double);
12397 int vec_all_nle (vector double, vector double);
12398 int vec_all_nlt (vector double, vector double);
12399 int vec_all_numeric (vector double);
12400 int vec_any_eq (vector double, vector double);
12401 int vec_any_ge (vector double, vector double);
12402 int vec_any_gt (vector double, vector double);
12403 int vec_any_le (vector double, vector double);
12404 int vec_any_lt (vector double, vector double);
12405 int vec_any_nan (vector double);
12406 int vec_any_ne (vector double, vector double);
12407 int vec_any_nge (vector double, vector double);
12408 int vec_any_ngt (vector double, vector double);
12409 int vec_any_nle (vector double, vector double);
12410 int vec_any_nlt (vector double, vector double);
12411 int vec_any_numeric (vector double);
12414 GCC provides a few other builtins on Powerpc to access certain instructions:
12416 float __builtin_recipdivf (float, float);
12417 float __builtin_rsqrtf (float);
12418 double __builtin_recipdiv (double, double);
12419 double __builtin_rsqrt (double);
12420 long __builtin_bpermd (long, long);
12421 int __builtin_bswap16 (int);
12424 The @code{vec_rsqrt}, @code{__builtin_rsqrt}, and
12425 @code{__builtin_rsqrtf} functions generate multiple instructions to
12426 implement the reciprocal sqrt functionality using reciprocal sqrt
12427 estimate instructions.
12429 The @code{__builtin_recipdiv}, and @code{__builtin_recipdivf}
12430 functions generate multiple instructions to implement division using
12431 the reciprocal estimate instructions.
12433 @node RX Built-in Functions
12434 @subsection RX Built-in Functions
12435 GCC supports some of the RX instructions which cannot be expressed in
12436 the C programming language via the use of built-in functions. The
12437 following functions are supported:
12439 @deftypefn {Built-in Function} void __builtin_rx_brk (void)
12440 Generates the @code{brk} machine instruction.
12443 @deftypefn {Built-in Function} void __builtin_rx_clrpsw (int)
12444 Generates the @code{clrpsw} machine instruction to clear the specified
12445 bit in the processor status word.
12448 @deftypefn {Built-in Function} void __builtin_rx_int (int)
12449 Generates the @code{int} machine instruction to generate an interrupt
12450 with the specified value.
12453 @deftypefn {Built-in Function} void __builtin_rx_machi (int, int)
12454 Generates the @code{machi} machine instruction to add the result of
12455 multiplying the top 16-bits of the two arguments into the
12459 @deftypefn {Built-in Function} void __builtin_rx_maclo (int, int)
12460 Generates the @code{maclo} machine instruction to add the result of
12461 multiplying the bottom 16-bits of the two arguments into the
12465 @deftypefn {Built-in Function} void __builtin_rx_mulhi (int, int)
12466 Generates the @code{mulhi} machine instruction to place the result of
12467 multiplying the top 16-bits of the two arguments into the
12471 @deftypefn {Built-in Function} void __builtin_rx_mullo (int, int)
12472 Generates the @code{mullo} machine instruction to place the result of
12473 multiplying the bottom 16-bits of the two arguments into the
12477 @deftypefn {Built-in Function} int __builtin_rx_mvfachi (void)
12478 Generates the @code{mvfachi} machine instruction to read the top
12479 32-bits of the accumulator.
12482 @deftypefn {Built-in Function} int __builtin_rx_mvfacmi (void)
12483 Generates the @code{mvfacmi} machine instruction to read the middle
12484 32-bits of the accumulator.
12487 @deftypefn {Built-in Function} int __builtin_rx_mvfc (int)
12488 Generates the @code{mvfc} machine instruction which reads the control
12489 register specified in its argument and returns its value.
12492 @deftypefn {Built-in Function} void __builtin_rx_mvtachi (int)
12493 Generates the @code{mvtachi} machine instruction to set the top
12494 32-bits of the accumulator.
12497 @deftypefn {Built-in Function} void __builtin_rx_mvtaclo (int)
12498 Generates the @code{mvtaclo} machine instruction to set the bottom
12499 32-bits of the accumulator.
12502 @deftypefn {Built-in Function} void __builtin_rx_mvtc (int reg, int val)
12503 Generates the @code{mvtc} machine instruction which sets control
12504 register number @code{reg} to @code{val}.
12507 @deftypefn {Built-in Function} void __builtin_rx_mvtipl (int)
12508 Generates the @code{mvtipl} machine instruction set the interrupt
12512 @deftypefn {Built-in Function} void __builtin_rx_racw (int)
12513 Generates the @code{racw} machine instruction to round the accumulator
12514 according to the specified mode.
12517 @deftypefn {Built-in Function} int __builtin_rx_revw (int)
12518 Generates the @code{revw} machine instruction which swaps the bytes in
12519 the argument so that bits 0--7 now occupy bits 8--15 and vice versa,
12520 and also bits 16--23 occupy bits 24--31 and vice versa.
12523 @deftypefn {Built-in Function} void __builtin_rx_rmpa (void)
12524 Generates the @code{rmpa} machine instruction which initiates a
12525 repeated multiply and accumulate sequence.
12528 @deftypefn {Built-in Function} void __builtin_rx_round (float)
12529 Generates the @code{round} machine instruction which returns the
12530 floating point argument rounded according to the current rounding mode
12531 set in the floating point status word register.
12534 @deftypefn {Built-in Function} int __builtin_rx_sat (int)
12535 Generates the @code{sat} machine instruction which returns the
12536 saturated value of the argument.
12539 @deftypefn {Built-in Function} void __builtin_rx_setpsw (int)
12540 Generates the @code{setpsw} machine instruction to set the specified
12541 bit in the processor status word.
12544 @deftypefn {Built-in Function} void __builtin_rx_wait (void)
12545 Generates the @code{wait} machine instruction.
12548 @node SPARC VIS Built-in Functions
12549 @subsection SPARC VIS Built-in Functions
12551 GCC supports SIMD operations on the SPARC using both the generic vector
12552 extensions (@pxref{Vector Extensions}) as well as built-in functions for
12553 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
12554 switch, the VIS extension is exposed as the following built-in functions:
12557 typedef int v2si __attribute__ ((vector_size (8)));
12558 typedef short v4hi __attribute__ ((vector_size (8)));
12559 typedef short v2hi __attribute__ ((vector_size (4)));
12560 typedef char v8qi __attribute__ ((vector_size (8)));
12561 typedef char v4qi __attribute__ ((vector_size (4)));
12563 void * __builtin_vis_alignaddr (void *, long);
12564 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
12565 v2si __builtin_vis_faligndatav2si (v2si, v2si);
12566 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
12567 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
12569 v4hi __builtin_vis_fexpand (v4qi);
12571 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
12572 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
12573 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
12574 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
12575 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
12576 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
12577 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
12579 v4qi __builtin_vis_fpack16 (v4hi);
12580 v8qi __builtin_vis_fpack32 (v2si, v2si);
12581 v2hi __builtin_vis_fpackfix (v2si);
12582 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
12584 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
12587 @node SPU Built-in Functions
12588 @subsection SPU Built-in Functions
12590 GCC provides extensions for the SPU processor as described in the
12591 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
12592 found at @uref{http://cell.scei.co.jp/} or
12593 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
12594 implementation differs in several ways.
12599 The optional extension of specifying vector constants in parentheses is
12603 A vector initializer requires no cast if the vector constant is of the
12604 same type as the variable it is initializing.
12607 If @code{signed} or @code{unsigned} is omitted, the signedness of the
12608 vector type is the default signedness of the base type. The default
12609 varies depending on the operating system, so a portable program should
12610 always specify the signedness.
12613 By default, the keyword @code{__vector} is added. The macro
12614 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
12618 GCC allows using a @code{typedef} name as the type specifier for a
12622 For C, overloaded functions are implemented with macros so the following
12626 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
12629 Since @code{spu_add} is a macro, the vector constant in the example
12630 is treated as four separate arguments. Wrap the entire argument in
12631 parentheses for this to work.
12634 The extended version of @code{__builtin_expect} is not supported.
12638 @emph{Note:} Only the interface described in the aforementioned
12639 specification is supported. Internally, GCC uses built-in functions to
12640 implement the required functionality, but these are not supported and
12641 are subject to change without notice.
12643 @node Target Format Checks
12644 @section Format Checks Specific to Particular Target Machines
12646 For some target machines, GCC supports additional options to the
12648 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
12651 * Solaris Format Checks::
12652 * Darwin Format Checks::
12655 @node Solaris Format Checks
12656 @subsection Solaris Format Checks
12658 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
12659 check. @code{cmn_err} accepts a subset of the standard @code{printf}
12660 conversions, and the two-argument @code{%b} conversion for displaying
12661 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
12663 @node Darwin Format Checks
12664 @subsection Darwin Format Checks
12666 Darwin targets support the @code{CFString} (or @code{__CFString__}) in the format
12667 attribute context. Declarations made with such attribution will be parsed for correct syntax
12668 and format argument types. However, parsing of the format string itself is currently undefined
12669 and will not be carried out by this version of the compiler.
12671 Additionally, @code{CFStringRefs} (defined by the @code{CoreFoundation} headers) may
12672 also be used as format arguments. Note that the relevant headers are only likely to be
12673 available on Darwin (OSX) installations. On such installations, the XCode and system
12674 documentation provide descriptions of @code{CFString}, @code{CFStringRefs} and
12675 associated functions.
12678 @section Pragmas Accepted by GCC
12680 @cindex @code{#pragma}
12682 GCC supports several types of pragmas, primarily in order to compile
12683 code originally written for other compilers. Note that in general
12684 we do not recommend the use of pragmas; @xref{Function Attributes},
12685 for further explanation.
12691 * RS/6000 and PowerPC Pragmas::
12693 * Solaris Pragmas::
12694 * Symbol-Renaming Pragmas::
12695 * Structure-Packing Pragmas::
12697 * Diagnostic Pragmas::
12698 * Visibility Pragmas::
12699 * Push/Pop Macro Pragmas::
12700 * Function Specific Option Pragmas::
12704 @subsection ARM Pragmas
12706 The ARM target defines pragmas for controlling the default addition of
12707 @code{long_call} and @code{short_call} attributes to functions.
12708 @xref{Function Attributes}, for information about the effects of these
12713 @cindex pragma, long_calls
12714 Set all subsequent functions to have the @code{long_call} attribute.
12716 @item no_long_calls
12717 @cindex pragma, no_long_calls
12718 Set all subsequent functions to have the @code{short_call} attribute.
12720 @item long_calls_off
12721 @cindex pragma, long_calls_off
12722 Do not affect the @code{long_call} or @code{short_call} attributes of
12723 subsequent functions.
12727 @subsection M32C Pragmas
12730 @item GCC memregs @var{number}
12731 @cindex pragma, memregs
12732 Overrides the command-line option @code{-memregs=} for the current
12733 file. Use with care! This pragma must be before any function in the
12734 file, and mixing different memregs values in different objects may
12735 make them incompatible. This pragma is useful when a
12736 performance-critical function uses a memreg for temporary values,
12737 as it may allow you to reduce the number of memregs used.
12739 @item ADDRESS @var{name} @var{address}
12740 @cindex pragma, address
12741 For any declared symbols matching @var{name}, this does three things
12742 to that symbol: it forces the symbol to be located at the given
12743 address (a number), it forces the symbol to be volatile, and it
12744 changes the symbol's scope to be static. This pragma exists for
12745 compatibility with other compilers, but note that the common
12746 @code{1234H} numeric syntax is not supported (use @code{0x1234}
12750 #pragma ADDRESS port3 0x103
12757 @subsection MeP Pragmas
12761 @item custom io_volatile (on|off)
12762 @cindex pragma, custom io_volatile
12763 Overrides the command line option @code{-mio-volatile} for the current
12764 file. Note that for compatibility with future GCC releases, this
12765 option should only be used once before any @code{io} variables in each
12768 @item GCC coprocessor available @var{registers}
12769 @cindex pragma, coprocessor available
12770 Specifies which coprocessor registers are available to the register
12771 allocator. @var{registers} may be a single register, register range
12772 separated by ellipses, or comma-separated list of those. Example:
12775 #pragma GCC coprocessor available $c0...$c10, $c28
12778 @item GCC coprocessor call_saved @var{registers}
12779 @cindex pragma, coprocessor call_saved
12780 Specifies which coprocessor registers are to be saved and restored by
12781 any function using them. @var{registers} may be a single register,
12782 register range separated by ellipses, or comma-separated list of
12786 #pragma GCC coprocessor call_saved $c4...$c6, $c31
12789 @item GCC coprocessor subclass '(A|B|C|D)' = @var{registers}
12790 @cindex pragma, coprocessor subclass
12791 Creates and defines a register class. These register classes can be
12792 used by inline @code{asm} constructs. @var{registers} may be a single
12793 register, register range separated by ellipses, or comma-separated
12794 list of those. Example:
12797 #pragma GCC coprocessor subclass 'B' = $c2, $c4, $c6
12799 asm ("cpfoo %0" : "=B" (x));
12802 @item GCC disinterrupt @var{name} , @var{name} @dots{}
12803 @cindex pragma, disinterrupt
12804 For the named functions, the compiler adds code to disable interrupts
12805 for the duration of those functions. Any functions so named, which
12806 are not encountered in the source, cause a warning that the pragma was
12807 not used. Examples:
12810 #pragma disinterrupt foo
12811 #pragma disinterrupt bar, grill
12812 int foo () @{ @dots{} @}
12815 @item GCC call @var{name} , @var{name} @dots{}
12816 @cindex pragma, call
12817 For the named functions, the compiler always uses a register-indirect
12818 call model when calling the named functions. Examples:
12827 @node RS/6000 and PowerPC Pragmas
12828 @subsection RS/6000 and PowerPC Pragmas
12830 The RS/6000 and PowerPC targets define one pragma for controlling
12831 whether or not the @code{longcall} attribute is added to function
12832 declarations by default. This pragma overrides the @option{-mlongcall}
12833 option, but not the @code{longcall} and @code{shortcall} attributes.
12834 @xref{RS/6000 and PowerPC Options}, for more information about when long
12835 calls are and are not necessary.
12839 @cindex pragma, longcall
12840 Apply the @code{longcall} attribute to all subsequent function
12844 Do not apply the @code{longcall} attribute to subsequent function
12848 @c Describe h8300 pragmas here.
12849 @c Describe sh pragmas here.
12850 @c Describe v850 pragmas here.
12852 @node Darwin Pragmas
12853 @subsection Darwin Pragmas
12855 The following pragmas are available for all architectures running the
12856 Darwin operating system. These are useful for compatibility with other
12860 @item mark @var{tokens}@dots{}
12861 @cindex pragma, mark
12862 This pragma is accepted, but has no effect.
12864 @item options align=@var{alignment}
12865 @cindex pragma, options align
12866 This pragma sets the alignment of fields in structures. The values of
12867 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
12868 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
12869 properly; to restore the previous setting, use @code{reset} for the
12872 @item segment @var{tokens}@dots{}
12873 @cindex pragma, segment
12874 This pragma is accepted, but has no effect.
12876 @item unused (@var{var} [, @var{var}]@dots{})
12877 @cindex pragma, unused
12878 This pragma declares variables to be possibly unused. GCC will not
12879 produce warnings for the listed variables. The effect is similar to
12880 that of the @code{unused} attribute, except that this pragma may appear
12881 anywhere within the variables' scopes.
12884 @node Solaris Pragmas
12885 @subsection Solaris Pragmas
12887 The Solaris target supports @code{#pragma redefine_extname}
12888 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
12889 @code{#pragma} directives for compatibility with the system compiler.
12892 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
12893 @cindex pragma, align
12895 Increase the minimum alignment of each @var{variable} to @var{alignment}.
12896 This is the same as GCC's @code{aligned} attribute @pxref{Variable
12897 Attributes}). Macro expansion occurs on the arguments to this pragma
12898 when compiling C and Objective-C@. It does not currently occur when
12899 compiling C++, but this is a bug which may be fixed in a future
12902 @item fini (@var{function} [, @var{function}]...)
12903 @cindex pragma, fini
12905 This pragma causes each listed @var{function} to be called after
12906 main, or during shared module unloading, by adding a call to the
12907 @code{.fini} section.
12909 @item init (@var{function} [, @var{function}]...)
12910 @cindex pragma, init
12912 This pragma causes each listed @var{function} to be called during
12913 initialization (before @code{main}) or during shared module loading, by
12914 adding a call to the @code{.init} section.
12918 @node Symbol-Renaming Pragmas
12919 @subsection Symbol-Renaming Pragmas
12921 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
12922 supports two @code{#pragma} directives which change the name used in
12923 assembly for a given declaration. @code{#pragma extern_prefix} is only
12924 available on platforms whose system headers need it. To get this effect
12925 on all platforms supported by GCC, use the asm labels extension (@pxref{Asm
12929 @item redefine_extname @var{oldname} @var{newname}
12930 @cindex pragma, redefine_extname
12932 This pragma gives the C function @var{oldname} the assembly symbol
12933 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
12934 will be defined if this pragma is available (currently on all platforms).
12936 @item extern_prefix @var{string}
12937 @cindex pragma, extern_prefix
12939 This pragma causes all subsequent external function and variable
12940 declarations to have @var{string} prepended to their assembly symbols.
12941 This effect may be terminated with another @code{extern_prefix} pragma
12942 whose argument is an empty string. The preprocessor macro
12943 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
12944 available (currently only on Tru64 UNIX)@.
12947 These pragmas and the asm labels extension interact in a complicated
12948 manner. Here are some corner cases you may want to be aware of.
12951 @item Both pragmas silently apply only to declarations with external
12952 linkage. Asm labels do not have this restriction.
12954 @item In C++, both pragmas silently apply only to declarations with
12955 ``C'' linkage. Again, asm labels do not have this restriction.
12957 @item If any of the three ways of changing the assembly name of a
12958 declaration is applied to a declaration whose assembly name has
12959 already been determined (either by a previous use of one of these
12960 features, or because the compiler needed the assembly name in order to
12961 generate code), and the new name is different, a warning issues and
12962 the name does not change.
12964 @item The @var{oldname} used by @code{#pragma redefine_extname} is
12965 always the C-language name.
12967 @item If @code{#pragma extern_prefix} is in effect, and a declaration
12968 occurs with an asm label attached, the prefix is silently ignored for
12971 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
12972 apply to the same declaration, whichever triggered first wins, and a
12973 warning issues if they contradict each other. (We would like to have
12974 @code{#pragma redefine_extname} always win, for consistency with asm
12975 labels, but if @code{#pragma extern_prefix} triggers first we have no
12976 way of knowing that that happened.)
12979 @node Structure-Packing Pragmas
12980 @subsection Structure-Packing Pragmas
12982 For compatibility with Microsoft Windows compilers, GCC supports a
12983 set of @code{#pragma} directives which change the maximum alignment of
12984 members of structures (other than zero-width bitfields), unions, and
12985 classes subsequently defined. The @var{n} value below always is required
12986 to be a small power of two and specifies the new alignment in bytes.
12989 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
12990 @item @code{#pragma pack()} sets the alignment to the one that was in
12991 effect when compilation started (see also command-line option
12992 @option{-fpack-struct[=@var{n}]} @pxref{Code Gen Options}).
12993 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
12994 setting on an internal stack and then optionally sets the new alignment.
12995 @item @code{#pragma pack(pop)} restores the alignment setting to the one
12996 saved at the top of the internal stack (and removes that stack entry).
12997 Note that @code{#pragma pack([@var{n}])} does not influence this internal
12998 stack; thus it is possible to have @code{#pragma pack(push)} followed by
12999 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
13000 @code{#pragma pack(pop)}.
13003 Some targets, e.g.@: i386 and powerpc, support the @code{ms_struct}
13004 @code{#pragma} which lays out a structure as the documented
13005 @code{__attribute__ ((ms_struct))}.
13007 @item @code{#pragma ms_struct on} turns on the layout for structures
13009 @item @code{#pragma ms_struct off} turns off the layout for structures
13011 @item @code{#pragma ms_struct reset} goes back to the default layout.
13015 @subsection Weak Pragmas
13017 For compatibility with SVR4, GCC supports a set of @code{#pragma}
13018 directives for declaring symbols to be weak, and defining weak
13022 @item #pragma weak @var{symbol}
13023 @cindex pragma, weak
13024 This pragma declares @var{symbol} to be weak, as if the declaration
13025 had the attribute of the same name. The pragma may appear before
13026 or after the declaration of @var{symbol}, but must appear before
13027 either its first use or its definition. It is not an error for
13028 @var{symbol} to never be defined at all.
13030 @item #pragma weak @var{symbol1} = @var{symbol2}
13031 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
13032 It is an error if @var{symbol2} is not defined in the current
13036 @node Diagnostic Pragmas
13037 @subsection Diagnostic Pragmas
13039 GCC allows the user to selectively enable or disable certain types of
13040 diagnostics, and change the kind of the diagnostic. For example, a
13041 project's policy might require that all sources compile with
13042 @option{-Werror} but certain files might have exceptions allowing
13043 specific types of warnings. Or, a project might selectively enable
13044 diagnostics and treat them as errors depending on which preprocessor
13045 macros are defined.
13048 @item #pragma GCC diagnostic @var{kind} @var{option}
13049 @cindex pragma, diagnostic
13051 Modifies the disposition of a diagnostic. Note that not all
13052 diagnostics are modifiable; at the moment only warnings (normally
13053 controlled by @samp{-W@dots{}}) can be controlled, and not all of them.
13054 Use @option{-fdiagnostics-show-option} to determine which diagnostics
13055 are controllable and which option controls them.
13057 @var{kind} is @samp{error} to treat this diagnostic as an error,
13058 @samp{warning} to treat it like a warning (even if @option{-Werror} is
13059 in effect), or @samp{ignored} if the diagnostic is to be ignored.
13060 @var{option} is a double quoted string which matches the command-line
13064 #pragma GCC diagnostic warning "-Wformat"
13065 #pragma GCC diagnostic error "-Wformat"
13066 #pragma GCC diagnostic ignored "-Wformat"
13069 Note that these pragmas override any command-line options. GCC keeps
13070 track of the location of each pragma, and issues diagnostics according
13071 to the state as of that point in the source file. Thus, pragmas occurring
13072 after a line do not affect diagnostics caused by that line.
13074 @item #pragma GCC diagnostic push
13075 @itemx #pragma GCC diagnostic pop
13077 Causes GCC to remember the state of the diagnostics as of each
13078 @code{push}, and restore to that point at each @code{pop}. If a
13079 @code{pop} has no matching @code{push}, the command line options are
13083 #pragma GCC diagnostic error "-Wuninitialized"
13084 foo(a); /* error is given for this one */
13085 #pragma GCC diagnostic push
13086 #pragma GCC diagnostic ignored "-Wuninitialized"
13087 foo(b); /* no diagnostic for this one */
13088 #pragma GCC diagnostic pop
13089 foo(c); /* error is given for this one */
13090 #pragma GCC diagnostic pop
13091 foo(d); /* depends on command line options */
13096 GCC also offers a simple mechanism for printing messages during
13100 @item #pragma message @var{string}
13101 @cindex pragma, diagnostic
13103 Prints @var{string} as a compiler message on compilation. The message
13104 is informational only, and is neither a compilation warning nor an error.
13107 #pragma message "Compiling " __FILE__ "..."
13110 @var{string} may be parenthesized, and is printed with location
13111 information. For example,
13114 #define DO_PRAGMA(x) _Pragma (#x)
13115 #define TODO(x) DO_PRAGMA(message ("TODO - " #x))
13117 TODO(Remember to fix this)
13120 prints @samp{/tmp/file.c:4: note: #pragma message:
13121 TODO - Remember to fix this}.
13125 @node Visibility Pragmas
13126 @subsection Visibility Pragmas
13129 @item #pragma GCC visibility push(@var{visibility})
13130 @itemx #pragma GCC visibility pop
13131 @cindex pragma, visibility
13133 This pragma allows the user to set the visibility for multiple
13134 declarations without having to give each a visibility attribute
13135 @xref{Function Attributes}, for more information about visibility and
13136 the attribute syntax.
13138 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
13139 declarations. Class members and template specializations are not
13140 affected; if you want to override the visibility for a particular
13141 member or instantiation, you must use an attribute.
13146 @node Push/Pop Macro Pragmas
13147 @subsection Push/Pop Macro Pragmas
13149 For compatibility with Microsoft Windows compilers, GCC supports
13150 @samp{#pragma push_macro(@var{"macro_name"})}
13151 and @samp{#pragma pop_macro(@var{"macro_name"})}.
13154 @item #pragma push_macro(@var{"macro_name"})
13155 @cindex pragma, push_macro
13156 This pragma saves the value of the macro named as @var{macro_name} to
13157 the top of the stack for this macro.
13159 @item #pragma pop_macro(@var{"macro_name"})
13160 @cindex pragma, pop_macro
13161 This pragma sets the value of the macro named as @var{macro_name} to
13162 the value on top of the stack for this macro. If the stack for
13163 @var{macro_name} is empty, the value of the macro remains unchanged.
13170 #pragma push_macro("X")
13173 #pragma pop_macro("X")
13177 In this example, the definition of X as 1 is saved by @code{#pragma
13178 push_macro} and restored by @code{#pragma pop_macro}.
13180 @node Function Specific Option Pragmas
13181 @subsection Function Specific Option Pragmas
13184 @item #pragma GCC target (@var{"string"}...)
13185 @cindex pragma GCC target
13187 This pragma allows you to set target specific options for functions
13188 defined later in the source file. One or more strings can be
13189 specified. Each function that is defined after this point will be as
13190 if @code{attribute((target("STRING")))} was specified for that
13191 function. The parenthesis around the options is optional.
13192 @xref{Function Attributes}, for more information about the
13193 @code{target} attribute and the attribute syntax.
13195 The @code{#pragma GCC target} attribute is not implemented in GCC versions earlier
13196 than 4.4 for the i386/x86_64 and 4.6 for the PowerPC backends. At
13197 present, it is not implemented for other backends.
13201 @item #pragma GCC optimize (@var{"string"}...)
13202 @cindex pragma GCC optimize
13204 This pragma allows you to set global optimization options for functions
13205 defined later in the source file. One or more strings can be
13206 specified. Each function that is defined after this point will be as
13207 if @code{attribute((optimize("STRING")))} was specified for that
13208 function. The parenthesis around the options is optional.
13209 @xref{Function Attributes}, for more information about the
13210 @code{optimize} attribute and the attribute syntax.
13212 The @samp{#pragma GCC optimize} pragma is not implemented in GCC
13213 versions earlier than 4.4.
13217 @item #pragma GCC push_options
13218 @itemx #pragma GCC pop_options
13219 @cindex pragma GCC push_options
13220 @cindex pragma GCC pop_options
13222 These pragmas maintain a stack of the current target and optimization
13223 options. It is intended for include files where you temporarily want
13224 to switch to using a different @samp{#pragma GCC target} or
13225 @samp{#pragma GCC optimize} and then to pop back to the previous
13228 The @samp{#pragma GCC push_options} and @samp{#pragma GCC pop_options}
13229 pragmas are not implemented in GCC versions earlier than 4.4.
13233 @item #pragma GCC reset_options
13234 @cindex pragma GCC reset_options
13236 This pragma clears the current @code{#pragma GCC target} and
13237 @code{#pragma GCC optimize} to use the default switches as specified
13238 on the command line.
13240 The @samp{#pragma GCC reset_options} pragma is not implemented in GCC
13241 versions earlier than 4.4.
13244 @node Unnamed Fields
13245 @section Unnamed struct/union fields within structs/unions
13246 @cindex @code{struct}
13247 @cindex @code{union}
13249 As permitted by ISO C1X and for compatibility with other compilers,
13250 GCC allows you to define
13251 a structure or union that contains, as fields, structures and unions
13252 without names. For example:
13265 In this example, the user would be able to access members of the unnamed
13266 union with code like @samp{foo.b}. Note that only unnamed structs and
13267 unions are allowed, you may not have, for example, an unnamed
13270 You must never create such structures that cause ambiguous field definitions.
13271 For example, this structure:
13282 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
13283 The compiler gives errors for such constructs.
13285 @opindex fms-extensions
13286 Unless @option{-fms-extensions} is used, the unnamed field must be a
13287 structure or union definition without a tag (for example, @samp{struct
13288 @{ int a; @};}), or a @code{typedef} name for such a structure or
13289 union. If @option{-fms-extensions} is used, the field may
13290 also be a definition with a tag such as @samp{struct foo @{ int a;
13291 @};}, a reference to a previously defined structure or union such as
13292 @samp{struct foo;}, or a reference to a @code{typedef} name for a
13293 previously defined structure or union type with a tag.
13295 @opindex fplan9-extensions
13296 The option @option{-fplan9-extensions} enables
13297 @option{-fms-extensions} as well as two other extensions. First, a
13298 pointer to a structure is automatically converted to a pointer to an
13299 anonymous field for assignments and function calls. For example:
13302 struct s1 @{ int a; @};
13303 struct s2 @{ struct s1; @};
13304 extern void f1 (struct s1 *);
13305 void f2 (struct s2 *p) @{ f1 (p); @}
13308 In the call to @code{f1} inside @code{f2}, the pointer @code{p} is
13309 converted into a pointer to the anonymous field.
13311 Second, when the type of an anonymous field is a @code{typedef} for a
13312 @code{struct} or @code{union}, code may refer to the field using the
13313 name of the @code{typedef}.
13316 typedef struct @{ int a; @} s1;
13317 struct s2 @{ s1; @};
13318 s1 f1 (struct s2 *p) @{ return p->s1; @}
13321 These usages are only permitted when they are not ambiguous.
13324 @section Thread-Local Storage
13325 @cindex Thread-Local Storage
13326 @cindex @acronym{TLS}
13327 @cindex @code{__thread}
13329 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
13330 are allocated such that there is one instance of the variable per extant
13331 thread. The run-time model GCC uses to implement this originates
13332 in the IA-64 processor-specific ABI, but has since been migrated
13333 to other processors as well. It requires significant support from
13334 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
13335 system libraries (@file{libc.so} and @file{libpthread.so}), so it
13336 is not available everywhere.
13338 At the user level, the extension is visible with a new storage
13339 class keyword: @code{__thread}. For example:
13343 extern __thread struct state s;
13344 static __thread char *p;
13347 The @code{__thread} specifier may be used alone, with the @code{extern}
13348 or @code{static} specifiers, but with no other storage class specifier.
13349 When used with @code{extern} or @code{static}, @code{__thread} must appear
13350 immediately after the other storage class specifier.
13352 The @code{__thread} specifier may be applied to any global, file-scoped
13353 static, function-scoped static, or static data member of a class. It may
13354 not be applied to block-scoped automatic or non-static data member.
13356 When the address-of operator is applied to a thread-local variable, it is
13357 evaluated at run-time and returns the address of the current thread's
13358 instance of that variable. An address so obtained may be used by any
13359 thread. When a thread terminates, any pointers to thread-local variables
13360 in that thread become invalid.
13362 No static initialization may refer to the address of a thread-local variable.
13364 In C++, if an initializer is present for a thread-local variable, it must
13365 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
13368 See @uref{http://people.redhat.com/drepper/tls.pdf,
13369 ELF Handling For Thread-Local Storage} for a detailed explanation of
13370 the four thread-local storage addressing models, and how the run-time
13371 is expected to function.
13374 * C99 Thread-Local Edits::
13375 * C++98 Thread-Local Edits::
13378 @node C99 Thread-Local Edits
13379 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
13381 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
13382 that document the exact semantics of the language extension.
13386 @cite{5.1.2 Execution environments}
13388 Add new text after paragraph 1
13391 Within either execution environment, a @dfn{thread} is a flow of
13392 control within a program. It is implementation defined whether
13393 or not there may be more than one thread associated with a program.
13394 It is implementation defined how threads beyond the first are
13395 created, the name and type of the function called at thread
13396 startup, and how threads may be terminated. However, objects
13397 with thread storage duration shall be initialized before thread
13402 @cite{6.2.4 Storage durations of objects}
13404 Add new text before paragraph 3
13407 An object whose identifier is declared with the storage-class
13408 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
13409 Its lifetime is the entire execution of the thread, and its
13410 stored value is initialized only once, prior to thread startup.
13414 @cite{6.4.1 Keywords}
13416 Add @code{__thread}.
13419 @cite{6.7.1 Storage-class specifiers}
13421 Add @code{__thread} to the list of storage class specifiers in
13424 Change paragraph 2 to
13427 With the exception of @code{__thread}, at most one storage-class
13428 specifier may be given [@dots{}]. The @code{__thread} specifier may
13429 be used alone, or immediately following @code{extern} or
13433 Add new text after paragraph 6
13436 The declaration of an identifier for a variable that has
13437 block scope that specifies @code{__thread} shall also
13438 specify either @code{extern} or @code{static}.
13440 The @code{__thread} specifier shall be used only with
13445 @node C++98 Thread-Local Edits
13446 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
13448 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
13449 that document the exact semantics of the language extension.
13453 @b{[intro.execution]}
13455 New text after paragraph 4
13458 A @dfn{thread} is a flow of control within the abstract machine.
13459 It is implementation defined whether or not there may be more than
13463 New text after paragraph 7
13466 It is unspecified whether additional action must be taken to
13467 ensure when and whether side effects are visible to other threads.
13473 Add @code{__thread}.
13476 @b{[basic.start.main]}
13478 Add after paragraph 5
13481 The thread that begins execution at the @code{main} function is called
13482 the @dfn{main thread}. It is implementation defined how functions
13483 beginning threads other than the main thread are designated or typed.
13484 A function so designated, as well as the @code{main} function, is called
13485 a @dfn{thread startup function}. It is implementation defined what
13486 happens if a thread startup function returns. It is implementation
13487 defined what happens to other threads when any thread calls @code{exit}.
13491 @b{[basic.start.init]}
13493 Add after paragraph 4
13496 The storage for an object of thread storage duration shall be
13497 statically initialized before the first statement of the thread startup
13498 function. An object of thread storage duration shall not require
13499 dynamic initialization.
13503 @b{[basic.start.term]}
13505 Add after paragraph 3
13508 The type of an object with thread storage duration shall not have a
13509 non-trivial destructor, nor shall it be an array type whose elements
13510 (directly or indirectly) have non-trivial destructors.
13516 Add ``thread storage duration'' to the list in paragraph 1.
13521 Thread, static, and automatic storage durations are associated with
13522 objects introduced by declarations [@dots{}].
13525 Add @code{__thread} to the list of specifiers in paragraph 3.
13528 @b{[basic.stc.thread]}
13530 New section before @b{[basic.stc.static]}
13533 The keyword @code{__thread} applied to a non-local object gives the
13534 object thread storage duration.
13536 A local variable or class data member declared both @code{static}
13537 and @code{__thread} gives the variable or member thread storage
13542 @b{[basic.stc.static]}
13547 All objects which have neither thread storage duration, dynamic
13548 storage duration nor are local [@dots{}].
13554 Add @code{__thread} to the list in paragraph 1.
13559 With the exception of @code{__thread}, at most one
13560 @var{storage-class-specifier} shall appear in a given
13561 @var{decl-specifier-seq}. The @code{__thread} specifier may
13562 be used alone, or immediately following the @code{extern} or
13563 @code{static} specifiers. [@dots{}]
13566 Add after paragraph 5
13569 The @code{__thread} specifier can be applied only to the names of objects
13570 and to anonymous unions.
13576 Add after paragraph 6
13579 Non-@code{static} members shall not be @code{__thread}.
13583 @node Binary constants
13584 @section Binary constants using the @samp{0b} prefix
13585 @cindex Binary constants using the @samp{0b} prefix
13587 Integer constants can be written as binary constants, consisting of a
13588 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
13589 @samp{0B}. This is particularly useful in environments that operate a
13590 lot on the bit-level (like microcontrollers).
13592 The following statements are identical:
13601 The type of these constants follows the same rules as for octal or
13602 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
13605 @node C++ Extensions
13606 @chapter Extensions to the C++ Language
13607 @cindex extensions, C++ language
13608 @cindex C++ language extensions
13610 The GNU compiler provides these extensions to the C++ language (and you
13611 can also use most of the C language extensions in your C++ programs). If you
13612 want to write code that checks whether these features are available, you can
13613 test for the GNU compiler the same way as for C programs: check for a
13614 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
13615 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
13616 Predefined Macros,cpp,The GNU C Preprocessor}).
13619 * C++ Volatiles:: What constitutes an access to a volatile object.
13620 * Restricted Pointers:: C99 restricted pointers and references.
13621 * Vague Linkage:: Where G++ puts inlines, vtables and such.
13622 * C++ Interface:: You can use a single C++ header file for both
13623 declarations and definitions.
13624 * Template Instantiation:: Methods for ensuring that exactly one copy of
13625 each needed template instantiation is emitted.
13626 * Bound member functions:: You can extract a function pointer to the
13627 method denoted by a @samp{->*} or @samp{.*} expression.
13628 * C++ Attributes:: Variable, function, and type attributes for C++ only.
13629 * Namespace Association:: Strong using-directives for namespace association.
13630 * Type Traits:: Compiler support for type traits
13631 * Java Exceptions:: Tweaking exception handling to work with Java.
13632 * Deprecated Features:: Things will disappear from g++.
13633 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
13636 @node C++ Volatiles
13637 @section When is a Volatile C++ Object Accessed?
13638 @cindex accessing volatiles
13639 @cindex volatile read
13640 @cindex volatile write
13641 @cindex volatile access
13643 The C++ standard differs from the C standard in its treatment of
13644 volatile objects. It fails to specify what constitutes a volatile
13645 access, except to say that C++ should behave in a similar manner to C
13646 with respect to volatiles, where possible. However, the different
13647 lvalueness of expressions between C and C++ complicate the behaviour.
13648 G++ behaves the same as GCC for volatile access, @xref{C
13649 Extensions,,Volatiles}, for a description of GCC's behaviour.
13651 The C and C++ language specifications differ when an object is
13652 accessed in a void context:
13655 volatile int *src = @var{somevalue};
13659 The C++ standard specifies that such expressions do not undergo lvalue
13660 to rvalue conversion, and that the type of the dereferenced object may
13661 be incomplete. The C++ standard does not specify explicitly that it
13662 is lvalue to rvalue conversion which is responsible for causing an
13663 access. There is reason to believe that it is, because otherwise
13664 certain simple expressions become undefined. However, because it
13665 would surprise most programmers, G++ treats dereferencing a pointer to
13666 volatile object of complete type as GCC would do for an equivalent
13667 type in C@. When the object has incomplete type, G++ issues a
13668 warning; if you wish to force an error, you must force a conversion to
13669 rvalue with, for instance, a static cast.
13671 When using a reference to volatile, G++ does not treat equivalent
13672 expressions as accesses to volatiles, but instead issues a warning that
13673 no volatile is accessed. The rationale for this is that otherwise it
13674 becomes difficult to determine where volatile access occur, and not
13675 possible to ignore the return value from functions returning volatile
13676 references. Again, if you wish to force a read, cast the reference to
13679 G++ implements the same behaviour as GCC does when assigning to a
13680 volatile object -- there is no reread of the assigned-to object, the
13681 assigned rvalue is reused. Note that in C++ assignment expressions
13682 are lvalues, and if used as an lvalue, the volatile object will be
13683 referred to. For instance, @var{vref} will refer to @var{vobj}, as
13684 expected, in the following example:
13688 volatile int &vref = vobj = @var{something};
13691 @node Restricted Pointers
13692 @section Restricting Pointer Aliasing
13693 @cindex restricted pointers
13694 @cindex restricted references
13695 @cindex restricted this pointer
13697 As with the C front end, G++ understands the C99 feature of restricted pointers,
13698 specified with the @code{__restrict__}, or @code{__restrict} type
13699 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
13700 language flag, @code{restrict} is not a keyword in C++.
13702 In addition to allowing restricted pointers, you can specify restricted
13703 references, which indicate that the reference is not aliased in the local
13707 void fn (int *__restrict__ rptr, int &__restrict__ rref)
13714 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
13715 @var{rref} refers to a (different) unaliased integer.
13717 You may also specify whether a member function's @var{this} pointer is
13718 unaliased by using @code{__restrict__} as a member function qualifier.
13721 void T::fn () __restrict__
13728 Within the body of @code{T::fn}, @var{this} will have the effective
13729 definition @code{T *__restrict__ const this}. Notice that the
13730 interpretation of a @code{__restrict__} member function qualifier is
13731 different to that of @code{const} or @code{volatile} qualifier, in that it
13732 is applied to the pointer rather than the object. This is consistent with
13733 other compilers which implement restricted pointers.
13735 As with all outermost parameter qualifiers, @code{__restrict__} is
13736 ignored in function definition matching. This means you only need to
13737 specify @code{__restrict__} in a function definition, rather than
13738 in a function prototype as well.
13740 @node Vague Linkage
13741 @section Vague Linkage
13742 @cindex vague linkage
13744 There are several constructs in C++ which require space in the object
13745 file but are not clearly tied to a single translation unit. We say that
13746 these constructs have ``vague linkage''. Typically such constructs are
13747 emitted wherever they are needed, though sometimes we can be more
13751 @item Inline Functions
13752 Inline functions are typically defined in a header file which can be
13753 included in many different compilations. Hopefully they can usually be
13754 inlined, but sometimes an out-of-line copy is necessary, if the address
13755 of the function is taken or if inlining fails. In general, we emit an
13756 out-of-line copy in all translation units where one is needed. As an
13757 exception, we only emit inline virtual functions with the vtable, since
13758 it will always require a copy.
13760 Local static variables and string constants used in an inline function
13761 are also considered to have vague linkage, since they must be shared
13762 between all inlined and out-of-line instances of the function.
13766 C++ virtual functions are implemented in most compilers using a lookup
13767 table, known as a vtable. The vtable contains pointers to the virtual
13768 functions provided by a class, and each object of the class contains a
13769 pointer to its vtable (or vtables, in some multiple-inheritance
13770 situations). If the class declares any non-inline, non-pure virtual
13771 functions, the first one is chosen as the ``key method'' for the class,
13772 and the vtable is only emitted in the translation unit where the key
13775 @emph{Note:} If the chosen key method is later defined as inline, the
13776 vtable will still be emitted in every translation unit which defines it.
13777 Make sure that any inline virtuals are declared inline in the class
13778 body, even if they are not defined there.
13780 @item @code{type_info} objects
13781 @cindex @code{type_info}
13783 C++ requires information about types to be written out in order to
13784 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
13785 For polymorphic classes (classes with virtual functions), the @samp{type_info}
13786 object is written out along with the vtable so that @samp{dynamic_cast}
13787 can determine the dynamic type of a class object at runtime. For all
13788 other types, we write out the @samp{type_info} object when it is used: when
13789 applying @samp{typeid} to an expression, throwing an object, or
13790 referring to a type in a catch clause or exception specification.
13792 @item Template Instantiations
13793 Most everything in this section also applies to template instantiations,
13794 but there are other options as well.
13795 @xref{Template Instantiation,,Where's the Template?}.
13799 When used with GNU ld version 2.8 or later on an ELF system such as
13800 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
13801 these constructs will be discarded at link time. This is known as
13804 On targets that don't support COMDAT, but do support weak symbols, GCC
13805 will use them. This way one copy will override all the others, but
13806 the unused copies will still take up space in the executable.
13808 For targets which do not support either COMDAT or weak symbols,
13809 most entities with vague linkage will be emitted as local symbols to
13810 avoid duplicate definition errors from the linker. This will not happen
13811 for local statics in inlines, however, as having multiple copies will
13812 almost certainly break things.
13814 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
13815 another way to control placement of these constructs.
13817 @node C++ Interface
13818 @section #pragma interface and implementation
13820 @cindex interface and implementation headers, C++
13821 @cindex C++ interface and implementation headers
13822 @cindex pragmas, interface and implementation
13824 @code{#pragma interface} and @code{#pragma implementation} provide the
13825 user with a way of explicitly directing the compiler to emit entities
13826 with vague linkage (and debugging information) in a particular
13829 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
13830 most cases, because of COMDAT support and the ``key method'' heuristic
13831 mentioned in @ref{Vague Linkage}. Using them can actually cause your
13832 program to grow due to unnecessary out-of-line copies of inline
13833 functions. Currently (3.4) the only benefit of these
13834 @code{#pragma}s is reduced duplication of debugging information, and
13835 that should be addressed soon on DWARF 2 targets with the use of
13839 @item #pragma interface
13840 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
13841 @kindex #pragma interface
13842 Use this directive in @emph{header files} that define object classes, to save
13843 space in most of the object files that use those classes. Normally,
13844 local copies of certain information (backup copies of inline member
13845 functions, debugging information, and the internal tables that implement
13846 virtual functions) must be kept in each object file that includes class
13847 definitions. You can use this pragma to avoid such duplication. When a
13848 header file containing @samp{#pragma interface} is included in a
13849 compilation, this auxiliary information will not be generated (unless
13850 the main input source file itself uses @samp{#pragma implementation}).
13851 Instead, the object files will contain references to be resolved at link
13854 The second form of this directive is useful for the case where you have
13855 multiple headers with the same name in different directories. If you
13856 use this form, you must specify the same string to @samp{#pragma
13859 @item #pragma implementation
13860 @itemx #pragma implementation "@var{objects}.h"
13861 @kindex #pragma implementation
13862 Use this pragma in a @emph{main input file}, when you want full output from
13863 included header files to be generated (and made globally visible). The
13864 included header file, in turn, should use @samp{#pragma interface}.
13865 Backup copies of inline member functions, debugging information, and the
13866 internal tables used to implement virtual functions are all generated in
13867 implementation files.
13869 @cindex implied @code{#pragma implementation}
13870 @cindex @code{#pragma implementation}, implied
13871 @cindex naming convention, implementation headers
13872 If you use @samp{#pragma implementation} with no argument, it applies to
13873 an include file with the same basename@footnote{A file's @dfn{basename}
13874 was the name stripped of all leading path information and of trailing
13875 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
13876 file. For example, in @file{allclass.cc}, giving just
13877 @samp{#pragma implementation}
13878 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
13880 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
13881 an implementation file whenever you would include it from
13882 @file{allclass.cc} even if you never specified @samp{#pragma
13883 implementation}. This was deemed to be more trouble than it was worth,
13884 however, and disabled.
13886 Use the string argument if you want a single implementation file to
13887 include code from multiple header files. (You must also use
13888 @samp{#include} to include the header file; @samp{#pragma
13889 implementation} only specifies how to use the file---it doesn't actually
13892 There is no way to split up the contents of a single header file into
13893 multiple implementation files.
13896 @cindex inlining and C++ pragmas
13897 @cindex C++ pragmas, effect on inlining
13898 @cindex pragmas in C++, effect on inlining
13899 @samp{#pragma implementation} and @samp{#pragma interface} also have an
13900 effect on function inlining.
13902 If you define a class in a header file marked with @samp{#pragma
13903 interface}, the effect on an inline function defined in that class is
13904 similar to an explicit @code{extern} declaration---the compiler emits
13905 no code at all to define an independent version of the function. Its
13906 definition is used only for inlining with its callers.
13908 @opindex fno-implement-inlines
13909 Conversely, when you include the same header file in a main source file
13910 that declares it as @samp{#pragma implementation}, the compiler emits
13911 code for the function itself; this defines a version of the function
13912 that can be found via pointers (or by callers compiled without
13913 inlining). If all calls to the function can be inlined, you can avoid
13914 emitting the function by compiling with @option{-fno-implement-inlines}.
13915 If any calls were not inlined, you will get linker errors.
13917 @node Template Instantiation
13918 @section Where's the Template?
13919 @cindex template instantiation
13921 C++ templates are the first language feature to require more
13922 intelligence from the environment than one usually finds on a UNIX
13923 system. Somehow the compiler and linker have to make sure that each
13924 template instance occurs exactly once in the executable if it is needed,
13925 and not at all otherwise. There are two basic approaches to this
13926 problem, which are referred to as the Borland model and the Cfront model.
13929 @item Borland model
13930 Borland C++ solved the template instantiation problem by adding the code
13931 equivalent of common blocks to their linker; the compiler emits template
13932 instances in each translation unit that uses them, and the linker
13933 collapses them together. The advantage of this model is that the linker
13934 only has to consider the object files themselves; there is no external
13935 complexity to worry about. This disadvantage is that compilation time
13936 is increased because the template code is being compiled repeatedly.
13937 Code written for this model tends to include definitions of all
13938 templates in the header file, since they must be seen to be
13942 The AT&T C++ translator, Cfront, solved the template instantiation
13943 problem by creating the notion of a template repository, an
13944 automatically maintained place where template instances are stored. A
13945 more modern version of the repository works as follows: As individual
13946 object files are built, the compiler places any template definitions and
13947 instantiations encountered in the repository. At link time, the link
13948 wrapper adds in the objects in the repository and compiles any needed
13949 instances that were not previously emitted. The advantages of this
13950 model are more optimal compilation speed and the ability to use the
13951 system linker; to implement the Borland model a compiler vendor also
13952 needs to replace the linker. The disadvantages are vastly increased
13953 complexity, and thus potential for error; for some code this can be
13954 just as transparent, but in practice it can been very difficult to build
13955 multiple programs in one directory and one program in multiple
13956 directories. Code written for this model tends to separate definitions
13957 of non-inline member templates into a separate file, which should be
13958 compiled separately.
13961 When used with GNU ld version 2.8 or later on an ELF system such as
13962 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
13963 Borland model. On other systems, G++ implements neither automatic
13966 A future version of G++ will support a hybrid model whereby the compiler
13967 will emit any instantiations for which the template definition is
13968 included in the compile, and store template definitions and
13969 instantiation context information into the object file for the rest.
13970 The link wrapper will extract that information as necessary and invoke
13971 the compiler to produce the remaining instantiations. The linker will
13972 then combine duplicate instantiations.
13974 In the mean time, you have the following options for dealing with
13975 template instantiations:
13980 Compile your template-using code with @option{-frepo}. The compiler will
13981 generate files with the extension @samp{.rpo} listing all of the
13982 template instantiations used in the corresponding object files which
13983 could be instantiated there; the link wrapper, @samp{collect2}, will
13984 then update the @samp{.rpo} files to tell the compiler where to place
13985 those instantiations and rebuild any affected object files. The
13986 link-time overhead is negligible after the first pass, as the compiler
13987 will continue to place the instantiations in the same files.
13989 This is your best option for application code written for the Borland
13990 model, as it will just work. Code written for the Cfront model will
13991 need to be modified so that the template definitions are available at
13992 one or more points of instantiation; usually this is as simple as adding
13993 @code{#include <tmethods.cc>} to the end of each template header.
13995 For library code, if you want the library to provide all of the template
13996 instantiations it needs, just try to link all of its object files
13997 together; the link will fail, but cause the instantiations to be
13998 generated as a side effect. Be warned, however, that this may cause
13999 conflicts if multiple libraries try to provide the same instantiations.
14000 For greater control, use explicit instantiation as described in the next
14004 @opindex fno-implicit-templates
14005 Compile your code with @option{-fno-implicit-templates} to disable the
14006 implicit generation of template instances, and explicitly instantiate
14007 all the ones you use. This approach requires more knowledge of exactly
14008 which instances you need than do the others, but it's less
14009 mysterious and allows greater control. You can scatter the explicit
14010 instantiations throughout your program, perhaps putting them in the
14011 translation units where the instances are used or the translation units
14012 that define the templates themselves; you can put all of the explicit
14013 instantiations you need into one big file; or you can create small files
14020 template class Foo<int>;
14021 template ostream& operator <<
14022 (ostream&, const Foo<int>&);
14025 for each of the instances you need, and create a template instantiation
14026 library from those.
14028 If you are using Cfront-model code, you can probably get away with not
14029 using @option{-fno-implicit-templates} when compiling files that don't
14030 @samp{#include} the member template definitions.
14032 If you use one big file to do the instantiations, you may want to
14033 compile it without @option{-fno-implicit-templates} so you get all of the
14034 instances required by your explicit instantiations (but not by any
14035 other files) without having to specify them as well.
14037 G++ has extended the template instantiation syntax given in the ISO
14038 standard to allow forward declaration of explicit instantiations
14039 (with @code{extern}), instantiation of the compiler support data for a
14040 template class (i.e.@: the vtable) without instantiating any of its
14041 members (with @code{inline}), and instantiation of only the static data
14042 members of a template class, without the support data or member
14043 functions (with (@code{static}):
14046 extern template int max (int, int);
14047 inline template class Foo<int>;
14048 static template class Foo<int>;
14052 Do nothing. Pretend G++ does implement automatic instantiation
14053 management. Code written for the Borland model will work fine, but
14054 each translation unit will contain instances of each of the templates it
14055 uses. In a large program, this can lead to an unacceptable amount of code
14059 @node Bound member functions
14060 @section Extracting the function pointer from a bound pointer to member function
14062 @cindex pointer to member function
14063 @cindex bound pointer to member function
14065 In C++, pointer to member functions (PMFs) are implemented using a wide
14066 pointer of sorts to handle all the possible call mechanisms; the PMF
14067 needs to store information about how to adjust the @samp{this} pointer,
14068 and if the function pointed to is virtual, where to find the vtable, and
14069 where in the vtable to look for the member function. If you are using
14070 PMFs in an inner loop, you should really reconsider that decision. If
14071 that is not an option, you can extract the pointer to the function that
14072 would be called for a given object/PMF pair and call it directly inside
14073 the inner loop, to save a bit of time.
14075 Note that you will still be paying the penalty for the call through a
14076 function pointer; on most modern architectures, such a call defeats the
14077 branch prediction features of the CPU@. This is also true of normal
14078 virtual function calls.
14080 The syntax for this extension is
14084 extern int (A::*fp)();
14085 typedef int (*fptr)(A *);
14087 fptr p = (fptr)(a.*fp);
14090 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
14091 no object is needed to obtain the address of the function. They can be
14092 converted to function pointers directly:
14095 fptr p1 = (fptr)(&A::foo);
14098 @opindex Wno-pmf-conversions
14099 You must specify @option{-Wno-pmf-conversions} to use this extension.
14101 @node C++ Attributes
14102 @section C++-Specific Variable, Function, and Type Attributes
14104 Some attributes only make sense for C++ programs.
14107 @item init_priority (@var{priority})
14108 @cindex @code{init_priority} attribute
14111 In Standard C++, objects defined at namespace scope are guaranteed to be
14112 initialized in an order in strict accordance with that of their definitions
14113 @emph{in a given translation unit}. No guarantee is made for initializations
14114 across translation units. However, GNU C++ allows users to control the
14115 order of initialization of objects defined at namespace scope with the
14116 @code{init_priority} attribute by specifying a relative @var{priority},
14117 a constant integral expression currently bounded between 101 and 65535
14118 inclusive. Lower numbers indicate a higher priority.
14120 In the following example, @code{A} would normally be created before
14121 @code{B}, but the @code{init_priority} attribute has reversed that order:
14124 Some_Class A __attribute__ ((init_priority (2000)));
14125 Some_Class B __attribute__ ((init_priority (543)));
14129 Note that the particular values of @var{priority} do not matter; only their
14132 @item java_interface
14133 @cindex @code{java_interface} attribute
14135 This type attribute informs C++ that the class is a Java interface. It may
14136 only be applied to classes declared within an @code{extern "Java"} block.
14137 Calls to methods declared in this interface will be dispatched using GCJ's
14138 interface table mechanism, instead of regular virtual table dispatch.
14142 See also @ref{Namespace Association}.
14144 @node Namespace Association
14145 @section Namespace Association
14147 @strong{Caution:} The semantics of this extension are not fully
14148 defined. Users should refrain from using this extension as its
14149 semantics may change subtly over time. It is possible that this
14150 extension will be removed in future versions of G++.
14152 A using-directive with @code{__attribute ((strong))} is stronger
14153 than a normal using-directive in two ways:
14157 Templates from the used namespace can be specialized and explicitly
14158 instantiated as though they were members of the using namespace.
14161 The using namespace is considered an associated namespace of all
14162 templates in the used namespace for purposes of argument-dependent
14166 The used namespace must be nested within the using namespace so that
14167 normal unqualified lookup works properly.
14169 This is useful for composing a namespace transparently from
14170 implementation namespaces. For example:
14175 template <class T> struct A @{ @};
14177 using namespace debug __attribute ((__strong__));
14178 template <> struct A<int> @{ @}; // @r{ok to specialize}
14180 template <class T> void f (A<T>);
14185 f (std::A<float>()); // @r{lookup finds} std::f
14191 @section Type Traits
14193 The C++ front-end implements syntactic extensions that allow to
14194 determine at compile time various characteristics of a type (or of a
14198 @item __has_nothrow_assign (type)
14199 If @code{type} is const qualified or is a reference type then the trait is
14200 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
14201 is true, else if @code{type} is a cv class or union type with copy assignment
14202 operators that are known not to throw an exception then the trait is true,
14203 else it is false. Requires: @code{type} shall be a complete type, an array
14204 type of unknown bound, or is a @code{void} type.
14206 @item __has_nothrow_copy (type)
14207 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
14208 @code{type} is a cv class or union type with copy constructors that
14209 are known not to throw an exception then the trait is true, else it is false.
14210 Requires: @code{type} shall be a complete type, an array type of
14211 unknown bound, or is a @code{void} type.
14213 @item __has_nothrow_constructor (type)
14214 If @code{__has_trivial_constructor (type)} is true then the trait is
14215 true, else if @code{type} is a cv class or union type (or array
14216 thereof) with a default constructor that is known not to throw an
14217 exception then the trait is true, else it is false. Requires:
14218 @code{type} shall be a complete type, an array type of unknown bound,
14219 or is a @code{void} type.
14221 @item __has_trivial_assign (type)
14222 If @code{type} is const qualified or is a reference type then the trait is
14223 false. Otherwise if @code{__is_pod (type)} is true then the trait is
14224 true, else if @code{type} is a cv class or union type with a trivial
14225 copy assignment ([class.copy]) then the trait is true, else it is
14226 false. Requires: @code{type} shall be a complete type, an array type
14227 of unknown bound, or is a @code{void} type.
14229 @item __has_trivial_copy (type)
14230 If @code{__is_pod (type)} is true or @code{type} is a reference type
14231 then the trait is true, else if @code{type} is a cv class or union type
14232 with a trivial copy constructor ([class.copy]) then the trait
14233 is true, else it is false. Requires: @code{type} shall be a complete
14234 type, an array type of unknown bound, or is a @code{void} type.
14236 @item __has_trivial_constructor (type)
14237 If @code{__is_pod (type)} is true then the trait is true, else if
14238 @code{type} is a cv class or union type (or array thereof) with a
14239 trivial default constructor ([class.ctor]) then the trait is true,
14240 else it is false. Requires: @code{type} shall be a complete type, an
14241 array type of unknown bound, or is a @code{void} type.
14243 @item __has_trivial_destructor (type)
14244 If @code{__is_pod (type)} is true or @code{type} is a reference type then
14245 the trait is true, else if @code{type} is a cv class or union type (or
14246 array thereof) with a trivial destructor ([class.dtor]) then the trait
14247 is true, else it is false. Requires: @code{type} shall be a complete
14248 type, an array type of unknown bound, or is a @code{void} type.
14250 @item __has_virtual_destructor (type)
14251 If @code{type} is a class type with a virtual destructor
14252 ([class.dtor]) then the trait is true, else it is false. Requires:
14253 @code{type} shall be a complete type, an array type of unknown bound,
14254 or is a @code{void} type.
14256 @item __is_abstract (type)
14257 If @code{type} is an abstract class ([class.abstract]) then the trait
14258 is true, else it is false. Requires: @code{type} shall be a complete
14259 type, an array type of unknown bound, or is a @code{void} type.
14261 @item __is_base_of (base_type, derived_type)
14262 If @code{base_type} is a base class of @code{derived_type}
14263 ([class.derived]) then the trait is true, otherwise it is false.
14264 Top-level cv qualifications of @code{base_type} and
14265 @code{derived_type} are ignored. For the purposes of this trait, a
14266 class type is considered is own base. Requires: if @code{__is_class
14267 (base_type)} and @code{__is_class (derived_type)} are true and
14268 @code{base_type} and @code{derived_type} are not the same type
14269 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
14270 type. Diagnostic is produced if this requirement is not met.
14272 @item __is_class (type)
14273 If @code{type} is a cv class type, and not a union type
14274 ([basic.compound]) the trait is true, else it is false.
14276 @item __is_empty (type)
14277 If @code{__is_class (type)} is false then the trait is false.
14278 Otherwise @code{type} is considered empty if and only if: @code{type}
14279 has no non-static data members, or all non-static data members, if
14280 any, are bit-fields of length 0, and @code{type} has no virtual
14281 members, and @code{type} has no virtual base classes, and @code{type}
14282 has no base classes @code{base_type} for which
14283 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
14284 be a complete type, an array type of unknown bound, or is a
14287 @item __is_enum (type)
14288 If @code{type} is a cv enumeration type ([basic.compound]) the trait is
14289 true, else it is false.
14291 @item __is_pod (type)
14292 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
14293 else it is false. Requires: @code{type} shall be a complete type,
14294 an array type of unknown bound, or is a @code{void} type.
14296 @item __is_polymorphic (type)
14297 If @code{type} is a polymorphic class ([class.virtual]) then the trait
14298 is true, else it is false. Requires: @code{type} shall be a complete
14299 type, an array type of unknown bound, or is a @code{void} type.
14301 @item __is_union (type)
14302 If @code{type} is a cv union type ([basic.compound]) the trait is
14303 true, else it is false.
14307 @node Java Exceptions
14308 @section Java Exceptions
14310 The Java language uses a slightly different exception handling model
14311 from C++. Normally, GNU C++ will automatically detect when you are
14312 writing C++ code that uses Java exceptions, and handle them
14313 appropriately. However, if C++ code only needs to execute destructors
14314 when Java exceptions are thrown through it, GCC will guess incorrectly.
14315 Sample problematic code is:
14318 struct S @{ ~S(); @};
14319 extern void bar(); // @r{is written in Java, and may throw exceptions}
14328 The usual effect of an incorrect guess is a link failure, complaining of
14329 a missing routine called @samp{__gxx_personality_v0}.
14331 You can inform the compiler that Java exceptions are to be used in a
14332 translation unit, irrespective of what it might think, by writing
14333 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
14334 @samp{#pragma} must appear before any functions that throw or catch
14335 exceptions, or run destructors when exceptions are thrown through them.
14337 You cannot mix Java and C++ exceptions in the same translation unit. It
14338 is believed to be safe to throw a C++ exception from one file through
14339 another file compiled for the Java exception model, or vice versa, but
14340 there may be bugs in this area.
14342 @node Deprecated Features
14343 @section Deprecated Features
14345 In the past, the GNU C++ compiler was extended to experiment with new
14346 features, at a time when the C++ language was still evolving. Now that
14347 the C++ standard is complete, some of those features are superseded by
14348 superior alternatives. Using the old features might cause a warning in
14349 some cases that the feature will be dropped in the future. In other
14350 cases, the feature might be gone already.
14352 While the list below is not exhaustive, it documents some of the options
14353 that are now deprecated:
14356 @item -fexternal-templates
14357 @itemx -falt-external-templates
14358 These are two of the many ways for G++ to implement template
14359 instantiation. @xref{Template Instantiation}. The C++ standard clearly
14360 defines how template definitions have to be organized across
14361 implementation units. G++ has an implicit instantiation mechanism that
14362 should work just fine for standard-conforming code.
14364 @item -fstrict-prototype
14365 @itemx -fno-strict-prototype
14366 Previously it was possible to use an empty prototype parameter list to
14367 indicate an unspecified number of parameters (like C), rather than no
14368 parameters, as C++ demands. This feature has been removed, except where
14369 it is required for backwards compatibility. @xref{Backwards Compatibility}.
14372 G++ allows a virtual function returning @samp{void *} to be overridden
14373 by one returning a different pointer type. This extension to the
14374 covariant return type rules is now deprecated and will be removed from a
14377 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
14378 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
14379 and are now removed from G++. Code using these operators should be
14380 modified to use @code{std::min} and @code{std::max} instead.
14382 The named return value extension has been deprecated, and is now
14385 The use of initializer lists with new expressions has been deprecated,
14386 and is now removed from G++.
14388 Floating and complex non-type template parameters have been deprecated,
14389 and are now removed from G++.
14391 The implicit typename extension has been deprecated and is now
14394 The use of default arguments in function pointers, function typedefs
14395 and other places where they are not permitted by the standard is
14396 deprecated and will be removed from a future version of G++.
14398 G++ allows floating-point literals to appear in integral constant expressions,
14399 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
14400 This extension is deprecated and will be removed from a future version.
14402 G++ allows static data members of const floating-point type to be declared
14403 with an initializer in a class definition. The standard only allows
14404 initializers for static members of const integral types and const
14405 enumeration types so this extension has been deprecated and will be removed
14406 from a future version.
14408 @node Backwards Compatibility
14409 @section Backwards Compatibility
14410 @cindex Backwards Compatibility
14411 @cindex ARM [Annotated C++ Reference Manual]
14413 Now that there is a definitive ISO standard C++, G++ has a specification
14414 to adhere to. The C++ language evolved over time, and features that
14415 used to be acceptable in previous drafts of the standard, such as the ARM
14416 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
14417 compilation of C++ written to such drafts, G++ contains some backwards
14418 compatibilities. @emph{All such backwards compatibility features are
14419 liable to disappear in future versions of G++.} They should be considered
14420 deprecated. @xref{Deprecated Features}.
14424 If a variable is declared at for scope, it used to remain in scope until
14425 the end of the scope which contained the for statement (rather than just
14426 within the for scope). G++ retains this, but issues a warning, if such a
14427 variable is accessed outside the for scope.
14429 @item Implicit C language
14430 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
14431 scope to set the language. On such systems, all header files are
14432 implicitly scoped inside a C language scope. Also, an empty prototype
14433 @code{()} will be treated as an unspecified number of arguments, rather
14434 than no arguments, as C++ demands.