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, 2011
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.
937 The @code{__float128} type is supported on hppa HP-UX targets.
940 @section Half-Precision Floating Point
941 @cindex half-precision floating point
942 @cindex @code{__fp16} data type
944 On ARM targets, GCC supports half-precision (16-bit) floating point via
945 the @code{__fp16} type. You must enable this type explicitly
946 with the @option{-mfp16-format} command-line option in order to use it.
948 ARM supports two incompatible representations for half-precision
949 floating-point values. You must choose one of the representations and
950 use it consistently in your program.
952 Specifying @option{-mfp16-format=ieee} selects the IEEE 754-2008 format.
953 This format can represent normalized values in the range of @math{2^{-14}} to 65504.
954 There are 11 bits of significand precision, approximately 3
957 Specifying @option{-mfp16-format=alternative} selects the ARM
958 alternative format. This representation is similar to the IEEE
959 format, but does not support infinities or NaNs. Instead, the range
960 of exponents is extended, so that this format can represent normalized
961 values in the range of @math{2^{-14}} to 131008.
963 The @code{__fp16} type is a storage format only. For purposes
964 of arithmetic and other operations, @code{__fp16} values in C or C++
965 expressions are automatically promoted to @code{float}. In addition,
966 you cannot declare a function with a return value or parameters
967 of type @code{__fp16}.
969 Note that conversions from @code{double} to @code{__fp16}
970 involve an intermediate conversion to @code{float}. Because
971 of rounding, this can sometimes produce a different result than a
974 ARM provides hardware support for conversions between
975 @code{__fp16} and @code{float} values
976 as an extension to VFP and NEON (Advanced SIMD). GCC generates
977 code using these hardware instructions if you compile with
978 options to select an FPU that provides them;
979 for example, @option{-mfpu=neon-fp16 -mfloat-abi=softfp},
980 in addition to the @option{-mfp16-format} option to select
981 a half-precision format.
983 Language-level support for the @code{__fp16} data type is
984 independent of whether GCC generates code using hardware floating-point
985 instructions. In cases where hardware support is not specified, GCC
986 implements conversions between @code{__fp16} and @code{float} values
990 @section Decimal Floating Types
991 @cindex decimal floating types
992 @cindex @code{_Decimal32} data type
993 @cindex @code{_Decimal64} data type
994 @cindex @code{_Decimal128} data type
995 @cindex @code{df} integer suffix
996 @cindex @code{dd} integer suffix
997 @cindex @code{dl} integer suffix
998 @cindex @code{DF} integer suffix
999 @cindex @code{DD} integer suffix
1000 @cindex @code{DL} integer suffix
1002 As an extension, the GNU C compiler supports decimal floating types as
1003 defined in the N1312 draft of ISO/IEC WDTR24732. Support for decimal
1004 floating types in GCC will evolve as the draft technical report changes.
1005 Calling conventions for any target might also change. Not all targets
1006 support decimal floating types.
1008 The decimal floating types are @code{_Decimal32}, @code{_Decimal64}, and
1009 @code{_Decimal128}. They use a radix of ten, unlike the floating types
1010 @code{float}, @code{double}, and @code{long double} whose radix is not
1011 specified by the C standard but is usually two.
1013 Support for decimal floating types includes the arithmetic operators
1014 add, subtract, multiply, divide; unary arithmetic operators;
1015 relational operators; equality operators; and conversions to and from
1016 integer and other floating types. Use a suffix @samp{df} or
1017 @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd}
1018 or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for
1021 GCC support of decimal float as specified by the draft technical report
1026 When the value of a decimal floating type cannot be represented in the
1027 integer type to which it is being converted, the result is undefined
1028 rather than the result value specified by the draft technical report.
1031 GCC does not provide the C library functionality associated with
1032 @file{math.h}, @file{fenv.h}, @file{stdio.h}, @file{stdlib.h}, and
1033 @file{wchar.h}, which must come from a separate C library implementation.
1034 Because of this the GNU C compiler does not define macro
1035 @code{__STDC_DEC_FP__} to indicate that the implementation conforms to
1036 the technical report.
1039 Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128}
1040 are supported by the DWARF2 debug information format.
1046 ISO C99 supports floating-point numbers written not only in the usual
1047 decimal notation, such as @code{1.55e1}, but also numbers such as
1048 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
1049 supports this in C90 mode (except in some cases when strictly
1050 conforming) and in C++. In that format the
1051 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
1052 mandatory. The exponent is a decimal number that indicates the power of
1053 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
1060 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
1061 is the same as @code{1.55e1}.
1063 Unlike for floating-point numbers in the decimal notation the exponent
1064 is always required in the hexadecimal notation. Otherwise the compiler
1065 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
1066 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
1067 extension for floating-point constants of type @code{float}.
1070 @section Fixed-Point Types
1071 @cindex fixed-point types
1072 @cindex @code{_Fract} data type
1073 @cindex @code{_Accum} data type
1074 @cindex @code{_Sat} data type
1075 @cindex @code{hr} fixed-suffix
1076 @cindex @code{r} fixed-suffix
1077 @cindex @code{lr} fixed-suffix
1078 @cindex @code{llr} fixed-suffix
1079 @cindex @code{uhr} fixed-suffix
1080 @cindex @code{ur} fixed-suffix
1081 @cindex @code{ulr} fixed-suffix
1082 @cindex @code{ullr} fixed-suffix
1083 @cindex @code{hk} fixed-suffix
1084 @cindex @code{k} fixed-suffix
1085 @cindex @code{lk} fixed-suffix
1086 @cindex @code{llk} fixed-suffix
1087 @cindex @code{uhk} fixed-suffix
1088 @cindex @code{uk} fixed-suffix
1089 @cindex @code{ulk} fixed-suffix
1090 @cindex @code{ullk} fixed-suffix
1091 @cindex @code{HR} fixed-suffix
1092 @cindex @code{R} fixed-suffix
1093 @cindex @code{LR} fixed-suffix
1094 @cindex @code{LLR} fixed-suffix
1095 @cindex @code{UHR} fixed-suffix
1096 @cindex @code{UR} fixed-suffix
1097 @cindex @code{ULR} fixed-suffix
1098 @cindex @code{ULLR} fixed-suffix
1099 @cindex @code{HK} fixed-suffix
1100 @cindex @code{K} fixed-suffix
1101 @cindex @code{LK} fixed-suffix
1102 @cindex @code{LLK} fixed-suffix
1103 @cindex @code{UHK} fixed-suffix
1104 @cindex @code{UK} fixed-suffix
1105 @cindex @code{ULK} fixed-suffix
1106 @cindex @code{ULLK} fixed-suffix
1108 As an extension, the GNU C compiler supports fixed-point types as
1109 defined in the N1169 draft of ISO/IEC DTR 18037. Support for fixed-point
1110 types in GCC will evolve as the draft technical report changes.
1111 Calling conventions for any target might also change. Not all targets
1112 support fixed-point types.
1114 The fixed-point types are
1115 @code{short _Fract},
1118 @code{long long _Fract},
1119 @code{unsigned short _Fract},
1120 @code{unsigned _Fract},
1121 @code{unsigned long _Fract},
1122 @code{unsigned long long _Fract},
1123 @code{_Sat short _Fract},
1125 @code{_Sat long _Fract},
1126 @code{_Sat long long _Fract},
1127 @code{_Sat unsigned short _Fract},
1128 @code{_Sat unsigned _Fract},
1129 @code{_Sat unsigned long _Fract},
1130 @code{_Sat unsigned long long _Fract},
1131 @code{short _Accum},
1134 @code{long long _Accum},
1135 @code{unsigned short _Accum},
1136 @code{unsigned _Accum},
1137 @code{unsigned long _Accum},
1138 @code{unsigned long long _Accum},
1139 @code{_Sat short _Accum},
1141 @code{_Sat long _Accum},
1142 @code{_Sat long long _Accum},
1143 @code{_Sat unsigned short _Accum},
1144 @code{_Sat unsigned _Accum},
1145 @code{_Sat unsigned long _Accum},
1146 @code{_Sat unsigned long long _Accum}.
1148 Fixed-point data values contain fractional and optional integral parts.
1149 The format of fixed-point data varies and depends on the target machine.
1151 Support for fixed-point types includes:
1154 prefix and postfix increment and decrement operators (@code{++}, @code{--})
1156 unary arithmetic operators (@code{+}, @code{-}, @code{!})
1158 binary arithmetic operators (@code{+}, @code{-}, @code{*}, @code{/})
1160 binary shift operators (@code{<<}, @code{>>})
1162 relational operators (@code{<}, @code{<=}, @code{>=}, @code{>})
1164 equality operators (@code{==}, @code{!=})
1166 assignment operators (@code{+=}, @code{-=}, @code{*=}, @code{/=},
1167 @code{<<=}, @code{>>=})
1169 conversions to and from integer, floating-point, or fixed-point types
1172 Use a suffix in a fixed-point literal constant:
1174 @item @samp{hr} or @samp{HR} for @code{short _Fract} and
1175 @code{_Sat short _Fract}
1176 @item @samp{r} or @samp{R} for @code{_Fract} and @code{_Sat _Fract}
1177 @item @samp{lr} or @samp{LR} for @code{long _Fract} and
1178 @code{_Sat long _Fract}
1179 @item @samp{llr} or @samp{LLR} for @code{long long _Fract} and
1180 @code{_Sat long long _Fract}
1181 @item @samp{uhr} or @samp{UHR} for @code{unsigned short _Fract} and
1182 @code{_Sat unsigned short _Fract}
1183 @item @samp{ur} or @samp{UR} for @code{unsigned _Fract} and
1184 @code{_Sat unsigned _Fract}
1185 @item @samp{ulr} or @samp{ULR} for @code{unsigned long _Fract} and
1186 @code{_Sat unsigned long _Fract}
1187 @item @samp{ullr} or @samp{ULLR} for @code{unsigned long long _Fract}
1188 and @code{_Sat unsigned long long _Fract}
1189 @item @samp{hk} or @samp{HK} for @code{short _Accum} and
1190 @code{_Sat short _Accum}
1191 @item @samp{k} or @samp{K} for @code{_Accum} and @code{_Sat _Accum}
1192 @item @samp{lk} or @samp{LK} for @code{long _Accum} and
1193 @code{_Sat long _Accum}
1194 @item @samp{llk} or @samp{LLK} for @code{long long _Accum} and
1195 @code{_Sat long long _Accum}
1196 @item @samp{uhk} or @samp{UHK} for @code{unsigned short _Accum} and
1197 @code{_Sat unsigned short _Accum}
1198 @item @samp{uk} or @samp{UK} for @code{unsigned _Accum} and
1199 @code{_Sat unsigned _Accum}
1200 @item @samp{ulk} or @samp{ULK} for @code{unsigned long _Accum} and
1201 @code{_Sat unsigned long _Accum}
1202 @item @samp{ullk} or @samp{ULLK} for @code{unsigned long long _Accum}
1203 and @code{_Sat unsigned long long _Accum}
1206 GCC support of fixed-point types as specified by the draft technical report
1211 Pragmas to control overflow and rounding behaviors are not implemented.
1214 Fixed-point types are supported by the DWARF2 debug information format.
1216 @node Named Address Spaces
1217 @section Named address spaces
1218 @cindex named address spaces
1220 As an extension, the GNU C compiler supports named address spaces as
1221 defined in the N1275 draft of ISO/IEC DTR 18037. Support for named
1222 address spaces in GCC will evolve as the draft technical report changes.
1223 Calling conventions for any target might also change. At present, only
1224 the SPU and M32C targets support other address spaces. On the SPU target, for
1225 example, variables may be declared as belonging to another address space
1226 by qualifying the type with the @code{__ea} address space identifier:
1232 When the variable @code{i} is accessed, the compiler will generate
1233 special code to access this variable. It may use runtime library
1234 support, or generate special machine instructions to access that address
1237 The @code{__ea} identifier may be used exactly like any other C type
1238 qualifier (e.g., @code{const} or @code{volatile}). See the N1275
1239 document for more details.
1241 On the M32C target, with the R8C and M16C cpu variants, variables
1242 qualified with @code{__far} are accessed using 32-bit addresses in
1243 order to access memory beyond the first 64k bytes. If @code{__far} is
1244 used with the M32CM or M32C cpu variants, it has no effect.
1247 @section Arrays of Length Zero
1248 @cindex arrays of length zero
1249 @cindex zero-length arrays
1250 @cindex length-zero arrays
1251 @cindex flexible array members
1253 Zero-length arrays are allowed in GNU C@. They are very useful as the
1254 last element of a structure which is really a header for a variable-length
1263 struct line *thisline = (struct line *)
1264 malloc (sizeof (struct line) + this_length);
1265 thisline->length = this_length;
1268 In ISO C90, you would have to give @code{contents} a length of 1, which
1269 means either you waste space or complicate the argument to @code{malloc}.
1271 In ISO C99, you would use a @dfn{flexible array member}, which is
1272 slightly different in syntax and semantics:
1276 Flexible array members are written as @code{contents[]} without
1280 Flexible array members have incomplete type, and so the @code{sizeof}
1281 operator may not be applied. As a quirk of the original implementation
1282 of zero-length arrays, @code{sizeof} evaluates to zero.
1285 Flexible array members may only appear as the last member of a
1286 @code{struct} that is otherwise non-empty.
1289 A structure containing a flexible array member, or a union containing
1290 such a structure (possibly recursively), may not be a member of a
1291 structure or an element of an array. (However, these uses are
1292 permitted by GCC as extensions.)
1295 GCC versions before 3.0 allowed zero-length arrays to be statically
1296 initialized, as if they were flexible arrays. In addition to those
1297 cases that were useful, it also allowed initializations in situations
1298 that would corrupt later data. Non-empty initialization of zero-length
1299 arrays is now treated like any case where there are more initializer
1300 elements than the array holds, in that a suitable warning about "excess
1301 elements in array" is given, and the excess elements (all of them, in
1302 this case) are ignored.
1304 Instead GCC allows static initialization of flexible array members.
1305 This is equivalent to defining a new structure containing the original
1306 structure followed by an array of sufficient size to contain the data.
1307 I.e.@: in the following, @code{f1} is constructed as if it were declared
1313 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
1316 struct f1 f1; int data[3];
1317 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1321 The convenience of this extension is that @code{f1} has the desired
1322 type, eliminating the need to consistently refer to @code{f2.f1}.
1324 This has symmetry with normal static arrays, in that an array of
1325 unknown size is also written with @code{[]}.
1327 Of course, this extension only makes sense if the extra data comes at
1328 the end of a top-level object, as otherwise we would be overwriting
1329 data at subsequent offsets. To avoid undue complication and confusion
1330 with initialization of deeply nested arrays, we simply disallow any
1331 non-empty initialization except when the structure is the top-level
1332 object. For example:
1335 struct foo @{ int x; int y[]; @};
1336 struct bar @{ struct foo z; @};
1338 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1339 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1340 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1341 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1344 @node Empty Structures
1345 @section Structures With No Members
1346 @cindex empty structures
1347 @cindex zero-size structures
1349 GCC permits a C structure to have no members:
1356 The structure will have size zero. In C++, empty structures are part
1357 of the language. G++ treats empty structures as if they had a single
1358 member of type @code{char}.
1360 @node Variable Length
1361 @section Arrays of Variable Length
1362 @cindex variable-length arrays
1363 @cindex arrays of variable length
1366 Variable-length automatic arrays are allowed in ISO C99, and as an
1367 extension GCC accepts them in C90 mode and in C++. These arrays are
1368 declared like any other automatic arrays, but with a length that is not
1369 a constant expression. The storage is allocated at the point of
1370 declaration and deallocated when the brace-level is exited. For
1375 concat_fopen (char *s1, char *s2, char *mode)
1377 char str[strlen (s1) + strlen (s2) + 1];
1380 return fopen (str, mode);
1384 @cindex scope of a variable length array
1385 @cindex variable-length array scope
1386 @cindex deallocating variable length arrays
1387 Jumping or breaking out of the scope of the array name deallocates the
1388 storage. Jumping into the scope is not allowed; you get an error
1391 @cindex @code{alloca} vs variable-length arrays
1392 You can use the function @code{alloca} to get an effect much like
1393 variable-length arrays. The function @code{alloca} is available in
1394 many other C implementations (but not in all). On the other hand,
1395 variable-length arrays are more elegant.
1397 There are other differences between these two methods. Space allocated
1398 with @code{alloca} exists until the containing @emph{function} returns.
1399 The space for a variable-length array is deallocated as soon as the array
1400 name's scope ends. (If you use both variable-length arrays and
1401 @code{alloca} in the same function, deallocation of a variable-length array
1402 will also deallocate anything more recently allocated with @code{alloca}.)
1404 You can also use variable-length arrays as arguments to functions:
1408 tester (int len, char data[len][len])
1414 The length of an array is computed once when the storage is allocated
1415 and is remembered for the scope of the array in case you access it with
1418 If you want to pass the array first and the length afterward, you can
1419 use a forward declaration in the parameter list---another GNU extension.
1423 tester (int len; char data[len][len], int len)
1429 @cindex parameter forward declaration
1430 The @samp{int len} before the semicolon is a @dfn{parameter forward
1431 declaration}, and it serves the purpose of making the name @code{len}
1432 known when the declaration of @code{data} is parsed.
1434 You can write any number of such parameter forward declarations in the
1435 parameter list. They can be separated by commas or semicolons, but the
1436 last one must end with a semicolon, which is followed by the ``real''
1437 parameter declarations. Each forward declaration must match a ``real''
1438 declaration in parameter name and data type. ISO C99 does not support
1439 parameter forward declarations.
1441 @node Variadic Macros
1442 @section Macros with a Variable Number of Arguments.
1443 @cindex variable number of arguments
1444 @cindex macro with variable arguments
1445 @cindex rest argument (in macro)
1446 @cindex variadic macros
1448 In the ISO C standard of 1999, a macro can be declared to accept a
1449 variable number of arguments much as a function can. The syntax for
1450 defining the macro is similar to that of a function. Here is an
1454 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1457 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1458 such a macro, it represents the zero or more tokens until the closing
1459 parenthesis that ends the invocation, including any commas. This set of
1460 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1461 wherever it appears. See the CPP manual for more information.
1463 GCC has long supported variadic macros, and used a different syntax that
1464 allowed you to give a name to the variable arguments just like any other
1465 argument. Here is an example:
1468 #define debug(format, args...) fprintf (stderr, format, args)
1471 This is in all ways equivalent to the ISO C example above, but arguably
1472 more readable and descriptive.
1474 GNU CPP has two further variadic macro extensions, and permits them to
1475 be used with either of the above forms of macro definition.
1477 In standard C, you are not allowed to leave the variable argument out
1478 entirely; but you are allowed to pass an empty argument. For example,
1479 this invocation is invalid in ISO C, because there is no comma after
1486 GNU CPP permits you to completely omit the variable arguments in this
1487 way. In the above examples, the compiler would complain, though since
1488 the expansion of the macro still has the extra comma after the format
1491 To help solve this problem, CPP behaves specially for variable arguments
1492 used with the token paste operator, @samp{##}. If instead you write
1495 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1498 and if the variable arguments are omitted or empty, the @samp{##}
1499 operator causes the preprocessor to remove the comma before it. If you
1500 do provide some variable arguments in your macro invocation, GNU CPP
1501 does not complain about the paste operation and instead places the
1502 variable arguments after the comma. Just like any other pasted macro
1503 argument, these arguments are not macro expanded.
1505 @node Escaped Newlines
1506 @section Slightly Looser Rules for Escaped Newlines
1507 @cindex escaped newlines
1508 @cindex newlines (escaped)
1510 Recently, the preprocessor has relaxed its treatment of escaped
1511 newlines. Previously, the newline had to immediately follow a
1512 backslash. The current implementation allows whitespace in the form
1513 of spaces, horizontal and vertical tabs, and form feeds between the
1514 backslash and the subsequent newline. The preprocessor issues a
1515 warning, but treats it as a valid escaped newline and combines the two
1516 lines to form a single logical line. This works within comments and
1517 tokens, as well as between tokens. Comments are @emph{not} treated as
1518 whitespace for the purposes of this relaxation, since they have not
1519 yet been replaced with spaces.
1522 @section Non-Lvalue Arrays May Have Subscripts
1523 @cindex subscripting
1524 @cindex arrays, non-lvalue
1526 @cindex subscripting and function values
1527 In ISO C99, arrays that are not lvalues still decay to pointers, and
1528 may be subscripted, although they may not be modified or used after
1529 the next sequence point and the unary @samp{&} operator may not be
1530 applied to them. As an extension, GCC allows such arrays to be
1531 subscripted in C90 mode, though otherwise they do not decay to
1532 pointers outside C99 mode. For example,
1533 this is valid in GNU C though not valid in C90:
1537 struct foo @{int a[4];@};
1543 return f().a[index];
1549 @section Arithmetic on @code{void}- and Function-Pointers
1550 @cindex void pointers, arithmetic
1551 @cindex void, size of pointer to
1552 @cindex function pointers, arithmetic
1553 @cindex function, size of pointer to
1555 In GNU C, addition and subtraction operations are supported on pointers to
1556 @code{void} and on pointers to functions. This is done by treating the
1557 size of a @code{void} or of a function as 1.
1559 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1560 and on function types, and returns 1.
1562 @opindex Wpointer-arith
1563 The option @option{-Wpointer-arith} requests a warning if these extensions
1567 @section Non-Constant Initializers
1568 @cindex initializers, non-constant
1569 @cindex non-constant initializers
1571 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1572 automatic variable are not required to be constant expressions in GNU C@.
1573 Here is an example of an initializer with run-time varying elements:
1576 foo (float f, float g)
1578 float beat_freqs[2] = @{ f-g, f+g @};
1583 @node Compound Literals
1584 @section Compound Literals
1585 @cindex constructor expressions
1586 @cindex initializations in expressions
1587 @cindex structures, constructor expression
1588 @cindex expressions, constructor
1589 @cindex compound literals
1590 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1592 ISO C99 supports compound literals. A compound literal looks like
1593 a cast containing an initializer. Its value is an object of the
1594 type specified in the cast, containing the elements specified in
1595 the initializer; it is an lvalue. As an extension, GCC supports
1596 compound literals in C90 mode and in C++.
1598 Usually, the specified type is a structure. Assume that
1599 @code{struct foo} and @code{structure} are declared as shown:
1602 struct foo @{int a; char b[2];@} structure;
1606 Here is an example of constructing a @code{struct foo} with a compound literal:
1609 structure = ((struct foo) @{x + y, 'a', 0@});
1613 This is equivalent to writing the following:
1617 struct foo temp = @{x + y, 'a', 0@};
1622 You can also construct an array. If all the elements of the compound literal
1623 are (made up of) simple constant expressions, suitable for use in
1624 initializers of objects of static storage duration, then the compound
1625 literal can be coerced to a pointer to its first element and used in
1626 such an initializer, as shown here:
1629 char **foo = (char *[]) @{ "x", "y", "z" @};
1632 Compound literals for scalar types and union types are
1633 also allowed, but then the compound literal is equivalent
1636 As a GNU extension, GCC allows initialization of objects with static storage
1637 duration by compound literals (which is not possible in ISO C99, because
1638 the initializer is not a constant).
1639 It is handled as if the object was initialized only with the bracket
1640 enclosed list if the types of the compound literal and the object match.
1641 The initializer list of the compound literal must be constant.
1642 If the object being initialized has array type of unknown size, the size is
1643 determined by compound literal size.
1646 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1647 static int y[] = (int []) @{1, 2, 3@};
1648 static int z[] = (int [3]) @{1@};
1652 The above lines are equivalent to the following:
1654 static struct foo x = @{1, 'a', 'b'@};
1655 static int y[] = @{1, 2, 3@};
1656 static int z[] = @{1, 0, 0@};
1659 @node Designated Inits
1660 @section Designated Initializers
1661 @cindex initializers with labeled elements
1662 @cindex labeled elements in initializers
1663 @cindex case labels in initializers
1664 @cindex designated initializers
1666 Standard C90 requires the elements of an initializer to appear in a fixed
1667 order, the same as the order of the elements in the array or structure
1670 In ISO C99 you can give the elements in any order, specifying the array
1671 indices or structure field names they apply to, and GNU C allows this as
1672 an extension in C90 mode as well. This extension is not
1673 implemented in GNU C++.
1675 To specify an array index, write
1676 @samp{[@var{index}] =} before the element value. For example,
1679 int a[6] = @{ [4] = 29, [2] = 15 @};
1686 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1690 The index values must be constant expressions, even if the array being
1691 initialized is automatic.
1693 An alternative syntax for this which has been obsolete since GCC 2.5 but
1694 GCC still accepts is to write @samp{[@var{index}]} before the element
1695 value, with no @samp{=}.
1697 To initialize a range of elements to the same value, write
1698 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1699 extension. For example,
1702 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1706 If the value in it has side-effects, the side-effects will happen only once,
1707 not for each initialized field by the range initializer.
1710 Note that the length of the array is the highest value specified
1713 In a structure initializer, specify the name of a field to initialize
1714 with @samp{.@var{fieldname} =} before the element value. For example,
1715 given the following structure,
1718 struct point @{ int x, y; @};
1722 the following initialization
1725 struct point p = @{ .y = yvalue, .x = xvalue @};
1732 struct point p = @{ xvalue, yvalue @};
1735 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1736 @samp{@var{fieldname}:}, as shown here:
1739 struct point p = @{ y: yvalue, x: xvalue @};
1743 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1744 @dfn{designator}. You can also use a designator (or the obsolete colon
1745 syntax) when initializing a union, to specify which element of the union
1746 should be used. For example,
1749 union foo @{ int i; double d; @};
1751 union foo f = @{ .d = 4 @};
1755 will convert 4 to a @code{double} to store it in the union using
1756 the second element. By contrast, casting 4 to type @code{union foo}
1757 would store it into the union as the integer @code{i}, since it is
1758 an integer. (@xref{Cast to Union}.)
1760 You can combine this technique of naming elements with ordinary C
1761 initialization of successive elements. Each initializer element that
1762 does not have a designator applies to the next consecutive element of the
1763 array or structure. For example,
1766 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1773 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1776 Labeling the elements of an array initializer is especially useful
1777 when the indices are characters or belong to an @code{enum} type.
1782 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1783 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1786 @cindex designator lists
1787 You can also write a series of @samp{.@var{fieldname}} and
1788 @samp{[@var{index}]} designators before an @samp{=} to specify a
1789 nested subobject to initialize; the list is taken relative to the
1790 subobject corresponding to the closest surrounding brace pair. For
1791 example, with the @samp{struct point} declaration above:
1794 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1798 If the same field is initialized multiple times, it will have value from
1799 the last initialization. If any such overridden initialization has
1800 side-effect, it is unspecified whether the side-effect happens or not.
1801 Currently, GCC will discard them and issue a warning.
1804 @section Case Ranges
1806 @cindex ranges in case statements
1808 You can specify a range of consecutive values in a single @code{case} label,
1812 case @var{low} ... @var{high}:
1816 This has the same effect as the proper number of individual @code{case}
1817 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1819 This feature is especially useful for ranges of ASCII character codes:
1825 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1826 it may be parsed wrong when you use it with integer values. For example,
1841 @section Cast to a Union Type
1842 @cindex cast to a union
1843 @cindex union, casting to a
1845 A cast to union type is similar to other casts, except that the type
1846 specified is a union type. You can specify the type either with
1847 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1848 a constructor though, not a cast, and hence does not yield an lvalue like
1849 normal casts. (@xref{Compound Literals}.)
1851 The types that may be cast to the union type are those of the members
1852 of the union. Thus, given the following union and variables:
1855 union foo @{ int i; double d; @};
1861 both @code{x} and @code{y} can be cast to type @code{union foo}.
1863 Using the cast as the right-hand side of an assignment to a variable of
1864 union type is equivalent to storing in a member of the union:
1869 u = (union foo) x @equiv{} u.i = x
1870 u = (union foo) y @equiv{} u.d = y
1873 You can also use the union cast as a function argument:
1876 void hack (union foo);
1878 hack ((union foo) x);
1881 @node Mixed Declarations
1882 @section Mixed Declarations and Code
1883 @cindex mixed declarations and code
1884 @cindex declarations, mixed with code
1885 @cindex code, mixed with declarations
1887 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1888 within compound statements. As an extension, GCC also allows this in
1889 C90 mode. For example, you could do:
1898 Each identifier is visible from where it is declared until the end of
1899 the enclosing block.
1901 @node Function Attributes
1902 @section Declaring Attributes of Functions
1903 @cindex function attributes
1904 @cindex declaring attributes of functions
1905 @cindex functions that never return
1906 @cindex functions that return more than once
1907 @cindex functions that have no side effects
1908 @cindex functions in arbitrary sections
1909 @cindex functions that behave like malloc
1910 @cindex @code{volatile} applied to function
1911 @cindex @code{const} applied to function
1912 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1913 @cindex functions with non-null pointer arguments
1914 @cindex functions that are passed arguments in registers on the 386
1915 @cindex functions that pop the argument stack on the 386
1916 @cindex functions that do not pop the argument stack on the 386
1917 @cindex functions that have different compilation options on the 386
1918 @cindex functions that have different optimization options
1919 @cindex functions that are dynamically resolved
1921 In GNU C, you declare certain things about functions called in your program
1922 which help the compiler optimize function calls and check your code more
1925 The keyword @code{__attribute__} allows you to specify special
1926 attributes when making a declaration. This keyword is followed by an
1927 attribute specification inside double parentheses. The following
1928 attributes are currently defined for functions on all targets:
1929 @code{aligned}, @code{alloc_size}, @code{noreturn},
1930 @code{returns_twice}, @code{noinline}, @code{noclone},
1931 @code{always_inline}, @code{flatten}, @code{pure}, @code{const},
1932 @code{nothrow}, @code{sentinel}, @code{format}, @code{format_arg},
1933 @code{no_instrument_function}, @code{no_split_stack},
1934 @code{section}, @code{constructor},
1935 @code{destructor}, @code{used}, @code{unused}, @code{deprecated},
1936 @code{weak}, @code{malloc}, @code{alias}, @code{ifunc},
1937 @code{warn_unused_result}, @code{nonnull}, @code{gnu_inline},
1938 @code{externally_visible}, @code{hot}, @code{cold}, @code{artificial},
1939 @code{error} and @code{warning}. Several other attributes are defined
1940 for functions on particular target systems. Other attributes,
1941 including @code{section} are supported for variables declarations
1942 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1944 GCC plugins may provide their own attributes.
1946 You may also specify attributes with @samp{__} preceding and following
1947 each keyword. This allows you to use them in header files without
1948 being concerned about a possible macro of the same name. For example,
1949 you may use @code{__noreturn__} instead of @code{noreturn}.
1951 @xref{Attribute Syntax}, for details of the exact syntax for using
1955 @c Keep this table alphabetized by attribute name. Treat _ as space.
1957 @item alias ("@var{target}")
1958 @cindex @code{alias} attribute
1959 The @code{alias} attribute causes the declaration to be emitted as an
1960 alias for another symbol, which must be specified. For instance,
1963 void __f () @{ /* @r{Do something.} */; @}
1964 void f () __attribute__ ((weak, alias ("__f")));
1967 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
1968 mangled name for the target must be used. It is an error if @samp{__f}
1969 is not defined in the same translation unit.
1971 Not all target machines support this attribute.
1973 @item aligned (@var{alignment})
1974 @cindex @code{aligned} attribute
1975 This attribute specifies a minimum alignment for the function,
1978 You cannot use this attribute to decrease the alignment of a function,
1979 only to increase it. However, when you explicitly specify a function
1980 alignment this will override the effect of the
1981 @option{-falign-functions} (@pxref{Optimize Options}) option for this
1984 Note that the effectiveness of @code{aligned} attributes may be
1985 limited by inherent limitations in your linker. On many systems, the
1986 linker is only able to arrange for functions to be aligned up to a
1987 certain maximum alignment. (For some linkers, the maximum supported
1988 alignment may be very very small.) See your linker documentation for
1989 further information.
1991 The @code{aligned} attribute can also be used for variables and fields
1992 (@pxref{Variable Attributes}.)
1995 @cindex @code{alloc_size} attribute
1996 The @code{alloc_size} attribute is used to tell the compiler that the
1997 function return value points to memory, where the size is given by
1998 one or two of the functions parameters. GCC uses this
1999 information to improve the correctness of @code{__builtin_object_size}.
2001 The function parameter(s) denoting the allocated size are specified by
2002 one or two integer arguments supplied to the attribute. The allocated size
2003 is either the value of the single function argument specified or the product
2004 of the two function arguments specified. Argument numbering starts at
2010 void* my_calloc(size_t, size_t) __attribute__((alloc_size(1,2)))
2011 void my_realloc(void*, size_t) __attribute__((alloc_size(2)))
2014 declares that my_calloc will return memory of the size given by
2015 the product of parameter 1 and 2 and that my_realloc will return memory
2016 of the size given by parameter 2.
2019 @cindex @code{always_inline} function attribute
2020 Generally, functions are not inlined unless optimization is specified.
2021 For functions declared inline, this attribute inlines the function even
2022 if no optimization level was specified.
2025 @cindex @code{gnu_inline} function attribute
2026 This attribute should be used with a function which is also declared
2027 with the @code{inline} keyword. It directs GCC to treat the function
2028 as if it were defined in gnu90 mode even when compiling in C99 or
2031 If the function is declared @code{extern}, then this definition of the
2032 function is used only for inlining. In no case is the function
2033 compiled as a standalone function, not even if you take its address
2034 explicitly. Such an address becomes an external reference, as if you
2035 had only declared the function, and had not defined it. This has
2036 almost the effect of a macro. The way to use this is to put a
2037 function definition in a header file with this attribute, and put
2038 another copy of the function, without @code{extern}, in a library
2039 file. The definition in the header file will cause most calls to the
2040 function to be inlined. If any uses of the function remain, they will
2041 refer to the single copy in the library. Note that the two
2042 definitions of the functions need not be precisely the same, although
2043 if they do not have the same effect your program may behave oddly.
2045 In C, if the function is neither @code{extern} nor @code{static}, then
2046 the function is compiled as a standalone function, as well as being
2047 inlined where possible.
2049 This is how GCC traditionally handled functions declared
2050 @code{inline}. Since ISO C99 specifies a different semantics for
2051 @code{inline}, this function attribute is provided as a transition
2052 measure and as a useful feature in its own right. This attribute is
2053 available in GCC 4.1.3 and later. It is available if either of the
2054 preprocessor macros @code{__GNUC_GNU_INLINE__} or
2055 @code{__GNUC_STDC_INLINE__} are defined. @xref{Inline,,An Inline
2056 Function is As Fast As a Macro}.
2058 In C++, this attribute does not depend on @code{extern} in any way,
2059 but it still requires the @code{inline} keyword to enable its special
2063 @cindex @code{artificial} function attribute
2064 This attribute is useful for small inline wrappers which if possible
2065 should appear during debugging as a unit, depending on the debug
2066 info format it will either mean marking the function as artificial
2067 or using the caller location for all instructions within the inlined
2071 @cindex interrupt handler functions
2072 When added to an interrupt handler with the M32C port, causes the
2073 prologue and epilogue to use bank switching to preserve the registers
2074 rather than saving them on the stack.
2077 @cindex @code{flatten} function attribute
2078 Generally, inlining into a function is limited. For a function marked with
2079 this attribute, every call inside this function will be inlined, if possible.
2080 Whether the function itself is considered for inlining depends on its size and
2081 the current inlining parameters.
2083 @item error ("@var{message}")
2084 @cindex @code{error} function attribute
2085 If this attribute is used on a function declaration and a call to such a function
2086 is not eliminated through dead code elimination or other optimizations, an error
2087 which will include @var{message} will be diagnosed. This is useful
2088 for compile time checking, especially together with @code{__builtin_constant_p}
2089 and inline functions where checking the inline function arguments is not
2090 possible through @code{extern char [(condition) ? 1 : -1];} tricks.
2091 While it is possible to leave the function undefined and thus invoke
2092 a link failure, when using this attribute the problem will be diagnosed
2093 earlier and with exact location of the call even in presence of inline
2094 functions or when not emitting debugging information.
2096 @item warning ("@var{message}")
2097 @cindex @code{warning} function attribute
2098 If this attribute is used on a function declaration and a call to such a function
2099 is not eliminated through dead code elimination or other optimizations, a warning
2100 which will include @var{message} will be diagnosed. This is useful
2101 for compile time checking, especially together with @code{__builtin_constant_p}
2102 and inline functions. While it is possible to define the function with
2103 a message in @code{.gnu.warning*} section, when using this attribute the problem
2104 will be diagnosed earlier and with exact location of the call even in presence
2105 of inline functions or when not emitting debugging information.
2108 @cindex functions that do pop the argument stack on the 386
2110 On the Intel 386, the @code{cdecl} attribute causes the compiler to
2111 assume that the calling function will pop off the stack space used to
2112 pass arguments. This is
2113 useful to override the effects of the @option{-mrtd} switch.
2116 @cindex @code{const} function attribute
2117 Many functions do not examine any values except their arguments, and
2118 have no effects except the return value. Basically this is just slightly
2119 more strict class than the @code{pure} attribute below, since function is not
2120 allowed to read global memory.
2122 @cindex pointer arguments
2123 Note that a function that has pointer arguments and examines the data
2124 pointed to must @emph{not} be declared @code{const}. Likewise, a
2125 function that calls a non-@code{const} function usually must not be
2126 @code{const}. It does not make sense for a @code{const} function to
2129 The attribute @code{const} is not implemented in GCC versions earlier
2130 than 2.5. An alternative way to declare that a function has no side
2131 effects, which works in the current version and in some older versions,
2135 typedef int intfn ();
2137 extern const intfn square;
2140 This approach does not work in GNU C++ from 2.6.0 on, since the language
2141 specifies that the @samp{const} must be attached to the return value.
2145 @itemx constructor (@var{priority})
2146 @itemx destructor (@var{priority})
2147 @cindex @code{constructor} function attribute
2148 @cindex @code{destructor} function attribute
2149 The @code{constructor} attribute causes the function to be called
2150 automatically before execution enters @code{main ()}. Similarly, the
2151 @code{destructor} attribute causes the function to be called
2152 automatically after @code{main ()} has completed or @code{exit ()} has
2153 been called. Functions with these attributes are useful for
2154 initializing data that will be used implicitly during the execution of
2157 You may provide an optional integer priority to control the order in
2158 which constructor and destructor functions are run. A constructor
2159 with a smaller priority number runs before a constructor with a larger
2160 priority number; the opposite relationship holds for destructors. So,
2161 if you have a constructor that allocates a resource and a destructor
2162 that deallocates the same resource, both functions typically have the
2163 same priority. The priorities for constructor and destructor
2164 functions are the same as those specified for namespace-scope C++
2165 objects (@pxref{C++ Attributes}).
2167 These attributes are not currently implemented for Objective-C@.
2170 @itemx deprecated (@var{msg})
2171 @cindex @code{deprecated} attribute.
2172 The @code{deprecated} attribute results in a warning if the function
2173 is used anywhere in the source file. This is useful when identifying
2174 functions that are expected to be removed in a future version of a
2175 program. The warning also includes the location of the declaration
2176 of the deprecated function, to enable users to easily find further
2177 information about why the function is deprecated, or what they should
2178 do instead. Note that the warnings only occurs for uses:
2181 int old_fn () __attribute__ ((deprecated));
2183 int (*fn_ptr)() = old_fn;
2186 results in a warning on line 3 but not line 2. The optional msg
2187 argument, which must be a string, will be printed in the warning if
2190 The @code{deprecated} attribute can also be used for variables and
2191 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2194 @cindex @code{disinterrupt} attribute
2195 On MeP targets, this attribute causes the compiler to emit
2196 instructions to disable interrupts for the duration of the given
2200 @cindex @code{__declspec(dllexport)}
2201 On Microsoft Windows targets and Symbian OS targets the
2202 @code{dllexport} attribute causes the compiler to provide a global
2203 pointer to a pointer in a DLL, so that it can be referenced with the
2204 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
2205 name is formed by combining @code{_imp__} and the function or variable
2208 You can use @code{__declspec(dllexport)} as a synonym for
2209 @code{__attribute__ ((dllexport))} for compatibility with other
2212 On systems that support the @code{visibility} attribute, this
2213 attribute also implies ``default'' visibility. It is an error to
2214 explicitly specify any other visibility.
2216 In previous versions of GCC, the @code{dllexport} attribute was ignored
2217 for inlined functions, unless the @option{-fkeep-inline-functions} flag
2218 had been used. The default behaviour now is to emit all dllexported
2219 inline functions; however, this can cause object file-size bloat, in
2220 which case the old behaviour can be restored by using
2221 @option{-fno-keep-inline-dllexport}.
2223 The attribute is also ignored for undefined symbols.
2225 When applied to C++ classes, the attribute marks defined non-inlined
2226 member functions and static data members as exports. Static consts
2227 initialized in-class are not marked unless they are also defined
2230 For Microsoft Windows targets there are alternative methods for
2231 including the symbol in the DLL's export table such as using a
2232 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
2233 the @option{--export-all} linker flag.
2236 @cindex @code{__declspec(dllimport)}
2237 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
2238 attribute causes the compiler to reference a function or variable via
2239 a global pointer to a pointer that is set up by the DLL exporting the
2240 symbol. The attribute implies @code{extern}. On Microsoft Windows
2241 targets, the pointer name is formed by combining @code{_imp__} and the
2242 function or variable name.
2244 You can use @code{__declspec(dllimport)} as a synonym for
2245 @code{__attribute__ ((dllimport))} for compatibility with other
2248 On systems that support the @code{visibility} attribute, this
2249 attribute also implies ``default'' visibility. It is an error to
2250 explicitly specify any other visibility.
2252 Currently, the attribute is ignored for inlined functions. If the
2253 attribute is applied to a symbol @emph{definition}, an error is reported.
2254 If a symbol previously declared @code{dllimport} is later defined, the
2255 attribute is ignored in subsequent references, and a warning is emitted.
2256 The attribute is also overridden by a subsequent declaration as
2259 When applied to C++ classes, the attribute marks non-inlined
2260 member functions and static data members as imports. However, the
2261 attribute is ignored for virtual methods to allow creation of vtables
2264 On the SH Symbian OS target the @code{dllimport} attribute also has
2265 another affect---it can cause the vtable and run-time type information
2266 for a class to be exported. This happens when the class has a
2267 dllimport'ed constructor or a non-inline, non-pure virtual function
2268 and, for either of those two conditions, the class also has an inline
2269 constructor or destructor and has a key function that is defined in
2270 the current translation unit.
2272 For Microsoft Windows based targets the use of the @code{dllimport}
2273 attribute on functions is not necessary, but provides a small
2274 performance benefit by eliminating a thunk in the DLL@. The use of the
2275 @code{dllimport} attribute on imported variables was required on older
2276 versions of the GNU linker, but can now be avoided by passing the
2277 @option{--enable-auto-import} switch to the GNU linker. As with
2278 functions, using the attribute for a variable eliminates a thunk in
2281 One drawback to using this attribute is that a pointer to a
2282 @emph{variable} marked as @code{dllimport} cannot be used as a constant
2283 address. However, a pointer to a @emph{function} with the
2284 @code{dllimport} attribute can be used as a constant initializer; in
2285 this case, the address of a stub function in the import lib is
2286 referenced. On Microsoft Windows targets, the attribute can be disabled
2287 for functions by setting the @option{-mnop-fun-dllimport} flag.
2290 @cindex eight bit data on the H8/300, H8/300H, and H8S
2291 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2292 variable should be placed into the eight bit data section.
2293 The compiler will generate more efficient code for certain operations
2294 on data in the eight bit data area. Note the eight bit data area is limited to
2297 You must use GAS and GLD from GNU binutils version 2.7 or later for
2298 this attribute to work correctly.
2300 @item exception_handler
2301 @cindex exception handler functions on the Blackfin processor
2302 Use this attribute on the Blackfin to indicate that the specified function
2303 is an exception handler. The compiler will generate function entry and
2304 exit sequences suitable for use in an exception handler when this
2305 attribute is present.
2307 @item externally_visible
2308 @cindex @code{externally_visible} attribute.
2309 This attribute, attached to a global variable or function, nullifies
2310 the effect of the @option{-fwhole-program} command-line option, so the
2311 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.
2314 @cindex functions which handle memory bank switching
2315 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
2316 use a calling convention that takes care of switching memory banks when
2317 entering and leaving a function. This calling convention is also the
2318 default when using the @option{-mlong-calls} option.
2320 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
2321 to call and return from a function.
2323 On 68HC11 the compiler will generate a sequence of instructions
2324 to invoke a board-specific routine to switch the memory bank and call the
2325 real function. The board-specific routine simulates a @code{call}.
2326 At the end of a function, it will jump to a board-specific routine
2327 instead of using @code{rts}. The board-specific return routine simulates
2330 On MeP targets this causes the compiler to use a calling convention
2331 which assumes the called function is too far away for the built-in
2334 @item fast_interrupt
2335 @cindex interrupt handler functions
2336 Use this attribute on the M32C and RX ports to indicate that the specified
2337 function is a fast interrupt handler. This is just like the
2338 @code{interrupt} attribute, except that @code{freit} is used to return
2339 instead of @code{reit}.
2342 @cindex functions that pop the argument stack on the 386
2343 On the Intel 386, the @code{fastcall} attribute causes the compiler to
2344 pass the first argument (if of integral type) in the register ECX and
2345 the second argument (if of integral type) in the register EDX@. Subsequent
2346 and other typed arguments are passed on the stack. The called function will
2347 pop the arguments off the stack. If the number of arguments is variable all
2348 arguments are pushed on the stack.
2351 @cindex functions that pop the argument stack on the 386
2352 On the Intel 386, the @code{thiscall} attribute causes the compiler to
2353 pass the first argument (if of integral type) in the register ECX.
2354 Subsequent and other typed arguments are passed on the stack. The called
2355 function will pop the arguments off the stack.
2356 If the number of arguments is variable all arguments are pushed on the
2358 The @code{thiscall} attribute is intended for C++ non-static member functions.
2359 As gcc extension this calling convention can be used for C-functions
2360 and for static member methods.
2362 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2363 @cindex @code{format} function attribute
2365 The @code{format} attribute specifies that a function takes @code{printf},
2366 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
2367 should be type-checked against a format string. For example, the
2372 my_printf (void *my_object, const char *my_format, ...)
2373 __attribute__ ((format (printf, 2, 3)));
2377 causes the compiler to check the arguments in calls to @code{my_printf}
2378 for consistency with the @code{printf} style format string argument
2381 The parameter @var{archetype} determines how the format string is
2382 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime},
2383 @code{gnu_printf}, @code{gnu_scanf}, @code{gnu_strftime} or
2384 @code{strfmon}. (You can also use @code{__printf__},
2385 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) On
2386 MinGW targets, @code{ms_printf}, @code{ms_scanf}, and
2387 @code{ms_strftime} are also present.
2388 @var{archtype} values such as @code{printf} refer to the formats accepted
2389 by the system's C run-time library, while @code{gnu_} values always refer
2390 to the formats accepted by the GNU C Library. On Microsoft Windows
2391 targets, @code{ms_} values refer to the formats accepted by the
2392 @file{msvcrt.dll} library.
2393 The parameter @var{string-index}
2394 specifies which argument is the format string argument (starting
2395 from 1), while @var{first-to-check} is the number of the first
2396 argument to check against the format string. For functions
2397 where the arguments are not available to be checked (such as
2398 @code{vprintf}), specify the third parameter as zero. In this case the
2399 compiler only checks the format string for consistency. For
2400 @code{strftime} formats, the third parameter is required to be zero.
2401 Since non-static C++ methods have an implicit @code{this} argument, the
2402 arguments of such methods should be counted from two, not one, when
2403 giving values for @var{string-index} and @var{first-to-check}.
2405 In the example above, the format string (@code{my_format}) is the second
2406 argument of the function @code{my_print}, and the arguments to check
2407 start with the third argument, so the correct parameters for the format
2408 attribute are 2 and 3.
2410 @opindex ffreestanding
2411 @opindex fno-builtin
2412 The @code{format} attribute allows you to identify your own functions
2413 which take format strings as arguments, so that GCC can check the
2414 calls to these functions for errors. The compiler always (unless
2415 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
2416 for the standard library functions @code{printf}, @code{fprintf},
2417 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2418 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2419 warnings are requested (using @option{-Wformat}), so there is no need to
2420 modify the header file @file{stdio.h}. In C99 mode, the functions
2421 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2422 @code{vsscanf} are also checked. Except in strictly conforming C
2423 standard modes, the X/Open function @code{strfmon} is also checked as
2424 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2425 @xref{C Dialect Options,,Options Controlling C Dialect}.
2427 For Objective-C dialects, @code{NSString} (or @code{__NSString__}) is
2428 recognized in the same context. Declarations including these format attributes
2429 will be parsed for correct syntax, however the result of checking of such format
2430 strings is not yet defined, and will not be carried out by this version of the
2433 The target may also provide additional types of format checks.
2434 @xref{Target Format Checks,,Format Checks Specific to Particular
2437 @item format_arg (@var{string-index})
2438 @cindex @code{format_arg} function attribute
2439 @opindex Wformat-nonliteral
2440 The @code{format_arg} attribute specifies that a function takes a format
2441 string for a @code{printf}, @code{scanf}, @code{strftime} or
2442 @code{strfmon} style function and modifies it (for example, to translate
2443 it into another language), so the result can be passed to a
2444 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2445 function (with the remaining arguments to the format function the same
2446 as they would have been for the unmodified string). For example, the
2451 my_dgettext (char *my_domain, const char *my_format)
2452 __attribute__ ((format_arg (2)));
2456 causes the compiler to check the arguments in calls to a @code{printf},
2457 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2458 format string argument is a call to the @code{my_dgettext} function, for
2459 consistency with the format string argument @code{my_format}. If the
2460 @code{format_arg} attribute had not been specified, all the compiler
2461 could tell in such calls to format functions would be that the format
2462 string argument is not constant; this would generate a warning when
2463 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2464 without the attribute.
2466 The parameter @var{string-index} specifies which argument is the format
2467 string argument (starting from one). Since non-static C++ methods have
2468 an implicit @code{this} argument, the arguments of such methods should
2469 be counted from two.
2471 The @code{format-arg} attribute allows you to identify your own
2472 functions which modify format strings, so that GCC can check the
2473 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2474 type function whose operands are a call to one of your own function.
2475 The compiler always treats @code{gettext}, @code{dgettext}, and
2476 @code{dcgettext} in this manner except when strict ISO C support is
2477 requested by @option{-ansi} or an appropriate @option{-std} option, or
2478 @option{-ffreestanding} or @option{-fno-builtin}
2479 is used. @xref{C Dialect Options,,Options
2480 Controlling C Dialect}.
2482 For Objective-C dialects, the @code{format-arg} attribute may refer to an
2483 @code{NSString} reference for compatibility with the @code{format} attribute
2486 The target may also allow additional types in @code{format-arg} attributes.
2487 @xref{Target Format Checks,,Format Checks Specific to Particular
2490 @item function_vector
2491 @cindex calling functions through the function vector on H8/300, M16C, M32C and SH2A processors
2492 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2493 function should be called through the function vector. Calling a
2494 function through the function vector will reduce code size, however;
2495 the function vector has a limited size (maximum 128 entries on the H8/300
2496 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2498 In SH2A target, this attribute declares a function to be called using the
2499 TBR relative addressing mode. The argument to this attribute is the entry
2500 number of the same function in a vector table containing all the TBR
2501 relative addressable functions. For the successful jump, register TBR
2502 should contain the start address of this TBR relative vector table.
2503 In the startup routine of the user application, user needs to care of this
2504 TBR register initialization. The TBR relative vector table can have at
2505 max 256 function entries. The jumps to these functions will be generated
2506 using a SH2A specific, non delayed branch instruction JSR/N @@(disp8,TBR).
2507 You must use GAS and GLD from GNU binutils version 2.7 or later for
2508 this attribute to work correctly.
2510 Please refer the example of M16C target, to see the use of this
2511 attribute while declaring a function,
2513 In an application, for a function being called once, this attribute will
2514 save at least 8 bytes of code; and if other successive calls are being
2515 made to the same function, it will save 2 bytes of code per each of these
2518 On M16C/M32C targets, the @code{function_vector} attribute declares a
2519 special page subroutine call function. Use of this attribute reduces
2520 the code size by 2 bytes for each call generated to the
2521 subroutine. The argument to the attribute is the vector number entry
2522 from the special page vector table which contains the 16 low-order
2523 bits of the subroutine's entry address. Each vector table has special
2524 page number (18 to 255) which are used in @code{jsrs} instruction.
2525 Jump addresses of the routines are generated by adding 0x0F0000 (in
2526 case of M16C targets) or 0xFF0000 (in case of M32C targets), to the 2
2527 byte addresses set in the vector table. Therefore you need to ensure
2528 that all the special page vector routines should get mapped within the
2529 address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF
2532 In the following example 2 bytes will be saved for each call to
2533 function @code{foo}.
2536 void foo (void) __attribute__((function_vector(0x18)));
2547 If functions are defined in one file and are called in another file,
2548 then be sure to write this declaration in both files.
2550 This attribute is ignored for R8C target.
2553 @cindex interrupt handler functions
2554 Use this attribute on the ARM, AVR, M32C, M32R/D, m68k, MeP, MIPS,
2555 RX and Xstormy16 ports to indicate that the specified function is an
2556 interrupt handler. The compiler will generate function entry and exit
2557 sequences suitable for use in an interrupt handler when this attribute
2560 Note, interrupt handlers for the Blackfin, H8/300, H8/300H, H8S, MicroBlaze,
2561 and SH processors can be specified via the @code{interrupt_handler} attribute.
2563 Note, on the AVR, interrupts will be enabled inside the function.
2565 Note, for the ARM, you can specify the kind of interrupt to be handled by
2566 adding an optional parameter to the interrupt attribute like this:
2569 void f () __attribute__ ((interrupt ("IRQ")));
2572 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2574 On ARMv7-M the interrupt type is ignored, and the attribute means the function
2575 may be called with a word aligned stack pointer.
2577 On MIPS targets, you can use the following attributes to modify the behavior
2578 of an interrupt handler:
2580 @item use_shadow_register_set
2581 @cindex @code{use_shadow_register_set} attribute
2582 Assume that the handler uses a shadow register set, instead of
2583 the main general-purpose registers.
2585 @item keep_interrupts_masked
2586 @cindex @code{keep_interrupts_masked} attribute
2587 Keep interrupts masked for the whole function. Without this attribute,
2588 GCC tries to reenable interrupts for as much of the function as it can.
2590 @item use_debug_exception_return
2591 @cindex @code{use_debug_exception_return} attribute
2592 Return using the @code{deret} instruction. Interrupt handlers that don't
2593 have this attribute return using @code{eret} instead.
2596 You can use any combination of these attributes, as shown below:
2598 void __attribute__ ((interrupt)) v0 ();
2599 void __attribute__ ((interrupt, use_shadow_register_set)) v1 ();
2600 void __attribute__ ((interrupt, keep_interrupts_masked)) v2 ();
2601 void __attribute__ ((interrupt, use_debug_exception_return)) v3 ();
2602 void __attribute__ ((interrupt, use_shadow_register_set,
2603 keep_interrupts_masked)) v4 ();
2604 void __attribute__ ((interrupt, use_shadow_register_set,
2605 use_debug_exception_return)) v5 ();
2606 void __attribute__ ((interrupt, keep_interrupts_masked,
2607 use_debug_exception_return)) v6 ();
2608 void __attribute__ ((interrupt, use_shadow_register_set,
2609 keep_interrupts_masked,
2610 use_debug_exception_return)) v7 ();
2613 @item ifunc ("@var{resolver}")
2614 @cindex @code{ifunc} attribute
2615 The @code{ifunc} attribute is used to mark a function as an indirect
2616 function using the STT_GNU_IFUNC symbol type extension to the ELF
2617 standard. This allows the resolution of the symbol value to be
2618 determined dynamically at load time, and an optimized version of the
2619 routine can be selected for the particular processor or other system
2620 characteristics determined then. To use this attribute, first define
2621 the implementation functions available, and a resolver function that
2622 returns a pointer to the selected implementation function. The
2623 implementation functions' declarations must match the API of the
2624 function being implemented, the resolver's declaration is be a
2625 function returning pointer to void function returning void:
2628 void *my_memcpy (void *dst, const void *src, size_t len)
2633 static void (*resolve_memcpy (void)) (void)
2635 return my_memcpy; // we'll just always select this routine
2639 The exported header file declaring the function the user calls would
2643 extern void *memcpy (void *, const void *, size_t);
2646 allowing the user to call this as a regular function, unaware of the
2647 implementation. Finally, the indirect function needs to be defined in
2648 the same translation unit as the resolver function:
2651 void *memcpy (void *, const void *, size_t)
2652 __attribute__ ((ifunc ("resolve_memcpy")));
2655 Indirect functions cannot be weak, and require a recent binutils (at
2656 least version 2.20.1), and GNU C library (at least version 2.11.1).
2658 @item interrupt_handler
2659 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
2660 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
2661 indicate that the specified function is an interrupt handler. The compiler
2662 will generate function entry and exit sequences suitable for use in an
2663 interrupt handler when this attribute is present.
2665 @item interrupt_thread
2666 @cindex interrupt thread functions on fido
2667 Use this attribute on fido, a subarchitecture of the m68k, to indicate
2668 that the specified function is an interrupt handler that is designed
2669 to run as a thread. The compiler omits generate prologue/epilogue
2670 sequences and replaces the return instruction with a @code{sleep}
2671 instruction. This attribute is available only on fido.
2674 @cindex interrupt service routines on ARM
2675 Use this attribute on ARM to write Interrupt Service Routines. This is an
2676 alias to the @code{interrupt} attribute above.
2679 @cindex User stack pointer in interrupts on the Blackfin
2680 When used together with @code{interrupt_handler}, @code{exception_handler}
2681 or @code{nmi_handler}, code will be generated to load the stack pointer
2682 from the USP register in the function prologue.
2685 @cindex @code{l1_text} function attribute
2686 This attribute specifies a function to be placed into L1 Instruction
2687 SRAM@. The function will be put into a specific section named @code{.l1.text}.
2688 With @option{-mfdpic}, function calls with a such function as the callee
2689 or caller will use inlined PLT.
2692 @cindex @code{l2} function attribute
2693 On the Blackfin, this attribute specifies a function to be placed into L2
2694 SRAM. The function will be put into a specific section named
2695 @code{.l1.text}. With @option{-mfdpic}, callers of such functions will use
2699 @cindex @code{leaf} function attribute
2700 Calls to external functions with this attribute must return to the current
2701 compilation unit only by return or by exception handling. In particular, leaf
2702 functions are not allowed to call callback function passed to it from the current
2703 compilation unit or directly call functions exported by the unit or longjmp
2704 into the unit. Leaf function might still call functions from other compilation
2705 units and thus they are not necessarily leaf in the sense that they contain no
2706 function calls at all.
2708 The attribute is intended for library functions to improve dataflow analysis.
2709 The compiler takes the hint that any data not escaping the current compilation unit can
2710 not be used or modified by the leaf function. For example, the @code{sin} function
2711 is a leaf function, but @code{qsort} is not.
2713 Note that leaf functions might invoke signals and signal handlers might be
2714 defined in the current compilation unit and use static variables. The only
2715 compliant way to write such a signal handler is to declare such variables
2718 The attribute has no effect on functions defined within the current compilation
2719 unit. This is to allow easy merging of multiple compilation units into one,
2720 for example, by using the link time optimization. For this reason the
2721 attribute is not allowed on types to annotate indirect calls.
2723 @item long_call/short_call
2724 @cindex indirect calls on ARM
2725 This attribute specifies how a particular function is called on
2726 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2727 command-line switch and @code{#pragma long_calls} settings. The
2728 @code{long_call} attribute indicates that the function might be far
2729 away from the call site and require a different (more expensive)
2730 calling sequence. The @code{short_call} attribute always places
2731 the offset to the function from the call site into the @samp{BL}
2732 instruction directly.
2734 @item longcall/shortcall
2735 @cindex functions called via pointer on the RS/6000 and PowerPC
2736 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
2737 indicates that the function might be far away from the call site and
2738 require a different (more expensive) calling sequence. The
2739 @code{shortcall} attribute indicates that the function is always close
2740 enough for the shorter calling sequence to be used. These attributes
2741 override both the @option{-mlongcall} switch and, on the RS/6000 and
2742 PowerPC, the @code{#pragma longcall} setting.
2744 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2745 calls are necessary.
2747 @item long_call/near/far
2748 @cindex indirect calls on MIPS
2749 These attributes specify how a particular function is called on MIPS@.
2750 The attributes override the @option{-mlong-calls} (@pxref{MIPS Options})
2751 command-line switch. The @code{long_call} and @code{far} attributes are
2752 synonyms, and cause the compiler to always call
2753 the function by first loading its address into a register, and then using
2754 the contents of that register. The @code{near} attribute has the opposite
2755 effect; it specifies that non-PIC calls should be made using the more
2756 efficient @code{jal} instruction.
2759 @cindex @code{malloc} attribute
2760 The @code{malloc} attribute is used to tell the compiler that a function
2761 may be treated as if any non-@code{NULL} pointer it returns cannot
2762 alias any other pointer valid when the function returns.
2763 This will often improve optimization.
2764 Standard functions with this property include @code{malloc} and
2765 @code{calloc}. @code{realloc}-like functions have this property as
2766 long as the old pointer is never referred to (including comparing it
2767 to the new pointer) after the function returns a non-@code{NULL}
2770 @item mips16/nomips16
2771 @cindex @code{mips16} attribute
2772 @cindex @code{nomips16} attribute
2774 On MIPS targets, you can use the @code{mips16} and @code{nomips16}
2775 function attributes to locally select or turn off MIPS16 code generation.
2776 A function with the @code{mips16} attribute is emitted as MIPS16 code,
2777 while MIPS16 code generation is disabled for functions with the
2778 @code{nomips16} attribute. These attributes override the
2779 @option{-mips16} and @option{-mno-mips16} options on the command line
2780 (@pxref{MIPS Options}).
2782 When compiling files containing mixed MIPS16 and non-MIPS16 code, the
2783 preprocessor symbol @code{__mips16} reflects the setting on the command line,
2784 not that within individual functions. Mixed MIPS16 and non-MIPS16 code
2785 may interact badly with some GCC extensions such as @code{__builtin_apply}
2786 (@pxref{Constructing Calls}).
2788 @item model (@var{model-name})
2789 @cindex function addressability on the M32R/D
2790 @cindex variable addressability on the IA-64
2792 On the M32R/D, use this attribute to set the addressability of an
2793 object, and of the code generated for a function. The identifier
2794 @var{model-name} is one of @code{small}, @code{medium}, or
2795 @code{large}, representing each of the code models.
2797 Small model objects live in the lower 16MB of memory (so that their
2798 addresses can be loaded with the @code{ld24} instruction), and are
2799 callable with the @code{bl} instruction.
2801 Medium model objects may live anywhere in the 32-bit address space (the
2802 compiler will generate @code{seth/add3} instructions to load their addresses),
2803 and are callable with the @code{bl} instruction.
2805 Large model objects may live anywhere in the 32-bit address space (the
2806 compiler will generate @code{seth/add3} instructions to load their addresses),
2807 and may not be reachable with the @code{bl} instruction (the compiler will
2808 generate the much slower @code{seth/add3/jl} instruction sequence).
2810 On IA-64, use this attribute to set the addressability of an object.
2811 At present, the only supported identifier for @var{model-name} is
2812 @code{small}, indicating addressability via ``small'' (22-bit)
2813 addresses (so that their addresses can be loaded with the @code{addl}
2814 instruction). Caveat: such addressing is by definition not position
2815 independent and hence this attribute must not be used for objects
2816 defined by shared libraries.
2818 @item ms_abi/sysv_abi
2819 @cindex @code{ms_abi} attribute
2820 @cindex @code{sysv_abi} attribute
2822 On 64-bit x86_64-*-* targets, you can use an ABI attribute to indicate
2823 which calling convention should be used for a function. The @code{ms_abi}
2824 attribute tells the compiler to use the Microsoft ABI, while the
2825 @code{sysv_abi} attribute tells the compiler to use the ABI used on
2826 GNU/Linux and other systems. The default is to use the Microsoft ABI
2827 when targeting Windows. On all other systems, the default is the AMD ABI.
2829 Note, the @code{ms_abi} attribute for Windows targets currently requires
2830 the @option{-maccumulate-outgoing-args} option.
2832 @item callee_pop_aggregate_return (@var{number})
2833 @cindex @code{callee_pop_aggregate_return} attribute
2835 On 32-bit i?86-*-* targets, you can control by those attribute for
2836 aggregate return in memory, if the caller is responsible to pop the hidden
2837 pointer together with the rest of the arguments - @var{number} equal to
2838 zero -, or if the callee is responsible to pop hidden pointer - @var{number}
2839 equal to one. The default i386 ABI assumes that the callee pops the
2840 stack for hidden pointer.
2842 @item ms_hook_prologue
2843 @cindex @code{ms_hook_prologue} attribute
2845 On 32 bit i[34567]86-*-* targets and 64 bit x86_64-*-* targets, you can use
2846 this function attribute to make gcc generate the "hot-patching" function
2847 prologue used in Win32 API functions in Microsoft Windows XP Service Pack 2
2851 @cindex function without a prologue/epilogue code
2852 Use this attribute on the ARM, AVR, MCORE, RX and SPU ports to indicate that
2853 the specified function does not need prologue/epilogue sequences generated by
2854 the compiler. It is up to the programmer to provide these sequences. The
2855 only statements that can be safely included in naked functions are
2856 @code{asm} statements that do not have operands. All other statements,
2857 including declarations of local variables, @code{if} statements, and so
2858 forth, should be avoided. Naked functions should be used to implement the
2859 body of an assembly function, while allowing the compiler to construct
2860 the requisite function declaration for the assembler.
2863 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2864 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2865 use the normal calling convention based on @code{jsr} and @code{rts}.
2866 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2869 On MeP targets this attribute causes the compiler to assume the called
2870 function is close enough to use the normal calling convention,
2871 overriding the @code{-mtf} command line option.
2874 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2875 Use this attribute together with @code{interrupt_handler},
2876 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2877 entry code should enable nested interrupts or exceptions.
2880 @cindex NMI handler functions on the Blackfin processor
2881 Use this attribute on the Blackfin to indicate that the specified function
2882 is an NMI handler. The compiler will generate function entry and
2883 exit sequences suitable for use in an NMI handler when this
2884 attribute is present.
2886 @item no_instrument_function
2887 @cindex @code{no_instrument_function} function attribute
2888 @opindex finstrument-functions
2889 If @option{-finstrument-functions} is given, profiling function calls will
2890 be generated at entry and exit of most user-compiled functions.
2891 Functions with this attribute will not be so instrumented.
2893 @item no_split_stack
2894 @cindex @code{no_split_stack} function attribute
2895 @opindex fsplit-stack
2896 If @option{-fsplit-stack} is given, functions will have a small
2897 prologue which decides whether to split the stack. Functions with the
2898 @code{no_split_stack} attribute will not have that prologue, and thus
2899 may run with only a small amount of stack space available.
2902 @cindex @code{noinline} function attribute
2903 This function attribute prevents a function from being considered for
2905 @c Don't enumerate the optimizations by name here; we try to be
2906 @c future-compatible with this mechanism.
2907 If the function does not have side-effects, there are optimizations
2908 other than inlining that causes function calls to be optimized away,
2909 although the function call is live. To keep such calls from being
2914 (@pxref{Extended Asm}) in the called function, to serve as a special
2918 @cindex @code{noclone} function attribute
2919 This function attribute prevents a function from being considered for
2920 cloning - a mechanism which produces specialized copies of functions
2921 and which is (currently) performed by interprocedural constant
2924 @item nonnull (@var{arg-index}, @dots{})
2925 @cindex @code{nonnull} function attribute
2926 The @code{nonnull} attribute specifies that some function parameters should
2927 be non-null pointers. For instance, the declaration:
2931 my_memcpy (void *dest, const void *src, size_t len)
2932 __attribute__((nonnull (1, 2)));
2936 causes the compiler to check that, in calls to @code{my_memcpy},
2937 arguments @var{dest} and @var{src} are non-null. If the compiler
2938 determines that a null pointer is passed in an argument slot marked
2939 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2940 is issued. The compiler may also choose to make optimizations based
2941 on the knowledge that certain function arguments will not be null.
2943 If no argument index list is given to the @code{nonnull} attribute,
2944 all pointer arguments are marked as non-null. To illustrate, the
2945 following declaration is equivalent to the previous example:
2949 my_memcpy (void *dest, const void *src, size_t len)
2950 __attribute__((nonnull));
2954 @cindex @code{noreturn} function attribute
2955 A few standard library functions, such as @code{abort} and @code{exit},
2956 cannot return. GCC knows this automatically. Some programs define
2957 their own functions that never return. You can declare them
2958 @code{noreturn} to tell the compiler this fact. For example,
2962 void fatal () __attribute__ ((noreturn));
2965 fatal (/* @r{@dots{}} */)
2967 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2973 The @code{noreturn} keyword tells the compiler to assume that
2974 @code{fatal} cannot return. It can then optimize without regard to what
2975 would happen if @code{fatal} ever did return. This makes slightly
2976 better code. More importantly, it helps avoid spurious warnings of
2977 uninitialized variables.
2979 The @code{noreturn} keyword does not affect the exceptional path when that
2980 applies: a @code{noreturn}-marked function may still return to the caller
2981 by throwing an exception or calling @code{longjmp}.
2983 Do not assume that registers saved by the calling function are
2984 restored before calling the @code{noreturn} function.
2986 It does not make sense for a @code{noreturn} function to have a return
2987 type other than @code{void}.
2989 The attribute @code{noreturn} is not implemented in GCC versions
2990 earlier than 2.5. An alternative way to declare that a function does
2991 not return, which works in the current version and in some older
2992 versions, is as follows:
2995 typedef void voidfn ();
2997 volatile voidfn fatal;
3000 This approach does not work in GNU C++.
3003 @cindex @code{nothrow} function attribute
3004 The @code{nothrow} attribute is used to inform the compiler that a
3005 function cannot throw an exception. For example, most functions in
3006 the standard C library can be guaranteed not to throw an exception
3007 with the notable exceptions of @code{qsort} and @code{bsearch} that
3008 take function pointer arguments. The @code{nothrow} attribute is not
3009 implemented in GCC versions earlier than 3.3.
3012 @cindex @code{optimize} function attribute
3013 The @code{optimize} attribute is used to specify that a function is to
3014 be compiled with different optimization options than specified on the
3015 command line. Arguments can either be numbers or strings. Numbers
3016 are assumed to be an optimization level. Strings that begin with
3017 @code{O} are assumed to be an optimization option, while other options
3018 are assumed to be used with a @code{-f} prefix. You can also use the
3019 @samp{#pragma GCC optimize} pragma to set the optimization options
3020 that affect more than one function.
3021 @xref{Function Specific Option Pragmas}, for details about the
3022 @samp{#pragma GCC optimize} pragma.
3024 This can be used for instance to have frequently executed functions
3025 compiled with more aggressive optimization options that produce faster
3026 and larger code, while other functions can be called with less
3030 @cindex @code{pcs} function attribute
3032 The @code{pcs} attribute can be used to control the calling convention
3033 used for a function on ARM. The attribute takes an argument that specifies
3034 the calling convention to use.
3036 When compiling using the AAPCS ABI (or a variant of that) then valid
3037 values for the argument are @code{"aapcs"} and @code{"aapcs-vfp"}. In
3038 order to use a variant other than @code{"aapcs"} then the compiler must
3039 be permitted to use the appropriate co-processor registers (i.e., the
3040 VFP registers must be available in order to use @code{"aapcs-vfp"}).
3044 /* Argument passed in r0, and result returned in r0+r1. */
3045 double f2d (float) __attribute__((pcs("aapcs")));
3048 Variadic functions always use the @code{"aapcs"} calling convention and
3049 the compiler will reject attempts to specify an alternative.
3052 @cindex @code{pure} function attribute
3053 Many functions have no effects except the return value and their
3054 return value depends only on the parameters and/or global variables.
3055 Such a function can be subject
3056 to common subexpression elimination and loop optimization just as an
3057 arithmetic operator would be. These functions should be declared
3058 with the attribute @code{pure}. For example,
3061 int square (int) __attribute__ ((pure));
3065 says that the hypothetical function @code{square} is safe to call
3066 fewer times than the program says.
3068 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
3069 Interesting non-pure functions are functions with infinite loops or those
3070 depending on volatile memory or other system resource, that may change between
3071 two consecutive calls (such as @code{feof} in a multithreading environment).
3073 The attribute @code{pure} is not implemented in GCC versions earlier
3077 @cindex @code{hot} function attribute
3078 The @code{hot} attribute is used to inform the compiler that a function is a
3079 hot spot of the compiled program. The function is optimized more aggressively
3080 and on many target it is placed into special subsection of the text section so
3081 all hot functions appears close together improving locality.
3083 When profile feedback is available, via @option{-fprofile-use}, hot functions
3084 are automatically detected and this attribute is ignored.
3086 The @code{hot} attribute is not implemented in GCC versions earlier
3090 @cindex @code{cold} function attribute
3091 The @code{cold} attribute is used to inform the compiler that a function is
3092 unlikely executed. The function is optimized for size rather than speed and on
3093 many targets it is placed into special subsection of the text section so all
3094 cold functions appears close together improving code locality of non-cold parts
3095 of program. The paths leading to call of cold functions within code are marked
3096 as unlikely by the branch prediction mechanism. It is thus useful to mark
3097 functions used to handle unlikely conditions, such as @code{perror}, as cold to
3098 improve optimization of hot functions that do call marked functions in rare
3101 When profile feedback is available, via @option{-fprofile-use}, hot functions
3102 are automatically detected and this attribute is ignored.
3104 The @code{cold} attribute is not implemented in GCC versions earlier than 4.3.
3106 @item regparm (@var{number})
3107 @cindex @code{regparm} attribute
3108 @cindex functions that are passed arguments in registers on the 386
3109 On the Intel 386, the @code{regparm} attribute causes the compiler to
3110 pass arguments number one to @var{number} if they are of integral type
3111 in registers EAX, EDX, and ECX instead of on the stack. Functions that
3112 take a variable number of arguments will continue to be passed all of their
3113 arguments on the stack.
3115 Beware that on some ELF systems this attribute is unsuitable for
3116 global functions in shared libraries with lazy binding (which is the
3117 default). Lazy binding will send the first call via resolving code in
3118 the loader, which might assume EAX, EDX and ECX can be clobbered, as
3119 per the standard calling conventions. Solaris 8 is affected by this.
3120 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
3121 safe since the loaders there save EAX, EDX and ECX. (Lazy binding can be
3122 disabled with the linker or the loader if desired, to avoid the
3126 @cindex @code{sseregparm} attribute
3127 On the Intel 386 with SSE support, the @code{sseregparm} attribute
3128 causes the compiler to pass up to 3 floating point arguments in
3129 SSE registers instead of on the stack. Functions that take a
3130 variable number of arguments will continue to pass all of their
3131 floating point arguments on the stack.
3133 @item force_align_arg_pointer
3134 @cindex @code{force_align_arg_pointer} attribute
3135 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
3136 applied to individual function definitions, generating an alternate
3137 prologue and epilogue that realigns the runtime stack if necessary.
3138 This supports mixing legacy codes that run with a 4-byte aligned stack
3139 with modern codes that keep a 16-byte stack for SSE compatibility.
3142 @cindex @code{resbank} attribute
3143 On the SH2A target, this attribute enables the high-speed register
3144 saving and restoration using a register bank for @code{interrupt_handler}
3145 routines. Saving to the bank is performed automatically after the CPU
3146 accepts an interrupt that uses a register bank.
3148 The nineteen 32-bit registers comprising general register R0 to R14,
3149 control register GBR, and system registers MACH, MACL, and PR and the
3150 vector table address offset are saved into a register bank. Register
3151 banks are stacked in first-in last-out (FILO) sequence. Restoration
3152 from the bank is executed by issuing a RESBANK instruction.
3155 @cindex @code{returns_twice} attribute
3156 The @code{returns_twice} attribute tells the compiler that a function may
3157 return more than one time. The compiler will ensure that all registers
3158 are dead before calling such a function and will emit a warning about
3159 the variables that may be clobbered after the second return from the
3160 function. Examples of such functions are @code{setjmp} and @code{vfork}.
3161 The @code{longjmp}-like counterpart of such function, if any, might need
3162 to be marked with the @code{noreturn} attribute.
3165 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
3166 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
3167 all registers except the stack pointer should be saved in the prologue
3168 regardless of whether they are used or not.
3170 @item save_volatiles
3171 @cindex save volatile registers on the MicroBlaze
3172 Use this attribute on the MicroBlaze to indicate that the function is
3173 an interrupt handler. All volatile registers (in addition to non-volatile
3174 registers) will be saved in the function prologue. If the function is a leaf
3175 function, only volatiles used by the function are saved. A normal function
3176 return is generated instead of a return from interrupt.
3178 @item section ("@var{section-name}")
3179 @cindex @code{section} function attribute
3180 Normally, the compiler places the code it generates in the @code{text} section.
3181 Sometimes, however, you need additional sections, or you need certain
3182 particular functions to appear in special sections. The @code{section}
3183 attribute specifies that a function lives in a particular section.
3184 For example, the declaration:
3187 extern void foobar (void) __attribute__ ((section ("bar")));
3191 puts the function @code{foobar} in the @code{bar} section.
3193 Some file formats do not support arbitrary sections so the @code{section}
3194 attribute is not available on all platforms.
3195 If you need to map the entire contents of a module to a particular
3196 section, consider using the facilities of the linker instead.
3199 @cindex @code{sentinel} function attribute
3200 This function attribute ensures that a parameter in a function call is
3201 an explicit @code{NULL}. The attribute is only valid on variadic
3202 functions. By default, the sentinel is located at position zero, the
3203 last parameter of the function call. If an optional integer position
3204 argument P is supplied to the attribute, the sentinel must be located at
3205 position P counting backwards from the end of the argument list.
3208 __attribute__ ((sentinel))
3210 __attribute__ ((sentinel(0)))
3213 The attribute is automatically set with a position of 0 for the built-in
3214 functions @code{execl} and @code{execlp}. The built-in function
3215 @code{execle} has the attribute set with a position of 1.
3217 A valid @code{NULL} in this context is defined as zero with any pointer
3218 type. If your system defines the @code{NULL} macro with an integer type
3219 then you need to add an explicit cast. GCC replaces @code{stddef.h}
3220 with a copy that redefines NULL appropriately.
3222 The warnings for missing or incorrect sentinels are enabled with
3226 See long_call/short_call.
3229 See longcall/shortcall.
3232 @cindex signal handler functions on the AVR processors
3233 Use this attribute on the AVR to indicate that the specified
3234 function is a signal handler. The compiler will generate function
3235 entry and exit sequences suitable for use in a signal handler when this
3236 attribute is present. Interrupts will be disabled inside the function.
3239 Use this attribute on the SH to indicate an @code{interrupt_handler}
3240 function should switch to an alternate stack. It expects a string
3241 argument that names a global variable holding the address of the
3246 void f () __attribute__ ((interrupt_handler,
3247 sp_switch ("alt_stack")));
3251 @cindex functions that pop the argument stack on the 386
3252 On the Intel 386, the @code{stdcall} attribute causes the compiler to
3253 assume that the called function will pop off the stack space used to
3254 pass arguments, unless it takes a variable number of arguments.
3256 @item syscall_linkage
3257 @cindex @code{syscall_linkage} attribute
3258 This attribute is used to modify the IA64 calling convention by marking
3259 all input registers as live at all function exits. This makes it possible
3260 to restart a system call after an interrupt without having to save/restore
3261 the input registers. This also prevents kernel data from leaking into
3265 @cindex @code{target} function attribute
3266 The @code{target} attribute is used to specify that a function is to
3267 be compiled with different target options than specified on the
3268 command line. This can be used for instance to have functions
3269 compiled with a different ISA (instruction set architecture) than the
3270 default. You can also use the @samp{#pragma GCC target} pragma to set
3271 more than one function to be compiled with specific target options.
3272 @xref{Function Specific Option Pragmas}, for details about the
3273 @samp{#pragma GCC target} pragma.
3275 For instance on a 386, you could compile one function with
3276 @code{target("sse4.1,arch=core2")} and another with
3277 @code{target("sse4a,arch=amdfam10")} that would be equivalent to
3278 compiling the first function with @option{-msse4.1} and
3279 @option{-march=core2} options, and the second function with
3280 @option{-msse4a} and @option{-march=amdfam10} options. It is up to the
3281 user to make sure that a function is only invoked on a machine that
3282 supports the particular ISA it was compiled for (for example by using
3283 @code{cpuid} on 386 to determine what feature bits and architecture
3287 int core2_func (void) __attribute__ ((__target__ ("arch=core2")));
3288 int sse3_func (void) __attribute__ ((__target__ ("sse3")));
3291 On the 386, the following options are allowed:
3296 @cindex @code{target("abm")} attribute
3297 Enable/disable the generation of the advanced bit instructions.
3301 @cindex @code{target("aes")} attribute
3302 Enable/disable the generation of the AES instructions.
3306 @cindex @code{target("mmx")} attribute
3307 Enable/disable the generation of the MMX instructions.
3311 @cindex @code{target("pclmul")} attribute
3312 Enable/disable the generation of the PCLMUL instructions.
3316 @cindex @code{target("popcnt")} attribute
3317 Enable/disable the generation of the POPCNT instruction.
3321 @cindex @code{target("sse")} attribute
3322 Enable/disable the generation of the SSE instructions.
3326 @cindex @code{target("sse2")} attribute
3327 Enable/disable the generation of the SSE2 instructions.
3331 @cindex @code{target("sse3")} attribute
3332 Enable/disable the generation of the SSE3 instructions.
3336 @cindex @code{target("sse4")} attribute
3337 Enable/disable the generation of the SSE4 instructions (both SSE4.1
3342 @cindex @code{target("sse4.1")} attribute
3343 Enable/disable the generation of the sse4.1 instructions.
3347 @cindex @code{target("sse4.2")} attribute
3348 Enable/disable the generation of the sse4.2 instructions.
3352 @cindex @code{target("sse4a")} attribute
3353 Enable/disable the generation of the SSE4A instructions.
3357 @cindex @code{target("fma4")} attribute
3358 Enable/disable the generation of the FMA4 instructions.
3362 @cindex @code{target("xop")} attribute
3363 Enable/disable the generation of the XOP instructions.
3367 @cindex @code{target("lwp")} attribute
3368 Enable/disable the generation of the LWP instructions.
3372 @cindex @code{target("ssse3")} attribute
3373 Enable/disable the generation of the SSSE3 instructions.
3377 @cindex @code{target("cld")} attribute
3378 Enable/disable the generation of the CLD before string moves.
3380 @item fancy-math-387
3381 @itemx no-fancy-math-387
3382 @cindex @code{target("fancy-math-387")} attribute
3383 Enable/disable the generation of the @code{sin}, @code{cos}, and
3384 @code{sqrt} instructions on the 387 floating point unit.
3387 @itemx no-fused-madd
3388 @cindex @code{target("fused-madd")} attribute
3389 Enable/disable the generation of the fused multiply/add instructions.
3393 @cindex @code{target("ieee-fp")} attribute
3394 Enable/disable the generation of floating point that depends on IEEE arithmetic.
3396 @item inline-all-stringops
3397 @itemx no-inline-all-stringops
3398 @cindex @code{target("inline-all-stringops")} attribute
3399 Enable/disable inlining of string operations.
3401 @item inline-stringops-dynamically
3402 @itemx no-inline-stringops-dynamically
3403 @cindex @code{target("inline-stringops-dynamically")} attribute
3404 Enable/disable the generation of the inline code to do small string
3405 operations and calling the library routines for large operations.
3407 @item align-stringops
3408 @itemx no-align-stringops
3409 @cindex @code{target("align-stringops")} attribute
3410 Do/do not align destination of inlined string operations.
3414 @cindex @code{target("recip")} attribute
3415 Enable/disable the generation of RCPSS, RCPPS, RSQRTSS and RSQRTPS
3416 instructions followed an additional Newton-Raphson step instead of
3417 doing a floating point division.
3419 @item arch=@var{ARCH}
3420 @cindex @code{target("arch=@var{ARCH}")} attribute
3421 Specify the architecture to generate code for in compiling the function.
3423 @item tune=@var{TUNE}
3424 @cindex @code{target("tune=@var{TUNE}")} attribute
3425 Specify the architecture to tune for in compiling the function.
3427 @item fpmath=@var{FPMATH}
3428 @cindex @code{target("fpmath=@var{FPMATH}")} attribute
3429 Specify which floating point unit to use. The
3430 @code{target("fpmath=sse,387")} option must be specified as
3431 @code{target("fpmath=sse+387")} because the comma would separate
3435 On the PowerPC, the following options are allowed:
3440 @cindex @code{target("altivec")} attribute
3441 Generate code that uses (does not use) AltiVec instructions. In
3442 32-bit code, you cannot enable Altivec instructions unless
3443 @option{-mabi=altivec} was used on the command line.
3447 @cindex @code{target("cmpb")} attribute
3448 Generate code that uses (does not use) the compare bytes instruction
3449 implemented on the POWER6 processor and other processors that support
3450 the PowerPC V2.05 architecture.
3454 @cindex @code{target("dlmzb")} attribute
3455 Generate code that uses (does not use) the string-search @samp{dlmzb}
3456 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
3457 generated by default when targetting those processors.
3461 @cindex @code{target("fprnd")} attribute
3462 Generate code that uses (does not use) the FP round to integer
3463 instructions implemented on the POWER5+ processor and other processors
3464 that support the PowerPC V2.03 architecture.
3468 @cindex @code{target("hard-dfp")} attribute
3469 Generate code that uses (does not use) the decimal floating point
3470 instructions implemented on some POWER processors.
3474 @cindex @code{target("isel")} attribute
3475 Generate code that uses (does not use) ISEL instruction.
3479 @cindex @code{target("mfcrf")} attribute
3480 Generate code that uses (does not use) the move from condition
3481 register field instruction implemented on the POWER4 processor and
3482 other processors that support the PowerPC V2.01 architecture.
3486 @cindex @code{target("mfpgpr")} attribute
3487 Generate code that uses (does not use) the FP move to/from general
3488 purpose register instructions implemented on the POWER6X processor and
3489 other processors that support the extended PowerPC V2.05 architecture.
3493 @cindex @code{target("mulhw")} attribute
3494 Generate code that uses (does not use) the half-word multiply and
3495 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
3496 These instructions are generated by default when targetting those
3501 @cindex @code{target("multiple")} attribute
3502 Generate code that uses (does not use) the load multiple word
3503 instructions and the store multiple word instructions.
3507 @cindex @code{target("update")} attribute
3508 Generate code that uses (does not use) the load or store instructions
3509 that update the base register to the address of the calculated memory
3514 @cindex @code{target("popcntb")} attribute
3515 Generate code that uses (does not use) the popcount and double
3516 precision FP reciprocal estimate instruction implemented on the POWER5
3517 processor and other processors that support the PowerPC V2.02
3522 @cindex @code{target("popcntd")} attribute
3523 Generate code that uses (does not use) the popcount instruction
3524 implemented on the POWER7 processor and other processors that support
3525 the PowerPC V2.06 architecture.
3527 @item powerpc-gfxopt
3528 @itemx no-powerpc-gfxopt
3529 @cindex @code{target("powerpc-gfxopt")} attribute
3530 Generate code that uses (does not use) the optional PowerPC
3531 architecture instructions in the Graphics group, including
3532 floating-point select.
3535 @itemx no-powerpc-gpopt
3536 @cindex @code{target("powerpc-gpopt")} attribute
3537 Generate code that uses (does not use) the optional PowerPC
3538 architecture instructions in the General Purpose group, including
3539 floating-point square root.
3541 @item recip-precision
3542 @itemx no-recip-precision
3543 @cindex @code{target("recip-precision")} attribute
3544 Assume (do not assume) that the reciprocal estimate instructions
3545 provide higher precision estimates than is mandated by the powerpc
3550 @cindex @code{target("string")} attribute
3551 Generate code that uses (does not use) the load string instructions
3552 and the store string word instructions to save multiple registers and
3553 do small block moves.
3557 @cindex @code{target("vsx")} attribute
3558 Generate code that uses (does not use) vector/scalar (VSX)
3559 instructions, and also enable the use of built-in functions that allow
3560 more direct access to the VSX instruction set. In 32-bit code, you
3561 cannot enable VSX or Altivec instructions unless
3562 @option{-mabi=altivec} was used on the command line.
3566 @cindex @code{target("friz")} attribute
3567 Generate (do not generate) the @code{friz} instruction when the
3568 @option{-funsafe-math-optimizations} option is used to optimize
3569 rounding a floating point value to 64-bit integer and back to floating
3570 point. The @code{friz} instruction does not return the same value if
3571 the floating point number is too large to fit in an integer.
3573 @item avoid-indexed-addresses
3574 @itemx no-avoid-indexed-addresses
3575 @cindex @code{target("avoid-indexed-addresses")} attribute
3576 Generate code that tries to avoid (not avoid) the use of indexed load
3577 or store instructions.
3581 @cindex @code{target("paired")} attribute
3582 Generate code that uses (does not use) the generation of PAIRED simd
3587 @cindex @code{target("longcall")} attribute
3588 Generate code that assumes (does not assume) that all calls are far
3589 away so that a longer more expensive calling sequence is required.
3592 @cindex @code{target("cpu=@var{CPU}")} attribute
3593 Specify the architecture to generate code for when compiling the
3594 function. If you select the @code{target("cpu=power7")} attribute when
3595 generating 32-bit code, VSX and Altivec instructions are not generated
3596 unless you use the @option{-mabi=altivec} option on the command line.
3598 @item tune=@var{TUNE}
3599 @cindex @code{target("tune=@var{TUNE}")} attribute
3600 Specify the architecture to tune for when compiling the function. If
3601 you do not specify the @code{target("tune=@var{TUNE}")} attribute and
3602 you do specify the @code{target("cpu=@var{CPU}")} attribute,
3603 compilation will tune for the @var{CPU} architecture, and not the
3604 default tuning specified on the command line.
3607 On the 386/x86_64 and PowerPC backends, you can use either multiple
3608 strings to specify multiple options, or you can separate the option
3609 with a comma (@code{,}).
3611 On the 386/x86_64 and PowerPC backends, the inliner will not inline a
3612 function that has different target options than the caller, unless the
3613 callee has a subset of the target options of the caller. For example
3614 a function declared with @code{target("sse3")} can inline a function
3615 with @code{target("sse2")}, since @code{-msse3} implies @code{-msse2}.
3617 The @code{target} attribute is not implemented in GCC versions earlier
3618 than 4.4 for the i386/x86_64 and 4.6 for the PowerPC backends. It is
3619 not currently implemented for other backends.
3622 @cindex tiny data section on the H8/300H and H8S
3623 Use this attribute on the H8/300H and H8S to indicate that the specified
3624 variable should be placed into the tiny data section.
3625 The compiler will generate more efficient code for loads and stores
3626 on data in the tiny data section. Note the tiny data area is limited to
3627 slightly under 32kbytes of data.
3630 Use this attribute on the SH for an @code{interrupt_handler} to return using
3631 @code{trapa} instead of @code{rte}. This attribute expects an integer
3632 argument specifying the trap number to be used.
3635 @cindex @code{unused} attribute.
3636 This attribute, attached to a function, means that the function is meant
3637 to be possibly unused. GCC will not produce a warning for this
3641 @cindex @code{used} attribute.
3642 This attribute, attached to a function, means that code must be emitted
3643 for the function even if it appears that the function is not referenced.
3644 This is useful, for example, when the function is referenced only in
3648 @cindex @code{version_id} attribute
3649 This IA64 HP-UX attribute, attached to a global variable or function, renames a
3650 symbol to contain a version string, thus allowing for function level
3651 versioning. HP-UX system header files may use version level functioning
3652 for some system calls.
3655 extern int foo () __attribute__((version_id ("20040821")));
3658 Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}.
3660 @item visibility ("@var{visibility_type}")
3661 @cindex @code{visibility} attribute
3662 This attribute affects the linkage of the declaration to which it is attached.
3663 There are four supported @var{visibility_type} values: default,
3664 hidden, protected or internal visibility.
3667 void __attribute__ ((visibility ("protected")))
3668 f () @{ /* @r{Do something.} */; @}
3669 int i __attribute__ ((visibility ("hidden")));
3672 The possible values of @var{visibility_type} correspond to the
3673 visibility settings in the ELF gABI.
3676 @c keep this list of visibilities in alphabetical order.
3679 Default visibility is the normal case for the object file format.
3680 This value is available for the visibility attribute to override other
3681 options that may change the assumed visibility of entities.
3683 On ELF, default visibility means that the declaration is visible to other
3684 modules and, in shared libraries, means that the declared entity may be
3687 On Darwin, default visibility means that the declaration is visible to
3690 Default visibility corresponds to ``external linkage'' in the language.
3693 Hidden visibility indicates that the entity declared will have a new
3694 form of linkage, which we'll call ``hidden linkage''. Two
3695 declarations of an object with hidden linkage refer to the same object
3696 if they are in the same shared object.
3699 Internal visibility is like hidden visibility, but with additional
3700 processor specific semantics. Unless otherwise specified by the
3701 psABI, GCC defines internal visibility to mean that a function is
3702 @emph{never} called from another module. Compare this with hidden
3703 functions which, while they cannot be referenced directly by other
3704 modules, can be referenced indirectly via function pointers. By
3705 indicating that a function cannot be called from outside the module,
3706 GCC may for instance omit the load of a PIC register since it is known
3707 that the calling function loaded the correct value.
3710 Protected visibility is like default visibility except that it
3711 indicates that references within the defining module will bind to the
3712 definition in that module. That is, the declared entity cannot be
3713 overridden by another module.
3717 All visibilities are supported on many, but not all, ELF targets
3718 (supported when the assembler supports the @samp{.visibility}
3719 pseudo-op). Default visibility is supported everywhere. Hidden
3720 visibility is supported on Darwin targets.
3722 The visibility attribute should be applied only to declarations which
3723 would otherwise have external linkage. The attribute should be applied
3724 consistently, so that the same entity should not be declared with
3725 different settings of the attribute.
3727 In C++, the visibility attribute applies to types as well as functions
3728 and objects, because in C++ types have linkage. A class must not have
3729 greater visibility than its non-static data member types and bases,
3730 and class members default to the visibility of their class. Also, a
3731 declaration without explicit visibility is limited to the visibility
3734 In C++, you can mark member functions and static member variables of a
3735 class with the visibility attribute. This is useful if you know a
3736 particular method or static member variable should only be used from
3737 one shared object; then you can mark it hidden while the rest of the
3738 class has default visibility. Care must be taken to avoid breaking
3739 the One Definition Rule; for example, it is usually not useful to mark
3740 an inline method as hidden without marking the whole class as hidden.
3742 A C++ namespace declaration can also have the visibility attribute.
3743 This attribute applies only to the particular namespace body, not to
3744 other definitions of the same namespace; it is equivalent to using
3745 @samp{#pragma GCC visibility} before and after the namespace
3746 definition (@pxref{Visibility Pragmas}).
3748 In C++, if a template argument has limited visibility, this
3749 restriction is implicitly propagated to the template instantiation.
3750 Otherwise, template instantiations and specializations default to the
3751 visibility of their template.
3753 If both the template and enclosing class have explicit visibility, the
3754 visibility from the template is used.
3757 @cindex @code{vliw} attribute
3758 On MeP, the @code{vliw} attribute tells the compiler to emit
3759 instructions in VLIW mode instead of core mode. Note that this
3760 attribute is not allowed unless a VLIW coprocessor has been configured
3761 and enabled through command line options.
3763 @item warn_unused_result
3764 @cindex @code{warn_unused_result} attribute
3765 The @code{warn_unused_result} attribute causes a warning to be emitted
3766 if a caller of the function with this attribute does not use its
3767 return value. This is useful for functions where not checking
3768 the result is either a security problem or always a bug, such as
3772 int fn () __attribute__ ((warn_unused_result));
3775 if (fn () < 0) return -1;
3781 results in warning on line 5.
3784 @cindex @code{weak} attribute
3785 The @code{weak} attribute causes the declaration to be emitted as a weak
3786 symbol rather than a global. This is primarily useful in defining
3787 library functions which can be overridden in user code, though it can
3788 also be used with non-function declarations. Weak symbols are supported
3789 for ELF targets, and also for a.out targets when using the GNU assembler
3793 @itemx weakref ("@var{target}")
3794 @cindex @code{weakref} attribute
3795 The @code{weakref} attribute marks a declaration as a weak reference.
3796 Without arguments, it should be accompanied by an @code{alias} attribute
3797 naming the target symbol. Optionally, the @var{target} may be given as
3798 an argument to @code{weakref} itself. In either case, @code{weakref}
3799 implicitly marks the declaration as @code{weak}. Without a
3800 @var{target}, given as an argument to @code{weakref} or to @code{alias},
3801 @code{weakref} is equivalent to @code{weak}.
3804 static int x() __attribute__ ((weakref ("y")));
3805 /* is equivalent to... */
3806 static int x() __attribute__ ((weak, weakref, alias ("y")));
3808 static int x() __attribute__ ((weakref));
3809 static int x() __attribute__ ((alias ("y")));
3812 A weak reference is an alias that does not by itself require a
3813 definition to be given for the target symbol. If the target symbol is
3814 only referenced through weak references, then it becomes a @code{weak}
3815 undefined symbol. If it is directly referenced, however, then such
3816 strong references prevail, and a definition will be required for the
3817 symbol, not necessarily in the same translation unit.
3819 The effect is equivalent to moving all references to the alias to a
3820 separate translation unit, renaming the alias to the aliased symbol,
3821 declaring it as weak, compiling the two separate translation units and
3822 performing a reloadable link on them.
3824 At present, a declaration to which @code{weakref} is attached can
3825 only be @code{static}.
3829 You can specify multiple attributes in a declaration by separating them
3830 by commas within the double parentheses or by immediately following an
3831 attribute declaration with another attribute declaration.
3833 @cindex @code{#pragma}, reason for not using
3834 @cindex pragma, reason for not using
3835 Some people object to the @code{__attribute__} feature, suggesting that
3836 ISO C's @code{#pragma} should be used instead. At the time
3837 @code{__attribute__} was designed, there were two reasons for not doing
3842 It is impossible to generate @code{#pragma} commands from a macro.
3845 There is no telling what the same @code{#pragma} might mean in another
3849 These two reasons applied to almost any application that might have been
3850 proposed for @code{#pragma}. It was basically a mistake to use
3851 @code{#pragma} for @emph{anything}.
3853 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
3854 to be generated from macros. In addition, a @code{#pragma GCC}
3855 namespace is now in use for GCC-specific pragmas. However, it has been
3856 found convenient to use @code{__attribute__} to achieve a natural
3857 attachment of attributes to their corresponding declarations, whereas
3858 @code{#pragma GCC} is of use for constructs that do not naturally form
3859 part of the grammar. @xref{Other Directives,,Miscellaneous
3860 Preprocessing Directives, cpp, The GNU C Preprocessor}.
3862 @node Attribute Syntax
3863 @section Attribute Syntax
3864 @cindex attribute syntax
3866 This section describes the syntax with which @code{__attribute__} may be
3867 used, and the constructs to which attribute specifiers bind, for the C
3868 language. Some details may vary for C++ and Objective-C@. Because of
3869 infelicities in the grammar for attributes, some forms described here
3870 may not be successfully parsed in all cases.
3872 There are some problems with the semantics of attributes in C++. For
3873 example, there are no manglings for attributes, although they may affect
3874 code generation, so problems may arise when attributed types are used in
3875 conjunction with templates or overloading. Similarly, @code{typeid}
3876 does not distinguish between types with different attributes. Support
3877 for attributes in C++ may be restricted in future to attributes on
3878 declarations only, but not on nested declarators.
3880 @xref{Function Attributes}, for details of the semantics of attributes
3881 applying to functions. @xref{Variable Attributes}, for details of the
3882 semantics of attributes applying to variables. @xref{Type Attributes},
3883 for details of the semantics of attributes applying to structure, union
3884 and enumerated types.
3886 An @dfn{attribute specifier} is of the form
3887 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
3888 is a possibly empty comma-separated sequence of @dfn{attributes}, where
3889 each attribute is one of the following:
3893 Empty. Empty attributes are ignored.
3896 A word (which may be an identifier such as @code{unused}, or a reserved
3897 word such as @code{const}).
3900 A word, followed by, in parentheses, parameters for the attribute.
3901 These parameters take one of the following forms:
3905 An identifier. For example, @code{mode} attributes use this form.
3908 An identifier followed by a comma and a non-empty comma-separated list
3909 of expressions. For example, @code{format} attributes use this form.
3912 A possibly empty comma-separated list of expressions. For example,
3913 @code{format_arg} attributes use this form with the list being a single
3914 integer constant expression, and @code{alias} attributes use this form
3915 with the list being a single string constant.
3919 An @dfn{attribute specifier list} is a sequence of one or more attribute
3920 specifiers, not separated by any other tokens.
3922 In GNU C, an attribute specifier list may appear after the colon following a
3923 label, other than a @code{case} or @code{default} label. The only
3924 attribute it makes sense to use after a label is @code{unused}. This
3925 feature is intended for code generated by programs which contains labels
3926 that may be unused but which is compiled with @option{-Wall}. It would
3927 not normally be appropriate to use in it human-written code, though it
3928 could be useful in cases where the code that jumps to the label is
3929 contained within an @code{#ifdef} conditional. GNU C++ only permits
3930 attributes on labels if the attribute specifier is immediately
3931 followed by a semicolon (i.e., the label applies to an empty
3932 statement). If the semicolon is missing, C++ label attributes are
3933 ambiguous, as it is permissible for a declaration, which could begin
3934 with an attribute list, to be labelled in C++. Declarations cannot be
3935 labelled in C90 or C99, so the ambiguity does not arise there.
3937 An attribute specifier list may appear as part of a @code{struct},
3938 @code{union} or @code{enum} specifier. It may go either immediately
3939 after the @code{struct}, @code{union} or @code{enum} keyword, or after
3940 the closing brace. The former syntax is preferred.
3941 Where attribute specifiers follow the closing brace, they are considered
3942 to relate to the structure, union or enumerated type defined, not to any
3943 enclosing declaration the type specifier appears in, and the type
3944 defined is not complete until after the attribute specifiers.
3945 @c Otherwise, there would be the following problems: a shift/reduce
3946 @c conflict between attributes binding the struct/union/enum and
3947 @c binding to the list of specifiers/qualifiers; and "aligned"
3948 @c attributes could use sizeof for the structure, but the size could be
3949 @c changed later by "packed" attributes.
3951 Otherwise, an attribute specifier appears as part of a declaration,
3952 counting declarations of unnamed parameters and type names, and relates
3953 to that declaration (which may be nested in another declaration, for
3954 example in the case of a parameter declaration), or to a particular declarator
3955 within a declaration. Where an
3956 attribute specifier is applied to a parameter declared as a function or
3957 an array, it should apply to the function or array rather than the
3958 pointer to which the parameter is implicitly converted, but this is not
3959 yet correctly implemented.
3961 Any list of specifiers and qualifiers at the start of a declaration may
3962 contain attribute specifiers, whether or not such a list may in that
3963 context contain storage class specifiers. (Some attributes, however,
3964 are essentially in the nature of storage class specifiers, and only make
3965 sense where storage class specifiers may be used; for example,
3966 @code{section}.) There is one necessary limitation to this syntax: the
3967 first old-style parameter declaration in a function definition cannot
3968 begin with an attribute specifier, because such an attribute applies to
3969 the function instead by syntax described below (which, however, is not
3970 yet implemented in this case). In some other cases, attribute
3971 specifiers are permitted by this grammar but not yet supported by the
3972 compiler. All attribute specifiers in this place relate to the
3973 declaration as a whole. In the obsolescent usage where a type of
3974 @code{int} is implied by the absence of type specifiers, such a list of
3975 specifiers and qualifiers may be an attribute specifier list with no
3976 other specifiers or qualifiers.
3978 At present, the first parameter in a function prototype must have some
3979 type specifier which is not an attribute specifier; this resolves an
3980 ambiguity in the interpretation of @code{void f(int
3981 (__attribute__((foo)) x))}, but is subject to change. At present, if
3982 the parentheses of a function declarator contain only attributes then
3983 those attributes are ignored, rather than yielding an error or warning
3984 or implying a single parameter of type int, but this is subject to
3987 An attribute specifier list may appear immediately before a declarator
3988 (other than the first) in a comma-separated list of declarators in a
3989 declaration of more than one identifier using a single list of
3990 specifiers and qualifiers. Such attribute specifiers apply
3991 only to the identifier before whose declarator they appear. For
3995 __attribute__((noreturn)) void d0 (void),
3996 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
4001 the @code{noreturn} attribute applies to all the functions
4002 declared; the @code{format} attribute only applies to @code{d1}.
4004 An attribute specifier list may appear immediately before the comma,
4005 @code{=} or semicolon terminating the declaration of an identifier other
4006 than a function definition. Such attribute specifiers apply
4007 to the declared object or function. Where an
4008 assembler name for an object or function is specified (@pxref{Asm
4009 Labels}), the attribute must follow the @code{asm}
4012 An attribute specifier list may, in future, be permitted to appear after
4013 the declarator in a function definition (before any old-style parameter
4014 declarations or the function body).
4016 Attribute specifiers may be mixed with type qualifiers appearing inside
4017 the @code{[]} of a parameter array declarator, in the C99 construct by
4018 which such qualifiers are applied to the pointer to which the array is
4019 implicitly converted. Such attribute specifiers apply to the pointer,
4020 not to the array, but at present this is not implemented and they are
4023 An attribute specifier list may appear at the start of a nested
4024 declarator. At present, there are some limitations in this usage: the
4025 attributes correctly apply to the declarator, but for most individual
4026 attributes the semantics this implies are not implemented.
4027 When attribute specifiers follow the @code{*} of a pointer
4028 declarator, they may be mixed with any type qualifiers present.
4029 The following describes the formal semantics of this syntax. It will make the
4030 most sense if you are familiar with the formal specification of
4031 declarators in the ISO C standard.
4033 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
4034 D1}, where @code{T} contains declaration specifiers that specify a type
4035 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
4036 contains an identifier @var{ident}. The type specified for @var{ident}
4037 for derived declarators whose type does not include an attribute
4038 specifier is as in the ISO C standard.
4040 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
4041 and the declaration @code{T D} specifies the type
4042 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
4043 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
4044 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
4046 If @code{D1} has the form @code{*
4047 @var{type-qualifier-and-attribute-specifier-list} D}, and the
4048 declaration @code{T D} specifies the type
4049 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
4050 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
4051 @var{type-qualifier-and-attribute-specifier-list} pointer to @var{Type}'' for
4057 void (__attribute__((noreturn)) ****f) (void);
4061 specifies the type ``pointer to pointer to pointer to pointer to
4062 non-returning function returning @code{void}''. As another example,
4065 char *__attribute__((aligned(8))) *f;
4069 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
4070 Note again that this does not work with most attributes; for example,
4071 the usage of @samp{aligned} and @samp{noreturn} attributes given above
4072 is not yet supported.
4074 For compatibility with existing code written for compiler versions that
4075 did not implement attributes on nested declarators, some laxity is
4076 allowed in the placing of attributes. If an attribute that only applies
4077 to types is applied to a declaration, it will be treated as applying to
4078 the type of that declaration. If an attribute that only applies to
4079 declarations is applied to the type of a declaration, it will be treated
4080 as applying to that declaration; and, for compatibility with code
4081 placing the attributes immediately before the identifier declared, such
4082 an attribute applied to a function return type will be treated as
4083 applying to the function type, and such an attribute applied to an array
4084 element type will be treated as applying to the array type. If an
4085 attribute that only applies to function types is applied to a
4086 pointer-to-function type, it will be treated as applying to the pointer
4087 target type; if such an attribute is applied to a function return type
4088 that is not a pointer-to-function type, it will be treated as applying
4089 to the function type.
4091 @node Function Prototypes
4092 @section Prototypes and Old-Style Function Definitions
4093 @cindex function prototype declarations
4094 @cindex old-style function definitions
4095 @cindex promotion of formal parameters
4097 GNU C extends ISO C to allow a function prototype to override a later
4098 old-style non-prototype definition. Consider the following example:
4101 /* @r{Use prototypes unless the compiler is old-fashioned.} */
4108 /* @r{Prototype function declaration.} */
4109 int isroot P((uid_t));
4111 /* @r{Old-style function definition.} */
4113 isroot (x) /* @r{??? lossage here ???} */
4120 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
4121 not allow this example, because subword arguments in old-style
4122 non-prototype definitions are promoted. Therefore in this example the
4123 function definition's argument is really an @code{int}, which does not
4124 match the prototype argument type of @code{short}.
4126 This restriction of ISO C makes it hard to write code that is portable
4127 to traditional C compilers, because the programmer does not know
4128 whether the @code{uid_t} type is @code{short}, @code{int}, or
4129 @code{long}. Therefore, in cases like these GNU C allows a prototype
4130 to override a later old-style definition. More precisely, in GNU C, a
4131 function prototype argument type overrides the argument type specified
4132 by a later old-style definition if the former type is the same as the
4133 latter type before promotion. Thus in GNU C the above example is
4134 equivalent to the following:
4147 GNU C++ does not support old-style function definitions, so this
4148 extension is irrelevant.
4151 @section C++ Style Comments
4153 @cindex C++ comments
4154 @cindex comments, C++ style
4156 In GNU C, you may use C++ style comments, which start with @samp{//} and
4157 continue until the end of the line. Many other C implementations allow
4158 such comments, and they are included in the 1999 C standard. However,
4159 C++ style comments are not recognized if you specify an @option{-std}
4160 option specifying a version of ISO C before C99, or @option{-ansi}
4161 (equivalent to @option{-std=c90}).
4164 @section Dollar Signs in Identifier Names
4166 @cindex dollar signs in identifier names
4167 @cindex identifier names, dollar signs in
4169 In GNU C, you may normally use dollar signs in identifier names.
4170 This is because many traditional C implementations allow such identifiers.
4171 However, dollar signs in identifiers are not supported on a few target
4172 machines, typically because the target assembler does not allow them.
4174 @node Character Escapes
4175 @section The Character @key{ESC} in Constants
4177 You can use the sequence @samp{\e} in a string or character constant to
4178 stand for the ASCII character @key{ESC}.
4180 @node Variable Attributes
4181 @section Specifying Attributes of Variables
4182 @cindex attribute of variables
4183 @cindex variable attributes
4185 The keyword @code{__attribute__} allows you to specify special
4186 attributes of variables or structure fields. This keyword is followed
4187 by an attribute specification inside double parentheses. Some
4188 attributes are currently defined generically for variables.
4189 Other attributes are defined for variables on particular target
4190 systems. Other attributes are available for functions
4191 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
4192 Other front ends might define more attributes
4193 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
4195 You may also specify attributes with @samp{__} preceding and following
4196 each keyword. This allows you to use them in header files without
4197 being concerned about a possible macro of the same name. For example,
4198 you may use @code{__aligned__} instead of @code{aligned}.
4200 @xref{Attribute Syntax}, for details of the exact syntax for using
4204 @cindex @code{aligned} attribute
4205 @item aligned (@var{alignment})
4206 This attribute specifies a minimum alignment for the variable or
4207 structure field, measured in bytes. For example, the declaration:
4210 int x __attribute__ ((aligned (16))) = 0;
4214 causes the compiler to allocate the global variable @code{x} on a
4215 16-byte boundary. On a 68040, this could be used in conjunction with
4216 an @code{asm} expression to access the @code{move16} instruction which
4217 requires 16-byte aligned operands.
4219 You can also specify the alignment of structure fields. For example, to
4220 create a double-word aligned @code{int} pair, you could write:
4223 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
4227 This is an alternative to creating a union with a @code{double} member
4228 that forces the union to be double-word aligned.
4230 As in the preceding examples, you can explicitly specify the alignment
4231 (in bytes) that you wish the compiler to use for a given variable or
4232 structure field. Alternatively, you can leave out the alignment factor
4233 and just ask the compiler to align a variable or field to the
4234 default alignment for the target architecture you are compiling for.
4235 The default alignment is sufficient for all scalar types, but may not be
4236 enough for all vector types on a target which supports vector operations.
4237 The default alignment is fixed for a particular target ABI.
4239 Gcc also provides a target specific macro @code{__BIGGEST_ALIGNMENT__},
4240 which is the largest alignment ever used for any data type on the
4241 target machine you are compiling for. For example, you could write:
4244 short array[3] __attribute__ ((aligned (__BIGGEST_ALIGNMENT__)));
4247 The compiler automatically sets the alignment for the declared
4248 variable or field to @code{__BIGGEST_ALIGNMENT__}. Doing this can
4249 often make copy operations more efficient, because the compiler can
4250 use whatever instructions copy the biggest chunks of memory when
4251 performing copies to or from the variables or fields that you have
4252 aligned this way. Note that the value of @code{__BIGGEST_ALIGNMENT__}
4253 may change depending on command line options.
4255 When used on a struct, or struct member, the @code{aligned} attribute can
4256 only increase the alignment; in order to decrease it, the @code{packed}
4257 attribute must be specified as well. When used as part of a typedef, the
4258 @code{aligned} attribute can both increase and decrease alignment, and
4259 specifying the @code{packed} attribute will generate a warning.
4261 Note that the effectiveness of @code{aligned} attributes may be limited
4262 by inherent limitations in your linker. On many systems, the linker is
4263 only able to arrange for variables to be aligned up to a certain maximum
4264 alignment. (For some linkers, the maximum supported alignment may
4265 be very very small.) If your linker is only able to align variables
4266 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4267 in an @code{__attribute__} will still only provide you with 8 byte
4268 alignment. See your linker documentation for further information.
4270 The @code{aligned} attribute can also be used for functions
4271 (@pxref{Function Attributes}.)
4273 @item cleanup (@var{cleanup_function})
4274 @cindex @code{cleanup} attribute
4275 The @code{cleanup} attribute runs a function when the variable goes
4276 out of scope. This attribute can only be applied to auto function
4277 scope variables; it may not be applied to parameters or variables
4278 with static storage duration. The function must take one parameter,
4279 a pointer to a type compatible with the variable. The return value
4280 of the function (if any) is ignored.
4282 If @option{-fexceptions} is enabled, then @var{cleanup_function}
4283 will be run during the stack unwinding that happens during the
4284 processing of the exception. Note that the @code{cleanup} attribute
4285 does not allow the exception to be caught, only to perform an action.
4286 It is undefined what happens if @var{cleanup_function} does not
4291 @cindex @code{common} attribute
4292 @cindex @code{nocommon} attribute
4295 The @code{common} attribute requests GCC to place a variable in
4296 ``common'' storage. The @code{nocommon} attribute requests the
4297 opposite---to allocate space for it directly.
4299 These attributes override the default chosen by the
4300 @option{-fno-common} and @option{-fcommon} flags respectively.
4303 @itemx deprecated (@var{msg})
4304 @cindex @code{deprecated} attribute
4305 The @code{deprecated} attribute results in a warning if the variable
4306 is used anywhere in the source file. This is useful when identifying
4307 variables that are expected to be removed in a future version of a
4308 program. The warning also includes the location of the declaration
4309 of the deprecated variable, to enable users to easily find further
4310 information about why the variable is deprecated, or what they should
4311 do instead. Note that the warning only occurs for uses:
4314 extern int old_var __attribute__ ((deprecated));
4316 int new_fn () @{ return old_var; @}
4319 results in a warning on line 3 but not line 2. The optional msg
4320 argument, which must be a string, will be printed in the warning if
4323 The @code{deprecated} attribute can also be used for functions and
4324 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
4326 @item mode (@var{mode})
4327 @cindex @code{mode} attribute
4328 This attribute specifies the data type for the declaration---whichever
4329 type corresponds to the mode @var{mode}. This in effect lets you
4330 request an integer or floating point type according to its width.
4332 You may also specify a mode of @samp{byte} or @samp{__byte__} to
4333 indicate the mode corresponding to a one-byte integer, @samp{word} or
4334 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
4335 or @samp{__pointer__} for the mode used to represent pointers.
4338 @cindex @code{packed} attribute
4339 The @code{packed} attribute specifies that a variable or structure field
4340 should have the smallest possible alignment---one byte for a variable,
4341 and one bit for a field, unless you specify a larger value with the
4342 @code{aligned} attribute.
4344 Here is a structure in which the field @code{x} is packed, so that it
4345 immediately follows @code{a}:
4351 int x[2] __attribute__ ((packed));
4355 @emph{Note:} The 4.1, 4.2 and 4.3 series of GCC ignore the
4356 @code{packed} attribute on bit-fields of type @code{char}. This has
4357 been fixed in GCC 4.4 but the change can lead to differences in the
4358 structure layout. See the documentation of
4359 @option{-Wpacked-bitfield-compat} for more information.
4361 @item section ("@var{section-name}")
4362 @cindex @code{section} variable attribute
4363 Normally, the compiler places the objects it generates in sections like
4364 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
4365 or you need certain particular variables to appear in special sections,
4366 for example to map to special hardware. The @code{section}
4367 attribute specifies that a variable (or function) lives in a particular
4368 section. For example, this small program uses several specific section names:
4371 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
4372 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
4373 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
4374 int init_data __attribute__ ((section ("INITDATA")));
4378 /* @r{Initialize stack pointer} */
4379 init_sp (stack + sizeof (stack));
4381 /* @r{Initialize initialized data} */
4382 memcpy (&init_data, &data, &edata - &data);
4384 /* @r{Turn on the serial ports} */
4391 Use the @code{section} attribute with
4392 @emph{global} variables and not @emph{local} variables,
4393 as shown in the example.
4395 You may use the @code{section} attribute with initialized or
4396 uninitialized global variables but the linker requires
4397 each object be defined once, with the exception that uninitialized
4398 variables tentatively go in the @code{common} (or @code{bss}) section
4399 and can be multiply ``defined''. Using the @code{section} attribute
4400 will change what section the variable goes into and may cause the
4401 linker to issue an error if an uninitialized variable has multiple
4402 definitions. You can force a variable to be initialized with the
4403 @option{-fno-common} flag or the @code{nocommon} attribute.
4405 Some file formats do not support arbitrary sections so the @code{section}
4406 attribute is not available on all platforms.
4407 If you need to map the entire contents of a module to a particular
4408 section, consider using the facilities of the linker instead.
4411 @cindex @code{shared} variable attribute
4412 On Microsoft Windows, in addition to putting variable definitions in a named
4413 section, the section can also be shared among all running copies of an
4414 executable or DLL@. For example, this small program defines shared data
4415 by putting it in a named section @code{shared} and marking the section
4419 int foo __attribute__((section ("shared"), shared)) = 0;
4424 /* @r{Read and write foo. All running
4425 copies see the same value.} */
4431 You may only use the @code{shared} attribute along with @code{section}
4432 attribute with a fully initialized global definition because of the way
4433 linkers work. See @code{section} attribute for more information.
4435 The @code{shared} attribute is only available on Microsoft Windows@.
4437 @item tls_model ("@var{tls_model}")
4438 @cindex @code{tls_model} attribute
4439 The @code{tls_model} attribute sets thread-local storage model
4440 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
4441 overriding @option{-ftls-model=} command-line switch on a per-variable
4443 The @var{tls_model} argument should be one of @code{global-dynamic},
4444 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
4446 Not all targets support this attribute.
4449 This attribute, attached to a variable, means that the variable is meant
4450 to be possibly unused. GCC will not produce a warning for this
4454 This attribute, attached to a variable, means that the variable must be
4455 emitted even if it appears that the variable is not referenced.
4457 @item vector_size (@var{bytes})
4458 This attribute specifies the vector size for the variable, measured in
4459 bytes. For example, the declaration:
4462 int foo __attribute__ ((vector_size (16)));
4466 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
4467 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
4468 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
4470 This attribute is only applicable to integral and float scalars,
4471 although arrays, pointers, and function return values are allowed in
4472 conjunction with this construct.
4474 Aggregates with this attribute are invalid, even if they are of the same
4475 size as a corresponding scalar. For example, the declaration:
4478 struct S @{ int a; @};
4479 struct S __attribute__ ((vector_size (16))) foo;
4483 is invalid even if the size of the structure is the same as the size of
4487 The @code{selectany} attribute causes an initialized global variable to
4488 have link-once semantics. When multiple definitions of the variable are
4489 encountered by the linker, the first is selected and the remainder are
4490 discarded. Following usage by the Microsoft compiler, the linker is told
4491 @emph{not} to warn about size or content differences of the multiple
4494 Although the primary usage of this attribute is for POD types, the
4495 attribute can also be applied to global C++ objects that are initialized
4496 by a constructor. In this case, the static initialization and destruction
4497 code for the object is emitted in each translation defining the object,
4498 but the calls to the constructor and destructor are protected by a
4499 link-once guard variable.
4501 The @code{selectany} attribute is only available on Microsoft Windows
4502 targets. You can use @code{__declspec (selectany)} as a synonym for
4503 @code{__attribute__ ((selectany))} for compatibility with other
4507 The @code{weak} attribute is described in @ref{Function Attributes}.
4510 The @code{dllimport} attribute is described in @ref{Function Attributes}.
4513 The @code{dllexport} attribute is described in @ref{Function Attributes}.
4517 @subsection Blackfin Variable Attributes
4519 Three attributes are currently defined for the Blackfin.
4525 @cindex @code{l1_data} variable attribute
4526 @cindex @code{l1_data_A} variable attribute
4527 @cindex @code{l1_data_B} variable attribute
4528 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
4529 Variables with @code{l1_data} attribute will be put into the specific section
4530 named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into
4531 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
4532 attribute will be put into the specific section named @code{.l1.data.B}.
4535 @cindex @code{l2} variable attribute
4536 Use this attribute on the Blackfin to place the variable into L2 SRAM.
4537 Variables with @code{l2} attribute will be put into the specific section
4538 named @code{.l2.data}.
4541 @subsection M32R/D Variable Attributes
4543 One attribute is currently defined for the M32R/D@.
4546 @item model (@var{model-name})
4547 @cindex variable addressability on the M32R/D
4548 Use this attribute on the M32R/D to set the addressability of an object.
4549 The identifier @var{model-name} is one of @code{small}, @code{medium},
4550 or @code{large}, representing each of the code models.
4552 Small model objects live in the lower 16MB of memory (so that their
4553 addresses can be loaded with the @code{ld24} instruction).
4555 Medium and large model objects may live anywhere in the 32-bit address space
4556 (the compiler will generate @code{seth/add3} instructions to load their
4560 @anchor{MeP Variable Attributes}
4561 @subsection MeP Variable Attributes
4563 The MeP target has a number of addressing modes and busses. The
4564 @code{near} space spans the standard memory space's first 16 megabytes
4565 (24 bits). The @code{far} space spans the entire 32-bit memory space.
4566 The @code{based} space is a 128 byte region in the memory space which
4567 is addressed relative to the @code{$tp} register. The @code{tiny}
4568 space is a 65536 byte region relative to the @code{$gp} register. In
4569 addition to these memory regions, the MeP target has a separate 16-bit
4570 control bus which is specified with @code{cb} attributes.
4575 Any variable with the @code{based} attribute will be assigned to the
4576 @code{.based} section, and will be accessed with relative to the
4577 @code{$tp} register.
4580 Likewise, the @code{tiny} attribute assigned variables to the
4581 @code{.tiny} section, relative to the @code{$gp} register.
4584 Variables with the @code{near} attribute are assumed to have addresses
4585 that fit in a 24-bit addressing mode. This is the default for large
4586 variables (@code{-mtiny=4} is the default) but this attribute can
4587 override @code{-mtiny=} for small variables, or override @code{-ml}.
4590 Variables with the @code{far} attribute are addressed using a full
4591 32-bit address. Since this covers the entire memory space, this
4592 allows modules to make no assumptions about where variables might be
4596 @itemx io (@var{addr})
4597 Variables with the @code{io} attribute are used to address
4598 memory-mapped peripherals. If an address is specified, the variable
4599 is assigned that address, else it is not assigned an address (it is
4600 assumed some other module will assign an address). Example:
4603 int timer_count __attribute__((io(0x123)));
4607 @itemx cb (@var{addr})
4608 Variables with the @code{cb} attribute are used to access the control
4609 bus, using special instructions. @code{addr} indicates the control bus
4613 int cpu_clock __attribute__((cb(0x123)));
4618 @anchor{i386 Variable Attributes}
4619 @subsection i386 Variable Attributes
4621 Two attributes are currently defined for i386 configurations:
4622 @code{ms_struct} and @code{gcc_struct}
4627 @cindex @code{ms_struct} attribute
4628 @cindex @code{gcc_struct} attribute
4630 If @code{packed} is used on a structure, or if bit-fields are used
4631 it may be that the Microsoft ABI packs them differently
4632 than GCC would normally pack them. Particularly when moving packed
4633 data between functions compiled with GCC and the native Microsoft compiler
4634 (either via function call or as data in a file), it may be necessary to access
4637 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4638 compilers to match the native Microsoft compiler.
4640 The Microsoft structure layout algorithm is fairly simple with the exception
4641 of the bitfield packing:
4643 The padding and alignment of members of structures and whether a bit field
4644 can straddle a storage-unit boundary
4647 @item Structure members are stored sequentially in the order in which they are
4648 declared: the first member has the lowest memory address and the last member
4651 @item Every data object has an alignment-requirement. The alignment-requirement
4652 for all data except structures, unions, and arrays is either the size of the
4653 object or the current packing size (specified with either the aligned attribute
4654 or the pack pragma), whichever is less. For structures, unions, and arrays,
4655 the alignment-requirement is the largest alignment-requirement of its members.
4656 Every object is allocated an offset so that:
4658 offset % alignment-requirement == 0
4660 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
4661 unit if the integral types are the same size and if the next bit field fits
4662 into the current allocation unit without crossing the boundary imposed by the
4663 common alignment requirements of the bit fields.
4666 Handling of zero-length bitfields:
4668 MSVC interprets zero-length bitfields in the following ways:
4671 @item If a zero-length bitfield is inserted between two bitfields that would
4672 normally be coalesced, the bitfields will not be coalesced.
4679 unsigned long bf_1 : 12;
4681 unsigned long bf_2 : 12;
4685 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
4686 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
4688 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
4689 alignment of the zero-length bitfield is greater than the member that follows it,
4690 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
4710 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
4711 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
4712 bitfield will not affect the alignment of @code{bar} or, as a result, the size
4715 Taking this into account, it is important to note the following:
4718 @item If a zero-length bitfield follows a normal bitfield, the type of the
4719 zero-length bitfield may affect the alignment of the structure as whole. For
4720 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
4721 normal bitfield, and is of type short.
4723 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
4724 still affect the alignment of the structure:
4734 Here, @code{t4} will take up 4 bytes.
4737 @item Zero-length bitfields following non-bitfield members are ignored:
4748 Here, @code{t5} will take up 2 bytes.
4752 @subsection PowerPC Variable Attributes
4754 Three attributes currently are defined for PowerPC configurations:
4755 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4757 For full documentation of the struct attributes please see the
4758 documentation in @ref{i386 Variable Attributes}.
4760 For documentation of @code{altivec} attribute please see the
4761 documentation in @ref{PowerPC Type Attributes}.
4763 @subsection SPU Variable Attributes
4765 The SPU supports the @code{spu_vector} attribute for variables. For
4766 documentation of this attribute please see the documentation in
4767 @ref{SPU Type Attributes}.
4769 @subsection Xstormy16 Variable Attributes
4771 One attribute is currently defined for xstormy16 configurations:
4776 @cindex @code{below100} attribute
4778 If a variable has the @code{below100} attribute (@code{BELOW100} is
4779 allowed also), GCC will place the variable in the first 0x100 bytes of
4780 memory and use special opcodes to access it. Such variables will be
4781 placed in either the @code{.bss_below100} section or the
4782 @code{.data_below100} section.
4786 @subsection AVR Variable Attributes
4790 @cindex @code{progmem} variable attribute
4791 The @code{progmem} attribute is used on the AVR to place data in the Program
4792 Memory address space. The AVR is a Harvard Architecture processor and data
4793 normally resides in the Data Memory address space.
4796 @node Type Attributes
4797 @section Specifying Attributes of Types
4798 @cindex attribute of types
4799 @cindex type attributes
4801 The keyword @code{__attribute__} allows you to specify special
4802 attributes of @code{struct} and @code{union} types when you define
4803 such types. This keyword is followed by an attribute specification
4804 inside double parentheses. Seven attributes are currently defined for
4805 types: @code{aligned}, @code{packed}, @code{transparent_union},
4806 @code{unused}, @code{deprecated}, @code{visibility}, and
4807 @code{may_alias}. Other attributes are defined for functions
4808 (@pxref{Function Attributes}) and for variables (@pxref{Variable
4811 You may also specify any one of these attributes with @samp{__}
4812 preceding and following its keyword. This allows you to use these
4813 attributes in header files without being concerned about a possible
4814 macro of the same name. For example, you may use @code{__aligned__}
4815 instead of @code{aligned}.
4817 You may specify type attributes in an enum, struct or union type
4818 declaration or definition, or for other types in a @code{typedef}
4821 For an enum, struct or union type, you may specify attributes either
4822 between the enum, struct or union tag and the name of the type, or
4823 just past the closing curly brace of the @emph{definition}. The
4824 former syntax is preferred.
4826 @xref{Attribute Syntax}, for details of the exact syntax for using
4830 @cindex @code{aligned} attribute
4831 @item aligned (@var{alignment})
4832 This attribute specifies a minimum alignment (in bytes) for variables
4833 of the specified type. For example, the declarations:
4836 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
4837 typedef int more_aligned_int __attribute__ ((aligned (8)));
4841 force the compiler to insure (as far as it can) that each variable whose
4842 type is @code{struct S} or @code{more_aligned_int} will be allocated and
4843 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
4844 variables of type @code{struct S} aligned to 8-byte boundaries allows
4845 the compiler to use the @code{ldd} and @code{std} (doubleword load and
4846 store) instructions when copying one variable of type @code{struct S} to
4847 another, thus improving run-time efficiency.
4849 Note that the alignment of any given @code{struct} or @code{union} type
4850 is required by the ISO C standard to be at least a perfect multiple of
4851 the lowest common multiple of the alignments of all of the members of
4852 the @code{struct} or @code{union} in question. This means that you @emph{can}
4853 effectively adjust the alignment of a @code{struct} or @code{union}
4854 type by attaching an @code{aligned} attribute to any one of the members
4855 of such a type, but the notation illustrated in the example above is a
4856 more obvious, intuitive, and readable way to request the compiler to
4857 adjust the alignment of an entire @code{struct} or @code{union} type.
4859 As in the preceding example, you can explicitly specify the alignment
4860 (in bytes) that you wish the compiler to use for a given @code{struct}
4861 or @code{union} type. Alternatively, you can leave out the alignment factor
4862 and just ask the compiler to align a type to the maximum
4863 useful alignment for the target machine you are compiling for. For
4864 example, you could write:
4867 struct S @{ short f[3]; @} __attribute__ ((aligned));
4870 Whenever you leave out the alignment factor in an @code{aligned}
4871 attribute specification, the compiler automatically sets the alignment
4872 for the type to the largest alignment which is ever used for any data
4873 type on the target machine you are compiling for. Doing this can often
4874 make copy operations more efficient, because the compiler can use
4875 whatever instructions copy the biggest chunks of memory when performing
4876 copies to or from the variables which have types that you have aligned
4879 In the example above, if the size of each @code{short} is 2 bytes, then
4880 the size of the entire @code{struct S} type is 6 bytes. The smallest
4881 power of two which is greater than or equal to that is 8, so the
4882 compiler sets the alignment for the entire @code{struct S} type to 8
4885 Note that although you can ask the compiler to select a time-efficient
4886 alignment for a given type and then declare only individual stand-alone
4887 objects of that type, the compiler's ability to select a time-efficient
4888 alignment is primarily useful only when you plan to create arrays of
4889 variables having the relevant (efficiently aligned) type. If you
4890 declare or use arrays of variables of an efficiently-aligned type, then
4891 it is likely that your program will also be doing pointer arithmetic (or
4892 subscripting, which amounts to the same thing) on pointers to the
4893 relevant type, and the code that the compiler generates for these
4894 pointer arithmetic operations will often be more efficient for
4895 efficiently-aligned types than for other types.
4897 The @code{aligned} attribute can only increase the alignment; but you
4898 can decrease it by specifying @code{packed} as well. See below.
4900 Note that the effectiveness of @code{aligned} attributes may be limited
4901 by inherent limitations in your linker. On many systems, the linker is
4902 only able to arrange for variables to be aligned up to a certain maximum
4903 alignment. (For some linkers, the maximum supported alignment may
4904 be very very small.) If your linker is only able to align variables
4905 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4906 in an @code{__attribute__} will still only provide you with 8 byte
4907 alignment. See your linker documentation for further information.
4910 This attribute, attached to @code{struct} or @code{union} type
4911 definition, specifies that each member (other than zero-width bitfields)
4912 of the structure or union is placed to minimize the memory required. When
4913 attached to an @code{enum} definition, it indicates that the smallest
4914 integral type should be used.
4916 @opindex fshort-enums
4917 Specifying this attribute for @code{struct} and @code{union} types is
4918 equivalent to specifying the @code{packed} attribute on each of the
4919 structure or union members. Specifying the @option{-fshort-enums}
4920 flag on the line is equivalent to specifying the @code{packed}
4921 attribute on all @code{enum} definitions.
4923 In the following example @code{struct my_packed_struct}'s members are
4924 packed closely together, but the internal layout of its @code{s} member
4925 is not packed---to do that, @code{struct my_unpacked_struct} would need to
4929 struct my_unpacked_struct
4935 struct __attribute__ ((__packed__)) my_packed_struct
4939 struct my_unpacked_struct s;
4943 You may only specify this attribute on the definition of an @code{enum},
4944 @code{struct} or @code{union}, not on a @code{typedef} which does not
4945 also define the enumerated type, structure or union.
4947 @item transparent_union
4948 This attribute, attached to a @code{union} type definition, indicates
4949 that any function parameter having that union type causes calls to that
4950 function to be treated in a special way.
4952 First, the argument corresponding to a transparent union type can be of
4953 any type in the union; no cast is required. Also, if the union contains
4954 a pointer type, the corresponding argument can be a null pointer
4955 constant or a void pointer expression; and if the union contains a void
4956 pointer type, the corresponding argument can be any pointer expression.
4957 If the union member type is a pointer, qualifiers like @code{const} on
4958 the referenced type must be respected, just as with normal pointer
4961 Second, the argument is passed to the function using the calling
4962 conventions of the first member of the transparent union, not the calling
4963 conventions of the union itself. All members of the union must have the
4964 same machine representation; this is necessary for this argument passing
4967 Transparent unions are designed for library functions that have multiple
4968 interfaces for compatibility reasons. For example, suppose the
4969 @code{wait} function must accept either a value of type @code{int *} to
4970 comply with Posix, or a value of type @code{union wait *} to comply with
4971 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
4972 @code{wait} would accept both kinds of arguments, but it would also
4973 accept any other pointer type and this would make argument type checking
4974 less useful. Instead, @code{<sys/wait.h>} might define the interface
4978 typedef union __attribute__ ((__transparent_union__))
4982 @} wait_status_ptr_t;
4984 pid_t wait (wait_status_ptr_t);
4987 This interface allows either @code{int *} or @code{union wait *}
4988 arguments to be passed, using the @code{int *} calling convention.
4989 The program can call @code{wait} with arguments of either type:
4992 int w1 () @{ int w; return wait (&w); @}
4993 int w2 () @{ union wait w; return wait (&w); @}
4996 With this interface, @code{wait}'s implementation might look like this:
4999 pid_t wait (wait_status_ptr_t p)
5001 return waitpid (-1, p.__ip, 0);
5006 When attached to a type (including a @code{union} or a @code{struct}),
5007 this attribute means that variables of that type are meant to appear
5008 possibly unused. GCC will not produce a warning for any variables of
5009 that type, even if the variable appears to do nothing. This is often
5010 the case with lock or thread classes, which are usually defined and then
5011 not referenced, but contain constructors and destructors that have
5012 nontrivial bookkeeping functions.
5015 @itemx deprecated (@var{msg})
5016 The @code{deprecated} attribute results in a warning if the type
5017 is used anywhere in the source file. This is useful when identifying
5018 types that are expected to be removed in a future version of a program.
5019 If possible, the warning also includes the location of the declaration
5020 of the deprecated type, to enable users to easily find further
5021 information about why the type is deprecated, or what they should do
5022 instead. Note that the warnings only occur for uses and then only
5023 if the type is being applied to an identifier that itself is not being
5024 declared as deprecated.
5027 typedef int T1 __attribute__ ((deprecated));
5031 typedef T1 T3 __attribute__ ((deprecated));
5032 T3 z __attribute__ ((deprecated));
5035 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
5036 warning is issued for line 4 because T2 is not explicitly
5037 deprecated. Line 5 has no warning because T3 is explicitly
5038 deprecated. Similarly for line 6. The optional msg
5039 argument, which must be a string, will be printed in the warning if
5042 The @code{deprecated} attribute can also be used for functions and
5043 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
5046 Accesses through pointers to types with this attribute are not subject
5047 to type-based alias analysis, but are instead assumed to be able to alias
5048 any other type of objects. In the context of 6.5/7 an lvalue expression
5049 dereferencing such a pointer is treated like having a character type.
5050 See @option{-fstrict-aliasing} for more information on aliasing issues.
5051 This extension exists to support some vector APIs, in which pointers to
5052 one vector type are permitted to alias pointers to a different vector type.
5054 Note that an object of a type with this attribute does not have any
5060 typedef short __attribute__((__may_alias__)) short_a;
5066 short_a *b = (short_a *) &a;
5070 if (a == 0x12345678)
5077 If you replaced @code{short_a} with @code{short} in the variable
5078 declaration, the above program would abort when compiled with
5079 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
5080 above in recent GCC versions.
5083 In C++, attribute visibility (@pxref{Function Attributes}) can also be
5084 applied to class, struct, union and enum types. Unlike other type
5085 attributes, the attribute must appear between the initial keyword and
5086 the name of the type; it cannot appear after the body of the type.
5088 Note that the type visibility is applied to vague linkage entities
5089 associated with the class (vtable, typeinfo node, etc.). In
5090 particular, if a class is thrown as an exception in one shared object
5091 and caught in another, the class must have default visibility.
5092 Otherwise the two shared objects will be unable to use the same
5093 typeinfo node and exception handling will break.
5097 @subsection ARM Type Attributes
5099 On those ARM targets that support @code{dllimport} (such as Symbian
5100 OS), you can use the @code{notshared} attribute to indicate that the
5101 virtual table and other similar data for a class should not be
5102 exported from a DLL@. For example:
5105 class __declspec(notshared) C @{
5107 __declspec(dllimport) C();
5111 __declspec(dllexport)
5115 In this code, @code{C::C} is exported from the current DLL, but the
5116 virtual table for @code{C} is not exported. (You can use
5117 @code{__attribute__} instead of @code{__declspec} if you prefer, but
5118 most Symbian OS code uses @code{__declspec}.)
5120 @anchor{MeP Type Attributes}
5121 @subsection MeP Type Attributes
5123 Many of the MeP variable attributes may be applied to types as well.
5124 Specifically, the @code{based}, @code{tiny}, @code{near}, and
5125 @code{far} attributes may be applied to either. The @code{io} and
5126 @code{cb} attributes may not be applied to types.
5128 @anchor{i386 Type Attributes}
5129 @subsection i386 Type Attributes
5131 Two attributes are currently defined for i386 configurations:
5132 @code{ms_struct} and @code{gcc_struct}.
5138 @cindex @code{ms_struct}
5139 @cindex @code{gcc_struct}
5141 If @code{packed} is used on a structure, or if bit-fields are used
5142 it may be that the Microsoft ABI packs them differently
5143 than GCC would normally pack them. Particularly when moving packed
5144 data between functions compiled with GCC and the native Microsoft compiler
5145 (either via function call or as data in a file), it may be necessary to access
5148 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
5149 compilers to match the native Microsoft compiler.
5152 To specify multiple attributes, separate them by commas within the
5153 double parentheses: for example, @samp{__attribute__ ((aligned (16),
5156 @anchor{PowerPC Type Attributes}
5157 @subsection PowerPC Type Attributes
5159 Three attributes currently are defined for PowerPC configurations:
5160 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
5162 For full documentation of the @code{ms_struct} and @code{gcc_struct}
5163 attributes please see the documentation in @ref{i386 Type Attributes}.
5165 The @code{altivec} attribute allows one to declare AltiVec vector data
5166 types supported by the AltiVec Programming Interface Manual. The
5167 attribute requires an argument to specify one of three vector types:
5168 @code{vector__}, @code{pixel__} (always followed by unsigned short),
5169 and @code{bool__} (always followed by unsigned).
5172 __attribute__((altivec(vector__)))
5173 __attribute__((altivec(pixel__))) unsigned short
5174 __attribute__((altivec(bool__))) unsigned
5177 These attributes mainly are intended to support the @code{__vector},
5178 @code{__pixel}, and @code{__bool} AltiVec keywords.
5180 @anchor{SPU Type Attributes}
5181 @subsection SPU Type Attributes
5183 The SPU supports the @code{spu_vector} attribute for types. This attribute
5184 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
5185 Language Extensions Specification. It is intended to support the
5186 @code{__vector} keyword.
5189 @section Inquiring on Alignment of Types or Variables
5191 @cindex type alignment
5192 @cindex variable alignment
5194 The keyword @code{__alignof__} allows you to inquire about how an object
5195 is aligned, or the minimum alignment usually required by a type. Its
5196 syntax is just like @code{sizeof}.
5198 For example, if the target machine requires a @code{double} value to be
5199 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
5200 This is true on many RISC machines. On more traditional machine
5201 designs, @code{__alignof__ (double)} is 4 or even 2.
5203 Some machines never actually require alignment; they allow reference to any
5204 data type even at an odd address. For these machines, @code{__alignof__}
5205 reports the smallest alignment that GCC will give the data type, usually as
5206 mandated by the target ABI.
5208 If the operand of @code{__alignof__} is an lvalue rather than a type,
5209 its value is the required alignment for its type, taking into account
5210 any minimum alignment specified with GCC's @code{__attribute__}
5211 extension (@pxref{Variable Attributes}). For example, after this
5215 struct foo @{ int x; char y; @} foo1;
5219 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
5220 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
5222 It is an error to ask for the alignment of an incomplete type.
5226 @section An Inline Function is As Fast As a Macro
5227 @cindex inline functions
5228 @cindex integrating function code
5230 @cindex macros, inline alternative
5232 By declaring a function inline, you can direct GCC to make
5233 calls to that function faster. One way GCC can achieve this is to
5234 integrate that function's code into the code for its callers. This
5235 makes execution faster by eliminating the function-call overhead; in
5236 addition, if any of the actual argument values are constant, their
5237 known values may permit simplifications at compile time so that not
5238 all of the inline function's code needs to be included. The effect on
5239 code size is less predictable; object code may be larger or smaller
5240 with function inlining, depending on the particular case. You can
5241 also direct GCC to try to integrate all ``simple enough'' functions
5242 into their callers with the option @option{-finline-functions}.
5244 GCC implements three different semantics of declaring a function
5245 inline. One is available with @option{-std=gnu89} or
5246 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
5247 on all inline declarations, another when
5248 @option{-std=c99}, @option{-std=c1x},
5249 @option{-std=gnu99} or @option{-std=gnu1x}
5250 (without @option{-fgnu89-inline}), and the third
5251 is used when compiling C++.
5253 To declare a function inline, use the @code{inline} keyword in its
5254 declaration, like this:
5264 If you are writing a header file to be included in ISO C90 programs, write
5265 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
5267 The three types of inlining behave similarly in two important cases:
5268 when the @code{inline} keyword is used on a @code{static} function,
5269 like the example above, and when a function is first declared without
5270 using the @code{inline} keyword and then is defined with
5271 @code{inline}, like this:
5274 extern int inc (int *a);
5282 In both of these common cases, the program behaves the same as if you
5283 had not used the @code{inline} keyword, except for its speed.
5285 @cindex inline functions, omission of
5286 @opindex fkeep-inline-functions
5287 When a function is both inline and @code{static}, if all calls to the
5288 function are integrated into the caller, and the function's address is
5289 never used, then the function's own assembler code is never referenced.
5290 In this case, GCC does not actually output assembler code for the
5291 function, unless you specify the option @option{-fkeep-inline-functions}.
5292 Some calls cannot be integrated for various reasons (in particular,
5293 calls that precede the function's definition cannot be integrated, and
5294 neither can recursive calls within the definition). If there is a
5295 nonintegrated call, then the function is compiled to assembler code as
5296 usual. The function must also be compiled as usual if the program
5297 refers to its address, because that can't be inlined.
5300 Note that certain usages in a function definition can make it unsuitable
5301 for inline substitution. Among these usages are: use of varargs, use of
5302 alloca, use of variable sized data types (@pxref{Variable Length}),
5303 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
5304 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
5305 will warn when a function marked @code{inline} could not be substituted,
5306 and will give the reason for the failure.
5308 @cindex automatic @code{inline} for C++ member fns
5309 @cindex @code{inline} automatic for C++ member fns
5310 @cindex member fns, automatically @code{inline}
5311 @cindex C++ member fns, automatically @code{inline}
5312 @opindex fno-default-inline
5313 As required by ISO C++, GCC considers member functions defined within
5314 the body of a class to be marked inline even if they are
5315 not explicitly declared with the @code{inline} keyword. You can
5316 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
5317 Options,,Options Controlling C++ Dialect}.
5319 GCC does not inline any functions when not optimizing unless you specify
5320 the @samp{always_inline} attribute for the function, like this:
5323 /* @r{Prototype.} */
5324 inline void foo (const char) __attribute__((always_inline));
5327 The remainder of this section is specific to GNU C90 inlining.
5329 @cindex non-static inline function
5330 When an inline function is not @code{static}, then the compiler must assume
5331 that there may be calls from other source files; since a global symbol can
5332 be defined only once in any program, the function must not be defined in
5333 the other source files, so the calls therein cannot be integrated.
5334 Therefore, a non-@code{static} inline function is always compiled on its
5335 own in the usual fashion.
5337 If you specify both @code{inline} and @code{extern} in the function
5338 definition, then the definition is used only for inlining. In no case
5339 is the function compiled on its own, not even if you refer to its
5340 address explicitly. Such an address becomes an external reference, as
5341 if you had only declared the function, and had not defined it.
5343 This combination of @code{inline} and @code{extern} has almost the
5344 effect of a macro. The way to use it is to put a function definition in
5345 a header file with these keywords, and put another copy of the
5346 definition (lacking @code{inline} and @code{extern}) in a library file.
5347 The definition in the header file will cause most calls to the function
5348 to be inlined. If any uses of the function remain, they will refer to
5349 the single copy in the library.
5352 @section When is a Volatile Object Accessed?
5353 @cindex accessing volatiles
5354 @cindex volatile read
5355 @cindex volatile write
5356 @cindex volatile access
5358 C has the concept of volatile objects. These are normally accessed by
5359 pointers and used for accessing hardware or inter-thread
5360 communication. The standard encourages compilers to refrain from
5361 optimizations concerning accesses to volatile objects, but leaves it
5362 implementation defined as to what constitutes a volatile access. The
5363 minimum requirement is that at a sequence point all previous accesses
5364 to volatile objects have stabilized and no subsequent accesses have
5365 occurred. Thus an implementation is free to reorder and combine
5366 volatile accesses which occur between sequence points, but cannot do
5367 so for accesses across a sequence point. The use of volatile does
5368 not allow you to violate the restriction on updating objects multiple
5369 times between two sequence points.
5371 Accesses to non-volatile objects are not ordered with respect to
5372 volatile accesses. You cannot use a volatile object as a memory
5373 barrier to order a sequence of writes to non-volatile memory. For
5377 int *ptr = @var{something};
5379 *ptr = @var{something};
5383 Unless @var{*ptr} and @var{vobj} can be aliased, it is not guaranteed
5384 that the write to @var{*ptr} will have occurred by the time the update
5385 of @var{vobj} has happened. If you need this guarantee, you must use
5386 a stronger memory barrier such as:
5389 int *ptr = @var{something};
5391 *ptr = @var{something};
5392 asm volatile ("" : : : "memory");
5396 A scalar volatile object is read when it is accessed in a void context:
5399 volatile int *src = @var{somevalue};
5403 Such expressions are rvalues, and GCC implements this as a
5404 read of the volatile object being pointed to.
5406 Assignments are also expressions and have an rvalue. However when
5407 assigning to a scalar volatile, the volatile object is not reread,
5408 regardless of whether the assignment expression's rvalue is used or
5409 not. If the assignment's rvalue is used, the value is that assigned
5410 to the volatile object. For instance, there is no read of @var{vobj}
5411 in all the following cases:
5416 vobj = @var{something};
5417 obj = vobj = @var{something};
5418 obj ? vobj = @var{onething} : vobj = @var{anotherthing};
5419 obj = (@var{something}, vobj = @var{anotherthing});
5422 If you need to read the volatile object after an assignment has
5423 occurred, you must use a separate expression with an intervening
5426 As bitfields are not individually addressable, volatile bitfields may
5427 be implicitly read when written to, or when adjacent bitfields are
5428 accessed. Bitfield operations may be optimized such that adjacent
5429 bitfields are only partially accessed, if they straddle a storage unit
5430 boundary. For these reasons it is unwise to use volatile bitfields to
5434 @section Assembler Instructions with C Expression Operands
5435 @cindex extended @code{asm}
5436 @cindex @code{asm} expressions
5437 @cindex assembler instructions
5440 In an assembler instruction using @code{asm}, you can specify the
5441 operands of the instruction using C expressions. This means you need not
5442 guess which registers or memory locations will contain the data you want
5445 You must specify an assembler instruction template much like what
5446 appears in a machine description, plus an operand constraint string for
5449 For example, here is how to use the 68881's @code{fsinx} instruction:
5452 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
5456 Here @code{angle} is the C expression for the input operand while
5457 @code{result} is that of the output operand. Each has @samp{"f"} as its
5458 operand constraint, saying that a floating point register is required.
5459 The @samp{=} in @samp{=f} indicates that the operand is an output; all
5460 output operands' constraints must use @samp{=}. The constraints use the
5461 same language used in the machine description (@pxref{Constraints}).
5463 Each operand is described by an operand-constraint string followed by
5464 the C expression in parentheses. A colon separates the assembler
5465 template from the first output operand and another separates the last
5466 output operand from the first input, if any. Commas separate the
5467 operands within each group. The total number of operands is currently
5468 limited to 30; this limitation may be lifted in some future version of
5471 If there are no output operands but there are input operands, you must
5472 place two consecutive colons surrounding the place where the output
5475 As of GCC version 3.1, it is also possible to specify input and output
5476 operands using symbolic names which can be referenced within the
5477 assembler code. These names are specified inside square brackets
5478 preceding the constraint string, and can be referenced inside the
5479 assembler code using @code{%[@var{name}]} instead of a percentage sign
5480 followed by the operand number. Using named operands the above example
5484 asm ("fsinx %[angle],%[output]"
5485 : [output] "=f" (result)
5486 : [angle] "f" (angle));
5490 Note that the symbolic operand names have no relation whatsoever to
5491 other C identifiers. You may use any name you like, even those of
5492 existing C symbols, but you must ensure that no two operands within the same
5493 assembler construct use the same symbolic name.
5495 Output operand expressions must be lvalues; the compiler can check this.
5496 The input operands need not be lvalues. The compiler cannot check
5497 whether the operands have data types that are reasonable for the
5498 instruction being executed. It does not parse the assembler instruction
5499 template and does not know what it means or even whether it is valid
5500 assembler input. The extended @code{asm} feature is most often used for
5501 machine instructions the compiler itself does not know exist. If
5502 the output expression cannot be directly addressed (for example, it is a
5503 bit-field), your constraint must allow a register. In that case, GCC
5504 will use the register as the output of the @code{asm}, and then store
5505 that register into the output.
5507 The ordinary output operands must be write-only; GCC will assume that
5508 the values in these operands before the instruction are dead and need
5509 not be generated. Extended asm supports input-output or read-write
5510 operands. Use the constraint character @samp{+} to indicate such an
5511 operand and list it with the output operands. You should only use
5512 read-write operands when the constraints for the operand (or the
5513 operand in which only some of the bits are to be changed) allow a
5516 You may, as an alternative, logically split its function into two
5517 separate operands, one input operand and one write-only output
5518 operand. The connection between them is expressed by constraints
5519 which say they need to be in the same location when the instruction
5520 executes. You can use the same C expression for both operands, or
5521 different expressions. For example, here we write the (fictitious)
5522 @samp{combine} instruction with @code{bar} as its read-only source
5523 operand and @code{foo} as its read-write destination:
5526 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
5530 The constraint @samp{"0"} for operand 1 says that it must occupy the
5531 same location as operand 0. A number in constraint is allowed only in
5532 an input operand and it must refer to an output operand.
5534 Only a number in the constraint can guarantee that one operand will be in
5535 the same place as another. The mere fact that @code{foo} is the value
5536 of both operands is not enough to guarantee that they will be in the
5537 same place in the generated assembler code. The following would not
5541 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
5544 Various optimizations or reloading could cause operands 0 and 1 to be in
5545 different registers; GCC knows no reason not to do so. For example, the
5546 compiler might find a copy of the value of @code{foo} in one register and
5547 use it for operand 1, but generate the output operand 0 in a different
5548 register (copying it afterward to @code{foo}'s own address). Of course,
5549 since the register for operand 1 is not even mentioned in the assembler
5550 code, the result will not work, but GCC can't tell that.
5552 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
5553 the operand number for a matching constraint. For example:
5556 asm ("cmoveq %1,%2,%[result]"
5557 : [result] "=r"(result)
5558 : "r" (test), "r"(new), "[result]"(old));
5561 Sometimes you need to make an @code{asm} operand be a specific register,
5562 but there's no matching constraint letter for that register @emph{by
5563 itself}. To force the operand into that register, use a local variable
5564 for the operand and specify the register in the variable declaration.
5565 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
5566 register constraint letter that matches the register:
5569 register int *p1 asm ("r0") = @dots{};
5570 register int *p2 asm ("r1") = @dots{};
5571 register int *result asm ("r0");
5572 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5575 @anchor{Example of asm with clobbered asm reg}
5576 In the above example, beware that a register that is call-clobbered by
5577 the target ABI will be overwritten by any function call in the
5578 assignment, including library calls for arithmetic operators.
5579 Also a register may be clobbered when generating some operations,
5580 like variable shift, memory copy or memory move on x86.
5581 Assuming it is a call-clobbered register, this may happen to @code{r0}
5582 above by the assignment to @code{p2}. If you have to use such a
5583 register, use temporary variables for expressions between the register
5588 register int *p1 asm ("r0") = @dots{};
5589 register int *p2 asm ("r1") = t1;
5590 register int *result asm ("r0");
5591 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5594 Some instructions clobber specific hard registers. To describe this,
5595 write a third colon after the input operands, followed by the names of
5596 the clobbered hard registers (given as strings). Here is a realistic
5597 example for the VAX:
5600 asm volatile ("movc3 %0,%1,%2"
5601 : /* @r{no outputs} */
5602 : "g" (from), "g" (to), "g" (count)
5603 : "r0", "r1", "r2", "r3", "r4", "r5");
5606 You may not write a clobber description in a way that overlaps with an
5607 input or output operand. For example, you may not have an operand
5608 describing a register class with one member if you mention that register
5609 in the clobber list. Variables declared to live in specific registers
5610 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
5611 have no part mentioned in the clobber description.
5612 There is no way for you to specify that an input
5613 operand is modified without also specifying it as an output
5614 operand. Note that if all the output operands you specify are for this
5615 purpose (and hence unused), you will then also need to specify
5616 @code{volatile} for the @code{asm} construct, as described below, to
5617 prevent GCC from deleting the @code{asm} statement as unused.
5619 If you refer to a particular hardware register from the assembler code,
5620 you will probably have to list the register after the third colon to
5621 tell the compiler the register's value is modified. In some assemblers,
5622 the register names begin with @samp{%}; to produce one @samp{%} in the
5623 assembler code, you must write @samp{%%} in the input.
5625 If your assembler instruction can alter the condition code register, add
5626 @samp{cc} to the list of clobbered registers. GCC on some machines
5627 represents the condition codes as a specific hardware register;
5628 @samp{cc} serves to name this register. On other machines, the
5629 condition code is handled differently, and specifying @samp{cc} has no
5630 effect. But it is valid no matter what the machine.
5632 If your assembler instructions access memory in an unpredictable
5633 fashion, add @samp{memory} to the list of clobbered registers. This
5634 will cause GCC to not keep memory values cached in registers across the
5635 assembler instruction and not optimize stores or loads to that memory.
5636 You will also want to add the @code{volatile} keyword if the memory
5637 affected is not listed in the inputs or outputs of the @code{asm}, as
5638 the @samp{memory} clobber does not count as a side-effect of the
5639 @code{asm}. If you know how large the accessed memory is, you can add
5640 it as input or output but if this is not known, you should add
5641 @samp{memory}. As an example, if you access ten bytes of a string, you
5642 can use a memory input like:
5645 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
5648 Note that in the following example the memory input is necessary,
5649 otherwise GCC might optimize the store to @code{x} away:
5656 asm ("magic stuff accessing an 'int' pointed to by '%1'"
5657 "=&d" (r) : "a" (y), "m" (*y));
5662 You can put multiple assembler instructions together in a single
5663 @code{asm} template, separated by the characters normally used in assembly
5664 code for the system. A combination that works in most places is a newline
5665 to break the line, plus a tab character to move to the instruction field
5666 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
5667 assembler allows semicolons as a line-breaking character. Note that some
5668 assembler dialects use semicolons to start a comment.
5669 The input operands are guaranteed not to use any of the clobbered
5670 registers, and neither will the output operands' addresses, so you can
5671 read and write the clobbered registers as many times as you like. Here
5672 is an example of multiple instructions in a template; it assumes the
5673 subroutine @code{_foo} accepts arguments in registers 9 and 10:
5676 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
5678 : "g" (from), "g" (to)
5682 Unless an output operand has the @samp{&} constraint modifier, GCC
5683 may allocate it in the same register as an unrelated input operand, on
5684 the assumption the inputs are consumed before the outputs are produced.
5685 This assumption may be false if the assembler code actually consists of
5686 more than one instruction. In such a case, use @samp{&} for each output
5687 operand that may not overlap an input. @xref{Modifiers}.
5689 If you want to test the condition code produced by an assembler
5690 instruction, you must include a branch and a label in the @code{asm}
5691 construct, as follows:
5694 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
5700 This assumes your assembler supports local labels, as the GNU assembler
5701 and most Unix assemblers do.
5703 Speaking of labels, jumps from one @code{asm} to another are not
5704 supported. The compiler's optimizers do not know about these jumps, and
5705 therefore they cannot take account of them when deciding how to
5706 optimize. @xref{Extended asm with goto}.
5708 @cindex macros containing @code{asm}
5709 Usually the most convenient way to use these @code{asm} instructions is to
5710 encapsulate them in macros that look like functions. For example,
5714 (@{ double __value, __arg = (x); \
5715 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
5720 Here the variable @code{__arg} is used to make sure that the instruction
5721 operates on a proper @code{double} value, and to accept only those
5722 arguments @code{x} which can convert automatically to a @code{double}.
5724 Another way to make sure the instruction operates on the correct data
5725 type is to use a cast in the @code{asm}. This is different from using a
5726 variable @code{__arg} in that it converts more different types. For
5727 example, if the desired type were @code{int}, casting the argument to
5728 @code{int} would accept a pointer with no complaint, while assigning the
5729 argument to an @code{int} variable named @code{__arg} would warn about
5730 using a pointer unless the caller explicitly casts it.
5732 If an @code{asm} has output operands, GCC assumes for optimization
5733 purposes the instruction has no side effects except to change the output
5734 operands. This does not mean instructions with a side effect cannot be
5735 used, but you must be careful, because the compiler may eliminate them
5736 if the output operands aren't used, or move them out of loops, or
5737 replace two with one if they constitute a common subexpression. Also,
5738 if your instruction does have a side effect on a variable that otherwise
5739 appears not to change, the old value of the variable may be reused later
5740 if it happens to be found in a register.
5742 You can prevent an @code{asm} instruction from being deleted
5743 by writing the keyword @code{volatile} after
5744 the @code{asm}. For example:
5747 #define get_and_set_priority(new) \
5749 asm volatile ("get_and_set_priority %0, %1" \
5750 : "=g" (__old) : "g" (new)); \
5755 The @code{volatile} keyword indicates that the instruction has
5756 important side-effects. GCC will not delete a volatile @code{asm} if
5757 it is reachable. (The instruction can still be deleted if GCC can
5758 prove that control-flow will never reach the location of the
5759 instruction.) Note that even a volatile @code{asm} instruction
5760 can be moved relative to other code, including across jump
5761 instructions. For example, on many targets there is a system
5762 register which can be set to control the rounding mode of
5763 floating point operations. You might try
5764 setting it with a volatile @code{asm}, like this PowerPC example:
5767 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
5772 This will not work reliably, as the compiler may move the addition back
5773 before the volatile @code{asm}. To make it work you need to add an
5774 artificial dependency to the @code{asm} referencing a variable in the code
5775 you don't want moved, for example:
5778 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
5782 Similarly, you can't expect a
5783 sequence of volatile @code{asm} instructions to remain perfectly
5784 consecutive. If you want consecutive output, use a single @code{asm}.
5785 Also, GCC will perform some optimizations across a volatile @code{asm}
5786 instruction; GCC does not ``forget everything'' when it encounters
5787 a volatile @code{asm} instruction the way some other compilers do.
5789 An @code{asm} instruction without any output operands will be treated
5790 identically to a volatile @code{asm} instruction.
5792 It is a natural idea to look for a way to give access to the condition
5793 code left by the assembler instruction. However, when we attempted to
5794 implement this, we found no way to make it work reliably. The problem
5795 is that output operands might need reloading, which would result in
5796 additional following ``store'' instructions. On most machines, these
5797 instructions would alter the condition code before there was time to
5798 test it. This problem doesn't arise for ordinary ``test'' and
5799 ``compare'' instructions because they don't have any output operands.
5801 For reasons similar to those described above, it is not possible to give
5802 an assembler instruction access to the condition code left by previous
5805 @anchor{Extended asm with goto}
5806 As of GCC version 4.5, @code{asm goto} may be used to have the assembly
5807 jump to one or more C labels. In this form, a fifth section after the
5808 clobber list contains a list of all C labels to which the assembly may jump.
5809 Each label operand is implicitly self-named. The @code{asm} is also assumed
5810 to fall through to the next statement.
5812 This form of @code{asm} is restricted to not have outputs. This is due
5813 to a internal restriction in the compiler that control transfer instructions
5814 cannot have outputs. This restriction on @code{asm goto} may be lifted
5815 in some future version of the compiler. In the mean time, @code{asm goto}
5816 may include a memory clobber, and so leave outputs in memory.
5822 asm goto ("frob %%r5, %1; jc %l[error]; mov (%2), %%r5"
5823 : : "r"(x), "r"(&y) : "r5", "memory" : error);
5830 In this (inefficient) example, the @code{frob} instruction sets the
5831 carry bit to indicate an error. The @code{jc} instruction detects
5832 this and branches to the @code{error} label. Finally, the output
5833 of the @code{frob} instruction (@code{%r5}) is stored into the memory
5834 for variable @code{y}, which is later read by the @code{return} statement.
5840 asm goto ("mfsr %%r1, 123; jmp %%r1;"
5841 ".pushsection doit_table;"
5842 ".long %l0, %l1, %l2, %l3;"
5844 : : : "r1" : label1, label2, label3, label4);
5845 __builtin_unreachable ();
5860 In this (also inefficient) example, the @code{mfsr} instruction reads
5861 an address from some out-of-band machine register, and the following
5862 @code{jmp} instruction branches to that address. The address read by
5863 the @code{mfsr} instruction is assumed to have been previously set via
5864 some application-specific mechanism to be one of the four values stored
5865 in the @code{doit_table} section. Finally, the @code{asm} is followed
5866 by a call to @code{__builtin_unreachable} to indicate that the @code{asm}
5867 does not in fact fall through.
5870 #define TRACE1(NUM) \
5872 asm goto ("0: nop;" \
5873 ".pushsection trace_table;" \
5876 : : : : trace#NUM); \
5877 if (0) @{ trace#NUM: trace(); @} \
5879 #define TRACE TRACE1(__COUNTER__)
5882 In this example (which in fact inspired the @code{asm goto} feature)
5883 we want on rare occasions to call the @code{trace} function; on other
5884 occasions we'd like to keep the overhead to the absolute minimum.
5885 The normal code path consists of a single @code{nop} instruction.
5886 However, we record the address of this @code{nop} together with the
5887 address of a label that calls the @code{trace} function. This allows
5888 the @code{nop} instruction to be patched at runtime to be an
5889 unconditional branch to the stored label. It is assumed that an
5890 optimizing compiler will move the labeled block out of line, to
5891 optimize the fall through path from the @code{asm}.
5893 If you are writing a header file that should be includable in ISO C
5894 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
5897 @subsection Size of an @code{asm}
5899 Some targets require that GCC track the size of each instruction used in
5900 order to generate correct code. Because the final length of an
5901 @code{asm} is only known by the assembler, GCC must make an estimate as
5902 to how big it will be. The estimate is formed by counting the number of
5903 statements in the pattern of the @code{asm} and multiplying that by the
5904 length of the longest instruction on that processor. Statements in the
5905 @code{asm} are identified by newline characters and whatever statement
5906 separator characters are supported by the assembler; on most processors
5907 this is the `@code{;}' character.
5909 Normally, GCC's estimate is perfectly adequate to ensure that correct
5910 code is generated, but it is possible to confuse the compiler if you use
5911 pseudo instructions or assembler macros that expand into multiple real
5912 instructions or if you use assembler directives that expand to more
5913 space in the object file than would be needed for a single instruction.
5914 If this happens then the assembler will produce a diagnostic saying that
5915 a label is unreachable.
5917 @subsection i386 floating point asm operands
5919 There are several rules on the usage of stack-like regs in
5920 asm_operands insns. These rules apply only to the operands that are
5925 Given a set of input regs that die in an asm_operands, it is
5926 necessary to know which are implicitly popped by the asm, and
5927 which must be explicitly popped by gcc.
5929 An input reg that is implicitly popped by the asm must be
5930 explicitly clobbered, unless it is constrained to match an
5934 For any input reg that is implicitly popped by an asm, it is
5935 necessary to know how to adjust the stack to compensate for the pop.
5936 If any non-popped input is closer to the top of the reg-stack than
5937 the implicitly popped reg, it would not be possible to know what the
5938 stack looked like---it's not clear how the rest of the stack ``slides
5941 All implicitly popped input regs must be closer to the top of
5942 the reg-stack than any input that is not implicitly popped.
5944 It is possible that if an input dies in an insn, reload might
5945 use the input reg for an output reload. Consider this example:
5948 asm ("foo" : "=t" (a) : "f" (b));
5951 This asm says that input B is not popped by the asm, and that
5952 the asm pushes a result onto the reg-stack, i.e., the stack is one
5953 deeper after the asm than it was before. But, it is possible that
5954 reload will think that it can use the same reg for both the input and
5955 the output, if input B dies in this insn.
5957 If any input operand uses the @code{f} constraint, all output reg
5958 constraints must use the @code{&} earlyclobber.
5960 The asm above would be written as
5963 asm ("foo" : "=&t" (a) : "f" (b));
5967 Some operands need to be in particular places on the stack. All
5968 output operands fall in this category---there is no other way to
5969 know which regs the outputs appear in unless the user indicates
5970 this in the constraints.
5972 Output operands must specifically indicate which reg an output
5973 appears in after an asm. @code{=f} is not allowed: the operand
5974 constraints must select a class with a single reg.
5977 Output operands may not be ``inserted'' between existing stack regs.
5978 Since no 387 opcode uses a read/write operand, all output operands
5979 are dead before the asm_operands, and are pushed by the asm_operands.
5980 It makes no sense to push anywhere but the top of the reg-stack.
5982 Output operands must start at the top of the reg-stack: output
5983 operands may not ``skip'' a reg.
5986 Some asm statements may need extra stack space for internal
5987 calculations. This can be guaranteed by clobbering stack registers
5988 unrelated to the inputs and outputs.
5992 Here are a couple of reasonable asms to want to write. This asm
5993 takes one input, which is internally popped, and produces two outputs.
5996 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
5999 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
6000 and replaces them with one output. The user must code the @code{st(1)}
6001 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
6004 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
6010 @section Controlling Names Used in Assembler Code
6011 @cindex assembler names for identifiers
6012 @cindex names used in assembler code
6013 @cindex identifiers, names in assembler code
6015 You can specify the name to be used in the assembler code for a C
6016 function or variable by writing the @code{asm} (or @code{__asm__})
6017 keyword after the declarator as follows:
6020 int foo asm ("myfoo") = 2;
6024 This specifies that the name to be used for the variable @code{foo} in
6025 the assembler code should be @samp{myfoo} rather than the usual
6028 On systems where an underscore is normally prepended to the name of a C
6029 function or variable, this feature allows you to define names for the
6030 linker that do not start with an underscore.
6032 It does not make sense to use this feature with a non-static local
6033 variable since such variables do not have assembler names. If you are
6034 trying to put the variable in a particular register, see @ref{Explicit
6035 Reg Vars}. GCC presently accepts such code with a warning, but will
6036 probably be changed to issue an error, rather than a warning, in the
6039 You cannot use @code{asm} in this way in a function @emph{definition}; but
6040 you can get the same effect by writing a declaration for the function
6041 before its definition and putting @code{asm} there, like this:
6044 extern func () asm ("FUNC");
6051 It is up to you to make sure that the assembler names you choose do not
6052 conflict with any other assembler symbols. Also, you must not use a
6053 register name; that would produce completely invalid assembler code. GCC
6054 does not as yet have the ability to store static variables in registers.
6055 Perhaps that will be added.
6057 @node Explicit Reg Vars
6058 @section Variables in Specified Registers
6059 @cindex explicit register variables
6060 @cindex variables in specified registers
6061 @cindex specified registers
6062 @cindex registers, global allocation
6064 GNU C allows you to put a few global variables into specified hardware
6065 registers. You can also specify the register in which an ordinary
6066 register variable should be allocated.
6070 Global register variables reserve registers throughout the program.
6071 This may be useful in programs such as programming language
6072 interpreters which have a couple of global variables that are accessed
6076 Local register variables in specific registers do not reserve the
6077 registers, except at the point where they are used as input or output
6078 operands in an @code{asm} statement and the @code{asm} statement itself is
6079 not deleted. The compiler's data flow analysis is capable of determining
6080 where the specified registers contain live values, and where they are
6081 available for other uses. Stores into local register variables may be deleted
6082 when they appear to be dead according to dataflow analysis. References
6083 to local register variables may be deleted or moved or simplified.
6085 These local variables are sometimes convenient for use with the extended
6086 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
6087 output of the assembler instruction directly into a particular register.
6088 (This will work provided the register you specify fits the constraints
6089 specified for that operand in the @code{asm}.)
6097 @node Global Reg Vars
6098 @subsection Defining Global Register Variables
6099 @cindex global register variables
6100 @cindex registers, global variables in
6102 You can define a global register variable in GNU C like this:
6105 register int *foo asm ("a5");
6109 Here @code{a5} is the name of the register which should be used. Choose a
6110 register which is normally saved and restored by function calls on your
6111 machine, so that library routines will not clobber it.
6113 Naturally the register name is cpu-dependent, so you would need to
6114 conditionalize your program according to cpu type. The register
6115 @code{a5} would be a good choice on a 68000 for a variable of pointer
6116 type. On machines with register windows, be sure to choose a ``global''
6117 register that is not affected magically by the function call mechanism.
6119 In addition, operating systems on one type of cpu may differ in how they
6120 name the registers; then you would need additional conditionals. For
6121 example, some 68000 operating systems call this register @code{%a5}.
6123 Eventually there may be a way of asking the compiler to choose a register
6124 automatically, but first we need to figure out how it should choose and
6125 how to enable you to guide the choice. No solution is evident.
6127 Defining a global register variable in a certain register reserves that
6128 register entirely for this use, at least within the current compilation.
6129 The register will not be allocated for any other purpose in the functions
6130 in the current compilation. The register will not be saved and restored by
6131 these functions. Stores into this register are never deleted even if they
6132 would appear to be dead, but references may be deleted or moved or
6135 It is not safe to access the global register variables from signal
6136 handlers, or from more than one thread of control, because the system
6137 library routines may temporarily use the register for other things (unless
6138 you recompile them specially for the task at hand).
6140 @cindex @code{qsort}, and global register variables
6141 It is not safe for one function that uses a global register variable to
6142 call another such function @code{foo} by way of a third function
6143 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
6144 different source file in which the variable wasn't declared). This is
6145 because @code{lose} might save the register and put some other value there.
6146 For example, you can't expect a global register variable to be available in
6147 the comparison-function that you pass to @code{qsort}, since @code{qsort}
6148 might have put something else in that register. (If you are prepared to
6149 recompile @code{qsort} with the same global register variable, you can
6150 solve this problem.)
6152 If you want to recompile @code{qsort} or other source files which do not
6153 actually use your global register variable, so that they will not use that
6154 register for any other purpose, then it suffices to specify the compiler
6155 option @option{-ffixed-@var{reg}}. You need not actually add a global
6156 register declaration to their source code.
6158 A function which can alter the value of a global register variable cannot
6159 safely be called from a function compiled without this variable, because it
6160 could clobber the value the caller expects to find there on return.
6161 Therefore, the function which is the entry point into the part of the
6162 program that uses the global register variable must explicitly save and
6163 restore the value which belongs to its caller.
6165 @cindex register variable after @code{longjmp}
6166 @cindex global register after @code{longjmp}
6167 @cindex value after @code{longjmp}
6170 On most machines, @code{longjmp} will restore to each global register
6171 variable the value it had at the time of the @code{setjmp}. On some
6172 machines, however, @code{longjmp} will not change the value of global
6173 register variables. To be portable, the function that called @code{setjmp}
6174 should make other arrangements to save the values of the global register
6175 variables, and to restore them in a @code{longjmp}. This way, the same
6176 thing will happen regardless of what @code{longjmp} does.
6178 All global register variable declarations must precede all function
6179 definitions. If such a declaration could appear after function
6180 definitions, the declaration would be too late to prevent the register from
6181 being used for other purposes in the preceding functions.
6183 Global register variables may not have initial values, because an
6184 executable file has no means to supply initial contents for a register.
6186 On the SPARC, there are reports that g3 @dots{} g7 are suitable
6187 registers, but certain library functions, such as @code{getwd}, as well
6188 as the subroutines for division and remainder, modify g3 and g4. g1 and
6189 g2 are local temporaries.
6191 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
6192 Of course, it will not do to use more than a few of those.
6194 @node Local Reg Vars
6195 @subsection Specifying Registers for Local Variables
6196 @cindex local variables, specifying registers
6197 @cindex specifying registers for local variables
6198 @cindex registers for local variables
6200 You can define a local register variable with a specified register
6204 register int *foo asm ("a5");
6208 Here @code{a5} is the name of the register which should be used. Note
6209 that this is the same syntax used for defining global register
6210 variables, but for a local variable it would appear within a function.
6212 Naturally the register name is cpu-dependent, but this is not a
6213 problem, since specific registers are most often useful with explicit
6214 assembler instructions (@pxref{Extended Asm}). Both of these things
6215 generally require that you conditionalize your program according to
6218 In addition, operating systems on one type of cpu may differ in how they
6219 name the registers; then you would need additional conditionals. For
6220 example, some 68000 operating systems call this register @code{%a5}.
6222 Defining such a register variable does not reserve the register; it
6223 remains available for other uses in places where flow control determines
6224 the variable's value is not live.
6226 This option does not guarantee that GCC will generate code that has
6227 this variable in the register you specify at all times. You may not
6228 code an explicit reference to this register in the @emph{assembler
6229 instruction template} part of an @code{asm} statement and assume it will
6230 always refer to this variable. However, using the variable as an
6231 @code{asm} @emph{operand} guarantees that the specified register is used
6234 Stores into local register variables may be deleted when they appear to be dead
6235 according to dataflow analysis. References to local register variables may
6236 be deleted or moved or simplified.
6238 As for global register variables, it's recommended that you choose a
6239 register which is normally saved and restored by function calls on
6240 your machine, so that library routines will not clobber it. A common
6241 pitfall is to initialize multiple call-clobbered registers with
6242 arbitrary expressions, where a function call or library call for an
6243 arithmetic operator will overwrite a register value from a previous
6244 assignment, for example @code{r0} below:
6246 register int *p1 asm ("r0") = @dots{};
6247 register int *p2 asm ("r1") = @dots{};
6249 In those cases, a solution is to use a temporary variable for
6250 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
6252 @node Alternate Keywords
6253 @section Alternate Keywords
6254 @cindex alternate keywords
6255 @cindex keywords, alternate
6257 @option{-ansi} and the various @option{-std} options disable certain
6258 keywords. This causes trouble when you want to use GNU C extensions, or
6259 a general-purpose header file that should be usable by all programs,
6260 including ISO C programs. The keywords @code{asm}, @code{typeof} and
6261 @code{inline} are not available in programs compiled with
6262 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
6263 program compiled with @option{-std=c99} or @option{-std=c1x}). The
6265 @code{restrict} is only available when @option{-std=gnu99} (which will
6266 eventually be the default) or @option{-std=c99} (or the equivalent
6267 @option{-std=iso9899:1999}), or an option for a later standard
6270 The way to solve these problems is to put @samp{__} at the beginning and
6271 end of each problematical keyword. For example, use @code{__asm__}
6272 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
6274 Other C compilers won't accept these alternative keywords; if you want to
6275 compile with another compiler, you can define the alternate keywords as
6276 macros to replace them with the customary keywords. It looks like this:
6284 @findex __extension__
6286 @option{-pedantic} and other options cause warnings for many GNU C extensions.
6288 prevent such warnings within one expression by writing
6289 @code{__extension__} before the expression. @code{__extension__} has no
6290 effect aside from this.
6292 @node Incomplete Enums
6293 @section Incomplete @code{enum} Types
6295 You can define an @code{enum} tag without specifying its possible values.
6296 This results in an incomplete type, much like what you get if you write
6297 @code{struct foo} without describing the elements. A later declaration
6298 which does specify the possible values completes the type.
6300 You can't allocate variables or storage using the type while it is
6301 incomplete. However, you can work with pointers to that type.
6303 This extension may not be very useful, but it makes the handling of
6304 @code{enum} more consistent with the way @code{struct} and @code{union}
6307 This extension is not supported by GNU C++.
6309 @node Function Names
6310 @section Function Names as Strings
6311 @cindex @code{__func__} identifier
6312 @cindex @code{__FUNCTION__} identifier
6313 @cindex @code{__PRETTY_FUNCTION__} identifier
6315 GCC provides three magic variables which hold the name of the current
6316 function, as a string. The first of these is @code{__func__}, which
6317 is part of the C99 standard:
6319 The identifier @code{__func__} is implicitly declared by the translator
6320 as if, immediately following the opening brace of each function
6321 definition, the declaration
6324 static const char __func__[] = "function-name";
6328 appeared, where function-name is the name of the lexically-enclosing
6329 function. This name is the unadorned name of the function.
6331 @code{__FUNCTION__} is another name for @code{__func__}. Older
6332 versions of GCC recognize only this name. However, it is not
6333 standardized. For maximum portability, we recommend you use
6334 @code{__func__}, but provide a fallback definition with the
6338 #if __STDC_VERSION__ < 199901L
6340 # define __func__ __FUNCTION__
6342 # define __func__ "<unknown>"
6347 In C, @code{__PRETTY_FUNCTION__} is yet another name for
6348 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
6349 the type signature of the function as well as its bare name. For
6350 example, this program:
6354 extern int printf (char *, ...);
6361 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
6362 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
6380 __PRETTY_FUNCTION__ = void a::sub(int)
6383 These identifiers are not preprocessor macros. In GCC 3.3 and
6384 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
6385 were treated as string literals; they could be used to initialize
6386 @code{char} arrays, and they could be concatenated with other string
6387 literals. GCC 3.4 and later treat them as variables, like
6388 @code{__func__}. In C++, @code{__FUNCTION__} and
6389 @code{__PRETTY_FUNCTION__} have always been variables.
6391 @node Return Address
6392 @section Getting the Return or Frame Address of a Function
6394 These functions may be used to get information about the callers of a
6397 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
6398 This function returns the return address of the current function, or of
6399 one of its callers. The @var{level} argument is number of frames to
6400 scan up the call stack. A value of @code{0} yields the return address
6401 of the current function, a value of @code{1} yields the return address
6402 of the caller of the current function, and so forth. When inlining
6403 the expected behavior is that the function will return the address of
6404 the function that will be returned to. To work around this behavior use
6405 the @code{noinline} function attribute.
6407 The @var{level} argument must be a constant integer.
6409 On some machines it may be impossible to determine the return address of
6410 any function other than the current one; in such cases, or when the top
6411 of the stack has been reached, this function will return @code{0} or a
6412 random value. In addition, @code{__builtin_frame_address} may be used
6413 to determine if the top of the stack has been reached.
6415 Additional post-processing of the returned value may be needed, see
6416 @code{__builtin_extract_return_address}.
6418 This function should only be used with a nonzero argument for debugging
6422 @deftypefn {Built-in Function} {void *} __builtin_extract_return_address (void *@var{addr})
6423 The address as returned by @code{__builtin_return_address} may have to be fed
6424 through this function to get the actual encoded address. For example, on the
6425 31-bit S/390 platform the highest bit has to be masked out, or on SPARC
6426 platforms an offset has to be added for the true next instruction to be
6429 If no fixup is needed, this function simply passes through @var{addr}.
6432 @deftypefn {Built-in Function} {void *} __builtin_frob_return_address (void *@var{addr})
6433 This function does the reverse of @code{__builtin_extract_return_address}.
6436 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
6437 This function is similar to @code{__builtin_return_address}, but it
6438 returns the address of the function frame rather than the return address
6439 of the function. Calling @code{__builtin_frame_address} with a value of
6440 @code{0} yields the frame address of the current function, a value of
6441 @code{1} yields the frame address of the caller of the current function,
6444 The frame is the area on the stack which holds local variables and saved
6445 registers. The frame address is normally the address of the first word
6446 pushed on to the stack by the function. However, the exact definition
6447 depends upon the processor and the calling convention. If the processor
6448 has a dedicated frame pointer register, and the function has a frame,
6449 then @code{__builtin_frame_address} will return the value of the frame
6452 On some machines it may be impossible to determine the frame address of
6453 any function other than the current one; in such cases, or when the top
6454 of the stack has been reached, this function will return @code{0} if
6455 the first frame pointer is properly initialized by the startup code.
6457 This function should only be used with a nonzero argument for debugging
6461 @node Vector Extensions
6462 @section Using vector instructions through built-in functions
6464 On some targets, the instruction set contains SIMD vector instructions that
6465 operate on multiple values contained in one large register at the same time.
6466 For example, on the i386 the MMX, 3DNow!@: and SSE extensions can be used
6469 The first step in using these extensions is to provide the necessary data
6470 types. This should be done using an appropriate @code{typedef}:
6473 typedef int v4si __attribute__ ((vector_size (16)));
6476 The @code{int} type specifies the base type, while the attribute specifies
6477 the vector size for the variable, measured in bytes. For example, the
6478 declaration above causes the compiler to set the mode for the @code{v4si}
6479 type to be 16 bytes wide and divided into @code{int} sized units. For
6480 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
6481 corresponding mode of @code{foo} will be @acronym{V4SI}.
6483 The @code{vector_size} attribute is only applicable to integral and
6484 float scalars, although arrays, pointers, and function return values
6485 are allowed in conjunction with this construct.
6487 All the basic integer types can be used as base types, both as signed
6488 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
6489 @code{long long}. In addition, @code{float} and @code{double} can be
6490 used to build floating-point vector types.
6492 Specifying a combination that is not valid for the current architecture
6493 will cause GCC to synthesize the instructions using a narrower mode.
6494 For example, if you specify a variable of type @code{V4SI} and your
6495 architecture does not allow for this specific SIMD type, GCC will
6496 produce code that uses 4 @code{SIs}.
6498 The types defined in this manner can be used with a subset of normal C
6499 operations. Currently, GCC will allow using the following operators
6500 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~, %}@.
6502 The operations behave like C++ @code{valarrays}. Addition is defined as
6503 the addition of the corresponding elements of the operands. For
6504 example, in the code below, each of the 4 elements in @var{a} will be
6505 added to the corresponding 4 elements in @var{b} and the resulting
6506 vector will be stored in @var{c}.
6509 typedef int v4si __attribute__ ((vector_size (16)));
6516 Subtraction, multiplication, division, and the logical operations
6517 operate in a similar manner. Likewise, the result of using the unary
6518 minus or complement operators on a vector type is a vector whose
6519 elements are the negative or complemented values of the corresponding
6520 elements in the operand.
6522 In C it is possible to use shifting operators @code{<<}, @code{>>} on
6523 integer-type vectors. The operation is defined as following: @code{@{a0,
6524 a1, @dots{}, an@} >> @{b0, b1, @dots{}, bn@} == @{a0 >> b0, a1 >> b1,
6525 @dots{}, an >> bn@}}@. Vector operands must have the same number of
6526 elements. Additionally second operands can be a scalar integer in which
6527 case the scalar is converted to the type used by the vector operand (with
6528 possible truncation) and each element of this new vector is the scalar's
6530 Consider the following code.
6533 typedef int v4si __attribute__ ((vector_size (16)));
6537 b = a >> 1; /* b = a >> @{1,1,1,1@}; */
6540 In C vectors can be subscripted as if the vector were an array with
6541 the same number of elements and base type. Out of bound accesses
6542 invoke undefined behavior at runtime. Warnings for out of bound
6543 accesses for vector subscription can be enabled with
6544 @option{-Warray-bounds}.
6546 You can declare variables and use them in function calls and returns, as
6547 well as in assignments and some casts. You can specify a vector type as
6548 a return type for a function. Vector types can also be used as function
6549 arguments. It is possible to cast from one vector type to another,
6550 provided they are of the same size (in fact, you can also cast vectors
6551 to and from other datatypes of the same size).
6553 You cannot operate between vectors of different lengths or different
6554 signedness without a cast.
6556 A port that supports hardware vector operations, usually provides a set
6557 of built-in functions that can be used to operate on vectors. For
6558 example, a function to add two vectors and multiply the result by a
6559 third could look like this:
6562 v4si f (v4si a, v4si b, v4si c)
6564 v4si tmp = __builtin_addv4si (a, b);
6565 return __builtin_mulv4si (tmp, c);
6572 @findex __builtin_offsetof
6574 GCC implements for both C and C++ a syntactic extension to implement
6575 the @code{offsetof} macro.
6579 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
6581 offsetof_member_designator:
6583 | offsetof_member_designator "." @code{identifier}
6584 | offsetof_member_designator "[" @code{expr} "]"
6587 This extension is sufficient such that
6590 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
6593 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
6594 may be dependent. In either case, @var{member} may consist of a single
6595 identifier, or a sequence of member accesses and array references.
6597 @node Atomic Builtins
6598 @section Built-in functions for atomic memory access
6600 The following builtins are intended to be compatible with those described
6601 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
6602 section 7.4. As such, they depart from the normal GCC practice of using
6603 the ``__builtin_'' prefix, and further that they are overloaded such that
6604 they work on multiple types.
6606 The definition given in the Intel documentation allows only for the use of
6607 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
6608 counterparts. GCC will allow any integral scalar or pointer type that is
6609 1, 2, 4 or 8 bytes in length.
6611 Not all operations are supported by all target processors. If a particular
6612 operation cannot be implemented on the target processor, a warning will be
6613 generated and a call an external function will be generated. The external
6614 function will carry the same name as the builtin, with an additional suffix
6615 @samp{_@var{n}} where @var{n} is the size of the data type.
6617 @c ??? Should we have a mechanism to suppress this warning? This is almost
6618 @c useful for implementing the operation under the control of an external
6621 In most cases, these builtins are considered a @dfn{full barrier}. That is,
6622 no memory operand will be moved across the operation, either forward or
6623 backward. Further, instructions will be issued as necessary to prevent the
6624 processor from speculating loads across the operation and from queuing stores
6625 after the operation.
6627 All of the routines are described in the Intel documentation to take
6628 ``an optional list of variables protected by the memory barrier''. It's
6629 not clear what is meant by that; it could mean that @emph{only} the
6630 following variables are protected, or it could mean that these variables
6631 should in addition be protected. At present GCC ignores this list and
6632 protects all variables which are globally accessible. If in the future
6633 we make some use of this list, an empty list will continue to mean all
6634 globally accessible variables.
6637 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
6638 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
6639 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
6640 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
6641 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
6642 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
6643 @findex __sync_fetch_and_add
6644 @findex __sync_fetch_and_sub
6645 @findex __sync_fetch_and_or
6646 @findex __sync_fetch_and_and
6647 @findex __sync_fetch_and_xor
6648 @findex __sync_fetch_and_nand
6649 These builtins perform the operation suggested by the name, and
6650 returns the value that had previously been in memory. That is,
6653 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
6654 @{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; @} // nand
6657 @emph{Note:} GCC 4.4 and later implement @code{__sync_fetch_and_nand}
6658 builtin as @code{*ptr = ~(tmp & value)} instead of @code{*ptr = ~tmp & value}.
6660 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
6661 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
6662 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
6663 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
6664 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
6665 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
6666 @findex __sync_add_and_fetch
6667 @findex __sync_sub_and_fetch
6668 @findex __sync_or_and_fetch
6669 @findex __sync_and_and_fetch
6670 @findex __sync_xor_and_fetch
6671 @findex __sync_nand_and_fetch
6672 These builtins perform the operation suggested by the name, and
6673 return the new value. That is,
6676 @{ *ptr @var{op}= value; return *ptr; @}
6677 @{ *ptr = ~(*ptr & value); return *ptr; @} // nand
6680 @emph{Note:} GCC 4.4 and later implement @code{__sync_nand_and_fetch}
6681 builtin as @code{*ptr = ~(*ptr & value)} instead of
6682 @code{*ptr = ~*ptr & value}.
6684 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6685 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6686 @findex __sync_bool_compare_and_swap
6687 @findex __sync_val_compare_and_swap
6688 These builtins perform an atomic compare and swap. That is, if the current
6689 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
6692 The ``bool'' version returns true if the comparison is successful and
6693 @var{newval} was written. The ``val'' version returns the contents
6694 of @code{*@var{ptr}} before the operation.
6696 @item __sync_synchronize (...)
6697 @findex __sync_synchronize
6698 This builtin issues a full memory barrier.
6700 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
6701 @findex __sync_lock_test_and_set
6702 This builtin, as described by Intel, is not a traditional test-and-set
6703 operation, but rather an atomic exchange operation. It writes @var{value}
6704 into @code{*@var{ptr}}, and returns the previous contents of
6707 Many targets have only minimal support for such locks, and do not support
6708 a full exchange operation. In this case, a target may support reduced
6709 functionality here by which the @emph{only} valid value to store is the
6710 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
6711 is implementation defined.
6713 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
6714 This means that references after the builtin cannot move to (or be
6715 speculated to) before the builtin, but previous memory stores may not
6716 be globally visible yet, and previous memory loads may not yet be
6719 @item void __sync_lock_release (@var{type} *ptr, ...)
6720 @findex __sync_lock_release
6721 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
6722 Normally this means writing the constant 0 to @code{*@var{ptr}}.
6724 This builtin is not a full barrier, but rather a @dfn{release barrier}.
6725 This means that all previous memory stores are globally visible, and all
6726 previous memory loads have been satisfied, but following memory reads
6727 are not prevented from being speculated to before the barrier.
6730 @node Object Size Checking
6731 @section Object Size Checking Builtins
6732 @findex __builtin_object_size
6733 @findex __builtin___memcpy_chk
6734 @findex __builtin___mempcpy_chk
6735 @findex __builtin___memmove_chk
6736 @findex __builtin___memset_chk
6737 @findex __builtin___strcpy_chk
6738 @findex __builtin___stpcpy_chk
6739 @findex __builtin___strncpy_chk
6740 @findex __builtin___strcat_chk
6741 @findex __builtin___strncat_chk
6742 @findex __builtin___sprintf_chk
6743 @findex __builtin___snprintf_chk
6744 @findex __builtin___vsprintf_chk
6745 @findex __builtin___vsnprintf_chk
6746 @findex __builtin___printf_chk
6747 @findex __builtin___vprintf_chk
6748 @findex __builtin___fprintf_chk
6749 @findex __builtin___vfprintf_chk
6751 GCC implements a limited buffer overflow protection mechanism
6752 that can prevent some buffer overflow attacks.
6754 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
6755 is a built-in construct that returns a constant number of bytes from
6756 @var{ptr} to the end of the object @var{ptr} pointer points to
6757 (if known at compile time). @code{__builtin_object_size} never evaluates
6758 its arguments for side-effects. If there are any side-effects in them, it
6759 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6760 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
6761 point to and all of them are known at compile time, the returned number
6762 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
6763 0 and minimum if nonzero. If it is not possible to determine which objects
6764 @var{ptr} points to at compile time, @code{__builtin_object_size} should
6765 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6766 for @var{type} 2 or 3.
6768 @var{type} is an integer constant from 0 to 3. If the least significant
6769 bit is clear, objects are whole variables, if it is set, a closest
6770 surrounding subobject is considered the object a pointer points to.
6771 The second bit determines if maximum or minimum of remaining bytes
6775 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
6776 char *p = &var.buf1[1], *q = &var.b;
6778 /* Here the object p points to is var. */
6779 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
6780 /* The subobject p points to is var.buf1. */
6781 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
6782 /* The object q points to is var. */
6783 assert (__builtin_object_size (q, 0)
6784 == (char *) (&var + 1) - (char *) &var.b);
6785 /* The subobject q points to is var.b. */
6786 assert (__builtin_object_size (q, 1) == sizeof (var.b));
6790 There are built-in functions added for many common string operation
6791 functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk}
6792 built-in is provided. This built-in has an additional last argument,
6793 which is the number of bytes remaining in object the @var{dest}
6794 argument points to or @code{(size_t) -1} if the size is not known.
6796 The built-in functions are optimized into the normal string functions
6797 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
6798 it is known at compile time that the destination object will not
6799 be overflown. If the compiler can determine at compile time the
6800 object will be always overflown, it issues a warning.
6802 The intended use can be e.g.
6806 #define bos0(dest) __builtin_object_size (dest, 0)
6807 #define memcpy(dest, src, n) \
6808 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
6812 /* It is unknown what object p points to, so this is optimized
6813 into plain memcpy - no checking is possible. */
6814 memcpy (p, "abcde", n);
6815 /* Destination is known and length too. It is known at compile
6816 time there will be no overflow. */
6817 memcpy (&buf[5], "abcde", 5);
6818 /* Destination is known, but the length is not known at compile time.
6819 This will result in __memcpy_chk call that can check for overflow
6821 memcpy (&buf[5], "abcde", n);
6822 /* Destination is known and it is known at compile time there will
6823 be overflow. There will be a warning and __memcpy_chk call that
6824 will abort the program at runtime. */
6825 memcpy (&buf[6], "abcde", 5);
6828 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
6829 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
6830 @code{strcat} and @code{strncat}.
6832 There are also checking built-in functions for formatted output functions.
6834 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
6835 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6836 const char *fmt, ...);
6837 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
6839 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6840 const char *fmt, va_list ap);
6843 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
6844 etc.@: functions and can contain implementation specific flags on what
6845 additional security measures the checking function might take, such as
6846 handling @code{%n} differently.
6848 The @var{os} argument is the object size @var{s} points to, like in the
6849 other built-in functions. There is a small difference in the behavior
6850 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
6851 optimized into the non-checking functions only if @var{flag} is 0, otherwise
6852 the checking function is called with @var{os} argument set to
6855 In addition to this, there are checking built-in functions
6856 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
6857 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
6858 These have just one additional argument, @var{flag}, right before
6859 format string @var{fmt}. If the compiler is able to optimize them to
6860 @code{fputc} etc.@: functions, it will, otherwise the checking function
6861 should be called and the @var{flag} argument passed to it.
6863 @node Other Builtins
6864 @section Other built-in functions provided by GCC
6865 @cindex built-in functions
6866 @findex __builtin_fpclassify
6867 @findex __builtin_isfinite
6868 @findex __builtin_isnormal
6869 @findex __builtin_isgreater
6870 @findex __builtin_isgreaterequal
6871 @findex __builtin_isinf_sign
6872 @findex __builtin_isless
6873 @findex __builtin_islessequal
6874 @findex __builtin_islessgreater
6875 @findex __builtin_isunordered
6876 @findex __builtin_powi
6877 @findex __builtin_powif
6878 @findex __builtin_powil
7036 @findex fprintf_unlocked
7038 @findex fputs_unlocked
7155 @findex printf_unlocked
7187 @findex significandf
7188 @findex significandl
7259 GCC provides a large number of built-in functions other than the ones
7260 mentioned above. Some of these are for internal use in the processing
7261 of exceptions or variable-length argument lists and will not be
7262 documented here because they may change from time to time; we do not
7263 recommend general use of these functions.
7265 The remaining functions are provided for optimization purposes.
7267 @opindex fno-builtin
7268 GCC includes built-in versions of many of the functions in the standard
7269 C library. The versions prefixed with @code{__builtin_} will always be
7270 treated as having the same meaning as the C library function even if you
7271 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
7272 Many of these functions are only optimized in certain cases; if they are
7273 not optimized in a particular case, a call to the library function will
7278 Outside strict ISO C mode (@option{-ansi}, @option{-std=c90},
7279 @option{-std=c99} or @option{-std=c1x}), the functions
7280 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
7281 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
7282 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
7283 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
7284 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
7285 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
7286 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
7287 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
7288 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
7289 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
7290 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
7291 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
7292 @code{signbitd64}, @code{signbitd128}, @code{significandf},
7293 @code{significandl}, @code{significand}, @code{sincosf},
7294 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
7295 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
7296 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
7297 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
7299 may be handled as built-in functions.
7300 All these functions have corresponding versions
7301 prefixed with @code{__builtin_}, which may be used even in strict C90
7304 The ISO C99 functions
7305 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
7306 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
7307 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
7308 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
7309 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
7310 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
7311 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
7312 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
7313 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
7314 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
7315 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
7316 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
7317 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
7318 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
7319 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
7320 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
7321 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
7322 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
7323 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
7324 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
7325 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
7326 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
7327 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
7328 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
7329 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
7330 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
7331 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
7332 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
7333 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
7334 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
7335 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
7336 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
7337 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
7338 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
7339 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
7340 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
7341 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
7342 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
7343 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
7344 are handled as built-in functions
7345 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
7347 There are also built-in versions of the ISO C99 functions
7348 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
7349 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
7350 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
7351 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
7352 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
7353 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
7354 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
7355 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
7356 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
7357 that are recognized in any mode since ISO C90 reserves these names for
7358 the purpose to which ISO C99 puts them. All these functions have
7359 corresponding versions prefixed with @code{__builtin_}.
7361 The ISO C94 functions
7362 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
7363 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
7364 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
7366 are handled as built-in functions
7367 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
7369 The ISO C90 functions
7370 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
7371 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
7372 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
7373 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
7374 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
7375 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
7376 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
7377 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
7378 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
7379 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
7380 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
7381 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
7382 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
7383 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
7384 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
7385 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
7386 are all recognized as built-in functions unless
7387 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
7388 is specified for an individual function). All of these functions have
7389 corresponding versions prefixed with @code{__builtin_}.
7391 GCC provides built-in versions of the ISO C99 floating point comparison
7392 macros that avoid raising exceptions for unordered operands. They have
7393 the same names as the standard macros ( @code{isgreater},
7394 @code{isgreaterequal}, @code{isless}, @code{islessequal},
7395 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
7396 prefixed. We intend for a library implementor to be able to simply
7397 @code{#define} each standard macro to its built-in equivalent.
7398 In the same fashion, GCC provides @code{fpclassify}, @code{isfinite},
7399 @code{isinf_sign} and @code{isnormal} built-ins used with
7400 @code{__builtin_} prefixed. The @code{isinf} and @code{isnan}
7401 builtins appear both with and without the @code{__builtin_} prefix.
7403 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
7405 You can use the built-in function @code{__builtin_types_compatible_p} to
7406 determine whether two types are the same.
7408 This built-in function returns 1 if the unqualified versions of the
7409 types @var{type1} and @var{type2} (which are types, not expressions) are
7410 compatible, 0 otherwise. The result of this built-in function can be
7411 used in integer constant expressions.
7413 This built-in function ignores top level qualifiers (e.g., @code{const},
7414 @code{volatile}). For example, @code{int} is equivalent to @code{const
7417 The type @code{int[]} and @code{int[5]} are compatible. On the other
7418 hand, @code{int} and @code{char *} are not compatible, even if the size
7419 of their types, on the particular architecture are the same. Also, the
7420 amount of pointer indirection is taken into account when determining
7421 similarity. Consequently, @code{short *} is not similar to
7422 @code{short **}. Furthermore, two types that are typedefed are
7423 considered compatible if their underlying types are compatible.
7425 An @code{enum} type is not considered to be compatible with another
7426 @code{enum} type even if both are compatible with the same integer
7427 type; this is what the C standard specifies.
7428 For example, @code{enum @{foo, bar@}} is not similar to
7429 @code{enum @{hot, dog@}}.
7431 You would typically use this function in code whose execution varies
7432 depending on the arguments' types. For example:
7437 typeof (x) tmp = (x); \
7438 if (__builtin_types_compatible_p (typeof (x), long double)) \
7439 tmp = foo_long_double (tmp); \
7440 else if (__builtin_types_compatible_p (typeof (x), double)) \
7441 tmp = foo_double (tmp); \
7442 else if (__builtin_types_compatible_p (typeof (x), float)) \
7443 tmp = foo_float (tmp); \
7450 @emph{Note:} This construct is only available for C@.
7454 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
7456 You can use the built-in function @code{__builtin_choose_expr} to
7457 evaluate code depending on the value of a constant expression. This
7458 built-in function returns @var{exp1} if @var{const_exp}, which is an
7459 integer constant expression, is nonzero. Otherwise it returns @var{exp2}.
7461 This built-in function is analogous to the @samp{? :} operator in C,
7462 except that the expression returned has its type unaltered by promotion
7463 rules. Also, the built-in function does not evaluate the expression
7464 that was not chosen. For example, if @var{const_exp} evaluates to true,
7465 @var{exp2} is not evaluated even if it has side-effects.
7467 This built-in function can return an lvalue if the chosen argument is an
7470 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
7471 type. Similarly, if @var{exp2} is returned, its return type is the same
7478 __builtin_choose_expr ( \
7479 __builtin_types_compatible_p (typeof (x), double), \
7481 __builtin_choose_expr ( \
7482 __builtin_types_compatible_p (typeof (x), float), \
7484 /* @r{The void expression results in a compile-time error} \
7485 @r{when assigning the result to something.} */ \
7489 @emph{Note:} This construct is only available for C@. Furthermore, the
7490 unused expression (@var{exp1} or @var{exp2} depending on the value of
7491 @var{const_exp}) may still generate syntax errors. This may change in
7496 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
7497 You can use the built-in function @code{__builtin_constant_p} to
7498 determine if a value is known to be constant at compile-time and hence
7499 that GCC can perform constant-folding on expressions involving that
7500 value. The argument of the function is the value to test. The function
7501 returns the integer 1 if the argument is known to be a compile-time
7502 constant and 0 if it is not known to be a compile-time constant. A
7503 return of 0 does not indicate that the value is @emph{not} a constant,
7504 but merely that GCC cannot prove it is a constant with the specified
7505 value of the @option{-O} option.
7507 You would typically use this function in an embedded application where
7508 memory was a critical resource. If you have some complex calculation,
7509 you may want it to be folded if it involves constants, but need to call
7510 a function if it does not. For example:
7513 #define Scale_Value(X) \
7514 (__builtin_constant_p (X) \
7515 ? ((X) * SCALE + OFFSET) : Scale (X))
7518 You may use this built-in function in either a macro or an inline
7519 function. However, if you use it in an inlined function and pass an
7520 argument of the function as the argument to the built-in, GCC will
7521 never return 1 when you call the inline function with a string constant
7522 or compound literal (@pxref{Compound Literals}) and will not return 1
7523 when you pass a constant numeric value to the inline function unless you
7524 specify the @option{-O} option.
7526 You may also use @code{__builtin_constant_p} in initializers for static
7527 data. For instance, you can write
7530 static const int table[] = @{
7531 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
7537 This is an acceptable initializer even if @var{EXPRESSION} is not a
7538 constant expression, including the case where
7539 @code{__builtin_constant_p} returns 1 because @var{EXPRESSION} can be
7540 folded to a constant but @var{EXPRESSION} contains operands that would
7541 not otherwise be permitted in a static initializer (for example,
7542 @code{0 && foo ()}). GCC must be more conservative about evaluating the
7543 built-in in this case, because it has no opportunity to perform
7546 Previous versions of GCC did not accept this built-in in data
7547 initializers. The earliest version where it is completely safe is
7551 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
7552 @opindex fprofile-arcs
7553 You may use @code{__builtin_expect} to provide the compiler with
7554 branch prediction information. In general, you should prefer to
7555 use actual profile feedback for this (@option{-fprofile-arcs}), as
7556 programmers are notoriously bad at predicting how their programs
7557 actually perform. However, there are applications in which this
7558 data is hard to collect.
7560 The return value is the value of @var{exp}, which should be an integral
7561 expression. The semantics of the built-in are that it is expected that
7562 @var{exp} == @var{c}. For example:
7565 if (__builtin_expect (x, 0))
7570 would indicate that we do not expect to call @code{foo}, since
7571 we expect @code{x} to be zero. Since you are limited to integral
7572 expressions for @var{exp}, you should use constructions such as
7575 if (__builtin_expect (ptr != NULL, 1))
7580 when testing pointer or floating-point values.
7583 @deftypefn {Built-in Function} void __builtin_trap (void)
7584 This function causes the program to exit abnormally. GCC implements
7585 this function by using a target-dependent mechanism (such as
7586 intentionally executing an illegal instruction) or by calling
7587 @code{abort}. The mechanism used may vary from release to release so
7588 you should not rely on any particular implementation.
7591 @deftypefn {Built-in Function} void __builtin_unreachable (void)
7592 If control flow reaches the point of the @code{__builtin_unreachable},
7593 the program is undefined. It is useful in situations where the
7594 compiler cannot deduce the unreachability of the code.
7596 One such case is immediately following an @code{asm} statement that
7597 will either never terminate, or one that transfers control elsewhere
7598 and never returns. In this example, without the
7599 @code{__builtin_unreachable}, GCC would issue a warning that control
7600 reaches the end of a non-void function. It would also generate code
7601 to return after the @code{asm}.
7604 int f (int c, int v)
7612 asm("jmp error_handler");
7613 __builtin_unreachable ();
7618 Because the @code{asm} statement unconditionally transfers control out
7619 of the function, control will never reach the end of the function
7620 body. The @code{__builtin_unreachable} is in fact unreachable and
7621 communicates this fact to the compiler.
7623 Another use for @code{__builtin_unreachable} is following a call a
7624 function that never returns but that is not declared
7625 @code{__attribute__((noreturn))}, as in this example:
7628 void function_that_never_returns (void);
7638 function_that_never_returns ();
7639 __builtin_unreachable ();
7646 @deftypefn {Built-in Function} void *__builtin_assume_aligned (const void *@var{exp}, size_t @var{align}, ...)
7647 This function returns its first argument, and allows the compiler
7648 to assume that the returned pointer is at least @var{align} bytes
7649 aligned. This built-in can have either two or three arguments,
7650 if it has three, the third argument should have integer type, and
7651 if it is non-zero means misalignment offset. For example:
7654 void *x = __builtin_assume_aligned (arg, 16);
7657 means that the compiler can assume x, set to arg, is at least
7658 16 byte aligned, while:
7661 void *x = __builtin_assume_aligned (arg, 32, 8);
7664 means that the compiler can assume for x, set to arg, that
7665 (char *) x - 8 is 32 byte aligned.
7668 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
7669 This function is used to flush the processor's instruction cache for
7670 the region of memory between @var{begin} inclusive and @var{end}
7671 exclusive. Some targets require that the instruction cache be
7672 flushed, after modifying memory containing code, in order to obtain
7673 deterministic behavior.
7675 If the target does not require instruction cache flushes,
7676 @code{__builtin___clear_cache} has no effect. Otherwise either
7677 instructions are emitted in-line to clear the instruction cache or a
7678 call to the @code{__clear_cache} function in libgcc is made.
7681 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
7682 This function is used to minimize cache-miss latency by moving data into
7683 a cache before it is accessed.
7684 You can insert calls to @code{__builtin_prefetch} into code for which
7685 you know addresses of data in memory that is likely to be accessed soon.
7686 If the target supports them, data prefetch instructions will be generated.
7687 If the prefetch is done early enough before the access then the data will
7688 be in the cache by the time it is accessed.
7690 The value of @var{addr} is the address of the memory to prefetch.
7691 There are two optional arguments, @var{rw} and @var{locality}.
7692 The value of @var{rw} is a compile-time constant one or zero; one
7693 means that the prefetch is preparing for a write to the memory address
7694 and zero, the default, means that the prefetch is preparing for a read.
7695 The value @var{locality} must be a compile-time constant integer between
7696 zero and three. A value of zero means that the data has no temporal
7697 locality, so it need not be left in the cache after the access. A value
7698 of three means that the data has a high degree of temporal locality and
7699 should be left in all levels of cache possible. Values of one and two
7700 mean, respectively, a low or moderate degree of temporal locality. The
7704 for (i = 0; i < n; i++)
7707 __builtin_prefetch (&a[i+j], 1, 1);
7708 __builtin_prefetch (&b[i+j], 0, 1);
7713 Data prefetch does not generate faults if @var{addr} is invalid, but
7714 the address expression itself must be valid. For example, a prefetch
7715 of @code{p->next} will not fault if @code{p->next} is not a valid
7716 address, but evaluation will fault if @code{p} is not a valid address.
7718 If the target does not support data prefetch, the address expression
7719 is evaluated if it includes side effects but no other code is generated
7720 and GCC does not issue a warning.
7723 @deftypefn {Built-in Function} double __builtin_huge_val (void)
7724 Returns a positive infinity, if supported by the floating-point format,
7725 else @code{DBL_MAX}. This function is suitable for implementing the
7726 ISO C macro @code{HUGE_VAL}.
7729 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
7730 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
7733 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
7734 Similar to @code{__builtin_huge_val}, except the return
7735 type is @code{long double}.
7738 @deftypefn {Built-in Function} int __builtin_fpclassify (int, int, int, int, int, ...)
7739 This built-in implements the C99 fpclassify functionality. The first
7740 five int arguments should be the target library's notion of the
7741 possible FP classes and are used for return values. They must be
7742 constant values and they must appear in this order: @code{FP_NAN},
7743 @code{FP_INFINITE}, @code{FP_NORMAL}, @code{FP_SUBNORMAL} and
7744 @code{FP_ZERO}. The ellipsis is for exactly one floating point value
7745 to classify. GCC treats the last argument as type-generic, which
7746 means it does not do default promotion from float to double.
7749 @deftypefn {Built-in Function} double __builtin_inf (void)
7750 Similar to @code{__builtin_huge_val}, except a warning is generated
7751 if the target floating-point format does not support infinities.
7754 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
7755 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
7758 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
7759 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
7762 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
7763 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
7766 @deftypefn {Built-in Function} float __builtin_inff (void)
7767 Similar to @code{__builtin_inf}, except the return type is @code{float}.
7768 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
7771 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
7772 Similar to @code{__builtin_inf}, except the return
7773 type is @code{long double}.
7776 @deftypefn {Built-in Function} int __builtin_isinf_sign (...)
7777 Similar to @code{isinf}, except the return value will be negative for
7778 an argument of @code{-Inf}. Note while the parameter list is an
7779 ellipsis, this function only accepts exactly one floating point
7780 argument. GCC treats this parameter as type-generic, which means it
7781 does not do default promotion from float to double.
7784 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
7785 This is an implementation of the ISO C99 function @code{nan}.
7787 Since ISO C99 defines this function in terms of @code{strtod}, which we
7788 do not implement, a description of the parsing is in order. The string
7789 is parsed as by @code{strtol}; that is, the base is recognized by
7790 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
7791 in the significand such that the least significant bit of the number
7792 is at the least significant bit of the significand. The number is
7793 truncated to fit the significand field provided. The significand is
7794 forced to be a quiet NaN@.
7796 This function, if given a string literal all of which would have been
7797 consumed by strtol, is evaluated early enough that it is considered a
7798 compile-time constant.
7801 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
7802 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
7805 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
7806 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
7809 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
7810 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
7813 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
7814 Similar to @code{__builtin_nan}, except the return type is @code{float}.
7817 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
7818 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
7821 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
7822 Similar to @code{__builtin_nan}, except the significand is forced
7823 to be a signaling NaN@. The @code{nans} function is proposed by
7824 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
7827 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
7828 Similar to @code{__builtin_nans}, except the return type is @code{float}.
7831 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
7832 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
7835 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
7836 Returns one plus the index of the least significant 1-bit of @var{x}, or
7837 if @var{x} is zero, returns zero.
7840 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
7841 Returns the number of leading 0-bits in @var{x}, starting at the most
7842 significant bit position. If @var{x} is 0, the result is undefined.
7845 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
7846 Returns the number of trailing 0-bits in @var{x}, starting at the least
7847 significant bit position. If @var{x} is 0, the result is undefined.
7850 @deftypefn {Built-in Function} int __builtin_clrsb (int x)
7851 Returns the number of leading redundant sign bits in @var{x}, i.e. the
7852 number of bits following the most significant bit which are identical
7853 to it. There are no special cases for 0 or other values.
7856 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
7857 Returns the number of 1-bits in @var{x}.
7860 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
7861 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
7865 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
7866 Similar to @code{__builtin_ffs}, except the argument type is
7867 @code{unsigned long}.
7870 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
7871 Similar to @code{__builtin_clz}, except the argument type is
7872 @code{unsigned long}.
7875 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
7876 Similar to @code{__builtin_ctz}, except the argument type is
7877 @code{unsigned long}.
7880 @deftypefn {Built-in Function} int __builtin_clrsbl (long)
7881 Similar to @code{__builtin_clrsb}, except the argument type is
7885 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
7886 Similar to @code{__builtin_popcount}, except the argument type is
7887 @code{unsigned long}.
7890 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
7891 Similar to @code{__builtin_parity}, except the argument type is
7892 @code{unsigned long}.
7895 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
7896 Similar to @code{__builtin_ffs}, except the argument type is
7897 @code{unsigned long long}.
7900 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
7901 Similar to @code{__builtin_clz}, except the argument type is
7902 @code{unsigned long long}.
7905 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
7906 Similar to @code{__builtin_ctz}, except the argument type is
7907 @code{unsigned long long}.
7910 @deftypefn {Built-in Function} int __builtin_clrsbll (long long)
7911 Similar to @code{__builtin_clrsb}, except the argument type is
7915 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
7916 Similar to @code{__builtin_popcount}, except the argument type is
7917 @code{unsigned long long}.
7920 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
7921 Similar to @code{__builtin_parity}, except the argument type is
7922 @code{unsigned long long}.
7925 @deftypefn {Built-in Function} double __builtin_powi (double, int)
7926 Returns the first argument raised to the power of the second. Unlike the
7927 @code{pow} function no guarantees about precision and rounding are made.
7930 @deftypefn {Built-in Function} float __builtin_powif (float, int)
7931 Similar to @code{__builtin_powi}, except the argument and return types
7935 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
7936 Similar to @code{__builtin_powi}, except the argument and return types
7937 are @code{long double}.
7940 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
7941 Returns @var{x} with the order of the bytes reversed; for example,
7942 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
7946 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
7947 Similar to @code{__builtin_bswap32}, except the argument and return types
7951 @node Target Builtins
7952 @section Built-in Functions Specific to Particular Target Machines
7954 On some target machines, GCC supports many built-in functions specific
7955 to those machines. Generally these generate calls to specific machine
7956 instructions, but allow the compiler to schedule those calls.
7959 * Alpha Built-in Functions::
7960 * ARM iWMMXt Built-in Functions::
7961 * ARM NEON Intrinsics::
7962 * AVR Built-in Functions::
7963 * Blackfin Built-in Functions::
7964 * FR-V Built-in Functions::
7965 * X86 Built-in Functions::
7966 * MIPS DSP Built-in Functions::
7967 * MIPS Paired-Single Support::
7968 * MIPS Loongson Built-in Functions::
7969 * Other MIPS Built-in Functions::
7970 * picoChip Built-in Functions::
7971 * PowerPC AltiVec/VSX Built-in Functions::
7972 * RX Built-in Functions::
7973 * SPARC VIS Built-in Functions::
7974 * SPU Built-in Functions::
7975 * TI C6X Built-in Functions::
7978 @node Alpha Built-in Functions
7979 @subsection Alpha Built-in Functions
7981 These built-in functions are available for the Alpha family of
7982 processors, depending on the command-line switches used.
7984 The following built-in functions are always available. They
7985 all generate the machine instruction that is part of the name.
7988 long __builtin_alpha_implver (void)
7989 long __builtin_alpha_rpcc (void)
7990 long __builtin_alpha_amask (long)
7991 long __builtin_alpha_cmpbge (long, long)
7992 long __builtin_alpha_extbl (long, long)
7993 long __builtin_alpha_extwl (long, long)
7994 long __builtin_alpha_extll (long, long)
7995 long __builtin_alpha_extql (long, long)
7996 long __builtin_alpha_extwh (long, long)
7997 long __builtin_alpha_extlh (long, long)
7998 long __builtin_alpha_extqh (long, long)
7999 long __builtin_alpha_insbl (long, long)
8000 long __builtin_alpha_inswl (long, long)
8001 long __builtin_alpha_insll (long, long)
8002 long __builtin_alpha_insql (long, long)
8003 long __builtin_alpha_inswh (long, long)
8004 long __builtin_alpha_inslh (long, long)
8005 long __builtin_alpha_insqh (long, long)
8006 long __builtin_alpha_mskbl (long, long)
8007 long __builtin_alpha_mskwl (long, long)
8008 long __builtin_alpha_mskll (long, long)
8009 long __builtin_alpha_mskql (long, long)
8010 long __builtin_alpha_mskwh (long, long)
8011 long __builtin_alpha_msklh (long, long)
8012 long __builtin_alpha_mskqh (long, long)
8013 long __builtin_alpha_umulh (long, long)
8014 long __builtin_alpha_zap (long, long)
8015 long __builtin_alpha_zapnot (long, long)
8018 The following built-in functions are always with @option{-mmax}
8019 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
8020 later. They all generate the machine instruction that is part
8024 long __builtin_alpha_pklb (long)
8025 long __builtin_alpha_pkwb (long)
8026 long __builtin_alpha_unpkbl (long)
8027 long __builtin_alpha_unpkbw (long)
8028 long __builtin_alpha_minub8 (long, long)
8029 long __builtin_alpha_minsb8 (long, long)
8030 long __builtin_alpha_minuw4 (long, long)
8031 long __builtin_alpha_minsw4 (long, long)
8032 long __builtin_alpha_maxub8 (long, long)
8033 long __builtin_alpha_maxsb8 (long, long)
8034 long __builtin_alpha_maxuw4 (long, long)
8035 long __builtin_alpha_maxsw4 (long, long)
8036 long __builtin_alpha_perr (long, long)
8039 The following built-in functions are always with @option{-mcix}
8040 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
8041 later. They all generate the machine instruction that is part
8045 long __builtin_alpha_cttz (long)
8046 long __builtin_alpha_ctlz (long)
8047 long __builtin_alpha_ctpop (long)
8050 The following builtins are available on systems that use the OSF/1
8051 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
8052 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
8053 @code{rdval} and @code{wrval}.
8056 void *__builtin_thread_pointer (void)
8057 void __builtin_set_thread_pointer (void *)
8060 @node ARM iWMMXt Built-in Functions
8061 @subsection ARM iWMMXt Built-in Functions
8063 These built-in functions are available for the ARM family of
8064 processors when the @option{-mcpu=iwmmxt} switch is used:
8067 typedef int v2si __attribute__ ((vector_size (8)));
8068 typedef short v4hi __attribute__ ((vector_size (8)));
8069 typedef char v8qi __attribute__ ((vector_size (8)));
8071 int __builtin_arm_getwcx (int)
8072 void __builtin_arm_setwcx (int, int)
8073 int __builtin_arm_textrmsb (v8qi, int)
8074 int __builtin_arm_textrmsh (v4hi, int)
8075 int __builtin_arm_textrmsw (v2si, int)
8076 int __builtin_arm_textrmub (v8qi, int)
8077 int __builtin_arm_textrmuh (v4hi, int)
8078 int __builtin_arm_textrmuw (v2si, int)
8079 v8qi __builtin_arm_tinsrb (v8qi, int)
8080 v4hi __builtin_arm_tinsrh (v4hi, int)
8081 v2si __builtin_arm_tinsrw (v2si, int)
8082 long long __builtin_arm_tmia (long long, int, int)
8083 long long __builtin_arm_tmiabb (long long, int, int)
8084 long long __builtin_arm_tmiabt (long long, int, int)
8085 long long __builtin_arm_tmiaph (long long, int, int)
8086 long long __builtin_arm_tmiatb (long long, int, int)
8087 long long __builtin_arm_tmiatt (long long, int, int)
8088 int __builtin_arm_tmovmskb (v8qi)
8089 int __builtin_arm_tmovmskh (v4hi)
8090 int __builtin_arm_tmovmskw (v2si)
8091 long long __builtin_arm_waccb (v8qi)
8092 long long __builtin_arm_wacch (v4hi)
8093 long long __builtin_arm_waccw (v2si)
8094 v8qi __builtin_arm_waddb (v8qi, v8qi)
8095 v8qi __builtin_arm_waddbss (v8qi, v8qi)
8096 v8qi __builtin_arm_waddbus (v8qi, v8qi)
8097 v4hi __builtin_arm_waddh (v4hi, v4hi)
8098 v4hi __builtin_arm_waddhss (v4hi, v4hi)
8099 v4hi __builtin_arm_waddhus (v4hi, v4hi)
8100 v2si __builtin_arm_waddw (v2si, v2si)
8101 v2si __builtin_arm_waddwss (v2si, v2si)
8102 v2si __builtin_arm_waddwus (v2si, v2si)
8103 v8qi __builtin_arm_walign (v8qi, v8qi, int)
8104 long long __builtin_arm_wand(long long, long long)
8105 long long __builtin_arm_wandn (long long, long long)
8106 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
8107 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
8108 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
8109 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
8110 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
8111 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
8112 v2si __builtin_arm_wcmpeqw (v2si, v2si)
8113 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
8114 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
8115 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
8116 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
8117 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
8118 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
8119 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
8120 long long __builtin_arm_wmacsz (v4hi, v4hi)
8121 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
8122 long long __builtin_arm_wmacuz (v4hi, v4hi)
8123 v4hi __builtin_arm_wmadds (v4hi, v4hi)
8124 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
8125 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
8126 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
8127 v2si __builtin_arm_wmaxsw (v2si, v2si)
8128 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
8129 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
8130 v2si __builtin_arm_wmaxuw (v2si, v2si)
8131 v8qi __builtin_arm_wminsb (v8qi, v8qi)
8132 v4hi __builtin_arm_wminsh (v4hi, v4hi)
8133 v2si __builtin_arm_wminsw (v2si, v2si)
8134 v8qi __builtin_arm_wminub (v8qi, v8qi)
8135 v4hi __builtin_arm_wminuh (v4hi, v4hi)
8136 v2si __builtin_arm_wminuw (v2si, v2si)
8137 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
8138 v4hi __builtin_arm_wmulul (v4hi, v4hi)
8139 v4hi __builtin_arm_wmulum (v4hi, v4hi)
8140 long long __builtin_arm_wor (long long, long long)
8141 v2si __builtin_arm_wpackdss (long long, long long)
8142 v2si __builtin_arm_wpackdus (long long, long long)
8143 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
8144 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
8145 v4hi __builtin_arm_wpackwss (v2si, v2si)
8146 v4hi __builtin_arm_wpackwus (v2si, v2si)
8147 long long __builtin_arm_wrord (long long, long long)
8148 long long __builtin_arm_wrordi (long long, int)
8149 v4hi __builtin_arm_wrorh (v4hi, long long)
8150 v4hi __builtin_arm_wrorhi (v4hi, int)
8151 v2si __builtin_arm_wrorw (v2si, long long)
8152 v2si __builtin_arm_wrorwi (v2si, int)
8153 v2si __builtin_arm_wsadb (v8qi, v8qi)
8154 v2si __builtin_arm_wsadbz (v8qi, v8qi)
8155 v2si __builtin_arm_wsadh (v4hi, v4hi)
8156 v2si __builtin_arm_wsadhz (v4hi, v4hi)
8157 v4hi __builtin_arm_wshufh (v4hi, int)
8158 long long __builtin_arm_wslld (long long, long long)
8159 long long __builtin_arm_wslldi (long long, int)
8160 v4hi __builtin_arm_wsllh (v4hi, long long)
8161 v4hi __builtin_arm_wsllhi (v4hi, int)
8162 v2si __builtin_arm_wsllw (v2si, long long)
8163 v2si __builtin_arm_wsllwi (v2si, int)
8164 long long __builtin_arm_wsrad (long long, long long)
8165 long long __builtin_arm_wsradi (long long, int)
8166 v4hi __builtin_arm_wsrah (v4hi, long long)
8167 v4hi __builtin_arm_wsrahi (v4hi, int)
8168 v2si __builtin_arm_wsraw (v2si, long long)
8169 v2si __builtin_arm_wsrawi (v2si, int)
8170 long long __builtin_arm_wsrld (long long, long long)
8171 long long __builtin_arm_wsrldi (long long, int)
8172 v4hi __builtin_arm_wsrlh (v4hi, long long)
8173 v4hi __builtin_arm_wsrlhi (v4hi, int)
8174 v2si __builtin_arm_wsrlw (v2si, long long)
8175 v2si __builtin_arm_wsrlwi (v2si, int)
8176 v8qi __builtin_arm_wsubb (v8qi, v8qi)
8177 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
8178 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
8179 v4hi __builtin_arm_wsubh (v4hi, v4hi)
8180 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
8181 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
8182 v2si __builtin_arm_wsubw (v2si, v2si)
8183 v2si __builtin_arm_wsubwss (v2si, v2si)
8184 v2si __builtin_arm_wsubwus (v2si, v2si)
8185 v4hi __builtin_arm_wunpckehsb (v8qi)
8186 v2si __builtin_arm_wunpckehsh (v4hi)
8187 long long __builtin_arm_wunpckehsw (v2si)
8188 v4hi __builtin_arm_wunpckehub (v8qi)
8189 v2si __builtin_arm_wunpckehuh (v4hi)
8190 long long __builtin_arm_wunpckehuw (v2si)
8191 v4hi __builtin_arm_wunpckelsb (v8qi)
8192 v2si __builtin_arm_wunpckelsh (v4hi)
8193 long long __builtin_arm_wunpckelsw (v2si)
8194 v4hi __builtin_arm_wunpckelub (v8qi)
8195 v2si __builtin_arm_wunpckeluh (v4hi)
8196 long long __builtin_arm_wunpckeluw (v2si)
8197 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
8198 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
8199 v2si __builtin_arm_wunpckihw (v2si, v2si)
8200 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
8201 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
8202 v2si __builtin_arm_wunpckilw (v2si, v2si)
8203 long long __builtin_arm_wxor (long long, long long)
8204 long long __builtin_arm_wzero ()
8207 @node ARM NEON Intrinsics
8208 @subsection ARM NEON Intrinsics
8210 These built-in intrinsics for the ARM Advanced SIMD extension are available
8211 when the @option{-mfpu=neon} switch is used:
8213 @include arm-neon-intrinsics.texi
8215 @node AVR Built-in Functions
8216 @subsection AVR Built-in Functions
8218 For each built-in function for AVR, there is an equally named,
8219 uppercase built-in macro defined. That way users can easily query if
8220 or if not a specific built-in is implemented or not. For example, if
8221 @code{__builtin_avr_nop} is available the macro
8222 @code{__BUILTIN_AVR_NOP} is defined to @code{1} and undefined otherwise.
8224 The following built-in functions map to the respective machine
8225 instruction, i.e. @code{nop}, @code{sei}, @code{cli}, @code{sleep},
8226 @code{wdr}, @code{swap}, @code{fmul}, @code{fmuls}
8227 resp. @code{fmulsu}. The three @code{fmul*} built-ins are implemented
8228 as library call if no hardware multiplier is available.
8231 void __builtin_avr_nop (void)
8232 void __builtin_avr_sei (void)
8233 void __builtin_avr_cli (void)
8234 void __builtin_avr_sleep (void)
8235 void __builtin_avr_wdr (void)
8236 unsigned char __builtin_avr_swap (unsigned char)
8237 unsigned int __builtin_avr_fmul (unsigned char, unsigned char)
8238 int __builtin_avr_fmuls (char, char)
8239 int __builtin_avr_fmulsu (char, unsigned char)
8242 In order to delay execution for a specific number of cycles, GCC
8245 void __builtin_avr_delay_cycles (unsigned long ticks)
8248 @code{ticks} is the number of ticks to delay execution. Note that this
8249 built-in does not take into account the effect of interrupts which
8250 might increase delay time. @code{ticks} must be a compile time
8251 integer constant; delays with a variable number of cycles are not supported.
8253 @node Blackfin Built-in Functions
8254 @subsection Blackfin Built-in Functions
8256 Currently, there are two Blackfin-specific built-in functions. These are
8257 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
8258 using inline assembly; by using these built-in functions the compiler can
8259 automatically add workarounds for hardware errata involving these
8260 instructions. These functions are named as follows:
8263 void __builtin_bfin_csync (void)
8264 void __builtin_bfin_ssync (void)
8267 @node FR-V Built-in Functions
8268 @subsection FR-V Built-in Functions
8270 GCC provides many FR-V-specific built-in functions. In general,
8271 these functions are intended to be compatible with those described
8272 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
8273 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
8274 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
8275 pointer rather than by value.
8277 Most of the functions are named after specific FR-V instructions.
8278 Such functions are said to be ``directly mapped'' and are summarized
8279 here in tabular form.
8283 * Directly-mapped Integer Functions::
8284 * Directly-mapped Media Functions::
8285 * Raw read/write Functions::
8286 * Other Built-in Functions::
8289 @node Argument Types
8290 @subsubsection Argument Types
8292 The arguments to the built-in functions can be divided into three groups:
8293 register numbers, compile-time constants and run-time values. In order
8294 to make this classification clear at a glance, the arguments and return
8295 values are given the following pseudo types:
8297 @multitable @columnfractions .20 .30 .15 .35
8298 @item Pseudo type @tab Real C type @tab Constant? @tab Description
8299 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
8300 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
8301 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
8302 @item @code{uw2} @tab @code{unsigned long long} @tab No
8303 @tab an unsigned doubleword
8304 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
8305 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
8306 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
8307 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
8310 These pseudo types are not defined by GCC, they are simply a notational
8311 convenience used in this manual.
8313 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
8314 and @code{sw2} are evaluated at run time. They correspond to
8315 register operands in the underlying FR-V instructions.
8317 @code{const} arguments represent immediate operands in the underlying
8318 FR-V instructions. They must be compile-time constants.
8320 @code{acc} arguments are evaluated at compile time and specify the number
8321 of an accumulator register. For example, an @code{acc} argument of 2
8322 will select the ACC2 register.
8324 @code{iacc} arguments are similar to @code{acc} arguments but specify the
8325 number of an IACC register. See @pxref{Other Built-in Functions}
8328 @node Directly-mapped Integer Functions
8329 @subsubsection Directly-mapped Integer Functions
8331 The functions listed below map directly to FR-V I-type instructions.
8333 @multitable @columnfractions .45 .32 .23
8334 @item Function prototype @tab Example usage @tab Assembly output
8335 @item @code{sw1 __ADDSS (sw1, sw1)}
8336 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
8337 @tab @code{ADDSS @var{a},@var{b},@var{c}}
8338 @item @code{sw1 __SCAN (sw1, sw1)}
8339 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
8340 @tab @code{SCAN @var{a},@var{b},@var{c}}
8341 @item @code{sw1 __SCUTSS (sw1)}
8342 @tab @code{@var{b} = __SCUTSS (@var{a})}
8343 @tab @code{SCUTSS @var{a},@var{b}}
8344 @item @code{sw1 __SLASS (sw1, sw1)}
8345 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
8346 @tab @code{SLASS @var{a},@var{b},@var{c}}
8347 @item @code{void __SMASS (sw1, sw1)}
8348 @tab @code{__SMASS (@var{a}, @var{b})}
8349 @tab @code{SMASS @var{a},@var{b}}
8350 @item @code{void __SMSSS (sw1, sw1)}
8351 @tab @code{__SMSSS (@var{a}, @var{b})}
8352 @tab @code{SMSSS @var{a},@var{b}}
8353 @item @code{void __SMU (sw1, sw1)}
8354 @tab @code{__SMU (@var{a}, @var{b})}
8355 @tab @code{SMU @var{a},@var{b}}
8356 @item @code{sw2 __SMUL (sw1, sw1)}
8357 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
8358 @tab @code{SMUL @var{a},@var{b},@var{c}}
8359 @item @code{sw1 __SUBSS (sw1, sw1)}
8360 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
8361 @tab @code{SUBSS @var{a},@var{b},@var{c}}
8362 @item @code{uw2 __UMUL (uw1, uw1)}
8363 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
8364 @tab @code{UMUL @var{a},@var{b},@var{c}}
8367 @node Directly-mapped Media Functions
8368 @subsubsection Directly-mapped Media Functions
8370 The functions listed below map directly to FR-V M-type instructions.
8372 @multitable @columnfractions .45 .32 .23
8373 @item Function prototype @tab Example usage @tab Assembly output
8374 @item @code{uw1 __MABSHS (sw1)}
8375 @tab @code{@var{b} = __MABSHS (@var{a})}
8376 @tab @code{MABSHS @var{a},@var{b}}
8377 @item @code{void __MADDACCS (acc, acc)}
8378 @tab @code{__MADDACCS (@var{b}, @var{a})}
8379 @tab @code{MADDACCS @var{a},@var{b}}
8380 @item @code{sw1 __MADDHSS (sw1, sw1)}
8381 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
8382 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
8383 @item @code{uw1 __MADDHUS (uw1, uw1)}
8384 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
8385 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
8386 @item @code{uw1 __MAND (uw1, uw1)}
8387 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
8388 @tab @code{MAND @var{a},@var{b},@var{c}}
8389 @item @code{void __MASACCS (acc, acc)}
8390 @tab @code{__MASACCS (@var{b}, @var{a})}
8391 @tab @code{MASACCS @var{a},@var{b}}
8392 @item @code{uw1 __MAVEH (uw1, uw1)}
8393 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
8394 @tab @code{MAVEH @var{a},@var{b},@var{c}}
8395 @item @code{uw2 __MBTOH (uw1)}
8396 @tab @code{@var{b} = __MBTOH (@var{a})}
8397 @tab @code{MBTOH @var{a},@var{b}}
8398 @item @code{void __MBTOHE (uw1 *, uw1)}
8399 @tab @code{__MBTOHE (&@var{b}, @var{a})}
8400 @tab @code{MBTOHE @var{a},@var{b}}
8401 @item @code{void __MCLRACC (acc)}
8402 @tab @code{__MCLRACC (@var{a})}
8403 @tab @code{MCLRACC @var{a}}
8404 @item @code{void __MCLRACCA (void)}
8405 @tab @code{__MCLRACCA ()}
8406 @tab @code{MCLRACCA}
8407 @item @code{uw1 __Mcop1 (uw1, uw1)}
8408 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
8409 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
8410 @item @code{uw1 __Mcop2 (uw1, uw1)}
8411 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
8412 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
8413 @item @code{uw1 __MCPLHI (uw2, const)}
8414 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
8415 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
8416 @item @code{uw1 __MCPLI (uw2, const)}
8417 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
8418 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
8419 @item @code{void __MCPXIS (acc, sw1, sw1)}
8420 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
8421 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
8422 @item @code{void __MCPXIU (acc, uw1, uw1)}
8423 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
8424 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
8425 @item @code{void __MCPXRS (acc, sw1, sw1)}
8426 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
8427 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
8428 @item @code{void __MCPXRU (acc, uw1, uw1)}
8429 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
8430 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
8431 @item @code{uw1 __MCUT (acc, uw1)}
8432 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
8433 @tab @code{MCUT @var{a},@var{b},@var{c}}
8434 @item @code{uw1 __MCUTSS (acc, sw1)}
8435 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
8436 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
8437 @item @code{void __MDADDACCS (acc, acc)}
8438 @tab @code{__MDADDACCS (@var{b}, @var{a})}
8439 @tab @code{MDADDACCS @var{a},@var{b}}
8440 @item @code{void __MDASACCS (acc, acc)}
8441 @tab @code{__MDASACCS (@var{b}, @var{a})}
8442 @tab @code{MDASACCS @var{a},@var{b}}
8443 @item @code{uw2 __MDCUTSSI (acc, const)}
8444 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
8445 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
8446 @item @code{uw2 __MDPACKH (uw2, uw2)}
8447 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
8448 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
8449 @item @code{uw2 __MDROTLI (uw2, const)}
8450 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
8451 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
8452 @item @code{void __MDSUBACCS (acc, acc)}
8453 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
8454 @tab @code{MDSUBACCS @var{a},@var{b}}
8455 @item @code{void __MDUNPACKH (uw1 *, uw2)}
8456 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
8457 @tab @code{MDUNPACKH @var{a},@var{b}}
8458 @item @code{uw2 __MEXPDHD (uw1, const)}
8459 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
8460 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
8461 @item @code{uw1 __MEXPDHW (uw1, const)}
8462 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
8463 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
8464 @item @code{uw1 __MHDSETH (uw1, const)}
8465 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
8466 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
8467 @item @code{sw1 __MHDSETS (const)}
8468 @tab @code{@var{b} = __MHDSETS (@var{a})}
8469 @tab @code{MHDSETS #@var{a},@var{b}}
8470 @item @code{uw1 __MHSETHIH (uw1, const)}
8471 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
8472 @tab @code{MHSETHIH #@var{a},@var{b}}
8473 @item @code{sw1 __MHSETHIS (sw1, const)}
8474 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
8475 @tab @code{MHSETHIS #@var{a},@var{b}}
8476 @item @code{uw1 __MHSETLOH (uw1, const)}
8477 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
8478 @tab @code{MHSETLOH #@var{a},@var{b}}
8479 @item @code{sw1 __MHSETLOS (sw1, const)}
8480 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
8481 @tab @code{MHSETLOS #@var{a},@var{b}}
8482 @item @code{uw1 __MHTOB (uw2)}
8483 @tab @code{@var{b} = __MHTOB (@var{a})}
8484 @tab @code{MHTOB @var{a},@var{b}}
8485 @item @code{void __MMACHS (acc, sw1, sw1)}
8486 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
8487 @tab @code{MMACHS @var{a},@var{b},@var{c}}
8488 @item @code{void __MMACHU (acc, uw1, uw1)}
8489 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
8490 @tab @code{MMACHU @var{a},@var{b},@var{c}}
8491 @item @code{void __MMRDHS (acc, sw1, sw1)}
8492 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
8493 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
8494 @item @code{void __MMRDHU (acc, uw1, uw1)}
8495 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
8496 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
8497 @item @code{void __MMULHS (acc, sw1, sw1)}
8498 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
8499 @tab @code{MMULHS @var{a},@var{b},@var{c}}
8500 @item @code{void __MMULHU (acc, uw1, uw1)}
8501 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
8502 @tab @code{MMULHU @var{a},@var{b},@var{c}}
8503 @item @code{void __MMULXHS (acc, sw1, sw1)}
8504 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
8505 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
8506 @item @code{void __MMULXHU (acc, uw1, uw1)}
8507 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
8508 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
8509 @item @code{uw1 __MNOT (uw1)}
8510 @tab @code{@var{b} = __MNOT (@var{a})}
8511 @tab @code{MNOT @var{a},@var{b}}
8512 @item @code{uw1 __MOR (uw1, uw1)}
8513 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
8514 @tab @code{MOR @var{a},@var{b},@var{c}}
8515 @item @code{uw1 __MPACKH (uh, uh)}
8516 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
8517 @tab @code{MPACKH @var{a},@var{b},@var{c}}
8518 @item @code{sw2 __MQADDHSS (sw2, sw2)}
8519 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
8520 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
8521 @item @code{uw2 __MQADDHUS (uw2, uw2)}
8522 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
8523 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
8524 @item @code{void __MQCPXIS (acc, sw2, sw2)}
8525 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
8526 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
8527 @item @code{void __MQCPXIU (acc, uw2, uw2)}
8528 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
8529 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
8530 @item @code{void __MQCPXRS (acc, sw2, sw2)}
8531 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
8532 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
8533 @item @code{void __MQCPXRU (acc, uw2, uw2)}
8534 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
8535 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
8536 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
8537 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
8538 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
8539 @item @code{sw2 __MQLMTHS (sw2, sw2)}
8540 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
8541 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
8542 @item @code{void __MQMACHS (acc, sw2, sw2)}
8543 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
8544 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
8545 @item @code{void __MQMACHU (acc, uw2, uw2)}
8546 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
8547 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
8548 @item @code{void __MQMACXHS (acc, sw2, sw2)}
8549 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
8550 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
8551 @item @code{void __MQMULHS (acc, sw2, sw2)}
8552 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
8553 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
8554 @item @code{void __MQMULHU (acc, uw2, uw2)}
8555 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
8556 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
8557 @item @code{void __MQMULXHS (acc, sw2, sw2)}
8558 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
8559 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
8560 @item @code{void __MQMULXHU (acc, uw2, uw2)}
8561 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
8562 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
8563 @item @code{sw2 __MQSATHS (sw2, sw2)}
8564 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
8565 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
8566 @item @code{uw2 __MQSLLHI (uw2, int)}
8567 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
8568 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
8569 @item @code{sw2 __MQSRAHI (sw2, int)}
8570 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
8571 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
8572 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
8573 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
8574 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
8575 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
8576 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
8577 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
8578 @item @code{void __MQXMACHS (acc, sw2, sw2)}
8579 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
8580 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
8581 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
8582 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
8583 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
8584 @item @code{uw1 __MRDACC (acc)}
8585 @tab @code{@var{b} = __MRDACC (@var{a})}
8586 @tab @code{MRDACC @var{a},@var{b}}
8587 @item @code{uw1 __MRDACCG (acc)}
8588 @tab @code{@var{b} = __MRDACCG (@var{a})}
8589 @tab @code{MRDACCG @var{a},@var{b}}
8590 @item @code{uw1 __MROTLI (uw1, const)}
8591 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
8592 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
8593 @item @code{uw1 __MROTRI (uw1, const)}
8594 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
8595 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
8596 @item @code{sw1 __MSATHS (sw1, sw1)}
8597 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
8598 @tab @code{MSATHS @var{a},@var{b},@var{c}}
8599 @item @code{uw1 __MSATHU (uw1, uw1)}
8600 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
8601 @tab @code{MSATHU @var{a},@var{b},@var{c}}
8602 @item @code{uw1 __MSLLHI (uw1, const)}
8603 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
8604 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
8605 @item @code{sw1 __MSRAHI (sw1, const)}
8606 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
8607 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
8608 @item @code{uw1 __MSRLHI (uw1, const)}
8609 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
8610 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
8611 @item @code{void __MSUBACCS (acc, acc)}
8612 @tab @code{__MSUBACCS (@var{b}, @var{a})}
8613 @tab @code{MSUBACCS @var{a},@var{b}}
8614 @item @code{sw1 __MSUBHSS (sw1, sw1)}
8615 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
8616 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
8617 @item @code{uw1 __MSUBHUS (uw1, uw1)}
8618 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
8619 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
8620 @item @code{void __MTRAP (void)}
8621 @tab @code{__MTRAP ()}
8623 @item @code{uw2 __MUNPACKH (uw1)}
8624 @tab @code{@var{b} = __MUNPACKH (@var{a})}
8625 @tab @code{MUNPACKH @var{a},@var{b}}
8626 @item @code{uw1 __MWCUT (uw2, uw1)}
8627 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
8628 @tab @code{MWCUT @var{a},@var{b},@var{c}}
8629 @item @code{void __MWTACC (acc, uw1)}
8630 @tab @code{__MWTACC (@var{b}, @var{a})}
8631 @tab @code{MWTACC @var{a},@var{b}}
8632 @item @code{void __MWTACCG (acc, uw1)}
8633 @tab @code{__MWTACCG (@var{b}, @var{a})}
8634 @tab @code{MWTACCG @var{a},@var{b}}
8635 @item @code{uw1 __MXOR (uw1, uw1)}
8636 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
8637 @tab @code{MXOR @var{a},@var{b},@var{c}}
8640 @node Raw read/write Functions
8641 @subsubsection Raw read/write Functions
8643 This sections describes built-in functions related to read and write
8644 instructions to access memory. These functions generate
8645 @code{membar} instructions to flush the I/O load and stores where
8646 appropriate, as described in Fujitsu's manual described above.
8650 @item unsigned char __builtin_read8 (void *@var{data})
8651 @item unsigned short __builtin_read16 (void *@var{data})
8652 @item unsigned long __builtin_read32 (void *@var{data})
8653 @item unsigned long long __builtin_read64 (void *@var{data})
8655 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
8656 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
8657 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
8658 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
8661 @node Other Built-in Functions
8662 @subsubsection Other Built-in Functions
8664 This section describes built-in functions that are not named after
8665 a specific FR-V instruction.
8668 @item sw2 __IACCreadll (iacc @var{reg})
8669 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
8670 for future expansion and must be 0.
8672 @item sw1 __IACCreadl (iacc @var{reg})
8673 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
8674 Other values of @var{reg} are rejected as invalid.
8676 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
8677 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
8678 is reserved for future expansion and must be 0.
8680 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
8681 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
8682 is 1. Other values of @var{reg} are rejected as invalid.
8684 @item void __data_prefetch0 (const void *@var{x})
8685 Use the @code{dcpl} instruction to load the contents of address @var{x}
8686 into the data cache.
8688 @item void __data_prefetch (const void *@var{x})
8689 Use the @code{nldub} instruction to load the contents of address @var{x}
8690 into the data cache. The instruction will be issued in slot I1@.
8693 @node X86 Built-in Functions
8694 @subsection X86 Built-in Functions
8696 These built-in functions are available for the i386 and x86-64 family
8697 of computers, depending on the command-line switches used.
8699 Note that, if you specify command-line switches such as @option{-msse},
8700 the compiler could use the extended instruction sets even if the built-ins
8701 are not used explicitly in the program. For this reason, applications
8702 which perform runtime CPU detection must compile separate files for each
8703 supported architecture, using the appropriate flags. In particular,
8704 the file containing the CPU detection code should be compiled without
8707 The following machine modes are available for use with MMX built-in functions
8708 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
8709 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
8710 vector of eight 8-bit integers. Some of the built-in functions operate on
8711 MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode.
8713 If 3DNow!@: extensions are enabled, @code{V2SF} is used as a mode for a vector
8714 of two 32-bit floating point values.
8716 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
8717 floating point values. Some instructions use a vector of four 32-bit
8718 integers, these use @code{V4SI}. Finally, some instructions operate on an
8719 entire vector register, interpreting it as a 128-bit integer, these use mode
8722 In 64-bit mode, the x86-64 family of processors uses additional built-in
8723 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
8724 floating point and @code{TC} 128-bit complex floating point values.
8726 The following floating point built-in functions are available in 64-bit
8727 mode. All of them implement the function that is part of the name.
8730 __float128 __builtin_fabsq (__float128)
8731 __float128 __builtin_copysignq (__float128, __float128)
8734 The following built-in function is always available.
8737 @item void __builtin_ia32_pause (void)
8738 Generates the @code{pause} machine instruction with a compiler memory
8742 The following floating point built-in functions are made available in the
8746 @item __float128 __builtin_infq (void)
8747 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
8748 @findex __builtin_infq
8750 @item __float128 __builtin_huge_valq (void)
8751 Similar to @code{__builtin_huge_val}, except the return type is @code{__float128}.
8752 @findex __builtin_huge_valq
8755 The following built-in functions are made available by @option{-mmmx}.
8756 All of them generate the machine instruction that is part of the name.
8759 v8qi __builtin_ia32_paddb (v8qi, v8qi)
8760 v4hi __builtin_ia32_paddw (v4hi, v4hi)
8761 v2si __builtin_ia32_paddd (v2si, v2si)
8762 v8qi __builtin_ia32_psubb (v8qi, v8qi)
8763 v4hi __builtin_ia32_psubw (v4hi, v4hi)
8764 v2si __builtin_ia32_psubd (v2si, v2si)
8765 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
8766 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
8767 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
8768 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
8769 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
8770 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
8771 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
8772 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
8773 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
8774 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
8775 di __builtin_ia32_pand (di, di)
8776 di __builtin_ia32_pandn (di,di)
8777 di __builtin_ia32_por (di, di)
8778 di __builtin_ia32_pxor (di, di)
8779 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
8780 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
8781 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
8782 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
8783 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
8784 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
8785 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
8786 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
8787 v2si __builtin_ia32_punpckhdq (v2si, v2si)
8788 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
8789 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
8790 v2si __builtin_ia32_punpckldq (v2si, v2si)
8791 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
8792 v4hi __builtin_ia32_packssdw (v2si, v2si)
8793 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
8795 v4hi __builtin_ia32_psllw (v4hi, v4hi)
8796 v2si __builtin_ia32_pslld (v2si, v2si)
8797 v1di __builtin_ia32_psllq (v1di, v1di)
8798 v4hi __builtin_ia32_psrlw (v4hi, v4hi)
8799 v2si __builtin_ia32_psrld (v2si, v2si)
8800 v1di __builtin_ia32_psrlq (v1di, v1di)
8801 v4hi __builtin_ia32_psraw (v4hi, v4hi)
8802 v2si __builtin_ia32_psrad (v2si, v2si)
8803 v4hi __builtin_ia32_psllwi (v4hi, int)
8804 v2si __builtin_ia32_pslldi (v2si, int)
8805 v1di __builtin_ia32_psllqi (v1di, int)
8806 v4hi __builtin_ia32_psrlwi (v4hi, int)
8807 v2si __builtin_ia32_psrldi (v2si, int)
8808 v1di __builtin_ia32_psrlqi (v1di, int)
8809 v4hi __builtin_ia32_psrawi (v4hi, int)
8810 v2si __builtin_ia32_psradi (v2si, int)
8814 The following built-in functions are made available either with
8815 @option{-msse}, or with a combination of @option{-m3dnow} and
8816 @option{-march=athlon}. All of them generate the machine
8817 instruction that is part of the name.
8820 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
8821 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
8822 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
8823 v1di __builtin_ia32_psadbw (v8qi, v8qi)
8824 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
8825 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
8826 v8qi __builtin_ia32_pminub (v8qi, v8qi)
8827 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
8828 int __builtin_ia32_pextrw (v4hi, int)
8829 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
8830 int __builtin_ia32_pmovmskb (v8qi)
8831 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
8832 void __builtin_ia32_movntq (di *, di)
8833 void __builtin_ia32_sfence (void)
8836 The following built-in functions are available when @option{-msse} is used.
8837 All of them generate the machine instruction that is part of the name.
8840 int __builtin_ia32_comieq (v4sf, v4sf)
8841 int __builtin_ia32_comineq (v4sf, v4sf)
8842 int __builtin_ia32_comilt (v4sf, v4sf)
8843 int __builtin_ia32_comile (v4sf, v4sf)
8844 int __builtin_ia32_comigt (v4sf, v4sf)
8845 int __builtin_ia32_comige (v4sf, v4sf)
8846 int __builtin_ia32_ucomieq (v4sf, v4sf)
8847 int __builtin_ia32_ucomineq (v4sf, v4sf)
8848 int __builtin_ia32_ucomilt (v4sf, v4sf)
8849 int __builtin_ia32_ucomile (v4sf, v4sf)
8850 int __builtin_ia32_ucomigt (v4sf, v4sf)
8851 int __builtin_ia32_ucomige (v4sf, v4sf)
8852 v4sf __builtin_ia32_addps (v4sf, v4sf)
8853 v4sf __builtin_ia32_subps (v4sf, v4sf)
8854 v4sf __builtin_ia32_mulps (v4sf, v4sf)
8855 v4sf __builtin_ia32_divps (v4sf, v4sf)
8856 v4sf __builtin_ia32_addss (v4sf, v4sf)
8857 v4sf __builtin_ia32_subss (v4sf, v4sf)
8858 v4sf __builtin_ia32_mulss (v4sf, v4sf)
8859 v4sf __builtin_ia32_divss (v4sf, v4sf)
8860 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
8861 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
8862 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
8863 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
8864 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
8865 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
8866 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
8867 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
8868 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
8869 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
8870 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
8871 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
8872 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
8873 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
8874 v4si __builtin_ia32_cmpless (v4sf, v4sf)
8875 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
8876 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
8877 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
8878 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
8879 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
8880 v4sf __builtin_ia32_maxps (v4sf, v4sf)
8881 v4sf __builtin_ia32_maxss (v4sf, v4sf)
8882 v4sf __builtin_ia32_minps (v4sf, v4sf)
8883 v4sf __builtin_ia32_minss (v4sf, v4sf)
8884 v4sf __builtin_ia32_andps (v4sf, v4sf)
8885 v4sf __builtin_ia32_andnps (v4sf, v4sf)
8886 v4sf __builtin_ia32_orps (v4sf, v4sf)
8887 v4sf __builtin_ia32_xorps (v4sf, v4sf)
8888 v4sf __builtin_ia32_movss (v4sf, v4sf)
8889 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
8890 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
8891 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
8892 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
8893 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
8894 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
8895 v2si __builtin_ia32_cvtps2pi (v4sf)
8896 int __builtin_ia32_cvtss2si (v4sf)
8897 v2si __builtin_ia32_cvttps2pi (v4sf)
8898 int __builtin_ia32_cvttss2si (v4sf)
8899 v4sf __builtin_ia32_rcpps (v4sf)
8900 v4sf __builtin_ia32_rsqrtps (v4sf)
8901 v4sf __builtin_ia32_sqrtps (v4sf)
8902 v4sf __builtin_ia32_rcpss (v4sf)
8903 v4sf __builtin_ia32_rsqrtss (v4sf)
8904 v4sf __builtin_ia32_sqrtss (v4sf)
8905 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
8906 void __builtin_ia32_movntps (float *, v4sf)
8907 int __builtin_ia32_movmskps (v4sf)
8910 The following built-in functions are available when @option{-msse} is used.
8913 @item v4sf __builtin_ia32_loadaps (float *)
8914 Generates the @code{movaps} machine instruction as a load from memory.
8915 @item void __builtin_ia32_storeaps (float *, v4sf)
8916 Generates the @code{movaps} machine instruction as a store to memory.
8917 @item v4sf __builtin_ia32_loadups (float *)
8918 Generates the @code{movups} machine instruction as a load from memory.
8919 @item void __builtin_ia32_storeups (float *, v4sf)
8920 Generates the @code{movups} machine instruction as a store to memory.
8921 @item v4sf __builtin_ia32_loadsss (float *)
8922 Generates the @code{movss} machine instruction as a load from memory.
8923 @item void __builtin_ia32_storess (float *, v4sf)
8924 Generates the @code{movss} machine instruction as a store to memory.
8925 @item v4sf __builtin_ia32_loadhps (v4sf, const v2sf *)
8926 Generates the @code{movhps} machine instruction as a load from memory.
8927 @item v4sf __builtin_ia32_loadlps (v4sf, const v2sf *)
8928 Generates the @code{movlps} machine instruction as a load from memory
8929 @item void __builtin_ia32_storehps (v2sf *, v4sf)
8930 Generates the @code{movhps} machine instruction as a store to memory.
8931 @item void __builtin_ia32_storelps (v2sf *, v4sf)
8932 Generates the @code{movlps} machine instruction as a store to memory.
8935 The following built-in functions are available when @option{-msse2} is used.
8936 All of them generate the machine instruction that is part of the name.
8939 int __builtin_ia32_comisdeq (v2df, v2df)
8940 int __builtin_ia32_comisdlt (v2df, v2df)
8941 int __builtin_ia32_comisdle (v2df, v2df)
8942 int __builtin_ia32_comisdgt (v2df, v2df)
8943 int __builtin_ia32_comisdge (v2df, v2df)
8944 int __builtin_ia32_comisdneq (v2df, v2df)
8945 int __builtin_ia32_ucomisdeq (v2df, v2df)
8946 int __builtin_ia32_ucomisdlt (v2df, v2df)
8947 int __builtin_ia32_ucomisdle (v2df, v2df)
8948 int __builtin_ia32_ucomisdgt (v2df, v2df)
8949 int __builtin_ia32_ucomisdge (v2df, v2df)
8950 int __builtin_ia32_ucomisdneq (v2df, v2df)
8951 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
8952 v2df __builtin_ia32_cmpltpd (v2df, v2df)
8953 v2df __builtin_ia32_cmplepd (v2df, v2df)
8954 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
8955 v2df __builtin_ia32_cmpgepd (v2df, v2df)
8956 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
8957 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
8958 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
8959 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
8960 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
8961 v2df __builtin_ia32_cmpngepd (v2df, v2df)
8962 v2df __builtin_ia32_cmpordpd (v2df, v2df)
8963 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
8964 v2df __builtin_ia32_cmpltsd (v2df, v2df)
8965 v2df __builtin_ia32_cmplesd (v2df, v2df)
8966 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
8967 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
8968 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
8969 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
8970 v2df __builtin_ia32_cmpordsd (v2df, v2df)
8971 v2di __builtin_ia32_paddq (v2di, v2di)
8972 v2di __builtin_ia32_psubq (v2di, v2di)
8973 v2df __builtin_ia32_addpd (v2df, v2df)
8974 v2df __builtin_ia32_subpd (v2df, v2df)
8975 v2df __builtin_ia32_mulpd (v2df, v2df)
8976 v2df __builtin_ia32_divpd (v2df, v2df)
8977 v2df __builtin_ia32_addsd (v2df, v2df)
8978 v2df __builtin_ia32_subsd (v2df, v2df)
8979 v2df __builtin_ia32_mulsd (v2df, v2df)
8980 v2df __builtin_ia32_divsd (v2df, v2df)
8981 v2df __builtin_ia32_minpd (v2df, v2df)
8982 v2df __builtin_ia32_maxpd (v2df, v2df)
8983 v2df __builtin_ia32_minsd (v2df, v2df)
8984 v2df __builtin_ia32_maxsd (v2df, v2df)
8985 v2df __builtin_ia32_andpd (v2df, v2df)
8986 v2df __builtin_ia32_andnpd (v2df, v2df)
8987 v2df __builtin_ia32_orpd (v2df, v2df)
8988 v2df __builtin_ia32_xorpd (v2df, v2df)
8989 v2df __builtin_ia32_movsd (v2df, v2df)
8990 v2df __builtin_ia32_unpckhpd (v2df, v2df)
8991 v2df __builtin_ia32_unpcklpd (v2df, v2df)
8992 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
8993 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
8994 v4si __builtin_ia32_paddd128 (v4si, v4si)
8995 v2di __builtin_ia32_paddq128 (v2di, v2di)
8996 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
8997 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
8998 v4si __builtin_ia32_psubd128 (v4si, v4si)
8999 v2di __builtin_ia32_psubq128 (v2di, v2di)
9000 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
9001 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
9002 v2di __builtin_ia32_pand128 (v2di, v2di)
9003 v2di __builtin_ia32_pandn128 (v2di, v2di)
9004 v2di __builtin_ia32_por128 (v2di, v2di)
9005 v2di __builtin_ia32_pxor128 (v2di, v2di)
9006 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
9007 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
9008 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
9009 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
9010 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
9011 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
9012 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
9013 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
9014 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
9015 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
9016 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
9017 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
9018 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
9019 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
9020 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
9021 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
9022 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
9023 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
9024 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
9025 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
9026 v16qi __builtin_ia32_packsswb128 (v8hi, v8hi)
9027 v8hi __builtin_ia32_packssdw128 (v4si, v4si)
9028 v16qi __builtin_ia32_packuswb128 (v8hi, v8hi)
9029 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
9030 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
9031 v2df __builtin_ia32_loadupd (double *)
9032 void __builtin_ia32_storeupd (double *, v2df)
9033 v2df __builtin_ia32_loadhpd (v2df, double const *)
9034 v2df __builtin_ia32_loadlpd (v2df, double const *)
9035 int __builtin_ia32_movmskpd (v2df)
9036 int __builtin_ia32_pmovmskb128 (v16qi)
9037 void __builtin_ia32_movnti (int *, int)
9038 void __builtin_ia32_movntpd (double *, v2df)
9039 void __builtin_ia32_movntdq (v2df *, v2df)
9040 v4si __builtin_ia32_pshufd (v4si, int)
9041 v8hi __builtin_ia32_pshuflw (v8hi, int)
9042 v8hi __builtin_ia32_pshufhw (v8hi, int)
9043 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
9044 v2df __builtin_ia32_sqrtpd (v2df)
9045 v2df __builtin_ia32_sqrtsd (v2df)
9046 v2df __builtin_ia32_shufpd (v2df, v2df, int)
9047 v2df __builtin_ia32_cvtdq2pd (v4si)
9048 v4sf __builtin_ia32_cvtdq2ps (v4si)
9049 v4si __builtin_ia32_cvtpd2dq (v2df)
9050 v2si __builtin_ia32_cvtpd2pi (v2df)
9051 v4sf __builtin_ia32_cvtpd2ps (v2df)
9052 v4si __builtin_ia32_cvttpd2dq (v2df)
9053 v2si __builtin_ia32_cvttpd2pi (v2df)
9054 v2df __builtin_ia32_cvtpi2pd (v2si)
9055 int __builtin_ia32_cvtsd2si (v2df)
9056 int __builtin_ia32_cvttsd2si (v2df)
9057 long long __builtin_ia32_cvtsd2si64 (v2df)
9058 long long __builtin_ia32_cvttsd2si64 (v2df)
9059 v4si __builtin_ia32_cvtps2dq (v4sf)
9060 v2df __builtin_ia32_cvtps2pd (v4sf)
9061 v4si __builtin_ia32_cvttps2dq (v4sf)
9062 v2df __builtin_ia32_cvtsi2sd (v2df, int)
9063 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
9064 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
9065 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
9066 void __builtin_ia32_clflush (const void *)
9067 void __builtin_ia32_lfence (void)
9068 void __builtin_ia32_mfence (void)
9069 v16qi __builtin_ia32_loaddqu (const char *)
9070 void __builtin_ia32_storedqu (char *, v16qi)
9071 v1di __builtin_ia32_pmuludq (v2si, v2si)
9072 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
9073 v8hi __builtin_ia32_psllw128 (v8hi, v8hi)
9074 v4si __builtin_ia32_pslld128 (v4si, v4si)
9075 v2di __builtin_ia32_psllq128 (v2di, v2di)
9076 v8hi __builtin_ia32_psrlw128 (v8hi, v8hi)
9077 v4si __builtin_ia32_psrld128 (v4si, v4si)
9078 v2di __builtin_ia32_psrlq128 (v2di, v2di)
9079 v8hi __builtin_ia32_psraw128 (v8hi, v8hi)
9080 v4si __builtin_ia32_psrad128 (v4si, v4si)
9081 v2di __builtin_ia32_pslldqi128 (v2di, int)
9082 v8hi __builtin_ia32_psllwi128 (v8hi, int)
9083 v4si __builtin_ia32_pslldi128 (v4si, int)
9084 v2di __builtin_ia32_psllqi128 (v2di, int)
9085 v2di __builtin_ia32_psrldqi128 (v2di, int)
9086 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
9087 v4si __builtin_ia32_psrldi128 (v4si, int)
9088 v2di __builtin_ia32_psrlqi128 (v2di, int)
9089 v8hi __builtin_ia32_psrawi128 (v8hi, int)
9090 v4si __builtin_ia32_psradi128 (v4si, int)
9091 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
9092 v2di __builtin_ia32_movq128 (v2di)
9095 The following built-in functions are available when @option{-msse3} is used.
9096 All of them generate the machine instruction that is part of the name.
9099 v2df __builtin_ia32_addsubpd (v2df, v2df)
9100 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
9101 v2df __builtin_ia32_haddpd (v2df, v2df)
9102 v4sf __builtin_ia32_haddps (v4sf, v4sf)
9103 v2df __builtin_ia32_hsubpd (v2df, v2df)
9104 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
9105 v16qi __builtin_ia32_lddqu (char const *)
9106 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
9107 v2df __builtin_ia32_movddup (v2df)
9108 v4sf __builtin_ia32_movshdup (v4sf)
9109 v4sf __builtin_ia32_movsldup (v4sf)
9110 void __builtin_ia32_mwait (unsigned int, unsigned int)
9113 The following built-in functions are available when @option{-msse3} is used.
9116 @item v2df __builtin_ia32_loadddup (double const *)
9117 Generates the @code{movddup} machine instruction as a load from memory.
9120 The following built-in functions are available when @option{-mssse3} is used.
9121 All of them generate the machine instruction that is part of the name
9125 v2si __builtin_ia32_phaddd (v2si, v2si)
9126 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
9127 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
9128 v2si __builtin_ia32_phsubd (v2si, v2si)
9129 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
9130 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
9131 v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi)
9132 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
9133 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
9134 v8qi __builtin_ia32_psignb (v8qi, v8qi)
9135 v2si __builtin_ia32_psignd (v2si, v2si)
9136 v4hi __builtin_ia32_psignw (v4hi, v4hi)
9137 v1di __builtin_ia32_palignr (v1di, v1di, int)
9138 v8qi __builtin_ia32_pabsb (v8qi)
9139 v2si __builtin_ia32_pabsd (v2si)
9140 v4hi __builtin_ia32_pabsw (v4hi)
9143 The following built-in functions are available when @option{-mssse3} is used.
9144 All of them generate the machine instruction that is part of the name
9148 v4si __builtin_ia32_phaddd128 (v4si, v4si)
9149 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
9150 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
9151 v4si __builtin_ia32_phsubd128 (v4si, v4si)
9152 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
9153 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
9154 v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
9155 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
9156 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
9157 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
9158 v4si __builtin_ia32_psignd128 (v4si, v4si)
9159 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
9160 v2di __builtin_ia32_palignr128 (v2di, v2di, int)
9161 v16qi __builtin_ia32_pabsb128 (v16qi)
9162 v4si __builtin_ia32_pabsd128 (v4si)
9163 v8hi __builtin_ia32_pabsw128 (v8hi)
9166 The following built-in functions are available when @option{-msse4.1} is
9167 used. All of them generate the machine instruction that is part of the
9171 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
9172 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
9173 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
9174 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
9175 v2df __builtin_ia32_dppd (v2df, v2df, const int)
9176 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
9177 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
9178 v2di __builtin_ia32_movntdqa (v2di *);
9179 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
9180 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
9181 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
9182 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
9183 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
9184 v8hi __builtin_ia32_phminposuw128 (v8hi)
9185 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
9186 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
9187 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
9188 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
9189 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
9190 v4si __builtin_ia32_pminsd128 (v4si, v4si)
9191 v4si __builtin_ia32_pminud128 (v4si, v4si)
9192 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
9193 v4si __builtin_ia32_pmovsxbd128 (v16qi)
9194 v2di __builtin_ia32_pmovsxbq128 (v16qi)
9195 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
9196 v2di __builtin_ia32_pmovsxdq128 (v4si)
9197 v4si __builtin_ia32_pmovsxwd128 (v8hi)
9198 v2di __builtin_ia32_pmovsxwq128 (v8hi)
9199 v4si __builtin_ia32_pmovzxbd128 (v16qi)
9200 v2di __builtin_ia32_pmovzxbq128 (v16qi)
9201 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
9202 v2di __builtin_ia32_pmovzxdq128 (v4si)
9203 v4si __builtin_ia32_pmovzxwd128 (v8hi)
9204 v2di __builtin_ia32_pmovzxwq128 (v8hi)
9205 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
9206 v4si __builtin_ia32_pmulld128 (v4si, v4si)
9207 int __builtin_ia32_ptestc128 (v2di, v2di)
9208 int __builtin_ia32_ptestnzc128 (v2di, v2di)
9209 int __builtin_ia32_ptestz128 (v2di, v2di)
9210 v2df __builtin_ia32_roundpd (v2df, const int)
9211 v4sf __builtin_ia32_roundps (v4sf, const int)
9212 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
9213 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
9216 The following built-in functions are available when @option{-msse4.1} is
9220 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
9221 Generates the @code{insertps} machine instruction.
9222 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
9223 Generates the @code{pextrb} machine instruction.
9224 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
9225 Generates the @code{pinsrb} machine instruction.
9226 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
9227 Generates the @code{pinsrd} machine instruction.
9228 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
9229 Generates the @code{pinsrq} machine instruction in 64bit mode.
9232 The following built-in functions are changed to generate new SSE4.1
9233 instructions when @option{-msse4.1} is used.
9236 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
9237 Generates the @code{extractps} machine instruction.
9238 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
9239 Generates the @code{pextrd} machine instruction.
9240 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
9241 Generates the @code{pextrq} machine instruction in 64bit mode.
9244 The following built-in functions are available when @option{-msse4.2} is
9245 used. All of them generate the machine instruction that is part of the
9249 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
9250 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
9251 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
9252 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
9253 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
9254 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
9255 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
9256 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
9257 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
9258 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
9259 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
9260 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
9261 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
9262 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
9263 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
9266 The following built-in functions are available when @option{-msse4.2} is
9270 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
9271 Generates the @code{crc32b} machine instruction.
9272 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
9273 Generates the @code{crc32w} machine instruction.
9274 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
9275 Generates the @code{crc32l} machine instruction.
9276 @item unsigned long long __builtin_ia32_crc32di (unsigned long long, unsigned long long)
9277 Generates the @code{crc32q} machine instruction.
9280 The following built-in functions are changed to generate new SSE4.2
9281 instructions when @option{-msse4.2} is used.
9284 @item int __builtin_popcount (unsigned int)
9285 Generates the @code{popcntl} machine instruction.
9286 @item int __builtin_popcountl (unsigned long)
9287 Generates the @code{popcntl} or @code{popcntq} machine instruction,
9288 depending on the size of @code{unsigned long}.
9289 @item int __builtin_popcountll (unsigned long long)
9290 Generates the @code{popcntq} machine instruction.
9293 The following built-in functions are available when @option{-mavx} is
9294 used. All of them generate the machine instruction that is part of the
9298 v4df __builtin_ia32_addpd256 (v4df,v4df)
9299 v8sf __builtin_ia32_addps256 (v8sf,v8sf)
9300 v4df __builtin_ia32_addsubpd256 (v4df,v4df)
9301 v8sf __builtin_ia32_addsubps256 (v8sf,v8sf)
9302 v4df __builtin_ia32_andnpd256 (v4df,v4df)
9303 v8sf __builtin_ia32_andnps256 (v8sf,v8sf)
9304 v4df __builtin_ia32_andpd256 (v4df,v4df)
9305 v8sf __builtin_ia32_andps256 (v8sf,v8sf)
9306 v4df __builtin_ia32_blendpd256 (v4df,v4df,int)
9307 v8sf __builtin_ia32_blendps256 (v8sf,v8sf,int)
9308 v4df __builtin_ia32_blendvpd256 (v4df,v4df,v4df)
9309 v8sf __builtin_ia32_blendvps256 (v8sf,v8sf,v8sf)
9310 v2df __builtin_ia32_cmppd (v2df,v2df,int)
9311 v4df __builtin_ia32_cmppd256 (v4df,v4df,int)
9312 v4sf __builtin_ia32_cmpps (v4sf,v4sf,int)
9313 v8sf __builtin_ia32_cmpps256 (v8sf,v8sf,int)
9314 v2df __builtin_ia32_cmpsd (v2df,v2df,int)
9315 v4sf __builtin_ia32_cmpss (v4sf,v4sf,int)
9316 v4df __builtin_ia32_cvtdq2pd256 (v4si)
9317 v8sf __builtin_ia32_cvtdq2ps256 (v8si)
9318 v4si __builtin_ia32_cvtpd2dq256 (v4df)
9319 v4sf __builtin_ia32_cvtpd2ps256 (v4df)
9320 v8si __builtin_ia32_cvtps2dq256 (v8sf)
9321 v4df __builtin_ia32_cvtps2pd256 (v4sf)
9322 v4si __builtin_ia32_cvttpd2dq256 (v4df)
9323 v8si __builtin_ia32_cvttps2dq256 (v8sf)
9324 v4df __builtin_ia32_divpd256 (v4df,v4df)
9325 v8sf __builtin_ia32_divps256 (v8sf,v8sf)
9326 v8sf __builtin_ia32_dpps256 (v8sf,v8sf,int)
9327 v4df __builtin_ia32_haddpd256 (v4df,v4df)
9328 v8sf __builtin_ia32_haddps256 (v8sf,v8sf)
9329 v4df __builtin_ia32_hsubpd256 (v4df,v4df)
9330 v8sf __builtin_ia32_hsubps256 (v8sf,v8sf)
9331 v32qi __builtin_ia32_lddqu256 (pcchar)
9332 v32qi __builtin_ia32_loaddqu256 (pcchar)
9333 v4df __builtin_ia32_loadupd256 (pcdouble)
9334 v8sf __builtin_ia32_loadups256 (pcfloat)
9335 v2df __builtin_ia32_maskloadpd (pcv2df,v2df)
9336 v4df __builtin_ia32_maskloadpd256 (pcv4df,v4df)
9337 v4sf __builtin_ia32_maskloadps (pcv4sf,v4sf)
9338 v8sf __builtin_ia32_maskloadps256 (pcv8sf,v8sf)
9339 void __builtin_ia32_maskstorepd (pv2df,v2df,v2df)
9340 void __builtin_ia32_maskstorepd256 (pv4df,v4df,v4df)
9341 void __builtin_ia32_maskstoreps (pv4sf,v4sf,v4sf)
9342 void __builtin_ia32_maskstoreps256 (pv8sf,v8sf,v8sf)
9343 v4df __builtin_ia32_maxpd256 (v4df,v4df)
9344 v8sf __builtin_ia32_maxps256 (v8sf,v8sf)
9345 v4df __builtin_ia32_minpd256 (v4df,v4df)
9346 v8sf __builtin_ia32_minps256 (v8sf,v8sf)
9347 v4df __builtin_ia32_movddup256 (v4df)
9348 int __builtin_ia32_movmskpd256 (v4df)
9349 int __builtin_ia32_movmskps256 (v8sf)
9350 v8sf __builtin_ia32_movshdup256 (v8sf)
9351 v8sf __builtin_ia32_movsldup256 (v8sf)
9352 v4df __builtin_ia32_mulpd256 (v4df,v4df)
9353 v8sf __builtin_ia32_mulps256 (v8sf,v8sf)
9354 v4df __builtin_ia32_orpd256 (v4df,v4df)
9355 v8sf __builtin_ia32_orps256 (v8sf,v8sf)
9356 v2df __builtin_ia32_pd_pd256 (v4df)
9357 v4df __builtin_ia32_pd256_pd (v2df)
9358 v4sf __builtin_ia32_ps_ps256 (v8sf)
9359 v8sf __builtin_ia32_ps256_ps (v4sf)
9360 int __builtin_ia32_ptestc256 (v4di,v4di,ptest)
9361 int __builtin_ia32_ptestnzc256 (v4di,v4di,ptest)
9362 int __builtin_ia32_ptestz256 (v4di,v4di,ptest)
9363 v8sf __builtin_ia32_rcpps256 (v8sf)
9364 v4df __builtin_ia32_roundpd256 (v4df,int)
9365 v8sf __builtin_ia32_roundps256 (v8sf,int)
9366 v8sf __builtin_ia32_rsqrtps_nr256 (v8sf)
9367 v8sf __builtin_ia32_rsqrtps256 (v8sf)
9368 v4df __builtin_ia32_shufpd256 (v4df,v4df,int)
9369 v8sf __builtin_ia32_shufps256 (v8sf,v8sf,int)
9370 v4si __builtin_ia32_si_si256 (v8si)
9371 v8si __builtin_ia32_si256_si (v4si)
9372 v4df __builtin_ia32_sqrtpd256 (v4df)
9373 v8sf __builtin_ia32_sqrtps_nr256 (v8sf)
9374 v8sf __builtin_ia32_sqrtps256 (v8sf)
9375 void __builtin_ia32_storedqu256 (pchar,v32qi)
9376 void __builtin_ia32_storeupd256 (pdouble,v4df)
9377 void __builtin_ia32_storeups256 (pfloat,v8sf)
9378 v4df __builtin_ia32_subpd256 (v4df,v4df)
9379 v8sf __builtin_ia32_subps256 (v8sf,v8sf)
9380 v4df __builtin_ia32_unpckhpd256 (v4df,v4df)
9381 v8sf __builtin_ia32_unpckhps256 (v8sf,v8sf)
9382 v4df __builtin_ia32_unpcklpd256 (v4df,v4df)
9383 v8sf __builtin_ia32_unpcklps256 (v8sf,v8sf)
9384 v4df __builtin_ia32_vbroadcastf128_pd256 (pcv2df)
9385 v8sf __builtin_ia32_vbroadcastf128_ps256 (pcv4sf)
9386 v4df __builtin_ia32_vbroadcastsd256 (pcdouble)
9387 v4sf __builtin_ia32_vbroadcastss (pcfloat)
9388 v8sf __builtin_ia32_vbroadcastss256 (pcfloat)
9389 v2df __builtin_ia32_vextractf128_pd256 (v4df,int)
9390 v4sf __builtin_ia32_vextractf128_ps256 (v8sf,int)
9391 v4si __builtin_ia32_vextractf128_si256 (v8si,int)
9392 v4df __builtin_ia32_vinsertf128_pd256 (v4df,v2df,int)
9393 v8sf __builtin_ia32_vinsertf128_ps256 (v8sf,v4sf,int)
9394 v8si __builtin_ia32_vinsertf128_si256 (v8si,v4si,int)
9395 v4df __builtin_ia32_vperm2f128_pd256 (v4df,v4df,int)
9396 v8sf __builtin_ia32_vperm2f128_ps256 (v8sf,v8sf,int)
9397 v8si __builtin_ia32_vperm2f128_si256 (v8si,v8si,int)
9398 v2df __builtin_ia32_vpermil2pd (v2df,v2df,v2di,int)
9399 v4df __builtin_ia32_vpermil2pd256 (v4df,v4df,v4di,int)
9400 v4sf __builtin_ia32_vpermil2ps (v4sf,v4sf,v4si,int)
9401 v8sf __builtin_ia32_vpermil2ps256 (v8sf,v8sf,v8si,int)
9402 v2df __builtin_ia32_vpermilpd (v2df,int)
9403 v4df __builtin_ia32_vpermilpd256 (v4df,int)
9404 v4sf __builtin_ia32_vpermilps (v4sf,int)
9405 v8sf __builtin_ia32_vpermilps256 (v8sf,int)
9406 v2df __builtin_ia32_vpermilvarpd (v2df,v2di)
9407 v4df __builtin_ia32_vpermilvarpd256 (v4df,v4di)
9408 v4sf __builtin_ia32_vpermilvarps (v4sf,v4si)
9409 v8sf __builtin_ia32_vpermilvarps256 (v8sf,v8si)
9410 int __builtin_ia32_vtestcpd (v2df,v2df,ptest)
9411 int __builtin_ia32_vtestcpd256 (v4df,v4df,ptest)
9412 int __builtin_ia32_vtestcps (v4sf,v4sf,ptest)
9413 int __builtin_ia32_vtestcps256 (v8sf,v8sf,ptest)
9414 int __builtin_ia32_vtestnzcpd (v2df,v2df,ptest)
9415 int __builtin_ia32_vtestnzcpd256 (v4df,v4df,ptest)
9416 int __builtin_ia32_vtestnzcps (v4sf,v4sf,ptest)
9417 int __builtin_ia32_vtestnzcps256 (v8sf,v8sf,ptest)
9418 int __builtin_ia32_vtestzpd (v2df,v2df,ptest)
9419 int __builtin_ia32_vtestzpd256 (v4df,v4df,ptest)
9420 int __builtin_ia32_vtestzps (v4sf,v4sf,ptest)
9421 int __builtin_ia32_vtestzps256 (v8sf,v8sf,ptest)
9422 void __builtin_ia32_vzeroall (void)
9423 void __builtin_ia32_vzeroupper (void)
9424 v4df __builtin_ia32_xorpd256 (v4df,v4df)
9425 v8sf __builtin_ia32_xorps256 (v8sf,v8sf)
9428 The following built-in functions are available when @option{-maes} is
9429 used. All of them generate the machine instruction that is part of the
9433 v2di __builtin_ia32_aesenc128 (v2di, v2di)
9434 v2di __builtin_ia32_aesenclast128 (v2di, v2di)
9435 v2di __builtin_ia32_aesdec128 (v2di, v2di)
9436 v2di __builtin_ia32_aesdeclast128 (v2di, v2di)
9437 v2di __builtin_ia32_aeskeygenassist128 (v2di, const int)
9438 v2di __builtin_ia32_aesimc128 (v2di)
9441 The following built-in function is available when @option{-mpclmul} is
9445 @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int)
9446 Generates the @code{pclmulqdq} machine instruction.
9449 The following built-in function is available when @option{-mfsgsbase} is
9450 used. All of them generate the machine instruction that is part of the
9454 unsigned int __builtin_ia32_rdfsbase32 (void)
9455 unsigned long long __builtin_ia32_rdfsbase64 (void)
9456 unsigned int __builtin_ia32_rdgsbase32 (void)
9457 unsigned long long __builtin_ia32_rdgsbase64 (void)
9458 void _writefsbase_u32 (unsigned int)
9459 void _writefsbase_u64 (unsigned long long)
9460 void _writegsbase_u32 (unsigned int)
9461 void _writegsbase_u64 (unsigned long long)
9464 The following built-in function is available when @option{-mrdrnd} is
9465 used. All of them generate the machine instruction that is part of the
9469 unsigned int __builtin_ia32_rdrand16_step (unsigned short *)
9470 unsigned int __builtin_ia32_rdrand32_step (unsigned int *)
9471 unsigned int __builtin_ia32_rdrand64_step (unsigned long long *)
9474 The following built-in functions are available when @option{-msse4a} is used.
9475 All of them generate the machine instruction that is part of the name.
9478 void __builtin_ia32_movntsd (double *, v2df)
9479 void __builtin_ia32_movntss (float *, v4sf)
9480 v2di __builtin_ia32_extrq (v2di, v16qi)
9481 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
9482 v2di __builtin_ia32_insertq (v2di, v2di)
9483 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
9486 The following built-in functions are available when @option{-mxop} is used.
9488 v2df __builtin_ia32_vfrczpd (v2df)
9489 v4sf __builtin_ia32_vfrczps (v4sf)
9490 v2df __builtin_ia32_vfrczsd (v2df, v2df)
9491 v4sf __builtin_ia32_vfrczss (v4sf, v4sf)
9492 v4df __builtin_ia32_vfrczpd256 (v4df)
9493 v8sf __builtin_ia32_vfrczps256 (v8sf)
9494 v2di __builtin_ia32_vpcmov (v2di, v2di, v2di)
9495 v2di __builtin_ia32_vpcmov_v2di (v2di, v2di, v2di)
9496 v4si __builtin_ia32_vpcmov_v4si (v4si, v4si, v4si)
9497 v8hi __builtin_ia32_vpcmov_v8hi (v8hi, v8hi, v8hi)
9498 v16qi __builtin_ia32_vpcmov_v16qi (v16qi, v16qi, v16qi)
9499 v2df __builtin_ia32_vpcmov_v2df (v2df, v2df, v2df)
9500 v4sf __builtin_ia32_vpcmov_v4sf (v4sf, v4sf, v4sf)
9501 v4di __builtin_ia32_vpcmov_v4di256 (v4di, v4di, v4di)
9502 v8si __builtin_ia32_vpcmov_v8si256 (v8si, v8si, v8si)
9503 v16hi __builtin_ia32_vpcmov_v16hi256 (v16hi, v16hi, v16hi)
9504 v32qi __builtin_ia32_vpcmov_v32qi256 (v32qi, v32qi, v32qi)
9505 v4df __builtin_ia32_vpcmov_v4df256 (v4df, v4df, v4df)
9506 v8sf __builtin_ia32_vpcmov_v8sf256 (v8sf, v8sf, v8sf)
9507 v16qi __builtin_ia32_vpcomeqb (v16qi, v16qi)
9508 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
9509 v4si __builtin_ia32_vpcomeqd (v4si, v4si)
9510 v2di __builtin_ia32_vpcomeqq (v2di, v2di)
9511 v16qi __builtin_ia32_vpcomequb (v16qi, v16qi)
9512 v4si __builtin_ia32_vpcomequd (v4si, v4si)
9513 v2di __builtin_ia32_vpcomequq (v2di, v2di)
9514 v8hi __builtin_ia32_vpcomequw (v8hi, v8hi)
9515 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
9516 v16qi __builtin_ia32_vpcomfalseb (v16qi, v16qi)
9517 v4si __builtin_ia32_vpcomfalsed (v4si, v4si)
9518 v2di __builtin_ia32_vpcomfalseq (v2di, v2di)
9519 v16qi __builtin_ia32_vpcomfalseub (v16qi, v16qi)
9520 v4si __builtin_ia32_vpcomfalseud (v4si, v4si)
9521 v2di __builtin_ia32_vpcomfalseuq (v2di, v2di)
9522 v8hi __builtin_ia32_vpcomfalseuw (v8hi, v8hi)
9523 v8hi __builtin_ia32_vpcomfalsew (v8hi, v8hi)
9524 v16qi __builtin_ia32_vpcomgeb (v16qi, v16qi)
9525 v4si __builtin_ia32_vpcomged (v4si, v4si)
9526 v2di __builtin_ia32_vpcomgeq (v2di, v2di)
9527 v16qi __builtin_ia32_vpcomgeub (v16qi, v16qi)
9528 v4si __builtin_ia32_vpcomgeud (v4si, v4si)
9529 v2di __builtin_ia32_vpcomgeuq (v2di, v2di)
9530 v8hi __builtin_ia32_vpcomgeuw (v8hi, v8hi)
9531 v8hi __builtin_ia32_vpcomgew (v8hi, v8hi)
9532 v16qi __builtin_ia32_vpcomgtb (v16qi, v16qi)
9533 v4si __builtin_ia32_vpcomgtd (v4si, v4si)
9534 v2di __builtin_ia32_vpcomgtq (v2di, v2di)
9535 v16qi __builtin_ia32_vpcomgtub (v16qi, v16qi)
9536 v4si __builtin_ia32_vpcomgtud (v4si, v4si)
9537 v2di __builtin_ia32_vpcomgtuq (v2di, v2di)
9538 v8hi __builtin_ia32_vpcomgtuw (v8hi, v8hi)
9539 v8hi __builtin_ia32_vpcomgtw (v8hi, v8hi)
9540 v16qi __builtin_ia32_vpcomleb (v16qi, v16qi)
9541 v4si __builtin_ia32_vpcomled (v4si, v4si)
9542 v2di __builtin_ia32_vpcomleq (v2di, v2di)
9543 v16qi __builtin_ia32_vpcomleub (v16qi, v16qi)
9544 v4si __builtin_ia32_vpcomleud (v4si, v4si)
9545 v2di __builtin_ia32_vpcomleuq (v2di, v2di)
9546 v8hi __builtin_ia32_vpcomleuw (v8hi, v8hi)
9547 v8hi __builtin_ia32_vpcomlew (v8hi, v8hi)
9548 v16qi __builtin_ia32_vpcomltb (v16qi, v16qi)
9549 v4si __builtin_ia32_vpcomltd (v4si, v4si)
9550 v2di __builtin_ia32_vpcomltq (v2di, v2di)
9551 v16qi __builtin_ia32_vpcomltub (v16qi, v16qi)
9552 v4si __builtin_ia32_vpcomltud (v4si, v4si)
9553 v2di __builtin_ia32_vpcomltuq (v2di, v2di)
9554 v8hi __builtin_ia32_vpcomltuw (v8hi, v8hi)
9555 v8hi __builtin_ia32_vpcomltw (v8hi, v8hi)
9556 v16qi __builtin_ia32_vpcomneb (v16qi, v16qi)
9557 v4si __builtin_ia32_vpcomned (v4si, v4si)
9558 v2di __builtin_ia32_vpcomneq (v2di, v2di)
9559 v16qi __builtin_ia32_vpcomneub (v16qi, v16qi)
9560 v4si __builtin_ia32_vpcomneud (v4si, v4si)
9561 v2di __builtin_ia32_vpcomneuq (v2di, v2di)
9562 v8hi __builtin_ia32_vpcomneuw (v8hi, v8hi)
9563 v8hi __builtin_ia32_vpcomnew (v8hi, v8hi)
9564 v16qi __builtin_ia32_vpcomtrueb (v16qi, v16qi)
9565 v4si __builtin_ia32_vpcomtrued (v4si, v4si)
9566 v2di __builtin_ia32_vpcomtrueq (v2di, v2di)
9567 v16qi __builtin_ia32_vpcomtrueub (v16qi, v16qi)
9568 v4si __builtin_ia32_vpcomtrueud (v4si, v4si)
9569 v2di __builtin_ia32_vpcomtrueuq (v2di, v2di)
9570 v8hi __builtin_ia32_vpcomtrueuw (v8hi, v8hi)
9571 v8hi __builtin_ia32_vpcomtruew (v8hi, v8hi)
9572 v4si __builtin_ia32_vphaddbd (v16qi)
9573 v2di __builtin_ia32_vphaddbq (v16qi)
9574 v8hi __builtin_ia32_vphaddbw (v16qi)
9575 v2di __builtin_ia32_vphadddq (v4si)
9576 v4si __builtin_ia32_vphaddubd (v16qi)
9577 v2di __builtin_ia32_vphaddubq (v16qi)
9578 v8hi __builtin_ia32_vphaddubw (v16qi)
9579 v2di __builtin_ia32_vphaddudq (v4si)
9580 v4si __builtin_ia32_vphadduwd (v8hi)
9581 v2di __builtin_ia32_vphadduwq (v8hi)
9582 v4si __builtin_ia32_vphaddwd (v8hi)
9583 v2di __builtin_ia32_vphaddwq (v8hi)
9584 v8hi __builtin_ia32_vphsubbw (v16qi)
9585 v2di __builtin_ia32_vphsubdq (v4si)
9586 v4si __builtin_ia32_vphsubwd (v8hi)
9587 v4si __builtin_ia32_vpmacsdd (v4si, v4si, v4si)
9588 v2di __builtin_ia32_vpmacsdqh (v4si, v4si, v2di)
9589 v2di __builtin_ia32_vpmacsdql (v4si, v4si, v2di)
9590 v4si __builtin_ia32_vpmacssdd (v4si, v4si, v4si)
9591 v2di __builtin_ia32_vpmacssdqh (v4si, v4si, v2di)
9592 v2di __builtin_ia32_vpmacssdql (v4si, v4si, v2di)
9593 v4si __builtin_ia32_vpmacsswd (v8hi, v8hi, v4si)
9594 v8hi __builtin_ia32_vpmacssww (v8hi, v8hi, v8hi)
9595 v4si __builtin_ia32_vpmacswd (v8hi, v8hi, v4si)
9596 v8hi __builtin_ia32_vpmacsww (v8hi, v8hi, v8hi)
9597 v4si __builtin_ia32_vpmadcsswd (v8hi, v8hi, v4si)
9598 v4si __builtin_ia32_vpmadcswd (v8hi, v8hi, v4si)
9599 v16qi __builtin_ia32_vpperm (v16qi, v16qi, v16qi)
9600 v16qi __builtin_ia32_vprotb (v16qi, v16qi)
9601 v4si __builtin_ia32_vprotd (v4si, v4si)
9602 v2di __builtin_ia32_vprotq (v2di, v2di)
9603 v8hi __builtin_ia32_vprotw (v8hi, v8hi)
9604 v16qi __builtin_ia32_vpshab (v16qi, v16qi)
9605 v4si __builtin_ia32_vpshad (v4si, v4si)
9606 v2di __builtin_ia32_vpshaq (v2di, v2di)
9607 v8hi __builtin_ia32_vpshaw (v8hi, v8hi)
9608 v16qi __builtin_ia32_vpshlb (v16qi, v16qi)
9609 v4si __builtin_ia32_vpshld (v4si, v4si)
9610 v2di __builtin_ia32_vpshlq (v2di, v2di)
9611 v8hi __builtin_ia32_vpshlw (v8hi, v8hi)
9614 The following built-in functions are available when @option{-mfma4} is used.
9615 All of them generate the machine instruction that is part of the name
9619 v2df __builtin_ia32_fmaddpd (v2df, v2df, v2df)
9620 v4sf __builtin_ia32_fmaddps (v4sf, v4sf, v4sf)
9621 v2df __builtin_ia32_fmaddsd (v2df, v2df, v2df)
9622 v4sf __builtin_ia32_fmaddss (v4sf, v4sf, v4sf)
9623 v2df __builtin_ia32_fmsubpd (v2df, v2df, v2df)
9624 v4sf __builtin_ia32_fmsubps (v4sf, v4sf, v4sf)
9625 v2df __builtin_ia32_fmsubsd (v2df, v2df, v2df)
9626 v4sf __builtin_ia32_fmsubss (v4sf, v4sf, v4sf)
9627 v2df __builtin_ia32_fnmaddpd (v2df, v2df, v2df)
9628 v4sf __builtin_ia32_fnmaddps (v4sf, v4sf, v4sf)
9629 v2df __builtin_ia32_fnmaddsd (v2df, v2df, v2df)
9630 v4sf __builtin_ia32_fnmaddss (v4sf, v4sf, v4sf)
9631 v2df __builtin_ia32_fnmsubpd (v2df, v2df, v2df)
9632 v4sf __builtin_ia32_fnmsubps (v4sf, v4sf, v4sf)
9633 v2df __builtin_ia32_fnmsubsd (v2df, v2df, v2df)
9634 v4sf __builtin_ia32_fnmsubss (v4sf, v4sf, v4sf)
9635 v2df __builtin_ia32_fmaddsubpd (v2df, v2df, v2df)
9636 v4sf __builtin_ia32_fmaddsubps (v4sf, v4sf, v4sf)
9637 v2df __builtin_ia32_fmsubaddpd (v2df, v2df, v2df)
9638 v4sf __builtin_ia32_fmsubaddps (v4sf, v4sf, v4sf)
9639 v4df __builtin_ia32_fmaddpd256 (v4df, v4df, v4df)
9640 v8sf __builtin_ia32_fmaddps256 (v8sf, v8sf, v8sf)
9641 v4df __builtin_ia32_fmsubpd256 (v4df, v4df, v4df)
9642 v8sf __builtin_ia32_fmsubps256 (v8sf, v8sf, v8sf)
9643 v4df __builtin_ia32_fnmaddpd256 (v4df, v4df, v4df)
9644 v8sf __builtin_ia32_fnmaddps256 (v8sf, v8sf, v8sf)
9645 v4df __builtin_ia32_fnmsubpd256 (v4df, v4df, v4df)
9646 v8sf __builtin_ia32_fnmsubps256 (v8sf, v8sf, v8sf)
9647 v4df __builtin_ia32_fmaddsubpd256 (v4df, v4df, v4df)
9648 v8sf __builtin_ia32_fmaddsubps256 (v8sf, v8sf, v8sf)
9649 v4df __builtin_ia32_fmsubaddpd256 (v4df, v4df, v4df)
9650 v8sf __builtin_ia32_fmsubaddps256 (v8sf, v8sf, v8sf)
9654 The following built-in functions are available when @option{-mlwp} is used.
9657 void __builtin_ia32_llwpcb16 (void *);
9658 void __builtin_ia32_llwpcb32 (void *);
9659 void __builtin_ia32_llwpcb64 (void *);
9660 void * __builtin_ia32_llwpcb16 (void);
9661 void * __builtin_ia32_llwpcb32 (void);
9662 void * __builtin_ia32_llwpcb64 (void);
9663 void __builtin_ia32_lwpval16 (unsigned short, unsigned int, unsigned short)
9664 void __builtin_ia32_lwpval32 (unsigned int, unsigned int, unsigned int)
9665 void __builtin_ia32_lwpval64 (unsigned __int64, unsigned int, unsigned int)
9666 unsigned char __builtin_ia32_lwpins16 (unsigned short, unsigned int, unsigned short)
9667 unsigned char __builtin_ia32_lwpins32 (unsigned int, unsigned int, unsigned int)
9668 unsigned char __builtin_ia32_lwpins64 (unsigned __int64, unsigned int, unsigned int)
9671 The following built-in functions are available when @option{-mbmi} is used.
9672 All of them generate the machine instruction that is part of the name.
9674 unsigned int __builtin_ia32_bextr_u32(unsigned int, unsigned int);
9675 unsigned long long __builtin_ia32_bextr_u64 (unsigned long long, unsigned long long);
9676 unsigned short __builtin_ia32_lzcnt_16(unsigned short);
9677 unsigned int __builtin_ia32_lzcnt_u32(unsigned int);
9678 unsigned long long __builtin_ia32_lzcnt_u64 (unsigned long long);
9681 The following built-in functions are available when @option{-mtbm} is used.
9682 Both of them generate the immediate form of the bextr machine instruction.
9684 unsigned int __builtin_ia32_bextri_u32 (unsigned int, const unsigned int);
9685 unsigned long long __builtin_ia32_bextri_u64 (unsigned long long, const unsigned long long);
9689 The following built-in functions are available when @option{-m3dnow} is used.
9690 All of them generate the machine instruction that is part of the name.
9693 void __builtin_ia32_femms (void)
9694 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
9695 v2si __builtin_ia32_pf2id (v2sf)
9696 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
9697 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
9698 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
9699 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
9700 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
9701 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
9702 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
9703 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
9704 v2sf __builtin_ia32_pfrcp (v2sf)
9705 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
9706 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
9707 v2sf __builtin_ia32_pfrsqrt (v2sf)
9708 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
9709 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
9710 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
9711 v2sf __builtin_ia32_pi2fd (v2si)
9712 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
9715 The following built-in functions are available when both @option{-m3dnow}
9716 and @option{-march=athlon} are used. All of them generate the machine
9717 instruction that is part of the name.
9720 v2si __builtin_ia32_pf2iw (v2sf)
9721 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
9722 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
9723 v2sf __builtin_ia32_pi2fw (v2si)
9724 v2sf __builtin_ia32_pswapdsf (v2sf)
9725 v2si __builtin_ia32_pswapdsi (v2si)
9728 @node MIPS DSP Built-in Functions
9729 @subsection MIPS DSP Built-in Functions
9731 The MIPS DSP Application-Specific Extension (ASE) includes new
9732 instructions that are designed to improve the performance of DSP and
9733 media applications. It provides instructions that operate on packed
9734 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
9736 GCC supports MIPS DSP operations using both the generic
9737 vector extensions (@pxref{Vector Extensions}) and a collection of
9738 MIPS-specific built-in functions. Both kinds of support are
9739 enabled by the @option{-mdsp} command-line option.
9741 Revision 2 of the ASE was introduced in the second half of 2006.
9742 This revision adds extra instructions to the original ASE, but is
9743 otherwise backwards-compatible with it. You can select revision 2
9744 using the command-line option @option{-mdspr2}; this option implies
9747 The SCOUNT and POS bits of the DSP control register are global. The
9748 WRDSP, EXTPDP, EXTPDPV and MTHLIP instructions modify the SCOUNT and
9749 POS bits. During optimization, the compiler will not delete these
9750 instructions and it will not delete calls to functions containing
9753 At present, GCC only provides support for operations on 32-bit
9754 vectors. The vector type associated with 8-bit integer data is
9755 usually called @code{v4i8}, the vector type associated with Q7
9756 is usually called @code{v4q7}, the vector type associated with 16-bit
9757 integer data is usually called @code{v2i16}, and the vector type
9758 associated with Q15 is usually called @code{v2q15}. They can be
9759 defined in C as follows:
9762 typedef signed char v4i8 __attribute__ ((vector_size(4)));
9763 typedef signed char v4q7 __attribute__ ((vector_size(4)));
9764 typedef short v2i16 __attribute__ ((vector_size(4)));
9765 typedef short v2q15 __attribute__ ((vector_size(4)));
9768 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
9769 initialized in the same way as aggregates. For example:
9772 v4i8 a = @{1, 2, 3, 4@};
9774 b = (v4i8) @{5, 6, 7, 8@};
9776 v2q15 c = @{0x0fcb, 0x3a75@};
9778 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
9781 @emph{Note:} The CPU's endianness determines the order in which values
9782 are packed. On little-endian targets, the first value is the least
9783 significant and the last value is the most significant. The opposite
9784 order applies to big-endian targets. For example, the code above will
9785 set the lowest byte of @code{a} to @code{1} on little-endian targets
9786 and @code{4} on big-endian targets.
9788 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
9789 representation. As shown in this example, the integer representation
9790 of a Q7 value can be obtained by multiplying the fractional value by
9791 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
9792 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
9795 The table below lists the @code{v4i8} and @code{v2q15} operations for which
9796 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
9797 and @code{c} and @code{d} are @code{v2q15} values.
9799 @multitable @columnfractions .50 .50
9800 @item C code @tab MIPS instruction
9801 @item @code{a + b} @tab @code{addu.qb}
9802 @item @code{c + d} @tab @code{addq.ph}
9803 @item @code{a - b} @tab @code{subu.qb}
9804 @item @code{c - d} @tab @code{subq.ph}
9807 The table below lists the @code{v2i16} operation for which
9808 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
9809 @code{v2i16} values.
9811 @multitable @columnfractions .50 .50
9812 @item C code @tab MIPS instruction
9813 @item @code{e * f} @tab @code{mul.ph}
9816 It is easier to describe the DSP built-in functions if we first define
9817 the following types:
9822 typedef unsigned int ui32;
9823 typedef long long a64;
9826 @code{q31} and @code{i32} are actually the same as @code{int}, but we
9827 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
9828 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
9829 @code{long long}, but we use @code{a64} to indicate values that will
9830 be placed in one of the four DSP accumulators (@code{$ac0},
9831 @code{$ac1}, @code{$ac2} or @code{$ac3}).
9833 Also, some built-in functions prefer or require immediate numbers as
9834 parameters, because the corresponding DSP instructions accept both immediate
9835 numbers and register operands, or accept immediate numbers only. The
9836 immediate parameters are listed as follows.
9845 imm_n32_31: -32 to 31.
9846 imm_n512_511: -512 to 511.
9849 The following built-in functions map directly to a particular MIPS DSP
9850 instruction. Please refer to the architecture specification
9851 for details on what each instruction does.
9854 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
9855 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
9856 q31 __builtin_mips_addq_s_w (q31, q31)
9857 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
9858 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
9859 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
9860 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
9861 q31 __builtin_mips_subq_s_w (q31, q31)
9862 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
9863 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
9864 i32 __builtin_mips_addsc (i32, i32)
9865 i32 __builtin_mips_addwc (i32, i32)
9866 i32 __builtin_mips_modsub (i32, i32)
9867 i32 __builtin_mips_raddu_w_qb (v4i8)
9868 v2q15 __builtin_mips_absq_s_ph (v2q15)
9869 q31 __builtin_mips_absq_s_w (q31)
9870 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
9871 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
9872 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
9873 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
9874 q31 __builtin_mips_preceq_w_phl (v2q15)
9875 q31 __builtin_mips_preceq_w_phr (v2q15)
9876 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
9877 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
9878 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
9879 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
9880 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
9881 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
9882 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
9883 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
9884 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
9885 v4i8 __builtin_mips_shll_qb (v4i8, i32)
9886 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
9887 v2q15 __builtin_mips_shll_ph (v2q15, i32)
9888 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
9889 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
9890 q31 __builtin_mips_shll_s_w (q31, imm0_31)
9891 q31 __builtin_mips_shll_s_w (q31, i32)
9892 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
9893 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
9894 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
9895 v2q15 __builtin_mips_shra_ph (v2q15, i32)
9896 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
9897 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
9898 q31 __builtin_mips_shra_r_w (q31, imm0_31)
9899 q31 __builtin_mips_shra_r_w (q31, i32)
9900 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
9901 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
9902 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
9903 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
9904 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
9905 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
9906 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
9907 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
9908 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
9909 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
9910 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
9911 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
9912 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
9913 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
9914 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
9915 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
9916 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
9917 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
9918 i32 __builtin_mips_bitrev (i32)
9919 i32 __builtin_mips_insv (i32, i32)
9920 v4i8 __builtin_mips_repl_qb (imm0_255)
9921 v4i8 __builtin_mips_repl_qb (i32)
9922 v2q15 __builtin_mips_repl_ph (imm_n512_511)
9923 v2q15 __builtin_mips_repl_ph (i32)
9924 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
9925 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
9926 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
9927 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
9928 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
9929 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
9930 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
9931 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
9932 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
9933 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
9934 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
9935 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
9936 i32 __builtin_mips_extr_w (a64, imm0_31)
9937 i32 __builtin_mips_extr_w (a64, i32)
9938 i32 __builtin_mips_extr_r_w (a64, imm0_31)
9939 i32 __builtin_mips_extr_s_h (a64, i32)
9940 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
9941 i32 __builtin_mips_extr_rs_w (a64, i32)
9942 i32 __builtin_mips_extr_s_h (a64, imm0_31)
9943 i32 __builtin_mips_extr_r_w (a64, i32)
9944 i32 __builtin_mips_extp (a64, imm0_31)
9945 i32 __builtin_mips_extp (a64, i32)
9946 i32 __builtin_mips_extpdp (a64, imm0_31)
9947 i32 __builtin_mips_extpdp (a64, i32)
9948 a64 __builtin_mips_shilo (a64, imm_n32_31)
9949 a64 __builtin_mips_shilo (a64, i32)
9950 a64 __builtin_mips_mthlip (a64, i32)
9951 void __builtin_mips_wrdsp (i32, imm0_63)
9952 i32 __builtin_mips_rddsp (imm0_63)
9953 i32 __builtin_mips_lbux (void *, i32)
9954 i32 __builtin_mips_lhx (void *, i32)
9955 i32 __builtin_mips_lwx (void *, i32)
9956 i32 __builtin_mips_bposge32 (void)
9957 a64 __builtin_mips_madd (a64, i32, i32);
9958 a64 __builtin_mips_maddu (a64, ui32, ui32);
9959 a64 __builtin_mips_msub (a64, i32, i32);
9960 a64 __builtin_mips_msubu (a64, ui32, ui32);
9961 a64 __builtin_mips_mult (i32, i32);
9962 a64 __builtin_mips_multu (ui32, ui32);
9965 The following built-in functions map directly to a particular MIPS DSP REV 2
9966 instruction. Please refer to the architecture specification
9967 for details on what each instruction does.
9970 v4q7 __builtin_mips_absq_s_qb (v4q7);
9971 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
9972 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
9973 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
9974 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
9975 i32 __builtin_mips_append (i32, i32, imm0_31);
9976 i32 __builtin_mips_balign (i32, i32, imm0_3);
9977 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
9978 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
9979 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
9980 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
9981 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
9982 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
9983 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
9984 q31 __builtin_mips_mulq_rs_w (q31, q31);
9985 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
9986 q31 __builtin_mips_mulq_s_w (q31, q31);
9987 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
9988 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
9989 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
9990 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
9991 i32 __builtin_mips_prepend (i32, i32, imm0_31);
9992 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
9993 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
9994 v4i8 __builtin_mips_shra_qb (v4i8, i32);
9995 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
9996 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
9997 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
9998 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
9999 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
10000 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
10001 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
10002 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
10003 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
10004 q31 __builtin_mips_addqh_w (q31, q31);
10005 q31 __builtin_mips_addqh_r_w (q31, q31);
10006 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
10007 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
10008 q31 __builtin_mips_subqh_w (q31, q31);
10009 q31 __builtin_mips_subqh_r_w (q31, q31);
10010 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
10011 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
10012 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
10013 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
10014 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
10015 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
10019 @node MIPS Paired-Single Support
10020 @subsection MIPS Paired-Single Support
10022 The MIPS64 architecture includes a number of instructions that
10023 operate on pairs of single-precision floating-point values.
10024 Each pair is packed into a 64-bit floating-point register,
10025 with one element being designated the ``upper half'' and
10026 the other being designated the ``lower half''.
10028 GCC supports paired-single operations using both the generic
10029 vector extensions (@pxref{Vector Extensions}) and a collection of
10030 MIPS-specific built-in functions. Both kinds of support are
10031 enabled by the @option{-mpaired-single} command-line option.
10033 The vector type associated with paired-single values is usually
10034 called @code{v2sf}. It can be defined in C as follows:
10037 typedef float v2sf __attribute__ ((vector_size (8)));
10040 @code{v2sf} values are initialized in the same way as aggregates.
10044 v2sf a = @{1.5, 9.1@};
10047 b = (v2sf) @{e, f@};
10050 @emph{Note:} The CPU's endianness determines which value is stored in
10051 the upper half of a register and which value is stored in the lower half.
10052 On little-endian targets, the first value is the lower one and the second
10053 value is the upper one. The opposite order applies to big-endian targets.
10054 For example, the code above will set the lower half of @code{a} to
10055 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
10057 @node MIPS Loongson Built-in Functions
10058 @subsection MIPS Loongson Built-in Functions
10060 GCC provides intrinsics to access the SIMD instructions provided by the
10061 ST Microelectronics Loongson-2E and -2F processors. These intrinsics,
10062 available after inclusion of the @code{loongson.h} header file,
10063 operate on the following 64-bit vector types:
10066 @item @code{uint8x8_t}, a vector of eight unsigned 8-bit integers;
10067 @item @code{uint16x4_t}, a vector of four unsigned 16-bit integers;
10068 @item @code{uint32x2_t}, a vector of two unsigned 32-bit integers;
10069 @item @code{int8x8_t}, a vector of eight signed 8-bit integers;
10070 @item @code{int16x4_t}, a vector of four signed 16-bit integers;
10071 @item @code{int32x2_t}, a vector of two signed 32-bit integers.
10074 The intrinsics provided are listed below; each is named after the
10075 machine instruction to which it corresponds, with suffixes added as
10076 appropriate to distinguish intrinsics that expand to the same machine
10077 instruction yet have different argument types. Refer to the architecture
10078 documentation for a description of the functionality of each
10082 int16x4_t packsswh (int32x2_t s, int32x2_t t);
10083 int8x8_t packsshb (int16x4_t s, int16x4_t t);
10084 uint8x8_t packushb (uint16x4_t s, uint16x4_t t);
10085 uint32x2_t paddw_u (uint32x2_t s, uint32x2_t t);
10086 uint16x4_t paddh_u (uint16x4_t s, uint16x4_t t);
10087 uint8x8_t paddb_u (uint8x8_t s, uint8x8_t t);
10088 int32x2_t paddw_s (int32x2_t s, int32x2_t t);
10089 int16x4_t paddh_s (int16x4_t s, int16x4_t t);
10090 int8x8_t paddb_s (int8x8_t s, int8x8_t t);
10091 uint64_t paddd_u (uint64_t s, uint64_t t);
10092 int64_t paddd_s (int64_t s, int64_t t);
10093 int16x4_t paddsh (int16x4_t s, int16x4_t t);
10094 int8x8_t paddsb (int8x8_t s, int8x8_t t);
10095 uint16x4_t paddush (uint16x4_t s, uint16x4_t t);
10096 uint8x8_t paddusb (uint8x8_t s, uint8x8_t t);
10097 uint64_t pandn_ud (uint64_t s, uint64_t t);
10098 uint32x2_t pandn_uw (uint32x2_t s, uint32x2_t t);
10099 uint16x4_t pandn_uh (uint16x4_t s, uint16x4_t t);
10100 uint8x8_t pandn_ub (uint8x8_t s, uint8x8_t t);
10101 int64_t pandn_sd (int64_t s, int64_t t);
10102 int32x2_t pandn_sw (int32x2_t s, int32x2_t t);
10103 int16x4_t pandn_sh (int16x4_t s, int16x4_t t);
10104 int8x8_t pandn_sb (int8x8_t s, int8x8_t t);
10105 uint16x4_t pavgh (uint16x4_t s, uint16x4_t t);
10106 uint8x8_t pavgb (uint8x8_t s, uint8x8_t t);
10107 uint32x2_t pcmpeqw_u (uint32x2_t s, uint32x2_t t);
10108 uint16x4_t pcmpeqh_u (uint16x4_t s, uint16x4_t t);
10109 uint8x8_t pcmpeqb_u (uint8x8_t s, uint8x8_t t);
10110 int32x2_t pcmpeqw_s (int32x2_t s, int32x2_t t);
10111 int16x4_t pcmpeqh_s (int16x4_t s, int16x4_t t);
10112 int8x8_t pcmpeqb_s (int8x8_t s, int8x8_t t);
10113 uint32x2_t pcmpgtw_u (uint32x2_t s, uint32x2_t t);
10114 uint16x4_t pcmpgth_u (uint16x4_t s, uint16x4_t t);
10115 uint8x8_t pcmpgtb_u (uint8x8_t s, uint8x8_t t);
10116 int32x2_t pcmpgtw_s (int32x2_t s, int32x2_t t);
10117 int16x4_t pcmpgth_s (int16x4_t s, int16x4_t t);
10118 int8x8_t pcmpgtb_s (int8x8_t s, int8x8_t t);
10119 uint16x4_t pextrh_u (uint16x4_t s, int field);
10120 int16x4_t pextrh_s (int16x4_t s, int field);
10121 uint16x4_t pinsrh_0_u (uint16x4_t s, uint16x4_t t);
10122 uint16x4_t pinsrh_1_u (uint16x4_t s, uint16x4_t t);
10123 uint16x4_t pinsrh_2_u (uint16x4_t s, uint16x4_t t);
10124 uint16x4_t pinsrh_3_u (uint16x4_t s, uint16x4_t t);
10125 int16x4_t pinsrh_0_s (int16x4_t s, int16x4_t t);
10126 int16x4_t pinsrh_1_s (int16x4_t s, int16x4_t t);
10127 int16x4_t pinsrh_2_s (int16x4_t s, int16x4_t t);
10128 int16x4_t pinsrh_3_s (int16x4_t s, int16x4_t t);
10129 int32x2_t pmaddhw (int16x4_t s, int16x4_t t);
10130 int16x4_t pmaxsh (int16x4_t s, int16x4_t t);
10131 uint8x8_t pmaxub (uint8x8_t s, uint8x8_t t);
10132 int16x4_t pminsh (int16x4_t s, int16x4_t t);
10133 uint8x8_t pminub (uint8x8_t s, uint8x8_t t);
10134 uint8x8_t pmovmskb_u (uint8x8_t s);
10135 int8x8_t pmovmskb_s (int8x8_t s);
10136 uint16x4_t pmulhuh (uint16x4_t s, uint16x4_t t);
10137 int16x4_t pmulhh (int16x4_t s, int16x4_t t);
10138 int16x4_t pmullh (int16x4_t s, int16x4_t t);
10139 int64_t pmuluw (uint32x2_t s, uint32x2_t t);
10140 uint8x8_t pasubub (uint8x8_t s, uint8x8_t t);
10141 uint16x4_t biadd (uint8x8_t s);
10142 uint16x4_t psadbh (uint8x8_t s, uint8x8_t t);
10143 uint16x4_t pshufh_u (uint16x4_t dest, uint16x4_t s, uint8_t order);
10144 int16x4_t pshufh_s (int16x4_t dest, int16x4_t s, uint8_t order);
10145 uint16x4_t psllh_u (uint16x4_t s, uint8_t amount);
10146 int16x4_t psllh_s (int16x4_t s, uint8_t amount);
10147 uint32x2_t psllw_u (uint32x2_t s, uint8_t amount);
10148 int32x2_t psllw_s (int32x2_t s, uint8_t amount);
10149 uint16x4_t psrlh_u (uint16x4_t s, uint8_t amount);
10150 int16x4_t psrlh_s (int16x4_t s, uint8_t amount);
10151 uint32x2_t psrlw_u (uint32x2_t s, uint8_t amount);
10152 int32x2_t psrlw_s (int32x2_t s, uint8_t amount);
10153 uint16x4_t psrah_u (uint16x4_t s, uint8_t amount);
10154 int16x4_t psrah_s (int16x4_t s, uint8_t amount);
10155 uint32x2_t psraw_u (uint32x2_t s, uint8_t amount);
10156 int32x2_t psraw_s (int32x2_t s, uint8_t amount);
10157 uint32x2_t psubw_u (uint32x2_t s, uint32x2_t t);
10158 uint16x4_t psubh_u (uint16x4_t s, uint16x4_t t);
10159 uint8x8_t psubb_u (uint8x8_t s, uint8x8_t t);
10160 int32x2_t psubw_s (int32x2_t s, int32x2_t t);
10161 int16x4_t psubh_s (int16x4_t s, int16x4_t t);
10162 int8x8_t psubb_s (int8x8_t s, int8x8_t t);
10163 uint64_t psubd_u (uint64_t s, uint64_t t);
10164 int64_t psubd_s (int64_t s, int64_t t);
10165 int16x4_t psubsh (int16x4_t s, int16x4_t t);
10166 int8x8_t psubsb (int8x8_t s, int8x8_t t);
10167 uint16x4_t psubush (uint16x4_t s, uint16x4_t t);
10168 uint8x8_t psubusb (uint8x8_t s, uint8x8_t t);
10169 uint32x2_t punpckhwd_u (uint32x2_t s, uint32x2_t t);
10170 uint16x4_t punpckhhw_u (uint16x4_t s, uint16x4_t t);
10171 uint8x8_t punpckhbh_u (uint8x8_t s, uint8x8_t t);
10172 int32x2_t punpckhwd_s (int32x2_t s, int32x2_t t);
10173 int16x4_t punpckhhw_s (int16x4_t s, int16x4_t t);
10174 int8x8_t punpckhbh_s (int8x8_t s, int8x8_t t);
10175 uint32x2_t punpcklwd_u (uint32x2_t s, uint32x2_t t);
10176 uint16x4_t punpcklhw_u (uint16x4_t s, uint16x4_t t);
10177 uint8x8_t punpcklbh_u (uint8x8_t s, uint8x8_t t);
10178 int32x2_t punpcklwd_s (int32x2_t s, int32x2_t t);
10179 int16x4_t punpcklhw_s (int16x4_t s, int16x4_t t);
10180 int8x8_t punpcklbh_s (int8x8_t s, int8x8_t t);
10184 * Paired-Single Arithmetic::
10185 * Paired-Single Built-in Functions::
10186 * MIPS-3D Built-in Functions::
10189 @node Paired-Single Arithmetic
10190 @subsubsection Paired-Single Arithmetic
10192 The table below lists the @code{v2sf} operations for which hardware
10193 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
10194 values and @code{x} is an integral value.
10196 @multitable @columnfractions .50 .50
10197 @item C code @tab MIPS instruction
10198 @item @code{a + b} @tab @code{add.ps}
10199 @item @code{a - b} @tab @code{sub.ps}
10200 @item @code{-a} @tab @code{neg.ps}
10201 @item @code{a * b} @tab @code{mul.ps}
10202 @item @code{a * b + c} @tab @code{madd.ps}
10203 @item @code{a * b - c} @tab @code{msub.ps}
10204 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
10205 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
10206 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
10209 Note that the multiply-accumulate instructions can be disabled
10210 using the command-line option @code{-mno-fused-madd}.
10212 @node Paired-Single Built-in Functions
10213 @subsubsection Paired-Single Built-in Functions
10215 The following paired-single functions map directly to a particular
10216 MIPS instruction. Please refer to the architecture specification
10217 for details on what each instruction does.
10220 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
10221 Pair lower lower (@code{pll.ps}).
10223 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
10224 Pair upper lower (@code{pul.ps}).
10226 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
10227 Pair lower upper (@code{plu.ps}).
10229 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
10230 Pair upper upper (@code{puu.ps}).
10232 @item v2sf __builtin_mips_cvt_ps_s (float, float)
10233 Convert pair to paired single (@code{cvt.ps.s}).
10235 @item float __builtin_mips_cvt_s_pl (v2sf)
10236 Convert pair lower to single (@code{cvt.s.pl}).
10238 @item float __builtin_mips_cvt_s_pu (v2sf)
10239 Convert pair upper to single (@code{cvt.s.pu}).
10241 @item v2sf __builtin_mips_abs_ps (v2sf)
10242 Absolute value (@code{abs.ps}).
10244 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
10245 Align variable (@code{alnv.ps}).
10247 @emph{Note:} The value of the third parameter must be 0 or 4
10248 modulo 8, otherwise the result will be unpredictable. Please read the
10249 instruction description for details.
10252 The following multi-instruction functions are also available.
10253 In each case, @var{cond} can be any of the 16 floating-point conditions:
10254 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
10255 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
10256 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
10259 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10260 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10261 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
10262 @code{movt.ps}/@code{movf.ps}).
10264 The @code{movt} functions return the value @var{x} computed by:
10267 c.@var{cond}.ps @var{cc},@var{a},@var{b}
10268 mov.ps @var{x},@var{c}
10269 movt.ps @var{x},@var{d},@var{cc}
10272 The @code{movf} functions are similar but use @code{movf.ps} instead
10275 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10276 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10277 Comparison of two paired-single values (@code{c.@var{cond}.ps},
10278 @code{bc1t}/@code{bc1f}).
10280 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
10281 and return either the upper or lower half of the result. For example:
10285 if (__builtin_mips_upper_c_eq_ps (a, b))
10286 upper_halves_are_equal ();
10288 upper_halves_are_unequal ();
10290 if (__builtin_mips_lower_c_eq_ps (a, b))
10291 lower_halves_are_equal ();
10293 lower_halves_are_unequal ();
10297 @node MIPS-3D Built-in Functions
10298 @subsubsection MIPS-3D Built-in Functions
10300 The MIPS-3D Application-Specific Extension (ASE) includes additional
10301 paired-single instructions that are designed to improve the performance
10302 of 3D graphics operations. Support for these instructions is controlled
10303 by the @option{-mips3d} command-line option.
10305 The functions listed below map directly to a particular MIPS-3D
10306 instruction. Please refer to the architecture specification for
10307 more details on what each instruction does.
10310 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
10311 Reduction add (@code{addr.ps}).
10313 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
10314 Reduction multiply (@code{mulr.ps}).
10316 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
10317 Convert paired single to paired word (@code{cvt.pw.ps}).
10319 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
10320 Convert paired word to paired single (@code{cvt.ps.pw}).
10322 @item float __builtin_mips_recip1_s (float)
10323 @itemx double __builtin_mips_recip1_d (double)
10324 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
10325 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
10327 @item float __builtin_mips_recip2_s (float, float)
10328 @itemx double __builtin_mips_recip2_d (double, double)
10329 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
10330 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
10332 @item float __builtin_mips_rsqrt1_s (float)
10333 @itemx double __builtin_mips_rsqrt1_d (double)
10334 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
10335 Reduced precision reciprocal square root (sequence step 1)
10336 (@code{rsqrt1.@var{fmt}}).
10338 @item float __builtin_mips_rsqrt2_s (float, float)
10339 @itemx double __builtin_mips_rsqrt2_d (double, double)
10340 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
10341 Reduced precision reciprocal square root (sequence step 2)
10342 (@code{rsqrt2.@var{fmt}}).
10345 The following multi-instruction functions are also available.
10346 In each case, @var{cond} can be any of the 16 floating-point conditions:
10347 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
10348 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
10349 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
10352 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
10353 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
10354 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
10355 @code{bc1t}/@code{bc1f}).
10357 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
10358 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
10363 if (__builtin_mips_cabs_eq_s (a, b))
10369 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10370 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10371 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
10372 @code{bc1t}/@code{bc1f}).
10374 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
10375 and return either the upper or lower half of the result. For example:
10379 if (__builtin_mips_upper_cabs_eq_ps (a, b))
10380 upper_halves_are_equal ();
10382 upper_halves_are_unequal ();
10384 if (__builtin_mips_lower_cabs_eq_ps (a, b))
10385 lower_halves_are_equal ();
10387 lower_halves_are_unequal ();
10390 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10391 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10392 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
10393 @code{movt.ps}/@code{movf.ps}).
10395 The @code{movt} functions return the value @var{x} computed by:
10398 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
10399 mov.ps @var{x},@var{c}
10400 movt.ps @var{x},@var{d},@var{cc}
10403 The @code{movf} functions are similar but use @code{movf.ps} instead
10406 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10407 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10408 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10409 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
10410 Comparison of two paired-single values
10411 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
10412 @code{bc1any2t}/@code{bc1any2f}).
10414 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
10415 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
10416 result is true and the @code{all} forms return true if both results are true.
10421 if (__builtin_mips_any_c_eq_ps (a, b))
10426 if (__builtin_mips_all_c_eq_ps (a, b))
10432 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10433 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10434 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10435 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
10436 Comparison of four paired-single values
10437 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
10438 @code{bc1any4t}/@code{bc1any4f}).
10440 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
10441 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
10442 The @code{any} forms return true if any of the four results are true
10443 and the @code{all} forms return true if all four results are true.
10448 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
10453 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
10460 @node picoChip Built-in Functions
10461 @subsection picoChip Built-in Functions
10463 GCC provides an interface to selected machine instructions from the
10464 picoChip instruction set.
10467 @item int __builtin_sbc (int @var{value})
10468 Sign bit count. Return the number of consecutive bits in @var{value}
10469 which have the same value as the sign-bit. The result is the number of
10470 leading sign bits minus one, giving the number of redundant sign bits in
10473 @item int __builtin_byteswap (int @var{value})
10474 Byte swap. Return the result of swapping the upper and lower bytes of
10477 @item int __builtin_brev (int @var{value})
10478 Bit reversal. Return the result of reversing the bits in
10479 @var{value}. Bit 15 is swapped with bit 0, bit 14 is swapped with bit 1,
10482 @item int __builtin_adds (int @var{x}, int @var{y})
10483 Saturating addition. Return the result of adding @var{x} and @var{y},
10484 storing the value 32767 if the result overflows.
10486 @item int __builtin_subs (int @var{x}, int @var{y})
10487 Saturating subtraction. Return the result of subtracting @var{y} from
10488 @var{x}, storing the value @minus{}32768 if the result overflows.
10490 @item void __builtin_halt (void)
10491 Halt. The processor will stop execution. This built-in is useful for
10492 implementing assertions.
10496 @node Other MIPS Built-in Functions
10497 @subsection Other MIPS Built-in Functions
10499 GCC provides other MIPS-specific built-in functions:
10502 @item void __builtin_mips_cache (int @var{op}, const volatile void *@var{addr})
10503 Insert a @samp{cache} instruction with operands @var{op} and @var{addr}.
10504 GCC defines the preprocessor macro @code{___GCC_HAVE_BUILTIN_MIPS_CACHE}
10505 when this function is available.
10508 @node PowerPC AltiVec/VSX Built-in Functions
10509 @subsection PowerPC AltiVec Built-in Functions
10511 GCC provides an interface for the PowerPC family of processors to access
10512 the AltiVec operations described in Motorola's AltiVec Programming
10513 Interface Manual. The interface is made available by including
10514 @code{<altivec.h>} and using @option{-maltivec} and
10515 @option{-mabi=altivec}. The interface supports the following vector
10519 vector unsigned char
10523 vector unsigned short
10524 vector signed short
10528 vector unsigned int
10534 If @option{-mvsx} is used the following additional vector types are
10538 vector unsigned long
10543 The long types are only implemented for 64-bit code generation, and
10544 the long type is only used in the floating point/integer conversion
10547 GCC's implementation of the high-level language interface available from
10548 C and C++ code differs from Motorola's documentation in several ways.
10553 A vector constant is a list of constant expressions within curly braces.
10556 A vector initializer requires no cast if the vector constant is of the
10557 same type as the variable it is initializing.
10560 If @code{signed} or @code{unsigned} is omitted, the signedness of the
10561 vector type is the default signedness of the base type. The default
10562 varies depending on the operating system, so a portable program should
10563 always specify the signedness.
10566 Compiling with @option{-maltivec} adds keywords @code{__vector},
10567 @code{vector}, @code{__pixel}, @code{pixel}, @code{__bool} and
10568 @code{bool}. When compiling ISO C, the context-sensitive substitution
10569 of the keywords @code{vector}, @code{pixel} and @code{bool} is
10570 disabled. To use them, you must include @code{<altivec.h>} instead.
10573 GCC allows using a @code{typedef} name as the type specifier for a
10577 For C, overloaded functions are implemented with macros so the following
10581 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
10584 Since @code{vec_add} is a macro, the vector constant in the example
10585 is treated as four separate arguments. Wrap the entire argument in
10586 parentheses for this to work.
10589 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
10590 Internally, GCC uses built-in functions to achieve the functionality in
10591 the aforementioned header file, but they are not supported and are
10592 subject to change without notice.
10594 The following interfaces are supported for the generic and specific
10595 AltiVec operations and the AltiVec predicates. In cases where there
10596 is a direct mapping between generic and specific operations, only the
10597 generic names are shown here, although the specific operations can also
10600 Arguments that are documented as @code{const int} require literal
10601 integral values within the range required for that operation.
10604 vector signed char vec_abs (vector signed char);
10605 vector signed short vec_abs (vector signed short);
10606 vector signed int vec_abs (vector signed int);
10607 vector float vec_abs (vector float);
10609 vector signed char vec_abss (vector signed char);
10610 vector signed short vec_abss (vector signed short);
10611 vector signed int vec_abss (vector signed int);
10613 vector signed char vec_add (vector bool char, vector signed char);
10614 vector signed char vec_add (vector signed char, vector bool char);
10615 vector signed char vec_add (vector signed char, vector signed char);
10616 vector unsigned char vec_add (vector bool char, vector unsigned char);
10617 vector unsigned char vec_add (vector unsigned char, vector bool char);
10618 vector unsigned char vec_add (vector unsigned char,
10619 vector unsigned char);
10620 vector signed short vec_add (vector bool short, vector signed short);
10621 vector signed short vec_add (vector signed short, vector bool short);
10622 vector signed short vec_add (vector signed short, vector signed short);
10623 vector unsigned short vec_add (vector bool short,
10624 vector unsigned short);
10625 vector unsigned short vec_add (vector unsigned short,
10626 vector bool short);
10627 vector unsigned short vec_add (vector unsigned short,
10628 vector unsigned short);
10629 vector signed int vec_add (vector bool int, vector signed int);
10630 vector signed int vec_add (vector signed int, vector bool int);
10631 vector signed int vec_add (vector signed int, vector signed int);
10632 vector unsigned int vec_add (vector bool int, vector unsigned int);
10633 vector unsigned int vec_add (vector unsigned int, vector bool int);
10634 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
10635 vector float vec_add (vector float, vector float);
10637 vector float vec_vaddfp (vector float, vector float);
10639 vector signed int vec_vadduwm (vector bool int, vector signed int);
10640 vector signed int vec_vadduwm (vector signed int, vector bool int);
10641 vector signed int vec_vadduwm (vector signed int, vector signed int);
10642 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
10643 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
10644 vector unsigned int vec_vadduwm (vector unsigned int,
10645 vector unsigned int);
10647 vector signed short vec_vadduhm (vector bool short,
10648 vector signed short);
10649 vector signed short vec_vadduhm (vector signed short,
10650 vector bool short);
10651 vector signed short vec_vadduhm (vector signed short,
10652 vector signed short);
10653 vector unsigned short vec_vadduhm (vector bool short,
10654 vector unsigned short);
10655 vector unsigned short vec_vadduhm (vector unsigned short,
10656 vector bool short);
10657 vector unsigned short vec_vadduhm (vector unsigned short,
10658 vector unsigned short);
10660 vector signed char vec_vaddubm (vector bool char, vector signed char);
10661 vector signed char vec_vaddubm (vector signed char, vector bool char);
10662 vector signed char vec_vaddubm (vector signed char, vector signed char);
10663 vector unsigned char vec_vaddubm (vector bool char,
10664 vector unsigned char);
10665 vector unsigned char vec_vaddubm (vector unsigned char,
10667 vector unsigned char vec_vaddubm (vector unsigned char,
10668 vector unsigned char);
10670 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
10672 vector unsigned char vec_adds (vector bool char, vector unsigned char);
10673 vector unsigned char vec_adds (vector unsigned char, vector bool char);
10674 vector unsigned char vec_adds (vector unsigned char,
10675 vector unsigned char);
10676 vector signed char vec_adds (vector bool char, vector signed char);
10677 vector signed char vec_adds (vector signed char, vector bool char);
10678 vector signed char vec_adds (vector signed char, vector signed char);
10679 vector unsigned short vec_adds (vector bool short,
10680 vector unsigned short);
10681 vector unsigned short vec_adds (vector unsigned short,
10682 vector bool short);
10683 vector unsigned short vec_adds (vector unsigned short,
10684 vector unsigned short);
10685 vector signed short vec_adds (vector bool short, vector signed short);
10686 vector signed short vec_adds (vector signed short, vector bool short);
10687 vector signed short vec_adds (vector signed short, vector signed short);
10688 vector unsigned int vec_adds (vector bool int, vector unsigned int);
10689 vector unsigned int vec_adds (vector unsigned int, vector bool int);
10690 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
10691 vector signed int vec_adds (vector bool int, vector signed int);
10692 vector signed int vec_adds (vector signed int, vector bool int);
10693 vector signed int vec_adds (vector signed int, vector signed int);
10695 vector signed int vec_vaddsws (vector bool int, vector signed int);
10696 vector signed int vec_vaddsws (vector signed int, vector bool int);
10697 vector signed int vec_vaddsws (vector signed int, vector signed int);
10699 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
10700 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
10701 vector unsigned int vec_vadduws (vector unsigned int,
10702 vector unsigned int);
10704 vector signed short vec_vaddshs (vector bool short,
10705 vector signed short);
10706 vector signed short vec_vaddshs (vector signed short,
10707 vector bool short);
10708 vector signed short vec_vaddshs (vector signed short,
10709 vector signed short);
10711 vector unsigned short vec_vadduhs (vector bool short,
10712 vector unsigned short);
10713 vector unsigned short vec_vadduhs (vector unsigned short,
10714 vector bool short);
10715 vector unsigned short vec_vadduhs (vector unsigned short,
10716 vector unsigned short);
10718 vector signed char vec_vaddsbs (vector bool char, vector signed char);
10719 vector signed char vec_vaddsbs (vector signed char, vector bool char);
10720 vector signed char vec_vaddsbs (vector signed char, vector signed char);
10722 vector unsigned char vec_vaddubs (vector bool char,
10723 vector unsigned char);
10724 vector unsigned char vec_vaddubs (vector unsigned char,
10726 vector unsigned char vec_vaddubs (vector unsigned char,
10727 vector unsigned char);
10729 vector float vec_and (vector float, vector float);
10730 vector float vec_and (vector float, vector bool int);
10731 vector float vec_and (vector bool int, vector float);
10732 vector bool int vec_and (vector bool int, vector bool int);
10733 vector signed int vec_and (vector bool int, vector signed int);
10734 vector signed int vec_and (vector signed int, vector bool int);
10735 vector signed int vec_and (vector signed int, vector signed int);
10736 vector unsigned int vec_and (vector bool int, vector unsigned int);
10737 vector unsigned int vec_and (vector unsigned int, vector bool int);
10738 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
10739 vector bool short vec_and (vector bool short, vector bool short);
10740 vector signed short vec_and (vector bool short, vector signed short);
10741 vector signed short vec_and (vector signed short, vector bool short);
10742 vector signed short vec_and (vector signed short, vector signed short);
10743 vector unsigned short vec_and (vector bool short,
10744 vector unsigned short);
10745 vector unsigned short vec_and (vector unsigned short,
10746 vector bool short);
10747 vector unsigned short vec_and (vector unsigned short,
10748 vector unsigned short);
10749 vector signed char vec_and (vector bool char, vector signed char);
10750 vector bool char vec_and (vector bool char, vector bool char);
10751 vector signed char vec_and (vector signed char, vector bool char);
10752 vector signed char vec_and (vector signed char, vector signed char);
10753 vector unsigned char vec_and (vector bool char, vector unsigned char);
10754 vector unsigned char vec_and (vector unsigned char, vector bool char);
10755 vector unsigned char vec_and (vector unsigned char,
10756 vector unsigned char);
10758 vector float vec_andc (vector float, vector float);
10759 vector float vec_andc (vector float, vector bool int);
10760 vector float vec_andc (vector bool int, vector float);
10761 vector bool int vec_andc (vector bool int, vector bool int);
10762 vector signed int vec_andc (vector bool int, vector signed int);
10763 vector signed int vec_andc (vector signed int, vector bool int);
10764 vector signed int vec_andc (vector signed int, vector signed int);
10765 vector unsigned int vec_andc (vector bool int, vector unsigned int);
10766 vector unsigned int vec_andc (vector unsigned int, vector bool int);
10767 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
10768 vector bool short vec_andc (vector bool short, vector bool short);
10769 vector signed short vec_andc (vector bool short, vector signed short);
10770 vector signed short vec_andc (vector signed short, vector bool short);
10771 vector signed short vec_andc (vector signed short, vector signed short);
10772 vector unsigned short vec_andc (vector bool short,
10773 vector unsigned short);
10774 vector unsigned short vec_andc (vector unsigned short,
10775 vector bool short);
10776 vector unsigned short vec_andc (vector unsigned short,
10777 vector unsigned short);
10778 vector signed char vec_andc (vector bool char, vector signed char);
10779 vector bool char vec_andc (vector bool char, vector bool char);
10780 vector signed char vec_andc (vector signed char, vector bool char);
10781 vector signed char vec_andc (vector signed char, vector signed char);
10782 vector unsigned char vec_andc (vector bool char, vector unsigned char);
10783 vector unsigned char vec_andc (vector unsigned char, vector bool char);
10784 vector unsigned char vec_andc (vector unsigned char,
10785 vector unsigned char);
10787 vector unsigned char vec_avg (vector unsigned char,
10788 vector unsigned char);
10789 vector signed char vec_avg (vector signed char, vector signed char);
10790 vector unsigned short vec_avg (vector unsigned short,
10791 vector unsigned short);
10792 vector signed short vec_avg (vector signed short, vector signed short);
10793 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
10794 vector signed int vec_avg (vector signed int, vector signed int);
10796 vector signed int vec_vavgsw (vector signed int, vector signed int);
10798 vector unsigned int vec_vavguw (vector unsigned int,
10799 vector unsigned int);
10801 vector signed short vec_vavgsh (vector signed short,
10802 vector signed short);
10804 vector unsigned short vec_vavguh (vector unsigned short,
10805 vector unsigned short);
10807 vector signed char vec_vavgsb (vector signed char, vector signed char);
10809 vector unsigned char vec_vavgub (vector unsigned char,
10810 vector unsigned char);
10812 vector float vec_copysign (vector float);
10814 vector float vec_ceil (vector float);
10816 vector signed int vec_cmpb (vector float, vector float);
10818 vector bool char vec_cmpeq (vector signed char, vector signed char);
10819 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
10820 vector bool short vec_cmpeq (vector signed short, vector signed short);
10821 vector bool short vec_cmpeq (vector unsigned short,
10822 vector unsigned short);
10823 vector bool int vec_cmpeq (vector signed int, vector signed int);
10824 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
10825 vector bool int vec_cmpeq (vector float, vector float);
10827 vector bool int vec_vcmpeqfp (vector float, vector float);
10829 vector bool int vec_vcmpequw (vector signed int, vector signed int);
10830 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
10832 vector bool short vec_vcmpequh (vector signed short,
10833 vector signed short);
10834 vector bool short vec_vcmpequh (vector unsigned short,
10835 vector unsigned short);
10837 vector bool char vec_vcmpequb (vector signed char, vector signed char);
10838 vector bool char vec_vcmpequb (vector unsigned char,
10839 vector unsigned char);
10841 vector bool int vec_cmpge (vector float, vector float);
10843 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
10844 vector bool char vec_cmpgt (vector signed char, vector signed char);
10845 vector bool short vec_cmpgt (vector unsigned short,
10846 vector unsigned short);
10847 vector bool short vec_cmpgt (vector signed short, vector signed short);
10848 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
10849 vector bool int vec_cmpgt (vector signed int, vector signed int);
10850 vector bool int vec_cmpgt (vector float, vector float);
10852 vector bool int vec_vcmpgtfp (vector float, vector float);
10854 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
10856 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
10858 vector bool short vec_vcmpgtsh (vector signed short,
10859 vector signed short);
10861 vector bool short vec_vcmpgtuh (vector unsigned short,
10862 vector unsigned short);
10864 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
10866 vector bool char vec_vcmpgtub (vector unsigned char,
10867 vector unsigned char);
10869 vector bool int vec_cmple (vector float, vector float);
10871 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
10872 vector bool char vec_cmplt (vector signed char, vector signed char);
10873 vector bool short vec_cmplt (vector unsigned short,
10874 vector unsigned short);
10875 vector bool short vec_cmplt (vector signed short, vector signed short);
10876 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
10877 vector bool int vec_cmplt (vector signed int, vector signed int);
10878 vector bool int vec_cmplt (vector float, vector float);
10880 vector float vec_ctf (vector unsigned int, const int);
10881 vector float vec_ctf (vector signed int, const int);
10883 vector float vec_vcfsx (vector signed int, const int);
10885 vector float vec_vcfux (vector unsigned int, const int);
10887 vector signed int vec_cts (vector float, const int);
10889 vector unsigned int vec_ctu (vector float, const int);
10891 void vec_dss (const int);
10893 void vec_dssall (void);
10895 void vec_dst (const vector unsigned char *, int, const int);
10896 void vec_dst (const vector signed char *, int, const int);
10897 void vec_dst (const vector bool char *, int, const int);
10898 void vec_dst (const vector unsigned short *, int, const int);
10899 void vec_dst (const vector signed short *, int, const int);
10900 void vec_dst (const vector bool short *, int, const int);
10901 void vec_dst (const vector pixel *, int, const int);
10902 void vec_dst (const vector unsigned int *, int, const int);
10903 void vec_dst (const vector signed int *, int, const int);
10904 void vec_dst (const vector bool int *, int, const int);
10905 void vec_dst (const vector float *, int, const int);
10906 void vec_dst (const unsigned char *, int, const int);
10907 void vec_dst (const signed char *, int, const int);
10908 void vec_dst (const unsigned short *, int, const int);
10909 void vec_dst (const short *, int, const int);
10910 void vec_dst (const unsigned int *, int, const int);
10911 void vec_dst (const int *, int, const int);
10912 void vec_dst (const unsigned long *, int, const int);
10913 void vec_dst (const long *, int, const int);
10914 void vec_dst (const float *, int, const int);
10916 void vec_dstst (const vector unsigned char *, int, const int);
10917 void vec_dstst (const vector signed char *, int, const int);
10918 void vec_dstst (const vector bool char *, int, const int);
10919 void vec_dstst (const vector unsigned short *, int, const int);
10920 void vec_dstst (const vector signed short *, int, const int);
10921 void vec_dstst (const vector bool short *, int, const int);
10922 void vec_dstst (const vector pixel *, int, const int);
10923 void vec_dstst (const vector unsigned int *, int, const int);
10924 void vec_dstst (const vector signed int *, int, const int);
10925 void vec_dstst (const vector bool int *, int, const int);
10926 void vec_dstst (const vector float *, int, const int);
10927 void vec_dstst (const unsigned char *, int, const int);
10928 void vec_dstst (const signed char *, int, const int);
10929 void vec_dstst (const unsigned short *, int, const int);
10930 void vec_dstst (const short *, int, const int);
10931 void vec_dstst (const unsigned int *, int, const int);
10932 void vec_dstst (const int *, int, const int);
10933 void vec_dstst (const unsigned long *, int, const int);
10934 void vec_dstst (const long *, int, const int);
10935 void vec_dstst (const float *, int, const int);
10937 void vec_dststt (const vector unsigned char *, int, const int);
10938 void vec_dststt (const vector signed char *, int, const int);
10939 void vec_dststt (const vector bool char *, int, const int);
10940 void vec_dststt (const vector unsigned short *, int, const int);
10941 void vec_dststt (const vector signed short *, int, const int);
10942 void vec_dststt (const vector bool short *, int, const int);
10943 void vec_dststt (const vector pixel *, int, const int);
10944 void vec_dststt (const vector unsigned int *, int, const int);
10945 void vec_dststt (const vector signed int *, int, const int);
10946 void vec_dststt (const vector bool int *, int, const int);
10947 void vec_dststt (const vector float *, int, const int);
10948 void vec_dststt (const unsigned char *, int, const int);
10949 void vec_dststt (const signed char *, int, const int);
10950 void vec_dststt (const unsigned short *, int, const int);
10951 void vec_dststt (const short *, int, const int);
10952 void vec_dststt (const unsigned int *, int, const int);
10953 void vec_dststt (const int *, int, const int);
10954 void vec_dststt (const unsigned long *, int, const int);
10955 void vec_dststt (const long *, int, const int);
10956 void vec_dststt (const float *, int, const int);
10958 void vec_dstt (const vector unsigned char *, int, const int);
10959 void vec_dstt (const vector signed char *, int, const int);
10960 void vec_dstt (const vector bool char *, int, const int);
10961 void vec_dstt (const vector unsigned short *, int, const int);
10962 void vec_dstt (const vector signed short *, int, const int);
10963 void vec_dstt (const vector bool short *, int, const int);
10964 void vec_dstt (const vector pixel *, int, const int);
10965 void vec_dstt (const vector unsigned int *, int, const int);
10966 void vec_dstt (const vector signed int *, int, const int);
10967 void vec_dstt (const vector bool int *, int, const int);
10968 void vec_dstt (const vector float *, int, const int);
10969 void vec_dstt (const unsigned char *, int, const int);
10970 void vec_dstt (const signed char *, int, const int);
10971 void vec_dstt (const unsigned short *, int, const int);
10972 void vec_dstt (const short *, int, const int);
10973 void vec_dstt (const unsigned int *, int, const int);
10974 void vec_dstt (const int *, int, const int);
10975 void vec_dstt (const unsigned long *, int, const int);
10976 void vec_dstt (const long *, int, const int);
10977 void vec_dstt (const float *, int, const int);
10979 vector float vec_expte (vector float);
10981 vector float vec_floor (vector float);
10983 vector float vec_ld (int, const vector float *);
10984 vector float vec_ld (int, const float *);
10985 vector bool int vec_ld (int, const vector bool int *);
10986 vector signed int vec_ld (int, const vector signed int *);
10987 vector signed int vec_ld (int, const int *);
10988 vector signed int vec_ld (int, const long *);
10989 vector unsigned int vec_ld (int, const vector unsigned int *);
10990 vector unsigned int vec_ld (int, const unsigned int *);
10991 vector unsigned int vec_ld (int, const unsigned long *);
10992 vector bool short vec_ld (int, const vector bool short *);
10993 vector pixel vec_ld (int, const vector pixel *);
10994 vector signed short vec_ld (int, const vector signed short *);
10995 vector signed short vec_ld (int, const short *);
10996 vector unsigned short vec_ld (int, const vector unsigned short *);
10997 vector unsigned short vec_ld (int, const unsigned short *);
10998 vector bool char vec_ld (int, const vector bool char *);
10999 vector signed char vec_ld (int, const vector signed char *);
11000 vector signed char vec_ld (int, const signed char *);
11001 vector unsigned char vec_ld (int, const vector unsigned char *);
11002 vector unsigned char vec_ld (int, const unsigned char *);
11004 vector signed char vec_lde (int, const signed char *);
11005 vector unsigned char vec_lde (int, const unsigned char *);
11006 vector signed short vec_lde (int, const short *);
11007 vector unsigned short vec_lde (int, const unsigned short *);
11008 vector float vec_lde (int, const float *);
11009 vector signed int vec_lde (int, const int *);
11010 vector unsigned int vec_lde (int, const unsigned int *);
11011 vector signed int vec_lde (int, const long *);
11012 vector unsigned int vec_lde (int, const unsigned long *);
11014 vector float vec_lvewx (int, float *);
11015 vector signed int vec_lvewx (int, int *);
11016 vector unsigned int vec_lvewx (int, unsigned int *);
11017 vector signed int vec_lvewx (int, long *);
11018 vector unsigned int vec_lvewx (int, unsigned long *);
11020 vector signed short vec_lvehx (int, short *);
11021 vector unsigned short vec_lvehx (int, unsigned short *);
11023 vector signed char vec_lvebx (int, char *);
11024 vector unsigned char vec_lvebx (int, unsigned char *);
11026 vector float vec_ldl (int, const vector float *);
11027 vector float vec_ldl (int, const float *);
11028 vector bool int vec_ldl (int, const vector bool int *);
11029 vector signed int vec_ldl (int, const vector signed int *);
11030 vector signed int vec_ldl (int, const int *);
11031 vector signed int vec_ldl (int, const long *);
11032 vector unsigned int vec_ldl (int, const vector unsigned int *);
11033 vector unsigned int vec_ldl (int, const unsigned int *);
11034 vector unsigned int vec_ldl (int, const unsigned long *);
11035 vector bool short vec_ldl (int, const vector bool short *);
11036 vector pixel vec_ldl (int, const vector pixel *);
11037 vector signed short vec_ldl (int, const vector signed short *);
11038 vector signed short vec_ldl (int, const short *);
11039 vector unsigned short vec_ldl (int, const vector unsigned short *);
11040 vector unsigned short vec_ldl (int, const unsigned short *);
11041 vector bool char vec_ldl (int, const vector bool char *);
11042 vector signed char vec_ldl (int, const vector signed char *);
11043 vector signed char vec_ldl (int, const signed char *);
11044 vector unsigned char vec_ldl (int, const vector unsigned char *);
11045 vector unsigned char vec_ldl (int, const unsigned char *);
11047 vector float vec_loge (vector float);
11049 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
11050 vector unsigned char vec_lvsl (int, const volatile signed char *);
11051 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
11052 vector unsigned char vec_lvsl (int, const volatile short *);
11053 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
11054 vector unsigned char vec_lvsl (int, const volatile int *);
11055 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
11056 vector unsigned char vec_lvsl (int, const volatile long *);
11057 vector unsigned char vec_lvsl (int, const volatile float *);
11059 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
11060 vector unsigned char vec_lvsr (int, const volatile signed char *);
11061 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
11062 vector unsigned char vec_lvsr (int, const volatile short *);
11063 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
11064 vector unsigned char vec_lvsr (int, const volatile int *);
11065 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
11066 vector unsigned char vec_lvsr (int, const volatile long *);
11067 vector unsigned char vec_lvsr (int, const volatile float *);
11069 vector float vec_madd (vector float, vector float, vector float);
11071 vector signed short vec_madds (vector signed short,
11072 vector signed short,
11073 vector signed short);
11075 vector unsigned char vec_max (vector bool char, vector unsigned char);
11076 vector unsigned char vec_max (vector unsigned char, vector bool char);
11077 vector unsigned char vec_max (vector unsigned char,
11078 vector unsigned char);
11079 vector signed char vec_max (vector bool char, vector signed char);
11080 vector signed char vec_max (vector signed char, vector bool char);
11081 vector signed char vec_max (vector signed char, vector signed char);
11082 vector unsigned short vec_max (vector bool short,
11083 vector unsigned short);
11084 vector unsigned short vec_max (vector unsigned short,
11085 vector bool short);
11086 vector unsigned short vec_max (vector unsigned short,
11087 vector unsigned short);
11088 vector signed short vec_max (vector bool short, vector signed short);
11089 vector signed short vec_max (vector signed short, vector bool short);
11090 vector signed short vec_max (vector signed short, vector signed short);
11091 vector unsigned int vec_max (vector bool int, vector unsigned int);
11092 vector unsigned int vec_max (vector unsigned int, vector bool int);
11093 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
11094 vector signed int vec_max (vector bool int, vector signed int);
11095 vector signed int vec_max (vector signed int, vector bool int);
11096 vector signed int vec_max (vector signed int, vector signed int);
11097 vector float vec_max (vector float, vector float);
11099 vector float vec_vmaxfp (vector float, vector float);
11101 vector signed int vec_vmaxsw (vector bool int, vector signed int);
11102 vector signed int vec_vmaxsw (vector signed int, vector bool int);
11103 vector signed int vec_vmaxsw (vector signed int, vector signed int);
11105 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
11106 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
11107 vector unsigned int vec_vmaxuw (vector unsigned int,
11108 vector unsigned int);
11110 vector signed short vec_vmaxsh (vector bool short, vector signed short);
11111 vector signed short vec_vmaxsh (vector signed short, vector bool short);
11112 vector signed short vec_vmaxsh (vector signed short,
11113 vector signed short);
11115 vector unsigned short vec_vmaxuh (vector bool short,
11116 vector unsigned short);
11117 vector unsigned short vec_vmaxuh (vector unsigned short,
11118 vector bool short);
11119 vector unsigned short vec_vmaxuh (vector unsigned short,
11120 vector unsigned short);
11122 vector signed char vec_vmaxsb (vector bool char, vector signed char);
11123 vector signed char vec_vmaxsb (vector signed char, vector bool char);
11124 vector signed char vec_vmaxsb (vector signed char, vector signed char);
11126 vector unsigned char vec_vmaxub (vector bool char,
11127 vector unsigned char);
11128 vector unsigned char vec_vmaxub (vector unsigned char,
11130 vector unsigned char vec_vmaxub (vector unsigned char,
11131 vector unsigned char);
11133 vector bool char vec_mergeh (vector bool char, vector bool char);
11134 vector signed char vec_mergeh (vector signed char, vector signed char);
11135 vector unsigned char vec_mergeh (vector unsigned char,
11136 vector unsigned char);
11137 vector bool short vec_mergeh (vector bool short, vector bool short);
11138 vector pixel vec_mergeh (vector pixel, vector pixel);
11139 vector signed short vec_mergeh (vector signed short,
11140 vector signed short);
11141 vector unsigned short vec_mergeh (vector unsigned short,
11142 vector unsigned short);
11143 vector float vec_mergeh (vector float, vector float);
11144 vector bool int vec_mergeh (vector bool int, vector bool int);
11145 vector signed int vec_mergeh (vector signed int, vector signed int);
11146 vector unsigned int vec_mergeh (vector unsigned int,
11147 vector unsigned int);
11149 vector float vec_vmrghw (vector float, vector float);
11150 vector bool int vec_vmrghw (vector bool int, vector bool int);
11151 vector signed int vec_vmrghw (vector signed int, vector signed int);
11152 vector unsigned int vec_vmrghw (vector unsigned int,
11153 vector unsigned int);
11155 vector bool short vec_vmrghh (vector bool short, vector bool short);
11156 vector signed short vec_vmrghh (vector signed short,
11157 vector signed short);
11158 vector unsigned short vec_vmrghh (vector unsigned short,
11159 vector unsigned short);
11160 vector pixel vec_vmrghh (vector pixel, vector pixel);
11162 vector bool char vec_vmrghb (vector bool char, vector bool char);
11163 vector signed char vec_vmrghb (vector signed char, vector signed char);
11164 vector unsigned char vec_vmrghb (vector unsigned char,
11165 vector unsigned char);
11167 vector bool char vec_mergel (vector bool char, vector bool char);
11168 vector signed char vec_mergel (vector signed char, vector signed char);
11169 vector unsigned char vec_mergel (vector unsigned char,
11170 vector unsigned char);
11171 vector bool short vec_mergel (vector bool short, vector bool short);
11172 vector pixel vec_mergel (vector pixel, vector pixel);
11173 vector signed short vec_mergel (vector signed short,
11174 vector signed short);
11175 vector unsigned short vec_mergel (vector unsigned short,
11176 vector unsigned short);
11177 vector float vec_mergel (vector float, vector float);
11178 vector bool int vec_mergel (vector bool int, vector bool int);
11179 vector signed int vec_mergel (vector signed int, vector signed int);
11180 vector unsigned int vec_mergel (vector unsigned int,
11181 vector unsigned int);
11183 vector float vec_vmrglw (vector float, vector float);
11184 vector signed int vec_vmrglw (vector signed int, vector signed int);
11185 vector unsigned int vec_vmrglw (vector unsigned int,
11186 vector unsigned int);
11187 vector bool int vec_vmrglw (vector bool int, vector bool int);
11189 vector bool short vec_vmrglh (vector bool short, vector bool short);
11190 vector signed short vec_vmrglh (vector signed short,
11191 vector signed short);
11192 vector unsigned short vec_vmrglh (vector unsigned short,
11193 vector unsigned short);
11194 vector pixel vec_vmrglh (vector pixel, vector pixel);
11196 vector bool char vec_vmrglb (vector bool char, vector bool char);
11197 vector signed char vec_vmrglb (vector signed char, vector signed char);
11198 vector unsigned char vec_vmrglb (vector unsigned char,
11199 vector unsigned char);
11201 vector unsigned short vec_mfvscr (void);
11203 vector unsigned char vec_min (vector bool char, vector unsigned char);
11204 vector unsigned char vec_min (vector unsigned char, vector bool char);
11205 vector unsigned char vec_min (vector unsigned char,
11206 vector unsigned char);
11207 vector signed char vec_min (vector bool char, vector signed char);
11208 vector signed char vec_min (vector signed char, vector bool char);
11209 vector signed char vec_min (vector signed char, vector signed char);
11210 vector unsigned short vec_min (vector bool short,
11211 vector unsigned short);
11212 vector unsigned short vec_min (vector unsigned short,
11213 vector bool short);
11214 vector unsigned short vec_min (vector unsigned short,
11215 vector unsigned short);
11216 vector signed short vec_min (vector bool short, vector signed short);
11217 vector signed short vec_min (vector signed short, vector bool short);
11218 vector signed short vec_min (vector signed short, vector signed short);
11219 vector unsigned int vec_min (vector bool int, vector unsigned int);
11220 vector unsigned int vec_min (vector unsigned int, vector bool int);
11221 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
11222 vector signed int vec_min (vector bool int, vector signed int);
11223 vector signed int vec_min (vector signed int, vector bool int);
11224 vector signed int vec_min (vector signed int, vector signed int);
11225 vector float vec_min (vector float, vector float);
11227 vector float vec_vminfp (vector float, vector float);
11229 vector signed int vec_vminsw (vector bool int, vector signed int);
11230 vector signed int vec_vminsw (vector signed int, vector bool int);
11231 vector signed int vec_vminsw (vector signed int, vector signed int);
11233 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
11234 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
11235 vector unsigned int vec_vminuw (vector unsigned int,
11236 vector unsigned int);
11238 vector signed short vec_vminsh (vector bool short, vector signed short);
11239 vector signed short vec_vminsh (vector signed short, vector bool short);
11240 vector signed short vec_vminsh (vector signed short,
11241 vector signed short);
11243 vector unsigned short vec_vminuh (vector bool short,
11244 vector unsigned short);
11245 vector unsigned short vec_vminuh (vector unsigned short,
11246 vector bool short);
11247 vector unsigned short vec_vminuh (vector unsigned short,
11248 vector unsigned short);
11250 vector signed char vec_vminsb (vector bool char, vector signed char);
11251 vector signed char vec_vminsb (vector signed char, vector bool char);
11252 vector signed char vec_vminsb (vector signed char, vector signed char);
11254 vector unsigned char vec_vminub (vector bool char,
11255 vector unsigned char);
11256 vector unsigned char vec_vminub (vector unsigned char,
11258 vector unsigned char vec_vminub (vector unsigned char,
11259 vector unsigned char);
11261 vector signed short vec_mladd (vector signed short,
11262 vector signed short,
11263 vector signed short);
11264 vector signed short vec_mladd (vector signed short,
11265 vector unsigned short,
11266 vector unsigned short);
11267 vector signed short vec_mladd (vector unsigned short,
11268 vector signed short,
11269 vector signed short);
11270 vector unsigned short vec_mladd (vector unsigned short,
11271 vector unsigned short,
11272 vector unsigned short);
11274 vector signed short vec_mradds (vector signed short,
11275 vector signed short,
11276 vector signed short);
11278 vector unsigned int vec_msum (vector unsigned char,
11279 vector unsigned char,
11280 vector unsigned int);
11281 vector signed int vec_msum (vector signed char,
11282 vector unsigned char,
11283 vector signed int);
11284 vector unsigned int vec_msum (vector unsigned short,
11285 vector unsigned short,
11286 vector unsigned int);
11287 vector signed int vec_msum (vector signed short,
11288 vector signed short,
11289 vector signed int);
11291 vector signed int vec_vmsumshm (vector signed short,
11292 vector signed short,
11293 vector signed int);
11295 vector unsigned int vec_vmsumuhm (vector unsigned short,
11296 vector unsigned short,
11297 vector unsigned int);
11299 vector signed int vec_vmsummbm (vector signed char,
11300 vector unsigned char,
11301 vector signed int);
11303 vector unsigned int vec_vmsumubm (vector unsigned char,
11304 vector unsigned char,
11305 vector unsigned int);
11307 vector unsigned int vec_msums (vector unsigned short,
11308 vector unsigned short,
11309 vector unsigned int);
11310 vector signed int vec_msums (vector signed short,
11311 vector signed short,
11312 vector signed int);
11314 vector signed int vec_vmsumshs (vector signed short,
11315 vector signed short,
11316 vector signed int);
11318 vector unsigned int vec_vmsumuhs (vector unsigned short,
11319 vector unsigned short,
11320 vector unsigned int);
11322 void vec_mtvscr (vector signed int);
11323 void vec_mtvscr (vector unsigned int);
11324 void vec_mtvscr (vector bool int);
11325 void vec_mtvscr (vector signed short);
11326 void vec_mtvscr (vector unsigned short);
11327 void vec_mtvscr (vector bool short);
11328 void vec_mtvscr (vector pixel);
11329 void vec_mtvscr (vector signed char);
11330 void vec_mtvscr (vector unsigned char);
11331 void vec_mtvscr (vector bool char);
11333 vector unsigned short vec_mule (vector unsigned char,
11334 vector unsigned char);
11335 vector signed short vec_mule (vector signed char,
11336 vector signed char);
11337 vector unsigned int vec_mule (vector unsigned short,
11338 vector unsigned short);
11339 vector signed int vec_mule (vector signed short, vector signed short);
11341 vector signed int vec_vmulesh (vector signed short,
11342 vector signed short);
11344 vector unsigned int vec_vmuleuh (vector unsigned short,
11345 vector unsigned short);
11347 vector signed short vec_vmulesb (vector signed char,
11348 vector signed char);
11350 vector unsigned short vec_vmuleub (vector unsigned char,
11351 vector unsigned char);
11353 vector unsigned short vec_mulo (vector unsigned char,
11354 vector unsigned char);
11355 vector signed short vec_mulo (vector signed char, vector signed char);
11356 vector unsigned int vec_mulo (vector unsigned short,
11357 vector unsigned short);
11358 vector signed int vec_mulo (vector signed short, vector signed short);
11360 vector signed int vec_vmulosh (vector signed short,
11361 vector signed short);
11363 vector unsigned int vec_vmulouh (vector unsigned short,
11364 vector unsigned short);
11366 vector signed short vec_vmulosb (vector signed char,
11367 vector signed char);
11369 vector unsigned short vec_vmuloub (vector unsigned char,
11370 vector unsigned char);
11372 vector float vec_nmsub (vector float, vector float, vector float);
11374 vector float vec_nor (vector float, vector float);
11375 vector signed int vec_nor (vector signed int, vector signed int);
11376 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
11377 vector bool int vec_nor (vector bool int, vector bool int);
11378 vector signed short vec_nor (vector signed short, vector signed short);
11379 vector unsigned short vec_nor (vector unsigned short,
11380 vector unsigned short);
11381 vector bool short vec_nor (vector bool short, vector bool short);
11382 vector signed char vec_nor (vector signed char, vector signed char);
11383 vector unsigned char vec_nor (vector unsigned char,
11384 vector unsigned char);
11385 vector bool char vec_nor (vector bool char, vector bool char);
11387 vector float vec_or (vector float, vector float);
11388 vector float vec_or (vector float, vector bool int);
11389 vector float vec_or (vector bool int, vector float);
11390 vector bool int vec_or (vector bool int, vector bool int);
11391 vector signed int vec_or (vector bool int, vector signed int);
11392 vector signed int vec_or (vector signed int, vector bool int);
11393 vector signed int vec_or (vector signed int, vector signed int);
11394 vector unsigned int vec_or (vector bool int, vector unsigned int);
11395 vector unsigned int vec_or (vector unsigned int, vector bool int);
11396 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
11397 vector bool short vec_or (vector bool short, vector bool short);
11398 vector signed short vec_or (vector bool short, vector signed short);
11399 vector signed short vec_or (vector signed short, vector bool short);
11400 vector signed short vec_or (vector signed short, vector signed short);
11401 vector unsigned short vec_or (vector bool short, vector unsigned short);
11402 vector unsigned short vec_or (vector unsigned short, vector bool short);
11403 vector unsigned short vec_or (vector unsigned short,
11404 vector unsigned short);
11405 vector signed char vec_or (vector bool char, vector signed char);
11406 vector bool char vec_or (vector bool char, vector bool char);
11407 vector signed char vec_or (vector signed char, vector bool char);
11408 vector signed char vec_or (vector signed char, vector signed char);
11409 vector unsigned char vec_or (vector bool char, vector unsigned char);
11410 vector unsigned char vec_or (vector unsigned char, vector bool char);
11411 vector unsigned char vec_or (vector unsigned char,
11412 vector unsigned char);
11414 vector signed char vec_pack (vector signed short, vector signed short);
11415 vector unsigned char vec_pack (vector unsigned short,
11416 vector unsigned short);
11417 vector bool char vec_pack (vector bool short, vector bool short);
11418 vector signed short vec_pack (vector signed int, vector signed int);
11419 vector unsigned short vec_pack (vector unsigned int,
11420 vector unsigned int);
11421 vector bool short vec_pack (vector bool int, vector bool int);
11423 vector bool short vec_vpkuwum (vector bool int, vector bool int);
11424 vector signed short vec_vpkuwum (vector signed int, vector signed int);
11425 vector unsigned short vec_vpkuwum (vector unsigned int,
11426 vector unsigned int);
11428 vector bool char vec_vpkuhum (vector bool short, vector bool short);
11429 vector signed char vec_vpkuhum (vector signed short,
11430 vector signed short);
11431 vector unsigned char vec_vpkuhum (vector unsigned short,
11432 vector unsigned short);
11434 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
11436 vector unsigned char vec_packs (vector unsigned short,
11437 vector unsigned short);
11438 vector signed char vec_packs (vector signed short, vector signed short);
11439 vector unsigned short vec_packs (vector unsigned int,
11440 vector unsigned int);
11441 vector signed short vec_packs (vector signed int, vector signed int);
11443 vector signed short vec_vpkswss (vector signed int, vector signed int);
11445 vector unsigned short vec_vpkuwus (vector unsigned int,
11446 vector unsigned int);
11448 vector signed char vec_vpkshss (vector signed short,
11449 vector signed short);
11451 vector unsigned char vec_vpkuhus (vector unsigned short,
11452 vector unsigned short);
11454 vector unsigned char vec_packsu (vector unsigned short,
11455 vector unsigned short);
11456 vector unsigned char vec_packsu (vector signed short,
11457 vector signed short);
11458 vector unsigned short vec_packsu (vector unsigned int,
11459 vector unsigned int);
11460 vector unsigned short vec_packsu (vector signed int, vector signed int);
11462 vector unsigned short vec_vpkswus (vector signed int,
11463 vector signed int);
11465 vector unsigned char vec_vpkshus (vector signed short,
11466 vector signed short);
11468 vector float vec_perm (vector float,
11470 vector unsigned char);
11471 vector signed int vec_perm (vector signed int,
11473 vector unsigned char);
11474 vector unsigned int vec_perm (vector unsigned int,
11475 vector unsigned int,
11476 vector unsigned char);
11477 vector bool int vec_perm (vector bool int,
11479 vector unsigned char);
11480 vector signed short vec_perm (vector signed short,
11481 vector signed short,
11482 vector unsigned char);
11483 vector unsigned short vec_perm (vector unsigned short,
11484 vector unsigned short,
11485 vector unsigned char);
11486 vector bool short vec_perm (vector bool short,
11488 vector unsigned char);
11489 vector pixel vec_perm (vector pixel,
11491 vector unsigned char);
11492 vector signed char vec_perm (vector signed char,
11493 vector signed char,
11494 vector unsigned char);
11495 vector unsigned char vec_perm (vector unsigned char,
11496 vector unsigned char,
11497 vector unsigned char);
11498 vector bool char vec_perm (vector bool char,
11500 vector unsigned char);
11502 vector float vec_re (vector float);
11504 vector signed char vec_rl (vector signed char,
11505 vector unsigned char);
11506 vector unsigned char vec_rl (vector unsigned char,
11507 vector unsigned char);
11508 vector signed short vec_rl (vector signed short, vector unsigned short);
11509 vector unsigned short vec_rl (vector unsigned short,
11510 vector unsigned short);
11511 vector signed int vec_rl (vector signed int, vector unsigned int);
11512 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
11514 vector signed int vec_vrlw (vector signed int, vector unsigned int);
11515 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
11517 vector signed short vec_vrlh (vector signed short,
11518 vector unsigned short);
11519 vector unsigned short vec_vrlh (vector unsigned short,
11520 vector unsigned short);
11522 vector signed char vec_vrlb (vector signed char, vector unsigned char);
11523 vector unsigned char vec_vrlb (vector unsigned char,
11524 vector unsigned char);
11526 vector float vec_round (vector float);
11528 vector float vec_recip (vector float, vector float);
11530 vector float vec_rsqrt (vector float);
11532 vector float vec_rsqrte (vector float);
11534 vector float vec_sel (vector float, vector float, vector bool int);
11535 vector float vec_sel (vector float, vector float, vector unsigned int);
11536 vector signed int vec_sel (vector signed int,
11539 vector signed int vec_sel (vector signed int,
11541 vector unsigned int);
11542 vector unsigned int vec_sel (vector unsigned int,
11543 vector unsigned int,
11545 vector unsigned int vec_sel (vector unsigned int,
11546 vector unsigned int,
11547 vector unsigned int);
11548 vector bool int vec_sel (vector bool int,
11551 vector bool int vec_sel (vector bool int,
11553 vector unsigned int);
11554 vector signed short vec_sel (vector signed short,
11555 vector signed short,
11556 vector bool short);
11557 vector signed short vec_sel (vector signed short,
11558 vector signed short,
11559 vector unsigned short);
11560 vector unsigned short vec_sel (vector unsigned short,
11561 vector unsigned short,
11562 vector bool short);
11563 vector unsigned short vec_sel (vector unsigned short,
11564 vector unsigned short,
11565 vector unsigned short);
11566 vector bool short vec_sel (vector bool short,
11568 vector bool short);
11569 vector bool short vec_sel (vector bool short,
11571 vector unsigned short);
11572 vector signed char vec_sel (vector signed char,
11573 vector signed char,
11575 vector signed char vec_sel (vector signed char,
11576 vector signed char,
11577 vector unsigned char);
11578 vector unsigned char vec_sel (vector unsigned char,
11579 vector unsigned char,
11581 vector unsigned char vec_sel (vector unsigned char,
11582 vector unsigned char,
11583 vector unsigned char);
11584 vector bool char vec_sel (vector bool char,
11587 vector bool char vec_sel (vector bool char,
11589 vector unsigned char);
11591 vector signed char vec_sl (vector signed char,
11592 vector unsigned char);
11593 vector unsigned char vec_sl (vector unsigned char,
11594 vector unsigned char);
11595 vector signed short vec_sl (vector signed short, vector unsigned short);
11596 vector unsigned short vec_sl (vector unsigned short,
11597 vector unsigned short);
11598 vector signed int vec_sl (vector signed int, vector unsigned int);
11599 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
11601 vector signed int vec_vslw (vector signed int, vector unsigned int);
11602 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
11604 vector signed short vec_vslh (vector signed short,
11605 vector unsigned short);
11606 vector unsigned short vec_vslh (vector unsigned short,
11607 vector unsigned short);
11609 vector signed char vec_vslb (vector signed char, vector unsigned char);
11610 vector unsigned char vec_vslb (vector unsigned char,
11611 vector unsigned char);
11613 vector float vec_sld (vector float, vector float, const int);
11614 vector signed int vec_sld (vector signed int,
11617 vector unsigned int vec_sld (vector unsigned int,
11618 vector unsigned int,
11620 vector bool int vec_sld (vector bool int,
11623 vector signed short vec_sld (vector signed short,
11624 vector signed short,
11626 vector unsigned short vec_sld (vector unsigned short,
11627 vector unsigned short,
11629 vector bool short vec_sld (vector bool short,
11632 vector pixel vec_sld (vector pixel,
11635 vector signed char vec_sld (vector signed char,
11636 vector signed char,
11638 vector unsigned char vec_sld (vector unsigned char,
11639 vector unsigned char,
11641 vector bool char vec_sld (vector bool char,
11645 vector signed int vec_sll (vector signed int,
11646 vector unsigned int);
11647 vector signed int vec_sll (vector signed int,
11648 vector unsigned short);
11649 vector signed int vec_sll (vector signed int,
11650 vector unsigned char);
11651 vector unsigned int vec_sll (vector unsigned int,
11652 vector unsigned int);
11653 vector unsigned int vec_sll (vector unsigned int,
11654 vector unsigned short);
11655 vector unsigned int vec_sll (vector unsigned int,
11656 vector unsigned char);
11657 vector bool int vec_sll (vector bool int,
11658 vector unsigned int);
11659 vector bool int vec_sll (vector bool int,
11660 vector unsigned short);
11661 vector bool int vec_sll (vector bool int,
11662 vector unsigned char);
11663 vector signed short vec_sll (vector signed short,
11664 vector unsigned int);
11665 vector signed short vec_sll (vector signed short,
11666 vector unsigned short);
11667 vector signed short vec_sll (vector signed short,
11668 vector unsigned char);
11669 vector unsigned short vec_sll (vector unsigned short,
11670 vector unsigned int);
11671 vector unsigned short vec_sll (vector unsigned short,
11672 vector unsigned short);
11673 vector unsigned short vec_sll (vector unsigned short,
11674 vector unsigned char);
11675 vector bool short vec_sll (vector bool short, vector unsigned int);
11676 vector bool short vec_sll (vector bool short, vector unsigned short);
11677 vector bool short vec_sll (vector bool short, vector unsigned char);
11678 vector pixel vec_sll (vector pixel, vector unsigned int);
11679 vector pixel vec_sll (vector pixel, vector unsigned short);
11680 vector pixel vec_sll (vector pixel, vector unsigned char);
11681 vector signed char vec_sll (vector signed char, vector unsigned int);
11682 vector signed char vec_sll (vector signed char, vector unsigned short);
11683 vector signed char vec_sll (vector signed char, vector unsigned char);
11684 vector unsigned char vec_sll (vector unsigned char,
11685 vector unsigned int);
11686 vector unsigned char vec_sll (vector unsigned char,
11687 vector unsigned short);
11688 vector unsigned char vec_sll (vector unsigned char,
11689 vector unsigned char);
11690 vector bool char vec_sll (vector bool char, vector unsigned int);
11691 vector bool char vec_sll (vector bool char, vector unsigned short);
11692 vector bool char vec_sll (vector bool char, vector unsigned char);
11694 vector float vec_slo (vector float, vector signed char);
11695 vector float vec_slo (vector float, vector unsigned char);
11696 vector signed int vec_slo (vector signed int, vector signed char);
11697 vector signed int vec_slo (vector signed int, vector unsigned char);
11698 vector unsigned int vec_slo (vector unsigned int, vector signed char);
11699 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
11700 vector signed short vec_slo (vector signed short, vector signed char);
11701 vector signed short vec_slo (vector signed short, vector unsigned char);
11702 vector unsigned short vec_slo (vector unsigned short,
11703 vector signed char);
11704 vector unsigned short vec_slo (vector unsigned short,
11705 vector unsigned char);
11706 vector pixel vec_slo (vector pixel, vector signed char);
11707 vector pixel vec_slo (vector pixel, vector unsigned char);
11708 vector signed char vec_slo (vector signed char, vector signed char);
11709 vector signed char vec_slo (vector signed char, vector unsigned char);
11710 vector unsigned char vec_slo (vector unsigned char, vector signed char);
11711 vector unsigned char vec_slo (vector unsigned char,
11712 vector unsigned char);
11714 vector signed char vec_splat (vector signed char, const int);
11715 vector unsigned char vec_splat (vector unsigned char, const int);
11716 vector bool char vec_splat (vector bool char, const int);
11717 vector signed short vec_splat (vector signed short, const int);
11718 vector unsigned short vec_splat (vector unsigned short, const int);
11719 vector bool short vec_splat (vector bool short, const int);
11720 vector pixel vec_splat (vector pixel, const int);
11721 vector float vec_splat (vector float, const int);
11722 vector signed int vec_splat (vector signed int, const int);
11723 vector unsigned int vec_splat (vector unsigned int, const int);
11724 vector bool int vec_splat (vector bool int, const int);
11726 vector float vec_vspltw (vector float, const int);
11727 vector signed int vec_vspltw (vector signed int, const int);
11728 vector unsigned int vec_vspltw (vector unsigned int, const int);
11729 vector bool int vec_vspltw (vector bool int, const int);
11731 vector bool short vec_vsplth (vector bool short, const int);
11732 vector signed short vec_vsplth (vector signed short, const int);
11733 vector unsigned short vec_vsplth (vector unsigned short, const int);
11734 vector pixel vec_vsplth (vector pixel, const int);
11736 vector signed char vec_vspltb (vector signed char, const int);
11737 vector unsigned char vec_vspltb (vector unsigned char, const int);
11738 vector bool char vec_vspltb (vector bool char, const int);
11740 vector signed char vec_splat_s8 (const int);
11742 vector signed short vec_splat_s16 (const int);
11744 vector signed int vec_splat_s32 (const int);
11746 vector unsigned char vec_splat_u8 (const int);
11748 vector unsigned short vec_splat_u16 (const int);
11750 vector unsigned int vec_splat_u32 (const int);
11752 vector signed char vec_sr (vector signed char, vector unsigned char);
11753 vector unsigned char vec_sr (vector unsigned char,
11754 vector unsigned char);
11755 vector signed short vec_sr (vector signed short,
11756 vector unsigned short);
11757 vector unsigned short vec_sr (vector unsigned short,
11758 vector unsigned short);
11759 vector signed int vec_sr (vector signed int, vector unsigned int);
11760 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
11762 vector signed int vec_vsrw (vector signed int, vector unsigned int);
11763 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
11765 vector signed short vec_vsrh (vector signed short,
11766 vector unsigned short);
11767 vector unsigned short vec_vsrh (vector unsigned short,
11768 vector unsigned short);
11770 vector signed char vec_vsrb (vector signed char, vector unsigned char);
11771 vector unsigned char vec_vsrb (vector unsigned char,
11772 vector unsigned char);
11774 vector signed char vec_sra (vector signed char, vector unsigned char);
11775 vector unsigned char vec_sra (vector unsigned char,
11776 vector unsigned char);
11777 vector signed short vec_sra (vector signed short,
11778 vector unsigned short);
11779 vector unsigned short vec_sra (vector unsigned short,
11780 vector unsigned short);
11781 vector signed int vec_sra (vector signed int, vector unsigned int);
11782 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
11784 vector signed int vec_vsraw (vector signed int, vector unsigned int);
11785 vector unsigned int vec_vsraw (vector unsigned int,
11786 vector unsigned int);
11788 vector signed short vec_vsrah (vector signed short,
11789 vector unsigned short);
11790 vector unsigned short vec_vsrah (vector unsigned short,
11791 vector unsigned short);
11793 vector signed char vec_vsrab (vector signed char, vector unsigned char);
11794 vector unsigned char vec_vsrab (vector unsigned char,
11795 vector unsigned char);
11797 vector signed int vec_srl (vector signed int, vector unsigned int);
11798 vector signed int vec_srl (vector signed int, vector unsigned short);
11799 vector signed int vec_srl (vector signed int, vector unsigned char);
11800 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
11801 vector unsigned int vec_srl (vector unsigned int,
11802 vector unsigned short);
11803 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
11804 vector bool int vec_srl (vector bool int, vector unsigned int);
11805 vector bool int vec_srl (vector bool int, vector unsigned short);
11806 vector bool int vec_srl (vector bool int, vector unsigned char);
11807 vector signed short vec_srl (vector signed short, vector unsigned int);
11808 vector signed short vec_srl (vector signed short,
11809 vector unsigned short);
11810 vector signed short vec_srl (vector signed short, vector unsigned char);
11811 vector unsigned short vec_srl (vector unsigned short,
11812 vector unsigned int);
11813 vector unsigned short vec_srl (vector unsigned short,
11814 vector unsigned short);
11815 vector unsigned short vec_srl (vector unsigned short,
11816 vector unsigned char);
11817 vector bool short vec_srl (vector bool short, vector unsigned int);
11818 vector bool short vec_srl (vector bool short, vector unsigned short);
11819 vector bool short vec_srl (vector bool short, vector unsigned char);
11820 vector pixel vec_srl (vector pixel, vector unsigned int);
11821 vector pixel vec_srl (vector pixel, vector unsigned short);
11822 vector pixel vec_srl (vector pixel, vector unsigned char);
11823 vector signed char vec_srl (vector signed char, vector unsigned int);
11824 vector signed char vec_srl (vector signed char, vector unsigned short);
11825 vector signed char vec_srl (vector signed char, vector unsigned char);
11826 vector unsigned char vec_srl (vector unsigned char,
11827 vector unsigned int);
11828 vector unsigned char vec_srl (vector unsigned char,
11829 vector unsigned short);
11830 vector unsigned char vec_srl (vector unsigned char,
11831 vector unsigned char);
11832 vector bool char vec_srl (vector bool char, vector unsigned int);
11833 vector bool char vec_srl (vector bool char, vector unsigned short);
11834 vector bool char vec_srl (vector bool char, vector unsigned char);
11836 vector float vec_sro (vector float, vector signed char);
11837 vector float vec_sro (vector float, vector unsigned char);
11838 vector signed int vec_sro (vector signed int, vector signed char);
11839 vector signed int vec_sro (vector signed int, vector unsigned char);
11840 vector unsigned int vec_sro (vector unsigned int, vector signed char);
11841 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
11842 vector signed short vec_sro (vector signed short, vector signed char);
11843 vector signed short vec_sro (vector signed short, vector unsigned char);
11844 vector unsigned short vec_sro (vector unsigned short,
11845 vector signed char);
11846 vector unsigned short vec_sro (vector unsigned short,
11847 vector unsigned char);
11848 vector pixel vec_sro (vector pixel, vector signed char);
11849 vector pixel vec_sro (vector pixel, vector unsigned char);
11850 vector signed char vec_sro (vector signed char, vector signed char);
11851 vector signed char vec_sro (vector signed char, vector unsigned char);
11852 vector unsigned char vec_sro (vector unsigned char, vector signed char);
11853 vector unsigned char vec_sro (vector unsigned char,
11854 vector unsigned char);
11856 void vec_st (vector float, int, vector float *);
11857 void vec_st (vector float, int, float *);
11858 void vec_st (vector signed int, int, vector signed int *);
11859 void vec_st (vector signed int, int, int *);
11860 void vec_st (vector unsigned int, int, vector unsigned int *);
11861 void vec_st (vector unsigned int, int, unsigned int *);
11862 void vec_st (vector bool int, int, vector bool int *);
11863 void vec_st (vector bool int, int, unsigned int *);
11864 void vec_st (vector bool int, int, int *);
11865 void vec_st (vector signed short, int, vector signed short *);
11866 void vec_st (vector signed short, int, short *);
11867 void vec_st (vector unsigned short, int, vector unsigned short *);
11868 void vec_st (vector unsigned short, int, unsigned short *);
11869 void vec_st (vector bool short, int, vector bool short *);
11870 void vec_st (vector bool short, int, unsigned short *);
11871 void vec_st (vector pixel, int, vector pixel *);
11872 void vec_st (vector pixel, int, unsigned short *);
11873 void vec_st (vector pixel, int, short *);
11874 void vec_st (vector bool short, int, short *);
11875 void vec_st (vector signed char, int, vector signed char *);
11876 void vec_st (vector signed char, int, signed char *);
11877 void vec_st (vector unsigned char, int, vector unsigned char *);
11878 void vec_st (vector unsigned char, int, unsigned char *);
11879 void vec_st (vector bool char, int, vector bool char *);
11880 void vec_st (vector bool char, int, unsigned char *);
11881 void vec_st (vector bool char, int, signed char *);
11883 void vec_ste (vector signed char, int, signed char *);
11884 void vec_ste (vector unsigned char, int, unsigned char *);
11885 void vec_ste (vector bool char, int, signed char *);
11886 void vec_ste (vector bool char, int, unsigned char *);
11887 void vec_ste (vector signed short, int, short *);
11888 void vec_ste (vector unsigned short, int, unsigned short *);
11889 void vec_ste (vector bool short, int, short *);
11890 void vec_ste (vector bool short, int, unsigned short *);
11891 void vec_ste (vector pixel, int, short *);
11892 void vec_ste (vector pixel, int, unsigned short *);
11893 void vec_ste (vector float, int, float *);
11894 void vec_ste (vector signed int, int, int *);
11895 void vec_ste (vector unsigned int, int, unsigned int *);
11896 void vec_ste (vector bool int, int, int *);
11897 void vec_ste (vector bool int, int, unsigned int *);
11899 void vec_stvewx (vector float, int, float *);
11900 void vec_stvewx (vector signed int, int, int *);
11901 void vec_stvewx (vector unsigned int, int, unsigned int *);
11902 void vec_stvewx (vector bool int, int, int *);
11903 void vec_stvewx (vector bool int, int, unsigned int *);
11905 void vec_stvehx (vector signed short, int, short *);
11906 void vec_stvehx (vector unsigned short, int, unsigned short *);
11907 void vec_stvehx (vector bool short, int, short *);
11908 void vec_stvehx (vector bool short, int, unsigned short *);
11909 void vec_stvehx (vector pixel, int, short *);
11910 void vec_stvehx (vector pixel, int, unsigned short *);
11912 void vec_stvebx (vector signed char, int, signed char *);
11913 void vec_stvebx (vector unsigned char, int, unsigned char *);
11914 void vec_stvebx (vector bool char, int, signed char *);
11915 void vec_stvebx (vector bool char, int, unsigned char *);
11917 void vec_stl (vector float, int, vector float *);
11918 void vec_stl (vector float, int, float *);
11919 void vec_stl (vector signed int, int, vector signed int *);
11920 void vec_stl (vector signed int, int, int *);
11921 void vec_stl (vector unsigned int, int, vector unsigned int *);
11922 void vec_stl (vector unsigned int, int, unsigned int *);
11923 void vec_stl (vector bool int, int, vector bool int *);
11924 void vec_stl (vector bool int, int, unsigned int *);
11925 void vec_stl (vector bool int, int, int *);
11926 void vec_stl (vector signed short, int, vector signed short *);
11927 void vec_stl (vector signed short, int, short *);
11928 void vec_stl (vector unsigned short, int, vector unsigned short *);
11929 void vec_stl (vector unsigned short, int, unsigned short *);
11930 void vec_stl (vector bool short, int, vector bool short *);
11931 void vec_stl (vector bool short, int, unsigned short *);
11932 void vec_stl (vector bool short, int, short *);
11933 void vec_stl (vector pixel, int, vector pixel *);
11934 void vec_stl (vector pixel, int, unsigned short *);
11935 void vec_stl (vector pixel, int, short *);
11936 void vec_stl (vector signed char, int, vector signed char *);
11937 void vec_stl (vector signed char, int, signed char *);
11938 void vec_stl (vector unsigned char, int, vector unsigned char *);
11939 void vec_stl (vector unsigned char, int, unsigned char *);
11940 void vec_stl (vector bool char, int, vector bool char *);
11941 void vec_stl (vector bool char, int, unsigned char *);
11942 void vec_stl (vector bool char, int, signed char *);
11944 vector signed char vec_sub (vector bool char, vector signed char);
11945 vector signed char vec_sub (vector signed char, vector bool char);
11946 vector signed char vec_sub (vector signed char, vector signed char);
11947 vector unsigned char vec_sub (vector bool char, vector unsigned char);
11948 vector unsigned char vec_sub (vector unsigned char, vector bool char);
11949 vector unsigned char vec_sub (vector unsigned char,
11950 vector unsigned char);
11951 vector signed short vec_sub (vector bool short, vector signed short);
11952 vector signed short vec_sub (vector signed short, vector bool short);
11953 vector signed short vec_sub (vector signed short, vector signed short);
11954 vector unsigned short vec_sub (vector bool short,
11955 vector unsigned short);
11956 vector unsigned short vec_sub (vector unsigned short,
11957 vector bool short);
11958 vector unsigned short vec_sub (vector unsigned short,
11959 vector unsigned short);
11960 vector signed int vec_sub (vector bool int, vector signed int);
11961 vector signed int vec_sub (vector signed int, vector bool int);
11962 vector signed int vec_sub (vector signed int, vector signed int);
11963 vector unsigned int vec_sub (vector bool int, vector unsigned int);
11964 vector unsigned int vec_sub (vector unsigned int, vector bool int);
11965 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
11966 vector float vec_sub (vector float, vector float);
11968 vector float vec_vsubfp (vector float, vector float);
11970 vector signed int vec_vsubuwm (vector bool int, vector signed int);
11971 vector signed int vec_vsubuwm (vector signed int, vector bool int);
11972 vector signed int vec_vsubuwm (vector signed int, vector signed int);
11973 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
11974 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
11975 vector unsigned int vec_vsubuwm (vector unsigned int,
11976 vector unsigned int);
11978 vector signed short vec_vsubuhm (vector bool short,
11979 vector signed short);
11980 vector signed short vec_vsubuhm (vector signed short,
11981 vector bool short);
11982 vector signed short vec_vsubuhm (vector signed short,
11983 vector signed short);
11984 vector unsigned short vec_vsubuhm (vector bool short,
11985 vector unsigned short);
11986 vector unsigned short vec_vsubuhm (vector unsigned short,
11987 vector bool short);
11988 vector unsigned short vec_vsubuhm (vector unsigned short,
11989 vector unsigned short);
11991 vector signed char vec_vsububm (vector bool char, vector signed char);
11992 vector signed char vec_vsububm (vector signed char, vector bool char);
11993 vector signed char vec_vsububm (vector signed char, vector signed char);
11994 vector unsigned char vec_vsububm (vector bool char,
11995 vector unsigned char);
11996 vector unsigned char vec_vsububm (vector unsigned char,
11998 vector unsigned char vec_vsububm (vector unsigned char,
11999 vector unsigned char);
12001 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
12003 vector unsigned char vec_subs (vector bool char, vector unsigned char);
12004 vector unsigned char vec_subs (vector unsigned char, vector bool char);
12005 vector unsigned char vec_subs (vector unsigned char,
12006 vector unsigned char);
12007 vector signed char vec_subs (vector bool char, vector signed char);
12008 vector signed char vec_subs (vector signed char, vector bool char);
12009 vector signed char vec_subs (vector signed char, vector signed char);
12010 vector unsigned short vec_subs (vector bool short,
12011 vector unsigned short);
12012 vector unsigned short vec_subs (vector unsigned short,
12013 vector bool short);
12014 vector unsigned short vec_subs (vector unsigned short,
12015 vector unsigned short);
12016 vector signed short vec_subs (vector bool short, vector signed short);
12017 vector signed short vec_subs (vector signed short, vector bool short);
12018 vector signed short vec_subs (vector signed short, vector signed short);
12019 vector unsigned int vec_subs (vector bool int, vector unsigned int);
12020 vector unsigned int vec_subs (vector unsigned int, vector bool int);
12021 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
12022 vector signed int vec_subs (vector bool int, vector signed int);
12023 vector signed int vec_subs (vector signed int, vector bool int);
12024 vector signed int vec_subs (vector signed int, vector signed int);
12026 vector signed int vec_vsubsws (vector bool int, vector signed int);
12027 vector signed int vec_vsubsws (vector signed int, vector bool int);
12028 vector signed int vec_vsubsws (vector signed int, vector signed int);
12030 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
12031 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
12032 vector unsigned int vec_vsubuws (vector unsigned int,
12033 vector unsigned int);
12035 vector signed short vec_vsubshs (vector bool short,
12036 vector signed short);
12037 vector signed short vec_vsubshs (vector signed short,
12038 vector bool short);
12039 vector signed short vec_vsubshs (vector signed short,
12040 vector signed short);
12042 vector unsigned short vec_vsubuhs (vector bool short,
12043 vector unsigned short);
12044 vector unsigned short vec_vsubuhs (vector unsigned short,
12045 vector bool short);
12046 vector unsigned short vec_vsubuhs (vector unsigned short,
12047 vector unsigned short);
12049 vector signed char vec_vsubsbs (vector bool char, vector signed char);
12050 vector signed char vec_vsubsbs (vector signed char, vector bool char);
12051 vector signed char vec_vsubsbs (vector signed char, vector signed char);
12053 vector unsigned char vec_vsububs (vector bool char,
12054 vector unsigned char);
12055 vector unsigned char vec_vsububs (vector unsigned char,
12057 vector unsigned char vec_vsububs (vector unsigned char,
12058 vector unsigned char);
12060 vector unsigned int vec_sum4s (vector unsigned char,
12061 vector unsigned int);
12062 vector signed int vec_sum4s (vector signed char, vector signed int);
12063 vector signed int vec_sum4s (vector signed short, vector signed int);
12065 vector signed int vec_vsum4shs (vector signed short, vector signed int);
12067 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
12069 vector unsigned int vec_vsum4ubs (vector unsigned char,
12070 vector unsigned int);
12072 vector signed int vec_sum2s (vector signed int, vector signed int);
12074 vector signed int vec_sums (vector signed int, vector signed int);
12076 vector float vec_trunc (vector float);
12078 vector signed short vec_unpackh (vector signed char);
12079 vector bool short vec_unpackh (vector bool char);
12080 vector signed int vec_unpackh (vector signed short);
12081 vector bool int vec_unpackh (vector bool short);
12082 vector unsigned int vec_unpackh (vector pixel);
12084 vector bool int vec_vupkhsh (vector bool short);
12085 vector signed int vec_vupkhsh (vector signed short);
12087 vector unsigned int vec_vupkhpx (vector pixel);
12089 vector bool short vec_vupkhsb (vector bool char);
12090 vector signed short vec_vupkhsb (vector signed char);
12092 vector signed short vec_unpackl (vector signed char);
12093 vector bool short vec_unpackl (vector bool char);
12094 vector unsigned int vec_unpackl (vector pixel);
12095 vector signed int vec_unpackl (vector signed short);
12096 vector bool int vec_unpackl (vector bool short);
12098 vector unsigned int vec_vupklpx (vector pixel);
12100 vector bool int vec_vupklsh (vector bool short);
12101 vector signed int vec_vupklsh (vector signed short);
12103 vector bool short vec_vupklsb (vector bool char);
12104 vector signed short vec_vupklsb (vector signed char);
12106 vector float vec_xor (vector float, vector float);
12107 vector float vec_xor (vector float, vector bool int);
12108 vector float vec_xor (vector bool int, vector float);
12109 vector bool int vec_xor (vector bool int, vector bool int);
12110 vector signed int vec_xor (vector bool int, vector signed int);
12111 vector signed int vec_xor (vector signed int, vector bool int);
12112 vector signed int vec_xor (vector signed int, vector signed int);
12113 vector unsigned int vec_xor (vector bool int, vector unsigned int);
12114 vector unsigned int vec_xor (vector unsigned int, vector bool int);
12115 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
12116 vector bool short vec_xor (vector bool short, vector bool short);
12117 vector signed short vec_xor (vector bool short, vector signed short);
12118 vector signed short vec_xor (vector signed short, vector bool short);
12119 vector signed short vec_xor (vector signed short, vector signed short);
12120 vector unsigned short vec_xor (vector bool short,
12121 vector unsigned short);
12122 vector unsigned short vec_xor (vector unsigned short,
12123 vector bool short);
12124 vector unsigned short vec_xor (vector unsigned short,
12125 vector unsigned short);
12126 vector signed char vec_xor (vector bool char, vector signed char);
12127 vector bool char vec_xor (vector bool char, vector bool char);
12128 vector signed char vec_xor (vector signed char, vector bool char);
12129 vector signed char vec_xor (vector signed char, vector signed char);
12130 vector unsigned char vec_xor (vector bool char, vector unsigned char);
12131 vector unsigned char vec_xor (vector unsigned char, vector bool char);
12132 vector unsigned char vec_xor (vector unsigned char,
12133 vector unsigned char);
12135 int vec_all_eq (vector signed char, vector bool char);
12136 int vec_all_eq (vector signed char, vector signed char);
12137 int vec_all_eq (vector unsigned char, vector bool char);
12138 int vec_all_eq (vector unsigned char, vector unsigned char);
12139 int vec_all_eq (vector bool char, vector bool char);
12140 int vec_all_eq (vector bool char, vector unsigned char);
12141 int vec_all_eq (vector bool char, vector signed char);
12142 int vec_all_eq (vector signed short, vector bool short);
12143 int vec_all_eq (vector signed short, vector signed short);
12144 int vec_all_eq (vector unsigned short, vector bool short);
12145 int vec_all_eq (vector unsigned short, vector unsigned short);
12146 int vec_all_eq (vector bool short, vector bool short);
12147 int vec_all_eq (vector bool short, vector unsigned short);
12148 int vec_all_eq (vector bool short, vector signed short);
12149 int vec_all_eq (vector pixel, vector pixel);
12150 int vec_all_eq (vector signed int, vector bool int);
12151 int vec_all_eq (vector signed int, vector signed int);
12152 int vec_all_eq (vector unsigned int, vector bool int);
12153 int vec_all_eq (vector unsigned int, vector unsigned int);
12154 int vec_all_eq (vector bool int, vector bool int);
12155 int vec_all_eq (vector bool int, vector unsigned int);
12156 int vec_all_eq (vector bool int, vector signed int);
12157 int vec_all_eq (vector float, vector float);
12159 int vec_all_ge (vector bool char, vector unsigned char);
12160 int vec_all_ge (vector unsigned char, vector bool char);
12161 int vec_all_ge (vector unsigned char, vector unsigned char);
12162 int vec_all_ge (vector bool char, vector signed char);
12163 int vec_all_ge (vector signed char, vector bool char);
12164 int vec_all_ge (vector signed char, vector signed char);
12165 int vec_all_ge (vector bool short, vector unsigned short);
12166 int vec_all_ge (vector unsigned short, vector bool short);
12167 int vec_all_ge (vector unsigned short, vector unsigned short);
12168 int vec_all_ge (vector signed short, vector signed short);
12169 int vec_all_ge (vector bool short, vector signed short);
12170 int vec_all_ge (vector signed short, vector bool short);
12171 int vec_all_ge (vector bool int, vector unsigned int);
12172 int vec_all_ge (vector unsigned int, vector bool int);
12173 int vec_all_ge (vector unsigned int, vector unsigned int);
12174 int vec_all_ge (vector bool int, vector signed int);
12175 int vec_all_ge (vector signed int, vector bool int);
12176 int vec_all_ge (vector signed int, vector signed int);
12177 int vec_all_ge (vector float, vector float);
12179 int vec_all_gt (vector bool char, vector unsigned char);
12180 int vec_all_gt (vector unsigned char, vector bool char);
12181 int vec_all_gt (vector unsigned char, vector unsigned char);
12182 int vec_all_gt (vector bool char, vector signed char);
12183 int vec_all_gt (vector signed char, vector bool char);
12184 int vec_all_gt (vector signed char, vector signed char);
12185 int vec_all_gt (vector bool short, vector unsigned short);
12186 int vec_all_gt (vector unsigned short, vector bool short);
12187 int vec_all_gt (vector unsigned short, vector unsigned short);
12188 int vec_all_gt (vector bool short, vector signed short);
12189 int vec_all_gt (vector signed short, vector bool short);
12190 int vec_all_gt (vector signed short, vector signed short);
12191 int vec_all_gt (vector bool int, vector unsigned int);
12192 int vec_all_gt (vector unsigned int, vector bool int);
12193 int vec_all_gt (vector unsigned int, vector unsigned int);
12194 int vec_all_gt (vector bool int, vector signed int);
12195 int vec_all_gt (vector signed int, vector bool int);
12196 int vec_all_gt (vector signed int, vector signed int);
12197 int vec_all_gt (vector float, vector float);
12199 int vec_all_in (vector float, vector float);
12201 int vec_all_le (vector bool char, vector unsigned char);
12202 int vec_all_le (vector unsigned char, vector bool char);
12203 int vec_all_le (vector unsigned char, vector unsigned char);
12204 int vec_all_le (vector bool char, vector signed char);
12205 int vec_all_le (vector signed char, vector bool char);
12206 int vec_all_le (vector signed char, vector signed char);
12207 int vec_all_le (vector bool short, vector unsigned short);
12208 int vec_all_le (vector unsigned short, vector bool short);
12209 int vec_all_le (vector unsigned short, vector unsigned short);
12210 int vec_all_le (vector bool short, vector signed short);
12211 int vec_all_le (vector signed short, vector bool short);
12212 int vec_all_le (vector signed short, vector signed short);
12213 int vec_all_le (vector bool int, vector unsigned int);
12214 int vec_all_le (vector unsigned int, vector bool int);
12215 int vec_all_le (vector unsigned int, vector unsigned int);
12216 int vec_all_le (vector bool int, vector signed int);
12217 int vec_all_le (vector signed int, vector bool int);
12218 int vec_all_le (vector signed int, vector signed int);
12219 int vec_all_le (vector float, vector float);
12221 int vec_all_lt (vector bool char, vector unsigned char);
12222 int vec_all_lt (vector unsigned char, vector bool char);
12223 int vec_all_lt (vector unsigned char, vector unsigned char);
12224 int vec_all_lt (vector bool char, vector signed char);
12225 int vec_all_lt (vector signed char, vector bool char);
12226 int vec_all_lt (vector signed char, vector signed char);
12227 int vec_all_lt (vector bool short, vector unsigned short);
12228 int vec_all_lt (vector unsigned short, vector bool short);
12229 int vec_all_lt (vector unsigned short, vector unsigned short);
12230 int vec_all_lt (vector bool short, vector signed short);
12231 int vec_all_lt (vector signed short, vector bool short);
12232 int vec_all_lt (vector signed short, vector signed short);
12233 int vec_all_lt (vector bool int, vector unsigned int);
12234 int vec_all_lt (vector unsigned int, vector bool int);
12235 int vec_all_lt (vector unsigned int, vector unsigned int);
12236 int vec_all_lt (vector bool int, vector signed int);
12237 int vec_all_lt (vector signed int, vector bool int);
12238 int vec_all_lt (vector signed int, vector signed int);
12239 int vec_all_lt (vector float, vector float);
12241 int vec_all_nan (vector float);
12243 int vec_all_ne (vector signed char, vector bool char);
12244 int vec_all_ne (vector signed char, vector signed char);
12245 int vec_all_ne (vector unsigned char, vector bool char);
12246 int vec_all_ne (vector unsigned char, vector unsigned char);
12247 int vec_all_ne (vector bool char, vector bool char);
12248 int vec_all_ne (vector bool char, vector unsigned char);
12249 int vec_all_ne (vector bool char, vector signed char);
12250 int vec_all_ne (vector signed short, vector bool short);
12251 int vec_all_ne (vector signed short, vector signed short);
12252 int vec_all_ne (vector unsigned short, vector bool short);
12253 int vec_all_ne (vector unsigned short, vector unsigned short);
12254 int vec_all_ne (vector bool short, vector bool short);
12255 int vec_all_ne (vector bool short, vector unsigned short);
12256 int vec_all_ne (vector bool short, vector signed short);
12257 int vec_all_ne (vector pixel, vector pixel);
12258 int vec_all_ne (vector signed int, vector bool int);
12259 int vec_all_ne (vector signed int, vector signed int);
12260 int vec_all_ne (vector unsigned int, vector bool int);
12261 int vec_all_ne (vector unsigned int, vector unsigned int);
12262 int vec_all_ne (vector bool int, vector bool int);
12263 int vec_all_ne (vector bool int, vector unsigned int);
12264 int vec_all_ne (vector bool int, vector signed int);
12265 int vec_all_ne (vector float, vector float);
12267 int vec_all_nge (vector float, vector float);
12269 int vec_all_ngt (vector float, vector float);
12271 int vec_all_nle (vector float, vector float);
12273 int vec_all_nlt (vector float, vector float);
12275 int vec_all_numeric (vector float);
12277 int vec_any_eq (vector signed char, vector bool char);
12278 int vec_any_eq (vector signed char, vector signed char);
12279 int vec_any_eq (vector unsigned char, vector bool char);
12280 int vec_any_eq (vector unsigned char, vector unsigned char);
12281 int vec_any_eq (vector bool char, vector bool char);
12282 int vec_any_eq (vector bool char, vector unsigned char);
12283 int vec_any_eq (vector bool char, vector signed char);
12284 int vec_any_eq (vector signed short, vector bool short);
12285 int vec_any_eq (vector signed short, vector signed short);
12286 int vec_any_eq (vector unsigned short, vector bool short);
12287 int vec_any_eq (vector unsigned short, vector unsigned short);
12288 int vec_any_eq (vector bool short, vector bool short);
12289 int vec_any_eq (vector bool short, vector unsigned short);
12290 int vec_any_eq (vector bool short, vector signed short);
12291 int vec_any_eq (vector pixel, vector pixel);
12292 int vec_any_eq (vector signed int, vector bool int);
12293 int vec_any_eq (vector signed int, vector signed int);
12294 int vec_any_eq (vector unsigned int, vector bool int);
12295 int vec_any_eq (vector unsigned int, vector unsigned int);
12296 int vec_any_eq (vector bool int, vector bool int);
12297 int vec_any_eq (vector bool int, vector unsigned int);
12298 int vec_any_eq (vector bool int, vector signed int);
12299 int vec_any_eq (vector float, vector float);
12301 int vec_any_ge (vector signed char, vector bool char);
12302 int vec_any_ge (vector unsigned char, vector bool char);
12303 int vec_any_ge (vector unsigned char, vector unsigned char);
12304 int vec_any_ge (vector signed char, vector signed char);
12305 int vec_any_ge (vector bool char, vector unsigned char);
12306 int vec_any_ge (vector bool char, vector signed char);
12307 int vec_any_ge (vector unsigned short, vector bool short);
12308 int vec_any_ge (vector unsigned short, vector unsigned short);
12309 int vec_any_ge (vector signed short, vector signed short);
12310 int vec_any_ge (vector signed short, vector bool short);
12311 int vec_any_ge (vector bool short, vector unsigned short);
12312 int vec_any_ge (vector bool short, vector signed short);
12313 int vec_any_ge (vector signed int, vector bool int);
12314 int vec_any_ge (vector unsigned int, vector bool int);
12315 int vec_any_ge (vector unsigned int, vector unsigned int);
12316 int vec_any_ge (vector signed int, vector signed int);
12317 int vec_any_ge (vector bool int, vector unsigned int);
12318 int vec_any_ge (vector bool int, vector signed int);
12319 int vec_any_ge (vector float, vector float);
12321 int vec_any_gt (vector bool char, vector unsigned char);
12322 int vec_any_gt (vector unsigned char, vector bool char);
12323 int vec_any_gt (vector unsigned char, vector unsigned char);
12324 int vec_any_gt (vector bool char, vector signed char);
12325 int vec_any_gt (vector signed char, vector bool char);
12326 int vec_any_gt (vector signed char, vector signed char);
12327 int vec_any_gt (vector bool short, vector unsigned short);
12328 int vec_any_gt (vector unsigned short, vector bool short);
12329 int vec_any_gt (vector unsigned short, vector unsigned short);
12330 int vec_any_gt (vector bool short, vector signed short);
12331 int vec_any_gt (vector signed short, vector bool short);
12332 int vec_any_gt (vector signed short, vector signed short);
12333 int vec_any_gt (vector bool int, vector unsigned int);
12334 int vec_any_gt (vector unsigned int, vector bool int);
12335 int vec_any_gt (vector unsigned int, vector unsigned int);
12336 int vec_any_gt (vector bool int, vector signed int);
12337 int vec_any_gt (vector signed int, vector bool int);
12338 int vec_any_gt (vector signed int, vector signed int);
12339 int vec_any_gt (vector float, vector float);
12341 int vec_any_le (vector bool char, vector unsigned char);
12342 int vec_any_le (vector unsigned char, vector bool char);
12343 int vec_any_le (vector unsigned char, vector unsigned char);
12344 int vec_any_le (vector bool char, vector signed char);
12345 int vec_any_le (vector signed char, vector bool char);
12346 int vec_any_le (vector signed char, vector signed char);
12347 int vec_any_le (vector bool short, vector unsigned short);
12348 int vec_any_le (vector unsigned short, vector bool short);
12349 int vec_any_le (vector unsigned short, vector unsigned short);
12350 int vec_any_le (vector bool short, vector signed short);
12351 int vec_any_le (vector signed short, vector bool short);
12352 int vec_any_le (vector signed short, vector signed short);
12353 int vec_any_le (vector bool int, vector unsigned int);
12354 int vec_any_le (vector unsigned int, vector bool int);
12355 int vec_any_le (vector unsigned int, vector unsigned int);
12356 int vec_any_le (vector bool int, vector signed int);
12357 int vec_any_le (vector signed int, vector bool int);
12358 int vec_any_le (vector signed int, vector signed int);
12359 int vec_any_le (vector float, vector float);
12361 int vec_any_lt (vector bool char, vector unsigned char);
12362 int vec_any_lt (vector unsigned char, vector bool char);
12363 int vec_any_lt (vector unsigned char, vector unsigned char);
12364 int vec_any_lt (vector bool char, vector signed char);
12365 int vec_any_lt (vector signed char, vector bool char);
12366 int vec_any_lt (vector signed char, vector signed char);
12367 int vec_any_lt (vector bool short, vector unsigned short);
12368 int vec_any_lt (vector unsigned short, vector bool short);
12369 int vec_any_lt (vector unsigned short, vector unsigned short);
12370 int vec_any_lt (vector bool short, vector signed short);
12371 int vec_any_lt (vector signed short, vector bool short);
12372 int vec_any_lt (vector signed short, vector signed short);
12373 int vec_any_lt (vector bool int, vector unsigned int);
12374 int vec_any_lt (vector unsigned int, vector bool int);
12375 int vec_any_lt (vector unsigned int, vector unsigned int);
12376 int vec_any_lt (vector bool int, vector signed int);
12377 int vec_any_lt (vector signed int, vector bool int);
12378 int vec_any_lt (vector signed int, vector signed int);
12379 int vec_any_lt (vector float, vector float);
12381 int vec_any_nan (vector float);
12383 int vec_any_ne (vector signed char, vector bool char);
12384 int vec_any_ne (vector signed char, vector signed char);
12385 int vec_any_ne (vector unsigned char, vector bool char);
12386 int vec_any_ne (vector unsigned char, vector unsigned char);
12387 int vec_any_ne (vector bool char, vector bool char);
12388 int vec_any_ne (vector bool char, vector unsigned char);
12389 int vec_any_ne (vector bool char, vector signed char);
12390 int vec_any_ne (vector signed short, vector bool short);
12391 int vec_any_ne (vector signed short, vector signed short);
12392 int vec_any_ne (vector unsigned short, vector bool short);
12393 int vec_any_ne (vector unsigned short, vector unsigned short);
12394 int vec_any_ne (vector bool short, vector bool short);
12395 int vec_any_ne (vector bool short, vector unsigned short);
12396 int vec_any_ne (vector bool short, vector signed short);
12397 int vec_any_ne (vector pixel, vector pixel);
12398 int vec_any_ne (vector signed int, vector bool int);
12399 int vec_any_ne (vector signed int, vector signed int);
12400 int vec_any_ne (vector unsigned int, vector bool int);
12401 int vec_any_ne (vector unsigned int, vector unsigned int);
12402 int vec_any_ne (vector bool int, vector bool int);
12403 int vec_any_ne (vector bool int, vector unsigned int);
12404 int vec_any_ne (vector bool int, vector signed int);
12405 int vec_any_ne (vector float, vector float);
12407 int vec_any_nge (vector float, vector float);
12409 int vec_any_ngt (vector float, vector float);
12411 int vec_any_nle (vector float, vector float);
12413 int vec_any_nlt (vector float, vector float);
12415 int vec_any_numeric (vector float);
12417 int vec_any_out (vector float, vector float);
12420 If the vector/scalar (VSX) instruction set is available, the following
12421 additional functions are available:
12424 vector double vec_abs (vector double);
12425 vector double vec_add (vector double, vector double);
12426 vector double vec_and (vector double, vector double);
12427 vector double vec_and (vector double, vector bool long);
12428 vector double vec_and (vector bool long, vector double);
12429 vector double vec_andc (vector double, vector double);
12430 vector double vec_andc (vector double, vector bool long);
12431 vector double vec_andc (vector bool long, vector double);
12432 vector double vec_ceil (vector double);
12433 vector bool long vec_cmpeq (vector double, vector double);
12434 vector bool long vec_cmpge (vector double, vector double);
12435 vector bool long vec_cmpgt (vector double, vector double);
12436 vector bool long vec_cmple (vector double, vector double);
12437 vector bool long vec_cmplt (vector double, vector double);
12438 vector float vec_div (vector float, vector float);
12439 vector double vec_div (vector double, vector double);
12440 vector double vec_floor (vector double);
12441 vector double vec_ld (int, const vector double *);
12442 vector double vec_ld (int, const double *);
12443 vector double vec_ldl (int, const vector double *);
12444 vector double vec_ldl (int, const double *);
12445 vector unsigned char vec_lvsl (int, const volatile double *);
12446 vector unsigned char vec_lvsr (int, const volatile double *);
12447 vector double vec_madd (vector double, vector double, vector double);
12448 vector double vec_max (vector double, vector double);
12449 vector double vec_min (vector double, vector double);
12450 vector float vec_msub (vector float, vector float, vector float);
12451 vector double vec_msub (vector double, vector double, vector double);
12452 vector float vec_mul (vector float, vector float);
12453 vector double vec_mul (vector double, vector double);
12454 vector float vec_nearbyint (vector float);
12455 vector double vec_nearbyint (vector double);
12456 vector float vec_nmadd (vector float, vector float, vector float);
12457 vector double vec_nmadd (vector double, vector double, vector double);
12458 vector double vec_nmsub (vector double, vector double, vector double);
12459 vector double vec_nor (vector double, vector double);
12460 vector double vec_or (vector double, vector double);
12461 vector double vec_or (vector double, vector bool long);
12462 vector double vec_or (vector bool long, vector double);
12463 vector double vec_perm (vector double,
12465 vector unsigned char);
12466 vector double vec_rint (vector double);
12467 vector double vec_recip (vector double, vector double);
12468 vector double vec_rsqrt (vector double);
12469 vector double vec_rsqrte (vector double);
12470 vector double vec_sel (vector double, vector double, vector bool long);
12471 vector double vec_sel (vector double, vector double, vector unsigned long);
12472 vector double vec_sub (vector double, vector double);
12473 vector float vec_sqrt (vector float);
12474 vector double vec_sqrt (vector double);
12475 void vec_st (vector double, int, vector double *);
12476 void vec_st (vector double, int, double *);
12477 vector double vec_trunc (vector double);
12478 vector double vec_xor (vector double, vector double);
12479 vector double vec_xor (vector double, vector bool long);
12480 vector double vec_xor (vector bool long, vector double);
12481 int vec_all_eq (vector double, vector double);
12482 int vec_all_ge (vector double, vector double);
12483 int vec_all_gt (vector double, vector double);
12484 int vec_all_le (vector double, vector double);
12485 int vec_all_lt (vector double, vector double);
12486 int vec_all_nan (vector double);
12487 int vec_all_ne (vector double, vector double);
12488 int vec_all_nge (vector double, vector double);
12489 int vec_all_ngt (vector double, vector double);
12490 int vec_all_nle (vector double, vector double);
12491 int vec_all_nlt (vector double, vector double);
12492 int vec_all_numeric (vector double);
12493 int vec_any_eq (vector double, vector double);
12494 int vec_any_ge (vector double, vector double);
12495 int vec_any_gt (vector double, vector double);
12496 int vec_any_le (vector double, vector double);
12497 int vec_any_lt (vector double, vector double);
12498 int vec_any_nan (vector double);
12499 int vec_any_ne (vector double, vector double);
12500 int vec_any_nge (vector double, vector double);
12501 int vec_any_ngt (vector double, vector double);
12502 int vec_any_nle (vector double, vector double);
12503 int vec_any_nlt (vector double, vector double);
12504 int vec_any_numeric (vector double);
12506 vector double vec_vsx_ld (int, const vector double *);
12507 vector double vec_vsx_ld (int, const double *);
12508 vector float vec_vsx_ld (int, const vector float *);
12509 vector float vec_vsx_ld (int, const float *);
12510 vector bool int vec_vsx_ld (int, const vector bool int *);
12511 vector signed int vec_vsx_ld (int, const vector signed int *);
12512 vector signed int vec_vsx_ld (int, const int *);
12513 vector signed int vec_vsx_ld (int, const long *);
12514 vector unsigned int vec_vsx_ld (int, const vector unsigned int *);
12515 vector unsigned int vec_vsx_ld (int, const unsigned int *);
12516 vector unsigned int vec_vsx_ld (int, const unsigned long *);
12517 vector bool short vec_vsx_ld (int, const vector bool short *);
12518 vector pixel vec_vsx_ld (int, const vector pixel *);
12519 vector signed short vec_vsx_ld (int, const vector signed short *);
12520 vector signed short vec_vsx_ld (int, const short *);
12521 vector unsigned short vec_vsx_ld (int, const vector unsigned short *);
12522 vector unsigned short vec_vsx_ld (int, const unsigned short *);
12523 vector bool char vec_vsx_ld (int, const vector bool char *);
12524 vector signed char vec_vsx_ld (int, const vector signed char *);
12525 vector signed char vec_vsx_ld (int, const signed char *);
12526 vector unsigned char vec_vsx_ld (int, const vector unsigned char *);
12527 vector unsigned char vec_vsx_ld (int, const unsigned char *);
12529 void vec_vsx_st (vector double, int, vector double *);
12530 void vec_vsx_st (vector double, int, double *);
12531 void vec_vsx_st (vector float, int, vector float *);
12532 void vec_vsx_st (vector float, int, float *);
12533 void vec_vsx_st (vector signed int, int, vector signed int *);
12534 void vec_vsx_st (vector signed int, int, int *);
12535 void vec_vsx_st (vector unsigned int, int, vector unsigned int *);
12536 void vec_vsx_st (vector unsigned int, int, unsigned int *);
12537 void vec_vsx_st (vector bool int, int, vector bool int *);
12538 void vec_vsx_st (vector bool int, int, unsigned int *);
12539 void vec_vsx_st (vector bool int, int, int *);
12540 void vec_vsx_st (vector signed short, int, vector signed short *);
12541 void vec_vsx_st (vector signed short, int, short *);
12542 void vec_vsx_st (vector unsigned short, int, vector unsigned short *);
12543 void vec_vsx_st (vector unsigned short, int, unsigned short *);
12544 void vec_vsx_st (vector bool short, int, vector bool short *);
12545 void vec_vsx_st (vector bool short, int, unsigned short *);
12546 void vec_vsx_st (vector pixel, int, vector pixel *);
12547 void vec_vsx_st (vector pixel, int, unsigned short *);
12548 void vec_vsx_st (vector pixel, int, short *);
12549 void vec_vsx_st (vector bool short, int, short *);
12550 void vec_vsx_st (vector signed char, int, vector signed char *);
12551 void vec_vsx_st (vector signed char, int, signed char *);
12552 void vec_vsx_st (vector unsigned char, int, vector unsigned char *);
12553 void vec_vsx_st (vector unsigned char, int, unsigned char *);
12554 void vec_vsx_st (vector bool char, int, vector bool char *);
12555 void vec_vsx_st (vector bool char, int, unsigned char *);
12556 void vec_vsx_st (vector bool char, int, signed char *);
12559 Note that the @samp{vec_ld} and @samp{vec_st} builtins will always
12560 generate the Altivec @samp{LVX} and @samp{STVX} instructions even
12561 if the VSX instruction set is available. The @samp{vec_vsx_ld} and
12562 @samp{vec_vsx_st} builtins will always generate the VSX @samp{LXVD2X},
12563 @samp{LXVW4X}, @samp{STXVD2X}, and @samp{STXVW4X} instructions.
12565 GCC provides a few other builtins on Powerpc to access certain instructions:
12567 float __builtin_recipdivf (float, float);
12568 float __builtin_rsqrtf (float);
12569 double __builtin_recipdiv (double, double);
12570 double __builtin_rsqrt (double);
12571 long __builtin_bpermd (long, long);
12572 int __builtin_bswap16 (int);
12575 The @code{vec_rsqrt}, @code{__builtin_rsqrt}, and
12576 @code{__builtin_rsqrtf} functions generate multiple instructions to
12577 implement the reciprocal sqrt functionality using reciprocal sqrt
12578 estimate instructions.
12580 The @code{__builtin_recipdiv}, and @code{__builtin_recipdivf}
12581 functions generate multiple instructions to implement division using
12582 the reciprocal estimate instructions.
12584 @node RX Built-in Functions
12585 @subsection RX Built-in Functions
12586 GCC supports some of the RX instructions which cannot be expressed in
12587 the C programming language via the use of built-in functions. The
12588 following functions are supported:
12590 @deftypefn {Built-in Function} void __builtin_rx_brk (void)
12591 Generates the @code{brk} machine instruction.
12594 @deftypefn {Built-in Function} void __builtin_rx_clrpsw (int)
12595 Generates the @code{clrpsw} machine instruction to clear the specified
12596 bit in the processor status word.
12599 @deftypefn {Built-in Function} void __builtin_rx_int (int)
12600 Generates the @code{int} machine instruction to generate an interrupt
12601 with the specified value.
12604 @deftypefn {Built-in Function} void __builtin_rx_machi (int, int)
12605 Generates the @code{machi} machine instruction to add the result of
12606 multiplying the top 16-bits of the two arguments into the
12610 @deftypefn {Built-in Function} void __builtin_rx_maclo (int, int)
12611 Generates the @code{maclo} machine instruction to add the result of
12612 multiplying the bottom 16-bits of the two arguments into the
12616 @deftypefn {Built-in Function} void __builtin_rx_mulhi (int, int)
12617 Generates the @code{mulhi} machine instruction to place the result of
12618 multiplying the top 16-bits of the two arguments into the
12622 @deftypefn {Built-in Function} void __builtin_rx_mullo (int, int)
12623 Generates the @code{mullo} machine instruction to place the result of
12624 multiplying the bottom 16-bits of the two arguments into the
12628 @deftypefn {Built-in Function} int __builtin_rx_mvfachi (void)
12629 Generates the @code{mvfachi} machine instruction to read the top
12630 32-bits of the accumulator.
12633 @deftypefn {Built-in Function} int __builtin_rx_mvfacmi (void)
12634 Generates the @code{mvfacmi} machine instruction to read the middle
12635 32-bits of the accumulator.
12638 @deftypefn {Built-in Function} int __builtin_rx_mvfc (int)
12639 Generates the @code{mvfc} machine instruction which reads the control
12640 register specified in its argument and returns its value.
12643 @deftypefn {Built-in Function} void __builtin_rx_mvtachi (int)
12644 Generates the @code{mvtachi} machine instruction to set the top
12645 32-bits of the accumulator.
12648 @deftypefn {Built-in Function} void __builtin_rx_mvtaclo (int)
12649 Generates the @code{mvtaclo} machine instruction to set the bottom
12650 32-bits of the accumulator.
12653 @deftypefn {Built-in Function} void __builtin_rx_mvtc (int reg, int val)
12654 Generates the @code{mvtc} machine instruction which sets control
12655 register number @code{reg} to @code{val}.
12658 @deftypefn {Built-in Function} void __builtin_rx_mvtipl (int)
12659 Generates the @code{mvtipl} machine instruction set the interrupt
12663 @deftypefn {Built-in Function} void __builtin_rx_racw (int)
12664 Generates the @code{racw} machine instruction to round the accumulator
12665 according to the specified mode.
12668 @deftypefn {Built-in Function} int __builtin_rx_revw (int)
12669 Generates the @code{revw} machine instruction which swaps the bytes in
12670 the argument so that bits 0--7 now occupy bits 8--15 and vice versa,
12671 and also bits 16--23 occupy bits 24--31 and vice versa.
12674 @deftypefn {Built-in Function} void __builtin_rx_rmpa (void)
12675 Generates the @code{rmpa} machine instruction which initiates a
12676 repeated multiply and accumulate sequence.
12679 @deftypefn {Built-in Function} void __builtin_rx_round (float)
12680 Generates the @code{round} machine instruction which returns the
12681 floating point argument rounded according to the current rounding mode
12682 set in the floating point status word register.
12685 @deftypefn {Built-in Function} int __builtin_rx_sat (int)
12686 Generates the @code{sat} machine instruction which returns the
12687 saturated value of the argument.
12690 @deftypefn {Built-in Function} void __builtin_rx_setpsw (int)
12691 Generates the @code{setpsw} machine instruction to set the specified
12692 bit in the processor status word.
12695 @deftypefn {Built-in Function} void __builtin_rx_wait (void)
12696 Generates the @code{wait} machine instruction.
12699 @node SPARC VIS Built-in Functions
12700 @subsection SPARC VIS Built-in Functions
12702 GCC supports SIMD operations on the SPARC using both the generic vector
12703 extensions (@pxref{Vector Extensions}) as well as built-in functions for
12704 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
12705 switch, the VIS extension is exposed as the following built-in functions:
12708 typedef int v2si __attribute__ ((vector_size (8)));
12709 typedef short v4hi __attribute__ ((vector_size (8)));
12710 typedef short v2hi __attribute__ ((vector_size (4)));
12711 typedef char v8qi __attribute__ ((vector_size (8)));
12712 typedef char v4qi __attribute__ ((vector_size (4)));
12714 void * __builtin_vis_alignaddr (void *, long);
12715 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
12716 v2si __builtin_vis_faligndatav2si (v2si, v2si);
12717 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
12718 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
12720 v4hi __builtin_vis_fexpand (v4qi);
12722 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
12723 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
12724 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
12725 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
12726 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
12727 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
12728 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
12730 v4qi __builtin_vis_fpack16 (v4hi);
12731 v8qi __builtin_vis_fpack32 (v2si, v2si);
12732 v2hi __builtin_vis_fpackfix (v2si);
12733 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
12735 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
12738 @node SPU Built-in Functions
12739 @subsection SPU Built-in Functions
12741 GCC provides extensions for the SPU processor as described in the
12742 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
12743 found at @uref{http://cell.scei.co.jp/} or
12744 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
12745 implementation differs in several ways.
12750 The optional extension of specifying vector constants in parentheses is
12754 A vector initializer requires no cast if the vector constant is of the
12755 same type as the variable it is initializing.
12758 If @code{signed} or @code{unsigned} is omitted, the signedness of the
12759 vector type is the default signedness of the base type. The default
12760 varies depending on the operating system, so a portable program should
12761 always specify the signedness.
12764 By default, the keyword @code{__vector} is added. The macro
12765 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
12769 GCC allows using a @code{typedef} name as the type specifier for a
12773 For C, overloaded functions are implemented with macros so the following
12777 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
12780 Since @code{spu_add} is a macro, the vector constant in the example
12781 is treated as four separate arguments. Wrap the entire argument in
12782 parentheses for this to work.
12785 The extended version of @code{__builtin_expect} is not supported.
12789 @emph{Note:} Only the interface described in the aforementioned
12790 specification is supported. Internally, GCC uses built-in functions to
12791 implement the required functionality, but these are not supported and
12792 are subject to change without notice.
12794 @node TI C6X Built-in Functions
12795 @subsection TI C6X Built-in Functions
12797 GCC provides intrinsics to access certain instructions of the TI C6X
12798 processors. These intrinsics, listed below, are available after
12799 inclusion of the @code{c6x_intrinsics.h} header file. They map directly
12800 to C6X instructions.
12804 int _sadd (int, int)
12805 int _ssub (int, int)
12806 int _sadd2 (int, int)
12807 int _ssub2 (int, int)
12808 long long _mpy2 (int, int)
12809 long long _smpy2 (int, int)
12810 int _add4 (int, int)
12811 int _sub4 (int, int)
12812 int _saddu4 (int, int)
12814 int _smpy (int, int)
12815 int _smpyh (int, int)
12816 int _smpyhl (int, int)
12817 int _smpylh (int, int)
12819 int _sshl (int, int)
12820 int _subc (int, int)
12822 int _avg2 (int, int)
12823 int _avgu4 (int, int)
12825 int _clrr (int, int)
12826 int _extr (int, int)
12827 int _extru (int, int)
12833 @node Target Format Checks
12834 @section Format Checks Specific to Particular Target Machines
12836 For some target machines, GCC supports additional options to the
12838 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
12841 * Solaris Format Checks::
12842 * Darwin Format Checks::
12845 @node Solaris Format Checks
12846 @subsection Solaris Format Checks
12848 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
12849 check. @code{cmn_err} accepts a subset of the standard @code{printf}
12850 conversions, and the two-argument @code{%b} conversion for displaying
12851 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
12853 @node Darwin Format Checks
12854 @subsection Darwin Format Checks
12856 Darwin targets support the @code{CFString} (or @code{__CFString__}) in the format
12857 attribute context. Declarations made with such attribution will be parsed for correct syntax
12858 and format argument types. However, parsing of the format string itself is currently undefined
12859 and will not be carried out by this version of the compiler.
12861 Additionally, @code{CFStringRefs} (defined by the @code{CoreFoundation} headers) may
12862 also be used as format arguments. Note that the relevant headers are only likely to be
12863 available on Darwin (OSX) installations. On such installations, the XCode and system
12864 documentation provide descriptions of @code{CFString}, @code{CFStringRefs} and
12865 associated functions.
12868 @section Pragmas Accepted by GCC
12870 @cindex @code{#pragma}
12872 GCC supports several types of pragmas, primarily in order to compile
12873 code originally written for other compilers. Note that in general
12874 we do not recommend the use of pragmas; @xref{Function Attributes},
12875 for further explanation.
12881 * RS/6000 and PowerPC Pragmas::
12883 * Solaris Pragmas::
12884 * Symbol-Renaming Pragmas::
12885 * Structure-Packing Pragmas::
12887 * Diagnostic Pragmas::
12888 * Visibility Pragmas::
12889 * Push/Pop Macro Pragmas::
12890 * Function Specific Option Pragmas::
12894 @subsection ARM Pragmas
12896 The ARM target defines pragmas for controlling the default addition of
12897 @code{long_call} and @code{short_call} attributes to functions.
12898 @xref{Function Attributes}, for information about the effects of these
12903 @cindex pragma, long_calls
12904 Set all subsequent functions to have the @code{long_call} attribute.
12906 @item no_long_calls
12907 @cindex pragma, no_long_calls
12908 Set all subsequent functions to have the @code{short_call} attribute.
12910 @item long_calls_off
12911 @cindex pragma, long_calls_off
12912 Do not affect the @code{long_call} or @code{short_call} attributes of
12913 subsequent functions.
12917 @subsection M32C Pragmas
12920 @item GCC memregs @var{number}
12921 @cindex pragma, memregs
12922 Overrides the command-line option @code{-memregs=} for the current
12923 file. Use with care! This pragma must be before any function in the
12924 file, and mixing different memregs values in different objects may
12925 make them incompatible. This pragma is useful when a
12926 performance-critical function uses a memreg for temporary values,
12927 as it may allow you to reduce the number of memregs used.
12929 @item ADDRESS @var{name} @var{address}
12930 @cindex pragma, address
12931 For any declared symbols matching @var{name}, this does three things
12932 to that symbol: it forces the symbol to be located at the given
12933 address (a number), it forces the symbol to be volatile, and it
12934 changes the symbol's scope to be static. This pragma exists for
12935 compatibility with other compilers, but note that the common
12936 @code{1234H} numeric syntax is not supported (use @code{0x1234}
12940 #pragma ADDRESS port3 0x103
12947 @subsection MeP Pragmas
12951 @item custom io_volatile (on|off)
12952 @cindex pragma, custom io_volatile
12953 Overrides the command line option @code{-mio-volatile} for the current
12954 file. Note that for compatibility with future GCC releases, this
12955 option should only be used once before any @code{io} variables in each
12958 @item GCC coprocessor available @var{registers}
12959 @cindex pragma, coprocessor available
12960 Specifies which coprocessor registers are available to the register
12961 allocator. @var{registers} may be a single register, register range
12962 separated by ellipses, or comma-separated list of those. Example:
12965 #pragma GCC coprocessor available $c0...$c10, $c28
12968 @item GCC coprocessor call_saved @var{registers}
12969 @cindex pragma, coprocessor call_saved
12970 Specifies which coprocessor registers are to be saved and restored by
12971 any function using them. @var{registers} may be a single register,
12972 register range separated by ellipses, or comma-separated list of
12976 #pragma GCC coprocessor call_saved $c4...$c6, $c31
12979 @item GCC coprocessor subclass '(A|B|C|D)' = @var{registers}
12980 @cindex pragma, coprocessor subclass
12981 Creates and defines a register class. These register classes can be
12982 used by inline @code{asm} constructs. @var{registers} may be a single
12983 register, register range separated by ellipses, or comma-separated
12984 list of those. Example:
12987 #pragma GCC coprocessor subclass 'B' = $c2, $c4, $c6
12989 asm ("cpfoo %0" : "=B" (x));
12992 @item GCC disinterrupt @var{name} , @var{name} @dots{}
12993 @cindex pragma, disinterrupt
12994 For the named functions, the compiler adds code to disable interrupts
12995 for the duration of those functions. Any functions so named, which
12996 are not encountered in the source, cause a warning that the pragma was
12997 not used. Examples:
13000 #pragma disinterrupt foo
13001 #pragma disinterrupt bar, grill
13002 int foo () @{ @dots{} @}
13005 @item GCC call @var{name} , @var{name} @dots{}
13006 @cindex pragma, call
13007 For the named functions, the compiler always uses a register-indirect
13008 call model when calling the named functions. Examples:
13017 @node RS/6000 and PowerPC Pragmas
13018 @subsection RS/6000 and PowerPC Pragmas
13020 The RS/6000 and PowerPC targets define one pragma for controlling
13021 whether or not the @code{longcall} attribute is added to function
13022 declarations by default. This pragma overrides the @option{-mlongcall}
13023 option, but not the @code{longcall} and @code{shortcall} attributes.
13024 @xref{RS/6000 and PowerPC Options}, for more information about when long
13025 calls are and are not necessary.
13029 @cindex pragma, longcall
13030 Apply the @code{longcall} attribute to all subsequent function
13034 Do not apply the @code{longcall} attribute to subsequent function
13038 @c Describe h8300 pragmas here.
13039 @c Describe sh pragmas here.
13040 @c Describe v850 pragmas here.
13042 @node Darwin Pragmas
13043 @subsection Darwin Pragmas
13045 The following pragmas are available for all architectures running the
13046 Darwin operating system. These are useful for compatibility with other
13050 @item mark @var{tokens}@dots{}
13051 @cindex pragma, mark
13052 This pragma is accepted, but has no effect.
13054 @item options align=@var{alignment}
13055 @cindex pragma, options align
13056 This pragma sets the alignment of fields in structures. The values of
13057 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
13058 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
13059 properly; to restore the previous setting, use @code{reset} for the
13062 @item segment @var{tokens}@dots{}
13063 @cindex pragma, segment
13064 This pragma is accepted, but has no effect.
13066 @item unused (@var{var} [, @var{var}]@dots{})
13067 @cindex pragma, unused
13068 This pragma declares variables to be possibly unused. GCC will not
13069 produce warnings for the listed variables. The effect is similar to
13070 that of the @code{unused} attribute, except that this pragma may appear
13071 anywhere within the variables' scopes.
13074 @node Solaris Pragmas
13075 @subsection Solaris Pragmas
13077 The Solaris target supports @code{#pragma redefine_extname}
13078 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
13079 @code{#pragma} directives for compatibility with the system compiler.
13082 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
13083 @cindex pragma, align
13085 Increase the minimum alignment of each @var{variable} to @var{alignment}.
13086 This is the same as GCC's @code{aligned} attribute @pxref{Variable
13087 Attributes}). Macro expansion occurs on the arguments to this pragma
13088 when compiling C and Objective-C@. It does not currently occur when
13089 compiling C++, but this is a bug which may be fixed in a future
13092 @item fini (@var{function} [, @var{function}]...)
13093 @cindex pragma, fini
13095 This pragma causes each listed @var{function} to be called after
13096 main, or during shared module unloading, by adding a call to the
13097 @code{.fini} section.
13099 @item init (@var{function} [, @var{function}]...)
13100 @cindex pragma, init
13102 This pragma causes each listed @var{function} to be called during
13103 initialization (before @code{main}) or during shared module loading, by
13104 adding a call to the @code{.init} section.
13108 @node Symbol-Renaming Pragmas
13109 @subsection Symbol-Renaming Pragmas
13111 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
13112 supports two @code{#pragma} directives which change the name used in
13113 assembly for a given declaration. @code{#pragma extern_prefix} is only
13114 available on platforms whose system headers need it. To get this effect
13115 on all platforms supported by GCC, use the asm labels extension (@pxref{Asm
13119 @item redefine_extname @var{oldname} @var{newname}
13120 @cindex pragma, redefine_extname
13122 This pragma gives the C function @var{oldname} the assembly symbol
13123 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
13124 will be defined if this pragma is available (currently on all platforms).
13126 @item extern_prefix @var{string}
13127 @cindex pragma, extern_prefix
13129 This pragma causes all subsequent external function and variable
13130 declarations to have @var{string} prepended to their assembly symbols.
13131 This effect may be terminated with another @code{extern_prefix} pragma
13132 whose argument is an empty string. The preprocessor macro
13133 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
13134 available (currently only on Tru64 UNIX)@.
13137 These pragmas and the asm labels extension interact in a complicated
13138 manner. Here are some corner cases you may want to be aware of.
13141 @item Both pragmas silently apply only to declarations with external
13142 linkage. Asm labels do not have this restriction.
13144 @item In C++, both pragmas silently apply only to declarations with
13145 ``C'' linkage. Again, asm labels do not have this restriction.
13147 @item If any of the three ways of changing the assembly name of a
13148 declaration is applied to a declaration whose assembly name has
13149 already been determined (either by a previous use of one of these
13150 features, or because the compiler needed the assembly name in order to
13151 generate code), and the new name is different, a warning issues and
13152 the name does not change.
13154 @item The @var{oldname} used by @code{#pragma redefine_extname} is
13155 always the C-language name.
13157 @item If @code{#pragma extern_prefix} is in effect, and a declaration
13158 occurs with an asm label attached, the prefix is silently ignored for
13161 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
13162 apply to the same declaration, whichever triggered first wins, and a
13163 warning issues if they contradict each other. (We would like to have
13164 @code{#pragma redefine_extname} always win, for consistency with asm
13165 labels, but if @code{#pragma extern_prefix} triggers first we have no
13166 way of knowing that that happened.)
13169 @node Structure-Packing Pragmas
13170 @subsection Structure-Packing Pragmas
13172 For compatibility with Microsoft Windows compilers, GCC supports a
13173 set of @code{#pragma} directives which change the maximum alignment of
13174 members of structures (other than zero-width bitfields), unions, and
13175 classes subsequently defined. The @var{n} value below always is required
13176 to be a small power of two and specifies the new alignment in bytes.
13179 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
13180 @item @code{#pragma pack()} sets the alignment to the one that was in
13181 effect when compilation started (see also command-line option
13182 @option{-fpack-struct[=@var{n}]} @pxref{Code Gen Options}).
13183 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
13184 setting on an internal stack and then optionally sets the new alignment.
13185 @item @code{#pragma pack(pop)} restores the alignment setting to the one
13186 saved at the top of the internal stack (and removes that stack entry).
13187 Note that @code{#pragma pack([@var{n}])} does not influence this internal
13188 stack; thus it is possible to have @code{#pragma pack(push)} followed by
13189 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
13190 @code{#pragma pack(pop)}.
13193 Some targets, e.g.@: i386 and powerpc, support the @code{ms_struct}
13194 @code{#pragma} which lays out a structure as the documented
13195 @code{__attribute__ ((ms_struct))}.
13197 @item @code{#pragma ms_struct on} turns on the layout for structures
13199 @item @code{#pragma ms_struct off} turns off the layout for structures
13201 @item @code{#pragma ms_struct reset} goes back to the default layout.
13205 @subsection Weak Pragmas
13207 For compatibility with SVR4, GCC supports a set of @code{#pragma}
13208 directives for declaring symbols to be weak, and defining weak
13212 @item #pragma weak @var{symbol}
13213 @cindex pragma, weak
13214 This pragma declares @var{symbol} to be weak, as if the declaration
13215 had the attribute of the same name. The pragma may appear before
13216 or after the declaration of @var{symbol}. It is not an error for
13217 @var{symbol} to never be defined at all.
13219 @item #pragma weak @var{symbol1} = @var{symbol2}
13220 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
13221 It is an error if @var{symbol2} is not defined in the current
13225 @node Diagnostic Pragmas
13226 @subsection Diagnostic Pragmas
13228 GCC allows the user to selectively enable or disable certain types of
13229 diagnostics, and change the kind of the diagnostic. For example, a
13230 project's policy might require that all sources compile with
13231 @option{-Werror} but certain files might have exceptions allowing
13232 specific types of warnings. Or, a project might selectively enable
13233 diagnostics and treat them as errors depending on which preprocessor
13234 macros are defined.
13237 @item #pragma GCC diagnostic @var{kind} @var{option}
13238 @cindex pragma, diagnostic
13240 Modifies the disposition of a diagnostic. Note that not all
13241 diagnostics are modifiable; at the moment only warnings (normally
13242 controlled by @samp{-W@dots{}}) can be controlled, and not all of them.
13243 Use @option{-fdiagnostics-show-option} to determine which diagnostics
13244 are controllable and which option controls them.
13246 @var{kind} is @samp{error} to treat this diagnostic as an error,
13247 @samp{warning} to treat it like a warning (even if @option{-Werror} is
13248 in effect), or @samp{ignored} if the diagnostic is to be ignored.
13249 @var{option} is a double quoted string which matches the command-line
13253 #pragma GCC diagnostic warning "-Wformat"
13254 #pragma GCC diagnostic error "-Wformat"
13255 #pragma GCC diagnostic ignored "-Wformat"
13258 Note that these pragmas override any command-line options. GCC keeps
13259 track of the location of each pragma, and issues diagnostics according
13260 to the state as of that point in the source file. Thus, pragmas occurring
13261 after a line do not affect diagnostics caused by that line.
13263 @item #pragma GCC diagnostic push
13264 @itemx #pragma GCC diagnostic pop
13266 Causes GCC to remember the state of the diagnostics as of each
13267 @code{push}, and restore to that point at each @code{pop}. If a
13268 @code{pop} has no matching @code{push}, the command line options are
13272 #pragma GCC diagnostic error "-Wuninitialized"
13273 foo(a); /* error is given for this one */
13274 #pragma GCC diagnostic push
13275 #pragma GCC diagnostic ignored "-Wuninitialized"
13276 foo(b); /* no diagnostic for this one */
13277 #pragma GCC diagnostic pop
13278 foo(c); /* error is given for this one */
13279 #pragma GCC diagnostic pop
13280 foo(d); /* depends on command line options */
13285 GCC also offers a simple mechanism for printing messages during
13289 @item #pragma message @var{string}
13290 @cindex pragma, diagnostic
13292 Prints @var{string} as a compiler message on compilation. The message
13293 is informational only, and is neither a compilation warning nor an error.
13296 #pragma message "Compiling " __FILE__ "..."
13299 @var{string} may be parenthesized, and is printed with location
13300 information. For example,
13303 #define DO_PRAGMA(x) _Pragma (#x)
13304 #define TODO(x) DO_PRAGMA(message ("TODO - " #x))
13306 TODO(Remember to fix this)
13309 prints @samp{/tmp/file.c:4: note: #pragma message:
13310 TODO - Remember to fix this}.
13314 @node Visibility Pragmas
13315 @subsection Visibility Pragmas
13318 @item #pragma GCC visibility push(@var{visibility})
13319 @itemx #pragma GCC visibility pop
13320 @cindex pragma, visibility
13322 This pragma allows the user to set the visibility for multiple
13323 declarations without having to give each a visibility attribute
13324 @xref{Function Attributes}, for more information about visibility and
13325 the attribute syntax.
13327 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
13328 declarations. Class members and template specializations are not
13329 affected; if you want to override the visibility for a particular
13330 member or instantiation, you must use an attribute.
13335 @node Push/Pop Macro Pragmas
13336 @subsection Push/Pop Macro Pragmas
13338 For compatibility with Microsoft Windows compilers, GCC supports
13339 @samp{#pragma push_macro(@var{"macro_name"})}
13340 and @samp{#pragma pop_macro(@var{"macro_name"})}.
13343 @item #pragma push_macro(@var{"macro_name"})
13344 @cindex pragma, push_macro
13345 This pragma saves the value of the macro named as @var{macro_name} to
13346 the top of the stack for this macro.
13348 @item #pragma pop_macro(@var{"macro_name"})
13349 @cindex pragma, pop_macro
13350 This pragma sets the value of the macro named as @var{macro_name} to
13351 the value on top of the stack for this macro. If the stack for
13352 @var{macro_name} is empty, the value of the macro remains unchanged.
13359 #pragma push_macro("X")
13362 #pragma pop_macro("X")
13366 In this example, the definition of X as 1 is saved by @code{#pragma
13367 push_macro} and restored by @code{#pragma pop_macro}.
13369 @node Function Specific Option Pragmas
13370 @subsection Function Specific Option Pragmas
13373 @item #pragma GCC target (@var{"string"}...)
13374 @cindex pragma GCC target
13376 This pragma allows you to set target specific options for functions
13377 defined later in the source file. One or more strings can be
13378 specified. Each function that is defined after this point will be as
13379 if @code{attribute((target("STRING")))} was specified for that
13380 function. The parenthesis around the options is optional.
13381 @xref{Function Attributes}, for more information about the
13382 @code{target} attribute and the attribute syntax.
13384 The @code{#pragma GCC target} attribute is not implemented in GCC versions earlier
13385 than 4.4 for the i386/x86_64 and 4.6 for the PowerPC backends. At
13386 present, it is not implemented for other backends.
13390 @item #pragma GCC optimize (@var{"string"}...)
13391 @cindex pragma GCC optimize
13393 This pragma allows you to set global optimization options for functions
13394 defined later in the source file. One or more strings can be
13395 specified. Each function that is defined after this point will be as
13396 if @code{attribute((optimize("STRING")))} was specified for that
13397 function. The parenthesis around the options is optional.
13398 @xref{Function Attributes}, for more information about the
13399 @code{optimize} attribute and the attribute syntax.
13401 The @samp{#pragma GCC optimize} pragma is not implemented in GCC
13402 versions earlier than 4.4.
13406 @item #pragma GCC push_options
13407 @itemx #pragma GCC pop_options
13408 @cindex pragma GCC push_options
13409 @cindex pragma GCC pop_options
13411 These pragmas maintain a stack of the current target and optimization
13412 options. It is intended for include files where you temporarily want
13413 to switch to using a different @samp{#pragma GCC target} or
13414 @samp{#pragma GCC optimize} and then to pop back to the previous
13417 The @samp{#pragma GCC push_options} and @samp{#pragma GCC pop_options}
13418 pragmas are not implemented in GCC versions earlier than 4.4.
13422 @item #pragma GCC reset_options
13423 @cindex pragma GCC reset_options
13425 This pragma clears the current @code{#pragma GCC target} and
13426 @code{#pragma GCC optimize} to use the default switches as specified
13427 on the command line.
13429 The @samp{#pragma GCC reset_options} pragma is not implemented in GCC
13430 versions earlier than 4.4.
13433 @node Unnamed Fields
13434 @section Unnamed struct/union fields within structs/unions
13435 @cindex @code{struct}
13436 @cindex @code{union}
13438 As permitted by ISO C1X and for compatibility with other compilers,
13439 GCC allows you to define
13440 a structure or union that contains, as fields, structures and unions
13441 without names. For example:
13454 In this example, the user would be able to access members of the unnamed
13455 union with code like @samp{foo.b}. Note that only unnamed structs and
13456 unions are allowed, you may not have, for example, an unnamed
13459 You must never create such structures that cause ambiguous field definitions.
13460 For example, this structure:
13471 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
13472 The compiler gives errors for such constructs.
13474 @opindex fms-extensions
13475 Unless @option{-fms-extensions} is used, the unnamed field must be a
13476 structure or union definition without a tag (for example, @samp{struct
13477 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
13478 also be a definition with a tag such as @samp{struct foo @{ int a;
13479 @};}, a reference to a previously defined structure or union such as
13480 @samp{struct foo;}, or a reference to a @code{typedef} name for a
13481 previously defined structure or union type.
13483 @opindex fplan9-extensions
13484 The option @option{-fplan9-extensions} enables
13485 @option{-fms-extensions} as well as two other extensions. First, a
13486 pointer to a structure is automatically converted to a pointer to an
13487 anonymous field for assignments and function calls. For example:
13490 struct s1 @{ int a; @};
13491 struct s2 @{ struct s1; @};
13492 extern void f1 (struct s1 *);
13493 void f2 (struct s2 *p) @{ f1 (p); @}
13496 In the call to @code{f1} inside @code{f2}, the pointer @code{p} is
13497 converted into a pointer to the anonymous field.
13499 Second, when the type of an anonymous field is a @code{typedef} for a
13500 @code{struct} or @code{union}, code may refer to the field using the
13501 name of the @code{typedef}.
13504 typedef struct @{ int a; @} s1;
13505 struct s2 @{ s1; @};
13506 s1 f1 (struct s2 *p) @{ return p->s1; @}
13509 These usages are only permitted when they are not ambiguous.
13512 @section Thread-Local Storage
13513 @cindex Thread-Local Storage
13514 @cindex @acronym{TLS}
13515 @cindex @code{__thread}
13517 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
13518 are allocated such that there is one instance of the variable per extant
13519 thread. The run-time model GCC uses to implement this originates
13520 in the IA-64 processor-specific ABI, but has since been migrated
13521 to other processors as well. It requires significant support from
13522 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
13523 system libraries (@file{libc.so} and @file{libpthread.so}), so it
13524 is not available everywhere.
13526 At the user level, the extension is visible with a new storage
13527 class keyword: @code{__thread}. For example:
13531 extern __thread struct state s;
13532 static __thread char *p;
13535 The @code{__thread} specifier may be used alone, with the @code{extern}
13536 or @code{static} specifiers, but with no other storage class specifier.
13537 When used with @code{extern} or @code{static}, @code{__thread} must appear
13538 immediately after the other storage class specifier.
13540 The @code{__thread} specifier may be applied to any global, file-scoped
13541 static, function-scoped static, or static data member of a class. It may
13542 not be applied to block-scoped automatic or non-static data member.
13544 When the address-of operator is applied to a thread-local variable, it is
13545 evaluated at run-time and returns the address of the current thread's
13546 instance of that variable. An address so obtained may be used by any
13547 thread. When a thread terminates, any pointers to thread-local variables
13548 in that thread become invalid.
13550 No static initialization may refer to the address of a thread-local variable.
13552 In C++, if an initializer is present for a thread-local variable, it must
13553 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
13556 See @uref{http://www.akkadia.org/drepper/tls.pdf,
13557 ELF Handling For Thread-Local Storage} for a detailed explanation of
13558 the four thread-local storage addressing models, and how the run-time
13559 is expected to function.
13562 * C99 Thread-Local Edits::
13563 * C++98 Thread-Local Edits::
13566 @node C99 Thread-Local Edits
13567 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
13569 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
13570 that document the exact semantics of the language extension.
13574 @cite{5.1.2 Execution environments}
13576 Add new text after paragraph 1
13579 Within either execution environment, a @dfn{thread} is a flow of
13580 control within a program. It is implementation defined whether
13581 or not there may be more than one thread associated with a program.
13582 It is implementation defined how threads beyond the first are
13583 created, the name and type of the function called at thread
13584 startup, and how threads may be terminated. However, objects
13585 with thread storage duration shall be initialized before thread
13590 @cite{6.2.4 Storage durations of objects}
13592 Add new text before paragraph 3
13595 An object whose identifier is declared with the storage-class
13596 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
13597 Its lifetime is the entire execution of the thread, and its
13598 stored value is initialized only once, prior to thread startup.
13602 @cite{6.4.1 Keywords}
13604 Add @code{__thread}.
13607 @cite{6.7.1 Storage-class specifiers}
13609 Add @code{__thread} to the list of storage class specifiers in
13612 Change paragraph 2 to
13615 With the exception of @code{__thread}, at most one storage-class
13616 specifier may be given [@dots{}]. The @code{__thread} specifier may
13617 be used alone, or immediately following @code{extern} or
13621 Add new text after paragraph 6
13624 The declaration of an identifier for a variable that has
13625 block scope that specifies @code{__thread} shall also
13626 specify either @code{extern} or @code{static}.
13628 The @code{__thread} specifier shall be used only with
13633 @node C++98 Thread-Local Edits
13634 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
13636 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
13637 that document the exact semantics of the language extension.
13641 @b{[intro.execution]}
13643 New text after paragraph 4
13646 A @dfn{thread} is a flow of control within the abstract machine.
13647 It is implementation defined whether or not there may be more than
13651 New text after paragraph 7
13654 It is unspecified whether additional action must be taken to
13655 ensure when and whether side effects are visible to other threads.
13661 Add @code{__thread}.
13664 @b{[basic.start.main]}
13666 Add after paragraph 5
13669 The thread that begins execution at the @code{main} function is called
13670 the @dfn{main thread}. It is implementation defined how functions
13671 beginning threads other than the main thread are designated or typed.
13672 A function so designated, as well as the @code{main} function, is called
13673 a @dfn{thread startup function}. It is implementation defined what
13674 happens if a thread startup function returns. It is implementation
13675 defined what happens to other threads when any thread calls @code{exit}.
13679 @b{[basic.start.init]}
13681 Add after paragraph 4
13684 The storage for an object of thread storage duration shall be
13685 statically initialized before the first statement of the thread startup
13686 function. An object of thread storage duration shall not require
13687 dynamic initialization.
13691 @b{[basic.start.term]}
13693 Add after paragraph 3
13696 The type of an object with thread storage duration shall not have a
13697 non-trivial destructor, nor shall it be an array type whose elements
13698 (directly or indirectly) have non-trivial destructors.
13704 Add ``thread storage duration'' to the list in paragraph 1.
13709 Thread, static, and automatic storage durations are associated with
13710 objects introduced by declarations [@dots{}].
13713 Add @code{__thread} to the list of specifiers in paragraph 3.
13716 @b{[basic.stc.thread]}
13718 New section before @b{[basic.stc.static]}
13721 The keyword @code{__thread} applied to a non-local object gives the
13722 object thread storage duration.
13724 A local variable or class data member declared both @code{static}
13725 and @code{__thread} gives the variable or member thread storage
13730 @b{[basic.stc.static]}
13735 All objects which have neither thread storage duration, dynamic
13736 storage duration nor are local [@dots{}].
13742 Add @code{__thread} to the list in paragraph 1.
13747 With the exception of @code{__thread}, at most one
13748 @var{storage-class-specifier} shall appear in a given
13749 @var{decl-specifier-seq}. The @code{__thread} specifier may
13750 be used alone, or immediately following the @code{extern} or
13751 @code{static} specifiers. [@dots{}]
13754 Add after paragraph 5
13757 The @code{__thread} specifier can be applied only to the names of objects
13758 and to anonymous unions.
13764 Add after paragraph 6
13767 Non-@code{static} members shall not be @code{__thread}.
13771 @node Binary constants
13772 @section Binary constants using the @samp{0b} prefix
13773 @cindex Binary constants using the @samp{0b} prefix
13775 Integer constants can be written as binary constants, consisting of a
13776 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
13777 @samp{0B}. This is particularly useful in environments that operate a
13778 lot on the bit-level (like microcontrollers).
13780 The following statements are identical:
13789 The type of these constants follows the same rules as for octal or
13790 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
13793 @node C++ Extensions
13794 @chapter Extensions to the C++ Language
13795 @cindex extensions, C++ language
13796 @cindex C++ language extensions
13798 The GNU compiler provides these extensions to the C++ language (and you
13799 can also use most of the C language extensions in your C++ programs). If you
13800 want to write code that checks whether these features are available, you can
13801 test for the GNU compiler the same way as for C programs: check for a
13802 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
13803 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
13804 Predefined Macros,cpp,The GNU C Preprocessor}).
13807 * C++ Volatiles:: What constitutes an access to a volatile object.
13808 * Restricted Pointers:: C99 restricted pointers and references.
13809 * Vague Linkage:: Where G++ puts inlines, vtables and such.
13810 * C++ Interface:: You can use a single C++ header file for both
13811 declarations and definitions.
13812 * Template Instantiation:: Methods for ensuring that exactly one copy of
13813 each needed template instantiation is emitted.
13814 * Bound member functions:: You can extract a function pointer to the
13815 method denoted by a @samp{->*} or @samp{.*} expression.
13816 * C++ Attributes:: Variable, function, and type attributes for C++ only.
13817 * Namespace Association:: Strong using-directives for namespace association.
13818 * Type Traits:: Compiler support for type traits
13819 * Java Exceptions:: Tweaking exception handling to work with Java.
13820 * Deprecated Features:: Things will disappear from g++.
13821 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
13824 @node C++ Volatiles
13825 @section When is a Volatile C++ Object Accessed?
13826 @cindex accessing volatiles
13827 @cindex volatile read
13828 @cindex volatile write
13829 @cindex volatile access
13831 The C++ standard differs from the C standard in its treatment of
13832 volatile objects. It fails to specify what constitutes a volatile
13833 access, except to say that C++ should behave in a similar manner to C
13834 with respect to volatiles, where possible. However, the different
13835 lvalueness of expressions between C and C++ complicate the behavior.
13836 G++ behaves the same as GCC for volatile access, @xref{C
13837 Extensions,,Volatiles}, for a description of GCC's behavior.
13839 The C and C++ language specifications differ when an object is
13840 accessed in a void context:
13843 volatile int *src = @var{somevalue};
13847 The C++ standard specifies that such expressions do not undergo lvalue
13848 to rvalue conversion, and that the type of the dereferenced object may
13849 be incomplete. The C++ standard does not specify explicitly that it
13850 is lvalue to rvalue conversion which is responsible for causing an
13851 access. There is reason to believe that it is, because otherwise
13852 certain simple expressions become undefined. However, because it
13853 would surprise most programmers, G++ treats dereferencing a pointer to
13854 volatile object of complete type as GCC would do for an equivalent
13855 type in C@. When the object has incomplete type, G++ issues a
13856 warning; if you wish to force an error, you must force a conversion to
13857 rvalue with, for instance, a static cast.
13859 When using a reference to volatile, G++ does not treat equivalent
13860 expressions as accesses to volatiles, but instead issues a warning that
13861 no volatile is accessed. The rationale for this is that otherwise it
13862 becomes difficult to determine where volatile access occur, and not
13863 possible to ignore the return value from functions returning volatile
13864 references. Again, if you wish to force a read, cast the reference to
13867 G++ implements the same behavior as GCC does when assigning to a
13868 volatile object -- there is no reread of the assigned-to object, the
13869 assigned rvalue is reused. Note that in C++ assignment expressions
13870 are lvalues, and if used as an lvalue, the volatile object will be
13871 referred to. For instance, @var{vref} will refer to @var{vobj}, as
13872 expected, in the following example:
13876 volatile int &vref = vobj = @var{something};
13879 @node Restricted Pointers
13880 @section Restricting Pointer Aliasing
13881 @cindex restricted pointers
13882 @cindex restricted references
13883 @cindex restricted this pointer
13885 As with the C front end, G++ understands the C99 feature of restricted pointers,
13886 specified with the @code{__restrict__}, or @code{__restrict} type
13887 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
13888 language flag, @code{restrict} is not a keyword in C++.
13890 In addition to allowing restricted pointers, you can specify restricted
13891 references, which indicate that the reference is not aliased in the local
13895 void fn (int *__restrict__ rptr, int &__restrict__ rref)
13902 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
13903 @var{rref} refers to a (different) unaliased integer.
13905 You may also specify whether a member function's @var{this} pointer is
13906 unaliased by using @code{__restrict__} as a member function qualifier.
13909 void T::fn () __restrict__
13916 Within the body of @code{T::fn}, @var{this} will have the effective
13917 definition @code{T *__restrict__ const this}. Notice that the
13918 interpretation of a @code{__restrict__} member function qualifier is
13919 different to that of @code{const} or @code{volatile} qualifier, in that it
13920 is applied to the pointer rather than the object. This is consistent with
13921 other compilers which implement restricted pointers.
13923 As with all outermost parameter qualifiers, @code{__restrict__} is
13924 ignored in function definition matching. This means you only need to
13925 specify @code{__restrict__} in a function definition, rather than
13926 in a function prototype as well.
13928 @node Vague Linkage
13929 @section Vague Linkage
13930 @cindex vague linkage
13932 There are several constructs in C++ which require space in the object
13933 file but are not clearly tied to a single translation unit. We say that
13934 these constructs have ``vague linkage''. Typically such constructs are
13935 emitted wherever they are needed, though sometimes we can be more
13939 @item Inline Functions
13940 Inline functions are typically defined in a header file which can be
13941 included in many different compilations. Hopefully they can usually be
13942 inlined, but sometimes an out-of-line copy is necessary, if the address
13943 of the function is taken or if inlining fails. In general, we emit an
13944 out-of-line copy in all translation units where one is needed. As an
13945 exception, we only emit inline virtual functions with the vtable, since
13946 it will always require a copy.
13948 Local static variables and string constants used in an inline function
13949 are also considered to have vague linkage, since they must be shared
13950 between all inlined and out-of-line instances of the function.
13954 C++ virtual functions are implemented in most compilers using a lookup
13955 table, known as a vtable. The vtable contains pointers to the virtual
13956 functions provided by a class, and each object of the class contains a
13957 pointer to its vtable (or vtables, in some multiple-inheritance
13958 situations). If the class declares any non-inline, non-pure virtual
13959 functions, the first one is chosen as the ``key method'' for the class,
13960 and the vtable is only emitted in the translation unit where the key
13963 @emph{Note:} If the chosen key method is later defined as inline, the
13964 vtable will still be emitted in every translation unit which defines it.
13965 Make sure that any inline virtuals are declared inline in the class
13966 body, even if they are not defined there.
13968 @item @code{type_info} objects
13969 @cindex @code{type_info}
13971 C++ requires information about types to be written out in order to
13972 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
13973 For polymorphic classes (classes with virtual functions), the @samp{type_info}
13974 object is written out along with the vtable so that @samp{dynamic_cast}
13975 can determine the dynamic type of a class object at runtime. For all
13976 other types, we write out the @samp{type_info} object when it is used: when
13977 applying @samp{typeid} to an expression, throwing an object, or
13978 referring to a type in a catch clause or exception specification.
13980 @item Template Instantiations
13981 Most everything in this section also applies to template instantiations,
13982 but there are other options as well.
13983 @xref{Template Instantiation,,Where's the Template?}.
13987 When used with GNU ld version 2.8 or later on an ELF system such as
13988 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
13989 these constructs will be discarded at link time. This is known as
13992 On targets that don't support COMDAT, but do support weak symbols, GCC
13993 will use them. This way one copy will override all the others, but
13994 the unused copies will still take up space in the executable.
13996 For targets which do not support either COMDAT or weak symbols,
13997 most entities with vague linkage will be emitted as local symbols to
13998 avoid duplicate definition errors from the linker. This will not happen
13999 for local statics in inlines, however, as having multiple copies will
14000 almost certainly break things.
14002 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
14003 another way to control placement of these constructs.
14005 @node C++ Interface
14006 @section #pragma interface and implementation
14008 @cindex interface and implementation headers, C++
14009 @cindex C++ interface and implementation headers
14010 @cindex pragmas, interface and implementation
14012 @code{#pragma interface} and @code{#pragma implementation} provide the
14013 user with a way of explicitly directing the compiler to emit entities
14014 with vague linkage (and debugging information) in a particular
14017 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
14018 most cases, because of COMDAT support and the ``key method'' heuristic
14019 mentioned in @ref{Vague Linkage}. Using them can actually cause your
14020 program to grow due to unnecessary out-of-line copies of inline
14021 functions. Currently (3.4) the only benefit of these
14022 @code{#pragma}s is reduced duplication of debugging information, and
14023 that should be addressed soon on DWARF 2 targets with the use of
14027 @item #pragma interface
14028 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
14029 @kindex #pragma interface
14030 Use this directive in @emph{header files} that define object classes, to save
14031 space in most of the object files that use those classes. Normally,
14032 local copies of certain information (backup copies of inline member
14033 functions, debugging information, and the internal tables that implement
14034 virtual functions) must be kept in each object file that includes class
14035 definitions. You can use this pragma to avoid such duplication. When a
14036 header file containing @samp{#pragma interface} is included in a
14037 compilation, this auxiliary information will not be generated (unless
14038 the main input source file itself uses @samp{#pragma implementation}).
14039 Instead, the object files will contain references to be resolved at link
14042 The second form of this directive is useful for the case where you have
14043 multiple headers with the same name in different directories. If you
14044 use this form, you must specify the same string to @samp{#pragma
14047 @item #pragma implementation
14048 @itemx #pragma implementation "@var{objects}.h"
14049 @kindex #pragma implementation
14050 Use this pragma in a @emph{main input file}, when you want full output from
14051 included header files to be generated (and made globally visible). The
14052 included header file, in turn, should use @samp{#pragma interface}.
14053 Backup copies of inline member functions, debugging information, and the
14054 internal tables used to implement virtual functions are all generated in
14055 implementation files.
14057 @cindex implied @code{#pragma implementation}
14058 @cindex @code{#pragma implementation}, implied
14059 @cindex naming convention, implementation headers
14060 If you use @samp{#pragma implementation} with no argument, it applies to
14061 an include file with the same basename@footnote{A file's @dfn{basename}
14062 was the name stripped of all leading path information and of trailing
14063 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
14064 file. For example, in @file{allclass.cc}, giving just
14065 @samp{#pragma implementation}
14066 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
14068 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
14069 an implementation file whenever you would include it from
14070 @file{allclass.cc} even if you never specified @samp{#pragma
14071 implementation}. This was deemed to be more trouble than it was worth,
14072 however, and disabled.
14074 Use the string argument if you want a single implementation file to
14075 include code from multiple header files. (You must also use
14076 @samp{#include} to include the header file; @samp{#pragma
14077 implementation} only specifies how to use the file---it doesn't actually
14080 There is no way to split up the contents of a single header file into
14081 multiple implementation files.
14084 @cindex inlining and C++ pragmas
14085 @cindex C++ pragmas, effect on inlining
14086 @cindex pragmas in C++, effect on inlining
14087 @samp{#pragma implementation} and @samp{#pragma interface} also have an
14088 effect on function inlining.
14090 If you define a class in a header file marked with @samp{#pragma
14091 interface}, the effect on an inline function defined in that class is
14092 similar to an explicit @code{extern} declaration---the compiler emits
14093 no code at all to define an independent version of the function. Its
14094 definition is used only for inlining with its callers.
14096 @opindex fno-implement-inlines
14097 Conversely, when you include the same header file in a main source file
14098 that declares it as @samp{#pragma implementation}, the compiler emits
14099 code for the function itself; this defines a version of the function
14100 that can be found via pointers (or by callers compiled without
14101 inlining). If all calls to the function can be inlined, you can avoid
14102 emitting the function by compiling with @option{-fno-implement-inlines}.
14103 If any calls were not inlined, you will get linker errors.
14105 @node Template Instantiation
14106 @section Where's the Template?
14107 @cindex template instantiation
14109 C++ templates are the first language feature to require more
14110 intelligence from the environment than one usually finds on a UNIX
14111 system. Somehow the compiler and linker have to make sure that each
14112 template instance occurs exactly once in the executable if it is needed,
14113 and not at all otherwise. There are two basic approaches to this
14114 problem, which are referred to as the Borland model and the Cfront model.
14117 @item Borland model
14118 Borland C++ solved the template instantiation problem by adding the code
14119 equivalent of common blocks to their linker; the compiler emits template
14120 instances in each translation unit that uses them, and the linker
14121 collapses them together. The advantage of this model is that the linker
14122 only has to consider the object files themselves; there is no external
14123 complexity to worry about. This disadvantage is that compilation time
14124 is increased because the template code is being compiled repeatedly.
14125 Code written for this model tends to include definitions of all
14126 templates in the header file, since they must be seen to be
14130 The AT&T C++ translator, Cfront, solved the template instantiation
14131 problem by creating the notion of a template repository, an
14132 automatically maintained place where template instances are stored. A
14133 more modern version of the repository works as follows: As individual
14134 object files are built, the compiler places any template definitions and
14135 instantiations encountered in the repository. At link time, the link
14136 wrapper adds in the objects in the repository and compiles any needed
14137 instances that were not previously emitted. The advantages of this
14138 model are more optimal compilation speed and the ability to use the
14139 system linker; to implement the Borland model a compiler vendor also
14140 needs to replace the linker. The disadvantages are vastly increased
14141 complexity, and thus potential for error; for some code this can be
14142 just as transparent, but in practice it can been very difficult to build
14143 multiple programs in one directory and one program in multiple
14144 directories. Code written for this model tends to separate definitions
14145 of non-inline member templates into a separate file, which should be
14146 compiled separately.
14149 When used with GNU ld version 2.8 or later on an ELF system such as
14150 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
14151 Borland model. On other systems, G++ implements neither automatic
14154 A future version of G++ will support a hybrid model whereby the compiler
14155 will emit any instantiations for which the template definition is
14156 included in the compile, and store template definitions and
14157 instantiation context information into the object file for the rest.
14158 The link wrapper will extract that information as necessary and invoke
14159 the compiler to produce the remaining instantiations. The linker will
14160 then combine duplicate instantiations.
14162 In the mean time, you have the following options for dealing with
14163 template instantiations:
14168 Compile your template-using code with @option{-frepo}. The compiler will
14169 generate files with the extension @samp{.rpo} listing all of the
14170 template instantiations used in the corresponding object files which
14171 could be instantiated there; the link wrapper, @samp{collect2}, will
14172 then update the @samp{.rpo} files to tell the compiler where to place
14173 those instantiations and rebuild any affected object files. The
14174 link-time overhead is negligible after the first pass, as the compiler
14175 will continue to place the instantiations in the same files.
14177 This is your best option for application code written for the Borland
14178 model, as it will just work. Code written for the Cfront model will
14179 need to be modified so that the template definitions are available at
14180 one or more points of instantiation; usually this is as simple as adding
14181 @code{#include <tmethods.cc>} to the end of each template header.
14183 For library code, if you want the library to provide all of the template
14184 instantiations it needs, just try to link all of its object files
14185 together; the link will fail, but cause the instantiations to be
14186 generated as a side effect. Be warned, however, that this may cause
14187 conflicts if multiple libraries try to provide the same instantiations.
14188 For greater control, use explicit instantiation as described in the next
14192 @opindex fno-implicit-templates
14193 Compile your code with @option{-fno-implicit-templates} to disable the
14194 implicit generation of template instances, and explicitly instantiate
14195 all the ones you use. This approach requires more knowledge of exactly
14196 which instances you need than do the others, but it's less
14197 mysterious and allows greater control. You can scatter the explicit
14198 instantiations throughout your program, perhaps putting them in the
14199 translation units where the instances are used or the translation units
14200 that define the templates themselves; you can put all of the explicit
14201 instantiations you need into one big file; or you can create small files
14208 template class Foo<int>;
14209 template ostream& operator <<
14210 (ostream&, const Foo<int>&);
14213 for each of the instances you need, and create a template instantiation
14214 library from those.
14216 If you are using Cfront-model code, you can probably get away with not
14217 using @option{-fno-implicit-templates} when compiling files that don't
14218 @samp{#include} the member template definitions.
14220 If you use one big file to do the instantiations, you may want to
14221 compile it without @option{-fno-implicit-templates} so you get all of the
14222 instances required by your explicit instantiations (but not by any
14223 other files) without having to specify them as well.
14225 G++ has extended the template instantiation syntax given in the ISO
14226 standard to allow forward declaration of explicit instantiations
14227 (with @code{extern}), instantiation of the compiler support data for a
14228 template class (i.e.@: the vtable) without instantiating any of its
14229 members (with @code{inline}), and instantiation of only the static data
14230 members of a template class, without the support data or member
14231 functions (with (@code{static}):
14234 extern template int max (int, int);
14235 inline template class Foo<int>;
14236 static template class Foo<int>;
14240 Do nothing. Pretend G++ does implement automatic instantiation
14241 management. Code written for the Borland model will work fine, but
14242 each translation unit will contain instances of each of the templates it
14243 uses. In a large program, this can lead to an unacceptable amount of code
14247 @node Bound member functions
14248 @section Extracting the function pointer from a bound pointer to member function
14250 @cindex pointer to member function
14251 @cindex bound pointer to member function
14253 In C++, pointer to member functions (PMFs) are implemented using a wide
14254 pointer of sorts to handle all the possible call mechanisms; the PMF
14255 needs to store information about how to adjust the @samp{this} pointer,
14256 and if the function pointed to is virtual, where to find the vtable, and
14257 where in the vtable to look for the member function. If you are using
14258 PMFs in an inner loop, you should really reconsider that decision. If
14259 that is not an option, you can extract the pointer to the function that
14260 would be called for a given object/PMF pair and call it directly inside
14261 the inner loop, to save a bit of time.
14263 Note that you will still be paying the penalty for the call through a
14264 function pointer; on most modern architectures, such a call defeats the
14265 branch prediction features of the CPU@. This is also true of normal
14266 virtual function calls.
14268 The syntax for this extension is
14272 extern int (A::*fp)();
14273 typedef int (*fptr)(A *);
14275 fptr p = (fptr)(a.*fp);
14278 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
14279 no object is needed to obtain the address of the function. They can be
14280 converted to function pointers directly:
14283 fptr p1 = (fptr)(&A::foo);
14286 @opindex Wno-pmf-conversions
14287 You must specify @option{-Wno-pmf-conversions} to use this extension.
14289 @node C++ Attributes
14290 @section C++-Specific Variable, Function, and Type Attributes
14292 Some attributes only make sense for C++ programs.
14295 @item init_priority (@var{priority})
14296 @cindex @code{init_priority} attribute
14299 In Standard C++, objects defined at namespace scope are guaranteed to be
14300 initialized in an order in strict accordance with that of their definitions
14301 @emph{in a given translation unit}. No guarantee is made for initializations
14302 across translation units. However, GNU C++ allows users to control the
14303 order of initialization of objects defined at namespace scope with the
14304 @code{init_priority} attribute by specifying a relative @var{priority},
14305 a constant integral expression currently bounded between 101 and 65535
14306 inclusive. Lower numbers indicate a higher priority.
14308 In the following example, @code{A} would normally be created before
14309 @code{B}, but the @code{init_priority} attribute has reversed that order:
14312 Some_Class A __attribute__ ((init_priority (2000)));
14313 Some_Class B __attribute__ ((init_priority (543)));
14317 Note that the particular values of @var{priority} do not matter; only their
14320 @item java_interface
14321 @cindex @code{java_interface} attribute
14323 This type attribute informs C++ that the class is a Java interface. It may
14324 only be applied to classes declared within an @code{extern "Java"} block.
14325 Calls to methods declared in this interface will be dispatched using GCJ's
14326 interface table mechanism, instead of regular virtual table dispatch.
14330 See also @ref{Namespace Association}.
14332 @node Namespace Association
14333 @section Namespace Association
14335 @strong{Caution:} The semantics of this extension are not fully
14336 defined. Users should refrain from using this extension as its
14337 semantics may change subtly over time. It is possible that this
14338 extension will be removed in future versions of G++.
14340 A using-directive with @code{__attribute ((strong))} is stronger
14341 than a normal using-directive in two ways:
14345 Templates from the used namespace can be specialized and explicitly
14346 instantiated as though they were members of the using namespace.
14349 The using namespace is considered an associated namespace of all
14350 templates in the used namespace for purposes of argument-dependent
14354 The used namespace must be nested within the using namespace so that
14355 normal unqualified lookup works properly.
14357 This is useful for composing a namespace transparently from
14358 implementation namespaces. For example:
14363 template <class T> struct A @{ @};
14365 using namespace debug __attribute ((__strong__));
14366 template <> struct A<int> @{ @}; // @r{ok to specialize}
14368 template <class T> void f (A<T>);
14373 f (std::A<float>()); // @r{lookup finds} std::f
14379 @section Type Traits
14381 The C++ front-end implements syntactic extensions that allow to
14382 determine at compile time various characteristics of a type (or of a
14386 @item __has_nothrow_assign (type)
14387 If @code{type} is const qualified or is a reference type then the trait is
14388 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
14389 is true, else if @code{type} is a cv class or union type with copy assignment
14390 operators that are known not to throw an exception then the trait is true,
14391 else it is false. Requires: @code{type} shall be a complete type,
14392 (possibly cv-qualified) @code{void}, or an array of unknown bound.
14394 @item __has_nothrow_copy (type)
14395 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
14396 @code{type} is a cv class or union type with copy constructors that
14397 are known not to throw an exception then the trait is true, else it is false.
14398 Requires: @code{type} shall be a complete type, (possibly cv-qualified)
14399 @code{void}, or an array of unknown bound.
14401 @item __has_nothrow_constructor (type)
14402 If @code{__has_trivial_constructor (type)} is true then the trait is
14403 true, else if @code{type} is a cv class or union type (or array
14404 thereof) with a default constructor that is known not to throw an
14405 exception then the trait is true, else it is false. Requires:
14406 @code{type} shall be a complete type, (possibly cv-qualified)
14407 @code{void}, or an array of unknown bound.
14409 @item __has_trivial_assign (type)
14410 If @code{type} is const qualified or is a reference type then the trait is
14411 false. Otherwise if @code{__is_pod (type)} is true then the trait is
14412 true, else if @code{type} is a cv class or union type with a trivial
14413 copy assignment ([class.copy]) then the trait is true, else it is
14414 false. Requires: @code{type} shall be a complete type, (possibly
14415 cv-qualified) @code{void}, or an array of unknown bound.
14417 @item __has_trivial_copy (type)
14418 If @code{__is_pod (type)} is true or @code{type} is a reference type
14419 then the trait is true, else if @code{type} is a cv class or union type
14420 with a trivial copy constructor ([class.copy]) then the trait
14421 is true, else it is false. Requires: @code{type} shall be a complete
14422 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
14424 @item __has_trivial_constructor (type)
14425 If @code{__is_pod (type)} is true then the trait is true, else if
14426 @code{type} is a cv class or union type (or array thereof) with a
14427 trivial default constructor ([class.ctor]) then the trait is true,
14428 else it is false. Requires: @code{type} shall be a complete
14429 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
14431 @item __has_trivial_destructor (type)
14432 If @code{__is_pod (type)} is true or @code{type} is a reference type then
14433 the trait is true, else if @code{type} is a cv class or union type (or
14434 array thereof) with a trivial destructor ([class.dtor]) then the trait
14435 is true, else it is false. Requires: @code{type} shall be a complete
14436 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
14438 @item __has_virtual_destructor (type)
14439 If @code{type} is a class type with a virtual destructor
14440 ([class.dtor]) then the trait is true, else it is false. Requires:
14441 @code{type} shall be a complete type, (possibly cv-qualified)
14442 @code{void}, or an array of unknown bound.
14444 @item __is_abstract (type)
14445 If @code{type} is an abstract class ([class.abstract]) then the trait
14446 is true, else it is false. Requires: @code{type} shall be a complete
14447 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
14449 @item __is_base_of (base_type, derived_type)
14450 If @code{base_type} is a base class of @code{derived_type}
14451 ([class.derived]) then the trait is true, otherwise it is false.
14452 Top-level cv qualifications of @code{base_type} and
14453 @code{derived_type} are ignored. For the purposes of this trait, a
14454 class type is considered is own base. Requires: if @code{__is_class
14455 (base_type)} and @code{__is_class (derived_type)} are true and
14456 @code{base_type} and @code{derived_type} are not the same type
14457 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
14458 type. Diagnostic is produced if this requirement is not met.
14460 @item __is_class (type)
14461 If @code{type} is a cv class type, and not a union type
14462 ([basic.compound]) the trait is true, else it is false.
14464 @item __is_empty (type)
14465 If @code{__is_class (type)} is false then the trait is false.
14466 Otherwise @code{type} is considered empty if and only if: @code{type}
14467 has no non-static data members, or all non-static data members, if
14468 any, are bit-fields of length 0, and @code{type} has no virtual
14469 members, and @code{type} has no virtual base classes, and @code{type}
14470 has no base classes @code{base_type} for which
14471 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
14472 be a complete type, (possibly cv-qualified) @code{void}, or an array
14475 @item __is_enum (type)
14476 If @code{type} is a cv enumeration type ([basic.compound]) the trait is
14477 true, else it is false.
14479 @item __is_literal_type (type)
14480 If @code{type} is a literal type ([basic.types]) the trait is
14481 true, else it is false. Requires: @code{type} shall be a complete type,
14482 (possibly cv-qualified) @code{void}, or an array of unknown bound.
14484 @item __is_pod (type)
14485 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
14486 else it is false. Requires: @code{type} shall be a complete type,
14487 (possibly cv-qualified) @code{void}, or an array of unknown bound.
14489 @item __is_polymorphic (type)
14490 If @code{type} is a polymorphic class ([class.virtual]) then the trait
14491 is true, else it is false. Requires: @code{type} shall be a complete
14492 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
14494 @item __is_standard_layout (type)
14495 If @code{type} is a standard-layout type ([basic.types]) the trait is
14496 true, else it is false. Requires: @code{type} shall be a complete
14497 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
14499 @item __is_trivial (type)
14500 If @code{type} is a trivial type ([basic.types]) the trait is
14501 true, else it is false. Requires: @code{type} shall be a complete
14502 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
14504 @item __is_union (type)
14505 If @code{type} is a cv union type ([basic.compound]) the trait is
14506 true, else it is false.
14508 @item __underlying_type (type)
14509 The underlying type of @code{type}. Requires: @code{type} shall be
14510 an enumeration type ([dcl.enum]).
14514 @node Java Exceptions
14515 @section Java Exceptions
14517 The Java language uses a slightly different exception handling model
14518 from C++. Normally, GNU C++ will automatically detect when you are
14519 writing C++ code that uses Java exceptions, and handle them
14520 appropriately. However, if C++ code only needs to execute destructors
14521 when Java exceptions are thrown through it, GCC will guess incorrectly.
14522 Sample problematic code is:
14525 struct S @{ ~S(); @};
14526 extern void bar(); // @r{is written in Java, and may throw exceptions}
14535 The usual effect of an incorrect guess is a link failure, complaining of
14536 a missing routine called @samp{__gxx_personality_v0}.
14538 You can inform the compiler that Java exceptions are to be used in a
14539 translation unit, irrespective of what it might think, by writing
14540 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
14541 @samp{#pragma} must appear before any functions that throw or catch
14542 exceptions, or run destructors when exceptions are thrown through them.
14544 You cannot mix Java and C++ exceptions in the same translation unit. It
14545 is believed to be safe to throw a C++ exception from one file through
14546 another file compiled for the Java exception model, or vice versa, but
14547 there may be bugs in this area.
14549 @node Deprecated Features
14550 @section Deprecated Features
14552 In the past, the GNU C++ compiler was extended to experiment with new
14553 features, at a time when the C++ language was still evolving. Now that
14554 the C++ standard is complete, some of those features are superseded by
14555 superior alternatives. Using the old features might cause a warning in
14556 some cases that the feature will be dropped in the future. In other
14557 cases, the feature might be gone already.
14559 While the list below is not exhaustive, it documents some of the options
14560 that are now deprecated:
14563 @item -fexternal-templates
14564 @itemx -falt-external-templates
14565 These are two of the many ways for G++ to implement template
14566 instantiation. @xref{Template Instantiation}. The C++ standard clearly
14567 defines how template definitions have to be organized across
14568 implementation units. G++ has an implicit instantiation mechanism that
14569 should work just fine for standard-conforming code.
14571 @item -fstrict-prototype
14572 @itemx -fno-strict-prototype
14573 Previously it was possible to use an empty prototype parameter list to
14574 indicate an unspecified number of parameters (like C), rather than no
14575 parameters, as C++ demands. This feature has been removed, except where
14576 it is required for backwards compatibility. @xref{Backwards Compatibility}.
14579 G++ allows a virtual function returning @samp{void *} to be overridden
14580 by one returning a different pointer type. This extension to the
14581 covariant return type rules is now deprecated and will be removed from a
14584 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
14585 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
14586 and are now removed from G++. Code using these operators should be
14587 modified to use @code{std::min} and @code{std::max} instead.
14589 The named return value extension has been deprecated, and is now
14592 The use of initializer lists with new expressions has been deprecated,
14593 and is now removed from G++.
14595 Floating and complex non-type template parameters have been deprecated,
14596 and are now removed from G++.
14598 The implicit typename extension has been deprecated and is now
14601 The use of default arguments in function pointers, function typedefs
14602 and other places where they are not permitted by the standard is
14603 deprecated and will be removed from a future version of G++.
14605 G++ allows floating-point literals to appear in integral constant expressions,
14606 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
14607 This extension is deprecated and will be removed from a future version.
14609 G++ allows static data members of const floating-point type to be declared
14610 with an initializer in a class definition. The standard only allows
14611 initializers for static members of const integral types and const
14612 enumeration types so this extension has been deprecated and will be removed
14613 from a future version.
14615 @node Backwards Compatibility
14616 @section Backwards Compatibility
14617 @cindex Backwards Compatibility
14618 @cindex ARM [Annotated C++ Reference Manual]
14620 Now that there is a definitive ISO standard C++, G++ has a specification
14621 to adhere to. The C++ language evolved over time, and features that
14622 used to be acceptable in previous drafts of the standard, such as the ARM
14623 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
14624 compilation of C++ written to such drafts, G++ contains some backwards
14625 compatibilities. @emph{All such backwards compatibility features are
14626 liable to disappear in future versions of G++.} They should be considered
14627 deprecated. @xref{Deprecated Features}.
14631 If a variable is declared at for scope, it used to remain in scope until
14632 the end of the scope which contained the for statement (rather than just
14633 within the for scope). G++ retains this, but issues a warning, if such a
14634 variable is accessed outside the for scope.
14636 @item Implicit C language
14637 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
14638 scope to set the language. On such systems, all header files are
14639 implicitly scoped inside a C language scope. Also, an empty prototype
14640 @code{()} will be treated as an unspecified number of arguments, rather
14641 than no arguments, as C++ demands.