1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1996, 1998, 1999, 2000, 2001,
2 @c 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
8 @chapter Extensions to the C Language Family
9 @cindex extensions, C language
10 @cindex C language extensions
13 GNU C provides several language features not found in ISO standard C@.
14 (The @option{-pedantic} option directs GCC to print a warning message if
15 any of these features is used.) To test for the availability of these
16 features in conditional compilation, check for a predefined macro
17 @code{__GNUC__}, which is always defined under GCC@.
19 These extensions are available in C and Objective-C@. Most of them are
20 also available in C++. @xref{C++ Extensions,,Extensions to the
21 C++ Language}, for extensions that apply @emph{only} to C++.
23 Some features that are in ISO C99 but not C89 or C++ are also, as
24 extensions, accepted by GCC in C89 mode and in C++.
27 * Statement Exprs:: Putting statements and declarations inside expressions.
28 * Local Labels:: Labels local to a block.
29 * Labels as Values:: Getting pointers to labels, and computed gotos.
30 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
31 * Constructing Calls:: Dispatching a call to another function.
32 * Typeof:: @code{typeof}: referring to the type of an expression.
33 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
34 * Long Long:: Double-word integers---@code{long long int}.
35 * Complex:: Data types for complex numbers.
36 * Floating Types:: Additional Floating Types.
37 * Decimal Float:: Decimal Floating Types.
38 * Hex Floats:: Hexadecimal floating-point constants.
39 * Fixed-Point:: Fixed-Point Types.
40 * Zero Length:: Zero-length arrays.
41 * Variable Length:: Arrays whose length is computed at run time.
42 * Empty Structures:: Structures with no members.
43 * Variadic Macros:: Macros with a variable number of arguments.
44 * Escaped Newlines:: Slightly looser rules for escaped newlines.
45 * Subscripting:: Any array can be subscripted, even if not an lvalue.
46 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
47 * Initializers:: Non-constant initializers.
48 * Compound Literals:: Compound literals give structures, unions
50 * Designated Inits:: Labeling elements of initializers.
51 * Cast to Union:: Casting to union type from any member of the union.
52 * Case Ranges:: `case 1 ... 9' and such.
53 * Mixed Declarations:: Mixing declarations and code.
54 * Function Attributes:: Declaring that functions have no side effects,
55 or that they can never return.
56 * Attribute Syntax:: Formal syntax for attributes.
57 * Function Prototypes:: Prototype declarations and old-style definitions.
58 * C++ Comments:: C++ comments are recognized.
59 * Dollar Signs:: Dollar sign is allowed in identifiers.
60 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
61 * Variable Attributes:: Specifying attributes of variables.
62 * Type Attributes:: Specifying attributes of types.
63 * Alignment:: Inquiring about the alignment of a type or variable.
64 * Inline:: Defining inline functions (as fast as macros).
65 * Extended Asm:: Assembler instructions with C expressions as operands.
66 (With them you can define ``built-in'' functions.)
67 * Constraints:: Constraints for asm operands
68 * Asm Labels:: Specifying the assembler name to use for a C symbol.
69 * Explicit Reg Vars:: Defining variables residing in specified registers.
70 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
71 * Incomplete Enums:: @code{enum foo;}, with details to follow.
72 * Function Names:: Printable strings which are the name of the current
74 * Return Address:: Getting the return or frame address of a function.
75 * Vector Extensions:: Using vector instructions through built-in functions.
76 * Offsetof:: Special syntax for implementing @code{offsetof}.
77 * Atomic Builtins:: Built-in functions for atomic memory access.
78 * Object Size Checking:: Built-in functions for limited buffer overflow
80 * Other Builtins:: Other built-in functions.
81 * Target Builtins:: Built-in functions specific to particular targets.
82 * Target Format Checks:: Format checks specific to particular targets.
83 * Pragmas:: Pragmas accepted by GCC.
84 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
85 * Thread-Local:: Per-thread variables.
86 * Binary constants:: Binary constants using the @samp{0b} prefix.
90 @section Statements and Declarations in Expressions
91 @cindex statements inside expressions
92 @cindex declarations inside expressions
93 @cindex expressions containing statements
94 @cindex macros, statements in expressions
96 @c the above section title wrapped and causes an underfull hbox.. i
97 @c changed it from "within" to "in". --mew 4feb93
98 A compound statement enclosed in parentheses may appear as an expression
99 in GNU C@. This allows you to use loops, switches, and local variables
100 within an expression.
102 Recall that a compound statement is a sequence of statements surrounded
103 by braces; in this construct, parentheses go around the braces. For
107 (@{ int y = foo (); int z;
114 is a valid (though slightly more complex than necessary) expression
115 for the absolute value of @code{foo ()}.
117 The last thing in the compound statement should be an expression
118 followed by a semicolon; the value of this subexpression serves as the
119 value of the entire construct. (If you use some other kind of statement
120 last within the braces, the construct has type @code{void}, and thus
121 effectively no value.)
123 This feature is especially useful in making macro definitions ``safe'' (so
124 that they evaluate each operand exactly once). For example, the
125 ``maximum'' function is commonly defined as a macro in standard C as
129 #define max(a,b) ((a) > (b) ? (a) : (b))
133 @cindex side effects, macro argument
134 But this definition computes either @var{a} or @var{b} twice, with bad
135 results if the operand has side effects. In GNU C, if you know the
136 type of the operands (here taken as @code{int}), you can define
137 the macro safely as follows:
140 #define maxint(a,b) \
141 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
144 Embedded statements are not allowed in constant expressions, such as
145 the value of an enumeration constant, the width of a bit-field, or
146 the initial value of a static variable.
148 If you don't know the type of the operand, you can still do this, but you
149 must use @code{typeof} (@pxref{Typeof}).
151 In G++, the result value of a statement expression undergoes array and
152 function pointer decay, and is returned by value to the enclosing
153 expression. For instance, if @code{A} is a class, then
162 will construct a temporary @code{A} object to hold the result of the
163 statement expression, and that will be used to invoke @code{Foo}.
164 Therefore the @code{this} pointer observed by @code{Foo} will not be the
167 Any temporaries created within a statement within a statement expression
168 will be destroyed at the statement's end. This makes statement
169 expressions inside macros slightly different from function calls. In
170 the latter case temporaries introduced during argument evaluation will
171 be destroyed at the end of the statement that includes the function
172 call. In the statement expression case they will be destroyed during
173 the statement expression. For instance,
176 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
177 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
187 will have different places where temporaries are destroyed. For the
188 @code{macro} case, the temporary @code{X} will be destroyed just after
189 the initialization of @code{b}. In the @code{function} case that
190 temporary will be destroyed when the function returns.
192 These considerations mean that it is probably a bad idea to use
193 statement-expressions of this form in header files that are designed to
194 work with C++. (Note that some versions of the GNU C Library contained
195 header files using statement-expression that lead to precisely this
198 Jumping into a statement expression with @code{goto} or using a
199 @code{switch} statement outside the statement expression with a
200 @code{case} or @code{default} label inside the statement expression is
201 not permitted. Jumping into a statement expression with a computed
202 @code{goto} (@pxref{Labels as Values}) yields undefined behavior.
203 Jumping out of a statement expression is permitted, but if the
204 statement expression is part of a larger expression then it is
205 unspecified which other subexpressions of that expression have been
206 evaluated except where the language definition requires certain
207 subexpressions to be evaluated before or after the statement
208 expression. In any case, as with a function call the evaluation of a
209 statement expression is not interleaved with the evaluation of other
210 parts of the containing expression. For example,
213 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
217 will call @code{foo} and @code{bar1} and will not call @code{baz} but
218 may or may not call @code{bar2}. If @code{bar2} is called, it will be
219 called after @code{foo} and before @code{bar1}
222 @section Locally Declared Labels
224 @cindex macros, local labels
226 GCC allows you to declare @dfn{local labels} in any nested block
227 scope. A local label is just like an ordinary label, but you can
228 only reference it (with a @code{goto} statement, or by taking its
229 address) within the block in which it was declared.
231 A local label declaration looks like this:
234 __label__ @var{label};
241 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
244 Local label declarations must come at the beginning of the block,
245 before any ordinary declarations or statements.
247 The label declaration defines the label @emph{name}, but does not define
248 the label itself. You must do this in the usual way, with
249 @code{@var{label}:}, within the statements of the statement expression.
251 The local label feature is useful for complex macros. If a macro
252 contains nested loops, a @code{goto} can be useful for breaking out of
253 them. However, an ordinary label whose scope is the whole function
254 cannot be used: if the macro can be expanded several times in one
255 function, the label will be multiply defined in that function. A
256 local label avoids this problem. For example:
259 #define SEARCH(value, array, target) \
262 typeof (target) _SEARCH_target = (target); \
263 typeof (*(array)) *_SEARCH_array = (array); \
266 for (i = 0; i < max; i++) \
267 for (j = 0; j < max; j++) \
268 if (_SEARCH_array[i][j] == _SEARCH_target) \
269 @{ (value) = i; goto found; @} \
275 This could also be written using a statement-expression:
278 #define SEARCH(array, target) \
281 typeof (target) _SEARCH_target = (target); \
282 typeof (*(array)) *_SEARCH_array = (array); \
285 for (i = 0; i < max; i++) \
286 for (j = 0; j < max; j++) \
287 if (_SEARCH_array[i][j] == _SEARCH_target) \
288 @{ value = i; goto found; @} \
295 Local label declarations also make the labels they declare visible to
296 nested functions, if there are any. @xref{Nested Functions}, for details.
298 @node Labels as Values
299 @section Labels as Values
300 @cindex labels as values
301 @cindex computed gotos
302 @cindex goto with computed label
303 @cindex address of a label
305 You can get the address of a label defined in the current function
306 (or a containing function) with the unary operator @samp{&&}. The
307 value has type @code{void *}. This value is a constant and can be used
308 wherever a constant of that type is valid. For example:
316 To use these values, you need to be able to jump to one. This is done
317 with the computed goto statement@footnote{The analogous feature in
318 Fortran is called an assigned goto, but that name seems inappropriate in
319 C, where one can do more than simply store label addresses in label
320 variables.}, @code{goto *@var{exp};}. For example,
327 Any expression of type @code{void *} is allowed.
329 One way of using these constants is in initializing a static array that
330 will serve as a jump table:
333 static void *array[] = @{ &&foo, &&bar, &&hack @};
336 Then you can select a label with indexing, like this:
343 Note that this does not check whether the subscript is in bounds---array
344 indexing in C never does that.
346 Such an array of label values serves a purpose much like that of the
347 @code{switch} statement. The @code{switch} statement is cleaner, so
348 use that rather than an array unless the problem does not fit a
349 @code{switch} statement very well.
351 Another use of label values is in an interpreter for threaded code.
352 The labels within the interpreter function can be stored in the
353 threaded code for super-fast dispatching.
355 You may not use this mechanism to jump to code in a different function.
356 If you do that, totally unpredictable things will happen. The best way to
357 avoid this is to store the label address only in automatic variables and
358 never pass it as an argument.
360 An alternate way to write the above example is
363 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
365 goto *(&&foo + array[i]);
369 This is more friendly to code living in shared libraries, as it reduces
370 the number of dynamic relocations that are needed, and by consequence,
371 allows the data to be read-only.
373 The @code{&&foo} expressions for the same label might have different values
374 if the containing function is inlined or cloned. If a program relies on
375 them being always the same, @code{__attribute__((__noinline__))} should
376 be used to prevent inlining. If @code{&&foo} is used
377 in a static variable initializer, inlining is forbidden.
379 @node Nested Functions
380 @section Nested Functions
381 @cindex nested functions
382 @cindex downward funargs
385 A @dfn{nested function} is a function defined inside another function.
386 (Nested functions are not supported for GNU C++.) The nested function's
387 name is local to the block where it is defined. For example, here we
388 define a nested function named @code{square}, and call it twice:
392 foo (double a, double b)
394 double square (double z) @{ return z * z; @}
396 return square (a) + square (b);
401 The nested function can access all the variables of the containing
402 function that are visible at the point of its definition. This is
403 called @dfn{lexical scoping}. For example, here we show a nested
404 function which uses an inherited variable named @code{offset}:
408 bar (int *array, int offset, int size)
410 int access (int *array, int index)
411 @{ return array[index + offset]; @}
414 for (i = 0; i < size; i++)
415 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
420 Nested function definitions are permitted within functions in the places
421 where variable definitions are allowed; that is, in any block, mixed
422 with the other declarations and statements in the block.
424 It is possible to call the nested function from outside the scope of its
425 name by storing its address or passing the address to another function:
428 hack (int *array, int size)
430 void store (int index, int value)
431 @{ array[index] = value; @}
433 intermediate (store, size);
437 Here, the function @code{intermediate} receives the address of
438 @code{store} as an argument. If @code{intermediate} calls @code{store},
439 the arguments given to @code{store} are used to store into @code{array}.
440 But this technique works only so long as the containing function
441 (@code{hack}, in this example) does not exit.
443 If you try to call the nested function through its address after the
444 containing function has exited, all hell will break loose. If you try
445 to call it after a containing scope level has exited, and if it refers
446 to some of the variables that are no longer in scope, you may be lucky,
447 but it's not wise to take the risk. If, however, the nested function
448 does not refer to anything that has gone out of scope, you should be
451 GCC implements taking the address of a nested function using a technique
452 called @dfn{trampolines}. A paper describing them is available as
455 @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}.
457 A nested function can jump to a label inherited from a containing
458 function, provided the label was explicitly declared in the containing
459 function (@pxref{Local Labels}). Such a jump returns instantly to the
460 containing function, exiting the nested function which did the
461 @code{goto} and any intermediate functions as well. Here is an example:
465 bar (int *array, int offset, int size)
468 int access (int *array, int index)
472 return array[index + offset];
476 for (i = 0; i < size; i++)
477 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
481 /* @r{Control comes here from @code{access}
482 if it detects an error.} */
489 A nested function always has no linkage. Declaring one with
490 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
491 before its definition, use @code{auto} (which is otherwise meaningless
492 for function declarations).
495 bar (int *array, int offset, int size)
498 auto int access (int *, int);
500 int access (int *array, int index)
504 return array[index + offset];
510 @node Constructing Calls
511 @section Constructing Function Calls
512 @cindex constructing calls
513 @cindex forwarding calls
515 Using the built-in functions described below, you can record
516 the arguments a function received, and call another function
517 with the same arguments, without knowing the number or types
520 You can also record the return value of that function call,
521 and later return that value, without knowing what data type
522 the function tried to return (as long as your caller expects
525 However, these built-in functions may interact badly with some
526 sophisticated features or other extensions of the language. It
527 is, therefore, not recommended to use them outside very simple
528 functions acting as mere forwarders for their arguments.
530 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
531 This built-in function returns a pointer to data
532 describing how to perform a call with the same arguments as were passed
533 to the current function.
535 The function saves the arg pointer register, structure value address,
536 and all registers that might be used to pass arguments to a function
537 into a block of memory allocated on the stack. Then it returns the
538 address of that block.
541 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
542 This built-in function invokes @var{function}
543 with a copy of the parameters described by @var{arguments}
546 The value of @var{arguments} should be the value returned by
547 @code{__builtin_apply_args}. The argument @var{size} specifies the size
548 of the stack argument data, in bytes.
550 This function returns a pointer to data describing
551 how to return whatever value was returned by @var{function}. The data
552 is saved in a block of memory allocated on the stack.
554 It is not always simple to compute the proper value for @var{size}. The
555 value is used by @code{__builtin_apply} to compute the amount of data
556 that should be pushed on the stack and copied from the incoming argument
560 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
561 This built-in function returns the value described by @var{result} from
562 the containing function. You should specify, for @var{result}, a value
563 returned by @code{__builtin_apply}.
566 @deftypefn {Built-in Function} __builtin_va_arg_pack ()
567 This built-in function represents all anonymous arguments of an inline
568 function. It can be used only in inline functions which will be always
569 inlined, never compiled as a separate function, such as those using
570 @code{__attribute__ ((__always_inline__))} or
571 @code{__attribute__ ((__gnu_inline__))} extern inline functions.
572 It must be only passed as last argument to some other function
573 with variable arguments. This is useful for writing small wrapper
574 inlines for variable argument functions, when using preprocessor
575 macros is undesirable. For example:
577 extern int myprintf (FILE *f, const char *format, ...);
578 extern inline __attribute__ ((__gnu_inline__)) int
579 myprintf (FILE *f, const char *format, ...)
581 int r = fprintf (f, "myprintf: ");
584 int s = fprintf (f, format, __builtin_va_arg_pack ());
592 @deftypefn {Built-in Function} __builtin_va_arg_pack_len ()
593 This built-in function returns the number of anonymous arguments of
594 an inline function. It can be used only in inline functions which
595 will be always inlined, never compiled as a separate function, such
596 as those using @code{__attribute__ ((__always_inline__))} or
597 @code{__attribute__ ((__gnu_inline__))} extern inline functions.
598 For example following will do link or runtime checking of open
599 arguments for optimized code:
602 extern inline __attribute__((__gnu_inline__)) int
603 myopen (const char *path, int oflag, ...)
605 if (__builtin_va_arg_pack_len () > 1)
606 warn_open_too_many_arguments ();
608 if (__builtin_constant_p (oflag))
610 if ((oflag & O_CREAT) != 0 && __builtin_va_arg_pack_len () < 1)
612 warn_open_missing_mode ();
613 return __open_2 (path, oflag);
615 return open (path, oflag, __builtin_va_arg_pack ());
618 if (__builtin_va_arg_pack_len () < 1)
619 return __open_2 (path, oflag);
621 return open (path, oflag, __builtin_va_arg_pack ());
628 @section Referring to a Type with @code{typeof}
631 @cindex macros, types of arguments
633 Another way to refer to the type of an expression is with @code{typeof}.
634 The syntax of using of this keyword looks like @code{sizeof}, but the
635 construct acts semantically like a type name defined with @code{typedef}.
637 There are two ways of writing the argument to @code{typeof}: with an
638 expression or with a type. Here is an example with an expression:
645 This assumes that @code{x} is an array of pointers to functions;
646 the type described is that of the values of the functions.
648 Here is an example with a typename as the argument:
655 Here the type described is that of pointers to @code{int}.
657 If you are writing a header file that must work when included in ISO C
658 programs, write @code{__typeof__} instead of @code{typeof}.
659 @xref{Alternate Keywords}.
661 A @code{typeof}-construct can be used anywhere a typedef name could be
662 used. For example, you can use it in a declaration, in a cast, or inside
663 of @code{sizeof} or @code{typeof}.
665 @code{typeof} is often useful in conjunction with the
666 statements-within-expressions feature. Here is how the two together can
667 be used to define a safe ``maximum'' macro that operates on any
668 arithmetic type and evaluates each of its arguments exactly once:
672 (@{ typeof (a) _a = (a); \
673 typeof (b) _b = (b); \
674 _a > _b ? _a : _b; @})
677 @cindex underscores in variables in macros
678 @cindex @samp{_} in variables in macros
679 @cindex local variables in macros
680 @cindex variables, local, in macros
681 @cindex macros, local variables in
683 The reason for using names that start with underscores for the local
684 variables is to avoid conflicts with variable names that occur within the
685 expressions that are substituted for @code{a} and @code{b}. Eventually we
686 hope to design a new form of declaration syntax that allows you to declare
687 variables whose scopes start only after their initializers; this will be a
688 more reliable way to prevent such conflicts.
691 Some more examples of the use of @code{typeof}:
695 This declares @code{y} with the type of what @code{x} points to.
702 This declares @code{y} as an array of such values.
709 This declares @code{y} as an array of pointers to characters:
712 typeof (typeof (char *)[4]) y;
716 It is equivalent to the following traditional C declaration:
722 To see the meaning of the declaration using @code{typeof}, and why it
723 might be a useful way to write, rewrite it with these macros:
726 #define pointer(T) typeof(T *)
727 #define array(T, N) typeof(T [N])
731 Now the declaration can be rewritten this way:
734 array (pointer (char), 4) y;
738 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
739 pointers to @code{char}.
742 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
743 a more limited extension which permitted one to write
746 typedef @var{T} = @var{expr};
750 with the effect of declaring @var{T} to have the type of the expression
751 @var{expr}. This extension does not work with GCC 3 (versions between
752 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
753 relies on it should be rewritten to use @code{typeof}:
756 typedef typeof(@var{expr}) @var{T};
760 This will work with all versions of GCC@.
763 @section Conditionals with Omitted Operands
764 @cindex conditional expressions, extensions
765 @cindex omitted middle-operands
766 @cindex middle-operands, omitted
767 @cindex extensions, @code{?:}
768 @cindex @code{?:} extensions
770 The middle operand in a conditional expression may be omitted. Then
771 if the first operand is nonzero, its value is the value of the conditional
774 Therefore, the expression
781 has the value of @code{x} if that is nonzero; otherwise, the value of
784 This example is perfectly equivalent to
790 @cindex side effect in ?:
791 @cindex ?: side effect
793 In this simple case, the ability to omit the middle operand is not
794 especially useful. When it becomes useful is when the first operand does,
795 or may (if it is a macro argument), contain a side effect. Then repeating
796 the operand in the middle would perform the side effect twice. Omitting
797 the middle operand uses the value already computed without the undesirable
798 effects of recomputing it.
801 @section Double-Word Integers
802 @cindex @code{long long} data types
803 @cindex double-word arithmetic
804 @cindex multiprecision arithmetic
805 @cindex @code{LL} integer suffix
806 @cindex @code{ULL} integer suffix
808 ISO C99 supports data types for integers that are at least 64 bits wide,
809 and as an extension GCC supports them in C89 mode and in C++.
810 Simply write @code{long long int} for a signed integer, or
811 @code{unsigned long long int} for an unsigned integer. To make an
812 integer constant of type @code{long long int}, add the suffix @samp{LL}
813 to the integer. To make an integer constant of type @code{unsigned long
814 long int}, add the suffix @samp{ULL} to the integer.
816 You can use these types in arithmetic like any other integer types.
817 Addition, subtraction, and bitwise boolean operations on these types
818 are open-coded on all types of machines. Multiplication is open-coded
819 if the machine supports fullword-to-doubleword a widening multiply
820 instruction. Division and shifts are open-coded only on machines that
821 provide special support. The operations that are not open-coded use
822 special library routines that come with GCC@.
824 There may be pitfalls when you use @code{long long} types for function
825 arguments, unless you declare function prototypes. If a function
826 expects type @code{int} for its argument, and you pass a value of type
827 @code{long long int}, confusion will result because the caller and the
828 subroutine will disagree about the number of bytes for the argument.
829 Likewise, if the function expects @code{long long int} and you pass
830 @code{int}. The best way to avoid such problems is to use prototypes.
833 @section Complex Numbers
834 @cindex complex numbers
835 @cindex @code{_Complex} keyword
836 @cindex @code{__complex__} keyword
838 ISO C99 supports complex floating data types, and as an extension GCC
839 supports them in C89 mode and in C++, and supports complex integer data
840 types which are not part of ISO C99. You can declare complex types
841 using the keyword @code{_Complex}. As an extension, the older GNU
842 keyword @code{__complex__} is also supported.
844 For example, @samp{_Complex double x;} declares @code{x} as a
845 variable whose real part and imaginary part are both of type
846 @code{double}. @samp{_Complex short int y;} declares @code{y} to
847 have real and imaginary parts of type @code{short int}; this is not
848 likely to be useful, but it shows that the set of complex types is
851 To write a constant with a complex data type, use the suffix @samp{i} or
852 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
853 has type @code{_Complex float} and @code{3i} has type
854 @code{_Complex int}. Such a constant always has a pure imaginary
855 value, but you can form any complex value you like by adding one to a
856 real constant. This is a GNU extension; if you have an ISO C99
857 conforming C library (such as GNU libc), and want to construct complex
858 constants of floating type, you should include @code{<complex.h>} and
859 use the macros @code{I} or @code{_Complex_I} instead.
861 @cindex @code{__real__} keyword
862 @cindex @code{__imag__} keyword
863 To extract the real part of a complex-valued expression @var{exp}, write
864 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
865 extract the imaginary part. This is a GNU extension; for values of
866 floating type, you should use the ISO C99 functions @code{crealf},
867 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
868 @code{cimagl}, declared in @code{<complex.h>} and also provided as
869 built-in functions by GCC@.
871 @cindex complex conjugation
872 The operator @samp{~} performs complex conjugation when used on a value
873 with a complex type. This is a GNU extension; for values of
874 floating type, you should use the ISO C99 functions @code{conjf},
875 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
876 provided as built-in functions by GCC@.
878 GCC can allocate complex automatic variables in a noncontiguous
879 fashion; it's even possible for the real part to be in a register while
880 the imaginary part is on the stack (or vice-versa). Only the DWARF2
881 debug info format can represent this, so use of DWARF2 is recommended.
882 If you are using the stabs debug info format, GCC describes a noncontiguous
883 complex variable as if it were two separate variables of noncomplex type.
884 If the variable's actual name is @code{foo}, the two fictitious
885 variables are named @code{foo$real} and @code{foo$imag}. You can
886 examine and set these two fictitious variables with your debugger.
889 @section Additional Floating Types
890 @cindex additional floating types
891 @cindex @code{__float80} data type
892 @cindex @code{__float128} data type
893 @cindex @code{w} floating point suffix
894 @cindex @code{q} floating point suffix
895 @cindex @code{W} floating point suffix
896 @cindex @code{Q} floating point suffix
898 As an extension, the GNU C compiler supports additional floating
899 types, @code{__float80} and @code{__float128} to support 80bit
900 (@code{XFmode}) and 128 bit (@code{TFmode}) floating types.
901 Support for additional types includes the arithmetic operators:
902 add, subtract, multiply, divide; unary arithmetic operators;
903 relational operators; equality operators; and conversions to and from
904 integer and other floating types. Use a suffix @samp{w} or @samp{W}
905 in a literal constant of type @code{__float80} and @samp{q} or @samp{Q}
906 for @code{_float128}. You can declare complex types using the
907 corresponding internal complex type, @code{XCmode} for @code{__float80}
908 type and @code{TCmode} for @code{__float128} type:
911 typedef _Complex float __attribute__((mode(TC))) _Complex128;
912 typedef _Complex float __attribute__((mode(XC))) _Complex80;
915 Not all targets support additional floating point types. @code{__float80}
916 is supported on i386, x86_64 and ia64 targets and target @code{__float128}
917 is supported on x86_64 and ia64 targets.
920 @section Decimal Floating Types
921 @cindex decimal floating types
922 @cindex @code{_Decimal32} data type
923 @cindex @code{_Decimal64} data type
924 @cindex @code{_Decimal128} data type
925 @cindex @code{df} integer suffix
926 @cindex @code{dd} integer suffix
927 @cindex @code{dl} integer suffix
928 @cindex @code{DF} integer suffix
929 @cindex @code{DD} integer suffix
930 @cindex @code{DL} integer suffix
932 As an extension, the GNU C compiler supports decimal floating types as
933 defined in the N1176 draft of ISO/IEC WDTR24732. Support for decimal
934 floating types in GCC will evolve as the draft technical report changes.
935 Calling conventions for any target might also change. Not all targets
936 support decimal floating types.
938 The decimal floating types are @code{_Decimal32}, @code{_Decimal64}, and
939 @code{_Decimal128}. They use a radix of ten, unlike the floating types
940 @code{float}, @code{double}, and @code{long double} whose radix is not
941 specified by the C standard but is usually two.
943 Support for decimal floating types includes the arithmetic operators
944 add, subtract, multiply, divide; unary arithmetic operators;
945 relational operators; equality operators; and conversions to and from
946 integer and other floating types. Use a suffix @samp{df} or
947 @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd}
948 or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for
951 GCC support of decimal float as specified by the draft technical report
956 Translation time data type (TTDT) is not supported.
959 When the value of a decimal floating type cannot be represented in the
960 integer type to which it is being converted, the result is undefined
961 rather than the result value specified by the draft technical report.
964 Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128}
965 are supported by the DWARF2 debug information format.
971 ISO C99 supports floating-point numbers written not only in the usual
972 decimal notation, such as @code{1.55e1}, but also numbers such as
973 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
974 supports this in C89 mode (except in some cases when strictly
975 conforming) and in C++. In that format the
976 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
977 mandatory. The exponent is a decimal number that indicates the power of
978 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
985 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
986 is the same as @code{1.55e1}.
988 Unlike for floating-point numbers in the decimal notation the exponent
989 is always required in the hexadecimal notation. Otherwise the compiler
990 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
991 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
992 extension for floating-point constants of type @code{float}.
995 @section Fixed-Point Types
996 @cindex fixed-point types
997 @cindex @code{_Fract} data type
998 @cindex @code{_Accum} data type
999 @cindex @code{_Sat} data type
1000 @cindex @code{hr} fixed-suffix
1001 @cindex @code{r} fixed-suffix
1002 @cindex @code{lr} fixed-suffix
1003 @cindex @code{llr} fixed-suffix
1004 @cindex @code{uhr} fixed-suffix
1005 @cindex @code{ur} fixed-suffix
1006 @cindex @code{ulr} fixed-suffix
1007 @cindex @code{ullr} fixed-suffix
1008 @cindex @code{hk} fixed-suffix
1009 @cindex @code{k} fixed-suffix
1010 @cindex @code{lk} fixed-suffix
1011 @cindex @code{llk} fixed-suffix
1012 @cindex @code{uhk} fixed-suffix
1013 @cindex @code{uk} fixed-suffix
1014 @cindex @code{ulk} fixed-suffix
1015 @cindex @code{ullk} fixed-suffix
1016 @cindex @code{HR} fixed-suffix
1017 @cindex @code{R} fixed-suffix
1018 @cindex @code{LR} fixed-suffix
1019 @cindex @code{LLR} fixed-suffix
1020 @cindex @code{UHR} fixed-suffix
1021 @cindex @code{UR} fixed-suffix
1022 @cindex @code{ULR} fixed-suffix
1023 @cindex @code{ULLR} fixed-suffix
1024 @cindex @code{HK} fixed-suffix
1025 @cindex @code{K} fixed-suffix
1026 @cindex @code{LK} fixed-suffix
1027 @cindex @code{LLK} fixed-suffix
1028 @cindex @code{UHK} fixed-suffix
1029 @cindex @code{UK} fixed-suffix
1030 @cindex @code{ULK} fixed-suffix
1031 @cindex @code{ULLK} fixed-suffix
1033 As an extension, the GNU C compiler supports fixed-point types as
1034 defined in the N1169 draft of ISO/IEC DTR 18037. Support for fixed-point
1035 types in GCC will evolve as the draft technical report changes.
1036 Calling conventions for any target might also change. Not all targets
1037 support fixed-point types.
1039 The fixed-point types are
1040 @code{short _Fract},
1043 @code{long long _Fract},
1044 @code{unsigned short _Fract},
1045 @code{unsigned _Fract},
1046 @code{unsigned long _Fract},
1047 @code{unsigned long long _Fract},
1048 @code{_Sat short _Fract},
1050 @code{_Sat long _Fract},
1051 @code{_Sat long long _Fract},
1052 @code{_Sat unsigned short _Fract},
1053 @code{_Sat unsigned _Fract},
1054 @code{_Sat unsigned long _Fract},
1055 @code{_Sat unsigned long long _Fract},
1056 @code{short _Accum},
1059 @code{long long _Accum},
1060 @code{unsigned short _Accum},
1061 @code{unsigned _Accum},
1062 @code{unsigned long _Accum},
1063 @code{unsigned long long _Accum},
1064 @code{_Sat short _Accum},
1066 @code{_Sat long _Accum},
1067 @code{_Sat long long _Accum},
1068 @code{_Sat unsigned short _Accum},
1069 @code{_Sat unsigned _Accum},
1070 @code{_Sat unsigned long _Accum},
1071 @code{_Sat unsigned long long _Accum}.
1072 Fixed-point data values contain fractional and optional integral parts.
1073 The format of fixed-point data varies and depends on the target machine.
1075 Support for fixed-point types includes prefix and postfix increment
1076 and decrement operators (@code{++}, @code{--}); unary arithmetic operators
1077 (@code{+}, @code{-}, @code{!}); binary arithmetic operators (@code{+},
1078 @code{-}, @code{*}, @code{/}); binary shift operators (@code{<<}, @code{>>});
1079 relational operators (@code{<}, @code{<=}, @code{>=}, @code{>});
1080 equality operators (@code{==}, @code{!=}); assignment operators
1081 (@code{+=}, @code{-=}, @code{*=}, @code{/=}, @code{<<=}, @code{>>=});
1082 and conversions to and from integer, floating-point, or fixed-point types.
1084 Use a suffix @samp{hr} or @samp{HR} in a literal constant of type
1085 @code{short _Fract} and @code{_Sat short _Fract},
1086 @samp{r} or @samp{R} for @code{_Fract} and @code{_Sat _Fract},
1087 @samp{lr} or @samp{LR} for @code{long _Fract} and @code{_Sat long _Fract},
1088 @samp{llr} or @samp{LLR} for @code{long long _Fract} and
1089 @code{_Sat long long _Fract},
1090 @samp{uhr} or @samp{UHR} for @code{unsigned short _Fract} and
1091 @code{_Sat unsigned short _Fract},
1092 @samp{ur} or @samp{UR} for @code{unsigned _Fract} and
1093 @code{_Sat unsigned _Fract},
1094 @samp{ulr} or @samp{ULR} for @code{unsigned long _Fract} and
1095 @code{_Sat unsigned long _Fract},
1096 @samp{ullr} or @samp{ULLR} for @code{unsigned long long _Fract}
1097 and @code{_Sat unsigned long long _Fract},
1098 @samp{hk} or @samp{HK} for @code{short _Accum} and @code{_Sat short _Accum},
1099 @samp{k} or @samp{K} for @code{_Accum} and @code{_Sat _Accum},
1100 @samp{lk} or @samp{LK} for @code{long _Accum} and @code{_Sat long _Accum},
1101 @samp{llk} or @samp{LLK} for @code{long long _Accum} and
1102 @code{_Sat long long _Accum},
1103 @samp{uhk} or @samp{UHK} for @code{unsigned short _Accum} and
1104 @code{_Sat unsigned short _Accum},
1105 @samp{uk} or @samp{UK} for @code{unsigned _Accum} and
1106 @code{_Sat unsigned _Accum},
1107 @samp{ulk} or @samp{ULK} for @code{unsigned long _Accum} and
1108 @code{_Sat unsigned long _Accum},
1109 and @samp{ullk} or @samp{ULLK} for @code{unsigned long long _Accum}
1110 and @code{_Sat unsigned long long _Accum}.
1112 GCC support of fixed-point types as specified by the draft technical report
1117 Pragmas to control overflow and rounding behaviors are not implemented.
1120 Fixed-point types are supported by the DWARF2 debug information format.
1123 @section Arrays of Length Zero
1124 @cindex arrays of length zero
1125 @cindex zero-length arrays
1126 @cindex length-zero arrays
1127 @cindex flexible array members
1129 Zero-length arrays are allowed in GNU C@. They are very useful as the
1130 last element of a structure which is really a header for a variable-length
1139 struct line *thisline = (struct line *)
1140 malloc (sizeof (struct line) + this_length);
1141 thisline->length = this_length;
1144 In ISO C90, you would have to give @code{contents} a length of 1, which
1145 means either you waste space or complicate the argument to @code{malloc}.
1147 In ISO C99, you would use a @dfn{flexible array member}, which is
1148 slightly different in syntax and semantics:
1152 Flexible array members are written as @code{contents[]} without
1156 Flexible array members have incomplete type, and so the @code{sizeof}
1157 operator may not be applied. As a quirk of the original implementation
1158 of zero-length arrays, @code{sizeof} evaluates to zero.
1161 Flexible array members may only appear as the last member of a
1162 @code{struct} that is otherwise non-empty.
1165 A structure containing a flexible array member, or a union containing
1166 such a structure (possibly recursively), may not be a member of a
1167 structure or an element of an array. (However, these uses are
1168 permitted by GCC as extensions.)
1171 GCC versions before 3.0 allowed zero-length arrays to be statically
1172 initialized, as if they were flexible arrays. In addition to those
1173 cases that were useful, it also allowed initializations in situations
1174 that would corrupt later data. Non-empty initialization of zero-length
1175 arrays is now treated like any case where there are more initializer
1176 elements than the array holds, in that a suitable warning about "excess
1177 elements in array" is given, and the excess elements (all of them, in
1178 this case) are ignored.
1180 Instead GCC allows static initialization of flexible array members.
1181 This is equivalent to defining a new structure containing the original
1182 structure followed by an array of sufficient size to contain the data.
1183 I.e.@: in the following, @code{f1} is constructed as if it were declared
1189 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
1192 struct f1 f1; int data[3];
1193 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1197 The convenience of this extension is that @code{f1} has the desired
1198 type, eliminating the need to consistently refer to @code{f2.f1}.
1200 This has symmetry with normal static arrays, in that an array of
1201 unknown size is also written with @code{[]}.
1203 Of course, this extension only makes sense if the extra data comes at
1204 the end of a top-level object, as otherwise we would be overwriting
1205 data at subsequent offsets. To avoid undue complication and confusion
1206 with initialization of deeply nested arrays, we simply disallow any
1207 non-empty initialization except when the structure is the top-level
1208 object. For example:
1211 struct foo @{ int x; int y[]; @};
1212 struct bar @{ struct foo z; @};
1214 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1215 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1216 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1217 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1220 @node Empty Structures
1221 @section Structures With No Members
1222 @cindex empty structures
1223 @cindex zero-size structures
1225 GCC permits a C structure to have no members:
1232 The structure will have size zero. In C++, empty structures are part
1233 of the language. G++ treats empty structures as if they had a single
1234 member of type @code{char}.
1236 @node Variable Length
1237 @section Arrays of Variable Length
1238 @cindex variable-length arrays
1239 @cindex arrays of variable length
1242 Variable-length automatic arrays are allowed in ISO C99, and as an
1243 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1244 implementation of variable-length arrays does not yet conform in detail
1245 to the ISO C99 standard.) These arrays are
1246 declared like any other automatic arrays, but with a length that is not
1247 a constant expression. The storage is allocated at the point of
1248 declaration and deallocated when the brace-level is exited. For
1253 concat_fopen (char *s1, char *s2, char *mode)
1255 char str[strlen (s1) + strlen (s2) + 1];
1258 return fopen (str, mode);
1262 @cindex scope of a variable length array
1263 @cindex variable-length array scope
1264 @cindex deallocating variable length arrays
1265 Jumping or breaking out of the scope of the array name deallocates the
1266 storage. Jumping into the scope is not allowed; you get an error
1269 @cindex @code{alloca} vs variable-length arrays
1270 You can use the function @code{alloca} to get an effect much like
1271 variable-length arrays. The function @code{alloca} is available in
1272 many other C implementations (but not in all). On the other hand,
1273 variable-length arrays are more elegant.
1275 There are other differences between these two methods. Space allocated
1276 with @code{alloca} exists until the containing @emph{function} returns.
1277 The space for a variable-length array is deallocated as soon as the array
1278 name's scope ends. (If you use both variable-length arrays and
1279 @code{alloca} in the same function, deallocation of a variable-length array
1280 will also deallocate anything more recently allocated with @code{alloca}.)
1282 You can also use variable-length arrays as arguments to functions:
1286 tester (int len, char data[len][len])
1292 The length of an array is computed once when the storage is allocated
1293 and is remembered for the scope of the array in case you access it with
1296 If you want to pass the array first and the length afterward, you can
1297 use a forward declaration in the parameter list---another GNU extension.
1301 tester (int len; char data[len][len], int len)
1307 @cindex parameter forward declaration
1308 The @samp{int len} before the semicolon is a @dfn{parameter forward
1309 declaration}, and it serves the purpose of making the name @code{len}
1310 known when the declaration of @code{data} is parsed.
1312 You can write any number of such parameter forward declarations in the
1313 parameter list. They can be separated by commas or semicolons, but the
1314 last one must end with a semicolon, which is followed by the ``real''
1315 parameter declarations. Each forward declaration must match a ``real''
1316 declaration in parameter name and data type. ISO C99 does not support
1317 parameter forward declarations.
1319 @node Variadic Macros
1320 @section Macros with a Variable Number of Arguments.
1321 @cindex variable number of arguments
1322 @cindex macro with variable arguments
1323 @cindex rest argument (in macro)
1324 @cindex variadic macros
1326 In the ISO C standard of 1999, a macro can be declared to accept a
1327 variable number of arguments much as a function can. The syntax for
1328 defining the macro is similar to that of a function. Here is an
1332 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1335 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1336 such a macro, it represents the zero or more tokens until the closing
1337 parenthesis that ends the invocation, including any commas. This set of
1338 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1339 wherever it appears. See the CPP manual for more information.
1341 GCC has long supported variadic macros, and used a different syntax that
1342 allowed you to give a name to the variable arguments just like any other
1343 argument. Here is an example:
1346 #define debug(format, args...) fprintf (stderr, format, args)
1349 This is in all ways equivalent to the ISO C example above, but arguably
1350 more readable and descriptive.
1352 GNU CPP has two further variadic macro extensions, and permits them to
1353 be used with either of the above forms of macro definition.
1355 In standard C, you are not allowed to leave the variable argument out
1356 entirely; but you are allowed to pass an empty argument. For example,
1357 this invocation is invalid in ISO C, because there is no comma after
1364 GNU CPP permits you to completely omit the variable arguments in this
1365 way. In the above examples, the compiler would complain, though since
1366 the expansion of the macro still has the extra comma after the format
1369 To help solve this problem, CPP behaves specially for variable arguments
1370 used with the token paste operator, @samp{##}. If instead you write
1373 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1376 and if the variable arguments are omitted or empty, the @samp{##}
1377 operator causes the preprocessor to remove the comma before it. If you
1378 do provide some variable arguments in your macro invocation, GNU CPP
1379 does not complain about the paste operation and instead places the
1380 variable arguments after the comma. Just like any other pasted macro
1381 argument, these arguments are not macro expanded.
1383 @node Escaped Newlines
1384 @section Slightly Looser Rules for Escaped Newlines
1385 @cindex escaped newlines
1386 @cindex newlines (escaped)
1388 Recently, the preprocessor has relaxed its treatment of escaped
1389 newlines. Previously, the newline had to immediately follow a
1390 backslash. The current implementation allows whitespace in the form
1391 of spaces, horizontal and vertical tabs, and form feeds between the
1392 backslash and the subsequent newline. The preprocessor issues a
1393 warning, but treats it as a valid escaped newline and combines the two
1394 lines to form a single logical line. This works within comments and
1395 tokens, as well as between tokens. Comments are @emph{not} treated as
1396 whitespace for the purposes of this relaxation, since they have not
1397 yet been replaced with spaces.
1400 @section Non-Lvalue Arrays May Have Subscripts
1401 @cindex subscripting
1402 @cindex arrays, non-lvalue
1404 @cindex subscripting and function values
1405 In ISO C99, arrays that are not lvalues still decay to pointers, and
1406 may be subscripted, although they may not be modified or used after
1407 the next sequence point and the unary @samp{&} operator may not be
1408 applied to them. As an extension, GCC allows such arrays to be
1409 subscripted in C89 mode, though otherwise they do not decay to
1410 pointers outside C99 mode. For example,
1411 this is valid in GNU C though not valid in C89:
1415 struct foo @{int a[4];@};
1421 return f().a[index];
1427 @section Arithmetic on @code{void}- and Function-Pointers
1428 @cindex void pointers, arithmetic
1429 @cindex void, size of pointer to
1430 @cindex function pointers, arithmetic
1431 @cindex function, size of pointer to
1433 In GNU C, addition and subtraction operations are supported on pointers to
1434 @code{void} and on pointers to functions. This is done by treating the
1435 size of a @code{void} or of a function as 1.
1437 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1438 and on function types, and returns 1.
1440 @opindex Wpointer-arith
1441 The option @option{-Wpointer-arith} requests a warning if these extensions
1445 @section Non-Constant Initializers
1446 @cindex initializers, non-constant
1447 @cindex non-constant initializers
1449 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1450 automatic variable are not required to be constant expressions in GNU C@.
1451 Here is an example of an initializer with run-time varying elements:
1454 foo (float f, float g)
1456 float beat_freqs[2] = @{ f-g, f+g @};
1461 @node Compound Literals
1462 @section Compound Literals
1463 @cindex constructor expressions
1464 @cindex initializations in expressions
1465 @cindex structures, constructor expression
1466 @cindex expressions, constructor
1467 @cindex compound literals
1468 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1470 ISO C99 supports compound literals. A compound literal looks like
1471 a cast containing an initializer. Its value is an object of the
1472 type specified in the cast, containing the elements specified in
1473 the initializer; it is an lvalue. As an extension, GCC supports
1474 compound literals in C89 mode and in C++.
1476 Usually, the specified type is a structure. Assume that
1477 @code{struct foo} and @code{structure} are declared as shown:
1480 struct foo @{int a; char b[2];@} structure;
1484 Here is an example of constructing a @code{struct foo} with a compound literal:
1487 structure = ((struct foo) @{x + y, 'a', 0@});
1491 This is equivalent to writing the following:
1495 struct foo temp = @{x + y, 'a', 0@};
1500 You can also construct an array. If all the elements of the compound literal
1501 are (made up of) simple constant expressions, suitable for use in
1502 initializers of objects of static storage duration, then the compound
1503 literal can be coerced to a pointer to its first element and used in
1504 such an initializer, as shown here:
1507 char **foo = (char *[]) @{ "x", "y", "z" @};
1510 Compound literals for scalar types and union types are is
1511 also allowed, but then the compound literal is equivalent
1514 As a GNU extension, GCC allows initialization of objects with static storage
1515 duration by compound literals (which is not possible in ISO C99, because
1516 the initializer is not a constant).
1517 It is handled as if the object was initialized only with the bracket
1518 enclosed list if the types of the compound literal and the object match.
1519 The initializer list of the compound literal must be constant.
1520 If the object being initialized has array type of unknown size, the size is
1521 determined by compound literal size.
1524 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1525 static int y[] = (int []) @{1, 2, 3@};
1526 static int z[] = (int [3]) @{1@};
1530 The above lines are equivalent to the following:
1532 static struct foo x = @{1, 'a', 'b'@};
1533 static int y[] = @{1, 2, 3@};
1534 static int z[] = @{1, 0, 0@};
1537 @node Designated Inits
1538 @section Designated Initializers
1539 @cindex initializers with labeled elements
1540 @cindex labeled elements in initializers
1541 @cindex case labels in initializers
1542 @cindex designated initializers
1544 Standard C89 requires the elements of an initializer to appear in a fixed
1545 order, the same as the order of the elements in the array or structure
1548 In ISO C99 you can give the elements in any order, specifying the array
1549 indices or structure field names they apply to, and GNU C allows this as
1550 an extension in C89 mode as well. This extension is not
1551 implemented in GNU C++.
1553 To specify an array index, write
1554 @samp{[@var{index}] =} before the element value. For example,
1557 int a[6] = @{ [4] = 29, [2] = 15 @};
1564 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1568 The index values must be constant expressions, even if the array being
1569 initialized is automatic.
1571 An alternative syntax for this which has been obsolete since GCC 2.5 but
1572 GCC still accepts is to write @samp{[@var{index}]} before the element
1573 value, with no @samp{=}.
1575 To initialize a range of elements to the same value, write
1576 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1577 extension. For example,
1580 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1584 If the value in it has side-effects, the side-effects will happen only once,
1585 not for each initialized field by the range initializer.
1588 Note that the length of the array is the highest value specified
1591 In a structure initializer, specify the name of a field to initialize
1592 with @samp{.@var{fieldname} =} before the element value. For example,
1593 given the following structure,
1596 struct point @{ int x, y; @};
1600 the following initialization
1603 struct point p = @{ .y = yvalue, .x = xvalue @};
1610 struct point p = @{ xvalue, yvalue @};
1613 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1614 @samp{@var{fieldname}:}, as shown here:
1617 struct point p = @{ y: yvalue, x: xvalue @};
1621 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1622 @dfn{designator}. You can also use a designator (or the obsolete colon
1623 syntax) when initializing a union, to specify which element of the union
1624 should be used. For example,
1627 union foo @{ int i; double d; @};
1629 union foo f = @{ .d = 4 @};
1633 will convert 4 to a @code{double} to store it in the union using
1634 the second element. By contrast, casting 4 to type @code{union foo}
1635 would store it into the union as the integer @code{i}, since it is
1636 an integer. (@xref{Cast to Union}.)
1638 You can combine this technique of naming elements with ordinary C
1639 initialization of successive elements. Each initializer element that
1640 does not have a designator applies to the next consecutive element of the
1641 array or structure. For example,
1644 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1651 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1654 Labeling the elements of an array initializer is especially useful
1655 when the indices are characters or belong to an @code{enum} type.
1660 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1661 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1664 @cindex designator lists
1665 You can also write a series of @samp{.@var{fieldname}} and
1666 @samp{[@var{index}]} designators before an @samp{=} to specify a
1667 nested subobject to initialize; the list is taken relative to the
1668 subobject corresponding to the closest surrounding brace pair. For
1669 example, with the @samp{struct point} declaration above:
1672 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1676 If the same field is initialized multiple times, it will have value from
1677 the last initialization. If any such overridden initialization has
1678 side-effect, it is unspecified whether the side-effect happens or not.
1679 Currently, GCC will discard them and issue a warning.
1682 @section Case Ranges
1684 @cindex ranges in case statements
1686 You can specify a range of consecutive values in a single @code{case} label,
1690 case @var{low} ... @var{high}:
1694 This has the same effect as the proper number of individual @code{case}
1695 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1697 This feature is especially useful for ranges of ASCII character codes:
1703 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1704 it may be parsed wrong when you use it with integer values. For example,
1719 @section Cast to a Union Type
1720 @cindex cast to a union
1721 @cindex union, casting to a
1723 A cast to union type is similar to other casts, except that the type
1724 specified is a union type. You can specify the type either with
1725 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1726 a constructor though, not a cast, and hence does not yield an lvalue like
1727 normal casts. (@xref{Compound Literals}.)
1729 The types that may be cast to the union type are those of the members
1730 of the union. Thus, given the following union and variables:
1733 union foo @{ int i; double d; @};
1739 both @code{x} and @code{y} can be cast to type @code{union foo}.
1741 Using the cast as the right-hand side of an assignment to a variable of
1742 union type is equivalent to storing in a member of the union:
1747 u = (union foo) x @equiv{} u.i = x
1748 u = (union foo) y @equiv{} u.d = y
1751 You can also use the union cast as a function argument:
1754 void hack (union foo);
1756 hack ((union foo) x);
1759 @node Mixed Declarations
1760 @section Mixed Declarations and Code
1761 @cindex mixed declarations and code
1762 @cindex declarations, mixed with code
1763 @cindex code, mixed with declarations
1765 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1766 within compound statements. As an extension, GCC also allows this in
1767 C89 mode. For example, you could do:
1776 Each identifier is visible from where it is declared until the end of
1777 the enclosing block.
1779 @node Function Attributes
1780 @section Declaring Attributes of Functions
1781 @cindex function attributes
1782 @cindex declaring attributes of functions
1783 @cindex functions that never return
1784 @cindex functions that return more than once
1785 @cindex functions that have no side effects
1786 @cindex functions in arbitrary sections
1787 @cindex functions that behave like malloc
1788 @cindex @code{volatile} applied to function
1789 @cindex @code{const} applied to function
1790 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1791 @cindex functions with non-null pointer arguments
1792 @cindex functions that are passed arguments in registers on the 386
1793 @cindex functions that pop the argument stack on the 386
1794 @cindex functions that do not pop the argument stack on the 386
1796 In GNU C, you declare certain things about functions called in your program
1797 which help the compiler optimize function calls and check your code more
1800 The keyword @code{__attribute__} allows you to specify special
1801 attributes when making a declaration. This keyword is followed by an
1802 attribute specification inside double parentheses. The following
1803 attributes are currently defined for functions on all targets:
1804 @code{aligned}, @code{alloc_size}, @code{noreturn},
1805 @code{returns_twice}, @code{noinline}, @code{always_inline},
1806 @code{flatten}, @code{pure}, @code{const}, @code{nothrow},
1807 @code{sentinel}, @code{format}, @code{format_arg},
1808 @code{no_instrument_function}, @code{section}, @code{constructor},
1809 @code{destructor}, @code{used}, @code{unused}, @code{deprecated},
1810 @code{weak}, @code{malloc}, @code{alias}, @code{warn_unused_result},
1811 @code{nonnull}, @code{gnu_inline}, @code{externally_visible},
1812 @code{hot}, @code{cold}, @code{artificial}, @code{error}
1814 Several other attributes are defined for functions on particular
1815 target systems. Other attributes, including @code{section} are
1816 supported for variables declarations (@pxref{Variable Attributes}) and
1817 for types (@pxref{Type Attributes}).
1819 You may also specify attributes with @samp{__} preceding and following
1820 each keyword. This allows you to use them in header files without
1821 being concerned about a possible macro of the same name. For example,
1822 you may use @code{__noreturn__} instead of @code{noreturn}.
1824 @xref{Attribute Syntax}, for details of the exact syntax for using
1828 @c Keep this table alphabetized by attribute name. Treat _ as space.
1830 @item alias ("@var{target}")
1831 @cindex @code{alias} attribute
1832 The @code{alias} attribute causes the declaration to be emitted as an
1833 alias for another symbol, which must be specified. For instance,
1836 void __f () @{ /* @r{Do something.} */; @}
1837 void f () __attribute__ ((weak, alias ("__f")));
1840 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
1841 mangled name for the target must be used. It is an error if @samp{__f}
1842 is not defined in the same translation unit.
1844 Not all target machines support this attribute.
1846 @item aligned (@var{alignment})
1847 @cindex @code{aligned} attribute
1848 This attribute specifies a minimum alignment for the function,
1851 You cannot use this attribute to decrease the alignment of a function,
1852 only to increase it. However, when you explicitly specify a function
1853 alignment this will override the effect of the
1854 @option{-falign-functions} (@pxref{Optimize Options}) option for this
1857 Note that the effectiveness of @code{aligned} attributes may be
1858 limited by inherent limitations in your linker. On many systems, the
1859 linker is only able to arrange for functions to be aligned up to a
1860 certain maximum alignment. (For some linkers, the maximum supported
1861 alignment may be very very small.) See your linker documentation for
1862 further information.
1864 The @code{aligned} attribute can also be used for variables and fields
1865 (@pxref{Variable Attributes}.)
1868 @cindex @code{alloc_size} attribute
1869 The @code{alloc_size} attribute is used to tell the compiler that the
1870 function return value points to memory, where the size is given by
1871 one or two of the functions parameters. GCC uses this
1872 information to improve the correctness of @code{__builtin_object_size}.
1874 The function parameter(s) denoting the allocated size are specified by
1875 one or two integer arguments supplied to the attribute. The allocated size
1876 is either the value of the single function argument specified or the product
1877 of the two function arguments specified. Argument numbering starts at
1883 void* my_calloc(size_t, size_t) __attribute__((alloc_size(1,2)))
1884 void my_realloc(void*, size_t) __attribute__((alloc_size(2)))
1887 declares that my_calloc will return memory of the size given by
1888 the product of parameter 1 and 2 and that my_realloc will return memory
1889 of the size given by parameter 2.
1892 @cindex @code{always_inline} function attribute
1893 Generally, functions are not inlined unless optimization is specified.
1894 For functions declared inline, this attribute inlines the function even
1895 if no optimization level was specified.
1898 @cindex @code{gnu_inline} function attribute
1899 This attribute should be used with a function which is also declared
1900 with the @code{inline} keyword. It directs GCC to treat the function
1901 as if it were defined in gnu89 mode even when compiling in C99 or
1904 If the function is declared @code{extern}, then this definition of the
1905 function is used only for inlining. In no case is the function
1906 compiled as a standalone function, not even if you take its address
1907 explicitly. Such an address becomes an external reference, as if you
1908 had only declared the function, and had not defined it. This has
1909 almost the effect of a macro. The way to use this is to put a
1910 function definition in a header file with this attribute, and put
1911 another copy of the function, without @code{extern}, in a library
1912 file. The definition in the header file will cause most calls to the
1913 function to be inlined. If any uses of the function remain, they will
1914 refer to the single copy in the library. Note that the two
1915 definitions of the functions need not be precisely the same, although
1916 if they do not have the same effect your program may behave oddly.
1918 In C, if the function is neither @code{extern} nor @code{static}, then
1919 the function is compiled as a standalone function, as well as being
1920 inlined where possible.
1922 This is how GCC traditionally handled functions declared
1923 @code{inline}. Since ISO C99 specifies a different semantics for
1924 @code{inline}, this function attribute is provided as a transition
1925 measure and as a useful feature in its own right. This attribute is
1926 available in GCC 4.1.3 and later. It is available if either of the
1927 preprocessor macros @code{__GNUC_GNU_INLINE__} or
1928 @code{__GNUC_STDC_INLINE__} are defined. @xref{Inline,,An Inline
1929 Function is As Fast As a Macro}.
1931 In C++, this attribute does not depend on @code{extern} in any way,
1932 but it still requires the @code{inline} keyword to enable its special
1935 @cindex @code{artificial} function attribute
1937 This attribute is useful for small inline wrappers which if possible
1938 should appear during debugging as a unit, depending on the debug
1939 info format it will either mean marking the function as artificial
1940 or using the caller location for all instructions within the inlined
1943 @cindex @code{flatten} function attribute
1945 Generally, inlining into a function is limited. For a function marked with
1946 this attribute, every call inside this function will be inlined, if possible.
1947 Whether the function itself is considered for inlining depends on its size and
1948 the current inlining parameters. The @code{flatten} attribute only works
1949 reliably in unit-at-a-time mode.
1951 @item error ("@var{message}")
1952 @cindex @code{error} function attribute
1953 If this attribute is used on a function declaration and a call to such a function
1954 is not eliminated through dead code elimination or other optimizations, an error
1955 which will include @var{message} will be diagnosed. This is useful
1956 for compile time checking, especially together with @code{__builtin_constant_p}
1957 and inline functions where checking the inline function arguments is not
1958 possible through @code{extern char [(condition) ? 1 : -1];} tricks.
1959 While it is possible to leave the function undefined and thus invoke
1960 a link failure, when using this attribute the problem will be diagnosed
1961 earlier and with exact location of the call even in presence of inline
1962 functions or when not emitting debugging information.
1964 @item warning ("@var{message}")
1965 @cindex @code{warning} function attribute
1966 If this attribute is used on a function declaration and a call to such a function
1967 is not eliminated through dead code elimination or other optimizations, a warning
1968 which will include @var{message} will be diagnosed. This is useful
1969 for compile time checking, especially together with @code{__builtin_constant_p}
1970 and inline functions. While it is possible to define the function with
1971 a message in @code{.gnu.warning*} section, when using this attribute the problem
1972 will be diagnosed earlier and with exact location of the call even in presence
1973 of inline functions or when not emitting debugging information.
1976 @cindex functions that do pop the argument stack on the 386
1978 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1979 assume that the calling function will pop off the stack space used to
1980 pass arguments. This is
1981 useful to override the effects of the @option{-mrtd} switch.
1984 @cindex @code{const} function attribute
1985 Many functions do not examine any values except their arguments, and
1986 have no effects except the return value. Basically this is just slightly
1987 more strict class than the @code{pure} attribute below, since function is not
1988 allowed to read global memory.
1990 @cindex pointer arguments
1991 Note that a function that has pointer arguments and examines the data
1992 pointed to must @emph{not} be declared @code{const}. Likewise, a
1993 function that calls a non-@code{const} function usually must not be
1994 @code{const}. It does not make sense for a @code{const} function to
1997 The attribute @code{const} is not implemented in GCC versions earlier
1998 than 2.5. An alternative way to declare that a function has no side
1999 effects, which works in the current version and in some older versions,
2003 typedef int intfn ();
2005 extern const intfn square;
2008 This approach does not work in GNU C++ from 2.6.0 on, since the language
2009 specifies that the @samp{const} must be attached to the return value.
2013 @itemx constructor (@var{priority})
2014 @itemx destructor (@var{priority})
2015 @cindex @code{constructor} function attribute
2016 @cindex @code{destructor} function attribute
2017 The @code{constructor} attribute causes the function to be called
2018 automatically before execution enters @code{main ()}. Similarly, the
2019 @code{destructor} attribute causes the function to be called
2020 automatically after @code{main ()} has completed or @code{exit ()} has
2021 been called. Functions with these attributes are useful for
2022 initializing data that will be used implicitly during the execution of
2025 You may provide an optional integer priority to control the order in
2026 which constructor and destructor functions are run. A constructor
2027 with a smaller priority number runs before a constructor with a larger
2028 priority number; the opposite relationship holds for destructors. So,
2029 if you have a constructor that allocates a resource and a destructor
2030 that deallocates the same resource, both functions typically have the
2031 same priority. The priorities for constructor and destructor
2032 functions are the same as those specified for namespace-scope C++
2033 objects (@pxref{C++ Attributes}).
2035 These attributes are not currently implemented for Objective-C@.
2038 @cindex @code{deprecated} attribute.
2039 The @code{deprecated} attribute results in a warning if the function
2040 is used anywhere in the source file. This is useful when identifying
2041 functions that are expected to be removed in a future version of a
2042 program. The warning also includes the location of the declaration
2043 of the deprecated function, to enable users to easily find further
2044 information about why the function is deprecated, or what they should
2045 do instead. Note that the warnings only occurs for uses:
2048 int old_fn () __attribute__ ((deprecated));
2050 int (*fn_ptr)() = old_fn;
2053 results in a warning on line 3 but not line 2.
2055 The @code{deprecated} attribute can also be used for variables and
2056 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2059 @cindex @code{__declspec(dllexport)}
2060 On Microsoft Windows targets and Symbian OS targets the
2061 @code{dllexport} attribute causes the compiler to provide a global
2062 pointer to a pointer in a DLL, so that it can be referenced with the
2063 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
2064 name is formed by combining @code{_imp__} and the function or variable
2067 You can use @code{__declspec(dllexport)} as a synonym for
2068 @code{__attribute__ ((dllexport))} for compatibility with other
2071 On systems that support the @code{visibility} attribute, this
2072 attribute also implies ``default'' visibility. It is an error to
2073 explicitly specify any other visibility.
2075 Currently, the @code{dllexport} attribute is ignored for inlined
2076 functions, unless the @option{-fkeep-inline-functions} flag has been
2077 used. The attribute is also ignored for undefined symbols.
2079 When applied to C++ classes, the attribute marks defined non-inlined
2080 member functions and static data members as exports. Static consts
2081 initialized in-class are not marked unless they are also defined
2084 For Microsoft Windows targets there are alternative methods for
2085 including the symbol in the DLL's export table such as using a
2086 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
2087 the @option{--export-all} linker flag.
2090 @cindex @code{__declspec(dllimport)}
2091 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
2092 attribute causes the compiler to reference a function or variable via
2093 a global pointer to a pointer that is set up by the DLL exporting the
2094 symbol. The attribute implies @code{extern}. On Microsoft Windows
2095 targets, the pointer name is formed by combining @code{_imp__} and the
2096 function or variable name.
2098 You can use @code{__declspec(dllimport)} as a synonym for
2099 @code{__attribute__ ((dllimport))} for compatibility with other
2102 On systems that support the @code{visibility} attribute, this
2103 attribute also implies ``default'' visibility. It is an error to
2104 explicitly specify any other visibility.
2106 Currently, the attribute is ignored for inlined functions. If the
2107 attribute is applied to a symbol @emph{definition}, an error is reported.
2108 If a symbol previously declared @code{dllimport} is later defined, the
2109 attribute is ignored in subsequent references, and a warning is emitted.
2110 The attribute is also overridden by a subsequent declaration as
2113 When applied to C++ classes, the attribute marks non-inlined
2114 member functions and static data members as imports. However, the
2115 attribute is ignored for virtual methods to allow creation of vtables
2118 On the SH Symbian OS target the @code{dllimport} attribute also has
2119 another affect---it can cause the vtable and run-time type information
2120 for a class to be exported. This happens when the class has a
2121 dllimport'ed constructor or a non-inline, non-pure virtual function
2122 and, for either of those two conditions, the class also has a inline
2123 constructor or destructor and has a key function that is defined in
2124 the current translation unit.
2126 For Microsoft Windows based targets the use of the @code{dllimport}
2127 attribute on functions is not necessary, but provides a small
2128 performance benefit by eliminating a thunk in the DLL@. The use of the
2129 @code{dllimport} attribute on imported variables was required on older
2130 versions of the GNU linker, but can now be avoided by passing the
2131 @option{--enable-auto-import} switch to the GNU linker. As with
2132 functions, using the attribute for a variable eliminates a thunk in
2135 One drawback to using this attribute is that a pointer to a
2136 @emph{variable} marked as @code{dllimport} cannot be used as a constant
2137 address. However, a pointer to a @emph{function} with the
2138 @code{dllimport} attribute can be used as a constant initializer; in
2139 this case, the address of a stub function in the import lib is
2140 referenced. On Microsoft Windows targets, the attribute can be disabled
2141 for functions by setting the @option{-mnop-fun-dllimport} flag.
2144 @cindex eight bit data on the H8/300, H8/300H, and H8S
2145 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2146 variable should be placed into the eight bit data section.
2147 The compiler will generate more efficient code for certain operations
2148 on data in the eight bit data area. Note the eight bit data area is limited to
2151 You must use GAS and GLD from GNU binutils version 2.7 or later for
2152 this attribute to work correctly.
2154 @item exception_handler
2155 @cindex exception handler functions on the Blackfin processor
2156 Use this attribute on the Blackfin to indicate that the specified function
2157 is an exception handler. The compiler will generate function entry and
2158 exit sequences suitable for use in an exception handler when this
2159 attribute is present.
2162 @cindex functions which handle memory bank switching
2163 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
2164 use a calling convention that takes care of switching memory banks when
2165 entering and leaving a function. This calling convention is also the
2166 default when using the @option{-mlong-calls} option.
2168 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
2169 to call and return from a function.
2171 On 68HC11 the compiler will generate a sequence of instructions
2172 to invoke a board-specific routine to switch the memory bank and call the
2173 real function. The board-specific routine simulates a @code{call}.
2174 At the end of a function, it will jump to a board-specific routine
2175 instead of using @code{rts}. The board-specific return routine simulates
2179 @cindex functions that pop the argument stack on the 386
2180 On the Intel 386, the @code{fastcall} attribute causes the compiler to
2181 pass the first argument (if of integral type) in the register ECX and
2182 the second argument (if of integral type) in the register EDX@. Subsequent
2183 and other typed arguments are passed on the stack. The called function will
2184 pop the arguments off the stack. If the number of arguments is variable all
2185 arguments are pushed on the stack.
2187 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2188 @cindex @code{format} function attribute
2190 The @code{format} attribute specifies that a function takes @code{printf},
2191 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
2192 should be type-checked against a format string. For example, the
2197 my_printf (void *my_object, const char *my_format, ...)
2198 __attribute__ ((format (printf, 2, 3)));
2202 causes the compiler to check the arguments in calls to @code{my_printf}
2203 for consistency with the @code{printf} style format string argument
2206 The parameter @var{archetype} determines how the format string is
2207 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime},
2208 @code{gnu_printf}, @code{gnu_scanf}, @code{gnu_strftime} or
2209 @code{strfmon}. (You can also use @code{__printf__},
2210 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) On
2211 MinGW targets, @code{ms_printf}, @code{ms_scanf}, and
2212 @code{ms_strftime} are also present.
2213 @var{archtype} values such as @code{printf} refer to the formats accepted
2214 by the system's C run-time library, while @code{gnu_} values always refer
2215 to the formats accepted by the GNU C Library. On Microsoft Windows
2216 targets, @code{ms_} values refer to the formats accepted by the
2217 @file{msvcrt.dll} library.
2218 The parameter @var{string-index}
2219 specifies which argument is the format string argument (starting
2220 from 1), while @var{first-to-check} is the number of the first
2221 argument to check against the format string. For functions
2222 where the arguments are not available to be checked (such as
2223 @code{vprintf}), specify the third parameter as zero. In this case the
2224 compiler only checks the format string for consistency. For
2225 @code{strftime} formats, the third parameter is required to be zero.
2226 Since non-static C++ methods have an implicit @code{this} argument, the
2227 arguments of such methods should be counted from two, not one, when
2228 giving values for @var{string-index} and @var{first-to-check}.
2230 In the example above, the format string (@code{my_format}) is the second
2231 argument of the function @code{my_print}, and the arguments to check
2232 start with the third argument, so the correct parameters for the format
2233 attribute are 2 and 3.
2235 @opindex ffreestanding
2236 @opindex fno-builtin
2237 The @code{format} attribute allows you to identify your own functions
2238 which take format strings as arguments, so that GCC can check the
2239 calls to these functions for errors. The compiler always (unless
2240 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
2241 for the standard library functions @code{printf}, @code{fprintf},
2242 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2243 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2244 warnings are requested (using @option{-Wformat}), so there is no need to
2245 modify the header file @file{stdio.h}. In C99 mode, the functions
2246 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2247 @code{vsscanf} are also checked. Except in strictly conforming C
2248 standard modes, the X/Open function @code{strfmon} is also checked as
2249 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2250 @xref{C Dialect Options,,Options Controlling C Dialect}.
2252 The target may provide additional types of format checks.
2253 @xref{Target Format Checks,,Format Checks Specific to Particular
2256 @item format_arg (@var{string-index})
2257 @cindex @code{format_arg} function attribute
2258 @opindex Wformat-nonliteral
2259 The @code{format_arg} attribute specifies that a function takes a format
2260 string for a @code{printf}, @code{scanf}, @code{strftime} or
2261 @code{strfmon} style function and modifies it (for example, to translate
2262 it into another language), so the result can be passed to a
2263 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2264 function (with the remaining arguments to the format function the same
2265 as they would have been for the unmodified string). For example, the
2270 my_dgettext (char *my_domain, const char *my_format)
2271 __attribute__ ((format_arg (2)));
2275 causes the compiler to check the arguments in calls to a @code{printf},
2276 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2277 format string argument is a call to the @code{my_dgettext} function, for
2278 consistency with the format string argument @code{my_format}. If the
2279 @code{format_arg} attribute had not been specified, all the compiler
2280 could tell in such calls to format functions would be that the format
2281 string argument is not constant; this would generate a warning when
2282 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2283 without the attribute.
2285 The parameter @var{string-index} specifies which argument is the format
2286 string argument (starting from one). Since non-static C++ methods have
2287 an implicit @code{this} argument, the arguments of such methods should
2288 be counted from two.
2290 The @code{format-arg} attribute allows you to identify your own
2291 functions which modify format strings, so that GCC can check the
2292 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2293 type function whose operands are a call to one of your own function.
2294 The compiler always treats @code{gettext}, @code{dgettext}, and
2295 @code{dcgettext} in this manner except when strict ISO C support is
2296 requested by @option{-ansi} or an appropriate @option{-std} option, or
2297 @option{-ffreestanding} or @option{-fno-builtin}
2298 is used. @xref{C Dialect Options,,Options
2299 Controlling C Dialect}.
2301 @item function_vector
2302 @cindex calling functions through the function vector on H8/300, M16C, M32C and SH2A processors
2303 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2304 function should be called through the function vector. Calling a
2305 function through the function vector will reduce code size, however;
2306 the function vector has a limited size (maximum 128 entries on the H8/300
2307 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2309 In SH2A target, this attribute declares a function to be called using the
2310 TBR relative addressing mode. The argument to this attribute is the entry
2311 number of the same function in a vector table containing all the TBR
2312 relative addressable functions. For the successful jump, register TBR
2313 should contain the start address of this TBR relative vector table.
2314 In the startup routine of the user application, user needs to care of this
2315 TBR register initialization. The TBR relative vector table can have at
2316 max 256 function entries. The jumps to these functions will be generated
2317 using a SH2A specific, non delayed branch instruction JSR/N @@(disp8,TBR).
2318 You must use GAS and GLD from GNU binutils version 2.7 or later for
2319 this attribute to work correctly.
2321 Please refer the example of M16C target, to see the use of this
2322 attribute while declaring a function,
2324 In an application, for a function being called once, this attribute will
2325 save at least 8 bytes of code; and if other successive calls are being
2326 made to the same function, it will save 2 bytes of code per each of these
2329 On M16C/M32C targets, the @code{function_vector} attribute declares a
2330 special page subroutine call function. Use of this attribute reduces
2331 the code size by 2 bytes for each call generated to the
2332 subroutine. The argument to the attribute is the vector number entry
2333 from the special page vector table which contains the 16 low-order
2334 bits of the subroutine's entry address. Each vector table has special
2335 page number (18 to 255) which are used in @code{jsrs} instruction.
2336 Jump addresses of the routines are generated by adding 0x0F0000 (in
2337 case of M16C targets) or 0xFF0000 (in case of M32C targets), to the 2
2338 byte addresses set in the vector table. Therefore you need to ensure
2339 that all the special page vector routines should get mapped within the
2340 address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF
2343 In the following example 2 bytes will be saved for each call to
2344 function @code{foo}.
2347 void foo (void) __attribute__((function_vector(0x18)));
2358 If functions are defined in one file and are called in another file,
2359 then be sure to write this declaration in both files.
2361 This attribute is ignored for R8C target.
2364 @cindex interrupt handler functions
2365 Use this attribute on the ARM, AVR, CRX, M32C, M32R/D, m68k, MS1,
2366 and Xstormy16 ports to indicate that the specified function is an
2367 interrupt handler. The compiler will generate function entry and exit
2368 sequences suitable for use in an interrupt handler when this attribute
2371 Note, interrupt handlers for the Blackfin, H8/300, H8/300H, H8S, and
2372 SH processors can be specified via the @code{interrupt_handler} attribute.
2374 Note, on the AVR, interrupts will be enabled inside the function.
2376 Note, for the ARM, you can specify the kind of interrupt to be handled by
2377 adding an optional parameter to the interrupt attribute like this:
2380 void f () __attribute__ ((interrupt ("IRQ")));
2383 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2385 On ARMv7-M the interrupt type is ignored, and the attribute means the function
2386 may be called with a word aligned stack pointer.
2388 @item interrupt_handler
2389 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
2390 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
2391 indicate that the specified function is an interrupt handler. The compiler
2392 will generate function entry and exit sequences suitable for use in an
2393 interrupt handler when this attribute is present.
2395 @item interrupt_thread
2396 @cindex interrupt thread functions on fido
2397 Use this attribute on fido, a subarchitecture of the m68k, to indicate
2398 that the specified function is an interrupt handler that is designed
2399 to run as a thread. The compiler omits generate prologue/epilogue
2400 sequences and replaces the return instruction with a @code{sleep}
2401 instruction. This attribute is available only on fido.
2404 @cindex User stack pointer in interrupts on the Blackfin
2405 When used together with @code{interrupt_handler}, @code{exception_handler}
2406 or @code{nmi_handler}, code will be generated to load the stack pointer
2407 from the USP register in the function prologue.
2410 @cindex @code{l1_text} function attribute
2411 This attribute specifies a function to be placed into L1 Instruction
2412 SRAM@. The function will be put into a specific section named @code{.l1.text}.
2413 With @option{-mfdpic}, function calls with a such function as the callee
2414 or caller will use inlined PLT.
2416 @item long_call/short_call
2417 @cindex indirect calls on ARM
2418 This attribute specifies how a particular function is called on
2419 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2420 command line switch and @code{#pragma long_calls} settings. The
2421 @code{long_call} attribute indicates that the function might be far
2422 away from the call site and require a different (more expensive)
2423 calling sequence. The @code{short_call} attribute always places
2424 the offset to the function from the call site into the @samp{BL}
2425 instruction directly.
2427 @item longcall/shortcall
2428 @cindex functions called via pointer on the RS/6000 and PowerPC
2429 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
2430 indicates that the function might be far away from the call site and
2431 require a different (more expensive) calling sequence. The
2432 @code{shortcall} attribute indicates that the function is always close
2433 enough for the shorter calling sequence to be used. These attributes
2434 override both the @option{-mlongcall} switch and, on the RS/6000 and
2435 PowerPC, the @code{#pragma longcall} setting.
2437 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2438 calls are necessary.
2440 @item long_call/near/far
2441 @cindex indirect calls on MIPS
2442 These attributes specify how a particular function is called on MIPS@.
2443 The attributes override the @option{-mlong-calls} (@pxref{MIPS Options})
2444 command-line switch. The @code{long_call} and @code{far} attributes are
2445 synonyms, and cause the compiler to always call
2446 the function by first loading its address into a register, and then using
2447 the contents of that register. The @code{near} attribute has the opposite
2448 effect; it specifies that non-PIC calls should be made using the more
2449 efficient @code{jal} instruction.
2452 @cindex @code{malloc} attribute
2453 The @code{malloc} attribute is used to tell the compiler that a function
2454 may be treated as if any non-@code{NULL} pointer it returns cannot
2455 alias any other pointer valid when the function returns.
2456 This will often improve optimization.
2457 Standard functions with this property include @code{malloc} and
2458 @code{calloc}. @code{realloc}-like functions have this property as
2459 long as the old pointer is never referred to (including comparing it
2460 to the new pointer) after the function returns a non-@code{NULL}
2463 @item mips16/nomips16
2464 @cindex @code{mips16} attribute
2465 @cindex @code{nomips16} attribute
2467 On MIPS targets, you can use the @code{mips16} and @code{nomips16}
2468 function attributes to locally select or turn off MIPS16 code generation.
2469 A function with the @code{mips16} attribute is emitted as MIPS16 code,
2470 while MIPS16 code generation is disabled for functions with the
2471 @code{nomips16} attribute. These attributes override the
2472 @option{-mips16} and @option{-mno-mips16} options on the command line
2473 (@pxref{MIPS Options}).
2475 When compiling files containing mixed MIPS16 and non-MIPS16 code, the
2476 preprocessor symbol @code{__mips16} reflects the setting on the command line,
2477 not that within individual functions. Mixed MIPS16 and non-MIPS16 code
2478 may interact badly with some GCC extensions such as @code{__builtin_apply}
2479 (@pxref{Constructing Calls}).
2481 @item model (@var{model-name})
2482 @cindex function addressability on the M32R/D
2483 @cindex variable addressability on the IA-64
2485 On the M32R/D, use this attribute to set the addressability of an
2486 object, and of the code generated for a function. The identifier
2487 @var{model-name} is one of @code{small}, @code{medium}, or
2488 @code{large}, representing each of the code models.
2490 Small model objects live in the lower 16MB of memory (so that their
2491 addresses can be loaded with the @code{ld24} instruction), and are
2492 callable with the @code{bl} instruction.
2494 Medium model objects may live anywhere in the 32-bit address space (the
2495 compiler will generate @code{seth/add3} instructions to load their addresses),
2496 and are callable with the @code{bl} instruction.
2498 Large model objects may live anywhere in the 32-bit address space (the
2499 compiler will generate @code{seth/add3} instructions to load their addresses),
2500 and may not be reachable with the @code{bl} instruction (the compiler will
2501 generate the much slower @code{seth/add3/jl} instruction sequence).
2503 On IA-64, use this attribute to set the addressability of an object.
2504 At present, the only supported identifier for @var{model-name} is
2505 @code{small}, indicating addressability via ``small'' (22-bit)
2506 addresses (so that their addresses can be loaded with the @code{addl}
2507 instruction). Caveat: such addressing is by definition not position
2508 independent and hence this attribute must not be used for objects
2509 defined by shared libraries.
2512 @cindex function without a prologue/epilogue code
2513 Use this attribute on the ARM, AVR, IP2K and SPU ports to indicate that
2514 the specified function does not need prologue/epilogue sequences generated by
2515 the compiler. It is up to the programmer to provide these sequences. The
2516 only statements that can be safely included in naked functions are
2517 @code{asm} statements that do not have operands. All other statements,
2518 including declarations of local variables, @code{if} statements, and so
2519 forth, should be avoided. Naked functions should be used to implement the
2520 body of an assembly function, while allowing the compiler to construct
2521 the requisite function declaration for the assembler.
2524 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2525 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2526 use the normal calling convention based on @code{jsr} and @code{rts}.
2527 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2531 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2532 Use this attribute together with @code{interrupt_handler},
2533 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2534 entry code should enable nested interrupts or exceptions.
2537 @cindex NMI handler functions on the Blackfin processor
2538 Use this attribute on the Blackfin to indicate that the specified function
2539 is an NMI handler. The compiler will generate function entry and
2540 exit sequences suitable for use in an NMI handler when this
2541 attribute is present.
2543 @item no_instrument_function
2544 @cindex @code{no_instrument_function} function attribute
2545 @opindex finstrument-functions
2546 If @option{-finstrument-functions} is given, profiling function calls will
2547 be generated at entry and exit of most user-compiled functions.
2548 Functions with this attribute will not be so instrumented.
2551 @cindex @code{noinline} function attribute
2552 This function attribute prevents a function from being considered for
2554 @c Don't enumerate the optimizations by name here; we try to be
2555 @c future-compatible with this mechanism.
2556 If the function does not have side-effects, there are optimizations
2557 other than inlining that causes function calls to be optimized away,
2558 although the function call is live. To keep such calls from being
2563 (@pxref{Extended Asm}) in the called function, to serve as a special
2566 @item nonnull (@var{arg-index}, @dots{})
2567 @cindex @code{nonnull} function attribute
2568 The @code{nonnull} attribute specifies that some function parameters should
2569 be non-null pointers. For instance, the declaration:
2573 my_memcpy (void *dest, const void *src, size_t len)
2574 __attribute__((nonnull (1, 2)));
2578 causes the compiler to check that, in calls to @code{my_memcpy},
2579 arguments @var{dest} and @var{src} are non-null. If the compiler
2580 determines that a null pointer is passed in an argument slot marked
2581 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2582 is issued. The compiler may also choose to make optimizations based
2583 on the knowledge that certain function arguments will not be null.
2585 If no argument index list is given to the @code{nonnull} attribute,
2586 all pointer arguments are marked as non-null. To illustrate, the
2587 following declaration is equivalent to the previous example:
2591 my_memcpy (void *dest, const void *src, size_t len)
2592 __attribute__((nonnull));
2596 @cindex @code{noreturn} function attribute
2597 A few standard library functions, such as @code{abort} and @code{exit},
2598 cannot return. GCC knows this automatically. Some programs define
2599 their own functions that never return. You can declare them
2600 @code{noreturn} to tell the compiler this fact. For example,
2604 void fatal () __attribute__ ((noreturn));
2607 fatal (/* @r{@dots{}} */)
2609 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2615 The @code{noreturn} keyword tells the compiler to assume that
2616 @code{fatal} cannot return. It can then optimize without regard to what
2617 would happen if @code{fatal} ever did return. This makes slightly
2618 better code. More importantly, it helps avoid spurious warnings of
2619 uninitialized variables.
2621 The @code{noreturn} keyword does not affect the exceptional path when that
2622 applies: a @code{noreturn}-marked function may still return to the caller
2623 by throwing an exception or calling @code{longjmp}.
2625 Do not assume that registers saved by the calling function are
2626 restored before calling the @code{noreturn} function.
2628 It does not make sense for a @code{noreturn} function to have a return
2629 type other than @code{void}.
2631 The attribute @code{noreturn} is not implemented in GCC versions
2632 earlier than 2.5. An alternative way to declare that a function does
2633 not return, which works in the current version and in some older
2634 versions, is as follows:
2637 typedef void voidfn ();
2639 volatile voidfn fatal;
2642 This approach does not work in GNU C++.
2645 @cindex @code{nothrow} function attribute
2646 The @code{nothrow} attribute is used to inform the compiler that a
2647 function cannot throw an exception. For example, most functions in
2648 the standard C library can be guaranteed not to throw an exception
2649 with the notable exceptions of @code{qsort} and @code{bsearch} that
2650 take function pointer arguments. The @code{nothrow} attribute is not
2651 implemented in GCC versions earlier than 3.3.
2654 @cindex @code{pure} function attribute
2655 Many functions have no effects except the return value and their
2656 return value depends only on the parameters and/or global variables.
2657 Such a function can be subject
2658 to common subexpression elimination and loop optimization just as an
2659 arithmetic operator would be. These functions should be declared
2660 with the attribute @code{pure}. For example,
2663 int square (int) __attribute__ ((pure));
2667 says that the hypothetical function @code{square} is safe to call
2668 fewer times than the program says.
2670 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2671 Interesting non-pure functions are functions with infinite loops or those
2672 depending on volatile memory or other system resource, that may change between
2673 two consecutive calls (such as @code{feof} in a multithreading environment).
2675 The attribute @code{pure} is not implemented in GCC versions earlier
2679 @cindex @code{hot} function attribute
2680 The @code{hot} attribute is used to inform the compiler that a function is a
2681 hot spot of the compiled program. The function is optimized more aggressively
2682 and on many target it is placed into special subsection of the text section so
2683 all hot functions appears close together improving locality.
2685 When profile feedback is available, via @option{-fprofile-use}, hot functions
2686 are automatically detected and this attribute is ignored.
2688 The @code{hot} attribute is not implemented in GCC versions earlier than 4.3.
2691 @cindex @code{cold} function attribute
2692 The @code{cold} attribute is used to inform the compiler that a function is
2693 unlikely executed. The function is optimized for size rather than speed and on
2694 many targets it is placed into special subsection of the text section so all
2695 cold functions appears close together improving code locality of non-cold parts
2696 of program. The paths leading to call of cold functions within code are marked
2697 as unlikely by the branch prediction mechanism. It is thus useful to mark
2698 functions used to handle unlikely conditions, such as @code{perror}, as cold to
2699 improve optimization of hot functions that do call marked functions in rare
2702 When profile feedback is available, via @option{-fprofile-use}, hot functions
2703 are automatically detected and this attribute is ignored.
2705 The @code{hot} attribute is not implemented in GCC versions earlier than 4.3.
2707 @item regparm (@var{number})
2708 @cindex @code{regparm} attribute
2709 @cindex functions that are passed arguments in registers on the 386
2710 On the Intel 386, the @code{regparm} attribute causes the compiler to
2711 pass arguments number one to @var{number} if they are of integral type
2712 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2713 take a variable number of arguments will continue to be passed all of their
2714 arguments on the stack.
2716 Beware that on some ELF systems this attribute is unsuitable for
2717 global functions in shared libraries with lazy binding (which is the
2718 default). Lazy binding will send the first call via resolving code in
2719 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2720 per the standard calling conventions. Solaris 8 is affected by this.
2721 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2722 safe since the loaders there save all registers. (Lazy binding can be
2723 disabled with the linker or the loader if desired, to avoid the
2727 @cindex @code{sseregparm} attribute
2728 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2729 causes the compiler to pass up to 3 floating point arguments in
2730 SSE registers instead of on the stack. Functions that take a
2731 variable number of arguments will continue to pass all of their
2732 floating point arguments on the stack.
2734 @item force_align_arg_pointer
2735 @cindex @code{force_align_arg_pointer} attribute
2736 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
2737 applied to individual function definitions, generating an alternate
2738 prologue and epilogue that realigns the runtime stack. This supports
2739 mixing legacy codes that run with a 4-byte aligned stack with modern
2740 codes that keep a 16-byte stack for SSE compatibility. The alternate
2741 prologue and epilogue are slower and bigger than the regular ones, and
2742 the alternate prologue requires a scratch register; this lowers the
2743 number of registers available if used in conjunction with the
2744 @code{regparm} attribute. The @code{force_align_arg_pointer}
2745 attribute is incompatible with nested functions; this is considered a
2749 @cindex @code{resbank} attribute
2750 On the SH2A target, this attribute enables the high-speed register
2751 saving and restoration using a register bank for @code{interrupt_handler}
2752 routines. Saving to the bank is performed automatcially after the CPU
2753 accepts an interrupt that uses a register bank.
2755 The nineteen 32-bit registers comprising general register R0 to R14,
2756 control register GBR, and system registers MACH, MACL, and PR and the
2757 vector table address offset are saved into a register bank. Register
2758 banks are stacked in first-in last-out (FILO) sequence. Restoration
2759 from the bank is executed by issuing a RESBANK instruction.
2762 @cindex @code{returns_twice} attribute
2763 The @code{returns_twice} attribute tells the compiler that a function may
2764 return more than one time. The compiler will ensure that all registers
2765 are dead before calling such a function and will emit a warning about
2766 the variables that may be clobbered after the second return from the
2767 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2768 The @code{longjmp}-like counterpart of such function, if any, might need
2769 to be marked with the @code{noreturn} attribute.
2772 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2773 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2774 all registers except the stack pointer should be saved in the prologue
2775 regardless of whether they are used or not.
2777 @item section ("@var{section-name}")
2778 @cindex @code{section} function attribute
2779 Normally, the compiler places the code it generates in the @code{text} section.
2780 Sometimes, however, you need additional sections, or you need certain
2781 particular functions to appear in special sections. The @code{section}
2782 attribute specifies that a function lives in a particular section.
2783 For example, the declaration:
2786 extern void foobar (void) __attribute__ ((section ("bar")));
2790 puts the function @code{foobar} in the @code{bar} section.
2792 Some file formats do not support arbitrary sections so the @code{section}
2793 attribute is not available on all platforms.
2794 If you need to map the entire contents of a module to a particular
2795 section, consider using the facilities of the linker instead.
2798 @cindex @code{sentinel} function attribute
2799 This function attribute ensures that a parameter in a function call is
2800 an explicit @code{NULL}. The attribute is only valid on variadic
2801 functions. By default, the sentinel is located at position zero, the
2802 last parameter of the function call. If an optional integer position
2803 argument P is supplied to the attribute, the sentinel must be located at
2804 position P counting backwards from the end of the argument list.
2807 __attribute__ ((sentinel))
2809 __attribute__ ((sentinel(0)))
2812 The attribute is automatically set with a position of 0 for the built-in
2813 functions @code{execl} and @code{execlp}. The built-in function
2814 @code{execle} has the attribute set with a position of 1.
2816 A valid @code{NULL} in this context is defined as zero with any pointer
2817 type. If your system defines the @code{NULL} macro with an integer type
2818 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2819 with a copy that redefines NULL appropriately.
2821 The warnings for missing or incorrect sentinels are enabled with
2825 See long_call/short_call.
2828 See longcall/shortcall.
2831 @cindex signal handler functions on the AVR processors
2832 Use this attribute on the AVR to indicate that the specified
2833 function is a signal handler. The compiler will generate function
2834 entry and exit sequences suitable for use in a signal handler when this
2835 attribute is present. Interrupts will be disabled inside the function.
2838 Use this attribute on the SH to indicate an @code{interrupt_handler}
2839 function should switch to an alternate stack. It expects a string
2840 argument that names a global variable holding the address of the
2845 void f () __attribute__ ((interrupt_handler,
2846 sp_switch ("alt_stack")));
2850 @cindex functions that pop the argument stack on the 386
2851 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2852 assume that the called function will pop off the stack space used to
2853 pass arguments, unless it takes a variable number of arguments.
2856 @cindex tiny data section on the H8/300H and H8S
2857 Use this attribute on the H8/300H and H8S to indicate that the specified
2858 variable should be placed into the tiny data section.
2859 The compiler will generate more efficient code for loads and stores
2860 on data in the tiny data section. Note the tiny data area is limited to
2861 slightly under 32kbytes of data.
2864 Use this attribute on the SH for an @code{interrupt_handler} to return using
2865 @code{trapa} instead of @code{rte}. This attribute expects an integer
2866 argument specifying the trap number to be used.
2869 @cindex @code{unused} attribute.
2870 This attribute, attached to a function, means that the function is meant
2871 to be possibly unused. GCC will not produce a warning for this
2875 @cindex @code{used} attribute.
2876 This attribute, attached to a function, means that code must be emitted
2877 for the function even if it appears that the function is not referenced.
2878 This is useful, for example, when the function is referenced only in
2882 @cindex @code{version_id} attribute on IA64 HP-UX
2883 This attribute, attached to a global variable or function, renames a
2884 symbol to contain a version string, thus allowing for function level
2885 versioning. HP-UX system header files may use version level functioning
2886 for some system calls.
2889 extern int foo () __attribute__((version_id ("20040821")));
2892 Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}.
2894 @item visibility ("@var{visibility_type}")
2895 @cindex @code{visibility} attribute
2896 This attribute affects the linkage of the declaration to which it is attached.
2897 There are four supported @var{visibility_type} values: default,
2898 hidden, protected or internal visibility.
2901 void __attribute__ ((visibility ("protected")))
2902 f () @{ /* @r{Do something.} */; @}
2903 int i __attribute__ ((visibility ("hidden")));
2906 The possible values of @var{visibility_type} correspond to the
2907 visibility settings in the ELF gABI.
2910 @c keep this list of visibilities in alphabetical order.
2913 Default visibility is the normal case for the object file format.
2914 This value is available for the visibility attribute to override other
2915 options that may change the assumed visibility of entities.
2917 On ELF, default visibility means that the declaration is visible to other
2918 modules and, in shared libraries, means that the declared entity may be
2921 On Darwin, default visibility means that the declaration is visible to
2924 Default visibility corresponds to ``external linkage'' in the language.
2927 Hidden visibility indicates that the entity declared will have a new
2928 form of linkage, which we'll call ``hidden linkage''. Two
2929 declarations of an object with hidden linkage refer to the same object
2930 if they are in the same shared object.
2933 Internal visibility is like hidden visibility, but with additional
2934 processor specific semantics. Unless otherwise specified by the
2935 psABI, GCC defines internal visibility to mean that a function is
2936 @emph{never} called from another module. Compare this with hidden
2937 functions which, while they cannot be referenced directly by other
2938 modules, can be referenced indirectly via function pointers. By
2939 indicating that a function cannot be called from outside the module,
2940 GCC may for instance omit the load of a PIC register since it is known
2941 that the calling function loaded the correct value.
2944 Protected visibility is like default visibility except that it
2945 indicates that references within the defining module will bind to the
2946 definition in that module. That is, the declared entity cannot be
2947 overridden by another module.
2951 All visibilities are supported on many, but not all, ELF targets
2952 (supported when the assembler supports the @samp{.visibility}
2953 pseudo-op). Default visibility is supported everywhere. Hidden
2954 visibility is supported on Darwin targets.
2956 The visibility attribute should be applied only to declarations which
2957 would otherwise have external linkage. The attribute should be applied
2958 consistently, so that the same entity should not be declared with
2959 different settings of the attribute.
2961 In C++, the visibility attribute applies to types as well as functions
2962 and objects, because in C++ types have linkage. A class must not have
2963 greater visibility than its non-static data member types and bases,
2964 and class members default to the visibility of their class. Also, a
2965 declaration without explicit visibility is limited to the visibility
2968 In C++, you can mark member functions and static member variables of a
2969 class with the visibility attribute. This is useful if if you know a
2970 particular method or static member variable should only be used from
2971 one shared object; then you can mark it hidden while the rest of the
2972 class has default visibility. Care must be taken to avoid breaking
2973 the One Definition Rule; for example, it is usually not useful to mark
2974 an inline method as hidden without marking the whole class as hidden.
2976 A C++ namespace declaration can also have the visibility attribute.
2977 This attribute applies only to the particular namespace body, not to
2978 other definitions of the same namespace; it is equivalent to using
2979 @samp{#pragma GCC visibility} before and after the namespace
2980 definition (@pxref{Visibility Pragmas}).
2982 In C++, if a template argument has limited visibility, this
2983 restriction is implicitly propagated to the template instantiation.
2984 Otherwise, template instantiations and specializations default to the
2985 visibility of their template.
2987 If both the template and enclosing class have explicit visibility, the
2988 visibility from the template is used.
2990 @item warn_unused_result
2991 @cindex @code{warn_unused_result} attribute
2992 The @code{warn_unused_result} attribute causes a warning to be emitted
2993 if a caller of the function with this attribute does not use its
2994 return value. This is useful for functions where not checking
2995 the result is either a security problem or always a bug, such as
2999 int fn () __attribute__ ((warn_unused_result));
3002 if (fn () < 0) return -1;
3008 results in warning on line 5.
3011 @cindex @code{weak} attribute
3012 The @code{weak} attribute causes the declaration to be emitted as a weak
3013 symbol rather than a global. This is primarily useful in defining
3014 library functions which can be overridden in user code, though it can
3015 also be used with non-function declarations. Weak symbols are supported
3016 for ELF targets, and also for a.out targets when using the GNU assembler
3020 @itemx weakref ("@var{target}")
3021 @cindex @code{weakref} attribute
3022 The @code{weakref} attribute marks a declaration as a weak reference.
3023 Without arguments, it should be accompanied by an @code{alias} attribute
3024 naming the target symbol. Optionally, the @var{target} may be given as
3025 an argument to @code{weakref} itself. In either case, @code{weakref}
3026 implicitly marks the declaration as @code{weak}. Without a
3027 @var{target}, given as an argument to @code{weakref} or to @code{alias},
3028 @code{weakref} is equivalent to @code{weak}.
3031 static int x() __attribute__ ((weakref ("y")));
3032 /* is equivalent to... */
3033 static int x() __attribute__ ((weak, weakref, alias ("y")));
3035 static int x() __attribute__ ((weakref));
3036 static int x() __attribute__ ((alias ("y")));
3039 A weak reference is an alias that does not by itself require a
3040 definition to be given for the target symbol. If the target symbol is
3041 only referenced through weak references, then the becomes a @code{weak}
3042 undefined symbol. If it is directly referenced, however, then such
3043 strong references prevail, and a definition will be required for the
3044 symbol, not necessarily in the same translation unit.
3046 The effect is equivalent to moving all references to the alias to a
3047 separate translation unit, renaming the alias to the aliased symbol,
3048 declaring it as weak, compiling the two separate translation units and
3049 performing a reloadable link on them.
3051 At present, a declaration to which @code{weakref} is attached can
3052 only be @code{static}.
3054 @item externally_visible
3055 @cindex @code{externally_visible} attribute.
3056 This attribute, attached to a global variable or function nullify
3057 effect of @option{-fwhole-program} command line option, so the object
3058 remain visible outside the current compilation unit
3062 You can specify multiple attributes in a declaration by separating them
3063 by commas within the double parentheses or by immediately following an
3064 attribute declaration with another attribute declaration.
3066 @cindex @code{#pragma}, reason for not using
3067 @cindex pragma, reason for not using
3068 Some people object to the @code{__attribute__} feature, suggesting that
3069 ISO C's @code{#pragma} should be used instead. At the time
3070 @code{__attribute__} was designed, there were two reasons for not doing
3075 It is impossible to generate @code{#pragma} commands from a macro.
3078 There is no telling what the same @code{#pragma} might mean in another
3082 These two reasons applied to almost any application that might have been
3083 proposed for @code{#pragma}. It was basically a mistake to use
3084 @code{#pragma} for @emph{anything}.
3086 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
3087 to be generated from macros. In addition, a @code{#pragma GCC}
3088 namespace is now in use for GCC-specific pragmas. However, it has been
3089 found convenient to use @code{__attribute__} to achieve a natural
3090 attachment of attributes to their corresponding declarations, whereas
3091 @code{#pragma GCC} is of use for constructs that do not naturally form
3092 part of the grammar. @xref{Other Directives,,Miscellaneous
3093 Preprocessing Directives, cpp, The GNU C Preprocessor}.
3095 @node Attribute Syntax
3096 @section Attribute Syntax
3097 @cindex attribute syntax
3099 This section describes the syntax with which @code{__attribute__} may be
3100 used, and the constructs to which attribute specifiers bind, for the C
3101 language. Some details may vary for C++ and Objective-C@. Because of
3102 infelicities in the grammar for attributes, some forms described here
3103 may not be successfully parsed in all cases.
3105 There are some problems with the semantics of attributes in C++. For
3106 example, there are no manglings for attributes, although they may affect
3107 code generation, so problems may arise when attributed types are used in
3108 conjunction with templates or overloading. Similarly, @code{typeid}
3109 does not distinguish between types with different attributes. Support
3110 for attributes in C++ may be restricted in future to attributes on
3111 declarations only, but not on nested declarators.
3113 @xref{Function Attributes}, for details of the semantics of attributes
3114 applying to functions. @xref{Variable Attributes}, for details of the
3115 semantics of attributes applying to variables. @xref{Type Attributes},
3116 for details of the semantics of attributes applying to structure, union
3117 and enumerated types.
3119 An @dfn{attribute specifier} is of the form
3120 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
3121 is a possibly empty comma-separated sequence of @dfn{attributes}, where
3122 each attribute is one of the following:
3126 Empty. Empty attributes are ignored.
3129 A word (which may be an identifier such as @code{unused}, or a reserved
3130 word such as @code{const}).
3133 A word, followed by, in parentheses, parameters for the attribute.
3134 These parameters take one of the following forms:
3138 An identifier. For example, @code{mode} attributes use this form.
3141 An identifier followed by a comma and a non-empty comma-separated list
3142 of expressions. For example, @code{format} attributes use this form.
3145 A possibly empty comma-separated list of expressions. For example,
3146 @code{format_arg} attributes use this form with the list being a single
3147 integer constant expression, and @code{alias} attributes use this form
3148 with the list being a single string constant.
3152 An @dfn{attribute specifier list} is a sequence of one or more attribute
3153 specifiers, not separated by any other tokens.
3155 In GNU C, an attribute specifier list may appear after the colon following a
3156 label, other than a @code{case} or @code{default} label. The only
3157 attribute it makes sense to use after a label is @code{unused}. This
3158 feature is intended for code generated by programs which contains labels
3159 that may be unused but which is compiled with @option{-Wall}. It would
3160 not normally be appropriate to use in it human-written code, though it
3161 could be useful in cases where the code that jumps to the label is
3162 contained within an @code{#ifdef} conditional. GNU C++ does not permit
3163 such placement of attribute lists, as it is permissible for a
3164 declaration, which could begin with an attribute list, to be labelled in
3165 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
3166 does not arise there.
3168 An attribute specifier list may appear as part of a @code{struct},
3169 @code{union} or @code{enum} specifier. It may go either immediately
3170 after the @code{struct}, @code{union} or @code{enum} keyword, or after
3171 the closing brace. The former syntax is preferred.
3172 Where attribute specifiers follow the closing brace, they are considered
3173 to relate to the structure, union or enumerated type defined, not to any
3174 enclosing declaration the type specifier appears in, and the type
3175 defined is not complete until after the attribute specifiers.
3176 @c Otherwise, there would be the following problems: a shift/reduce
3177 @c conflict between attributes binding the struct/union/enum and
3178 @c binding to the list of specifiers/qualifiers; and "aligned"
3179 @c attributes could use sizeof for the structure, but the size could be
3180 @c changed later by "packed" attributes.
3182 Otherwise, an attribute specifier appears as part of a declaration,
3183 counting declarations of unnamed parameters and type names, and relates
3184 to that declaration (which may be nested in another declaration, for
3185 example in the case of a parameter declaration), or to a particular declarator
3186 within a declaration. Where an
3187 attribute specifier is applied to a parameter declared as a function or
3188 an array, it should apply to the function or array rather than the
3189 pointer to which the parameter is implicitly converted, but this is not
3190 yet correctly implemented.
3192 Any list of specifiers and qualifiers at the start of a declaration may
3193 contain attribute specifiers, whether or not such a list may in that
3194 context contain storage class specifiers. (Some attributes, however,
3195 are essentially in the nature of storage class specifiers, and only make
3196 sense where storage class specifiers may be used; for example,
3197 @code{section}.) There is one necessary limitation to this syntax: the
3198 first old-style parameter declaration in a function definition cannot
3199 begin with an attribute specifier, because such an attribute applies to
3200 the function instead by syntax described below (which, however, is not
3201 yet implemented in this case). In some other cases, attribute
3202 specifiers are permitted by this grammar but not yet supported by the
3203 compiler. All attribute specifiers in this place relate to the
3204 declaration as a whole. In the obsolescent usage where a type of
3205 @code{int} is implied by the absence of type specifiers, such a list of
3206 specifiers and qualifiers may be an attribute specifier list with no
3207 other specifiers or qualifiers.
3209 At present, the first parameter in a function prototype must have some
3210 type specifier which is not an attribute specifier; this resolves an
3211 ambiguity in the interpretation of @code{void f(int
3212 (__attribute__((foo)) x))}, but is subject to change. At present, if
3213 the parentheses of a function declarator contain only attributes then
3214 those attributes are ignored, rather than yielding an error or warning
3215 or implying a single parameter of type int, but this is subject to
3218 An attribute specifier list may appear immediately before a declarator
3219 (other than the first) in a comma-separated list of declarators in a
3220 declaration of more than one identifier using a single list of
3221 specifiers and qualifiers. Such attribute specifiers apply
3222 only to the identifier before whose declarator they appear. For
3226 __attribute__((noreturn)) void d0 (void),
3227 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
3232 the @code{noreturn} attribute applies to all the functions
3233 declared; the @code{format} attribute only applies to @code{d1}.
3235 An attribute specifier list may appear immediately before the comma,
3236 @code{=} or semicolon terminating the declaration of an identifier other
3237 than a function definition. Such attribute specifiers apply
3238 to the declared object or function. Where an
3239 assembler name for an object or function is specified (@pxref{Asm
3240 Labels}), the attribute must follow the @code{asm}
3243 An attribute specifier list may, in future, be permitted to appear after
3244 the declarator in a function definition (before any old-style parameter
3245 declarations or the function body).
3247 Attribute specifiers may be mixed with type qualifiers appearing inside
3248 the @code{[]} of a parameter array declarator, in the C99 construct by
3249 which such qualifiers are applied to the pointer to which the array is
3250 implicitly converted. Such attribute specifiers apply to the pointer,
3251 not to the array, but at present this is not implemented and they are
3254 An attribute specifier list may appear at the start of a nested
3255 declarator. At present, there are some limitations in this usage: the
3256 attributes correctly apply to the declarator, but for most individual
3257 attributes the semantics this implies are not implemented.
3258 When attribute specifiers follow the @code{*} of a pointer
3259 declarator, they may be mixed with any type qualifiers present.
3260 The following describes the formal semantics of this syntax. It will make the
3261 most sense if you are familiar with the formal specification of
3262 declarators in the ISO C standard.
3264 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
3265 D1}, where @code{T} contains declaration specifiers that specify a type
3266 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
3267 contains an identifier @var{ident}. The type specified for @var{ident}
3268 for derived declarators whose type does not include an attribute
3269 specifier is as in the ISO C standard.
3271 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
3272 and the declaration @code{T D} specifies the type
3273 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3274 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3275 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
3277 If @code{D1} has the form @code{*
3278 @var{type-qualifier-and-attribute-specifier-list} D}, and the
3279 declaration @code{T D} specifies the type
3280 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3281 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3282 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
3288 void (__attribute__((noreturn)) ****f) (void);
3292 specifies the type ``pointer to pointer to pointer to pointer to
3293 non-returning function returning @code{void}''. As another example,
3296 char *__attribute__((aligned(8))) *f;
3300 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
3301 Note again that this does not work with most attributes; for example,
3302 the usage of @samp{aligned} and @samp{noreturn} attributes given above
3303 is not yet supported.
3305 For compatibility with existing code written for compiler versions that
3306 did not implement attributes on nested declarators, some laxity is
3307 allowed in the placing of attributes. If an attribute that only applies
3308 to types is applied to a declaration, it will be treated as applying to
3309 the type of that declaration. If an attribute that only applies to
3310 declarations is applied to the type of a declaration, it will be treated
3311 as applying to that declaration; and, for compatibility with code
3312 placing the attributes immediately before the identifier declared, such
3313 an attribute applied to a function return type will be treated as
3314 applying to the function type, and such an attribute applied to an array
3315 element type will be treated as applying to the array type. If an
3316 attribute that only applies to function types is applied to a
3317 pointer-to-function type, it will be treated as applying to the pointer
3318 target type; if such an attribute is applied to a function return type
3319 that is not a pointer-to-function type, it will be treated as applying
3320 to the function type.
3322 @node Function Prototypes
3323 @section Prototypes and Old-Style Function Definitions
3324 @cindex function prototype declarations
3325 @cindex old-style function definitions
3326 @cindex promotion of formal parameters
3328 GNU C extends ISO C to allow a function prototype to override a later
3329 old-style non-prototype definition. Consider the following example:
3332 /* @r{Use prototypes unless the compiler is old-fashioned.} */
3339 /* @r{Prototype function declaration.} */
3340 int isroot P((uid_t));
3342 /* @r{Old-style function definition.} */
3344 isroot (x) /* @r{??? lossage here ???} */
3351 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
3352 not allow this example, because subword arguments in old-style
3353 non-prototype definitions are promoted. Therefore in this example the
3354 function definition's argument is really an @code{int}, which does not
3355 match the prototype argument type of @code{short}.
3357 This restriction of ISO C makes it hard to write code that is portable
3358 to traditional C compilers, because the programmer does not know
3359 whether the @code{uid_t} type is @code{short}, @code{int}, or
3360 @code{long}. Therefore, in cases like these GNU C allows a prototype
3361 to override a later old-style definition. More precisely, in GNU C, a
3362 function prototype argument type overrides the argument type specified
3363 by a later old-style definition if the former type is the same as the
3364 latter type before promotion. Thus in GNU C the above example is
3365 equivalent to the following:
3378 GNU C++ does not support old-style function definitions, so this
3379 extension is irrelevant.
3382 @section C++ Style Comments
3384 @cindex C++ comments
3385 @cindex comments, C++ style
3387 In GNU C, you may use C++ style comments, which start with @samp{//} and
3388 continue until the end of the line. Many other C implementations allow
3389 such comments, and they are included in the 1999 C standard. However,
3390 C++ style comments are not recognized if you specify an @option{-std}
3391 option specifying a version of ISO C before C99, or @option{-ansi}
3392 (equivalent to @option{-std=c89}).
3395 @section Dollar Signs in Identifier Names
3397 @cindex dollar signs in identifier names
3398 @cindex identifier names, dollar signs in
3400 In GNU C, you may normally use dollar signs in identifier names.
3401 This is because many traditional C implementations allow such identifiers.
3402 However, dollar signs in identifiers are not supported on a few target
3403 machines, typically because the target assembler does not allow them.
3405 @node Character Escapes
3406 @section The Character @key{ESC} in Constants
3408 You can use the sequence @samp{\e} in a string or character constant to
3409 stand for the ASCII character @key{ESC}.
3412 @section Inquiring on Alignment of Types or Variables
3414 @cindex type alignment
3415 @cindex variable alignment
3417 The keyword @code{__alignof__} allows you to inquire about how an object
3418 is aligned, or the minimum alignment usually required by a type. Its
3419 syntax is just like @code{sizeof}.
3421 For example, if the target machine requires a @code{double} value to be
3422 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
3423 This is true on many RISC machines. On more traditional machine
3424 designs, @code{__alignof__ (double)} is 4 or even 2.
3426 Some machines never actually require alignment; they allow reference to any
3427 data type even at an odd address. For these machines, @code{__alignof__}
3428 reports the smallest alignment that GCC will give the data type, usually as
3429 mandated by the target ABI.
3431 If the operand of @code{__alignof__} is an lvalue rather than a type,
3432 its value is the required alignment for its type, taking into account
3433 any minimum alignment specified with GCC's @code{__attribute__}
3434 extension (@pxref{Variable Attributes}). For example, after this
3438 struct foo @{ int x; char y; @} foo1;
3442 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
3443 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
3445 It is an error to ask for the alignment of an incomplete type.
3447 @node Variable Attributes
3448 @section Specifying Attributes of Variables
3449 @cindex attribute of variables
3450 @cindex variable attributes
3452 The keyword @code{__attribute__} allows you to specify special
3453 attributes of variables or structure fields. This keyword is followed
3454 by an attribute specification inside double parentheses. Some
3455 attributes are currently defined generically for variables.
3456 Other attributes are defined for variables on particular target
3457 systems. Other attributes are available for functions
3458 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
3459 Other front ends might define more attributes
3460 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
3462 You may also specify attributes with @samp{__} preceding and following
3463 each keyword. This allows you to use them in header files without
3464 being concerned about a possible macro of the same name. For example,
3465 you may use @code{__aligned__} instead of @code{aligned}.
3467 @xref{Attribute Syntax}, for details of the exact syntax for using
3471 @cindex @code{aligned} attribute
3472 @item aligned (@var{alignment})
3473 This attribute specifies a minimum alignment for the variable or
3474 structure field, measured in bytes. For example, the declaration:
3477 int x __attribute__ ((aligned (16))) = 0;
3481 causes the compiler to allocate the global variable @code{x} on a
3482 16-byte boundary. On a 68040, this could be used in conjunction with
3483 an @code{asm} expression to access the @code{move16} instruction which
3484 requires 16-byte aligned operands.
3486 You can also specify the alignment of structure fields. For example, to
3487 create a double-word aligned @code{int} pair, you could write:
3490 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
3494 This is an alternative to creating a union with a @code{double} member
3495 that forces the union to be double-word aligned.
3497 As in the preceding examples, you can explicitly specify the alignment
3498 (in bytes) that you wish the compiler to use for a given variable or
3499 structure field. Alternatively, you can leave out the alignment factor
3500 and just ask the compiler to align a variable or field to the maximum
3501 useful alignment for the target machine you are compiling for. For
3502 example, you could write:
3505 short array[3] __attribute__ ((aligned));
3508 Whenever you leave out the alignment factor in an @code{aligned} attribute
3509 specification, the compiler automatically sets the alignment for the declared
3510 variable or field to the largest alignment which is ever used for any data
3511 type on the target machine you are compiling for. Doing this can often make
3512 copy operations more efficient, because the compiler can use whatever
3513 instructions copy the biggest chunks of memory when performing copies to
3514 or from the variables or fields that you have aligned this way.
3516 When used on a struct, or struct member, the @code{aligned} attribute can
3517 only increase the alignment; in order to decrease it, the @code{packed}
3518 attribute must be specified as well. When used as part of a typedef, the
3519 @code{aligned} attribute can both increase and decrease alignment, and
3520 specifying the @code{packed} attribute will generate a warning.
3522 Note that the effectiveness of @code{aligned} attributes may be limited
3523 by inherent limitations in your linker. On many systems, the linker is
3524 only able to arrange for variables to be aligned up to a certain maximum
3525 alignment. (For some linkers, the maximum supported alignment may
3526 be very very small.) If your linker is only able to align variables
3527 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3528 in an @code{__attribute__} will still only provide you with 8 byte
3529 alignment. See your linker documentation for further information.
3531 The @code{aligned} attribute can also be used for functions
3532 (@pxref{Function Attributes}.)
3534 @item cleanup (@var{cleanup_function})
3535 @cindex @code{cleanup} attribute
3536 The @code{cleanup} attribute runs a function when the variable goes
3537 out of scope. This attribute can only be applied to auto function
3538 scope variables; it may not be applied to parameters or variables
3539 with static storage duration. The function must take one parameter,
3540 a pointer to a type compatible with the variable. The return value
3541 of the function (if any) is ignored.
3543 If @option{-fexceptions} is enabled, then @var{cleanup_function}
3544 will be run during the stack unwinding that happens during the
3545 processing of the exception. Note that the @code{cleanup} attribute
3546 does not allow the exception to be caught, only to perform an action.
3547 It is undefined what happens if @var{cleanup_function} does not
3552 @cindex @code{common} attribute
3553 @cindex @code{nocommon} attribute
3556 The @code{common} attribute requests GCC to place a variable in
3557 ``common'' storage. The @code{nocommon} attribute requests the
3558 opposite---to allocate space for it directly.
3560 These attributes override the default chosen by the
3561 @option{-fno-common} and @option{-fcommon} flags respectively.
3564 @cindex @code{deprecated} attribute
3565 The @code{deprecated} attribute results in a warning if the variable
3566 is used anywhere in the source file. This is useful when identifying
3567 variables that are expected to be removed in a future version of a
3568 program. The warning also includes the location of the declaration
3569 of the deprecated variable, to enable users to easily find further
3570 information about why the variable is deprecated, or what they should
3571 do instead. Note that the warning only occurs for uses:
3574 extern int old_var __attribute__ ((deprecated));
3576 int new_fn () @{ return old_var; @}
3579 results in a warning on line 3 but not line 2.
3581 The @code{deprecated} attribute can also be used for functions and
3582 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3584 @item mode (@var{mode})
3585 @cindex @code{mode} attribute
3586 This attribute specifies the data type for the declaration---whichever
3587 type corresponds to the mode @var{mode}. This in effect lets you
3588 request an integer or floating point type according to its width.
3590 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3591 indicate the mode corresponding to a one-byte integer, @samp{word} or
3592 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3593 or @samp{__pointer__} for the mode used to represent pointers.
3596 @cindex @code{packed} attribute
3597 The @code{packed} attribute specifies that a variable or structure field
3598 should have the smallest possible alignment---one byte for a variable,
3599 and one bit for a field, unless you specify a larger value with the
3600 @code{aligned} attribute.
3602 Here is a structure in which the field @code{x} is packed, so that it
3603 immediately follows @code{a}:
3609 int x[2] __attribute__ ((packed));
3613 @item section ("@var{section-name}")
3614 @cindex @code{section} variable attribute
3615 Normally, the compiler places the objects it generates in sections like
3616 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3617 or you need certain particular variables to appear in special sections,
3618 for example to map to special hardware. The @code{section}
3619 attribute specifies that a variable (or function) lives in a particular
3620 section. For example, this small program uses several specific section names:
3623 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3624 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3625 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3626 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3630 /* @r{Initialize stack pointer} */
3631 init_sp (stack + sizeof (stack));
3633 /* @r{Initialize initialized data} */
3634 memcpy (&init_data, &data, &edata - &data);
3636 /* @r{Turn on the serial ports} */
3643 Use the @code{section} attribute with an @emph{initialized} definition
3644 of a @emph{global} variable, as shown in the example. GCC issues
3645 a warning and otherwise ignores the @code{section} attribute in
3646 uninitialized variable declarations.
3648 You may only use the @code{section} attribute with a fully initialized
3649 global definition because of the way linkers work. The linker requires
3650 each object be defined once, with the exception that uninitialized
3651 variables tentatively go in the @code{common} (or @code{bss}) section
3652 and can be multiply ``defined''. You can force a variable to be
3653 initialized with the @option{-fno-common} flag or the @code{nocommon}
3656 Some file formats do not support arbitrary sections so the @code{section}
3657 attribute is not available on all platforms.
3658 If you need to map the entire contents of a module to a particular
3659 section, consider using the facilities of the linker instead.
3662 @cindex @code{shared} variable attribute
3663 On Microsoft Windows, in addition to putting variable definitions in a named
3664 section, the section can also be shared among all running copies of an
3665 executable or DLL@. For example, this small program defines shared data
3666 by putting it in a named section @code{shared} and marking the section
3670 int foo __attribute__((section ("shared"), shared)) = 0;
3675 /* @r{Read and write foo. All running
3676 copies see the same value.} */
3682 You may only use the @code{shared} attribute along with @code{section}
3683 attribute with a fully initialized global definition because of the way
3684 linkers work. See @code{section} attribute for more information.
3686 The @code{shared} attribute is only available on Microsoft Windows@.
3688 @item tls_model ("@var{tls_model}")
3689 @cindex @code{tls_model} attribute
3690 The @code{tls_model} attribute sets thread-local storage model
3691 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3692 overriding @option{-ftls-model=} command line switch on a per-variable
3694 The @var{tls_model} argument should be one of @code{global-dynamic},
3695 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3697 Not all targets support this attribute.
3700 This attribute, attached to a variable, means that the variable is meant
3701 to be possibly unused. GCC will not produce a warning for this
3705 This attribute, attached to a variable, means that the variable must be
3706 emitted even if it appears that the variable is not referenced.
3708 @item vector_size (@var{bytes})
3709 This attribute specifies the vector size for the variable, measured in
3710 bytes. For example, the declaration:
3713 int foo __attribute__ ((vector_size (16)));
3717 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3718 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3719 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3721 This attribute is only applicable to integral and float scalars,
3722 although arrays, pointers, and function return values are allowed in
3723 conjunction with this construct.
3725 Aggregates with this attribute are invalid, even if they are of the same
3726 size as a corresponding scalar. For example, the declaration:
3729 struct S @{ int a; @};
3730 struct S __attribute__ ((vector_size (16))) foo;
3734 is invalid even if the size of the structure is the same as the size of
3738 The @code{selectany} attribute causes an initialized global variable to
3739 have link-once semantics. When multiple definitions of the variable are
3740 encountered by the linker, the first is selected and the remainder are
3741 discarded. Following usage by the Microsoft compiler, the linker is told
3742 @emph{not} to warn about size or content differences of the multiple
3745 Although the primary usage of this attribute is for POD types, the
3746 attribute can also be applied to global C++ objects that are initialized
3747 by a constructor. In this case, the static initialization and destruction
3748 code for the object is emitted in each translation defining the object,
3749 but the calls to the constructor and destructor are protected by a
3750 link-once guard variable.
3752 The @code{selectany} attribute is only available on Microsoft Windows
3753 targets. You can use @code{__declspec (selectany)} as a synonym for
3754 @code{__attribute__ ((selectany))} for compatibility with other
3758 The @code{weak} attribute is described in @xref{Function Attributes}.
3761 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3764 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3768 @subsection Blackfin Variable Attributes
3770 Three attributes are currently defined for the Blackfin.
3776 @cindex @code{l1_data} variable attribute
3777 @cindex @code{l1_data_A} variable attribute
3778 @cindex @code{l1_data_B} variable attribute
3779 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
3780 Variables with @code{l1_data} attribute will be put into the specific section
3781 named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into
3782 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
3783 attribute will be put into the specific section named @code{.l1.data.B}.
3786 @subsection M32R/D Variable Attributes
3788 One attribute is currently defined for the M32R/D@.
3791 @item model (@var{model-name})
3792 @cindex variable addressability on the M32R/D
3793 Use this attribute on the M32R/D to set the addressability of an object.
3794 The identifier @var{model-name} is one of @code{small}, @code{medium},
3795 or @code{large}, representing each of the code models.
3797 Small model objects live in the lower 16MB of memory (so that their
3798 addresses can be loaded with the @code{ld24} instruction).
3800 Medium and large model objects may live anywhere in the 32-bit address space
3801 (the compiler will generate @code{seth/add3} instructions to load their
3805 @anchor{i386 Variable Attributes}
3806 @subsection i386 Variable Attributes
3808 Two attributes are currently defined for i386 configurations:
3809 @code{ms_struct} and @code{gcc_struct}
3814 @cindex @code{ms_struct} attribute
3815 @cindex @code{gcc_struct} attribute
3817 If @code{packed} is used on a structure, or if bit-fields are used
3818 it may be that the Microsoft ABI packs them differently
3819 than GCC would normally pack them. Particularly when moving packed
3820 data between functions compiled with GCC and the native Microsoft compiler
3821 (either via function call or as data in a file), it may be necessary to access
3824 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3825 compilers to match the native Microsoft compiler.
3827 The Microsoft structure layout algorithm is fairly simple with the exception
3828 of the bitfield packing:
3830 The padding and alignment of members of structures and whether a bit field
3831 can straddle a storage-unit boundary
3834 @item Structure members are stored sequentially in the order in which they are
3835 declared: the first member has the lowest memory address and the last member
3838 @item Every data object has an alignment-requirement. The alignment-requirement
3839 for all data except structures, unions, and arrays is either the size of the
3840 object or the current packing size (specified with either the aligned attribute
3841 or the pack pragma), whichever is less. For structures, unions, and arrays,
3842 the alignment-requirement is the largest alignment-requirement of its members.
3843 Every object is allocated an offset so that:
3845 offset % alignment-requirement == 0
3847 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
3848 unit if the integral types are the same size and if the next bit field fits
3849 into the current allocation unit without crossing the boundary imposed by the
3850 common alignment requirements of the bit fields.
3853 Handling of zero-length bitfields:
3855 MSVC interprets zero-length bitfields in the following ways:
3858 @item If a zero-length bitfield is inserted between two bitfields that would
3859 normally be coalesced, the bitfields will not be coalesced.
3866 unsigned long bf_1 : 12;
3868 unsigned long bf_2 : 12;
3872 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
3873 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
3875 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
3876 alignment of the zero-length bitfield is greater than the member that follows it,
3877 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
3897 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
3898 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
3899 bitfield will not affect the alignment of @code{bar} or, as a result, the size
3902 Taking this into account, it is important to note the following:
3905 @item If a zero-length bitfield follows a normal bitfield, the type of the
3906 zero-length bitfield may affect the alignment of the structure as whole. For
3907 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
3908 normal bitfield, and is of type short.
3910 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
3911 still affect the alignment of the structure:
3921 Here, @code{t4} will take up 4 bytes.
3924 @item Zero-length bitfields following non-bitfield members are ignored:
3935 Here, @code{t5} will take up 2 bytes.
3939 @subsection PowerPC Variable Attributes
3941 Three attributes currently are defined for PowerPC configurations:
3942 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
3944 For full documentation of the struct attributes please see the
3945 documentation in the @xref{i386 Variable Attributes}, section.
3947 For documentation of @code{altivec} attribute please see the
3948 documentation in the @xref{PowerPC Type Attributes}, section.
3950 @subsection SPU Variable Attributes
3952 The SPU supports the @code{spu_vector} attribute for variables. For
3953 documentation of this attribute please see the documentation in the
3954 @xref{SPU Type Attributes}, section.
3956 @subsection Xstormy16 Variable Attributes
3958 One attribute is currently defined for xstormy16 configurations:
3963 @cindex @code{below100} attribute
3965 If a variable has the @code{below100} attribute (@code{BELOW100} is
3966 allowed also), GCC will place the variable in the first 0x100 bytes of
3967 memory and use special opcodes to access it. Such variables will be
3968 placed in either the @code{.bss_below100} section or the
3969 @code{.data_below100} section.
3973 @subsection AVR Variable Attributes
3977 @cindex @code{progmem} variable attribute
3978 The @code{progmem} attribute is used on the AVR to place data in the Program
3979 Memory address space. The AVR is a Harvard Architecture processor and data
3980 normally resides in the Data Memory address space.
3983 @node Type Attributes
3984 @section Specifying Attributes of Types
3985 @cindex attribute of types
3986 @cindex type attributes
3988 The keyword @code{__attribute__} allows you to specify special
3989 attributes of @code{struct} and @code{union} types when you define
3990 such types. This keyword is followed by an attribute specification
3991 inside double parentheses. Seven attributes are currently defined for
3992 types: @code{aligned}, @code{packed}, @code{transparent_union},
3993 @code{unused}, @code{deprecated}, @code{visibility}, and
3994 @code{may_alias}. Other attributes are defined for functions
3995 (@pxref{Function Attributes}) and for variables (@pxref{Variable
3998 You may also specify any one of these attributes with @samp{__}
3999 preceding and following its keyword. This allows you to use these
4000 attributes in header files without being concerned about a possible
4001 macro of the same name. For example, you may use @code{__aligned__}
4002 instead of @code{aligned}.
4004 You may specify type attributes in an enum, struct or union type
4005 declaration or definition, or for other types in a @code{typedef}
4008 For an enum, struct or union type, you may specify attributes either
4009 between the enum, struct or union tag and the name of the type, or
4010 just past the closing curly brace of the @emph{definition}. The
4011 former syntax is preferred.
4013 @xref{Attribute Syntax}, for details of the exact syntax for using
4017 @cindex @code{aligned} attribute
4018 @item aligned (@var{alignment})
4019 This attribute specifies a minimum alignment (in bytes) for variables
4020 of the specified type. For example, the declarations:
4023 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
4024 typedef int more_aligned_int __attribute__ ((aligned (8)));
4028 force the compiler to insure (as far as it can) that each variable whose
4029 type is @code{struct S} or @code{more_aligned_int} will be allocated and
4030 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
4031 variables of type @code{struct S} aligned to 8-byte boundaries allows
4032 the compiler to use the @code{ldd} and @code{std} (doubleword load and
4033 store) instructions when copying one variable of type @code{struct S} to
4034 another, thus improving run-time efficiency.
4036 Note that the alignment of any given @code{struct} or @code{union} type
4037 is required by the ISO C standard to be at least a perfect multiple of
4038 the lowest common multiple of the alignments of all of the members of
4039 the @code{struct} or @code{union} in question. This means that you @emph{can}
4040 effectively adjust the alignment of a @code{struct} or @code{union}
4041 type by attaching an @code{aligned} attribute to any one of the members
4042 of such a type, but the notation illustrated in the example above is a
4043 more obvious, intuitive, and readable way to request the compiler to
4044 adjust the alignment of an entire @code{struct} or @code{union} type.
4046 As in the preceding example, you can explicitly specify the alignment
4047 (in bytes) that you wish the compiler to use for a given @code{struct}
4048 or @code{union} type. Alternatively, you can leave out the alignment factor
4049 and just ask the compiler to align a type to the maximum
4050 useful alignment for the target machine you are compiling for. For
4051 example, you could write:
4054 struct S @{ short f[3]; @} __attribute__ ((aligned));
4057 Whenever you leave out the alignment factor in an @code{aligned}
4058 attribute specification, the compiler automatically sets the alignment
4059 for the type to the largest alignment which is ever used for any data
4060 type on the target machine you are compiling for. Doing this can often
4061 make copy operations more efficient, because the compiler can use
4062 whatever instructions copy the biggest chunks of memory when performing
4063 copies to or from the variables which have types that you have aligned
4066 In the example above, if the size of each @code{short} is 2 bytes, then
4067 the size of the entire @code{struct S} type is 6 bytes. The smallest
4068 power of two which is greater than or equal to that is 8, so the
4069 compiler sets the alignment for the entire @code{struct S} type to 8
4072 Note that although you can ask the compiler to select a time-efficient
4073 alignment for a given type and then declare only individual stand-alone
4074 objects of that type, the compiler's ability to select a time-efficient
4075 alignment is primarily useful only when you plan to create arrays of
4076 variables having the relevant (efficiently aligned) type. If you
4077 declare or use arrays of variables of an efficiently-aligned type, then
4078 it is likely that your program will also be doing pointer arithmetic (or
4079 subscripting, which amounts to the same thing) on pointers to the
4080 relevant type, and the code that the compiler generates for these
4081 pointer arithmetic operations will often be more efficient for
4082 efficiently-aligned types than for other types.
4084 The @code{aligned} attribute can only increase the alignment; but you
4085 can decrease it by specifying @code{packed} as well. See below.
4087 Note that the effectiveness of @code{aligned} attributes may be limited
4088 by inherent limitations in your linker. On many systems, the linker is
4089 only able to arrange for variables to be aligned up to a certain maximum
4090 alignment. (For some linkers, the maximum supported alignment may
4091 be very very small.) If your linker is only able to align variables
4092 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4093 in an @code{__attribute__} will still only provide you with 8 byte
4094 alignment. See your linker documentation for further information.
4097 This attribute, attached to @code{struct} or @code{union} type
4098 definition, specifies that each member (other than zero-width bitfields)
4099 of the structure or union is placed to minimize the memory required. When
4100 attached to an @code{enum} definition, it indicates that the smallest
4101 integral type should be used.
4103 @opindex fshort-enums
4104 Specifying this attribute for @code{struct} and @code{union} types is
4105 equivalent to specifying the @code{packed} attribute on each of the
4106 structure or union members. Specifying the @option{-fshort-enums}
4107 flag on the line is equivalent to specifying the @code{packed}
4108 attribute on all @code{enum} definitions.
4110 In the following example @code{struct my_packed_struct}'s members are
4111 packed closely together, but the internal layout of its @code{s} member
4112 is not packed---to do that, @code{struct my_unpacked_struct} would need to
4116 struct my_unpacked_struct
4122 struct __attribute__ ((__packed__)) my_packed_struct
4126 struct my_unpacked_struct s;
4130 You may only specify this attribute on the definition of a @code{enum},
4131 @code{struct} or @code{union}, not on a @code{typedef} which does not
4132 also define the enumerated type, structure or union.
4134 @item transparent_union
4135 This attribute, attached to a @code{union} type definition, indicates
4136 that any function parameter having that union type causes calls to that
4137 function to be treated in a special way.
4139 First, the argument corresponding to a transparent union type can be of
4140 any type in the union; no cast is required. Also, if the union contains
4141 a pointer type, the corresponding argument can be a null pointer
4142 constant or a void pointer expression; and if the union contains a void
4143 pointer type, the corresponding argument can be any pointer expression.
4144 If the union member type is a pointer, qualifiers like @code{const} on
4145 the referenced type must be respected, just as with normal pointer
4148 Second, the argument is passed to the function using the calling
4149 conventions of the first member of the transparent union, not the calling
4150 conventions of the union itself. All members of the union must have the
4151 same machine representation; this is necessary for this argument passing
4154 Transparent unions are designed for library functions that have multiple
4155 interfaces for compatibility reasons. For example, suppose the
4156 @code{wait} function must accept either a value of type @code{int *} to
4157 comply with Posix, or a value of type @code{union wait *} to comply with
4158 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
4159 @code{wait} would accept both kinds of arguments, but it would also
4160 accept any other pointer type and this would make argument type checking
4161 less useful. Instead, @code{<sys/wait.h>} might define the interface
4165 typedef union __attribute__ ((__transparent_union__))
4169 @} wait_status_ptr_t;
4171 pid_t wait (wait_status_ptr_t);
4174 This interface allows either @code{int *} or @code{union wait *}
4175 arguments to be passed, using the @code{int *} calling convention.
4176 The program can call @code{wait} with arguments of either type:
4179 int w1 () @{ int w; return wait (&w); @}
4180 int w2 () @{ union wait w; return wait (&w); @}
4183 With this interface, @code{wait}'s implementation might look like this:
4186 pid_t wait (wait_status_ptr_t p)
4188 return waitpid (-1, p.__ip, 0);
4193 When attached to a type (including a @code{union} or a @code{struct}),
4194 this attribute means that variables of that type are meant to appear
4195 possibly unused. GCC will not produce a warning for any variables of
4196 that type, even if the variable appears to do nothing. This is often
4197 the case with lock or thread classes, which are usually defined and then
4198 not referenced, but contain constructors and destructors that have
4199 nontrivial bookkeeping functions.
4202 The @code{deprecated} attribute results in a warning if the type
4203 is used anywhere in the source file. This is useful when identifying
4204 types that are expected to be removed in a future version of a program.
4205 If possible, the warning also includes the location of the declaration
4206 of the deprecated type, to enable users to easily find further
4207 information about why the type is deprecated, or what they should do
4208 instead. Note that the warnings only occur for uses and then only
4209 if the type is being applied to an identifier that itself is not being
4210 declared as deprecated.
4213 typedef int T1 __attribute__ ((deprecated));
4217 typedef T1 T3 __attribute__ ((deprecated));
4218 T3 z __attribute__ ((deprecated));
4221 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
4222 warning is issued for line 4 because T2 is not explicitly
4223 deprecated. Line 5 has no warning because T3 is explicitly
4224 deprecated. Similarly for line 6.
4226 The @code{deprecated} attribute can also be used for functions and
4227 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
4230 Accesses to objects with types with this attribute are not subjected to
4231 type-based alias analysis, but are instead assumed to be able to alias
4232 any other type of objects, just like the @code{char} type. See
4233 @option{-fstrict-aliasing} for more information on aliasing issues.
4238 typedef short __attribute__((__may_alias__)) short_a;
4244 short_a *b = (short_a *) &a;
4248 if (a == 0x12345678)
4255 If you replaced @code{short_a} with @code{short} in the variable
4256 declaration, the above program would abort when compiled with
4257 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
4258 above in recent GCC versions.
4261 In C++, attribute visibility (@pxref{Function Attributes}) can also be
4262 applied to class, struct, union and enum types. Unlike other type
4263 attributes, the attribute must appear between the initial keyword and
4264 the name of the type; it cannot appear after the body of the type.
4266 Note that the type visibility is applied to vague linkage entities
4267 associated with the class (vtable, typeinfo node, etc.). In
4268 particular, if a class is thrown as an exception in one shared object
4269 and caught in another, the class must have default visibility.
4270 Otherwise the two shared objects will be unable to use the same
4271 typeinfo node and exception handling will break.
4273 @subsection ARM Type Attributes
4275 On those ARM targets that support @code{dllimport} (such as Symbian
4276 OS), you can use the @code{notshared} attribute to indicate that the
4277 virtual table and other similar data for a class should not be
4278 exported from a DLL@. For example:
4281 class __declspec(notshared) C @{
4283 __declspec(dllimport) C();
4287 __declspec(dllexport)
4291 In this code, @code{C::C} is exported from the current DLL, but the
4292 virtual table for @code{C} is not exported. (You can use
4293 @code{__attribute__} instead of @code{__declspec} if you prefer, but
4294 most Symbian OS code uses @code{__declspec}.)
4296 @anchor{i386 Type Attributes}
4297 @subsection i386 Type Attributes
4299 Two attributes are currently defined for i386 configurations:
4300 @code{ms_struct} and @code{gcc_struct}
4304 @cindex @code{ms_struct}
4305 @cindex @code{gcc_struct}
4307 If @code{packed} is used on a structure, or if bit-fields are used
4308 it may be that the Microsoft ABI packs them differently
4309 than GCC would normally pack them. Particularly when moving packed
4310 data between functions compiled with GCC and the native Microsoft compiler
4311 (either via function call or as data in a file), it may be necessary to access
4314 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4315 compilers to match the native Microsoft compiler.
4318 To specify multiple attributes, separate them by commas within the
4319 double parentheses: for example, @samp{__attribute__ ((aligned (16),
4322 @anchor{PowerPC Type Attributes}
4323 @subsection PowerPC Type Attributes
4325 Three attributes currently are defined for PowerPC configurations:
4326 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4328 For full documentation of the struct attributes please see the
4329 documentation in the @xref{i386 Type Attributes}, section.
4331 The @code{altivec} attribute allows one to declare AltiVec vector data
4332 types supported by the AltiVec Programming Interface Manual. The
4333 attribute requires an argument to specify one of three vector types:
4334 @code{vector__}, @code{pixel__} (always followed by unsigned short),
4335 and @code{bool__} (always followed by unsigned).
4338 __attribute__((altivec(vector__)))
4339 __attribute__((altivec(pixel__))) unsigned short
4340 __attribute__((altivec(bool__))) unsigned
4343 These attributes mainly are intended to support the @code{__vector},
4344 @code{__pixel}, and @code{__bool} AltiVec keywords.
4346 @anchor{SPU Type Attributes}
4347 @subsection SPU Type Attributes
4349 The SPU supports the @code{spu_vector} attribute for types. This attribute
4350 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
4351 Language Extensions Specification. It is intended to support the
4352 @code{__vector} keyword.
4356 @section An Inline Function is As Fast As a Macro
4357 @cindex inline functions
4358 @cindex integrating function code
4360 @cindex macros, inline alternative
4362 By declaring a function inline, you can direct GCC to make
4363 calls to that function faster. One way GCC can achieve this is to
4364 integrate that function's code into the code for its callers. This
4365 makes execution faster by eliminating the function-call overhead; in
4366 addition, if any of the actual argument values are constant, their
4367 known values may permit simplifications at compile time so that not
4368 all of the inline function's code needs to be included. The effect on
4369 code size is less predictable; object code may be larger or smaller
4370 with function inlining, depending on the particular case. You can
4371 also direct GCC to try to integrate all ``simple enough'' functions
4372 into their callers with the option @option{-finline-functions}.
4374 GCC implements three different semantics of declaring a function
4375 inline. One is available with @option{-std=gnu89} or
4376 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
4377 on all inline declarations, another when @option{-std=c99} or
4378 @option{-std=gnu99} (without @option{-fgnu89-inline}), and the third
4379 is used when compiling C++.
4381 To declare a function inline, use the @code{inline} keyword in its
4382 declaration, like this:
4392 If you are writing a header file to be included in ISO C89 programs, write
4393 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
4395 The three types of inlining behave similarly in two important cases:
4396 when the @code{inline} keyword is used on a @code{static} function,
4397 like the example above, and when a function is first declared without
4398 using the @code{inline} keyword and then is defined with
4399 @code{inline}, like this:
4402 extern int inc (int *a);
4410 In both of these common cases, the program behaves the same as if you
4411 had not used the @code{inline} keyword, except for its speed.
4413 @cindex inline functions, omission of
4414 @opindex fkeep-inline-functions
4415 When a function is both inline and @code{static}, if all calls to the
4416 function are integrated into the caller, and the function's address is
4417 never used, then the function's own assembler code is never referenced.
4418 In this case, GCC does not actually output assembler code for the
4419 function, unless you specify the option @option{-fkeep-inline-functions}.
4420 Some calls cannot be integrated for various reasons (in particular,
4421 calls that precede the function's definition cannot be integrated, and
4422 neither can recursive calls within the definition). If there is a
4423 nonintegrated call, then the function is compiled to assembler code as
4424 usual. The function must also be compiled as usual if the program
4425 refers to its address, because that can't be inlined.
4428 Note that certain usages in a function definition can make it unsuitable
4429 for inline substitution. Among these usages are: use of varargs, use of
4430 alloca, use of variable sized data types (@pxref{Variable Length}),
4431 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
4432 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
4433 will warn when a function marked @code{inline} could not be substituted,
4434 and will give the reason for the failure.
4436 @cindex automatic @code{inline} for C++ member fns
4437 @cindex @code{inline} automatic for C++ member fns
4438 @cindex member fns, automatically @code{inline}
4439 @cindex C++ member fns, automatically @code{inline}
4440 @opindex fno-default-inline
4441 As required by ISO C++, GCC considers member functions defined within
4442 the body of a class to be marked inline even if they are
4443 not explicitly declared with the @code{inline} keyword. You can
4444 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
4445 Options,,Options Controlling C++ Dialect}.
4447 GCC does not inline any functions when not optimizing unless you specify
4448 the @samp{always_inline} attribute for the function, like this:
4451 /* @r{Prototype.} */
4452 inline void foo (const char) __attribute__((always_inline));
4455 The remainder of this section is specific to GNU C89 inlining.
4457 @cindex non-static inline function
4458 When an inline function is not @code{static}, then the compiler must assume
4459 that there may be calls from other source files; since a global symbol can
4460 be defined only once in any program, the function must not be defined in
4461 the other source files, so the calls therein cannot be integrated.
4462 Therefore, a non-@code{static} inline function is always compiled on its
4463 own in the usual fashion.
4465 If you specify both @code{inline} and @code{extern} in the function
4466 definition, then the definition is used only for inlining. In no case
4467 is the function compiled on its own, not even if you refer to its
4468 address explicitly. Such an address becomes an external reference, as
4469 if you had only declared the function, and had not defined it.
4471 This combination of @code{inline} and @code{extern} has almost the
4472 effect of a macro. The way to use it is to put a function definition in
4473 a header file with these keywords, and put another copy of the
4474 definition (lacking @code{inline} and @code{extern}) in a library file.
4475 The definition in the header file will cause most calls to the function
4476 to be inlined. If any uses of the function remain, they will refer to
4477 the single copy in the library.
4480 @section Assembler Instructions with C Expression Operands
4481 @cindex extended @code{asm}
4482 @cindex @code{asm} expressions
4483 @cindex assembler instructions
4486 In an assembler instruction using @code{asm}, you can specify the
4487 operands of the instruction using C expressions. This means you need not
4488 guess which registers or memory locations will contain the data you want
4491 You must specify an assembler instruction template much like what
4492 appears in a machine description, plus an operand constraint string for
4495 For example, here is how to use the 68881's @code{fsinx} instruction:
4498 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
4502 Here @code{angle} is the C expression for the input operand while
4503 @code{result} is that of the output operand. Each has @samp{"f"} as its
4504 operand constraint, saying that a floating point register is required.
4505 The @samp{=} in @samp{=f} indicates that the operand is an output; all
4506 output operands' constraints must use @samp{=}. The constraints use the
4507 same language used in the machine description (@pxref{Constraints}).
4509 Each operand is described by an operand-constraint string followed by
4510 the C expression in parentheses. A colon separates the assembler
4511 template from the first output operand and another separates the last
4512 output operand from the first input, if any. Commas separate the
4513 operands within each group. The total number of operands is currently
4514 limited to 30; this limitation may be lifted in some future version of
4517 If there are no output operands but there are input operands, you must
4518 place two consecutive colons surrounding the place where the output
4521 As of GCC version 3.1, it is also possible to specify input and output
4522 operands using symbolic names which can be referenced within the
4523 assembler code. These names are specified inside square brackets
4524 preceding the constraint string, and can be referenced inside the
4525 assembler code using @code{%[@var{name}]} instead of a percentage sign
4526 followed by the operand number. Using named operands the above example
4530 asm ("fsinx %[angle],%[output]"
4531 : [output] "=f" (result)
4532 : [angle] "f" (angle));
4536 Note that the symbolic operand names have no relation whatsoever to
4537 other C identifiers. You may use any name you like, even those of
4538 existing C symbols, but you must ensure that no two operands within the same
4539 assembler construct use the same symbolic name.
4541 Output operand expressions must be lvalues; the compiler can check this.
4542 The input operands need not be lvalues. The compiler cannot check
4543 whether the operands have data types that are reasonable for the
4544 instruction being executed. It does not parse the assembler instruction
4545 template and does not know what it means or even whether it is valid
4546 assembler input. The extended @code{asm} feature is most often used for
4547 machine instructions the compiler itself does not know exist. If
4548 the output expression cannot be directly addressed (for example, it is a
4549 bit-field), your constraint must allow a register. In that case, GCC
4550 will use the register as the output of the @code{asm}, and then store
4551 that register into the output.
4553 The ordinary output operands must be write-only; GCC will assume that
4554 the values in these operands before the instruction are dead and need
4555 not be generated. Extended asm supports input-output or read-write
4556 operands. Use the constraint character @samp{+} to indicate such an
4557 operand and list it with the output operands. You should only use
4558 read-write operands when the constraints for the operand (or the
4559 operand in which only some of the bits are to be changed) allow a
4562 You may, as an alternative, logically split its function into two
4563 separate operands, one input operand and one write-only output
4564 operand. The connection between them is expressed by constraints
4565 which say they need to be in the same location when the instruction
4566 executes. You can use the same C expression for both operands, or
4567 different expressions. For example, here we write the (fictitious)
4568 @samp{combine} instruction with @code{bar} as its read-only source
4569 operand and @code{foo} as its read-write destination:
4572 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
4576 The constraint @samp{"0"} for operand 1 says that it must occupy the
4577 same location as operand 0. A number in constraint is allowed only in
4578 an input operand and it must refer to an output operand.
4580 Only a number in the constraint can guarantee that one operand will be in
4581 the same place as another. The mere fact that @code{foo} is the value
4582 of both operands is not enough to guarantee that they will be in the
4583 same place in the generated assembler code. The following would not
4587 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
4590 Various optimizations or reloading could cause operands 0 and 1 to be in
4591 different registers; GCC knows no reason not to do so. For example, the
4592 compiler might find a copy of the value of @code{foo} in one register and
4593 use it for operand 1, but generate the output operand 0 in a different
4594 register (copying it afterward to @code{foo}'s own address). Of course,
4595 since the register for operand 1 is not even mentioned in the assembler
4596 code, the result will not work, but GCC can't tell that.
4598 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
4599 the operand number for a matching constraint. For example:
4602 asm ("cmoveq %1,%2,%[result]"
4603 : [result] "=r"(result)
4604 : "r" (test), "r"(new), "[result]"(old));
4607 Sometimes you need to make an @code{asm} operand be a specific register,
4608 but there's no matching constraint letter for that register @emph{by
4609 itself}. To force the operand into that register, use a local variable
4610 for the operand and specify the register in the variable declaration.
4611 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
4612 register constraint letter that matches the register:
4615 register int *p1 asm ("r0") = @dots{};
4616 register int *p2 asm ("r1") = @dots{};
4617 register int *result asm ("r0");
4618 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4621 @anchor{Example of asm with clobbered asm reg}
4622 In the above example, beware that a register that is call-clobbered by
4623 the target ABI will be overwritten by any function call in the
4624 assignment, including library calls for arithmetic operators.
4625 Assuming it is a call-clobbered register, this may happen to @code{r0}
4626 above by the assignment to @code{p2}. If you have to use such a
4627 register, use temporary variables for expressions between the register
4632 register int *p1 asm ("r0") = @dots{};
4633 register int *p2 asm ("r1") = t1;
4634 register int *result asm ("r0");
4635 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4638 Some instructions clobber specific hard registers. To describe this,
4639 write a third colon after the input operands, followed by the names of
4640 the clobbered hard registers (given as strings). Here is a realistic
4641 example for the VAX:
4644 asm volatile ("movc3 %0,%1,%2"
4645 : /* @r{no outputs} */
4646 : "g" (from), "g" (to), "g" (count)
4647 : "r0", "r1", "r2", "r3", "r4", "r5");
4650 You may not write a clobber description in a way that overlaps with an
4651 input or output operand. For example, you may not have an operand
4652 describing a register class with one member if you mention that register
4653 in the clobber list. Variables declared to live in specific registers
4654 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
4655 have no part mentioned in the clobber description.
4656 There is no way for you to specify that an input
4657 operand is modified without also specifying it as an output
4658 operand. Note that if all the output operands you specify are for this
4659 purpose (and hence unused), you will then also need to specify
4660 @code{volatile} for the @code{asm} construct, as described below, to
4661 prevent GCC from deleting the @code{asm} statement as unused.
4663 If you refer to a particular hardware register from the assembler code,
4664 you will probably have to list the register after the third colon to
4665 tell the compiler the register's value is modified. In some assemblers,
4666 the register names begin with @samp{%}; to produce one @samp{%} in the
4667 assembler code, you must write @samp{%%} in the input.
4669 If your assembler instruction can alter the condition code register, add
4670 @samp{cc} to the list of clobbered registers. GCC on some machines
4671 represents the condition codes as a specific hardware register;
4672 @samp{cc} serves to name this register. On other machines, the
4673 condition code is handled differently, and specifying @samp{cc} has no
4674 effect. But it is valid no matter what the machine.
4676 If your assembler instructions access memory in an unpredictable
4677 fashion, add @samp{memory} to the list of clobbered registers. This
4678 will cause GCC to not keep memory values cached in registers across the
4679 assembler instruction and not optimize stores or loads to that memory.
4680 You will also want to add the @code{volatile} keyword if the memory
4681 affected is not listed in the inputs or outputs of the @code{asm}, as
4682 the @samp{memory} clobber does not count as a side-effect of the
4683 @code{asm}. If you know how large the accessed memory is, you can add
4684 it as input or output but if this is not known, you should add
4685 @samp{memory}. As an example, if you access ten bytes of a string, you
4686 can use a memory input like:
4689 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
4692 Note that in the following example the memory input is necessary,
4693 otherwise GCC might optimize the store to @code{x} away:
4700 asm ("magic stuff accessing an 'int' pointed to by '%1'"
4701 "=&d" (r) : "a" (y), "m" (*y));
4706 You can put multiple assembler instructions together in a single
4707 @code{asm} template, separated by the characters normally used in assembly
4708 code for the system. A combination that works in most places is a newline
4709 to break the line, plus a tab character to move to the instruction field
4710 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
4711 assembler allows semicolons as a line-breaking character. Note that some
4712 assembler dialects use semicolons to start a comment.
4713 The input operands are guaranteed not to use any of the clobbered
4714 registers, and neither will the output operands' addresses, so you can
4715 read and write the clobbered registers as many times as you like. Here
4716 is an example of multiple instructions in a template; it assumes the
4717 subroutine @code{_foo} accepts arguments in registers 9 and 10:
4720 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
4722 : "g" (from), "g" (to)
4726 Unless an output operand has the @samp{&} constraint modifier, GCC
4727 may allocate it in the same register as an unrelated input operand, on
4728 the assumption the inputs are consumed before the outputs are produced.
4729 This assumption may be false if the assembler code actually consists of
4730 more than one instruction. In such a case, use @samp{&} for each output
4731 operand that may not overlap an input. @xref{Modifiers}.
4733 If you want to test the condition code produced by an assembler
4734 instruction, you must include a branch and a label in the @code{asm}
4735 construct, as follows:
4738 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
4744 This assumes your assembler supports local labels, as the GNU assembler
4745 and most Unix assemblers do.
4747 Speaking of labels, jumps from one @code{asm} to another are not
4748 supported. The compiler's optimizers do not know about these jumps, and
4749 therefore they cannot take account of them when deciding how to
4752 @cindex macros containing @code{asm}
4753 Usually the most convenient way to use these @code{asm} instructions is to
4754 encapsulate them in macros that look like functions. For example,
4758 (@{ double __value, __arg = (x); \
4759 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
4764 Here the variable @code{__arg} is used to make sure that the instruction
4765 operates on a proper @code{double} value, and to accept only those
4766 arguments @code{x} which can convert automatically to a @code{double}.
4768 Another way to make sure the instruction operates on the correct data
4769 type is to use a cast in the @code{asm}. This is different from using a
4770 variable @code{__arg} in that it converts more different types. For
4771 example, if the desired type were @code{int}, casting the argument to
4772 @code{int} would accept a pointer with no complaint, while assigning the
4773 argument to an @code{int} variable named @code{__arg} would warn about
4774 using a pointer unless the caller explicitly casts it.
4776 If an @code{asm} has output operands, GCC assumes for optimization
4777 purposes the instruction has no side effects except to change the output
4778 operands. This does not mean instructions with a side effect cannot be
4779 used, but you must be careful, because the compiler may eliminate them
4780 if the output operands aren't used, or move them out of loops, or
4781 replace two with one if they constitute a common subexpression. Also,
4782 if your instruction does have a side effect on a variable that otherwise
4783 appears not to change, the old value of the variable may be reused later
4784 if it happens to be found in a register.
4786 You can prevent an @code{asm} instruction from being deleted
4787 by writing the keyword @code{volatile} after
4788 the @code{asm}. For example:
4791 #define get_and_set_priority(new) \
4793 asm volatile ("get_and_set_priority %0, %1" \
4794 : "=g" (__old) : "g" (new)); \
4799 The @code{volatile} keyword indicates that the instruction has
4800 important side-effects. GCC will not delete a volatile @code{asm} if
4801 it is reachable. (The instruction can still be deleted if GCC can
4802 prove that control-flow will never reach the location of the
4803 instruction.) Note that even a volatile @code{asm} instruction
4804 can be moved relative to other code, including across jump
4805 instructions. For example, on many targets there is a system
4806 register which can be set to control the rounding mode of
4807 floating point operations. You might try
4808 setting it with a volatile @code{asm}, like this PowerPC example:
4811 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
4816 This will not work reliably, as the compiler may move the addition back
4817 before the volatile @code{asm}. To make it work you need to add an
4818 artificial dependency to the @code{asm} referencing a variable in the code
4819 you don't want moved, for example:
4822 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
4826 Similarly, you can't expect a
4827 sequence of volatile @code{asm} instructions to remain perfectly
4828 consecutive. If you want consecutive output, use a single @code{asm}.
4829 Also, GCC will perform some optimizations across a volatile @code{asm}
4830 instruction; GCC does not ``forget everything'' when it encounters
4831 a volatile @code{asm} instruction the way some other compilers do.
4833 An @code{asm} instruction without any output operands will be treated
4834 identically to a volatile @code{asm} instruction.
4836 It is a natural idea to look for a way to give access to the condition
4837 code left by the assembler instruction. However, when we attempted to
4838 implement this, we found no way to make it work reliably. The problem
4839 is that output operands might need reloading, which would result in
4840 additional following ``store'' instructions. On most machines, these
4841 instructions would alter the condition code before there was time to
4842 test it. This problem doesn't arise for ordinary ``test'' and
4843 ``compare'' instructions because they don't have any output operands.
4845 For reasons similar to those described above, it is not possible to give
4846 an assembler instruction access to the condition code left by previous
4849 If you are writing a header file that should be includable in ISO C
4850 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
4853 @subsection Size of an @code{asm}
4855 Some targets require that GCC track the size of each instruction used in
4856 order to generate correct code. Because the final length of an
4857 @code{asm} is only known by the assembler, GCC must make an estimate as
4858 to how big it will be. The estimate is formed by counting the number of
4859 statements in the pattern of the @code{asm} and multiplying that by the
4860 length of the longest instruction on that processor. Statements in the
4861 @code{asm} are identified by newline characters and whatever statement
4862 separator characters are supported by the assembler; on most processors
4863 this is the `@code{;}' character.
4865 Normally, GCC's estimate is perfectly adequate to ensure that correct
4866 code is generated, but it is possible to confuse the compiler if you use
4867 pseudo instructions or assembler macros that expand into multiple real
4868 instructions or if you use assembler directives that expand to more
4869 space in the object file than would be needed for a single instruction.
4870 If this happens then the assembler will produce a diagnostic saying that
4871 a label is unreachable.
4873 @subsection i386 floating point asm operands
4875 There are several rules on the usage of stack-like regs in
4876 asm_operands insns. These rules apply only to the operands that are
4881 Given a set of input regs that die in an asm_operands, it is
4882 necessary to know which are implicitly popped by the asm, and
4883 which must be explicitly popped by gcc.
4885 An input reg that is implicitly popped by the asm must be
4886 explicitly clobbered, unless it is constrained to match an
4890 For any input reg that is implicitly popped by an asm, it is
4891 necessary to know how to adjust the stack to compensate for the pop.
4892 If any non-popped input is closer to the top of the reg-stack than
4893 the implicitly popped reg, it would not be possible to know what the
4894 stack looked like---it's not clear how the rest of the stack ``slides
4897 All implicitly popped input regs must be closer to the top of
4898 the reg-stack than any input that is not implicitly popped.
4900 It is possible that if an input dies in an insn, reload might
4901 use the input reg for an output reload. Consider this example:
4904 asm ("foo" : "=t" (a) : "f" (b));
4907 This asm says that input B is not popped by the asm, and that
4908 the asm pushes a result onto the reg-stack, i.e., the stack is one
4909 deeper after the asm than it was before. But, it is possible that
4910 reload will think that it can use the same reg for both the input and
4911 the output, if input B dies in this insn.
4913 If any input operand uses the @code{f} constraint, all output reg
4914 constraints must use the @code{&} earlyclobber.
4916 The asm above would be written as
4919 asm ("foo" : "=&t" (a) : "f" (b));
4923 Some operands need to be in particular places on the stack. All
4924 output operands fall in this category---there is no other way to
4925 know which regs the outputs appear in unless the user indicates
4926 this in the constraints.
4928 Output operands must specifically indicate which reg an output
4929 appears in after an asm. @code{=f} is not allowed: the operand
4930 constraints must select a class with a single reg.
4933 Output operands may not be ``inserted'' between existing stack regs.
4934 Since no 387 opcode uses a read/write operand, all output operands
4935 are dead before the asm_operands, and are pushed by the asm_operands.
4936 It makes no sense to push anywhere but the top of the reg-stack.
4938 Output operands must start at the top of the reg-stack: output
4939 operands may not ``skip'' a reg.
4942 Some asm statements may need extra stack space for internal
4943 calculations. This can be guaranteed by clobbering stack registers
4944 unrelated to the inputs and outputs.
4948 Here are a couple of reasonable asms to want to write. This asm
4949 takes one input, which is internally popped, and produces two outputs.
4952 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4955 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4956 and replaces them with one output. The user must code the @code{st(1)}
4957 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4960 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4966 @section Controlling Names Used in Assembler Code
4967 @cindex assembler names for identifiers
4968 @cindex names used in assembler code
4969 @cindex identifiers, names in assembler code
4971 You can specify the name to be used in the assembler code for a C
4972 function or variable by writing the @code{asm} (or @code{__asm__})
4973 keyword after the declarator as follows:
4976 int foo asm ("myfoo") = 2;
4980 This specifies that the name to be used for the variable @code{foo} in
4981 the assembler code should be @samp{myfoo} rather than the usual
4984 On systems where an underscore is normally prepended to the name of a C
4985 function or variable, this feature allows you to define names for the
4986 linker that do not start with an underscore.
4988 It does not make sense to use this feature with a non-static local
4989 variable since such variables do not have assembler names. If you are
4990 trying to put the variable in a particular register, see @ref{Explicit
4991 Reg Vars}. GCC presently accepts such code with a warning, but will
4992 probably be changed to issue an error, rather than a warning, in the
4995 You cannot use @code{asm} in this way in a function @emph{definition}; but
4996 you can get the same effect by writing a declaration for the function
4997 before its definition and putting @code{asm} there, like this:
5000 extern func () asm ("FUNC");
5007 It is up to you to make sure that the assembler names you choose do not
5008 conflict with any other assembler symbols. Also, you must not use a
5009 register name; that would produce completely invalid assembler code. GCC
5010 does not as yet have the ability to store static variables in registers.
5011 Perhaps that will be added.
5013 @node Explicit Reg Vars
5014 @section Variables in Specified Registers
5015 @cindex explicit register variables
5016 @cindex variables in specified registers
5017 @cindex specified registers
5018 @cindex registers, global allocation
5020 GNU C allows you to put a few global variables into specified hardware
5021 registers. You can also specify the register in which an ordinary
5022 register variable should be allocated.
5026 Global register variables reserve registers throughout the program.
5027 This may be useful in programs such as programming language
5028 interpreters which have a couple of global variables that are accessed
5032 Local register variables in specific registers do not reserve the
5033 registers, except at the point where they are used as input or output
5034 operands in an @code{asm} statement and the @code{asm} statement itself is
5035 not deleted. The compiler's data flow analysis is capable of determining
5036 where the specified registers contain live values, and where they are
5037 available for other uses. Stores into local register variables may be deleted
5038 when they appear to be dead according to dataflow analysis. References
5039 to local register variables may be deleted or moved or simplified.
5041 These local variables are sometimes convenient for use with the extended
5042 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
5043 output of the assembler instruction directly into a particular register.
5044 (This will work provided the register you specify fits the constraints
5045 specified for that operand in the @code{asm}.)
5053 @node Global Reg Vars
5054 @subsection Defining Global Register Variables
5055 @cindex global register variables
5056 @cindex registers, global variables in
5058 You can define a global register variable in GNU C like this:
5061 register int *foo asm ("a5");
5065 Here @code{a5} is the name of the register which should be used. Choose a
5066 register which is normally saved and restored by function calls on your
5067 machine, so that library routines will not clobber it.
5069 Naturally the register name is cpu-dependent, so you would need to
5070 conditionalize your program according to cpu type. The register
5071 @code{a5} would be a good choice on a 68000 for a variable of pointer
5072 type. On machines with register windows, be sure to choose a ``global''
5073 register that is not affected magically by the function call mechanism.
5075 In addition, operating systems on one type of cpu may differ in how they
5076 name the registers; then you would need additional conditionals. For
5077 example, some 68000 operating systems call this register @code{%a5}.
5079 Eventually there may be a way of asking the compiler to choose a register
5080 automatically, but first we need to figure out how it should choose and
5081 how to enable you to guide the choice. No solution is evident.
5083 Defining a global register variable in a certain register reserves that
5084 register entirely for this use, at least within the current compilation.
5085 The register will not be allocated for any other purpose in the functions
5086 in the current compilation. The register will not be saved and restored by
5087 these functions. Stores into this register are never deleted even if they
5088 would appear to be dead, but references may be deleted or moved or
5091 It is not safe to access the global register variables from signal
5092 handlers, or from more than one thread of control, because the system
5093 library routines may temporarily use the register for other things (unless
5094 you recompile them specially for the task at hand).
5096 @cindex @code{qsort}, and global register variables
5097 It is not safe for one function that uses a global register variable to
5098 call another such function @code{foo} by way of a third function
5099 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
5100 different source file in which the variable wasn't declared). This is
5101 because @code{lose} might save the register and put some other value there.
5102 For example, you can't expect a global register variable to be available in
5103 the comparison-function that you pass to @code{qsort}, since @code{qsort}
5104 might have put something else in that register. (If you are prepared to
5105 recompile @code{qsort} with the same global register variable, you can
5106 solve this problem.)
5108 If you want to recompile @code{qsort} or other source files which do not
5109 actually use your global register variable, so that they will not use that
5110 register for any other purpose, then it suffices to specify the compiler
5111 option @option{-ffixed-@var{reg}}. You need not actually add a global
5112 register declaration to their source code.
5114 A function which can alter the value of a global register variable cannot
5115 safely be called from a function compiled without this variable, because it
5116 could clobber the value the caller expects to find there on return.
5117 Therefore, the function which is the entry point into the part of the
5118 program that uses the global register variable must explicitly save and
5119 restore the value which belongs to its caller.
5121 @cindex register variable after @code{longjmp}
5122 @cindex global register after @code{longjmp}
5123 @cindex value after @code{longjmp}
5126 On most machines, @code{longjmp} will restore to each global register
5127 variable the value it had at the time of the @code{setjmp}. On some
5128 machines, however, @code{longjmp} will not change the value of global
5129 register variables. To be portable, the function that called @code{setjmp}
5130 should make other arrangements to save the values of the global register
5131 variables, and to restore them in a @code{longjmp}. This way, the same
5132 thing will happen regardless of what @code{longjmp} does.
5134 All global register variable declarations must precede all function
5135 definitions. If such a declaration could appear after function
5136 definitions, the declaration would be too late to prevent the register from
5137 being used for other purposes in the preceding functions.
5139 Global register variables may not have initial values, because an
5140 executable file has no means to supply initial contents for a register.
5142 On the SPARC, there are reports that g3 @dots{} g7 are suitable
5143 registers, but certain library functions, such as @code{getwd}, as well
5144 as the subroutines for division and remainder, modify g3 and g4. g1 and
5145 g2 are local temporaries.
5147 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
5148 Of course, it will not do to use more than a few of those.
5150 @node Local Reg Vars
5151 @subsection Specifying Registers for Local Variables
5152 @cindex local variables, specifying registers
5153 @cindex specifying registers for local variables
5154 @cindex registers for local variables
5156 You can define a local register variable with a specified register
5160 register int *foo asm ("a5");
5164 Here @code{a5} is the name of the register which should be used. Note
5165 that this is the same syntax used for defining global register
5166 variables, but for a local variable it would appear within a function.
5168 Naturally the register name is cpu-dependent, but this is not a
5169 problem, since specific registers are most often useful with explicit
5170 assembler instructions (@pxref{Extended Asm}). Both of these things
5171 generally require that you conditionalize your program according to
5174 In addition, operating systems on one type of cpu may differ in how they
5175 name the registers; then you would need additional conditionals. For
5176 example, some 68000 operating systems call this register @code{%a5}.
5178 Defining such a register variable does not reserve the register; it
5179 remains available for other uses in places where flow control determines
5180 the variable's value is not live.
5182 This option does not guarantee that GCC will generate code that has
5183 this variable in the register you specify at all times. You may not
5184 code an explicit reference to this register in the @emph{assembler
5185 instruction template} part of an @code{asm} statement and assume it will
5186 always refer to this variable. However, using the variable as an
5187 @code{asm} @emph{operand} guarantees that the specified register is used
5190 Stores into local register variables may be deleted when they appear to be dead
5191 according to dataflow analysis. References to local register variables may
5192 be deleted or moved or simplified.
5194 As for global register variables, it's recommended that you choose a
5195 register which is normally saved and restored by function calls on
5196 your machine, so that library routines will not clobber it. A common
5197 pitfall is to initialize multiple call-clobbered registers with
5198 arbitrary expressions, where a function call or library call for an
5199 arithmetic operator will overwrite a register value from a previous
5200 assignment, for example @code{r0} below:
5202 register int *p1 asm ("r0") = @dots{};
5203 register int *p2 asm ("r1") = @dots{};
5205 In those cases, a solution is to use a temporary variable for
5206 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
5208 @node Alternate Keywords
5209 @section Alternate Keywords
5210 @cindex alternate keywords
5211 @cindex keywords, alternate
5213 @option{-ansi} and the various @option{-std} options disable certain
5214 keywords. This causes trouble when you want to use GNU C extensions, or
5215 a general-purpose header file that should be usable by all programs,
5216 including ISO C programs. The keywords @code{asm}, @code{typeof} and
5217 @code{inline} are not available in programs compiled with
5218 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
5219 program compiled with @option{-std=c99}). The ISO C99 keyword
5220 @code{restrict} is only available when @option{-std=gnu99} (which will
5221 eventually be the default) or @option{-std=c99} (or the equivalent
5222 @option{-std=iso9899:1999}) is used.
5224 The way to solve these problems is to put @samp{__} at the beginning and
5225 end of each problematical keyword. For example, use @code{__asm__}
5226 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
5228 Other C compilers won't accept these alternative keywords; if you want to
5229 compile with another compiler, you can define the alternate keywords as
5230 macros to replace them with the customary keywords. It looks like this:
5238 @findex __extension__
5240 @option{-pedantic} and other options cause warnings for many GNU C extensions.
5242 prevent such warnings within one expression by writing
5243 @code{__extension__} before the expression. @code{__extension__} has no
5244 effect aside from this.
5246 @node Incomplete Enums
5247 @section Incomplete @code{enum} Types
5249 You can define an @code{enum} tag without specifying its possible values.
5250 This results in an incomplete type, much like what you get if you write
5251 @code{struct foo} without describing the elements. A later declaration
5252 which does specify the possible values completes the type.
5254 You can't allocate variables or storage using the type while it is
5255 incomplete. However, you can work with pointers to that type.
5257 This extension may not be very useful, but it makes the handling of
5258 @code{enum} more consistent with the way @code{struct} and @code{union}
5261 This extension is not supported by GNU C++.
5263 @node Function Names
5264 @section Function Names as Strings
5265 @cindex @code{__func__} identifier
5266 @cindex @code{__FUNCTION__} identifier
5267 @cindex @code{__PRETTY_FUNCTION__} identifier
5269 GCC provides three magic variables which hold the name of the current
5270 function, as a string. The first of these is @code{__func__}, which
5271 is part of the C99 standard:
5274 The identifier @code{__func__} is implicitly declared by the translator
5275 as if, immediately following the opening brace of each function
5276 definition, the declaration
5279 static const char __func__[] = "function-name";
5282 appeared, where function-name is the name of the lexically-enclosing
5283 function. This name is the unadorned name of the function.
5286 @code{__FUNCTION__} is another name for @code{__func__}. Older
5287 versions of GCC recognize only this name. However, it is not
5288 standardized. For maximum portability, we recommend you use
5289 @code{__func__}, but provide a fallback definition with the
5293 #if __STDC_VERSION__ < 199901L
5295 # define __func__ __FUNCTION__
5297 # define __func__ "<unknown>"
5302 In C, @code{__PRETTY_FUNCTION__} is yet another name for
5303 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
5304 the type signature of the function as well as its bare name. For
5305 example, this program:
5309 extern int printf (char *, ...);
5316 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
5317 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
5335 __PRETTY_FUNCTION__ = void a::sub(int)
5338 These identifiers are not preprocessor macros. In GCC 3.3 and
5339 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
5340 were treated as string literals; they could be used to initialize
5341 @code{char} arrays, and they could be concatenated with other string
5342 literals. GCC 3.4 and later treat them as variables, like
5343 @code{__func__}. In C++, @code{__FUNCTION__} and
5344 @code{__PRETTY_FUNCTION__} have always been variables.
5346 @node Return Address
5347 @section Getting the Return or Frame Address of a Function
5349 These functions may be used to get information about the callers of a
5352 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
5353 This function returns the return address of the current function, or of
5354 one of its callers. The @var{level} argument is number of frames to
5355 scan up the call stack. A value of @code{0} yields the return address
5356 of the current function, a value of @code{1} yields the return address
5357 of the caller of the current function, and so forth. When inlining
5358 the expected behavior is that the function will return the address of
5359 the function that will be returned to. To work around this behavior use
5360 the @code{noinline} function attribute.
5362 The @var{level} argument must be a constant integer.
5364 On some machines it may be impossible to determine the return address of
5365 any function other than the current one; in such cases, or when the top
5366 of the stack has been reached, this function will return @code{0} or a
5367 random value. In addition, @code{__builtin_frame_address} may be used
5368 to determine if the top of the stack has been reached.
5370 This function should only be used with a nonzero argument for debugging
5374 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
5375 This function is similar to @code{__builtin_return_address}, but it
5376 returns the address of the function frame rather than the return address
5377 of the function. Calling @code{__builtin_frame_address} with a value of
5378 @code{0} yields the frame address of the current function, a value of
5379 @code{1} yields the frame address of the caller of the current function,
5382 The frame is the area on the stack which holds local variables and saved
5383 registers. The frame address is normally the address of the first word
5384 pushed on to the stack by the function. However, the exact definition
5385 depends upon the processor and the calling convention. If the processor
5386 has a dedicated frame pointer register, and the function has a frame,
5387 then @code{__builtin_frame_address} will return the value of the frame
5390 On some machines it may be impossible to determine the frame address of
5391 any function other than the current one; in such cases, or when the top
5392 of the stack has been reached, this function will return @code{0} if
5393 the first frame pointer is properly initialized by the startup code.
5395 This function should only be used with a nonzero argument for debugging
5399 @node Vector Extensions
5400 @section Using vector instructions through built-in functions
5402 On some targets, the instruction set contains SIMD vector instructions that
5403 operate on multiple values contained in one large register at the same time.
5404 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
5407 The first step in using these extensions is to provide the necessary data
5408 types. This should be done using an appropriate @code{typedef}:
5411 typedef int v4si __attribute__ ((vector_size (16)));
5414 The @code{int} type specifies the base type, while the attribute specifies
5415 the vector size for the variable, measured in bytes. For example, the
5416 declaration above causes the compiler to set the mode for the @code{v4si}
5417 type to be 16 bytes wide and divided into @code{int} sized units. For
5418 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
5419 corresponding mode of @code{foo} will be @acronym{V4SI}.
5421 The @code{vector_size} attribute is only applicable to integral and
5422 float scalars, although arrays, pointers, and function return values
5423 are allowed in conjunction with this construct.
5425 All the basic integer types can be used as base types, both as signed
5426 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
5427 @code{long long}. In addition, @code{float} and @code{double} can be
5428 used to build floating-point vector types.
5430 Specifying a combination that is not valid for the current architecture
5431 will cause GCC to synthesize the instructions using a narrower mode.
5432 For example, if you specify a variable of type @code{V4SI} and your
5433 architecture does not allow for this specific SIMD type, GCC will
5434 produce code that uses 4 @code{SIs}.
5436 The types defined in this manner can be used with a subset of normal C
5437 operations. Currently, GCC will allow using the following operators
5438 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
5440 The operations behave like C++ @code{valarrays}. Addition is defined as
5441 the addition of the corresponding elements of the operands. For
5442 example, in the code below, each of the 4 elements in @var{a} will be
5443 added to the corresponding 4 elements in @var{b} and the resulting
5444 vector will be stored in @var{c}.
5447 typedef int v4si __attribute__ ((vector_size (16)));
5454 Subtraction, multiplication, division, and the logical operations
5455 operate in a similar manner. Likewise, the result of using the unary
5456 minus or complement operators on a vector type is a vector whose
5457 elements are the negative or complemented values of the corresponding
5458 elements in the operand.
5460 You can declare variables and use them in function calls and returns, as
5461 well as in assignments and some casts. You can specify a vector type as
5462 a return type for a function. Vector types can also be used as function
5463 arguments. It is possible to cast from one vector type to another,
5464 provided they are of the same size (in fact, you can also cast vectors
5465 to and from other datatypes of the same size).
5467 You cannot operate between vectors of different lengths or different
5468 signedness without a cast.
5470 A port that supports hardware vector operations, usually provides a set
5471 of built-in functions that can be used to operate on vectors. For
5472 example, a function to add two vectors and multiply the result by a
5473 third could look like this:
5476 v4si f (v4si a, v4si b, v4si c)
5478 v4si tmp = __builtin_addv4si (a, b);
5479 return __builtin_mulv4si (tmp, c);
5486 @findex __builtin_offsetof
5488 GCC implements for both C and C++ a syntactic extension to implement
5489 the @code{offsetof} macro.
5493 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
5495 offsetof_member_designator:
5497 | offsetof_member_designator "." @code{identifier}
5498 | offsetof_member_designator "[" @code{expr} "]"
5501 This extension is sufficient such that
5504 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
5507 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
5508 may be dependent. In either case, @var{member} may consist of a single
5509 identifier, or a sequence of member accesses and array references.
5511 @node Atomic Builtins
5512 @section Built-in functions for atomic memory access
5514 The following builtins are intended to be compatible with those described
5515 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
5516 section 7.4. As such, they depart from the normal GCC practice of using
5517 the ``__builtin_'' prefix, and further that they are overloaded such that
5518 they work on multiple types.
5520 The definition given in the Intel documentation allows only for the use of
5521 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
5522 counterparts. GCC will allow any integral scalar or pointer type that is
5523 1, 2, 4 or 8 bytes in length.
5525 Not all operations are supported by all target processors. If a particular
5526 operation cannot be implemented on the target processor, a warning will be
5527 generated and a call an external function will be generated. The external
5528 function will carry the same name as the builtin, with an additional suffix
5529 @samp{_@var{n}} where @var{n} is the size of the data type.
5531 @c ??? Should we have a mechanism to suppress this warning? This is almost
5532 @c useful for implementing the operation under the control of an external
5535 In most cases, these builtins are considered a @dfn{full barrier}. That is,
5536 no memory operand will be moved across the operation, either forward or
5537 backward. Further, instructions will be issued as necessary to prevent the
5538 processor from speculating loads across the operation and from queuing stores
5539 after the operation.
5541 All of the routines are are described in the Intel documentation to take
5542 ``an optional list of variables protected by the memory barrier''. It's
5543 not clear what is meant by that; it could mean that @emph{only} the
5544 following variables are protected, or it could mean that these variables
5545 should in addition be protected. At present GCC ignores this list and
5546 protects all variables which are globally accessible. If in the future
5547 we make some use of this list, an empty list will continue to mean all
5548 globally accessible variables.
5551 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
5552 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
5553 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
5554 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
5555 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
5556 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
5557 @findex __sync_fetch_and_add
5558 @findex __sync_fetch_and_sub
5559 @findex __sync_fetch_and_or
5560 @findex __sync_fetch_and_and
5561 @findex __sync_fetch_and_xor
5562 @findex __sync_fetch_and_nand
5563 These builtins perform the operation suggested by the name, and
5564 returns the value that had previously been in memory. That is,
5567 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
5568 @{ tmp = *ptr; *ptr = ~tmp & value; return tmp; @} // nand
5571 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
5572 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
5573 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
5574 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
5575 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
5576 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
5577 @findex __sync_add_and_fetch
5578 @findex __sync_sub_and_fetch
5579 @findex __sync_or_and_fetch
5580 @findex __sync_and_and_fetch
5581 @findex __sync_xor_and_fetch
5582 @findex __sync_nand_and_fetch
5583 These builtins perform the operation suggested by the name, and
5584 return the new value. That is,
5587 @{ *ptr @var{op}= value; return *ptr; @}
5588 @{ *ptr = ~*ptr & value; return *ptr; @} // nand
5591 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5592 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5593 @findex __sync_bool_compare_and_swap
5594 @findex __sync_val_compare_and_swap
5595 These builtins perform an atomic compare and swap. That is, if the current
5596 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
5599 The ``bool'' version returns true if the comparison is successful and
5600 @var{newval} was written. The ``val'' version returns the contents
5601 of @code{*@var{ptr}} before the operation.
5603 @item __sync_synchronize (...)
5604 @findex __sync_synchronize
5605 This builtin issues a full memory barrier.
5607 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
5608 @findex __sync_lock_test_and_set
5609 This builtin, as described by Intel, is not a traditional test-and-set
5610 operation, but rather an atomic exchange operation. It writes @var{value}
5611 into @code{*@var{ptr}}, and returns the previous contents of
5614 Many targets have only minimal support for such locks, and do not support
5615 a full exchange operation. In this case, a target may support reduced
5616 functionality here by which the @emph{only} valid value to store is the
5617 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
5618 is implementation defined.
5620 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
5621 This means that references after the builtin cannot move to (or be
5622 speculated to) before the builtin, but previous memory stores may not
5623 be globally visible yet, and previous memory loads may not yet be
5626 @item void __sync_lock_release (@var{type} *ptr, ...)
5627 @findex __sync_lock_release
5628 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
5629 Normally this means writing the constant 0 to @code{*@var{ptr}}.
5631 This builtin is not a full barrier, but rather a @dfn{release barrier}.
5632 This means that all previous memory stores are globally visible, and all
5633 previous memory loads have been satisfied, but following memory reads
5634 are not prevented from being speculated to before the barrier.
5637 @node Object Size Checking
5638 @section Object Size Checking Builtins
5639 @findex __builtin_object_size
5640 @findex __builtin___memcpy_chk
5641 @findex __builtin___mempcpy_chk
5642 @findex __builtin___memmove_chk
5643 @findex __builtin___memset_chk
5644 @findex __builtin___strcpy_chk
5645 @findex __builtin___stpcpy_chk
5646 @findex __builtin___strncpy_chk
5647 @findex __builtin___strcat_chk
5648 @findex __builtin___strncat_chk
5649 @findex __builtin___sprintf_chk
5650 @findex __builtin___snprintf_chk
5651 @findex __builtin___vsprintf_chk
5652 @findex __builtin___vsnprintf_chk
5653 @findex __builtin___printf_chk
5654 @findex __builtin___vprintf_chk
5655 @findex __builtin___fprintf_chk
5656 @findex __builtin___vfprintf_chk
5658 GCC implements a limited buffer overflow protection mechanism
5659 that can prevent some buffer overflow attacks.
5661 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
5662 is a built-in construct that returns a constant number of bytes from
5663 @var{ptr} to the end of the object @var{ptr} pointer points to
5664 (if known at compile time). @code{__builtin_object_size} never evaluates
5665 its arguments for side-effects. If there are any side-effects in them, it
5666 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5667 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
5668 point to and all of them are known at compile time, the returned number
5669 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
5670 0 and minimum if nonzero. If it is not possible to determine which objects
5671 @var{ptr} points to at compile time, @code{__builtin_object_size} should
5672 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5673 for @var{type} 2 or 3.
5675 @var{type} is an integer constant from 0 to 3. If the least significant
5676 bit is clear, objects are whole variables, if it is set, a closest
5677 surrounding subobject is considered the object a pointer points to.
5678 The second bit determines if maximum or minimum of remaining bytes
5682 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
5683 char *p = &var.buf1[1], *q = &var.b;
5685 /* Here the object p points to is var. */
5686 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
5687 /* The subobject p points to is var.buf1. */
5688 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
5689 /* The object q points to is var. */
5690 assert (__builtin_object_size (q, 0)
5691 == (char *) (&var + 1) - (char *) &var.b);
5692 /* The subobject q points to is var.b. */
5693 assert (__builtin_object_size (q, 1) == sizeof (var.b));
5697 There are built-in functions added for many common string operation
5698 functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk}
5699 built-in is provided. This built-in has an additional last argument,
5700 which is the number of bytes remaining in object the @var{dest}
5701 argument points to or @code{(size_t) -1} if the size is not known.
5703 The built-in functions are optimized into the normal string functions
5704 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
5705 it is known at compile time that the destination object will not
5706 be overflown. If the compiler can determine at compile time the
5707 object will be always overflown, it issues a warning.
5709 The intended use can be e.g.
5713 #define bos0(dest) __builtin_object_size (dest, 0)
5714 #define memcpy(dest, src, n) \
5715 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
5719 /* It is unknown what object p points to, so this is optimized
5720 into plain memcpy - no checking is possible. */
5721 memcpy (p, "abcde", n);
5722 /* Destination is known and length too. It is known at compile
5723 time there will be no overflow. */
5724 memcpy (&buf[5], "abcde", 5);
5725 /* Destination is known, but the length is not known at compile time.
5726 This will result in __memcpy_chk call that can check for overflow
5728 memcpy (&buf[5], "abcde", n);
5729 /* Destination is known and it is known at compile time there will
5730 be overflow. There will be a warning and __memcpy_chk call that
5731 will abort the program at runtime. */
5732 memcpy (&buf[6], "abcde", 5);
5735 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
5736 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
5737 @code{strcat} and @code{strncat}.
5739 There are also checking built-in functions for formatted output functions.
5741 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
5742 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5743 const char *fmt, ...);
5744 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
5746 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5747 const char *fmt, va_list ap);
5750 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
5751 etc.@: functions and can contain implementation specific flags on what
5752 additional security measures the checking function might take, such as
5753 handling @code{%n} differently.
5755 The @var{os} argument is the object size @var{s} points to, like in the
5756 other built-in functions. There is a small difference in the behavior
5757 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
5758 optimized into the non-checking functions only if @var{flag} is 0, otherwise
5759 the checking function is called with @var{os} argument set to
5762 In addition to this, there are checking built-in functions
5763 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
5764 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
5765 These have just one additional argument, @var{flag}, right before
5766 format string @var{fmt}. If the compiler is able to optimize them to
5767 @code{fputc} etc.@: functions, it will, otherwise the checking function
5768 should be called and the @var{flag} argument passed to it.
5770 @node Other Builtins
5771 @section Other built-in functions provided by GCC
5772 @cindex built-in functions
5773 @findex __builtin_fpclassify
5774 @findex __builtin_isfinite
5775 @findex __builtin_isnormal
5776 @findex __builtin_isgreater
5777 @findex __builtin_isgreaterequal
5778 @findex __builtin_isinf_sign
5779 @findex __builtin_isless
5780 @findex __builtin_islessequal
5781 @findex __builtin_islessgreater
5782 @findex __builtin_isunordered
5783 @findex __builtin_powi
5784 @findex __builtin_powif
5785 @findex __builtin_powil
5943 @findex fprintf_unlocked
5945 @findex fputs_unlocked
6062 @findex printf_unlocked
6094 @findex significandf
6095 @findex significandl
6166 GCC provides a large number of built-in functions other than the ones
6167 mentioned above. Some of these are for internal use in the processing
6168 of exceptions or variable-length argument lists and will not be
6169 documented here because they may change from time to time; we do not
6170 recommend general use of these functions.
6172 The remaining functions are provided for optimization purposes.
6174 @opindex fno-builtin
6175 GCC includes built-in versions of many of the functions in the standard
6176 C library. The versions prefixed with @code{__builtin_} will always be
6177 treated as having the same meaning as the C library function even if you
6178 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
6179 Many of these functions are only optimized in certain cases; if they are
6180 not optimized in a particular case, a call to the library function will
6185 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
6186 @option{-std=c99}), the functions
6187 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
6188 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
6189 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
6190 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
6191 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
6192 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
6193 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
6194 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
6195 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
6196 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
6197 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
6198 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
6199 @code{signbitd64}, @code{signbitd128}, @code{significandf},
6200 @code{significandl}, @code{significand}, @code{sincosf},
6201 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
6202 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
6203 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
6204 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
6206 may be handled as built-in functions.
6207 All these functions have corresponding versions
6208 prefixed with @code{__builtin_}, which may be used even in strict C89
6211 The ISO C99 functions
6212 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
6213 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
6214 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
6215 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
6216 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
6217 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
6218 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
6219 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
6220 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
6221 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
6222 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
6223 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
6224 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
6225 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
6226 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
6227 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
6228 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
6229 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
6230 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
6231 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
6232 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
6233 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
6234 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
6235 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
6236 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
6237 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
6238 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
6239 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
6240 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
6241 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
6242 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
6243 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
6244 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
6245 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
6246 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
6247 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
6248 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
6249 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
6250 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
6251 are handled as built-in functions
6252 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
6254 There are also built-in versions of the ISO C99 functions
6255 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
6256 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
6257 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
6258 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
6259 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
6260 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
6261 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
6262 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
6263 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
6264 that are recognized in any mode since ISO C90 reserves these names for
6265 the purpose to which ISO C99 puts them. All these functions have
6266 corresponding versions prefixed with @code{__builtin_}.
6268 The ISO C94 functions
6269 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
6270 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
6271 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
6273 are handled as built-in functions
6274 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
6276 The ISO C90 functions
6277 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
6278 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
6279 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
6280 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
6281 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
6282 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
6283 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
6284 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
6285 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
6286 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
6287 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
6288 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
6289 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
6290 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
6291 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
6292 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
6293 are all recognized as built-in functions unless
6294 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
6295 is specified for an individual function). All of these functions have
6296 corresponding versions prefixed with @code{__builtin_}.
6298 GCC provides built-in versions of the ISO C99 floating point comparison
6299 macros that avoid raising exceptions for unordered operands. They have
6300 the same names as the standard macros ( @code{isgreater},
6301 @code{isgreaterequal}, @code{isless}, @code{islessequal},
6302 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
6303 prefixed. We intend for a library implementor to be able to simply
6304 @code{#define} each standard macro to its built-in equivalent.
6305 In the same fashion, GCC provides @code{fpclassify}, @code{isfinite},
6306 @code{isinf_sign} and @code{isnormal} built-ins used with
6307 @code{__builtin_} prefixed. The @code{isinf} and @code{isnan}
6308 builtins appear both with and without the @code{__builtin_} prefix.
6310 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
6312 You can use the built-in function @code{__builtin_types_compatible_p} to
6313 determine whether two types are the same.
6315 This built-in function returns 1 if the unqualified versions of the
6316 types @var{type1} and @var{type2} (which are types, not expressions) are
6317 compatible, 0 otherwise. The result of this built-in function can be
6318 used in integer constant expressions.
6320 This built-in function ignores top level qualifiers (e.g., @code{const},
6321 @code{volatile}). For example, @code{int} is equivalent to @code{const
6324 The type @code{int[]} and @code{int[5]} are compatible. On the other
6325 hand, @code{int} and @code{char *} are not compatible, even if the size
6326 of their types, on the particular architecture are the same. Also, the
6327 amount of pointer indirection is taken into account when determining
6328 similarity. Consequently, @code{short *} is not similar to
6329 @code{short **}. Furthermore, two types that are typedefed are
6330 considered compatible if their underlying types are compatible.
6332 An @code{enum} type is not considered to be compatible with another
6333 @code{enum} type even if both are compatible with the same integer
6334 type; this is what the C standard specifies.
6335 For example, @code{enum @{foo, bar@}} is not similar to
6336 @code{enum @{hot, dog@}}.
6338 You would typically use this function in code whose execution varies
6339 depending on the arguments' types. For example:
6344 typeof (x) tmp = (x); \
6345 if (__builtin_types_compatible_p (typeof (x), long double)) \
6346 tmp = foo_long_double (tmp); \
6347 else if (__builtin_types_compatible_p (typeof (x), double)) \
6348 tmp = foo_double (tmp); \
6349 else if (__builtin_types_compatible_p (typeof (x), float)) \
6350 tmp = foo_float (tmp); \
6357 @emph{Note:} This construct is only available for C@.
6361 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
6363 You can use the built-in function @code{__builtin_choose_expr} to
6364 evaluate code depending on the value of a constant expression. This
6365 built-in function returns @var{exp1} if @var{const_exp}, which is a
6366 constant expression that must be able to be determined at compile time,
6367 is nonzero. Otherwise it returns 0.
6369 This built-in function is analogous to the @samp{? :} operator in C,
6370 except that the expression returned has its type unaltered by promotion
6371 rules. Also, the built-in function does not evaluate the expression
6372 that was not chosen. For example, if @var{const_exp} evaluates to true,
6373 @var{exp2} is not evaluated even if it has side-effects.
6375 This built-in function can return an lvalue if the chosen argument is an
6378 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
6379 type. Similarly, if @var{exp2} is returned, its return type is the same
6386 __builtin_choose_expr ( \
6387 __builtin_types_compatible_p (typeof (x), double), \
6389 __builtin_choose_expr ( \
6390 __builtin_types_compatible_p (typeof (x), float), \
6392 /* @r{The void expression results in a compile-time error} \
6393 @r{when assigning the result to something.} */ \
6397 @emph{Note:} This construct is only available for C@. Furthermore, the
6398 unused expression (@var{exp1} or @var{exp2} depending on the value of
6399 @var{const_exp}) may still generate syntax errors. This may change in
6404 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
6405 You can use the built-in function @code{__builtin_constant_p} to
6406 determine if a value is known to be constant at compile-time and hence
6407 that GCC can perform constant-folding on expressions involving that
6408 value. The argument of the function is the value to test. The function
6409 returns the integer 1 if the argument is known to be a compile-time
6410 constant and 0 if it is not known to be a compile-time constant. A
6411 return of 0 does not indicate that the value is @emph{not} a constant,
6412 but merely that GCC cannot prove it is a constant with the specified
6413 value of the @option{-O} option.
6415 You would typically use this function in an embedded application where
6416 memory was a critical resource. If you have some complex calculation,
6417 you may want it to be folded if it involves constants, but need to call
6418 a function if it does not. For example:
6421 #define Scale_Value(X) \
6422 (__builtin_constant_p (X) \
6423 ? ((X) * SCALE + OFFSET) : Scale (X))
6426 You may use this built-in function in either a macro or an inline
6427 function. However, if you use it in an inlined function and pass an
6428 argument of the function as the argument to the built-in, GCC will
6429 never return 1 when you call the inline function with a string constant
6430 or compound literal (@pxref{Compound Literals}) and will not return 1
6431 when you pass a constant numeric value to the inline function unless you
6432 specify the @option{-O} option.
6434 You may also use @code{__builtin_constant_p} in initializers for static
6435 data. For instance, you can write
6438 static const int table[] = @{
6439 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
6445 This is an acceptable initializer even if @var{EXPRESSION} is not a
6446 constant expression. GCC must be more conservative about evaluating the
6447 built-in in this case, because it has no opportunity to perform
6450 Previous versions of GCC did not accept this built-in in data
6451 initializers. The earliest version where it is completely safe is
6455 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
6456 @opindex fprofile-arcs
6457 You may use @code{__builtin_expect} to provide the compiler with
6458 branch prediction information. In general, you should prefer to
6459 use actual profile feedback for this (@option{-fprofile-arcs}), as
6460 programmers are notoriously bad at predicting how their programs
6461 actually perform. However, there are applications in which this
6462 data is hard to collect.
6464 The return value is the value of @var{exp}, which should be an integral
6465 expression. The semantics of the built-in are that it is expected that
6466 @var{exp} == @var{c}. For example:
6469 if (__builtin_expect (x, 0))
6474 would indicate that we do not expect to call @code{foo}, since
6475 we expect @code{x} to be zero. Since you are limited to integral
6476 expressions for @var{exp}, you should use constructions such as
6479 if (__builtin_expect (ptr != NULL, 1))
6484 when testing pointer or floating-point values.
6487 @deftypefn {Built-in Function} void __builtin_trap (void)
6488 This function causes the program to exit abnormally. GCC implements
6489 this function by using a target-dependent mechanism (such as
6490 intentionally executing an illegal instruction) or by calling
6491 @code{abort}. The mechanism used may vary from release to release so
6492 you should not rely on any particular implementation.
6495 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
6496 This function is used to flush the processor's instruction cache for
6497 the region of memory between @var{begin} inclusive and @var{end}
6498 exclusive. Some targets require that the instruction cache be
6499 flushed, after modifying memory containing code, in order to obtain
6500 deterministic behavior.
6502 If the target does not require instruction cache flushes,
6503 @code{__builtin___clear_cache} has no effect. Otherwise either
6504 instructions are emitted in-line to clear the instruction cache or a
6505 call to the @code{__clear_cache} function in libgcc is made.
6508 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
6509 This function is used to minimize cache-miss latency by moving data into
6510 a cache before it is accessed.
6511 You can insert calls to @code{__builtin_prefetch} into code for which
6512 you know addresses of data in memory that is likely to be accessed soon.
6513 If the target supports them, data prefetch instructions will be generated.
6514 If the prefetch is done early enough before the access then the data will
6515 be in the cache by the time it is accessed.
6517 The value of @var{addr} is the address of the memory to prefetch.
6518 There are two optional arguments, @var{rw} and @var{locality}.
6519 The value of @var{rw} is a compile-time constant one or zero; one
6520 means that the prefetch is preparing for a write to the memory address
6521 and zero, the default, means that the prefetch is preparing for a read.
6522 The value @var{locality} must be a compile-time constant integer between
6523 zero and three. A value of zero means that the data has no temporal
6524 locality, so it need not be left in the cache after the access. A value
6525 of three means that the data has a high degree of temporal locality and
6526 should be left in all levels of cache possible. Values of one and two
6527 mean, respectively, a low or moderate degree of temporal locality. The
6531 for (i = 0; i < n; i++)
6534 __builtin_prefetch (&a[i+j], 1, 1);
6535 __builtin_prefetch (&b[i+j], 0, 1);
6540 Data prefetch does not generate faults if @var{addr} is invalid, but
6541 the address expression itself must be valid. For example, a prefetch
6542 of @code{p->next} will not fault if @code{p->next} is not a valid
6543 address, but evaluation will fault if @code{p} is not a valid address.
6545 If the target does not support data prefetch, the address expression
6546 is evaluated if it includes side effects but no other code is generated
6547 and GCC does not issue a warning.
6550 @deftypefn {Built-in Function} double __builtin_huge_val (void)
6551 Returns a positive infinity, if supported by the floating-point format,
6552 else @code{DBL_MAX}. This function is suitable for implementing the
6553 ISO C macro @code{HUGE_VAL}.
6556 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
6557 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
6560 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
6561 Similar to @code{__builtin_huge_val}, except the return
6562 type is @code{long double}.
6565 @deftypefn {Built-in Function} int __builtin_fpclassify (int, int, int, int, int, ...)
6566 This built-in implements the C99 fpclassify functionality. The first
6567 five int arguments should be the target library's notion of the
6568 possible FP classes and are used for return values. They must be
6569 constant values and they must appear in this order: @code{FP_NAN},
6570 @code{FP_INF}, @code{FP_NORMAL}, @code{FP_SUBNORMAL} and
6571 @code{FP_ZERO}. The ellipsis is for exactly one floating point value
6572 to classify. GCC treats the last argument as type-generic, which
6573 means it does not do default promotion from float to double.
6576 @deftypefn {Built-in Function} double __builtin_inf (void)
6577 Similar to @code{__builtin_huge_val}, except a warning is generated
6578 if the target floating-point format does not support infinities.
6581 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
6582 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
6585 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
6586 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
6589 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
6590 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
6593 @deftypefn {Built-in Function} float __builtin_inff (void)
6594 Similar to @code{__builtin_inf}, except the return type is @code{float}.
6595 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
6598 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
6599 Similar to @code{__builtin_inf}, except the return
6600 type is @code{long double}.
6603 @deftypefn {Built-in Function} int __builtin_isinf_sign (...)
6604 Similar to @code{isinf}, except the return value will be negative for
6605 an argument of @code{-Inf}. Note while the parameter list is an
6606 ellipsis, this function only accepts exactly one floating point
6607 argument. GCC treats this parameter as type-generic, which means it
6608 does not do default promotion from float to double.
6611 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
6612 This is an implementation of the ISO C99 function @code{nan}.
6614 Since ISO C99 defines this function in terms of @code{strtod}, which we
6615 do not implement, a description of the parsing is in order. The string
6616 is parsed as by @code{strtol}; that is, the base is recognized by
6617 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
6618 in the significand such that the least significant bit of the number
6619 is at the least significant bit of the significand. The number is
6620 truncated to fit the significand field provided. The significand is
6621 forced to be a quiet NaN@.
6623 This function, if given a string literal all of which would have been
6624 consumed by strtol, is evaluated early enough that it is considered a
6625 compile-time constant.
6628 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
6629 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
6632 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
6633 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
6636 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
6637 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
6640 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
6641 Similar to @code{__builtin_nan}, except the return type is @code{float}.
6644 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
6645 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
6648 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
6649 Similar to @code{__builtin_nan}, except the significand is forced
6650 to be a signaling NaN@. The @code{nans} function is proposed by
6651 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
6654 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
6655 Similar to @code{__builtin_nans}, except the return type is @code{float}.
6658 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
6659 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
6662 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
6663 Returns one plus the index of the least significant 1-bit of @var{x}, or
6664 if @var{x} is zero, returns zero.
6667 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
6668 Returns the number of leading 0-bits in @var{x}, starting at the most
6669 significant bit position. If @var{x} is 0, the result is undefined.
6672 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
6673 Returns the number of trailing 0-bits in @var{x}, starting at the least
6674 significant bit position. If @var{x} is 0, the result is undefined.
6677 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
6678 Returns the number of 1-bits in @var{x}.
6681 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
6682 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
6686 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
6687 Similar to @code{__builtin_ffs}, except the argument type is
6688 @code{unsigned long}.
6691 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
6692 Similar to @code{__builtin_clz}, except the argument type is
6693 @code{unsigned long}.
6696 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
6697 Similar to @code{__builtin_ctz}, except the argument type is
6698 @code{unsigned long}.
6701 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
6702 Similar to @code{__builtin_popcount}, except the argument type is
6703 @code{unsigned long}.
6706 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
6707 Similar to @code{__builtin_parity}, except the argument type is
6708 @code{unsigned long}.
6711 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
6712 Similar to @code{__builtin_ffs}, except the argument type is
6713 @code{unsigned long long}.
6716 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
6717 Similar to @code{__builtin_clz}, except the argument type is
6718 @code{unsigned long long}.
6721 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
6722 Similar to @code{__builtin_ctz}, except the argument type is
6723 @code{unsigned long long}.
6726 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
6727 Similar to @code{__builtin_popcount}, except the argument type is
6728 @code{unsigned long long}.
6731 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
6732 Similar to @code{__builtin_parity}, except the argument type is
6733 @code{unsigned long long}.
6736 @deftypefn {Built-in Function} double __builtin_powi (double, int)
6737 Returns the first argument raised to the power of the second. Unlike the
6738 @code{pow} function no guarantees about precision and rounding are made.
6741 @deftypefn {Built-in Function} float __builtin_powif (float, int)
6742 Similar to @code{__builtin_powi}, except the argument and return types
6746 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
6747 Similar to @code{__builtin_powi}, except the argument and return types
6748 are @code{long double}.
6751 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
6752 Returns @var{x} with the order of the bytes reversed; for example,
6753 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
6757 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
6758 Similar to @code{__builtin_bswap32}, except the argument and return types
6762 @node Target Builtins
6763 @section Built-in Functions Specific to Particular Target Machines
6765 On some target machines, GCC supports many built-in functions specific
6766 to those machines. Generally these generate calls to specific machine
6767 instructions, but allow the compiler to schedule those calls.
6770 * Alpha Built-in Functions::
6771 * ARM iWMMXt Built-in Functions::
6772 * ARM NEON Intrinsics::
6773 * Blackfin Built-in Functions::
6774 * FR-V Built-in Functions::
6775 * X86 Built-in Functions::
6776 * MIPS DSP Built-in Functions::
6777 * MIPS Paired-Single Support::
6778 * PowerPC AltiVec Built-in Functions::
6779 * SPARC VIS Built-in Functions::
6780 * SPU Built-in Functions::
6783 @node Alpha Built-in Functions
6784 @subsection Alpha Built-in Functions
6786 These built-in functions are available for the Alpha family of
6787 processors, depending on the command-line switches used.
6789 The following built-in functions are always available. They
6790 all generate the machine instruction that is part of the name.
6793 long __builtin_alpha_implver (void)
6794 long __builtin_alpha_rpcc (void)
6795 long __builtin_alpha_amask (long)
6796 long __builtin_alpha_cmpbge (long, long)
6797 long __builtin_alpha_extbl (long, long)
6798 long __builtin_alpha_extwl (long, long)
6799 long __builtin_alpha_extll (long, long)
6800 long __builtin_alpha_extql (long, long)
6801 long __builtin_alpha_extwh (long, long)
6802 long __builtin_alpha_extlh (long, long)
6803 long __builtin_alpha_extqh (long, long)
6804 long __builtin_alpha_insbl (long, long)
6805 long __builtin_alpha_inswl (long, long)
6806 long __builtin_alpha_insll (long, long)
6807 long __builtin_alpha_insql (long, long)
6808 long __builtin_alpha_inswh (long, long)
6809 long __builtin_alpha_inslh (long, long)
6810 long __builtin_alpha_insqh (long, long)
6811 long __builtin_alpha_mskbl (long, long)
6812 long __builtin_alpha_mskwl (long, long)
6813 long __builtin_alpha_mskll (long, long)
6814 long __builtin_alpha_mskql (long, long)
6815 long __builtin_alpha_mskwh (long, long)
6816 long __builtin_alpha_msklh (long, long)
6817 long __builtin_alpha_mskqh (long, long)
6818 long __builtin_alpha_umulh (long, long)
6819 long __builtin_alpha_zap (long, long)
6820 long __builtin_alpha_zapnot (long, long)
6823 The following built-in functions are always with @option{-mmax}
6824 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
6825 later. They all generate the machine instruction that is part
6829 long __builtin_alpha_pklb (long)
6830 long __builtin_alpha_pkwb (long)
6831 long __builtin_alpha_unpkbl (long)
6832 long __builtin_alpha_unpkbw (long)
6833 long __builtin_alpha_minub8 (long, long)
6834 long __builtin_alpha_minsb8 (long, long)
6835 long __builtin_alpha_minuw4 (long, long)
6836 long __builtin_alpha_minsw4 (long, long)
6837 long __builtin_alpha_maxub8 (long, long)
6838 long __builtin_alpha_maxsb8 (long, long)
6839 long __builtin_alpha_maxuw4 (long, long)
6840 long __builtin_alpha_maxsw4 (long, long)
6841 long __builtin_alpha_perr (long, long)
6844 The following built-in functions are always with @option{-mcix}
6845 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
6846 later. They all generate the machine instruction that is part
6850 long __builtin_alpha_cttz (long)
6851 long __builtin_alpha_ctlz (long)
6852 long __builtin_alpha_ctpop (long)
6855 The following builtins are available on systems that use the OSF/1
6856 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
6857 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
6858 @code{rdval} and @code{wrval}.
6861 void *__builtin_thread_pointer (void)
6862 void __builtin_set_thread_pointer (void *)
6865 @node ARM iWMMXt Built-in Functions
6866 @subsection ARM iWMMXt Built-in Functions
6868 These built-in functions are available for the ARM family of
6869 processors when the @option{-mcpu=iwmmxt} switch is used:
6872 typedef int v2si __attribute__ ((vector_size (8)));
6873 typedef short v4hi __attribute__ ((vector_size (8)));
6874 typedef char v8qi __attribute__ ((vector_size (8)));
6876 int __builtin_arm_getwcx (int)
6877 void __builtin_arm_setwcx (int, int)
6878 int __builtin_arm_textrmsb (v8qi, int)
6879 int __builtin_arm_textrmsh (v4hi, int)
6880 int __builtin_arm_textrmsw (v2si, int)
6881 int __builtin_arm_textrmub (v8qi, int)
6882 int __builtin_arm_textrmuh (v4hi, int)
6883 int __builtin_arm_textrmuw (v2si, int)
6884 v8qi __builtin_arm_tinsrb (v8qi, int)
6885 v4hi __builtin_arm_tinsrh (v4hi, int)
6886 v2si __builtin_arm_tinsrw (v2si, int)
6887 long long __builtin_arm_tmia (long long, int, int)
6888 long long __builtin_arm_tmiabb (long long, int, int)
6889 long long __builtin_arm_tmiabt (long long, int, int)
6890 long long __builtin_arm_tmiaph (long long, int, int)
6891 long long __builtin_arm_tmiatb (long long, int, int)
6892 long long __builtin_arm_tmiatt (long long, int, int)
6893 int __builtin_arm_tmovmskb (v8qi)
6894 int __builtin_arm_tmovmskh (v4hi)
6895 int __builtin_arm_tmovmskw (v2si)
6896 long long __builtin_arm_waccb (v8qi)
6897 long long __builtin_arm_wacch (v4hi)
6898 long long __builtin_arm_waccw (v2si)
6899 v8qi __builtin_arm_waddb (v8qi, v8qi)
6900 v8qi __builtin_arm_waddbss (v8qi, v8qi)
6901 v8qi __builtin_arm_waddbus (v8qi, v8qi)
6902 v4hi __builtin_arm_waddh (v4hi, v4hi)
6903 v4hi __builtin_arm_waddhss (v4hi, v4hi)
6904 v4hi __builtin_arm_waddhus (v4hi, v4hi)
6905 v2si __builtin_arm_waddw (v2si, v2si)
6906 v2si __builtin_arm_waddwss (v2si, v2si)
6907 v2si __builtin_arm_waddwus (v2si, v2si)
6908 v8qi __builtin_arm_walign (v8qi, v8qi, int)
6909 long long __builtin_arm_wand(long long, long long)
6910 long long __builtin_arm_wandn (long long, long long)
6911 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
6912 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
6913 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
6914 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
6915 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
6916 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
6917 v2si __builtin_arm_wcmpeqw (v2si, v2si)
6918 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
6919 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
6920 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
6921 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
6922 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
6923 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
6924 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
6925 long long __builtin_arm_wmacsz (v4hi, v4hi)
6926 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
6927 long long __builtin_arm_wmacuz (v4hi, v4hi)
6928 v4hi __builtin_arm_wmadds (v4hi, v4hi)
6929 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
6930 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
6931 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
6932 v2si __builtin_arm_wmaxsw (v2si, v2si)
6933 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
6934 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
6935 v2si __builtin_arm_wmaxuw (v2si, v2si)
6936 v8qi __builtin_arm_wminsb (v8qi, v8qi)
6937 v4hi __builtin_arm_wminsh (v4hi, v4hi)
6938 v2si __builtin_arm_wminsw (v2si, v2si)
6939 v8qi __builtin_arm_wminub (v8qi, v8qi)
6940 v4hi __builtin_arm_wminuh (v4hi, v4hi)
6941 v2si __builtin_arm_wminuw (v2si, v2si)
6942 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
6943 v4hi __builtin_arm_wmulul (v4hi, v4hi)
6944 v4hi __builtin_arm_wmulum (v4hi, v4hi)
6945 long long __builtin_arm_wor (long long, long long)
6946 v2si __builtin_arm_wpackdss (long long, long long)
6947 v2si __builtin_arm_wpackdus (long long, long long)
6948 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
6949 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
6950 v4hi __builtin_arm_wpackwss (v2si, v2si)
6951 v4hi __builtin_arm_wpackwus (v2si, v2si)
6952 long long __builtin_arm_wrord (long long, long long)
6953 long long __builtin_arm_wrordi (long long, int)
6954 v4hi __builtin_arm_wrorh (v4hi, long long)
6955 v4hi __builtin_arm_wrorhi (v4hi, int)
6956 v2si __builtin_arm_wrorw (v2si, long long)
6957 v2si __builtin_arm_wrorwi (v2si, int)
6958 v2si __builtin_arm_wsadb (v8qi, v8qi)
6959 v2si __builtin_arm_wsadbz (v8qi, v8qi)
6960 v2si __builtin_arm_wsadh (v4hi, v4hi)
6961 v2si __builtin_arm_wsadhz (v4hi, v4hi)
6962 v4hi __builtin_arm_wshufh (v4hi, int)
6963 long long __builtin_arm_wslld (long long, long long)
6964 long long __builtin_arm_wslldi (long long, int)
6965 v4hi __builtin_arm_wsllh (v4hi, long long)
6966 v4hi __builtin_arm_wsllhi (v4hi, int)
6967 v2si __builtin_arm_wsllw (v2si, long long)
6968 v2si __builtin_arm_wsllwi (v2si, int)
6969 long long __builtin_arm_wsrad (long long, long long)
6970 long long __builtin_arm_wsradi (long long, int)
6971 v4hi __builtin_arm_wsrah (v4hi, long long)
6972 v4hi __builtin_arm_wsrahi (v4hi, int)
6973 v2si __builtin_arm_wsraw (v2si, long long)
6974 v2si __builtin_arm_wsrawi (v2si, int)
6975 long long __builtin_arm_wsrld (long long, long long)
6976 long long __builtin_arm_wsrldi (long long, int)
6977 v4hi __builtin_arm_wsrlh (v4hi, long long)
6978 v4hi __builtin_arm_wsrlhi (v4hi, int)
6979 v2si __builtin_arm_wsrlw (v2si, long long)
6980 v2si __builtin_arm_wsrlwi (v2si, int)
6981 v8qi __builtin_arm_wsubb (v8qi, v8qi)
6982 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
6983 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
6984 v4hi __builtin_arm_wsubh (v4hi, v4hi)
6985 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
6986 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
6987 v2si __builtin_arm_wsubw (v2si, v2si)
6988 v2si __builtin_arm_wsubwss (v2si, v2si)
6989 v2si __builtin_arm_wsubwus (v2si, v2si)
6990 v4hi __builtin_arm_wunpckehsb (v8qi)
6991 v2si __builtin_arm_wunpckehsh (v4hi)
6992 long long __builtin_arm_wunpckehsw (v2si)
6993 v4hi __builtin_arm_wunpckehub (v8qi)
6994 v2si __builtin_arm_wunpckehuh (v4hi)
6995 long long __builtin_arm_wunpckehuw (v2si)
6996 v4hi __builtin_arm_wunpckelsb (v8qi)
6997 v2si __builtin_arm_wunpckelsh (v4hi)
6998 long long __builtin_arm_wunpckelsw (v2si)
6999 v4hi __builtin_arm_wunpckelub (v8qi)
7000 v2si __builtin_arm_wunpckeluh (v4hi)
7001 long long __builtin_arm_wunpckeluw (v2si)
7002 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
7003 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
7004 v2si __builtin_arm_wunpckihw (v2si, v2si)
7005 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
7006 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
7007 v2si __builtin_arm_wunpckilw (v2si, v2si)
7008 long long __builtin_arm_wxor (long long, long long)
7009 long long __builtin_arm_wzero ()
7012 @node ARM NEON Intrinsics
7013 @subsection ARM NEON Intrinsics
7015 These built-in intrinsics for the ARM Advanced SIMD extension are available
7016 when the @option{-mfpu=neon} switch is used:
7018 @include arm-neon-intrinsics.texi
7020 @node Blackfin Built-in Functions
7021 @subsection Blackfin Built-in Functions
7023 Currently, there are two Blackfin-specific built-in functions. These are
7024 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
7025 using inline assembly; by using these built-in functions the compiler can
7026 automatically add workarounds for hardware errata involving these
7027 instructions. These functions are named as follows:
7030 void __builtin_bfin_csync (void)
7031 void __builtin_bfin_ssync (void)
7034 @node FR-V Built-in Functions
7035 @subsection FR-V Built-in Functions
7037 GCC provides many FR-V-specific built-in functions. In general,
7038 these functions are intended to be compatible with those described
7039 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
7040 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
7041 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
7042 pointer rather than by value.
7044 Most of the functions are named after specific FR-V instructions.
7045 Such functions are said to be ``directly mapped'' and are summarized
7046 here in tabular form.
7050 * Directly-mapped Integer Functions::
7051 * Directly-mapped Media Functions::
7052 * Raw read/write Functions::
7053 * Other Built-in Functions::
7056 @node Argument Types
7057 @subsubsection Argument Types
7059 The arguments to the built-in functions can be divided into three groups:
7060 register numbers, compile-time constants and run-time values. In order
7061 to make this classification clear at a glance, the arguments and return
7062 values are given the following pseudo types:
7064 @multitable @columnfractions .20 .30 .15 .35
7065 @item Pseudo type @tab Real C type @tab Constant? @tab Description
7066 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
7067 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
7068 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
7069 @item @code{uw2} @tab @code{unsigned long long} @tab No
7070 @tab an unsigned doubleword
7071 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
7072 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
7073 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
7074 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
7077 These pseudo types are not defined by GCC, they are simply a notational
7078 convenience used in this manual.
7080 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
7081 and @code{sw2} are evaluated at run time. They correspond to
7082 register operands in the underlying FR-V instructions.
7084 @code{const} arguments represent immediate operands in the underlying
7085 FR-V instructions. They must be compile-time constants.
7087 @code{acc} arguments are evaluated at compile time and specify the number
7088 of an accumulator register. For example, an @code{acc} argument of 2
7089 will select the ACC2 register.
7091 @code{iacc} arguments are similar to @code{acc} arguments but specify the
7092 number of an IACC register. See @pxref{Other Built-in Functions}
7095 @node Directly-mapped Integer Functions
7096 @subsubsection Directly-mapped Integer Functions
7098 The functions listed below map directly to FR-V I-type instructions.
7100 @multitable @columnfractions .45 .32 .23
7101 @item Function prototype @tab Example usage @tab Assembly output
7102 @item @code{sw1 __ADDSS (sw1, sw1)}
7103 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
7104 @tab @code{ADDSS @var{a},@var{b},@var{c}}
7105 @item @code{sw1 __SCAN (sw1, sw1)}
7106 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
7107 @tab @code{SCAN @var{a},@var{b},@var{c}}
7108 @item @code{sw1 __SCUTSS (sw1)}
7109 @tab @code{@var{b} = __SCUTSS (@var{a})}
7110 @tab @code{SCUTSS @var{a},@var{b}}
7111 @item @code{sw1 __SLASS (sw1, sw1)}
7112 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
7113 @tab @code{SLASS @var{a},@var{b},@var{c}}
7114 @item @code{void __SMASS (sw1, sw1)}
7115 @tab @code{__SMASS (@var{a}, @var{b})}
7116 @tab @code{SMASS @var{a},@var{b}}
7117 @item @code{void __SMSSS (sw1, sw1)}
7118 @tab @code{__SMSSS (@var{a}, @var{b})}
7119 @tab @code{SMSSS @var{a},@var{b}}
7120 @item @code{void __SMU (sw1, sw1)}
7121 @tab @code{__SMU (@var{a}, @var{b})}
7122 @tab @code{SMU @var{a},@var{b}}
7123 @item @code{sw2 __SMUL (sw1, sw1)}
7124 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
7125 @tab @code{SMUL @var{a},@var{b},@var{c}}
7126 @item @code{sw1 __SUBSS (sw1, sw1)}
7127 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
7128 @tab @code{SUBSS @var{a},@var{b},@var{c}}
7129 @item @code{uw2 __UMUL (uw1, uw1)}
7130 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
7131 @tab @code{UMUL @var{a},@var{b},@var{c}}
7134 @node Directly-mapped Media Functions
7135 @subsubsection Directly-mapped Media Functions
7137 The functions listed below map directly to FR-V M-type instructions.
7139 @multitable @columnfractions .45 .32 .23
7140 @item Function prototype @tab Example usage @tab Assembly output
7141 @item @code{uw1 __MABSHS (sw1)}
7142 @tab @code{@var{b} = __MABSHS (@var{a})}
7143 @tab @code{MABSHS @var{a},@var{b}}
7144 @item @code{void __MADDACCS (acc, acc)}
7145 @tab @code{__MADDACCS (@var{b}, @var{a})}
7146 @tab @code{MADDACCS @var{a},@var{b}}
7147 @item @code{sw1 __MADDHSS (sw1, sw1)}
7148 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
7149 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
7150 @item @code{uw1 __MADDHUS (uw1, uw1)}
7151 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
7152 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
7153 @item @code{uw1 __MAND (uw1, uw1)}
7154 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
7155 @tab @code{MAND @var{a},@var{b},@var{c}}
7156 @item @code{void __MASACCS (acc, acc)}
7157 @tab @code{__MASACCS (@var{b}, @var{a})}
7158 @tab @code{MASACCS @var{a},@var{b}}
7159 @item @code{uw1 __MAVEH (uw1, uw1)}
7160 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
7161 @tab @code{MAVEH @var{a},@var{b},@var{c}}
7162 @item @code{uw2 __MBTOH (uw1)}
7163 @tab @code{@var{b} = __MBTOH (@var{a})}
7164 @tab @code{MBTOH @var{a},@var{b}}
7165 @item @code{void __MBTOHE (uw1 *, uw1)}
7166 @tab @code{__MBTOHE (&@var{b}, @var{a})}
7167 @tab @code{MBTOHE @var{a},@var{b}}
7168 @item @code{void __MCLRACC (acc)}
7169 @tab @code{__MCLRACC (@var{a})}
7170 @tab @code{MCLRACC @var{a}}
7171 @item @code{void __MCLRACCA (void)}
7172 @tab @code{__MCLRACCA ()}
7173 @tab @code{MCLRACCA}
7174 @item @code{uw1 __Mcop1 (uw1, uw1)}
7175 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
7176 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
7177 @item @code{uw1 __Mcop2 (uw1, uw1)}
7178 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
7179 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
7180 @item @code{uw1 __MCPLHI (uw2, const)}
7181 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
7182 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
7183 @item @code{uw1 __MCPLI (uw2, const)}
7184 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
7185 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
7186 @item @code{void __MCPXIS (acc, sw1, sw1)}
7187 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
7188 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
7189 @item @code{void __MCPXIU (acc, uw1, uw1)}
7190 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
7191 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
7192 @item @code{void __MCPXRS (acc, sw1, sw1)}
7193 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
7194 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
7195 @item @code{void __MCPXRU (acc, uw1, uw1)}
7196 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
7197 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
7198 @item @code{uw1 __MCUT (acc, uw1)}
7199 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
7200 @tab @code{MCUT @var{a},@var{b},@var{c}}
7201 @item @code{uw1 __MCUTSS (acc, sw1)}
7202 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
7203 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
7204 @item @code{void __MDADDACCS (acc, acc)}
7205 @tab @code{__MDADDACCS (@var{b}, @var{a})}
7206 @tab @code{MDADDACCS @var{a},@var{b}}
7207 @item @code{void __MDASACCS (acc, acc)}
7208 @tab @code{__MDASACCS (@var{b}, @var{a})}
7209 @tab @code{MDASACCS @var{a},@var{b}}
7210 @item @code{uw2 __MDCUTSSI (acc, const)}
7211 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
7212 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
7213 @item @code{uw2 __MDPACKH (uw2, uw2)}
7214 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
7215 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
7216 @item @code{uw2 __MDROTLI (uw2, const)}
7217 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
7218 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
7219 @item @code{void __MDSUBACCS (acc, acc)}
7220 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
7221 @tab @code{MDSUBACCS @var{a},@var{b}}
7222 @item @code{void __MDUNPACKH (uw1 *, uw2)}
7223 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
7224 @tab @code{MDUNPACKH @var{a},@var{b}}
7225 @item @code{uw2 __MEXPDHD (uw1, const)}
7226 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
7227 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
7228 @item @code{uw1 __MEXPDHW (uw1, const)}
7229 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
7230 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
7231 @item @code{uw1 __MHDSETH (uw1, const)}
7232 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
7233 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
7234 @item @code{sw1 __MHDSETS (const)}
7235 @tab @code{@var{b} = __MHDSETS (@var{a})}
7236 @tab @code{MHDSETS #@var{a},@var{b}}
7237 @item @code{uw1 __MHSETHIH (uw1, const)}
7238 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
7239 @tab @code{MHSETHIH #@var{a},@var{b}}
7240 @item @code{sw1 __MHSETHIS (sw1, const)}
7241 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
7242 @tab @code{MHSETHIS #@var{a},@var{b}}
7243 @item @code{uw1 __MHSETLOH (uw1, const)}
7244 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
7245 @tab @code{MHSETLOH #@var{a},@var{b}}
7246 @item @code{sw1 __MHSETLOS (sw1, const)}
7247 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
7248 @tab @code{MHSETLOS #@var{a},@var{b}}
7249 @item @code{uw1 __MHTOB (uw2)}
7250 @tab @code{@var{b} = __MHTOB (@var{a})}
7251 @tab @code{MHTOB @var{a},@var{b}}
7252 @item @code{void __MMACHS (acc, sw1, sw1)}
7253 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
7254 @tab @code{MMACHS @var{a},@var{b},@var{c}}
7255 @item @code{void __MMACHU (acc, uw1, uw1)}
7256 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
7257 @tab @code{MMACHU @var{a},@var{b},@var{c}}
7258 @item @code{void __MMRDHS (acc, sw1, sw1)}
7259 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
7260 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
7261 @item @code{void __MMRDHU (acc, uw1, uw1)}
7262 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
7263 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
7264 @item @code{void __MMULHS (acc, sw1, sw1)}
7265 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
7266 @tab @code{MMULHS @var{a},@var{b},@var{c}}
7267 @item @code{void __MMULHU (acc, uw1, uw1)}
7268 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
7269 @tab @code{MMULHU @var{a},@var{b},@var{c}}
7270 @item @code{void __MMULXHS (acc, sw1, sw1)}
7271 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
7272 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
7273 @item @code{void __MMULXHU (acc, uw1, uw1)}
7274 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
7275 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
7276 @item @code{uw1 __MNOT (uw1)}
7277 @tab @code{@var{b} = __MNOT (@var{a})}
7278 @tab @code{MNOT @var{a},@var{b}}
7279 @item @code{uw1 __MOR (uw1, uw1)}
7280 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
7281 @tab @code{MOR @var{a},@var{b},@var{c}}
7282 @item @code{uw1 __MPACKH (uh, uh)}
7283 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
7284 @tab @code{MPACKH @var{a},@var{b},@var{c}}
7285 @item @code{sw2 __MQADDHSS (sw2, sw2)}
7286 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
7287 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
7288 @item @code{uw2 __MQADDHUS (uw2, uw2)}
7289 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
7290 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
7291 @item @code{void __MQCPXIS (acc, sw2, sw2)}
7292 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
7293 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
7294 @item @code{void __MQCPXIU (acc, uw2, uw2)}
7295 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
7296 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
7297 @item @code{void __MQCPXRS (acc, sw2, sw2)}
7298 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
7299 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
7300 @item @code{void __MQCPXRU (acc, uw2, uw2)}
7301 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
7302 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
7303 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
7304 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
7305 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
7306 @item @code{sw2 __MQLMTHS (sw2, sw2)}
7307 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
7308 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
7309 @item @code{void __MQMACHS (acc, sw2, sw2)}
7310 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
7311 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
7312 @item @code{void __MQMACHU (acc, uw2, uw2)}
7313 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
7314 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
7315 @item @code{void __MQMACXHS (acc, sw2, sw2)}
7316 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
7317 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
7318 @item @code{void __MQMULHS (acc, sw2, sw2)}
7319 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
7320 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
7321 @item @code{void __MQMULHU (acc, uw2, uw2)}
7322 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
7323 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
7324 @item @code{void __MQMULXHS (acc, sw2, sw2)}
7325 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
7326 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
7327 @item @code{void __MQMULXHU (acc, uw2, uw2)}
7328 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
7329 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
7330 @item @code{sw2 __MQSATHS (sw2, sw2)}
7331 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
7332 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
7333 @item @code{uw2 __MQSLLHI (uw2, int)}
7334 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
7335 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
7336 @item @code{sw2 __MQSRAHI (sw2, int)}
7337 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
7338 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
7339 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
7340 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
7341 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
7342 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
7343 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
7344 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
7345 @item @code{void __MQXMACHS (acc, sw2, sw2)}
7346 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
7347 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
7348 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
7349 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
7350 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
7351 @item @code{uw1 __MRDACC (acc)}
7352 @tab @code{@var{b} = __MRDACC (@var{a})}
7353 @tab @code{MRDACC @var{a},@var{b}}
7354 @item @code{uw1 __MRDACCG (acc)}
7355 @tab @code{@var{b} = __MRDACCG (@var{a})}
7356 @tab @code{MRDACCG @var{a},@var{b}}
7357 @item @code{uw1 __MROTLI (uw1, const)}
7358 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
7359 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
7360 @item @code{uw1 __MROTRI (uw1, const)}
7361 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
7362 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
7363 @item @code{sw1 __MSATHS (sw1, sw1)}
7364 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
7365 @tab @code{MSATHS @var{a},@var{b},@var{c}}
7366 @item @code{uw1 __MSATHU (uw1, uw1)}
7367 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
7368 @tab @code{MSATHU @var{a},@var{b},@var{c}}
7369 @item @code{uw1 __MSLLHI (uw1, const)}
7370 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
7371 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
7372 @item @code{sw1 __MSRAHI (sw1, const)}
7373 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
7374 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
7375 @item @code{uw1 __MSRLHI (uw1, const)}
7376 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
7377 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
7378 @item @code{void __MSUBACCS (acc, acc)}
7379 @tab @code{__MSUBACCS (@var{b}, @var{a})}
7380 @tab @code{MSUBACCS @var{a},@var{b}}
7381 @item @code{sw1 __MSUBHSS (sw1, sw1)}
7382 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
7383 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
7384 @item @code{uw1 __MSUBHUS (uw1, uw1)}
7385 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
7386 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
7387 @item @code{void __MTRAP (void)}
7388 @tab @code{__MTRAP ()}
7390 @item @code{uw2 __MUNPACKH (uw1)}
7391 @tab @code{@var{b} = __MUNPACKH (@var{a})}
7392 @tab @code{MUNPACKH @var{a},@var{b}}
7393 @item @code{uw1 __MWCUT (uw2, uw1)}
7394 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
7395 @tab @code{MWCUT @var{a},@var{b},@var{c}}
7396 @item @code{void __MWTACC (acc, uw1)}
7397 @tab @code{__MWTACC (@var{b}, @var{a})}
7398 @tab @code{MWTACC @var{a},@var{b}}
7399 @item @code{void __MWTACCG (acc, uw1)}
7400 @tab @code{__MWTACCG (@var{b}, @var{a})}
7401 @tab @code{MWTACCG @var{a},@var{b}}
7402 @item @code{uw1 __MXOR (uw1, uw1)}
7403 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
7404 @tab @code{MXOR @var{a},@var{b},@var{c}}
7407 @node Raw read/write Functions
7408 @subsubsection Raw read/write Functions
7410 This sections describes built-in functions related to read and write
7411 instructions to access memory. These functions generate
7412 @code{membar} instructions to flush the I/O load and stores where
7413 appropriate, as described in Fujitsu's manual described above.
7417 @item unsigned char __builtin_read8 (void *@var{data})
7418 @item unsigned short __builtin_read16 (void *@var{data})
7419 @item unsigned long __builtin_read32 (void *@var{data})
7420 @item unsigned long long __builtin_read64 (void *@var{data})
7422 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
7423 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
7424 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
7425 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
7428 @node Other Built-in Functions
7429 @subsubsection Other Built-in Functions
7431 This section describes built-in functions that are not named after
7432 a specific FR-V instruction.
7435 @item sw2 __IACCreadll (iacc @var{reg})
7436 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
7437 for future expansion and must be 0.
7439 @item sw1 __IACCreadl (iacc @var{reg})
7440 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
7441 Other values of @var{reg} are rejected as invalid.
7443 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
7444 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
7445 is reserved for future expansion and must be 0.
7447 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
7448 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
7449 is 1. Other values of @var{reg} are rejected as invalid.
7451 @item void __data_prefetch0 (const void *@var{x})
7452 Use the @code{dcpl} instruction to load the contents of address @var{x}
7453 into the data cache.
7455 @item void __data_prefetch (const void *@var{x})
7456 Use the @code{nldub} instruction to load the contents of address @var{x}
7457 into the data cache. The instruction will be issued in slot I1@.
7460 @node X86 Built-in Functions
7461 @subsection X86 Built-in Functions
7463 These built-in functions are available for the i386 and x86-64 family
7464 of computers, depending on the command-line switches used.
7466 Note that, if you specify command-line switches such as @option{-msse},
7467 the compiler could use the extended instruction sets even if the built-ins
7468 are not used explicitly in the program. For this reason, applications
7469 which perform runtime CPU detection must compile separate files for each
7470 supported architecture, using the appropriate flags. In particular,
7471 the file containing the CPU detection code should be compiled without
7474 The following machine modes are available for use with MMX built-in functions
7475 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
7476 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
7477 vector of eight 8-bit integers. Some of the built-in functions operate on
7478 MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode.
7480 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
7481 of two 32-bit floating point values.
7483 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
7484 floating point values. Some instructions use a vector of four 32-bit
7485 integers, these use @code{V4SI}. Finally, some instructions operate on an
7486 entire vector register, interpreting it as a 128-bit integer, these use mode
7489 In 64-bit mode, the x86-64 family of processors uses additional built-in
7490 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
7491 floating point and @code{TC} 128-bit complex floating point values.
7493 The following floating point built-in functions are available in 64-bit
7494 mode. All of them implement the function that is part of the name.
7497 __float128 __builtin_fabsq (__float128)
7498 __float128 __builtin_copysignq (__float128, __float128)
7501 The following floating point built-in functions are made available in the
7505 @item __float128 __builtin_infq (void)
7506 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
7509 The following built-in functions are made available by @option{-mmmx}.
7510 All of them generate the machine instruction that is part of the name.
7513 v8qi __builtin_ia32_paddb (v8qi, v8qi)
7514 v4hi __builtin_ia32_paddw (v4hi, v4hi)
7515 v2si __builtin_ia32_paddd (v2si, v2si)
7516 v8qi __builtin_ia32_psubb (v8qi, v8qi)
7517 v4hi __builtin_ia32_psubw (v4hi, v4hi)
7518 v2si __builtin_ia32_psubd (v2si, v2si)
7519 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
7520 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
7521 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
7522 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
7523 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
7524 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
7525 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
7526 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
7527 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
7528 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
7529 di __builtin_ia32_pand (di, di)
7530 di __builtin_ia32_pandn (di,di)
7531 di __builtin_ia32_por (di, di)
7532 di __builtin_ia32_pxor (di, di)
7533 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7534 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7535 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7536 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7537 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7538 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7539 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7540 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7541 v2si __builtin_ia32_punpckhdq (v2si, v2si)
7542 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7543 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7544 v2si __builtin_ia32_punpckldq (v2si, v2si)
7545 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7546 v4hi __builtin_ia32_packssdw (v2si, v2si)
7547 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7549 v4hi __builtin_ia32_psllw (v4hi, v4hi)
7550 v2si __builtin_ia32_pslld (v2si, v2si)
7551 v1di __builtin_ia32_psllq (v1di, v1di)
7552 v4hi __builtin_ia32_psrlw (v4hi, v4hi)
7553 v2si __builtin_ia32_psrld (v2si, v2si)
7554 v1di __builtin_ia32_psrlq (v1di, v1di)
7555 v4hi __builtin_ia32_psraw (v4hi, v4hi)
7556 v2si __builtin_ia32_psrad (v2si, v2si)
7557 v4hi __builtin_ia32_psllwi (v4hi, int)
7558 v2si __builtin_ia32_pslldi (v2si, int)
7559 v1di __builtin_ia32_psllqi (v1di, int)
7560 v4hi __builtin_ia32_psrlwi (v4hi, int)
7561 v2si __builtin_ia32_psrldi (v2si, int)
7562 v1di __builtin_ia32_psrlqi (v1di, int)
7563 v4hi __builtin_ia32_psrawi (v4hi, int)
7564 v2si __builtin_ia32_psradi (v2si, int)
7568 The following built-in functions are made available either with
7569 @option{-msse}, or with a combination of @option{-m3dnow} and
7570 @option{-march=athlon}. All of them generate the machine
7571 instruction that is part of the name.
7574 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7575 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7576 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7577 v1di __builtin_ia32_psadbw (v8qi, v8qi)
7578 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7579 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7580 v8qi __builtin_ia32_pminub (v8qi, v8qi)
7581 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7582 int __builtin_ia32_pextrw (v4hi, int)
7583 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7584 int __builtin_ia32_pmovmskb (v8qi)
7585 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7586 void __builtin_ia32_movntq (di *, di)
7587 void __builtin_ia32_sfence (void)
7590 The following built-in functions are available when @option{-msse} is used.
7591 All of them generate the machine instruction that is part of the name.
7594 int __builtin_ia32_comieq (v4sf, v4sf)
7595 int __builtin_ia32_comineq (v4sf, v4sf)
7596 int __builtin_ia32_comilt (v4sf, v4sf)
7597 int __builtin_ia32_comile (v4sf, v4sf)
7598 int __builtin_ia32_comigt (v4sf, v4sf)
7599 int __builtin_ia32_comige (v4sf, v4sf)
7600 int __builtin_ia32_ucomieq (v4sf, v4sf)
7601 int __builtin_ia32_ucomineq (v4sf, v4sf)
7602 int __builtin_ia32_ucomilt (v4sf, v4sf)
7603 int __builtin_ia32_ucomile (v4sf, v4sf)
7604 int __builtin_ia32_ucomigt (v4sf, v4sf)
7605 int __builtin_ia32_ucomige (v4sf, v4sf)
7606 v4sf __builtin_ia32_addps (v4sf, v4sf)
7607 v4sf __builtin_ia32_subps (v4sf, v4sf)
7608 v4sf __builtin_ia32_mulps (v4sf, v4sf)
7609 v4sf __builtin_ia32_divps (v4sf, v4sf)
7610 v4sf __builtin_ia32_addss (v4sf, v4sf)
7611 v4sf __builtin_ia32_subss (v4sf, v4sf)
7612 v4sf __builtin_ia32_mulss (v4sf, v4sf)
7613 v4sf __builtin_ia32_divss (v4sf, v4sf)
7614 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
7615 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
7616 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
7617 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
7618 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
7619 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
7620 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
7621 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
7622 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
7623 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
7624 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
7625 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
7626 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
7627 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
7628 v4si __builtin_ia32_cmpless (v4sf, v4sf)
7629 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
7630 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
7631 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
7632 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
7633 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
7634 v4sf __builtin_ia32_maxps (v4sf, v4sf)
7635 v4sf __builtin_ia32_maxss (v4sf, v4sf)
7636 v4sf __builtin_ia32_minps (v4sf, v4sf)
7637 v4sf __builtin_ia32_minss (v4sf, v4sf)
7638 v4sf __builtin_ia32_andps (v4sf, v4sf)
7639 v4sf __builtin_ia32_andnps (v4sf, v4sf)
7640 v4sf __builtin_ia32_orps (v4sf, v4sf)
7641 v4sf __builtin_ia32_xorps (v4sf, v4sf)
7642 v4sf __builtin_ia32_movss (v4sf, v4sf)
7643 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
7644 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
7645 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
7646 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
7647 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
7648 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
7649 v2si __builtin_ia32_cvtps2pi (v4sf)
7650 int __builtin_ia32_cvtss2si (v4sf)
7651 v2si __builtin_ia32_cvttps2pi (v4sf)
7652 int __builtin_ia32_cvttss2si (v4sf)
7653 v4sf __builtin_ia32_rcpps (v4sf)
7654 v4sf __builtin_ia32_rsqrtps (v4sf)
7655 v4sf __builtin_ia32_sqrtps (v4sf)
7656 v4sf __builtin_ia32_rcpss (v4sf)
7657 v4sf __builtin_ia32_rsqrtss (v4sf)
7658 v4sf __builtin_ia32_sqrtss (v4sf)
7659 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
7660 void __builtin_ia32_movntps (float *, v4sf)
7661 int __builtin_ia32_movmskps (v4sf)
7664 The following built-in functions are available when @option{-msse} is used.
7667 @item v4sf __builtin_ia32_loadaps (float *)
7668 Generates the @code{movaps} machine instruction as a load from memory.
7669 @item void __builtin_ia32_storeaps (float *, v4sf)
7670 Generates the @code{movaps} machine instruction as a store to memory.
7671 @item v4sf __builtin_ia32_loadups (float *)
7672 Generates the @code{movups} machine instruction as a load from memory.
7673 @item void __builtin_ia32_storeups (float *, v4sf)
7674 Generates the @code{movups} machine instruction as a store to memory.
7675 @item v4sf __builtin_ia32_loadsss (float *)
7676 Generates the @code{movss} machine instruction as a load from memory.
7677 @item void __builtin_ia32_storess (float *, v4sf)
7678 Generates the @code{movss} machine instruction as a store to memory.
7679 @item v4sf __builtin_ia32_loadhps (v4sf, const v2sf *)
7680 Generates the @code{movhps} machine instruction as a load from memory.
7681 @item v4sf __builtin_ia32_loadlps (v4sf, const v2sf *)
7682 Generates the @code{movlps} machine instruction as a load from memory
7683 @item void __builtin_ia32_storehps (v2sf *, v4sf)
7684 Generates the @code{movhps} machine instruction as a store to memory.
7685 @item void __builtin_ia32_storelps (v2sf *, v4sf)
7686 Generates the @code{movlps} machine instruction as a store to memory.
7689 The following built-in functions are available when @option{-msse2} is used.
7690 All of them generate the machine instruction that is part of the name.
7693 int __builtin_ia32_comisdeq (v2df, v2df)
7694 int __builtin_ia32_comisdlt (v2df, v2df)
7695 int __builtin_ia32_comisdle (v2df, v2df)
7696 int __builtin_ia32_comisdgt (v2df, v2df)
7697 int __builtin_ia32_comisdge (v2df, v2df)
7698 int __builtin_ia32_comisdneq (v2df, v2df)
7699 int __builtin_ia32_ucomisdeq (v2df, v2df)
7700 int __builtin_ia32_ucomisdlt (v2df, v2df)
7701 int __builtin_ia32_ucomisdle (v2df, v2df)
7702 int __builtin_ia32_ucomisdgt (v2df, v2df)
7703 int __builtin_ia32_ucomisdge (v2df, v2df)
7704 int __builtin_ia32_ucomisdneq (v2df, v2df)
7705 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
7706 v2df __builtin_ia32_cmpltpd (v2df, v2df)
7707 v2df __builtin_ia32_cmplepd (v2df, v2df)
7708 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
7709 v2df __builtin_ia32_cmpgepd (v2df, v2df)
7710 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
7711 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
7712 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
7713 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
7714 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
7715 v2df __builtin_ia32_cmpngepd (v2df, v2df)
7716 v2df __builtin_ia32_cmpordpd (v2df, v2df)
7717 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
7718 v2df __builtin_ia32_cmpltsd (v2df, v2df)
7719 v2df __builtin_ia32_cmplesd (v2df, v2df)
7720 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
7721 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
7722 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
7723 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
7724 v2df __builtin_ia32_cmpordsd (v2df, v2df)
7725 v2di __builtin_ia32_paddq (v2di, v2di)
7726 v2di __builtin_ia32_psubq (v2di, v2di)
7727 v2df __builtin_ia32_addpd (v2df, v2df)
7728 v2df __builtin_ia32_subpd (v2df, v2df)
7729 v2df __builtin_ia32_mulpd (v2df, v2df)
7730 v2df __builtin_ia32_divpd (v2df, v2df)
7731 v2df __builtin_ia32_addsd (v2df, v2df)
7732 v2df __builtin_ia32_subsd (v2df, v2df)
7733 v2df __builtin_ia32_mulsd (v2df, v2df)
7734 v2df __builtin_ia32_divsd (v2df, v2df)
7735 v2df __builtin_ia32_minpd (v2df, v2df)
7736 v2df __builtin_ia32_maxpd (v2df, v2df)
7737 v2df __builtin_ia32_minsd (v2df, v2df)
7738 v2df __builtin_ia32_maxsd (v2df, v2df)
7739 v2df __builtin_ia32_andpd (v2df, v2df)
7740 v2df __builtin_ia32_andnpd (v2df, v2df)
7741 v2df __builtin_ia32_orpd (v2df, v2df)
7742 v2df __builtin_ia32_xorpd (v2df, v2df)
7743 v2df __builtin_ia32_movsd (v2df, v2df)
7744 v2df __builtin_ia32_unpckhpd (v2df, v2df)
7745 v2df __builtin_ia32_unpcklpd (v2df, v2df)
7746 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
7747 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
7748 v4si __builtin_ia32_paddd128 (v4si, v4si)
7749 v2di __builtin_ia32_paddq128 (v2di, v2di)
7750 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
7751 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
7752 v4si __builtin_ia32_psubd128 (v4si, v4si)
7753 v2di __builtin_ia32_psubq128 (v2di, v2di)
7754 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
7755 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
7756 v2di __builtin_ia32_pand128 (v2di, v2di)
7757 v2di __builtin_ia32_pandn128 (v2di, v2di)
7758 v2di __builtin_ia32_por128 (v2di, v2di)
7759 v2di __builtin_ia32_pxor128 (v2di, v2di)
7760 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
7761 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
7762 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
7763 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
7764 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
7765 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
7766 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
7767 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
7768 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
7769 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
7770 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
7771 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
7772 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
7773 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
7774 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
7775 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
7776 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
7777 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
7778 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
7779 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
7780 v16qi __builtin_ia32_packsswb128 (v8hi, v8hi)
7781 v8hi __builtin_ia32_packssdw128 (v4si, v4si)
7782 v16qi __builtin_ia32_packuswb128 (v8hi, v8hi)
7783 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
7784 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
7785 v2df __builtin_ia32_loadupd (double *)
7786 void __builtin_ia32_storeupd (double *, v2df)
7787 v2df __builtin_ia32_loadhpd (v2df, double const *)
7788 v2df __builtin_ia32_loadlpd (v2df, double const *)
7789 int __builtin_ia32_movmskpd (v2df)
7790 int __builtin_ia32_pmovmskb128 (v16qi)
7791 void __builtin_ia32_movnti (int *, int)
7792 void __builtin_ia32_movntpd (double *, v2df)
7793 void __builtin_ia32_movntdq (v2df *, v2df)
7794 v4si __builtin_ia32_pshufd (v4si, int)
7795 v8hi __builtin_ia32_pshuflw (v8hi, int)
7796 v8hi __builtin_ia32_pshufhw (v8hi, int)
7797 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
7798 v2df __builtin_ia32_sqrtpd (v2df)
7799 v2df __builtin_ia32_sqrtsd (v2df)
7800 v2df __builtin_ia32_shufpd (v2df, v2df, int)
7801 v2df __builtin_ia32_cvtdq2pd (v4si)
7802 v4sf __builtin_ia32_cvtdq2ps (v4si)
7803 v4si __builtin_ia32_cvtpd2dq (v2df)
7804 v2si __builtin_ia32_cvtpd2pi (v2df)
7805 v4sf __builtin_ia32_cvtpd2ps (v2df)
7806 v4si __builtin_ia32_cvttpd2dq (v2df)
7807 v2si __builtin_ia32_cvttpd2pi (v2df)
7808 v2df __builtin_ia32_cvtpi2pd (v2si)
7809 int __builtin_ia32_cvtsd2si (v2df)
7810 int __builtin_ia32_cvttsd2si (v2df)
7811 long long __builtin_ia32_cvtsd2si64 (v2df)
7812 long long __builtin_ia32_cvttsd2si64 (v2df)
7813 v4si __builtin_ia32_cvtps2dq (v4sf)
7814 v2df __builtin_ia32_cvtps2pd (v4sf)
7815 v4si __builtin_ia32_cvttps2dq (v4sf)
7816 v2df __builtin_ia32_cvtsi2sd (v2df, int)
7817 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
7818 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
7819 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
7820 void __builtin_ia32_clflush (const void *)
7821 void __builtin_ia32_lfence (void)
7822 void __builtin_ia32_mfence (void)
7823 v16qi __builtin_ia32_loaddqu (const char *)
7824 void __builtin_ia32_storedqu (char *, v16qi)
7825 v1di __builtin_ia32_pmuludq (v2si, v2si)
7826 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
7827 v8hi __builtin_ia32_psllw128 (v8hi, v8hi)
7828 v4si __builtin_ia32_pslld128 (v4si, v4si)
7829 v2di __builtin_ia32_psllq128 (v2di, v2di)
7830 v8hi __builtin_ia32_psrlw128 (v8hi, v8hi)
7831 v4si __builtin_ia32_psrld128 (v4si, v4si)
7832 v2di __builtin_ia32_psrlq128 (v2di, v2di)
7833 v8hi __builtin_ia32_psraw128 (v8hi, v8hi)
7834 v4si __builtin_ia32_psrad128 (v4si, v4si)
7835 v2di __builtin_ia32_pslldqi128 (v2di, int)
7836 v8hi __builtin_ia32_psllwi128 (v8hi, int)
7837 v4si __builtin_ia32_pslldi128 (v4si, int)
7838 v2di __builtin_ia32_psllqi128 (v2di, int)
7839 v2di __builtin_ia32_psrldqi128 (v2di, int)
7840 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
7841 v4si __builtin_ia32_psrldi128 (v4si, int)
7842 v2di __builtin_ia32_psrlqi128 (v2di, int)
7843 v8hi __builtin_ia32_psrawi128 (v8hi, int)
7844 v4si __builtin_ia32_psradi128 (v4si, int)
7845 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
7848 The following built-in functions are available when @option{-msse3} is used.
7849 All of them generate the machine instruction that is part of the name.
7852 v2df __builtin_ia32_addsubpd (v2df, v2df)
7853 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
7854 v2df __builtin_ia32_haddpd (v2df, v2df)
7855 v4sf __builtin_ia32_haddps (v4sf, v4sf)
7856 v2df __builtin_ia32_hsubpd (v2df, v2df)
7857 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
7858 v16qi __builtin_ia32_lddqu (char const *)
7859 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
7860 v2df __builtin_ia32_movddup (v2df)
7861 v4sf __builtin_ia32_movshdup (v4sf)
7862 v4sf __builtin_ia32_movsldup (v4sf)
7863 void __builtin_ia32_mwait (unsigned int, unsigned int)
7866 The following built-in functions are available when @option{-msse3} is used.
7869 @item v2df __builtin_ia32_loadddup (double const *)
7870 Generates the @code{movddup} machine instruction as a load from memory.
7873 The following built-in functions are available when @option{-mssse3} is used.
7874 All of them generate the machine instruction that is part of the name
7878 v2si __builtin_ia32_phaddd (v2si, v2si)
7879 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
7880 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
7881 v2si __builtin_ia32_phsubd (v2si, v2si)
7882 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
7883 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
7884 v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi)
7885 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
7886 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
7887 v8qi __builtin_ia32_psignb (v8qi, v8qi)
7888 v2si __builtin_ia32_psignd (v2si, v2si)
7889 v4hi __builtin_ia32_psignw (v4hi, v4hi)
7890 v1di __builtin_ia32_palignr (v1di, v1di, int)
7891 v8qi __builtin_ia32_pabsb (v8qi)
7892 v2si __builtin_ia32_pabsd (v2si)
7893 v4hi __builtin_ia32_pabsw (v4hi)
7896 The following built-in functions are available when @option{-mssse3} is used.
7897 All of them generate the machine instruction that is part of the name
7901 v4si __builtin_ia32_phaddd128 (v4si, v4si)
7902 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
7903 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
7904 v4si __builtin_ia32_phsubd128 (v4si, v4si)
7905 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
7906 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
7907 v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
7908 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
7909 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
7910 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
7911 v4si __builtin_ia32_psignd128 (v4si, v4si)
7912 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
7913 v2di __builtin_ia32_palignr128 (v2di, v2di, int)
7914 v16qi __builtin_ia32_pabsb128 (v16qi)
7915 v4si __builtin_ia32_pabsd128 (v4si)
7916 v8hi __builtin_ia32_pabsw128 (v8hi)
7919 The following built-in functions are available when @option{-msse4.1} is
7920 used. All of them generate the machine instruction that is part of the
7924 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
7925 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
7926 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
7927 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
7928 v2df __builtin_ia32_dppd (v2df, v2df, const int)
7929 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
7930 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
7931 v2di __builtin_ia32_movntdqa (v2di *);
7932 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
7933 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
7934 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
7935 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
7936 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
7937 v8hi __builtin_ia32_phminposuw128 (v8hi)
7938 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
7939 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
7940 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
7941 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
7942 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
7943 v4si __builtin_ia32_pminsd128 (v4si, v4si)
7944 v4si __builtin_ia32_pminud128 (v4si, v4si)
7945 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
7946 v4si __builtin_ia32_pmovsxbd128 (v16qi)
7947 v2di __builtin_ia32_pmovsxbq128 (v16qi)
7948 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
7949 v2di __builtin_ia32_pmovsxdq128 (v4si)
7950 v4si __builtin_ia32_pmovsxwd128 (v8hi)
7951 v2di __builtin_ia32_pmovsxwq128 (v8hi)
7952 v4si __builtin_ia32_pmovzxbd128 (v16qi)
7953 v2di __builtin_ia32_pmovzxbq128 (v16qi)
7954 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
7955 v2di __builtin_ia32_pmovzxdq128 (v4si)
7956 v4si __builtin_ia32_pmovzxwd128 (v8hi)
7957 v2di __builtin_ia32_pmovzxwq128 (v8hi)
7958 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
7959 v4si __builtin_ia32_pmulld128 (v4si, v4si)
7960 int __builtin_ia32_ptestc128 (v2di, v2di)
7961 int __builtin_ia32_ptestnzc128 (v2di, v2di)
7962 int __builtin_ia32_ptestz128 (v2di, v2di)
7963 v2df __builtin_ia32_roundpd (v2df, const int)
7964 v4sf __builtin_ia32_roundps (v4sf, const int)
7965 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
7966 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
7969 The following built-in functions are available when @option{-msse4.1} is
7973 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
7974 Generates the @code{insertps} machine instruction.
7975 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
7976 Generates the @code{pextrb} machine instruction.
7977 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
7978 Generates the @code{pinsrb} machine instruction.
7979 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
7980 Generates the @code{pinsrd} machine instruction.
7981 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
7982 Generates the @code{pinsrq} machine instruction in 64bit mode.
7985 The following built-in functions are changed to generate new SSE4.1
7986 instructions when @option{-msse4.1} is used.
7989 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
7990 Generates the @code{extractps} machine instruction.
7991 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
7992 Generates the @code{pextrd} machine instruction.
7993 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
7994 Generates the @code{pextrq} machine instruction in 64bit mode.
7997 The following built-in functions are available when @option{-msse4.2} is
7998 used. All of them generate the machine instruction that is part of the
8002 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
8003 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
8004 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
8005 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
8006 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
8007 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
8008 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
8009 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
8010 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
8011 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
8012 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
8013 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
8014 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
8015 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
8016 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
8019 The following built-in functions are available when @option{-msse4.2} is
8023 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
8024 Generates the @code{crc32b} machine instruction.
8025 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
8026 Generates the @code{crc32w} machine instruction.
8027 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
8028 Generates the @code{crc32l} machine instruction.
8029 @item unsigned long long __builtin_ia32_crc32di (unsigned long long, unsigned long long)
8032 The following built-in functions are changed to generate new SSE4.2
8033 instructions when @option{-msse4.2} is used.
8036 @item int __builtin_popcount (unsigned int)
8037 Generates the @code{popcntl} machine instruction.
8038 @item int __builtin_popcountl (unsigned long)
8039 Generates the @code{popcntl} or @code{popcntq} machine instruction,
8040 depending on the size of @code{unsigned long}.
8041 @item int __builtin_popcountll (unsigned long long)
8042 Generates the @code{popcntq} machine instruction.
8045 The following built-in functions are available when @option{-maes} is
8046 used. All of them generate the machine instruction that is part of the
8050 v2di __builtin_ia32_aesenc128 (v2di, v2di)
8051 v2di __builtin_ia32_aesenclast128 (v2di, v2di)
8052 v2di __builtin_ia32_aesdec128 (v2di, v2di)
8053 v2di __builtin_ia32_aesdeclast128 (v2di, v2di)
8054 v2di __builtin_ia32_aeskeygenassist128 (v2di, const int)
8055 v2di __builtin_ia32_aesimc128 (v2di)
8058 The following built-in function is available when @option{-mpclmul} is
8062 @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int)
8063 Generates the @code{pclmulqdq} machine instruction.
8066 The following built-in functions are available when @option{-msse4a} is used.
8067 All of them generate the machine instruction that is part of the name.
8070 void __builtin_ia32_movntsd (double *, v2df)
8071 void __builtin_ia32_movntss (float *, v4sf)
8072 v2di __builtin_ia32_extrq (v2di, v16qi)
8073 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
8074 v2di __builtin_ia32_insertq (v2di, v2di)
8075 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
8078 The following built-in functions are available when @option{-msse5} is used.
8079 All of them generate the machine instruction that is part of the name
8083 v2df __builtin_ia32_comeqpd (v2df, v2df)
8084 v2df __builtin_ia32_comeqps (v2df, v2df)
8085 v4sf __builtin_ia32_comeqsd (v4sf, v4sf)
8086 v4sf __builtin_ia32_comeqss (v4sf, v4sf)
8087 v2df __builtin_ia32_comfalsepd (v2df, v2df)
8088 v2df __builtin_ia32_comfalseps (v2df, v2df)
8089 v4sf __builtin_ia32_comfalsesd (v4sf, v4sf)
8090 v4sf __builtin_ia32_comfalsess (v4sf, v4sf)
8091 v2df __builtin_ia32_comgepd (v2df, v2df)
8092 v2df __builtin_ia32_comgeps (v2df, v2df)
8093 v4sf __builtin_ia32_comgesd (v4sf, v4sf)
8094 v4sf __builtin_ia32_comgess (v4sf, v4sf)
8095 v2df __builtin_ia32_comgtpd (v2df, v2df)
8096 v2df __builtin_ia32_comgtps (v2df, v2df)
8097 v4sf __builtin_ia32_comgtsd (v4sf, v4sf)
8098 v4sf __builtin_ia32_comgtss (v4sf, v4sf)
8099 v2df __builtin_ia32_comlepd (v2df, v2df)
8100 v2df __builtin_ia32_comleps (v2df, v2df)
8101 v4sf __builtin_ia32_comlesd (v4sf, v4sf)
8102 v4sf __builtin_ia32_comless (v4sf, v4sf)
8103 v2df __builtin_ia32_comltpd (v2df, v2df)
8104 v2df __builtin_ia32_comltps (v2df, v2df)
8105 v4sf __builtin_ia32_comltsd (v4sf, v4sf)
8106 v4sf __builtin_ia32_comltss (v4sf, v4sf)
8107 v2df __builtin_ia32_comnepd (v2df, v2df)
8108 v2df __builtin_ia32_comneps (v2df, v2df)
8109 v4sf __builtin_ia32_comnesd (v4sf, v4sf)
8110 v4sf __builtin_ia32_comness (v4sf, v4sf)
8111 v2df __builtin_ia32_comordpd (v2df, v2df)
8112 v2df __builtin_ia32_comordps (v2df, v2df)
8113 v4sf __builtin_ia32_comordsd (v4sf, v4sf)
8114 v4sf __builtin_ia32_comordss (v4sf, v4sf)
8115 v2df __builtin_ia32_comtruepd (v2df, v2df)
8116 v2df __builtin_ia32_comtrueps (v2df, v2df)
8117 v4sf __builtin_ia32_comtruesd (v4sf, v4sf)
8118 v4sf __builtin_ia32_comtruess (v4sf, v4sf)
8119 v2df __builtin_ia32_comueqpd (v2df, v2df)
8120 v2df __builtin_ia32_comueqps (v2df, v2df)
8121 v4sf __builtin_ia32_comueqsd (v4sf, v4sf)
8122 v4sf __builtin_ia32_comueqss (v4sf, v4sf)
8123 v2df __builtin_ia32_comugepd (v2df, v2df)
8124 v2df __builtin_ia32_comugeps (v2df, v2df)
8125 v4sf __builtin_ia32_comugesd (v4sf, v4sf)
8126 v4sf __builtin_ia32_comugess (v4sf, v4sf)
8127 v2df __builtin_ia32_comugtpd (v2df, v2df)
8128 v2df __builtin_ia32_comugtps (v2df, v2df)
8129 v4sf __builtin_ia32_comugtsd (v4sf, v4sf)
8130 v4sf __builtin_ia32_comugtss (v4sf, v4sf)
8131 v2df __builtin_ia32_comulepd (v2df, v2df)
8132 v2df __builtin_ia32_comuleps (v2df, v2df)
8133 v4sf __builtin_ia32_comulesd (v4sf, v4sf)
8134 v4sf __builtin_ia32_comuless (v4sf, v4sf)
8135 v2df __builtin_ia32_comultpd (v2df, v2df)
8136 v2df __builtin_ia32_comultps (v2df, v2df)
8137 v4sf __builtin_ia32_comultsd (v4sf, v4sf)
8138 v4sf __builtin_ia32_comultss (v4sf, v4sf)
8139 v2df __builtin_ia32_comunepd (v2df, v2df)
8140 v2df __builtin_ia32_comuneps (v2df, v2df)
8141 v4sf __builtin_ia32_comunesd (v4sf, v4sf)
8142 v4sf __builtin_ia32_comuness (v4sf, v4sf)
8143 v2df __builtin_ia32_comunordpd (v2df, v2df)
8144 v2df __builtin_ia32_comunordps (v2df, v2df)
8145 v4sf __builtin_ia32_comunordsd (v4sf, v4sf)
8146 v4sf __builtin_ia32_comunordss (v4sf, v4sf)
8147 v2df __builtin_ia32_fmaddpd (v2df, v2df, v2df)
8148 v4sf __builtin_ia32_fmaddps (v4sf, v4sf, v4sf)
8149 v2df __builtin_ia32_fmaddsd (v2df, v2df, v2df)
8150 v4sf __builtin_ia32_fmaddss (v4sf, v4sf, v4sf)
8151 v2df __builtin_ia32_fmsubpd (v2df, v2df, v2df)
8152 v4sf __builtin_ia32_fmsubps (v4sf, v4sf, v4sf)
8153 v2df __builtin_ia32_fmsubsd (v2df, v2df, v2df)
8154 v4sf __builtin_ia32_fmsubss (v4sf, v4sf, v4sf)
8155 v2df __builtin_ia32_fnmaddpd (v2df, v2df, v2df)
8156 v4sf __builtin_ia32_fnmaddps (v4sf, v4sf, v4sf)
8157 v2df __builtin_ia32_fnmaddsd (v2df, v2df, v2df)
8158 v4sf __builtin_ia32_fnmaddss (v4sf, v4sf, v4sf)
8159 v2df __builtin_ia32_fnmsubpd (v2df, v2df, v2df)
8160 v4sf __builtin_ia32_fnmsubps (v4sf, v4sf, v4sf)
8161 v2df __builtin_ia32_fnmsubsd (v2df, v2df, v2df)
8162 v4sf __builtin_ia32_fnmsubss (v4sf, v4sf, v4sf)
8163 v2df __builtin_ia32_frczpd (v2df)
8164 v4sf __builtin_ia32_frczps (v4sf)
8165 v2df __builtin_ia32_frczsd (v2df, v2df)
8166 v4sf __builtin_ia32_frczss (v4sf, v4sf)
8167 v2di __builtin_ia32_pcmov (v2di, v2di, v2di)
8168 v2di __builtin_ia32_pcmov_v2di (v2di, v2di, v2di)
8169 v4si __builtin_ia32_pcmov_v4si (v4si, v4si, v4si)
8170 v8hi __builtin_ia32_pcmov_v8hi (v8hi, v8hi, v8hi)
8171 v16qi __builtin_ia32_pcmov_v16qi (v16qi, v16qi, v16qi)
8172 v2df __builtin_ia32_pcmov_v2df (v2df, v2df, v2df)
8173 v4sf __builtin_ia32_pcmov_v4sf (v4sf, v4sf, v4sf)
8174 v16qi __builtin_ia32_pcomeqb (v16qi, v16qi)
8175 v8hi __builtin_ia32_pcomeqw (v8hi, v8hi)
8176 v4si __builtin_ia32_pcomeqd (v4si, v4si)
8177 v2di __builtin_ia32_pcomeqq (v2di, v2di)
8178 v16qi __builtin_ia32_pcomequb (v16qi, v16qi)
8179 v4si __builtin_ia32_pcomequd (v4si, v4si)
8180 v2di __builtin_ia32_pcomequq (v2di, v2di)
8181 v8hi __builtin_ia32_pcomequw (v8hi, v8hi)
8182 v8hi __builtin_ia32_pcomeqw (v8hi, v8hi)
8183 v16qi __builtin_ia32_pcomfalseb (v16qi, v16qi)
8184 v4si __builtin_ia32_pcomfalsed (v4si, v4si)
8185 v2di __builtin_ia32_pcomfalseq (v2di, v2di)
8186 v16qi __builtin_ia32_pcomfalseub (v16qi, v16qi)
8187 v4si __builtin_ia32_pcomfalseud (v4si, v4si)
8188 v2di __builtin_ia32_pcomfalseuq (v2di, v2di)
8189 v8hi __builtin_ia32_pcomfalseuw (v8hi, v8hi)
8190 v8hi __builtin_ia32_pcomfalsew (v8hi, v8hi)
8191 v16qi __builtin_ia32_pcomgeb (v16qi, v16qi)
8192 v4si __builtin_ia32_pcomged (v4si, v4si)
8193 v2di __builtin_ia32_pcomgeq (v2di, v2di)
8194 v16qi __builtin_ia32_pcomgeub (v16qi, v16qi)
8195 v4si __builtin_ia32_pcomgeud (v4si, v4si)
8196 v2di __builtin_ia32_pcomgeuq (v2di, v2di)
8197 v8hi __builtin_ia32_pcomgeuw (v8hi, v8hi)
8198 v8hi __builtin_ia32_pcomgew (v8hi, v8hi)
8199 v16qi __builtin_ia32_pcomgtb (v16qi, v16qi)
8200 v4si __builtin_ia32_pcomgtd (v4si, v4si)
8201 v2di __builtin_ia32_pcomgtq (v2di, v2di)
8202 v16qi __builtin_ia32_pcomgtub (v16qi, v16qi)
8203 v4si __builtin_ia32_pcomgtud (v4si, v4si)
8204 v2di __builtin_ia32_pcomgtuq (v2di, v2di)
8205 v8hi __builtin_ia32_pcomgtuw (v8hi, v8hi)
8206 v8hi __builtin_ia32_pcomgtw (v8hi, v8hi)
8207 v16qi __builtin_ia32_pcomleb (v16qi, v16qi)
8208 v4si __builtin_ia32_pcomled (v4si, v4si)
8209 v2di __builtin_ia32_pcomleq (v2di, v2di)
8210 v16qi __builtin_ia32_pcomleub (v16qi, v16qi)
8211 v4si __builtin_ia32_pcomleud (v4si, v4si)
8212 v2di __builtin_ia32_pcomleuq (v2di, v2di)
8213 v8hi __builtin_ia32_pcomleuw (v8hi, v8hi)
8214 v8hi __builtin_ia32_pcomlew (v8hi, v8hi)
8215 v16qi __builtin_ia32_pcomltb (v16qi, v16qi)
8216 v4si __builtin_ia32_pcomltd (v4si, v4si)
8217 v2di __builtin_ia32_pcomltq (v2di, v2di)
8218 v16qi __builtin_ia32_pcomltub (v16qi, v16qi)
8219 v4si __builtin_ia32_pcomltud (v4si, v4si)
8220 v2di __builtin_ia32_pcomltuq (v2di, v2di)
8221 v8hi __builtin_ia32_pcomltuw (v8hi, v8hi)
8222 v8hi __builtin_ia32_pcomltw (v8hi, v8hi)
8223 v16qi __builtin_ia32_pcomneb (v16qi, v16qi)
8224 v4si __builtin_ia32_pcomned (v4si, v4si)
8225 v2di __builtin_ia32_pcomneq (v2di, v2di)
8226 v16qi __builtin_ia32_pcomneub (v16qi, v16qi)
8227 v4si __builtin_ia32_pcomneud (v4si, v4si)
8228 v2di __builtin_ia32_pcomneuq (v2di, v2di)
8229 v8hi __builtin_ia32_pcomneuw (v8hi, v8hi)
8230 v8hi __builtin_ia32_pcomnew (v8hi, v8hi)
8231 v16qi __builtin_ia32_pcomtrueb (v16qi, v16qi)
8232 v4si __builtin_ia32_pcomtrued (v4si, v4si)
8233 v2di __builtin_ia32_pcomtrueq (v2di, v2di)
8234 v16qi __builtin_ia32_pcomtrueub (v16qi, v16qi)
8235 v4si __builtin_ia32_pcomtrueud (v4si, v4si)
8236 v2di __builtin_ia32_pcomtrueuq (v2di, v2di)
8237 v8hi __builtin_ia32_pcomtrueuw (v8hi, v8hi)
8238 v8hi __builtin_ia32_pcomtruew (v8hi, v8hi)
8239 v4df __builtin_ia32_permpd (v2df, v2df, v16qi)
8240 v4sf __builtin_ia32_permps (v4sf, v4sf, v16qi)
8241 v4si __builtin_ia32_phaddbd (v16qi)
8242 v2di __builtin_ia32_phaddbq (v16qi)
8243 v8hi __builtin_ia32_phaddbw (v16qi)
8244 v2di __builtin_ia32_phadddq (v4si)
8245 v4si __builtin_ia32_phaddubd (v16qi)
8246 v2di __builtin_ia32_phaddubq (v16qi)
8247 v8hi __builtin_ia32_phaddubw (v16qi)
8248 v2di __builtin_ia32_phaddudq (v4si)
8249 v4si __builtin_ia32_phadduwd (v8hi)
8250 v2di __builtin_ia32_phadduwq (v8hi)
8251 v4si __builtin_ia32_phaddwd (v8hi)
8252 v2di __builtin_ia32_phaddwq (v8hi)
8253 v8hi __builtin_ia32_phsubbw (v16qi)
8254 v2di __builtin_ia32_phsubdq (v4si)
8255 v4si __builtin_ia32_phsubwd (v8hi)
8256 v4si __builtin_ia32_pmacsdd (v4si, v4si, v4si)
8257 v2di __builtin_ia32_pmacsdqh (v4si, v4si, v2di)
8258 v2di __builtin_ia32_pmacsdql (v4si, v4si, v2di)
8259 v4si __builtin_ia32_pmacssdd (v4si, v4si, v4si)
8260 v2di __builtin_ia32_pmacssdqh (v4si, v4si, v2di)
8261 v2di __builtin_ia32_pmacssdql (v4si, v4si, v2di)
8262 v4si __builtin_ia32_pmacsswd (v8hi, v8hi, v4si)
8263 v8hi __builtin_ia32_pmacssww (v8hi, v8hi, v8hi)
8264 v4si __builtin_ia32_pmacswd (v8hi, v8hi, v4si)
8265 v8hi __builtin_ia32_pmacsww (v8hi, v8hi, v8hi)
8266 v4si __builtin_ia32_pmadcsswd (v8hi, v8hi, v4si)
8267 v4si __builtin_ia32_pmadcswd (v8hi, v8hi, v4si)
8268 v16qi __builtin_ia32_pperm (v16qi, v16qi, v16qi)
8269 v16qi __builtin_ia32_protb (v16qi, v16qi)
8270 v4si __builtin_ia32_protd (v4si, v4si)
8271 v2di __builtin_ia32_protq (v2di, v2di)
8272 v8hi __builtin_ia32_protw (v8hi, v8hi)
8273 v16qi __builtin_ia32_pshab (v16qi, v16qi)
8274 v4si __builtin_ia32_pshad (v4si, v4si)
8275 v2di __builtin_ia32_pshaq (v2di, v2di)
8276 v8hi __builtin_ia32_pshaw (v8hi, v8hi)
8277 v16qi __builtin_ia32_pshlb (v16qi, v16qi)
8278 v4si __builtin_ia32_pshld (v4si, v4si)
8279 v2di __builtin_ia32_pshlq (v2di, v2di)
8280 v8hi __builtin_ia32_pshlw (v8hi, v8hi)
8283 The following builtin-in functions are available when @option{-msse5}
8284 is used. The second argument must be an integer constant and generate
8285 the machine instruction that is part of the name with the @samp{_imm}
8289 v16qi __builtin_ia32_protb_imm (v16qi, int)
8290 v4si __builtin_ia32_protd_imm (v4si, int)
8291 v2di __builtin_ia32_protq_imm (v2di, int)
8292 v8hi __builtin_ia32_protw_imm (v8hi, int)
8295 The following built-in functions are available when @option{-m3dnow} is used.
8296 All of them generate the machine instruction that is part of the name.
8299 void __builtin_ia32_femms (void)
8300 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
8301 v2si __builtin_ia32_pf2id (v2sf)
8302 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
8303 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
8304 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
8305 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
8306 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
8307 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
8308 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
8309 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
8310 v2sf __builtin_ia32_pfrcp (v2sf)
8311 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
8312 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
8313 v2sf __builtin_ia32_pfrsqrt (v2sf)
8314 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
8315 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
8316 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
8317 v2sf __builtin_ia32_pi2fd (v2si)
8318 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
8321 The following built-in functions are available when both @option{-m3dnow}
8322 and @option{-march=athlon} are used. All of them generate the machine
8323 instruction that is part of the name.
8326 v2si __builtin_ia32_pf2iw (v2sf)
8327 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
8328 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
8329 v2sf __builtin_ia32_pi2fw (v2si)
8330 v2sf __builtin_ia32_pswapdsf (v2sf)
8331 v2si __builtin_ia32_pswapdsi (v2si)
8334 @node MIPS DSP Built-in Functions
8335 @subsection MIPS DSP Built-in Functions
8337 The MIPS DSP Application-Specific Extension (ASE) includes new
8338 instructions that are designed to improve the performance of DSP and
8339 media applications. It provides instructions that operate on packed
8340 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
8342 GCC supports MIPS DSP operations using both the generic
8343 vector extensions (@pxref{Vector Extensions}) and a collection of
8344 MIPS-specific built-in functions. Both kinds of support are
8345 enabled by the @option{-mdsp} command-line option.
8347 Revision 2 of the ASE was introduced in the second half of 2006.
8348 This revision adds extra instructions to the original ASE, but is
8349 otherwise backwards-compatible with it. You can select revision 2
8350 using the command-line option @option{-mdspr2}; this option implies
8353 At present, GCC only provides support for operations on 32-bit
8354 vectors. The vector type associated with 8-bit integer data is
8355 usually called @code{v4i8}, the vector type associated with Q7
8356 is usually called @code{v4q7}, the vector type associated with 16-bit
8357 integer data is usually called @code{v2i16}, and the vector type
8358 associated with Q15 is usually called @code{v2q15}. They can be
8359 defined in C as follows:
8362 typedef signed char v4i8 __attribute__ ((vector_size(4)));
8363 typedef signed char v4q7 __attribute__ ((vector_size(4)));
8364 typedef short v2i16 __attribute__ ((vector_size(4)));
8365 typedef short v2q15 __attribute__ ((vector_size(4)));
8368 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
8369 initialized in the same way as aggregates. For example:
8372 v4i8 a = @{1, 2, 3, 4@};
8374 b = (v4i8) @{5, 6, 7, 8@};
8376 v2q15 c = @{0x0fcb, 0x3a75@};
8378 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
8381 @emph{Note:} The CPU's endianness determines the order in which values
8382 are packed. On little-endian targets, the first value is the least
8383 significant and the last value is the most significant. The opposite
8384 order applies to big-endian targets. For example, the code above will
8385 set the lowest byte of @code{a} to @code{1} on little-endian targets
8386 and @code{4} on big-endian targets.
8388 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
8389 representation. As shown in this example, the integer representation
8390 of a Q7 value can be obtained by multiplying the fractional value by
8391 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
8392 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
8395 The table below lists the @code{v4i8} and @code{v2q15} operations for which
8396 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
8397 and @code{c} and @code{d} are @code{v2q15} values.
8399 @multitable @columnfractions .50 .50
8400 @item C code @tab MIPS instruction
8401 @item @code{a + b} @tab @code{addu.qb}
8402 @item @code{c + d} @tab @code{addq.ph}
8403 @item @code{a - b} @tab @code{subu.qb}
8404 @item @code{c - d} @tab @code{subq.ph}
8407 The table below lists the @code{v2i16} operation for which
8408 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
8409 @code{v2i16} values.
8411 @multitable @columnfractions .50 .50
8412 @item C code @tab MIPS instruction
8413 @item @code{e * f} @tab @code{mul.ph}
8416 It is easier to describe the DSP built-in functions if we first define
8417 the following types:
8422 typedef unsigned int ui32;
8423 typedef long long a64;
8426 @code{q31} and @code{i32} are actually the same as @code{int}, but we
8427 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
8428 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
8429 @code{long long}, but we use @code{a64} to indicate values that will
8430 be placed in one of the four DSP accumulators (@code{$ac0},
8431 @code{$ac1}, @code{$ac2} or @code{$ac3}).
8433 Also, some built-in functions prefer or require immediate numbers as
8434 parameters, because the corresponding DSP instructions accept both immediate
8435 numbers and register operands, or accept immediate numbers only. The
8436 immediate parameters are listed as follows.
8445 imm_n32_31: -32 to 31.
8446 imm_n512_511: -512 to 511.
8449 The following built-in functions map directly to a particular MIPS DSP
8450 instruction. Please refer to the architecture specification
8451 for details on what each instruction does.
8454 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
8455 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
8456 q31 __builtin_mips_addq_s_w (q31, q31)
8457 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
8458 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
8459 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
8460 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
8461 q31 __builtin_mips_subq_s_w (q31, q31)
8462 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
8463 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
8464 i32 __builtin_mips_addsc (i32, i32)
8465 i32 __builtin_mips_addwc (i32, i32)
8466 i32 __builtin_mips_modsub (i32, i32)
8467 i32 __builtin_mips_raddu_w_qb (v4i8)
8468 v2q15 __builtin_mips_absq_s_ph (v2q15)
8469 q31 __builtin_mips_absq_s_w (q31)
8470 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
8471 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
8472 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
8473 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
8474 q31 __builtin_mips_preceq_w_phl (v2q15)
8475 q31 __builtin_mips_preceq_w_phr (v2q15)
8476 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
8477 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
8478 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
8479 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
8480 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
8481 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
8482 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
8483 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
8484 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
8485 v4i8 __builtin_mips_shll_qb (v4i8, i32)
8486 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
8487 v2q15 __builtin_mips_shll_ph (v2q15, i32)
8488 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
8489 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
8490 q31 __builtin_mips_shll_s_w (q31, imm0_31)
8491 q31 __builtin_mips_shll_s_w (q31, i32)
8492 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
8493 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
8494 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
8495 v2q15 __builtin_mips_shra_ph (v2q15, i32)
8496 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
8497 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
8498 q31 __builtin_mips_shra_r_w (q31, imm0_31)
8499 q31 __builtin_mips_shra_r_w (q31, i32)
8500 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
8501 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
8502 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
8503 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
8504 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
8505 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
8506 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
8507 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
8508 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
8509 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
8510 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
8511 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
8512 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
8513 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
8514 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
8515 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
8516 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
8517 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
8518 i32 __builtin_mips_bitrev (i32)
8519 i32 __builtin_mips_insv (i32, i32)
8520 v4i8 __builtin_mips_repl_qb (imm0_255)
8521 v4i8 __builtin_mips_repl_qb (i32)
8522 v2q15 __builtin_mips_repl_ph (imm_n512_511)
8523 v2q15 __builtin_mips_repl_ph (i32)
8524 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
8525 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
8526 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
8527 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
8528 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
8529 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
8530 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
8531 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
8532 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
8533 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
8534 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
8535 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
8536 i32 __builtin_mips_extr_w (a64, imm0_31)
8537 i32 __builtin_mips_extr_w (a64, i32)
8538 i32 __builtin_mips_extr_r_w (a64, imm0_31)
8539 i32 __builtin_mips_extr_s_h (a64, i32)
8540 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
8541 i32 __builtin_mips_extr_rs_w (a64, i32)
8542 i32 __builtin_mips_extr_s_h (a64, imm0_31)
8543 i32 __builtin_mips_extr_r_w (a64, i32)
8544 i32 __builtin_mips_extp (a64, imm0_31)
8545 i32 __builtin_mips_extp (a64, i32)
8546 i32 __builtin_mips_extpdp (a64, imm0_31)
8547 i32 __builtin_mips_extpdp (a64, i32)
8548 a64 __builtin_mips_shilo (a64, imm_n32_31)
8549 a64 __builtin_mips_shilo (a64, i32)
8550 a64 __builtin_mips_mthlip (a64, i32)
8551 void __builtin_mips_wrdsp (i32, imm0_63)
8552 i32 __builtin_mips_rddsp (imm0_63)
8553 i32 __builtin_mips_lbux (void *, i32)
8554 i32 __builtin_mips_lhx (void *, i32)
8555 i32 __builtin_mips_lwx (void *, i32)
8556 i32 __builtin_mips_bposge32 (void)
8559 The following built-in functions map directly to a particular MIPS DSP REV 2
8560 instruction. Please refer to the architecture specification
8561 for details on what each instruction does.
8564 v4q7 __builtin_mips_absq_s_qb (v4q7);
8565 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
8566 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
8567 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
8568 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
8569 i32 __builtin_mips_append (i32, i32, imm0_31);
8570 i32 __builtin_mips_balign (i32, i32, imm0_3);
8571 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
8572 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
8573 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
8574 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
8575 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
8576 a64 __builtin_mips_madd (a64, i32, i32);
8577 a64 __builtin_mips_maddu (a64, ui32, ui32);
8578 a64 __builtin_mips_msub (a64, i32, i32);
8579 a64 __builtin_mips_msubu (a64, ui32, ui32);
8580 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
8581 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
8582 q31 __builtin_mips_mulq_rs_w (q31, q31);
8583 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
8584 q31 __builtin_mips_mulq_s_w (q31, q31);
8585 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
8586 a64 __builtin_mips_mult (i32, i32);
8587 a64 __builtin_mips_multu (ui32, ui32);
8588 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
8589 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
8590 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
8591 i32 __builtin_mips_prepend (i32, i32, imm0_31);
8592 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
8593 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
8594 v4i8 __builtin_mips_shra_qb (v4i8, i32);
8595 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
8596 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
8597 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
8598 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
8599 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
8600 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
8601 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
8602 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
8603 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
8604 q31 __builtin_mips_addqh_w (q31, q31);
8605 q31 __builtin_mips_addqh_r_w (q31, q31);
8606 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
8607 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
8608 q31 __builtin_mips_subqh_w (q31, q31);
8609 q31 __builtin_mips_subqh_r_w (q31, q31);
8610 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
8611 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
8612 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
8613 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
8614 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
8615 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
8619 @node MIPS Paired-Single Support
8620 @subsection MIPS Paired-Single Support
8622 The MIPS64 architecture includes a number of instructions that
8623 operate on pairs of single-precision floating-point values.
8624 Each pair is packed into a 64-bit floating-point register,
8625 with one element being designated the ``upper half'' and
8626 the other being designated the ``lower half''.
8628 GCC supports paired-single operations using both the generic
8629 vector extensions (@pxref{Vector Extensions}) and a collection of
8630 MIPS-specific built-in functions. Both kinds of support are
8631 enabled by the @option{-mpaired-single} command-line option.
8633 The vector type associated with paired-single values is usually
8634 called @code{v2sf}. It can be defined in C as follows:
8637 typedef float v2sf __attribute__ ((vector_size (8)));
8640 @code{v2sf} values are initialized in the same way as aggregates.
8644 v2sf a = @{1.5, 9.1@};
8647 b = (v2sf) @{e, f@};
8650 @emph{Note:} The CPU's endianness determines which value is stored in
8651 the upper half of a register and which value is stored in the lower half.
8652 On little-endian targets, the first value is the lower one and the second
8653 value is the upper one. The opposite order applies to big-endian targets.
8654 For example, the code above will set the lower half of @code{a} to
8655 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
8658 * Paired-Single Arithmetic::
8659 * Paired-Single Built-in Functions::
8660 * MIPS-3D Built-in Functions::
8663 @node Paired-Single Arithmetic
8664 @subsubsection Paired-Single Arithmetic
8666 The table below lists the @code{v2sf} operations for which hardware
8667 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
8668 values and @code{x} is an integral value.
8670 @multitable @columnfractions .50 .50
8671 @item C code @tab MIPS instruction
8672 @item @code{a + b} @tab @code{add.ps}
8673 @item @code{a - b} @tab @code{sub.ps}
8674 @item @code{-a} @tab @code{neg.ps}
8675 @item @code{a * b} @tab @code{mul.ps}
8676 @item @code{a * b + c} @tab @code{madd.ps}
8677 @item @code{a * b - c} @tab @code{msub.ps}
8678 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
8679 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
8680 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
8683 Note that the multiply-accumulate instructions can be disabled
8684 using the command-line option @code{-mno-fused-madd}.
8686 @node Paired-Single Built-in Functions
8687 @subsubsection Paired-Single Built-in Functions
8689 The following paired-single functions map directly to a particular
8690 MIPS instruction. Please refer to the architecture specification
8691 for details on what each instruction does.
8694 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
8695 Pair lower lower (@code{pll.ps}).
8697 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
8698 Pair upper lower (@code{pul.ps}).
8700 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
8701 Pair lower upper (@code{plu.ps}).
8703 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
8704 Pair upper upper (@code{puu.ps}).
8706 @item v2sf __builtin_mips_cvt_ps_s (float, float)
8707 Convert pair to paired single (@code{cvt.ps.s}).
8709 @item float __builtin_mips_cvt_s_pl (v2sf)
8710 Convert pair lower to single (@code{cvt.s.pl}).
8712 @item float __builtin_mips_cvt_s_pu (v2sf)
8713 Convert pair upper to single (@code{cvt.s.pu}).
8715 @item v2sf __builtin_mips_abs_ps (v2sf)
8716 Absolute value (@code{abs.ps}).
8718 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
8719 Align variable (@code{alnv.ps}).
8721 @emph{Note:} The value of the third parameter must be 0 or 4
8722 modulo 8, otherwise the result will be unpredictable. Please read the
8723 instruction description for details.
8726 The following multi-instruction functions are also available.
8727 In each case, @var{cond} can be any of the 16 floating-point conditions:
8728 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
8729 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
8730 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
8733 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8734 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8735 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
8736 @code{movt.ps}/@code{movf.ps}).
8738 The @code{movt} functions return the value @var{x} computed by:
8741 c.@var{cond}.ps @var{cc},@var{a},@var{b}
8742 mov.ps @var{x},@var{c}
8743 movt.ps @var{x},@var{d},@var{cc}
8746 The @code{movf} functions are similar but use @code{movf.ps} instead
8749 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8750 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8751 Comparison of two paired-single values (@code{c.@var{cond}.ps},
8752 @code{bc1t}/@code{bc1f}).
8754 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
8755 and return either the upper or lower half of the result. For example:
8759 if (__builtin_mips_upper_c_eq_ps (a, b))
8760 upper_halves_are_equal ();
8762 upper_halves_are_unequal ();
8764 if (__builtin_mips_lower_c_eq_ps (a, b))
8765 lower_halves_are_equal ();
8767 lower_halves_are_unequal ();
8771 @node MIPS-3D Built-in Functions
8772 @subsubsection MIPS-3D Built-in Functions
8774 The MIPS-3D Application-Specific Extension (ASE) includes additional
8775 paired-single instructions that are designed to improve the performance
8776 of 3D graphics operations. Support for these instructions is controlled
8777 by the @option{-mips3d} command-line option.
8779 The functions listed below map directly to a particular MIPS-3D
8780 instruction. Please refer to the architecture specification for
8781 more details on what each instruction does.
8784 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
8785 Reduction add (@code{addr.ps}).
8787 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
8788 Reduction multiply (@code{mulr.ps}).
8790 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
8791 Convert paired single to paired word (@code{cvt.pw.ps}).
8793 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
8794 Convert paired word to paired single (@code{cvt.ps.pw}).
8796 @item float __builtin_mips_recip1_s (float)
8797 @itemx double __builtin_mips_recip1_d (double)
8798 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
8799 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
8801 @item float __builtin_mips_recip2_s (float, float)
8802 @itemx double __builtin_mips_recip2_d (double, double)
8803 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
8804 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
8806 @item float __builtin_mips_rsqrt1_s (float)
8807 @itemx double __builtin_mips_rsqrt1_d (double)
8808 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
8809 Reduced precision reciprocal square root (sequence step 1)
8810 (@code{rsqrt1.@var{fmt}}).
8812 @item float __builtin_mips_rsqrt2_s (float, float)
8813 @itemx double __builtin_mips_rsqrt2_d (double, double)
8814 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
8815 Reduced precision reciprocal square root (sequence step 2)
8816 (@code{rsqrt2.@var{fmt}}).
8819 The following multi-instruction functions are also available.
8820 In each case, @var{cond} can be any of the 16 floating-point conditions:
8821 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
8822 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
8823 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
8826 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
8827 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
8828 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
8829 @code{bc1t}/@code{bc1f}).
8831 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
8832 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
8837 if (__builtin_mips_cabs_eq_s (a, b))
8843 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8844 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8845 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
8846 @code{bc1t}/@code{bc1f}).
8848 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
8849 and return either the upper or lower half of the result. For example:
8853 if (__builtin_mips_upper_cabs_eq_ps (a, b))
8854 upper_halves_are_equal ();
8856 upper_halves_are_unequal ();
8858 if (__builtin_mips_lower_cabs_eq_ps (a, b))
8859 lower_halves_are_equal ();
8861 lower_halves_are_unequal ();
8864 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8865 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8866 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
8867 @code{movt.ps}/@code{movf.ps}).
8869 The @code{movt} functions return the value @var{x} computed by:
8872 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
8873 mov.ps @var{x},@var{c}
8874 movt.ps @var{x},@var{d},@var{cc}
8877 The @code{movf} functions are similar but use @code{movf.ps} instead
8880 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8881 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8882 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8883 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8884 Comparison of two paired-single values
8885 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
8886 @code{bc1any2t}/@code{bc1any2f}).
8888 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
8889 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
8890 result is true and the @code{all} forms return true if both results are true.
8895 if (__builtin_mips_any_c_eq_ps (a, b))
8900 if (__builtin_mips_all_c_eq_ps (a, b))
8906 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8907 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8908 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8909 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8910 Comparison of four paired-single values
8911 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
8912 @code{bc1any4t}/@code{bc1any4f}).
8914 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
8915 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
8916 The @code{any} forms return true if any of the four results are true
8917 and the @code{all} forms return true if all four results are true.
8922 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
8927 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
8934 @node PowerPC AltiVec Built-in Functions
8935 @subsection PowerPC AltiVec Built-in Functions
8937 GCC provides an interface for the PowerPC family of processors to access
8938 the AltiVec operations described in Motorola's AltiVec Programming
8939 Interface Manual. The interface is made available by including
8940 @code{<altivec.h>} and using @option{-maltivec} and
8941 @option{-mabi=altivec}. The interface supports the following vector
8945 vector unsigned char
8949 vector unsigned short
8960 GCC's implementation of the high-level language interface available from
8961 C and C++ code differs from Motorola's documentation in several ways.
8966 A vector constant is a list of constant expressions within curly braces.
8969 A vector initializer requires no cast if the vector constant is of the
8970 same type as the variable it is initializing.
8973 If @code{signed} or @code{unsigned} is omitted, the signedness of the
8974 vector type is the default signedness of the base type. The default
8975 varies depending on the operating system, so a portable program should
8976 always specify the signedness.
8979 Compiling with @option{-maltivec} adds keywords @code{__vector},
8980 @code{__pixel}, and @code{__bool}. Macros @option{vector},
8981 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
8985 GCC allows using a @code{typedef} name as the type specifier for a
8989 For C, overloaded functions are implemented with macros so the following
8993 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
8996 Since @code{vec_add} is a macro, the vector constant in the example
8997 is treated as four separate arguments. Wrap the entire argument in
8998 parentheses for this to work.
9001 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
9002 Internally, GCC uses built-in functions to achieve the functionality in
9003 the aforementioned header file, but they are not supported and are
9004 subject to change without notice.
9006 The following interfaces are supported for the generic and specific
9007 AltiVec operations and the AltiVec predicates. In cases where there
9008 is a direct mapping between generic and specific operations, only the
9009 generic names are shown here, although the specific operations can also
9012 Arguments that are documented as @code{const int} require literal
9013 integral values within the range required for that operation.
9016 vector signed char vec_abs (vector signed char);
9017 vector signed short vec_abs (vector signed short);
9018 vector signed int vec_abs (vector signed int);
9019 vector float vec_abs (vector float);
9021 vector signed char vec_abss (vector signed char);
9022 vector signed short vec_abss (vector signed short);
9023 vector signed int vec_abss (vector signed int);
9025 vector signed char vec_add (vector bool char, vector signed char);
9026 vector signed char vec_add (vector signed char, vector bool char);
9027 vector signed char vec_add (vector signed char, vector signed char);
9028 vector unsigned char vec_add (vector bool char, vector unsigned char);
9029 vector unsigned char vec_add (vector unsigned char, vector bool char);
9030 vector unsigned char vec_add (vector unsigned char,
9031 vector unsigned char);
9032 vector signed short vec_add (vector bool short, vector signed short);
9033 vector signed short vec_add (vector signed short, vector bool short);
9034 vector signed short vec_add (vector signed short, vector signed short);
9035 vector unsigned short vec_add (vector bool short,
9036 vector unsigned short);
9037 vector unsigned short vec_add (vector unsigned short,
9039 vector unsigned short vec_add (vector unsigned short,
9040 vector unsigned short);
9041 vector signed int vec_add (vector bool int, vector signed int);
9042 vector signed int vec_add (vector signed int, vector bool int);
9043 vector signed int vec_add (vector signed int, vector signed int);
9044 vector unsigned int vec_add (vector bool int, vector unsigned int);
9045 vector unsigned int vec_add (vector unsigned int, vector bool int);
9046 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
9047 vector float vec_add (vector float, vector float);
9049 vector float vec_vaddfp (vector float, vector float);
9051 vector signed int vec_vadduwm (vector bool int, vector signed int);
9052 vector signed int vec_vadduwm (vector signed int, vector bool int);
9053 vector signed int vec_vadduwm (vector signed int, vector signed int);
9054 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
9055 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
9056 vector unsigned int vec_vadduwm (vector unsigned int,
9057 vector unsigned int);
9059 vector signed short vec_vadduhm (vector bool short,
9060 vector signed short);
9061 vector signed short vec_vadduhm (vector signed short,
9063 vector signed short vec_vadduhm (vector signed short,
9064 vector signed short);
9065 vector unsigned short vec_vadduhm (vector bool short,
9066 vector unsigned short);
9067 vector unsigned short vec_vadduhm (vector unsigned short,
9069 vector unsigned short vec_vadduhm (vector unsigned short,
9070 vector unsigned short);
9072 vector signed char vec_vaddubm (vector bool char, vector signed char);
9073 vector signed char vec_vaddubm (vector signed char, vector bool char);
9074 vector signed char vec_vaddubm (vector signed char, vector signed char);
9075 vector unsigned char vec_vaddubm (vector bool char,
9076 vector unsigned char);
9077 vector unsigned char vec_vaddubm (vector unsigned char,
9079 vector unsigned char vec_vaddubm (vector unsigned char,
9080 vector unsigned char);
9082 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
9084 vector unsigned char vec_adds (vector bool char, vector unsigned char);
9085 vector unsigned char vec_adds (vector unsigned char, vector bool char);
9086 vector unsigned char vec_adds (vector unsigned char,
9087 vector unsigned char);
9088 vector signed char vec_adds (vector bool char, vector signed char);
9089 vector signed char vec_adds (vector signed char, vector bool char);
9090 vector signed char vec_adds (vector signed char, vector signed char);
9091 vector unsigned short vec_adds (vector bool short,
9092 vector unsigned short);
9093 vector unsigned short vec_adds (vector unsigned short,
9095 vector unsigned short vec_adds (vector unsigned short,
9096 vector unsigned short);
9097 vector signed short vec_adds (vector bool short, vector signed short);
9098 vector signed short vec_adds (vector signed short, vector bool short);
9099 vector signed short vec_adds (vector signed short, vector signed short);
9100 vector unsigned int vec_adds (vector bool int, vector unsigned int);
9101 vector unsigned int vec_adds (vector unsigned int, vector bool int);
9102 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
9103 vector signed int vec_adds (vector bool int, vector signed int);
9104 vector signed int vec_adds (vector signed int, vector bool int);
9105 vector signed int vec_adds (vector signed int, vector signed int);
9107 vector signed int vec_vaddsws (vector bool int, vector signed int);
9108 vector signed int vec_vaddsws (vector signed int, vector bool int);
9109 vector signed int vec_vaddsws (vector signed int, vector signed int);
9111 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
9112 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
9113 vector unsigned int vec_vadduws (vector unsigned int,
9114 vector unsigned int);
9116 vector signed short vec_vaddshs (vector bool short,
9117 vector signed short);
9118 vector signed short vec_vaddshs (vector signed short,
9120 vector signed short vec_vaddshs (vector signed short,
9121 vector signed short);
9123 vector unsigned short vec_vadduhs (vector bool short,
9124 vector unsigned short);
9125 vector unsigned short vec_vadduhs (vector unsigned short,
9127 vector unsigned short vec_vadduhs (vector unsigned short,
9128 vector unsigned short);
9130 vector signed char vec_vaddsbs (vector bool char, vector signed char);
9131 vector signed char vec_vaddsbs (vector signed char, vector bool char);
9132 vector signed char vec_vaddsbs (vector signed char, vector signed char);
9134 vector unsigned char vec_vaddubs (vector bool char,
9135 vector unsigned char);
9136 vector unsigned char vec_vaddubs (vector unsigned char,
9138 vector unsigned char vec_vaddubs (vector unsigned char,
9139 vector unsigned char);
9141 vector float vec_and (vector float, vector float);
9142 vector float vec_and (vector float, vector bool int);
9143 vector float vec_and (vector bool int, vector float);
9144 vector bool int vec_and (vector bool int, vector bool int);
9145 vector signed int vec_and (vector bool int, vector signed int);
9146 vector signed int vec_and (vector signed int, vector bool int);
9147 vector signed int vec_and (vector signed int, vector signed int);
9148 vector unsigned int vec_and (vector bool int, vector unsigned int);
9149 vector unsigned int vec_and (vector unsigned int, vector bool int);
9150 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
9151 vector bool short vec_and (vector bool short, vector bool short);
9152 vector signed short vec_and (vector bool short, vector signed short);
9153 vector signed short vec_and (vector signed short, vector bool short);
9154 vector signed short vec_and (vector signed short, vector signed short);
9155 vector unsigned short vec_and (vector bool short,
9156 vector unsigned short);
9157 vector unsigned short vec_and (vector unsigned short,
9159 vector unsigned short vec_and (vector unsigned short,
9160 vector unsigned short);
9161 vector signed char vec_and (vector bool char, vector signed char);
9162 vector bool char vec_and (vector bool char, vector bool char);
9163 vector signed char vec_and (vector signed char, vector bool char);
9164 vector signed char vec_and (vector signed char, vector signed char);
9165 vector unsigned char vec_and (vector bool char, vector unsigned char);
9166 vector unsigned char vec_and (vector unsigned char, vector bool char);
9167 vector unsigned char vec_and (vector unsigned char,
9168 vector unsigned char);
9170 vector float vec_andc (vector float, vector float);
9171 vector float vec_andc (vector float, vector bool int);
9172 vector float vec_andc (vector bool int, vector float);
9173 vector bool int vec_andc (vector bool int, vector bool int);
9174 vector signed int vec_andc (vector bool int, vector signed int);
9175 vector signed int vec_andc (vector signed int, vector bool int);
9176 vector signed int vec_andc (vector signed int, vector signed int);
9177 vector unsigned int vec_andc (vector bool int, vector unsigned int);
9178 vector unsigned int vec_andc (vector unsigned int, vector bool int);
9179 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
9180 vector bool short vec_andc (vector bool short, vector bool short);
9181 vector signed short vec_andc (vector bool short, vector signed short);
9182 vector signed short vec_andc (vector signed short, vector bool short);
9183 vector signed short vec_andc (vector signed short, vector signed short);
9184 vector unsigned short vec_andc (vector bool short,
9185 vector unsigned short);
9186 vector unsigned short vec_andc (vector unsigned short,
9188 vector unsigned short vec_andc (vector unsigned short,
9189 vector unsigned short);
9190 vector signed char vec_andc (vector bool char, vector signed char);
9191 vector bool char vec_andc (vector bool char, vector bool char);
9192 vector signed char vec_andc (vector signed char, vector bool char);
9193 vector signed char vec_andc (vector signed char, vector signed char);
9194 vector unsigned char vec_andc (vector bool char, vector unsigned char);
9195 vector unsigned char vec_andc (vector unsigned char, vector bool char);
9196 vector unsigned char vec_andc (vector unsigned char,
9197 vector unsigned char);
9199 vector unsigned char vec_avg (vector unsigned char,
9200 vector unsigned char);
9201 vector signed char vec_avg (vector signed char, vector signed char);
9202 vector unsigned short vec_avg (vector unsigned short,
9203 vector unsigned short);
9204 vector signed short vec_avg (vector signed short, vector signed short);
9205 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
9206 vector signed int vec_avg (vector signed int, vector signed int);
9208 vector signed int vec_vavgsw (vector signed int, vector signed int);
9210 vector unsigned int vec_vavguw (vector unsigned int,
9211 vector unsigned int);
9213 vector signed short vec_vavgsh (vector signed short,
9214 vector signed short);
9216 vector unsigned short vec_vavguh (vector unsigned short,
9217 vector unsigned short);
9219 vector signed char vec_vavgsb (vector signed char, vector signed char);
9221 vector unsigned char vec_vavgub (vector unsigned char,
9222 vector unsigned char);
9224 vector float vec_ceil (vector float);
9226 vector signed int vec_cmpb (vector float, vector float);
9228 vector bool char vec_cmpeq (vector signed char, vector signed char);
9229 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
9230 vector bool short vec_cmpeq (vector signed short, vector signed short);
9231 vector bool short vec_cmpeq (vector unsigned short,
9232 vector unsigned short);
9233 vector bool int vec_cmpeq (vector signed int, vector signed int);
9234 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
9235 vector bool int vec_cmpeq (vector float, vector float);
9237 vector bool int vec_vcmpeqfp (vector float, vector float);
9239 vector bool int vec_vcmpequw (vector signed int, vector signed int);
9240 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
9242 vector bool short vec_vcmpequh (vector signed short,
9243 vector signed short);
9244 vector bool short vec_vcmpequh (vector unsigned short,
9245 vector unsigned short);
9247 vector bool char vec_vcmpequb (vector signed char, vector signed char);
9248 vector bool char vec_vcmpequb (vector unsigned char,
9249 vector unsigned char);
9251 vector bool int vec_cmpge (vector float, vector float);
9253 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
9254 vector bool char vec_cmpgt (vector signed char, vector signed char);
9255 vector bool short vec_cmpgt (vector unsigned short,
9256 vector unsigned short);
9257 vector bool short vec_cmpgt (vector signed short, vector signed short);
9258 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
9259 vector bool int vec_cmpgt (vector signed int, vector signed int);
9260 vector bool int vec_cmpgt (vector float, vector float);
9262 vector bool int vec_vcmpgtfp (vector float, vector float);
9264 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
9266 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
9268 vector bool short vec_vcmpgtsh (vector signed short,
9269 vector signed short);
9271 vector bool short vec_vcmpgtuh (vector unsigned short,
9272 vector unsigned short);
9274 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
9276 vector bool char vec_vcmpgtub (vector unsigned char,
9277 vector unsigned char);
9279 vector bool int vec_cmple (vector float, vector float);
9281 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
9282 vector bool char vec_cmplt (vector signed char, vector signed char);
9283 vector bool short vec_cmplt (vector unsigned short,
9284 vector unsigned short);
9285 vector bool short vec_cmplt (vector signed short, vector signed short);
9286 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
9287 vector bool int vec_cmplt (vector signed int, vector signed int);
9288 vector bool int vec_cmplt (vector float, vector float);
9290 vector float vec_ctf (vector unsigned int, const int);
9291 vector float vec_ctf (vector signed int, const int);
9293 vector float vec_vcfsx (vector signed int, const int);
9295 vector float vec_vcfux (vector unsigned int, const int);
9297 vector signed int vec_cts (vector float, const int);
9299 vector unsigned int vec_ctu (vector float, const int);
9301 void vec_dss (const int);
9303 void vec_dssall (void);
9305 void vec_dst (const vector unsigned char *, int, const int);
9306 void vec_dst (const vector signed char *, int, const int);
9307 void vec_dst (const vector bool char *, int, const int);
9308 void vec_dst (const vector unsigned short *, int, const int);
9309 void vec_dst (const vector signed short *, int, const int);
9310 void vec_dst (const vector bool short *, int, const int);
9311 void vec_dst (const vector pixel *, int, const int);
9312 void vec_dst (const vector unsigned int *, int, const int);
9313 void vec_dst (const vector signed int *, int, const int);
9314 void vec_dst (const vector bool int *, int, const int);
9315 void vec_dst (const vector float *, int, const int);
9316 void vec_dst (const unsigned char *, int, const int);
9317 void vec_dst (const signed char *, int, const int);
9318 void vec_dst (const unsigned short *, int, const int);
9319 void vec_dst (const short *, int, const int);
9320 void vec_dst (const unsigned int *, int, const int);
9321 void vec_dst (const int *, int, const int);
9322 void vec_dst (const unsigned long *, int, const int);
9323 void vec_dst (const long *, int, const int);
9324 void vec_dst (const float *, int, const int);
9326 void vec_dstst (const vector unsigned char *, int, const int);
9327 void vec_dstst (const vector signed char *, int, const int);
9328 void vec_dstst (const vector bool char *, int, const int);
9329 void vec_dstst (const vector unsigned short *, int, const int);
9330 void vec_dstst (const vector signed short *, int, const int);
9331 void vec_dstst (const vector bool short *, int, const int);
9332 void vec_dstst (const vector pixel *, int, const int);
9333 void vec_dstst (const vector unsigned int *, int, const int);
9334 void vec_dstst (const vector signed int *, int, const int);
9335 void vec_dstst (const vector bool int *, int, const int);
9336 void vec_dstst (const vector float *, int, const int);
9337 void vec_dstst (const unsigned char *, int, const int);
9338 void vec_dstst (const signed char *, int, const int);
9339 void vec_dstst (const unsigned short *, int, const int);
9340 void vec_dstst (const short *, int, const int);
9341 void vec_dstst (const unsigned int *, int, const int);
9342 void vec_dstst (const int *, int, const int);
9343 void vec_dstst (const unsigned long *, int, const int);
9344 void vec_dstst (const long *, int, const int);
9345 void vec_dstst (const float *, int, const int);
9347 void vec_dststt (const vector unsigned char *, int, const int);
9348 void vec_dststt (const vector signed char *, int, const int);
9349 void vec_dststt (const vector bool char *, int, const int);
9350 void vec_dststt (const vector unsigned short *, int, const int);
9351 void vec_dststt (const vector signed short *, int, const int);
9352 void vec_dststt (const vector bool short *, int, const int);
9353 void vec_dststt (const vector pixel *, int, const int);
9354 void vec_dststt (const vector unsigned int *, int, const int);
9355 void vec_dststt (const vector signed int *, int, const int);
9356 void vec_dststt (const vector bool int *, int, const int);
9357 void vec_dststt (const vector float *, int, const int);
9358 void vec_dststt (const unsigned char *, int, const int);
9359 void vec_dststt (const signed char *, int, const int);
9360 void vec_dststt (const unsigned short *, int, const int);
9361 void vec_dststt (const short *, int, const int);
9362 void vec_dststt (const unsigned int *, int, const int);
9363 void vec_dststt (const int *, int, const int);
9364 void vec_dststt (const unsigned long *, int, const int);
9365 void vec_dststt (const long *, int, const int);
9366 void vec_dststt (const float *, int, const int);
9368 void vec_dstt (const vector unsigned char *, int, const int);
9369 void vec_dstt (const vector signed char *, int, const int);
9370 void vec_dstt (const vector bool char *, int, const int);
9371 void vec_dstt (const vector unsigned short *, int, const int);
9372 void vec_dstt (const vector signed short *, int, const int);
9373 void vec_dstt (const vector bool short *, int, const int);
9374 void vec_dstt (const vector pixel *, int, const int);
9375 void vec_dstt (const vector unsigned int *, int, const int);
9376 void vec_dstt (const vector signed int *, int, const int);
9377 void vec_dstt (const vector bool int *, int, const int);
9378 void vec_dstt (const vector float *, int, const int);
9379 void vec_dstt (const unsigned char *, int, const int);
9380 void vec_dstt (const signed char *, int, const int);
9381 void vec_dstt (const unsigned short *, int, const int);
9382 void vec_dstt (const short *, int, const int);
9383 void vec_dstt (const unsigned int *, int, const int);
9384 void vec_dstt (const int *, int, const int);
9385 void vec_dstt (const unsigned long *, int, const int);
9386 void vec_dstt (const long *, int, const int);
9387 void vec_dstt (const float *, int, const int);
9389 vector float vec_expte (vector float);
9391 vector float vec_floor (vector float);
9393 vector float vec_ld (int, const vector float *);
9394 vector float vec_ld (int, const float *);
9395 vector bool int vec_ld (int, const vector bool int *);
9396 vector signed int vec_ld (int, const vector signed int *);
9397 vector signed int vec_ld (int, const int *);
9398 vector signed int vec_ld (int, const long *);
9399 vector unsigned int vec_ld (int, const vector unsigned int *);
9400 vector unsigned int vec_ld (int, const unsigned int *);
9401 vector unsigned int vec_ld (int, const unsigned long *);
9402 vector bool short vec_ld (int, const vector bool short *);
9403 vector pixel vec_ld (int, const vector pixel *);
9404 vector signed short vec_ld (int, const vector signed short *);
9405 vector signed short vec_ld (int, const short *);
9406 vector unsigned short vec_ld (int, const vector unsigned short *);
9407 vector unsigned short vec_ld (int, const unsigned short *);
9408 vector bool char vec_ld (int, const vector bool char *);
9409 vector signed char vec_ld (int, const vector signed char *);
9410 vector signed char vec_ld (int, const signed char *);
9411 vector unsigned char vec_ld (int, const vector unsigned char *);
9412 vector unsigned char vec_ld (int, const unsigned char *);
9414 vector signed char vec_lde (int, const signed char *);
9415 vector unsigned char vec_lde (int, const unsigned char *);
9416 vector signed short vec_lde (int, const short *);
9417 vector unsigned short vec_lde (int, const unsigned short *);
9418 vector float vec_lde (int, const float *);
9419 vector signed int vec_lde (int, const int *);
9420 vector unsigned int vec_lde (int, const unsigned int *);
9421 vector signed int vec_lde (int, const long *);
9422 vector unsigned int vec_lde (int, const unsigned long *);
9424 vector float vec_lvewx (int, float *);
9425 vector signed int vec_lvewx (int, int *);
9426 vector unsigned int vec_lvewx (int, unsigned int *);
9427 vector signed int vec_lvewx (int, long *);
9428 vector unsigned int vec_lvewx (int, unsigned long *);
9430 vector signed short vec_lvehx (int, short *);
9431 vector unsigned short vec_lvehx (int, unsigned short *);
9433 vector signed char vec_lvebx (int, char *);
9434 vector unsigned char vec_lvebx (int, unsigned char *);
9436 vector float vec_ldl (int, const vector float *);
9437 vector float vec_ldl (int, const float *);
9438 vector bool int vec_ldl (int, const vector bool int *);
9439 vector signed int vec_ldl (int, const vector signed int *);
9440 vector signed int vec_ldl (int, const int *);
9441 vector signed int vec_ldl (int, const long *);
9442 vector unsigned int vec_ldl (int, const vector unsigned int *);
9443 vector unsigned int vec_ldl (int, const unsigned int *);
9444 vector unsigned int vec_ldl (int, const unsigned long *);
9445 vector bool short vec_ldl (int, const vector bool short *);
9446 vector pixel vec_ldl (int, const vector pixel *);
9447 vector signed short vec_ldl (int, const vector signed short *);
9448 vector signed short vec_ldl (int, const short *);
9449 vector unsigned short vec_ldl (int, const vector unsigned short *);
9450 vector unsigned short vec_ldl (int, const unsigned short *);
9451 vector bool char vec_ldl (int, const vector bool char *);
9452 vector signed char vec_ldl (int, const vector signed char *);
9453 vector signed char vec_ldl (int, const signed char *);
9454 vector unsigned char vec_ldl (int, const vector unsigned char *);
9455 vector unsigned char vec_ldl (int, const unsigned char *);
9457 vector float vec_loge (vector float);
9459 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
9460 vector unsigned char vec_lvsl (int, const volatile signed char *);
9461 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
9462 vector unsigned char vec_lvsl (int, const volatile short *);
9463 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
9464 vector unsigned char vec_lvsl (int, const volatile int *);
9465 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
9466 vector unsigned char vec_lvsl (int, const volatile long *);
9467 vector unsigned char vec_lvsl (int, const volatile float *);
9469 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
9470 vector unsigned char vec_lvsr (int, const volatile signed char *);
9471 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
9472 vector unsigned char vec_lvsr (int, const volatile short *);
9473 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
9474 vector unsigned char vec_lvsr (int, const volatile int *);
9475 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
9476 vector unsigned char vec_lvsr (int, const volatile long *);
9477 vector unsigned char vec_lvsr (int, const volatile float *);
9479 vector float vec_madd (vector float, vector float, vector float);
9481 vector signed short vec_madds (vector signed short,
9482 vector signed short,
9483 vector signed short);
9485 vector unsigned char vec_max (vector bool char, vector unsigned char);
9486 vector unsigned char vec_max (vector unsigned char, vector bool char);
9487 vector unsigned char vec_max (vector unsigned char,
9488 vector unsigned char);
9489 vector signed char vec_max (vector bool char, vector signed char);
9490 vector signed char vec_max (vector signed char, vector bool char);
9491 vector signed char vec_max (vector signed char, vector signed char);
9492 vector unsigned short vec_max (vector bool short,
9493 vector unsigned short);
9494 vector unsigned short vec_max (vector unsigned short,
9496 vector unsigned short vec_max (vector unsigned short,
9497 vector unsigned short);
9498 vector signed short vec_max (vector bool short, vector signed short);
9499 vector signed short vec_max (vector signed short, vector bool short);
9500 vector signed short vec_max (vector signed short, vector signed short);
9501 vector unsigned int vec_max (vector bool int, vector unsigned int);
9502 vector unsigned int vec_max (vector unsigned int, vector bool int);
9503 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
9504 vector signed int vec_max (vector bool int, vector signed int);
9505 vector signed int vec_max (vector signed int, vector bool int);
9506 vector signed int vec_max (vector signed int, vector signed int);
9507 vector float vec_max (vector float, vector float);
9509 vector float vec_vmaxfp (vector float, vector float);
9511 vector signed int vec_vmaxsw (vector bool int, vector signed int);
9512 vector signed int vec_vmaxsw (vector signed int, vector bool int);
9513 vector signed int vec_vmaxsw (vector signed int, vector signed int);
9515 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
9516 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
9517 vector unsigned int vec_vmaxuw (vector unsigned int,
9518 vector unsigned int);
9520 vector signed short vec_vmaxsh (vector bool short, vector signed short);
9521 vector signed short vec_vmaxsh (vector signed short, vector bool short);
9522 vector signed short vec_vmaxsh (vector signed short,
9523 vector signed short);
9525 vector unsigned short vec_vmaxuh (vector bool short,
9526 vector unsigned short);
9527 vector unsigned short vec_vmaxuh (vector unsigned short,
9529 vector unsigned short vec_vmaxuh (vector unsigned short,
9530 vector unsigned short);
9532 vector signed char vec_vmaxsb (vector bool char, vector signed char);
9533 vector signed char vec_vmaxsb (vector signed char, vector bool char);
9534 vector signed char vec_vmaxsb (vector signed char, vector signed char);
9536 vector unsigned char vec_vmaxub (vector bool char,
9537 vector unsigned char);
9538 vector unsigned char vec_vmaxub (vector unsigned char,
9540 vector unsigned char vec_vmaxub (vector unsigned char,
9541 vector unsigned char);
9543 vector bool char vec_mergeh (vector bool char, vector bool char);
9544 vector signed char vec_mergeh (vector signed char, vector signed char);
9545 vector unsigned char vec_mergeh (vector unsigned char,
9546 vector unsigned char);
9547 vector bool short vec_mergeh (vector bool short, vector bool short);
9548 vector pixel vec_mergeh (vector pixel, vector pixel);
9549 vector signed short vec_mergeh (vector signed short,
9550 vector signed short);
9551 vector unsigned short vec_mergeh (vector unsigned short,
9552 vector unsigned short);
9553 vector float vec_mergeh (vector float, vector float);
9554 vector bool int vec_mergeh (vector bool int, vector bool int);
9555 vector signed int vec_mergeh (vector signed int, vector signed int);
9556 vector unsigned int vec_mergeh (vector unsigned int,
9557 vector unsigned int);
9559 vector float vec_vmrghw (vector float, vector float);
9560 vector bool int vec_vmrghw (vector bool int, vector bool int);
9561 vector signed int vec_vmrghw (vector signed int, vector signed int);
9562 vector unsigned int vec_vmrghw (vector unsigned int,
9563 vector unsigned int);
9565 vector bool short vec_vmrghh (vector bool short, vector bool short);
9566 vector signed short vec_vmrghh (vector signed short,
9567 vector signed short);
9568 vector unsigned short vec_vmrghh (vector unsigned short,
9569 vector unsigned short);
9570 vector pixel vec_vmrghh (vector pixel, vector pixel);
9572 vector bool char vec_vmrghb (vector bool char, vector bool char);
9573 vector signed char vec_vmrghb (vector signed char, vector signed char);
9574 vector unsigned char vec_vmrghb (vector unsigned char,
9575 vector unsigned char);
9577 vector bool char vec_mergel (vector bool char, vector bool char);
9578 vector signed char vec_mergel (vector signed char, vector signed char);
9579 vector unsigned char vec_mergel (vector unsigned char,
9580 vector unsigned char);
9581 vector bool short vec_mergel (vector bool short, vector bool short);
9582 vector pixel vec_mergel (vector pixel, vector pixel);
9583 vector signed short vec_mergel (vector signed short,
9584 vector signed short);
9585 vector unsigned short vec_mergel (vector unsigned short,
9586 vector unsigned short);
9587 vector float vec_mergel (vector float, vector float);
9588 vector bool int vec_mergel (vector bool int, vector bool int);
9589 vector signed int vec_mergel (vector signed int, vector signed int);
9590 vector unsigned int vec_mergel (vector unsigned int,
9591 vector unsigned int);
9593 vector float vec_vmrglw (vector float, vector float);
9594 vector signed int vec_vmrglw (vector signed int, vector signed int);
9595 vector unsigned int vec_vmrglw (vector unsigned int,
9596 vector unsigned int);
9597 vector bool int vec_vmrglw (vector bool int, vector bool int);
9599 vector bool short vec_vmrglh (vector bool short, vector bool short);
9600 vector signed short vec_vmrglh (vector signed short,
9601 vector signed short);
9602 vector unsigned short vec_vmrglh (vector unsigned short,
9603 vector unsigned short);
9604 vector pixel vec_vmrglh (vector pixel, vector pixel);
9606 vector bool char vec_vmrglb (vector bool char, vector bool char);
9607 vector signed char vec_vmrglb (vector signed char, vector signed char);
9608 vector unsigned char vec_vmrglb (vector unsigned char,
9609 vector unsigned char);
9611 vector unsigned short vec_mfvscr (void);
9613 vector unsigned char vec_min (vector bool char, vector unsigned char);
9614 vector unsigned char vec_min (vector unsigned char, vector bool char);
9615 vector unsigned char vec_min (vector unsigned char,
9616 vector unsigned char);
9617 vector signed char vec_min (vector bool char, vector signed char);
9618 vector signed char vec_min (vector signed char, vector bool char);
9619 vector signed char vec_min (vector signed char, vector signed char);
9620 vector unsigned short vec_min (vector bool short,
9621 vector unsigned short);
9622 vector unsigned short vec_min (vector unsigned short,
9624 vector unsigned short vec_min (vector unsigned short,
9625 vector unsigned short);
9626 vector signed short vec_min (vector bool short, vector signed short);
9627 vector signed short vec_min (vector signed short, vector bool short);
9628 vector signed short vec_min (vector signed short, vector signed short);
9629 vector unsigned int vec_min (vector bool int, vector unsigned int);
9630 vector unsigned int vec_min (vector unsigned int, vector bool int);
9631 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
9632 vector signed int vec_min (vector bool int, vector signed int);
9633 vector signed int vec_min (vector signed int, vector bool int);
9634 vector signed int vec_min (vector signed int, vector signed int);
9635 vector float vec_min (vector float, vector float);
9637 vector float vec_vminfp (vector float, vector float);
9639 vector signed int vec_vminsw (vector bool int, vector signed int);
9640 vector signed int vec_vminsw (vector signed int, vector bool int);
9641 vector signed int vec_vminsw (vector signed int, vector signed int);
9643 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
9644 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
9645 vector unsigned int vec_vminuw (vector unsigned int,
9646 vector unsigned int);
9648 vector signed short vec_vminsh (vector bool short, vector signed short);
9649 vector signed short vec_vminsh (vector signed short, vector bool short);
9650 vector signed short vec_vminsh (vector signed short,
9651 vector signed short);
9653 vector unsigned short vec_vminuh (vector bool short,
9654 vector unsigned short);
9655 vector unsigned short vec_vminuh (vector unsigned short,
9657 vector unsigned short vec_vminuh (vector unsigned short,
9658 vector unsigned short);
9660 vector signed char vec_vminsb (vector bool char, vector signed char);
9661 vector signed char vec_vminsb (vector signed char, vector bool char);
9662 vector signed char vec_vminsb (vector signed char, vector signed char);
9664 vector unsigned char vec_vminub (vector bool char,
9665 vector unsigned char);
9666 vector unsigned char vec_vminub (vector unsigned char,
9668 vector unsigned char vec_vminub (vector unsigned char,
9669 vector unsigned char);
9671 vector signed short vec_mladd (vector signed short,
9672 vector signed short,
9673 vector signed short);
9674 vector signed short vec_mladd (vector signed short,
9675 vector unsigned short,
9676 vector unsigned short);
9677 vector signed short vec_mladd (vector unsigned short,
9678 vector signed short,
9679 vector signed short);
9680 vector unsigned short vec_mladd (vector unsigned short,
9681 vector unsigned short,
9682 vector unsigned short);
9684 vector signed short vec_mradds (vector signed short,
9685 vector signed short,
9686 vector signed short);
9688 vector unsigned int vec_msum (vector unsigned char,
9689 vector unsigned char,
9690 vector unsigned int);
9691 vector signed int vec_msum (vector signed char,
9692 vector unsigned char,
9694 vector unsigned int vec_msum (vector unsigned short,
9695 vector unsigned short,
9696 vector unsigned int);
9697 vector signed int vec_msum (vector signed short,
9698 vector signed short,
9701 vector signed int vec_vmsumshm (vector signed short,
9702 vector signed short,
9705 vector unsigned int vec_vmsumuhm (vector unsigned short,
9706 vector unsigned short,
9707 vector unsigned int);
9709 vector signed int vec_vmsummbm (vector signed char,
9710 vector unsigned char,
9713 vector unsigned int vec_vmsumubm (vector unsigned char,
9714 vector unsigned char,
9715 vector unsigned int);
9717 vector unsigned int vec_msums (vector unsigned short,
9718 vector unsigned short,
9719 vector unsigned int);
9720 vector signed int vec_msums (vector signed short,
9721 vector signed short,
9724 vector signed int vec_vmsumshs (vector signed short,
9725 vector signed short,
9728 vector unsigned int vec_vmsumuhs (vector unsigned short,
9729 vector unsigned short,
9730 vector unsigned int);
9732 void vec_mtvscr (vector signed int);
9733 void vec_mtvscr (vector unsigned int);
9734 void vec_mtvscr (vector bool int);
9735 void vec_mtvscr (vector signed short);
9736 void vec_mtvscr (vector unsigned short);
9737 void vec_mtvscr (vector bool short);
9738 void vec_mtvscr (vector pixel);
9739 void vec_mtvscr (vector signed char);
9740 void vec_mtvscr (vector unsigned char);
9741 void vec_mtvscr (vector bool char);
9743 vector unsigned short vec_mule (vector unsigned char,
9744 vector unsigned char);
9745 vector signed short vec_mule (vector signed char,
9746 vector signed char);
9747 vector unsigned int vec_mule (vector unsigned short,
9748 vector unsigned short);
9749 vector signed int vec_mule (vector signed short, vector signed short);
9751 vector signed int vec_vmulesh (vector signed short,
9752 vector signed short);
9754 vector unsigned int vec_vmuleuh (vector unsigned short,
9755 vector unsigned short);
9757 vector signed short vec_vmulesb (vector signed char,
9758 vector signed char);
9760 vector unsigned short vec_vmuleub (vector unsigned char,
9761 vector unsigned char);
9763 vector unsigned short vec_mulo (vector unsigned char,
9764 vector unsigned char);
9765 vector signed short vec_mulo (vector signed char, vector signed char);
9766 vector unsigned int vec_mulo (vector unsigned short,
9767 vector unsigned short);
9768 vector signed int vec_mulo (vector signed short, vector signed short);
9770 vector signed int vec_vmulosh (vector signed short,
9771 vector signed short);
9773 vector unsigned int vec_vmulouh (vector unsigned short,
9774 vector unsigned short);
9776 vector signed short vec_vmulosb (vector signed char,
9777 vector signed char);
9779 vector unsigned short vec_vmuloub (vector unsigned char,
9780 vector unsigned char);
9782 vector float vec_nmsub (vector float, vector float, vector float);
9784 vector float vec_nor (vector float, vector float);
9785 vector signed int vec_nor (vector signed int, vector signed int);
9786 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
9787 vector bool int vec_nor (vector bool int, vector bool int);
9788 vector signed short vec_nor (vector signed short, vector signed short);
9789 vector unsigned short vec_nor (vector unsigned short,
9790 vector unsigned short);
9791 vector bool short vec_nor (vector bool short, vector bool short);
9792 vector signed char vec_nor (vector signed char, vector signed char);
9793 vector unsigned char vec_nor (vector unsigned char,
9794 vector unsigned char);
9795 vector bool char vec_nor (vector bool char, vector bool char);
9797 vector float vec_or (vector float, vector float);
9798 vector float vec_or (vector float, vector bool int);
9799 vector float vec_or (vector bool int, vector float);
9800 vector bool int vec_or (vector bool int, vector bool int);
9801 vector signed int vec_or (vector bool int, vector signed int);
9802 vector signed int vec_or (vector signed int, vector bool int);
9803 vector signed int vec_or (vector signed int, vector signed int);
9804 vector unsigned int vec_or (vector bool int, vector unsigned int);
9805 vector unsigned int vec_or (vector unsigned int, vector bool int);
9806 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
9807 vector bool short vec_or (vector bool short, vector bool short);
9808 vector signed short vec_or (vector bool short, vector signed short);
9809 vector signed short vec_or (vector signed short, vector bool short);
9810 vector signed short vec_or (vector signed short, vector signed short);
9811 vector unsigned short vec_or (vector bool short, vector unsigned short);
9812 vector unsigned short vec_or (vector unsigned short, vector bool short);
9813 vector unsigned short vec_or (vector unsigned short,
9814 vector unsigned short);
9815 vector signed char vec_or (vector bool char, vector signed char);
9816 vector bool char vec_or (vector bool char, vector bool char);
9817 vector signed char vec_or (vector signed char, vector bool char);
9818 vector signed char vec_or (vector signed char, vector signed char);
9819 vector unsigned char vec_or (vector bool char, vector unsigned char);
9820 vector unsigned char vec_or (vector unsigned char, vector bool char);
9821 vector unsigned char vec_or (vector unsigned char,
9822 vector unsigned char);
9824 vector signed char vec_pack (vector signed short, vector signed short);
9825 vector unsigned char vec_pack (vector unsigned short,
9826 vector unsigned short);
9827 vector bool char vec_pack (vector bool short, vector bool short);
9828 vector signed short vec_pack (vector signed int, vector signed int);
9829 vector unsigned short vec_pack (vector unsigned int,
9830 vector unsigned int);
9831 vector bool short vec_pack (vector bool int, vector bool int);
9833 vector bool short vec_vpkuwum (vector bool int, vector bool int);
9834 vector signed short vec_vpkuwum (vector signed int, vector signed int);
9835 vector unsigned short vec_vpkuwum (vector unsigned int,
9836 vector unsigned int);
9838 vector bool char vec_vpkuhum (vector bool short, vector bool short);
9839 vector signed char vec_vpkuhum (vector signed short,
9840 vector signed short);
9841 vector unsigned char vec_vpkuhum (vector unsigned short,
9842 vector unsigned short);
9844 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
9846 vector unsigned char vec_packs (vector unsigned short,
9847 vector unsigned short);
9848 vector signed char vec_packs (vector signed short, vector signed short);
9849 vector unsigned short vec_packs (vector unsigned int,
9850 vector unsigned int);
9851 vector signed short vec_packs (vector signed int, vector signed int);
9853 vector signed short vec_vpkswss (vector signed int, vector signed int);
9855 vector unsigned short vec_vpkuwus (vector unsigned int,
9856 vector unsigned int);
9858 vector signed char vec_vpkshss (vector signed short,
9859 vector signed short);
9861 vector unsigned char vec_vpkuhus (vector unsigned short,
9862 vector unsigned short);
9864 vector unsigned char vec_packsu (vector unsigned short,
9865 vector unsigned short);
9866 vector unsigned char vec_packsu (vector signed short,
9867 vector signed short);
9868 vector unsigned short vec_packsu (vector unsigned int,
9869 vector unsigned int);
9870 vector unsigned short vec_packsu (vector signed int, vector signed int);
9872 vector unsigned short vec_vpkswus (vector signed int,
9875 vector unsigned char vec_vpkshus (vector signed short,
9876 vector signed short);
9878 vector float vec_perm (vector float,
9880 vector unsigned char);
9881 vector signed int vec_perm (vector signed int,
9883 vector unsigned char);
9884 vector unsigned int vec_perm (vector unsigned int,
9885 vector unsigned int,
9886 vector unsigned char);
9887 vector bool int vec_perm (vector bool int,
9889 vector unsigned char);
9890 vector signed short vec_perm (vector signed short,
9891 vector signed short,
9892 vector unsigned char);
9893 vector unsigned short vec_perm (vector unsigned short,
9894 vector unsigned short,
9895 vector unsigned char);
9896 vector bool short vec_perm (vector bool short,
9898 vector unsigned char);
9899 vector pixel vec_perm (vector pixel,
9901 vector unsigned char);
9902 vector signed char vec_perm (vector signed char,
9904 vector unsigned char);
9905 vector unsigned char vec_perm (vector unsigned char,
9906 vector unsigned char,
9907 vector unsigned char);
9908 vector bool char vec_perm (vector bool char,
9910 vector unsigned char);
9912 vector float vec_re (vector float);
9914 vector signed char vec_rl (vector signed char,
9915 vector unsigned char);
9916 vector unsigned char vec_rl (vector unsigned char,
9917 vector unsigned char);
9918 vector signed short vec_rl (vector signed short, vector unsigned short);
9919 vector unsigned short vec_rl (vector unsigned short,
9920 vector unsigned short);
9921 vector signed int vec_rl (vector signed int, vector unsigned int);
9922 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
9924 vector signed int vec_vrlw (vector signed int, vector unsigned int);
9925 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
9927 vector signed short vec_vrlh (vector signed short,
9928 vector unsigned short);
9929 vector unsigned short vec_vrlh (vector unsigned short,
9930 vector unsigned short);
9932 vector signed char vec_vrlb (vector signed char, vector unsigned char);
9933 vector unsigned char vec_vrlb (vector unsigned char,
9934 vector unsigned char);
9936 vector float vec_round (vector float);
9938 vector float vec_rsqrte (vector float);
9940 vector float vec_sel (vector float, vector float, vector bool int);
9941 vector float vec_sel (vector float, vector float, vector unsigned int);
9942 vector signed int vec_sel (vector signed int,
9945 vector signed int vec_sel (vector signed int,
9947 vector unsigned int);
9948 vector unsigned int vec_sel (vector unsigned int,
9949 vector unsigned int,
9951 vector unsigned int vec_sel (vector unsigned int,
9952 vector unsigned int,
9953 vector unsigned int);
9954 vector bool int vec_sel (vector bool int,
9957 vector bool int vec_sel (vector bool int,
9959 vector unsigned int);
9960 vector signed short vec_sel (vector signed short,
9961 vector signed short,
9963 vector signed short vec_sel (vector signed short,
9964 vector signed short,
9965 vector unsigned short);
9966 vector unsigned short vec_sel (vector unsigned short,
9967 vector unsigned short,
9969 vector unsigned short vec_sel (vector unsigned short,
9970 vector unsigned short,
9971 vector unsigned short);
9972 vector bool short vec_sel (vector bool short,
9975 vector bool short vec_sel (vector bool short,
9977 vector unsigned short);
9978 vector signed char vec_sel (vector signed char,
9981 vector signed char vec_sel (vector signed char,
9983 vector unsigned char);
9984 vector unsigned char vec_sel (vector unsigned char,
9985 vector unsigned char,
9987 vector unsigned char vec_sel (vector unsigned char,
9988 vector unsigned char,
9989 vector unsigned char);
9990 vector bool char vec_sel (vector bool char,
9993 vector bool char vec_sel (vector bool char,
9995 vector unsigned char);
9997 vector signed char vec_sl (vector signed char,
9998 vector unsigned char);
9999 vector unsigned char vec_sl (vector unsigned char,
10000 vector unsigned char);
10001 vector signed short vec_sl (vector signed short, vector unsigned short);
10002 vector unsigned short vec_sl (vector unsigned short,
10003 vector unsigned short);
10004 vector signed int vec_sl (vector signed int, vector unsigned int);
10005 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
10007 vector signed int vec_vslw (vector signed int, vector unsigned int);
10008 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
10010 vector signed short vec_vslh (vector signed short,
10011 vector unsigned short);
10012 vector unsigned short vec_vslh (vector unsigned short,
10013 vector unsigned short);
10015 vector signed char vec_vslb (vector signed char, vector unsigned char);
10016 vector unsigned char vec_vslb (vector unsigned char,
10017 vector unsigned char);
10019 vector float vec_sld (vector float, vector float, const int);
10020 vector signed int vec_sld (vector signed int,
10023 vector unsigned int vec_sld (vector unsigned int,
10024 vector unsigned int,
10026 vector bool int vec_sld (vector bool int,
10029 vector signed short vec_sld (vector signed short,
10030 vector signed short,
10032 vector unsigned short vec_sld (vector unsigned short,
10033 vector unsigned short,
10035 vector bool short vec_sld (vector bool short,
10038 vector pixel vec_sld (vector pixel,
10041 vector signed char vec_sld (vector signed char,
10042 vector signed char,
10044 vector unsigned char vec_sld (vector unsigned char,
10045 vector unsigned char,
10047 vector bool char vec_sld (vector bool char,
10051 vector signed int vec_sll (vector signed int,
10052 vector unsigned int);
10053 vector signed int vec_sll (vector signed int,
10054 vector unsigned short);
10055 vector signed int vec_sll (vector signed int,
10056 vector unsigned char);
10057 vector unsigned int vec_sll (vector unsigned int,
10058 vector unsigned int);
10059 vector unsigned int vec_sll (vector unsigned int,
10060 vector unsigned short);
10061 vector unsigned int vec_sll (vector unsigned int,
10062 vector unsigned char);
10063 vector bool int vec_sll (vector bool int,
10064 vector unsigned int);
10065 vector bool int vec_sll (vector bool int,
10066 vector unsigned short);
10067 vector bool int vec_sll (vector bool int,
10068 vector unsigned char);
10069 vector signed short vec_sll (vector signed short,
10070 vector unsigned int);
10071 vector signed short vec_sll (vector signed short,
10072 vector unsigned short);
10073 vector signed short vec_sll (vector signed short,
10074 vector unsigned char);
10075 vector unsigned short vec_sll (vector unsigned short,
10076 vector unsigned int);
10077 vector unsigned short vec_sll (vector unsigned short,
10078 vector unsigned short);
10079 vector unsigned short vec_sll (vector unsigned short,
10080 vector unsigned char);
10081 vector bool short vec_sll (vector bool short, vector unsigned int);
10082 vector bool short vec_sll (vector bool short, vector unsigned short);
10083 vector bool short vec_sll (vector bool short, vector unsigned char);
10084 vector pixel vec_sll (vector pixel, vector unsigned int);
10085 vector pixel vec_sll (vector pixel, vector unsigned short);
10086 vector pixel vec_sll (vector pixel, vector unsigned char);
10087 vector signed char vec_sll (vector signed char, vector unsigned int);
10088 vector signed char vec_sll (vector signed char, vector unsigned short);
10089 vector signed char vec_sll (vector signed char, vector unsigned char);
10090 vector unsigned char vec_sll (vector unsigned char,
10091 vector unsigned int);
10092 vector unsigned char vec_sll (vector unsigned char,
10093 vector unsigned short);
10094 vector unsigned char vec_sll (vector unsigned char,
10095 vector unsigned char);
10096 vector bool char vec_sll (vector bool char, vector unsigned int);
10097 vector bool char vec_sll (vector bool char, vector unsigned short);
10098 vector bool char vec_sll (vector bool char, vector unsigned char);
10100 vector float vec_slo (vector float, vector signed char);
10101 vector float vec_slo (vector float, vector unsigned char);
10102 vector signed int vec_slo (vector signed int, vector signed char);
10103 vector signed int vec_slo (vector signed int, vector unsigned char);
10104 vector unsigned int vec_slo (vector unsigned int, vector signed char);
10105 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
10106 vector signed short vec_slo (vector signed short, vector signed char);
10107 vector signed short vec_slo (vector signed short, vector unsigned char);
10108 vector unsigned short vec_slo (vector unsigned short,
10109 vector signed char);
10110 vector unsigned short vec_slo (vector unsigned short,
10111 vector unsigned char);
10112 vector pixel vec_slo (vector pixel, vector signed char);
10113 vector pixel vec_slo (vector pixel, vector unsigned char);
10114 vector signed char vec_slo (vector signed char, vector signed char);
10115 vector signed char vec_slo (vector signed char, vector unsigned char);
10116 vector unsigned char vec_slo (vector unsigned char, vector signed char);
10117 vector unsigned char vec_slo (vector unsigned char,
10118 vector unsigned char);
10120 vector signed char vec_splat (vector signed char, const int);
10121 vector unsigned char vec_splat (vector unsigned char, const int);
10122 vector bool char vec_splat (vector bool char, const int);
10123 vector signed short vec_splat (vector signed short, const int);
10124 vector unsigned short vec_splat (vector unsigned short, const int);
10125 vector bool short vec_splat (vector bool short, const int);
10126 vector pixel vec_splat (vector pixel, const int);
10127 vector float vec_splat (vector float, const int);
10128 vector signed int vec_splat (vector signed int, const int);
10129 vector unsigned int vec_splat (vector unsigned int, const int);
10130 vector bool int vec_splat (vector bool int, const int);
10132 vector float vec_vspltw (vector float, const int);
10133 vector signed int vec_vspltw (vector signed int, const int);
10134 vector unsigned int vec_vspltw (vector unsigned int, const int);
10135 vector bool int vec_vspltw (vector bool int, const int);
10137 vector bool short vec_vsplth (vector bool short, const int);
10138 vector signed short vec_vsplth (vector signed short, const int);
10139 vector unsigned short vec_vsplth (vector unsigned short, const int);
10140 vector pixel vec_vsplth (vector pixel, const int);
10142 vector signed char vec_vspltb (vector signed char, const int);
10143 vector unsigned char vec_vspltb (vector unsigned char, const int);
10144 vector bool char vec_vspltb (vector bool char, const int);
10146 vector signed char vec_splat_s8 (const int);
10148 vector signed short vec_splat_s16 (const int);
10150 vector signed int vec_splat_s32 (const int);
10152 vector unsigned char vec_splat_u8 (const int);
10154 vector unsigned short vec_splat_u16 (const int);
10156 vector unsigned int vec_splat_u32 (const int);
10158 vector signed char vec_sr (vector signed char, vector unsigned char);
10159 vector unsigned char vec_sr (vector unsigned char,
10160 vector unsigned char);
10161 vector signed short vec_sr (vector signed short,
10162 vector unsigned short);
10163 vector unsigned short vec_sr (vector unsigned short,
10164 vector unsigned short);
10165 vector signed int vec_sr (vector signed int, vector unsigned int);
10166 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
10168 vector signed int vec_vsrw (vector signed int, vector unsigned int);
10169 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
10171 vector signed short vec_vsrh (vector signed short,
10172 vector unsigned short);
10173 vector unsigned short vec_vsrh (vector unsigned short,
10174 vector unsigned short);
10176 vector signed char vec_vsrb (vector signed char, vector unsigned char);
10177 vector unsigned char vec_vsrb (vector unsigned char,
10178 vector unsigned char);
10180 vector signed char vec_sra (vector signed char, vector unsigned char);
10181 vector unsigned char vec_sra (vector unsigned char,
10182 vector unsigned char);
10183 vector signed short vec_sra (vector signed short,
10184 vector unsigned short);
10185 vector unsigned short vec_sra (vector unsigned short,
10186 vector unsigned short);
10187 vector signed int vec_sra (vector signed int, vector unsigned int);
10188 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
10190 vector signed int vec_vsraw (vector signed int, vector unsigned int);
10191 vector unsigned int vec_vsraw (vector unsigned int,
10192 vector unsigned int);
10194 vector signed short vec_vsrah (vector signed short,
10195 vector unsigned short);
10196 vector unsigned short vec_vsrah (vector unsigned short,
10197 vector unsigned short);
10199 vector signed char vec_vsrab (vector signed char, vector unsigned char);
10200 vector unsigned char vec_vsrab (vector unsigned char,
10201 vector unsigned char);
10203 vector signed int vec_srl (vector signed int, vector unsigned int);
10204 vector signed int vec_srl (vector signed int, vector unsigned short);
10205 vector signed int vec_srl (vector signed int, vector unsigned char);
10206 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
10207 vector unsigned int vec_srl (vector unsigned int,
10208 vector unsigned short);
10209 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
10210 vector bool int vec_srl (vector bool int, vector unsigned int);
10211 vector bool int vec_srl (vector bool int, vector unsigned short);
10212 vector bool int vec_srl (vector bool int, vector unsigned char);
10213 vector signed short vec_srl (vector signed short, vector unsigned int);
10214 vector signed short vec_srl (vector signed short,
10215 vector unsigned short);
10216 vector signed short vec_srl (vector signed short, vector unsigned char);
10217 vector unsigned short vec_srl (vector unsigned short,
10218 vector unsigned int);
10219 vector unsigned short vec_srl (vector unsigned short,
10220 vector unsigned short);
10221 vector unsigned short vec_srl (vector unsigned short,
10222 vector unsigned char);
10223 vector bool short vec_srl (vector bool short, vector unsigned int);
10224 vector bool short vec_srl (vector bool short, vector unsigned short);
10225 vector bool short vec_srl (vector bool short, vector unsigned char);
10226 vector pixel vec_srl (vector pixel, vector unsigned int);
10227 vector pixel vec_srl (vector pixel, vector unsigned short);
10228 vector pixel vec_srl (vector pixel, vector unsigned char);
10229 vector signed char vec_srl (vector signed char, vector unsigned int);
10230 vector signed char vec_srl (vector signed char, vector unsigned short);
10231 vector signed char vec_srl (vector signed char, vector unsigned char);
10232 vector unsigned char vec_srl (vector unsigned char,
10233 vector unsigned int);
10234 vector unsigned char vec_srl (vector unsigned char,
10235 vector unsigned short);
10236 vector unsigned char vec_srl (vector unsigned char,
10237 vector unsigned char);
10238 vector bool char vec_srl (vector bool char, vector unsigned int);
10239 vector bool char vec_srl (vector bool char, vector unsigned short);
10240 vector bool char vec_srl (vector bool char, vector unsigned char);
10242 vector float vec_sro (vector float, vector signed char);
10243 vector float vec_sro (vector float, vector unsigned char);
10244 vector signed int vec_sro (vector signed int, vector signed char);
10245 vector signed int vec_sro (vector signed int, vector unsigned char);
10246 vector unsigned int vec_sro (vector unsigned int, vector signed char);
10247 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
10248 vector signed short vec_sro (vector signed short, vector signed char);
10249 vector signed short vec_sro (vector signed short, vector unsigned char);
10250 vector unsigned short vec_sro (vector unsigned short,
10251 vector signed char);
10252 vector unsigned short vec_sro (vector unsigned short,
10253 vector unsigned char);
10254 vector pixel vec_sro (vector pixel, vector signed char);
10255 vector pixel vec_sro (vector pixel, vector unsigned char);
10256 vector signed char vec_sro (vector signed char, vector signed char);
10257 vector signed char vec_sro (vector signed char, vector unsigned char);
10258 vector unsigned char vec_sro (vector unsigned char, vector signed char);
10259 vector unsigned char vec_sro (vector unsigned char,
10260 vector unsigned char);
10262 void vec_st (vector float, int, vector float *);
10263 void vec_st (vector float, int, float *);
10264 void vec_st (vector signed int, int, vector signed int *);
10265 void vec_st (vector signed int, int, int *);
10266 void vec_st (vector unsigned int, int, vector unsigned int *);
10267 void vec_st (vector unsigned int, int, unsigned int *);
10268 void vec_st (vector bool int, int, vector bool int *);
10269 void vec_st (vector bool int, int, unsigned int *);
10270 void vec_st (vector bool int, int, int *);
10271 void vec_st (vector signed short, int, vector signed short *);
10272 void vec_st (vector signed short, int, short *);
10273 void vec_st (vector unsigned short, int, vector unsigned short *);
10274 void vec_st (vector unsigned short, int, unsigned short *);
10275 void vec_st (vector bool short, int, vector bool short *);
10276 void vec_st (vector bool short, int, unsigned short *);
10277 void vec_st (vector pixel, int, vector pixel *);
10278 void vec_st (vector pixel, int, unsigned short *);
10279 void vec_st (vector pixel, int, short *);
10280 void vec_st (vector bool short, int, short *);
10281 void vec_st (vector signed char, int, vector signed char *);
10282 void vec_st (vector signed char, int, signed char *);
10283 void vec_st (vector unsigned char, int, vector unsigned char *);
10284 void vec_st (vector unsigned char, int, unsigned char *);
10285 void vec_st (vector bool char, int, vector bool char *);
10286 void vec_st (vector bool char, int, unsigned char *);
10287 void vec_st (vector bool char, int, signed char *);
10289 void vec_ste (vector signed char, int, signed char *);
10290 void vec_ste (vector unsigned char, int, unsigned char *);
10291 void vec_ste (vector bool char, int, signed char *);
10292 void vec_ste (vector bool char, int, unsigned char *);
10293 void vec_ste (vector signed short, int, short *);
10294 void vec_ste (vector unsigned short, int, unsigned short *);
10295 void vec_ste (vector bool short, int, short *);
10296 void vec_ste (vector bool short, int, unsigned short *);
10297 void vec_ste (vector pixel, int, short *);
10298 void vec_ste (vector pixel, int, unsigned short *);
10299 void vec_ste (vector float, int, float *);
10300 void vec_ste (vector signed int, int, int *);
10301 void vec_ste (vector unsigned int, int, unsigned int *);
10302 void vec_ste (vector bool int, int, int *);
10303 void vec_ste (vector bool int, int, unsigned int *);
10305 void vec_stvewx (vector float, int, float *);
10306 void vec_stvewx (vector signed int, int, int *);
10307 void vec_stvewx (vector unsigned int, int, unsigned int *);
10308 void vec_stvewx (vector bool int, int, int *);
10309 void vec_stvewx (vector bool int, int, unsigned int *);
10311 void vec_stvehx (vector signed short, int, short *);
10312 void vec_stvehx (vector unsigned short, int, unsigned short *);
10313 void vec_stvehx (vector bool short, int, short *);
10314 void vec_stvehx (vector bool short, int, unsigned short *);
10315 void vec_stvehx (vector pixel, int, short *);
10316 void vec_stvehx (vector pixel, int, unsigned short *);
10318 void vec_stvebx (vector signed char, int, signed char *);
10319 void vec_stvebx (vector unsigned char, int, unsigned char *);
10320 void vec_stvebx (vector bool char, int, signed char *);
10321 void vec_stvebx (vector bool char, int, unsigned char *);
10323 void vec_stl (vector float, int, vector float *);
10324 void vec_stl (vector float, int, float *);
10325 void vec_stl (vector signed int, int, vector signed int *);
10326 void vec_stl (vector signed int, int, int *);
10327 void vec_stl (vector unsigned int, int, vector unsigned int *);
10328 void vec_stl (vector unsigned int, int, unsigned int *);
10329 void vec_stl (vector bool int, int, vector bool int *);
10330 void vec_stl (vector bool int, int, unsigned int *);
10331 void vec_stl (vector bool int, int, int *);
10332 void vec_stl (vector signed short, int, vector signed short *);
10333 void vec_stl (vector signed short, int, short *);
10334 void vec_stl (vector unsigned short, int, vector unsigned short *);
10335 void vec_stl (vector unsigned short, int, unsigned short *);
10336 void vec_stl (vector bool short, int, vector bool short *);
10337 void vec_stl (vector bool short, int, unsigned short *);
10338 void vec_stl (vector bool short, int, short *);
10339 void vec_stl (vector pixel, int, vector pixel *);
10340 void vec_stl (vector pixel, int, unsigned short *);
10341 void vec_stl (vector pixel, int, short *);
10342 void vec_stl (vector signed char, int, vector signed char *);
10343 void vec_stl (vector signed char, int, signed char *);
10344 void vec_stl (vector unsigned char, int, vector unsigned char *);
10345 void vec_stl (vector unsigned char, int, unsigned char *);
10346 void vec_stl (vector bool char, int, vector bool char *);
10347 void vec_stl (vector bool char, int, unsigned char *);
10348 void vec_stl (vector bool char, int, signed char *);
10350 vector signed char vec_sub (vector bool char, vector signed char);
10351 vector signed char vec_sub (vector signed char, vector bool char);
10352 vector signed char vec_sub (vector signed char, vector signed char);
10353 vector unsigned char vec_sub (vector bool char, vector unsigned char);
10354 vector unsigned char vec_sub (vector unsigned char, vector bool char);
10355 vector unsigned char vec_sub (vector unsigned char,
10356 vector unsigned char);
10357 vector signed short vec_sub (vector bool short, vector signed short);
10358 vector signed short vec_sub (vector signed short, vector bool short);
10359 vector signed short vec_sub (vector signed short, vector signed short);
10360 vector unsigned short vec_sub (vector bool short,
10361 vector unsigned short);
10362 vector unsigned short vec_sub (vector unsigned short,
10363 vector bool short);
10364 vector unsigned short vec_sub (vector unsigned short,
10365 vector unsigned short);
10366 vector signed int vec_sub (vector bool int, vector signed int);
10367 vector signed int vec_sub (vector signed int, vector bool int);
10368 vector signed int vec_sub (vector signed int, vector signed int);
10369 vector unsigned int vec_sub (vector bool int, vector unsigned int);
10370 vector unsigned int vec_sub (vector unsigned int, vector bool int);
10371 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
10372 vector float vec_sub (vector float, vector float);
10374 vector float vec_vsubfp (vector float, vector float);
10376 vector signed int vec_vsubuwm (vector bool int, vector signed int);
10377 vector signed int vec_vsubuwm (vector signed int, vector bool int);
10378 vector signed int vec_vsubuwm (vector signed int, vector signed int);
10379 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
10380 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
10381 vector unsigned int vec_vsubuwm (vector unsigned int,
10382 vector unsigned int);
10384 vector signed short vec_vsubuhm (vector bool short,
10385 vector signed short);
10386 vector signed short vec_vsubuhm (vector signed short,
10387 vector bool short);
10388 vector signed short vec_vsubuhm (vector signed short,
10389 vector signed short);
10390 vector unsigned short vec_vsubuhm (vector bool short,
10391 vector unsigned short);
10392 vector unsigned short vec_vsubuhm (vector unsigned short,
10393 vector bool short);
10394 vector unsigned short vec_vsubuhm (vector unsigned short,
10395 vector unsigned short);
10397 vector signed char vec_vsububm (vector bool char, vector signed char);
10398 vector signed char vec_vsububm (vector signed char, vector bool char);
10399 vector signed char vec_vsububm (vector signed char, vector signed char);
10400 vector unsigned char vec_vsububm (vector bool char,
10401 vector unsigned char);
10402 vector unsigned char vec_vsububm (vector unsigned char,
10404 vector unsigned char vec_vsububm (vector unsigned char,
10405 vector unsigned char);
10407 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
10409 vector unsigned char vec_subs (vector bool char, vector unsigned char);
10410 vector unsigned char vec_subs (vector unsigned char, vector bool char);
10411 vector unsigned char vec_subs (vector unsigned char,
10412 vector unsigned char);
10413 vector signed char vec_subs (vector bool char, vector signed char);
10414 vector signed char vec_subs (vector signed char, vector bool char);
10415 vector signed char vec_subs (vector signed char, vector signed char);
10416 vector unsigned short vec_subs (vector bool short,
10417 vector unsigned short);
10418 vector unsigned short vec_subs (vector unsigned short,
10419 vector bool short);
10420 vector unsigned short vec_subs (vector unsigned short,
10421 vector unsigned short);
10422 vector signed short vec_subs (vector bool short, vector signed short);
10423 vector signed short vec_subs (vector signed short, vector bool short);
10424 vector signed short vec_subs (vector signed short, vector signed short);
10425 vector unsigned int vec_subs (vector bool int, vector unsigned int);
10426 vector unsigned int vec_subs (vector unsigned int, vector bool int);
10427 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
10428 vector signed int vec_subs (vector bool int, vector signed int);
10429 vector signed int vec_subs (vector signed int, vector bool int);
10430 vector signed int vec_subs (vector signed int, vector signed int);
10432 vector signed int vec_vsubsws (vector bool int, vector signed int);
10433 vector signed int vec_vsubsws (vector signed int, vector bool int);
10434 vector signed int vec_vsubsws (vector signed int, vector signed int);
10436 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
10437 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
10438 vector unsigned int vec_vsubuws (vector unsigned int,
10439 vector unsigned int);
10441 vector signed short vec_vsubshs (vector bool short,
10442 vector signed short);
10443 vector signed short vec_vsubshs (vector signed short,
10444 vector bool short);
10445 vector signed short vec_vsubshs (vector signed short,
10446 vector signed short);
10448 vector unsigned short vec_vsubuhs (vector bool short,
10449 vector unsigned short);
10450 vector unsigned short vec_vsubuhs (vector unsigned short,
10451 vector bool short);
10452 vector unsigned short vec_vsubuhs (vector unsigned short,
10453 vector unsigned short);
10455 vector signed char vec_vsubsbs (vector bool char, vector signed char);
10456 vector signed char vec_vsubsbs (vector signed char, vector bool char);
10457 vector signed char vec_vsubsbs (vector signed char, vector signed char);
10459 vector unsigned char vec_vsububs (vector bool char,
10460 vector unsigned char);
10461 vector unsigned char vec_vsububs (vector unsigned char,
10463 vector unsigned char vec_vsububs (vector unsigned char,
10464 vector unsigned char);
10466 vector unsigned int vec_sum4s (vector unsigned char,
10467 vector unsigned int);
10468 vector signed int vec_sum4s (vector signed char, vector signed int);
10469 vector signed int vec_sum4s (vector signed short, vector signed int);
10471 vector signed int vec_vsum4shs (vector signed short, vector signed int);
10473 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
10475 vector unsigned int vec_vsum4ubs (vector unsigned char,
10476 vector unsigned int);
10478 vector signed int vec_sum2s (vector signed int, vector signed int);
10480 vector signed int vec_sums (vector signed int, vector signed int);
10482 vector float vec_trunc (vector float);
10484 vector signed short vec_unpackh (vector signed char);
10485 vector bool short vec_unpackh (vector bool char);
10486 vector signed int vec_unpackh (vector signed short);
10487 vector bool int vec_unpackh (vector bool short);
10488 vector unsigned int vec_unpackh (vector pixel);
10490 vector bool int vec_vupkhsh (vector bool short);
10491 vector signed int vec_vupkhsh (vector signed short);
10493 vector unsigned int vec_vupkhpx (vector pixel);
10495 vector bool short vec_vupkhsb (vector bool char);
10496 vector signed short vec_vupkhsb (vector signed char);
10498 vector signed short vec_unpackl (vector signed char);
10499 vector bool short vec_unpackl (vector bool char);
10500 vector unsigned int vec_unpackl (vector pixel);
10501 vector signed int vec_unpackl (vector signed short);
10502 vector bool int vec_unpackl (vector bool short);
10504 vector unsigned int vec_vupklpx (vector pixel);
10506 vector bool int vec_vupklsh (vector bool short);
10507 vector signed int vec_vupklsh (vector signed short);
10509 vector bool short vec_vupklsb (vector bool char);
10510 vector signed short vec_vupklsb (vector signed char);
10512 vector float vec_xor (vector float, vector float);
10513 vector float vec_xor (vector float, vector bool int);
10514 vector float vec_xor (vector bool int, vector float);
10515 vector bool int vec_xor (vector bool int, vector bool int);
10516 vector signed int vec_xor (vector bool int, vector signed int);
10517 vector signed int vec_xor (vector signed int, vector bool int);
10518 vector signed int vec_xor (vector signed int, vector signed int);
10519 vector unsigned int vec_xor (vector bool int, vector unsigned int);
10520 vector unsigned int vec_xor (vector unsigned int, vector bool int);
10521 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
10522 vector bool short vec_xor (vector bool short, vector bool short);
10523 vector signed short vec_xor (vector bool short, vector signed short);
10524 vector signed short vec_xor (vector signed short, vector bool short);
10525 vector signed short vec_xor (vector signed short, vector signed short);
10526 vector unsigned short vec_xor (vector bool short,
10527 vector unsigned short);
10528 vector unsigned short vec_xor (vector unsigned short,
10529 vector bool short);
10530 vector unsigned short vec_xor (vector unsigned short,
10531 vector unsigned short);
10532 vector signed char vec_xor (vector bool char, vector signed char);
10533 vector bool char vec_xor (vector bool char, vector bool char);
10534 vector signed char vec_xor (vector signed char, vector bool char);
10535 vector signed char vec_xor (vector signed char, vector signed char);
10536 vector unsigned char vec_xor (vector bool char, vector unsigned char);
10537 vector unsigned char vec_xor (vector unsigned char, vector bool char);
10538 vector unsigned char vec_xor (vector unsigned char,
10539 vector unsigned char);
10541 int vec_all_eq (vector signed char, vector bool char);
10542 int vec_all_eq (vector signed char, vector signed char);
10543 int vec_all_eq (vector unsigned char, vector bool char);
10544 int vec_all_eq (vector unsigned char, vector unsigned char);
10545 int vec_all_eq (vector bool char, vector bool char);
10546 int vec_all_eq (vector bool char, vector unsigned char);
10547 int vec_all_eq (vector bool char, vector signed char);
10548 int vec_all_eq (vector signed short, vector bool short);
10549 int vec_all_eq (vector signed short, vector signed short);
10550 int vec_all_eq (vector unsigned short, vector bool short);
10551 int vec_all_eq (vector unsigned short, vector unsigned short);
10552 int vec_all_eq (vector bool short, vector bool short);
10553 int vec_all_eq (vector bool short, vector unsigned short);
10554 int vec_all_eq (vector bool short, vector signed short);
10555 int vec_all_eq (vector pixel, vector pixel);
10556 int vec_all_eq (vector signed int, vector bool int);
10557 int vec_all_eq (vector signed int, vector signed int);
10558 int vec_all_eq (vector unsigned int, vector bool int);
10559 int vec_all_eq (vector unsigned int, vector unsigned int);
10560 int vec_all_eq (vector bool int, vector bool int);
10561 int vec_all_eq (vector bool int, vector unsigned int);
10562 int vec_all_eq (vector bool int, vector signed int);
10563 int vec_all_eq (vector float, vector float);
10565 int vec_all_ge (vector bool char, vector unsigned char);
10566 int vec_all_ge (vector unsigned char, vector bool char);
10567 int vec_all_ge (vector unsigned char, vector unsigned char);
10568 int vec_all_ge (vector bool char, vector signed char);
10569 int vec_all_ge (vector signed char, vector bool char);
10570 int vec_all_ge (vector signed char, vector signed char);
10571 int vec_all_ge (vector bool short, vector unsigned short);
10572 int vec_all_ge (vector unsigned short, vector bool short);
10573 int vec_all_ge (vector unsigned short, vector unsigned short);
10574 int vec_all_ge (vector signed short, vector signed short);
10575 int vec_all_ge (vector bool short, vector signed short);
10576 int vec_all_ge (vector signed short, vector bool short);
10577 int vec_all_ge (vector bool int, vector unsigned int);
10578 int vec_all_ge (vector unsigned int, vector bool int);
10579 int vec_all_ge (vector unsigned int, vector unsigned int);
10580 int vec_all_ge (vector bool int, vector signed int);
10581 int vec_all_ge (vector signed int, vector bool int);
10582 int vec_all_ge (vector signed int, vector signed int);
10583 int vec_all_ge (vector float, vector float);
10585 int vec_all_gt (vector bool char, vector unsigned char);
10586 int vec_all_gt (vector unsigned char, vector bool char);
10587 int vec_all_gt (vector unsigned char, vector unsigned char);
10588 int vec_all_gt (vector bool char, vector signed char);
10589 int vec_all_gt (vector signed char, vector bool char);
10590 int vec_all_gt (vector signed char, vector signed char);
10591 int vec_all_gt (vector bool short, vector unsigned short);
10592 int vec_all_gt (vector unsigned short, vector bool short);
10593 int vec_all_gt (vector unsigned short, vector unsigned short);
10594 int vec_all_gt (vector bool short, vector signed short);
10595 int vec_all_gt (vector signed short, vector bool short);
10596 int vec_all_gt (vector signed short, vector signed short);
10597 int vec_all_gt (vector bool int, vector unsigned int);
10598 int vec_all_gt (vector unsigned int, vector bool int);
10599 int vec_all_gt (vector unsigned int, vector unsigned int);
10600 int vec_all_gt (vector bool int, vector signed int);
10601 int vec_all_gt (vector signed int, vector bool int);
10602 int vec_all_gt (vector signed int, vector signed int);
10603 int vec_all_gt (vector float, vector float);
10605 int vec_all_in (vector float, vector float);
10607 int vec_all_le (vector bool char, vector unsigned char);
10608 int vec_all_le (vector unsigned char, vector bool char);
10609 int vec_all_le (vector unsigned char, vector unsigned char);
10610 int vec_all_le (vector bool char, vector signed char);
10611 int vec_all_le (vector signed char, vector bool char);
10612 int vec_all_le (vector signed char, vector signed char);
10613 int vec_all_le (vector bool short, vector unsigned short);
10614 int vec_all_le (vector unsigned short, vector bool short);
10615 int vec_all_le (vector unsigned short, vector unsigned short);
10616 int vec_all_le (vector bool short, vector signed short);
10617 int vec_all_le (vector signed short, vector bool short);
10618 int vec_all_le (vector signed short, vector signed short);
10619 int vec_all_le (vector bool int, vector unsigned int);
10620 int vec_all_le (vector unsigned int, vector bool int);
10621 int vec_all_le (vector unsigned int, vector unsigned int);
10622 int vec_all_le (vector bool int, vector signed int);
10623 int vec_all_le (vector signed int, vector bool int);
10624 int vec_all_le (vector signed int, vector signed int);
10625 int vec_all_le (vector float, vector float);
10627 int vec_all_lt (vector bool char, vector unsigned char);
10628 int vec_all_lt (vector unsigned char, vector bool char);
10629 int vec_all_lt (vector unsigned char, vector unsigned char);
10630 int vec_all_lt (vector bool char, vector signed char);
10631 int vec_all_lt (vector signed char, vector bool char);
10632 int vec_all_lt (vector signed char, vector signed char);
10633 int vec_all_lt (vector bool short, vector unsigned short);
10634 int vec_all_lt (vector unsigned short, vector bool short);
10635 int vec_all_lt (vector unsigned short, vector unsigned short);
10636 int vec_all_lt (vector bool short, vector signed short);
10637 int vec_all_lt (vector signed short, vector bool short);
10638 int vec_all_lt (vector signed short, vector signed short);
10639 int vec_all_lt (vector bool int, vector unsigned int);
10640 int vec_all_lt (vector unsigned int, vector bool int);
10641 int vec_all_lt (vector unsigned int, vector unsigned int);
10642 int vec_all_lt (vector bool int, vector signed int);
10643 int vec_all_lt (vector signed int, vector bool int);
10644 int vec_all_lt (vector signed int, vector signed int);
10645 int vec_all_lt (vector float, vector float);
10647 int vec_all_nan (vector float);
10649 int vec_all_ne (vector signed char, vector bool char);
10650 int vec_all_ne (vector signed char, vector signed char);
10651 int vec_all_ne (vector unsigned char, vector bool char);
10652 int vec_all_ne (vector unsigned char, vector unsigned char);
10653 int vec_all_ne (vector bool char, vector bool char);
10654 int vec_all_ne (vector bool char, vector unsigned char);
10655 int vec_all_ne (vector bool char, vector signed char);
10656 int vec_all_ne (vector signed short, vector bool short);
10657 int vec_all_ne (vector signed short, vector signed short);
10658 int vec_all_ne (vector unsigned short, vector bool short);
10659 int vec_all_ne (vector unsigned short, vector unsigned short);
10660 int vec_all_ne (vector bool short, vector bool short);
10661 int vec_all_ne (vector bool short, vector unsigned short);
10662 int vec_all_ne (vector bool short, vector signed short);
10663 int vec_all_ne (vector pixel, vector pixel);
10664 int vec_all_ne (vector signed int, vector bool int);
10665 int vec_all_ne (vector signed int, vector signed int);
10666 int vec_all_ne (vector unsigned int, vector bool int);
10667 int vec_all_ne (vector unsigned int, vector unsigned int);
10668 int vec_all_ne (vector bool int, vector bool int);
10669 int vec_all_ne (vector bool int, vector unsigned int);
10670 int vec_all_ne (vector bool int, vector signed int);
10671 int vec_all_ne (vector float, vector float);
10673 int vec_all_nge (vector float, vector float);
10675 int vec_all_ngt (vector float, vector float);
10677 int vec_all_nle (vector float, vector float);
10679 int vec_all_nlt (vector float, vector float);
10681 int vec_all_numeric (vector float);
10683 int vec_any_eq (vector signed char, vector bool char);
10684 int vec_any_eq (vector signed char, vector signed char);
10685 int vec_any_eq (vector unsigned char, vector bool char);
10686 int vec_any_eq (vector unsigned char, vector unsigned char);
10687 int vec_any_eq (vector bool char, vector bool char);
10688 int vec_any_eq (vector bool char, vector unsigned char);
10689 int vec_any_eq (vector bool char, vector signed char);
10690 int vec_any_eq (vector signed short, vector bool short);
10691 int vec_any_eq (vector signed short, vector signed short);
10692 int vec_any_eq (vector unsigned short, vector bool short);
10693 int vec_any_eq (vector unsigned short, vector unsigned short);
10694 int vec_any_eq (vector bool short, vector bool short);
10695 int vec_any_eq (vector bool short, vector unsigned short);
10696 int vec_any_eq (vector bool short, vector signed short);
10697 int vec_any_eq (vector pixel, vector pixel);
10698 int vec_any_eq (vector signed int, vector bool int);
10699 int vec_any_eq (vector signed int, vector signed int);
10700 int vec_any_eq (vector unsigned int, vector bool int);
10701 int vec_any_eq (vector unsigned int, vector unsigned int);
10702 int vec_any_eq (vector bool int, vector bool int);
10703 int vec_any_eq (vector bool int, vector unsigned int);
10704 int vec_any_eq (vector bool int, vector signed int);
10705 int vec_any_eq (vector float, vector float);
10707 int vec_any_ge (vector signed char, vector bool char);
10708 int vec_any_ge (vector unsigned char, vector bool char);
10709 int vec_any_ge (vector unsigned char, vector unsigned char);
10710 int vec_any_ge (vector signed char, vector signed char);
10711 int vec_any_ge (vector bool char, vector unsigned char);
10712 int vec_any_ge (vector bool char, vector signed char);
10713 int vec_any_ge (vector unsigned short, vector bool short);
10714 int vec_any_ge (vector unsigned short, vector unsigned short);
10715 int vec_any_ge (vector signed short, vector signed short);
10716 int vec_any_ge (vector signed short, vector bool short);
10717 int vec_any_ge (vector bool short, vector unsigned short);
10718 int vec_any_ge (vector bool short, vector signed short);
10719 int vec_any_ge (vector signed int, vector bool int);
10720 int vec_any_ge (vector unsigned int, vector bool int);
10721 int vec_any_ge (vector unsigned int, vector unsigned int);
10722 int vec_any_ge (vector signed int, vector signed int);
10723 int vec_any_ge (vector bool int, vector unsigned int);
10724 int vec_any_ge (vector bool int, vector signed int);
10725 int vec_any_ge (vector float, vector float);
10727 int vec_any_gt (vector bool char, vector unsigned char);
10728 int vec_any_gt (vector unsigned char, vector bool char);
10729 int vec_any_gt (vector unsigned char, vector unsigned char);
10730 int vec_any_gt (vector bool char, vector signed char);
10731 int vec_any_gt (vector signed char, vector bool char);
10732 int vec_any_gt (vector signed char, vector signed char);
10733 int vec_any_gt (vector bool short, vector unsigned short);
10734 int vec_any_gt (vector unsigned short, vector bool short);
10735 int vec_any_gt (vector unsigned short, vector unsigned short);
10736 int vec_any_gt (vector bool short, vector signed short);
10737 int vec_any_gt (vector signed short, vector bool short);
10738 int vec_any_gt (vector signed short, vector signed short);
10739 int vec_any_gt (vector bool int, vector unsigned int);
10740 int vec_any_gt (vector unsigned int, vector bool int);
10741 int vec_any_gt (vector unsigned int, vector unsigned int);
10742 int vec_any_gt (vector bool int, vector signed int);
10743 int vec_any_gt (vector signed int, vector bool int);
10744 int vec_any_gt (vector signed int, vector signed int);
10745 int vec_any_gt (vector float, vector float);
10747 int vec_any_le (vector bool char, vector unsigned char);
10748 int vec_any_le (vector unsigned char, vector bool char);
10749 int vec_any_le (vector unsigned char, vector unsigned char);
10750 int vec_any_le (vector bool char, vector signed char);
10751 int vec_any_le (vector signed char, vector bool char);
10752 int vec_any_le (vector signed char, vector signed char);
10753 int vec_any_le (vector bool short, vector unsigned short);
10754 int vec_any_le (vector unsigned short, vector bool short);
10755 int vec_any_le (vector unsigned short, vector unsigned short);
10756 int vec_any_le (vector bool short, vector signed short);
10757 int vec_any_le (vector signed short, vector bool short);
10758 int vec_any_le (vector signed short, vector signed short);
10759 int vec_any_le (vector bool int, vector unsigned int);
10760 int vec_any_le (vector unsigned int, vector bool int);
10761 int vec_any_le (vector unsigned int, vector unsigned int);
10762 int vec_any_le (vector bool int, vector signed int);
10763 int vec_any_le (vector signed int, vector bool int);
10764 int vec_any_le (vector signed int, vector signed int);
10765 int vec_any_le (vector float, vector float);
10767 int vec_any_lt (vector bool char, vector unsigned char);
10768 int vec_any_lt (vector unsigned char, vector bool char);
10769 int vec_any_lt (vector unsigned char, vector unsigned char);
10770 int vec_any_lt (vector bool char, vector signed char);
10771 int vec_any_lt (vector signed char, vector bool char);
10772 int vec_any_lt (vector signed char, vector signed char);
10773 int vec_any_lt (vector bool short, vector unsigned short);
10774 int vec_any_lt (vector unsigned short, vector bool short);
10775 int vec_any_lt (vector unsigned short, vector unsigned short);
10776 int vec_any_lt (vector bool short, vector signed short);
10777 int vec_any_lt (vector signed short, vector bool short);
10778 int vec_any_lt (vector signed short, vector signed short);
10779 int vec_any_lt (vector bool int, vector unsigned int);
10780 int vec_any_lt (vector unsigned int, vector bool int);
10781 int vec_any_lt (vector unsigned int, vector unsigned int);
10782 int vec_any_lt (vector bool int, vector signed int);
10783 int vec_any_lt (vector signed int, vector bool int);
10784 int vec_any_lt (vector signed int, vector signed int);
10785 int vec_any_lt (vector float, vector float);
10787 int vec_any_nan (vector float);
10789 int vec_any_ne (vector signed char, vector bool char);
10790 int vec_any_ne (vector signed char, vector signed char);
10791 int vec_any_ne (vector unsigned char, vector bool char);
10792 int vec_any_ne (vector unsigned char, vector unsigned char);
10793 int vec_any_ne (vector bool char, vector bool char);
10794 int vec_any_ne (vector bool char, vector unsigned char);
10795 int vec_any_ne (vector bool char, vector signed char);
10796 int vec_any_ne (vector signed short, vector bool short);
10797 int vec_any_ne (vector signed short, vector signed short);
10798 int vec_any_ne (vector unsigned short, vector bool short);
10799 int vec_any_ne (vector unsigned short, vector unsigned short);
10800 int vec_any_ne (vector bool short, vector bool short);
10801 int vec_any_ne (vector bool short, vector unsigned short);
10802 int vec_any_ne (vector bool short, vector signed short);
10803 int vec_any_ne (vector pixel, vector pixel);
10804 int vec_any_ne (vector signed int, vector bool int);
10805 int vec_any_ne (vector signed int, vector signed int);
10806 int vec_any_ne (vector unsigned int, vector bool int);
10807 int vec_any_ne (vector unsigned int, vector unsigned int);
10808 int vec_any_ne (vector bool int, vector bool int);
10809 int vec_any_ne (vector bool int, vector unsigned int);
10810 int vec_any_ne (vector bool int, vector signed int);
10811 int vec_any_ne (vector float, vector float);
10813 int vec_any_nge (vector float, vector float);
10815 int vec_any_ngt (vector float, vector float);
10817 int vec_any_nle (vector float, vector float);
10819 int vec_any_nlt (vector float, vector float);
10821 int vec_any_numeric (vector float);
10823 int vec_any_out (vector float, vector float);
10826 @node SPARC VIS Built-in Functions
10827 @subsection SPARC VIS Built-in Functions
10829 GCC supports SIMD operations on the SPARC using both the generic vector
10830 extensions (@pxref{Vector Extensions}) as well as built-in functions for
10831 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
10832 switch, the VIS extension is exposed as the following built-in functions:
10835 typedef int v2si __attribute__ ((vector_size (8)));
10836 typedef short v4hi __attribute__ ((vector_size (8)));
10837 typedef short v2hi __attribute__ ((vector_size (4)));
10838 typedef char v8qi __attribute__ ((vector_size (8)));
10839 typedef char v4qi __attribute__ ((vector_size (4)));
10841 void * __builtin_vis_alignaddr (void *, long);
10842 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
10843 v2si __builtin_vis_faligndatav2si (v2si, v2si);
10844 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
10845 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
10847 v4hi __builtin_vis_fexpand (v4qi);
10849 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
10850 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
10851 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
10852 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
10853 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
10854 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
10855 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
10857 v4qi __builtin_vis_fpack16 (v4hi);
10858 v8qi __builtin_vis_fpack32 (v2si, v2si);
10859 v2hi __builtin_vis_fpackfix (v2si);
10860 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
10862 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
10865 @node SPU Built-in Functions
10866 @subsection SPU Built-in Functions
10868 GCC provides extensions for the SPU processor as described in the
10869 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
10870 found at @uref{http://cell.scei.co.jp/} or
10871 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
10872 implementation differs in several ways.
10877 The optional extension of specifying vector constants in parentheses is
10881 A vector initializer requires no cast if the vector constant is of the
10882 same type as the variable it is initializing.
10885 If @code{signed} or @code{unsigned} is omitted, the signedness of the
10886 vector type is the default signedness of the base type. The default
10887 varies depending on the operating system, so a portable program should
10888 always specify the signedness.
10891 By default, the keyword @code{__vector} is added. The macro
10892 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
10896 GCC allows using a @code{typedef} name as the type specifier for a
10900 For C, overloaded functions are implemented with macros so the following
10904 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
10907 Since @code{spu_add} is a macro, the vector constant in the example
10908 is treated as four separate arguments. Wrap the entire argument in
10909 parentheses for this to work.
10912 The extended version of @code{__builtin_expect} is not supported.
10916 @emph{Note:} Only the interface described in the aforementioned
10917 specification is supported. Internally, GCC uses built-in functions to
10918 implement the required functionality, but these are not supported and
10919 are subject to change without notice.
10921 @node Target Format Checks
10922 @section Format Checks Specific to Particular Target Machines
10924 For some target machines, GCC supports additional options to the
10926 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
10929 * Solaris Format Checks::
10932 @node Solaris Format Checks
10933 @subsection Solaris Format Checks
10935 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
10936 check. @code{cmn_err} accepts a subset of the standard @code{printf}
10937 conversions, and the two-argument @code{%b} conversion for displaying
10938 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
10941 @section Pragmas Accepted by GCC
10945 GCC supports several types of pragmas, primarily in order to compile
10946 code originally written for other compilers. Note that in general
10947 we do not recommend the use of pragmas; @xref{Function Attributes},
10948 for further explanation.
10953 * RS/6000 and PowerPC Pragmas::
10955 * Solaris Pragmas::
10956 * Symbol-Renaming Pragmas::
10957 * Structure-Packing Pragmas::
10959 * Diagnostic Pragmas::
10960 * Visibility Pragmas::
10961 * Push/Pop Macro Pragmas::
10965 @subsection ARM Pragmas
10967 The ARM target defines pragmas for controlling the default addition of
10968 @code{long_call} and @code{short_call} attributes to functions.
10969 @xref{Function Attributes}, for information about the effects of these
10974 @cindex pragma, long_calls
10975 Set all subsequent functions to have the @code{long_call} attribute.
10977 @item no_long_calls
10978 @cindex pragma, no_long_calls
10979 Set all subsequent functions to have the @code{short_call} attribute.
10981 @item long_calls_off
10982 @cindex pragma, long_calls_off
10983 Do not affect the @code{long_call} or @code{short_call} attributes of
10984 subsequent functions.
10988 @subsection M32C Pragmas
10991 @item memregs @var{number}
10992 @cindex pragma, memregs
10993 Overrides the command line option @code{-memregs=} for the current
10994 file. Use with care! This pragma must be before any function in the
10995 file, and mixing different memregs values in different objects may
10996 make them incompatible. This pragma is useful when a
10997 performance-critical function uses a memreg for temporary values,
10998 as it may allow you to reduce the number of memregs used.
11002 @node RS/6000 and PowerPC Pragmas
11003 @subsection RS/6000 and PowerPC Pragmas
11005 The RS/6000 and PowerPC targets define one pragma for controlling
11006 whether or not the @code{longcall} attribute is added to function
11007 declarations by default. This pragma overrides the @option{-mlongcall}
11008 option, but not the @code{longcall} and @code{shortcall} attributes.
11009 @xref{RS/6000 and PowerPC Options}, for more information about when long
11010 calls are and are not necessary.
11014 @cindex pragma, longcall
11015 Apply the @code{longcall} attribute to all subsequent function
11019 Do not apply the @code{longcall} attribute to subsequent function
11023 @c Describe h8300 pragmas here.
11024 @c Describe sh pragmas here.
11025 @c Describe v850 pragmas here.
11027 @node Darwin Pragmas
11028 @subsection Darwin Pragmas
11030 The following pragmas are available for all architectures running the
11031 Darwin operating system. These are useful for compatibility with other
11035 @item mark @var{tokens}@dots{}
11036 @cindex pragma, mark
11037 This pragma is accepted, but has no effect.
11039 @item options align=@var{alignment}
11040 @cindex pragma, options align
11041 This pragma sets the alignment of fields in structures. The values of
11042 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
11043 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
11044 properly; to restore the previous setting, use @code{reset} for the
11047 @item segment @var{tokens}@dots{}
11048 @cindex pragma, segment
11049 This pragma is accepted, but has no effect.
11051 @item unused (@var{var} [, @var{var}]@dots{})
11052 @cindex pragma, unused
11053 This pragma declares variables to be possibly unused. GCC will not
11054 produce warnings for the listed variables. The effect is similar to
11055 that of the @code{unused} attribute, except that this pragma may appear
11056 anywhere within the variables' scopes.
11059 @node Solaris Pragmas
11060 @subsection Solaris Pragmas
11062 The Solaris target supports @code{#pragma redefine_extname}
11063 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
11064 @code{#pragma} directives for compatibility with the system compiler.
11067 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
11068 @cindex pragma, align
11070 Increase the minimum alignment of each @var{variable} to @var{alignment}.
11071 This is the same as GCC's @code{aligned} attribute @pxref{Variable
11072 Attributes}). Macro expansion occurs on the arguments to this pragma
11073 when compiling C and Objective-C@. It does not currently occur when
11074 compiling C++, but this is a bug which may be fixed in a future
11077 @item fini (@var{function} [, @var{function}]...)
11078 @cindex pragma, fini
11080 This pragma causes each listed @var{function} to be called after
11081 main, or during shared module unloading, by adding a call to the
11082 @code{.fini} section.
11084 @item init (@var{function} [, @var{function}]...)
11085 @cindex pragma, init
11087 This pragma causes each listed @var{function} to be called during
11088 initialization (before @code{main}) or during shared module loading, by
11089 adding a call to the @code{.init} section.
11093 @node Symbol-Renaming Pragmas
11094 @subsection Symbol-Renaming Pragmas
11096 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
11097 supports two @code{#pragma} directives which change the name used in
11098 assembly for a given declaration. These pragmas are only available on
11099 platforms whose system headers need them. To get this effect on all
11100 platforms supported by GCC, use the asm labels extension (@pxref{Asm
11104 @item redefine_extname @var{oldname} @var{newname}
11105 @cindex pragma, redefine_extname
11107 This pragma gives the C function @var{oldname} the assembly symbol
11108 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
11109 will be defined if this pragma is available (currently only on
11112 @item extern_prefix @var{string}
11113 @cindex pragma, extern_prefix
11115 This pragma causes all subsequent external function and variable
11116 declarations to have @var{string} prepended to their assembly symbols.
11117 This effect may be terminated with another @code{extern_prefix} pragma
11118 whose argument is an empty string. The preprocessor macro
11119 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
11120 available (currently only on Tru64 UNIX)@.
11123 These pragmas and the asm labels extension interact in a complicated
11124 manner. Here are some corner cases you may want to be aware of.
11127 @item Both pragmas silently apply only to declarations with external
11128 linkage. Asm labels do not have this restriction.
11130 @item In C++, both pragmas silently apply only to declarations with
11131 ``C'' linkage. Again, asm labels do not have this restriction.
11133 @item If any of the three ways of changing the assembly name of a
11134 declaration is applied to a declaration whose assembly name has
11135 already been determined (either by a previous use of one of these
11136 features, or because the compiler needed the assembly name in order to
11137 generate code), and the new name is different, a warning issues and
11138 the name does not change.
11140 @item The @var{oldname} used by @code{#pragma redefine_extname} is
11141 always the C-language name.
11143 @item If @code{#pragma extern_prefix} is in effect, and a declaration
11144 occurs with an asm label attached, the prefix is silently ignored for
11147 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
11148 apply to the same declaration, whichever triggered first wins, and a
11149 warning issues if they contradict each other. (We would like to have
11150 @code{#pragma redefine_extname} always win, for consistency with asm
11151 labels, but if @code{#pragma extern_prefix} triggers first we have no
11152 way of knowing that that happened.)
11155 @node Structure-Packing Pragmas
11156 @subsection Structure-Packing Pragmas
11158 For compatibility with Microsoft Windows compilers, GCC supports a
11159 set of @code{#pragma} directives which change the maximum alignment of
11160 members of structures (other than zero-width bitfields), unions, and
11161 classes subsequently defined. The @var{n} value below always is required
11162 to be a small power of two and specifies the new alignment in bytes.
11165 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
11166 @item @code{#pragma pack()} sets the alignment to the one that was in
11167 effect when compilation started (see also command line option
11168 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
11169 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
11170 setting on an internal stack and then optionally sets the new alignment.
11171 @item @code{#pragma pack(pop)} restores the alignment setting to the one
11172 saved at the top of the internal stack (and removes that stack entry).
11173 Note that @code{#pragma pack([@var{n}])} does not influence this internal
11174 stack; thus it is possible to have @code{#pragma pack(push)} followed by
11175 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
11176 @code{#pragma pack(pop)}.
11179 Some targets, e.g.@: i386 and powerpc, support the @code{ms_struct}
11180 @code{#pragma} which lays out a structure as the documented
11181 @code{__attribute__ ((ms_struct))}.
11183 @item @code{#pragma ms_struct on} turns on the layout for structures
11185 @item @code{#pragma ms_struct off} turns off the layout for structures
11187 @item @code{#pragma ms_struct reset} goes back to the default layout.
11191 @subsection Weak Pragmas
11193 For compatibility with SVR4, GCC supports a set of @code{#pragma}
11194 directives for declaring symbols to be weak, and defining weak
11198 @item #pragma weak @var{symbol}
11199 @cindex pragma, weak
11200 This pragma declares @var{symbol} to be weak, as if the declaration
11201 had the attribute of the same name. The pragma may appear before
11202 or after the declaration of @var{symbol}, but must appear before
11203 either its first use or its definition. It is not an error for
11204 @var{symbol} to never be defined at all.
11206 @item #pragma weak @var{symbol1} = @var{symbol2}
11207 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
11208 It is an error if @var{symbol2} is not defined in the current
11212 @node Diagnostic Pragmas
11213 @subsection Diagnostic Pragmas
11215 GCC allows the user to selectively enable or disable certain types of
11216 diagnostics, and change the kind of the diagnostic. For example, a
11217 project's policy might require that all sources compile with
11218 @option{-Werror} but certain files might have exceptions allowing
11219 specific types of warnings. Or, a project might selectively enable
11220 diagnostics and treat them as errors depending on which preprocessor
11221 macros are defined.
11224 @item #pragma GCC diagnostic @var{kind} @var{option}
11225 @cindex pragma, diagnostic
11227 Modifies the disposition of a diagnostic. Note that not all
11228 diagnostics are modifiable; at the moment only warnings (normally
11229 controlled by @samp{-W@dots{}}) can be controlled, and not all of them.
11230 Use @option{-fdiagnostics-show-option} to determine which diagnostics
11231 are controllable and which option controls them.
11233 @var{kind} is @samp{error} to treat this diagnostic as an error,
11234 @samp{warning} to treat it like a warning (even if @option{-Werror} is
11235 in effect), or @samp{ignored} if the diagnostic is to be ignored.
11236 @var{option} is a double quoted string which matches the command line
11240 #pragma GCC diagnostic warning "-Wformat"
11241 #pragma GCC diagnostic error "-Wformat"
11242 #pragma GCC diagnostic ignored "-Wformat"
11245 Note that these pragmas override any command line options. Also,
11246 while it is syntactically valid to put these pragmas anywhere in your
11247 sources, the only supported location for them is before any data or
11248 functions are defined. Doing otherwise may result in unpredictable
11249 results depending on how the optimizer manages your sources. If the
11250 same option is listed multiple times, the last one specified is the
11251 one that is in effect. This pragma is not intended to be a general
11252 purpose replacement for command line options, but for implementing
11253 strict control over project policies.
11257 @node Visibility Pragmas
11258 @subsection Visibility Pragmas
11261 @item #pragma GCC visibility push(@var{visibility})
11262 @itemx #pragma GCC visibility pop
11263 @cindex pragma, visibility
11265 This pragma allows the user to set the visibility for multiple
11266 declarations without having to give each a visibility attribute
11267 @xref{Function Attributes}, for more information about visibility and
11268 the attribute syntax.
11270 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
11271 declarations. Class members and template specializations are not
11272 affected; if you want to override the visibility for a particular
11273 member or instantiation, you must use an attribute.
11278 @node Push/Pop Macro Pragmas
11279 @subsection Push/Pop Macro Pragmas
11281 For compatibility with Microsoft Windows compilers, GCC supports
11282 @samp{#pragma push_macro(@var{"macro_name"})}
11283 and @samp{#pragma pop_macro(@var{"macro_name"})}.
11286 @item #pragma push_macro(@var{"macro_name"})
11287 @cindex pragma, push_macro
11288 This pragma saves the value of the macro named as @var{macro_name} to
11289 the top of the stack for this macro.
11291 @item #pragma pop_macro(@var{"macro_name"})
11292 @cindex pragma, pop_macro
11293 This pragma sets the value of the macro named as @var{macro_name} to
11294 the value on top of the stack for this macro. If the stack for
11295 @var{macro_name} is empty, the value of the macro remains unchanged.
11302 #pragma push_macro("X")
11305 #pragma pop_macro("X")
11309 In this example, the definition of X as 1 is saved by @code{#pragma
11310 push_macro} and restored by @code{#pragma pop_macro}.
11312 @node Unnamed Fields
11313 @section Unnamed struct/union fields within structs/unions
11317 For compatibility with other compilers, GCC allows you to define
11318 a structure or union that contains, as fields, structures and unions
11319 without names. For example:
11332 In this example, the user would be able to access members of the unnamed
11333 union with code like @samp{foo.b}. Note that only unnamed structs and
11334 unions are allowed, you may not have, for example, an unnamed
11337 You must never create such structures that cause ambiguous field definitions.
11338 For example, this structure:
11349 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
11350 Such constructs are not supported and must be avoided. In the future,
11351 such constructs may be detected and treated as compilation errors.
11353 @opindex fms-extensions
11354 Unless @option{-fms-extensions} is used, the unnamed field must be a
11355 structure or union definition without a tag (for example, @samp{struct
11356 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
11357 also be a definition with a tag such as @samp{struct foo @{ int a;
11358 @};}, a reference to a previously defined structure or union such as
11359 @samp{struct foo;}, or a reference to a @code{typedef} name for a
11360 previously defined structure or union type.
11363 @section Thread-Local Storage
11364 @cindex Thread-Local Storage
11365 @cindex @acronym{TLS}
11368 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
11369 are allocated such that there is one instance of the variable per extant
11370 thread. The run-time model GCC uses to implement this originates
11371 in the IA-64 processor-specific ABI, but has since been migrated
11372 to other processors as well. It requires significant support from
11373 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
11374 system libraries (@file{libc.so} and @file{libpthread.so}), so it
11375 is not available everywhere.
11377 At the user level, the extension is visible with a new storage
11378 class keyword: @code{__thread}. For example:
11382 extern __thread struct state s;
11383 static __thread char *p;
11386 The @code{__thread} specifier may be used alone, with the @code{extern}
11387 or @code{static} specifiers, but with no other storage class specifier.
11388 When used with @code{extern} or @code{static}, @code{__thread} must appear
11389 immediately after the other storage class specifier.
11391 The @code{__thread} specifier may be applied to any global, file-scoped
11392 static, function-scoped static, or static data member of a class. It may
11393 not be applied to block-scoped automatic or non-static data member.
11395 When the address-of operator is applied to a thread-local variable, it is
11396 evaluated at run-time and returns the address of the current thread's
11397 instance of that variable. An address so obtained may be used by any
11398 thread. When a thread terminates, any pointers to thread-local variables
11399 in that thread become invalid.
11401 No static initialization may refer to the address of a thread-local variable.
11403 In C++, if an initializer is present for a thread-local variable, it must
11404 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
11407 See @uref{http://people.redhat.com/drepper/tls.pdf,
11408 ELF Handling For Thread-Local Storage} for a detailed explanation of
11409 the four thread-local storage addressing models, and how the run-time
11410 is expected to function.
11413 * C99 Thread-Local Edits::
11414 * C++98 Thread-Local Edits::
11417 @node C99 Thread-Local Edits
11418 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
11420 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
11421 that document the exact semantics of the language extension.
11425 @cite{5.1.2 Execution environments}
11427 Add new text after paragraph 1
11430 Within either execution environment, a @dfn{thread} is a flow of
11431 control within a program. It is implementation defined whether
11432 or not there may be more than one thread associated with a program.
11433 It is implementation defined how threads beyond the first are
11434 created, the name and type of the function called at thread
11435 startup, and how threads may be terminated. However, objects
11436 with thread storage duration shall be initialized before thread
11441 @cite{6.2.4 Storage durations of objects}
11443 Add new text before paragraph 3
11446 An object whose identifier is declared with the storage-class
11447 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
11448 Its lifetime is the entire execution of the thread, and its
11449 stored value is initialized only once, prior to thread startup.
11453 @cite{6.4.1 Keywords}
11455 Add @code{__thread}.
11458 @cite{6.7.1 Storage-class specifiers}
11460 Add @code{__thread} to the list of storage class specifiers in
11463 Change paragraph 2 to
11466 With the exception of @code{__thread}, at most one storage-class
11467 specifier may be given [@dots{}]. The @code{__thread} specifier may
11468 be used alone, or immediately following @code{extern} or
11472 Add new text after paragraph 6
11475 The declaration of an identifier for a variable that has
11476 block scope that specifies @code{__thread} shall also
11477 specify either @code{extern} or @code{static}.
11479 The @code{__thread} specifier shall be used only with
11484 @node C++98 Thread-Local Edits
11485 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
11487 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
11488 that document the exact semantics of the language extension.
11492 @b{[intro.execution]}
11494 New text after paragraph 4
11497 A @dfn{thread} is a flow of control within the abstract machine.
11498 It is implementation defined whether or not there may be more than
11502 New text after paragraph 7
11505 It is unspecified whether additional action must be taken to
11506 ensure when and whether side effects are visible to other threads.
11512 Add @code{__thread}.
11515 @b{[basic.start.main]}
11517 Add after paragraph 5
11520 The thread that begins execution at the @code{main} function is called
11521 the @dfn{main thread}. It is implementation defined how functions
11522 beginning threads other than the main thread are designated or typed.
11523 A function so designated, as well as the @code{main} function, is called
11524 a @dfn{thread startup function}. It is implementation defined what
11525 happens if a thread startup function returns. It is implementation
11526 defined what happens to other threads when any thread calls @code{exit}.
11530 @b{[basic.start.init]}
11532 Add after paragraph 4
11535 The storage for an object of thread storage duration shall be
11536 statically initialized before the first statement of the thread startup
11537 function. An object of thread storage duration shall not require
11538 dynamic initialization.
11542 @b{[basic.start.term]}
11544 Add after paragraph 3
11547 The type of an object with thread storage duration shall not have a
11548 non-trivial destructor, nor shall it be an array type whose elements
11549 (directly or indirectly) have non-trivial destructors.
11555 Add ``thread storage duration'' to the list in paragraph 1.
11560 Thread, static, and automatic storage durations are associated with
11561 objects introduced by declarations [@dots{}].
11564 Add @code{__thread} to the list of specifiers in paragraph 3.
11567 @b{[basic.stc.thread]}
11569 New section before @b{[basic.stc.static]}
11572 The keyword @code{__thread} applied to a non-local object gives the
11573 object thread storage duration.
11575 A local variable or class data member declared both @code{static}
11576 and @code{__thread} gives the variable or member thread storage
11581 @b{[basic.stc.static]}
11586 All objects which have neither thread storage duration, dynamic
11587 storage duration nor are local [@dots{}].
11593 Add @code{__thread} to the list in paragraph 1.
11598 With the exception of @code{__thread}, at most one
11599 @var{storage-class-specifier} shall appear in a given
11600 @var{decl-specifier-seq}. The @code{__thread} specifier may
11601 be used alone, or immediately following the @code{extern} or
11602 @code{static} specifiers. [@dots{}]
11605 Add after paragraph 5
11608 The @code{__thread} specifier can be applied only to the names of objects
11609 and to anonymous unions.
11615 Add after paragraph 6
11618 Non-@code{static} members shall not be @code{__thread}.
11622 @node Binary constants
11623 @section Binary constants using the @samp{0b} prefix
11624 @cindex Binary constants using the @samp{0b} prefix
11626 Integer constants can be written as binary constants, consisting of a
11627 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
11628 @samp{0B}. This is particularly useful in environments that operate a
11629 lot on the bit-level (like microcontrollers).
11631 The following statements are identical:
11640 The type of these constants follows the same rules as for octal or
11641 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
11644 @node C++ Extensions
11645 @chapter Extensions to the C++ Language
11646 @cindex extensions, C++ language
11647 @cindex C++ language extensions
11649 The GNU compiler provides these extensions to the C++ language (and you
11650 can also use most of the C language extensions in your C++ programs). If you
11651 want to write code that checks whether these features are available, you can
11652 test for the GNU compiler the same way as for C programs: check for a
11653 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
11654 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
11655 Predefined Macros,cpp,The GNU C Preprocessor}).
11658 * Volatiles:: What constitutes an access to a volatile object.
11659 * Restricted Pointers:: C99 restricted pointers and references.
11660 * Vague Linkage:: Where G++ puts inlines, vtables and such.
11661 * C++ Interface:: You can use a single C++ header file for both
11662 declarations and definitions.
11663 * Template Instantiation:: Methods for ensuring that exactly one copy of
11664 each needed template instantiation is emitted.
11665 * Bound member functions:: You can extract a function pointer to the
11666 method denoted by a @samp{->*} or @samp{.*} expression.
11667 * C++ Attributes:: Variable, function, and type attributes for C++ only.
11668 * Namespace Association:: Strong using-directives for namespace association.
11669 * Type Traits:: Compiler support for type traits
11670 * Java Exceptions:: Tweaking exception handling to work with Java.
11671 * Deprecated Features:: Things will disappear from g++.
11672 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
11676 @section When is a Volatile Object Accessed?
11677 @cindex accessing volatiles
11678 @cindex volatile read
11679 @cindex volatile write
11680 @cindex volatile access
11682 Both the C and C++ standard have the concept of volatile objects. These
11683 are normally accessed by pointers and used for accessing hardware. The
11684 standards encourage compilers to refrain from optimizations concerning
11685 accesses to volatile objects. The C standard leaves it implementation
11686 defined as to what constitutes a volatile access. The C++ standard omits
11687 to specify this, except to say that C++ should behave in a similar manner
11688 to C with respect to volatiles, where possible. The minimum either
11689 standard specifies is that at a sequence point all previous accesses to
11690 volatile objects have stabilized and no subsequent accesses have
11691 occurred. Thus an implementation is free to reorder and combine
11692 volatile accesses which occur between sequence points, but cannot do so
11693 for accesses across a sequence point. The use of volatiles does not
11694 allow you to violate the restriction on updating objects multiple times
11695 within a sequence point.
11697 @xref{Qualifiers implementation, , Volatile qualifier and the C compiler}.
11699 The behavior differs slightly between C and C++ in the non-obvious cases:
11702 volatile int *src = @var{somevalue};
11706 With C, such expressions are rvalues, and GCC interprets this either as a
11707 read of the volatile object being pointed to or only as request to evaluate
11708 the side-effects. The C++ standard specifies that such expressions do not
11709 undergo lvalue to rvalue conversion, and that the type of the dereferenced
11710 object may be incomplete. The C++ standard does not specify explicitly
11711 that it is this lvalue to rvalue conversion which may be responsible for
11712 causing an access. However, there is reason to believe that it is,
11713 because otherwise certain simple expressions become undefined. However,
11714 because it would surprise most programmers, G++ treats dereferencing a
11715 pointer to volatile object of complete type when the value is unused as
11716 GCC would do for an equivalent type in C@. When the object has incomplete
11717 type, G++ issues a warning; if you wish to force an error, you must
11718 force a conversion to rvalue with, for instance, a static cast.
11720 When using a reference to volatile, G++ does not treat equivalent
11721 expressions as accesses to volatiles, but instead issues a warning that
11722 no volatile is accessed. The rationale for this is that otherwise it
11723 becomes difficult to determine where volatile access occur, and not
11724 possible to ignore the return value from functions returning volatile
11725 references. Again, if you wish to force a read, cast the reference to
11728 @node Restricted Pointers
11729 @section Restricting Pointer Aliasing
11730 @cindex restricted pointers
11731 @cindex restricted references
11732 @cindex restricted this pointer
11734 As with the C front end, G++ understands the C99 feature of restricted pointers,
11735 specified with the @code{__restrict__}, or @code{__restrict} type
11736 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
11737 language flag, @code{restrict} is not a keyword in C++.
11739 In addition to allowing restricted pointers, you can specify restricted
11740 references, which indicate that the reference is not aliased in the local
11744 void fn (int *__restrict__ rptr, int &__restrict__ rref)
11751 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
11752 @var{rref} refers to a (different) unaliased integer.
11754 You may also specify whether a member function's @var{this} pointer is
11755 unaliased by using @code{__restrict__} as a member function qualifier.
11758 void T::fn () __restrict__
11765 Within the body of @code{T::fn}, @var{this} will have the effective
11766 definition @code{T *__restrict__ const this}. Notice that the
11767 interpretation of a @code{__restrict__} member function qualifier is
11768 different to that of @code{const} or @code{volatile} qualifier, in that it
11769 is applied to the pointer rather than the object. This is consistent with
11770 other compilers which implement restricted pointers.
11772 As with all outermost parameter qualifiers, @code{__restrict__} is
11773 ignored in function definition matching. This means you only need to
11774 specify @code{__restrict__} in a function definition, rather than
11775 in a function prototype as well.
11777 @node Vague Linkage
11778 @section Vague Linkage
11779 @cindex vague linkage
11781 There are several constructs in C++ which require space in the object
11782 file but are not clearly tied to a single translation unit. We say that
11783 these constructs have ``vague linkage''. Typically such constructs are
11784 emitted wherever they are needed, though sometimes we can be more
11788 @item Inline Functions
11789 Inline functions are typically defined in a header file which can be
11790 included in many different compilations. Hopefully they can usually be
11791 inlined, but sometimes an out-of-line copy is necessary, if the address
11792 of the function is taken or if inlining fails. In general, we emit an
11793 out-of-line copy in all translation units where one is needed. As an
11794 exception, we only emit inline virtual functions with the vtable, since
11795 it will always require a copy.
11797 Local static variables and string constants used in an inline function
11798 are also considered to have vague linkage, since they must be shared
11799 between all inlined and out-of-line instances of the function.
11803 C++ virtual functions are implemented in most compilers using a lookup
11804 table, known as a vtable. The vtable contains pointers to the virtual
11805 functions provided by a class, and each object of the class contains a
11806 pointer to its vtable (or vtables, in some multiple-inheritance
11807 situations). If the class declares any non-inline, non-pure virtual
11808 functions, the first one is chosen as the ``key method'' for the class,
11809 and the vtable is only emitted in the translation unit where the key
11812 @emph{Note:} If the chosen key method is later defined as inline, the
11813 vtable will still be emitted in every translation unit which defines it.
11814 Make sure that any inline virtuals are declared inline in the class
11815 body, even if they are not defined there.
11817 @item type_info objects
11820 C++ requires information about types to be written out in order to
11821 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
11822 For polymorphic classes (classes with virtual functions), the type_info
11823 object is written out along with the vtable so that @samp{dynamic_cast}
11824 can determine the dynamic type of a class object at runtime. For all
11825 other types, we write out the type_info object when it is used: when
11826 applying @samp{typeid} to an expression, throwing an object, or
11827 referring to a type in a catch clause or exception specification.
11829 @item Template Instantiations
11830 Most everything in this section also applies to template instantiations,
11831 but there are other options as well.
11832 @xref{Template Instantiation,,Where's the Template?}.
11836 When used with GNU ld version 2.8 or later on an ELF system such as
11837 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
11838 these constructs will be discarded at link time. This is known as
11841 On targets that don't support COMDAT, but do support weak symbols, GCC
11842 will use them. This way one copy will override all the others, but
11843 the unused copies will still take up space in the executable.
11845 For targets which do not support either COMDAT or weak symbols,
11846 most entities with vague linkage will be emitted as local symbols to
11847 avoid duplicate definition errors from the linker. This will not happen
11848 for local statics in inlines, however, as having multiple copies will
11849 almost certainly break things.
11851 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
11852 another way to control placement of these constructs.
11854 @node C++ Interface
11855 @section #pragma interface and implementation
11857 @cindex interface and implementation headers, C++
11858 @cindex C++ interface and implementation headers
11859 @cindex pragmas, interface and implementation
11861 @code{#pragma interface} and @code{#pragma implementation} provide the
11862 user with a way of explicitly directing the compiler to emit entities
11863 with vague linkage (and debugging information) in a particular
11866 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
11867 most cases, because of COMDAT support and the ``key method'' heuristic
11868 mentioned in @ref{Vague Linkage}. Using them can actually cause your
11869 program to grow due to unnecessary out-of-line copies of inline
11870 functions. Currently (3.4) the only benefit of these
11871 @code{#pragma}s is reduced duplication of debugging information, and
11872 that should be addressed soon on DWARF 2 targets with the use of
11876 @item #pragma interface
11877 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
11878 @kindex #pragma interface
11879 Use this directive in @emph{header files} that define object classes, to save
11880 space in most of the object files that use those classes. Normally,
11881 local copies of certain information (backup copies of inline member
11882 functions, debugging information, and the internal tables that implement
11883 virtual functions) must be kept in each object file that includes class
11884 definitions. You can use this pragma to avoid such duplication. When a
11885 header file containing @samp{#pragma interface} is included in a
11886 compilation, this auxiliary information will not be generated (unless
11887 the main input source file itself uses @samp{#pragma implementation}).
11888 Instead, the object files will contain references to be resolved at link
11891 The second form of this directive is useful for the case where you have
11892 multiple headers with the same name in different directories. If you
11893 use this form, you must specify the same string to @samp{#pragma
11896 @item #pragma implementation
11897 @itemx #pragma implementation "@var{objects}.h"
11898 @kindex #pragma implementation
11899 Use this pragma in a @emph{main input file}, when you want full output from
11900 included header files to be generated (and made globally visible). The
11901 included header file, in turn, should use @samp{#pragma interface}.
11902 Backup copies of inline member functions, debugging information, and the
11903 internal tables used to implement virtual functions are all generated in
11904 implementation files.
11906 @cindex implied @code{#pragma implementation}
11907 @cindex @code{#pragma implementation}, implied
11908 @cindex naming convention, implementation headers
11909 If you use @samp{#pragma implementation} with no argument, it applies to
11910 an include file with the same basename@footnote{A file's @dfn{basename}
11911 was the name stripped of all leading path information and of trailing
11912 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
11913 file. For example, in @file{allclass.cc}, giving just
11914 @samp{#pragma implementation}
11915 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
11917 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
11918 an implementation file whenever you would include it from
11919 @file{allclass.cc} even if you never specified @samp{#pragma
11920 implementation}. This was deemed to be more trouble than it was worth,
11921 however, and disabled.
11923 Use the string argument if you want a single implementation file to
11924 include code from multiple header files. (You must also use
11925 @samp{#include} to include the header file; @samp{#pragma
11926 implementation} only specifies how to use the file---it doesn't actually
11929 There is no way to split up the contents of a single header file into
11930 multiple implementation files.
11933 @cindex inlining and C++ pragmas
11934 @cindex C++ pragmas, effect on inlining
11935 @cindex pragmas in C++, effect on inlining
11936 @samp{#pragma implementation} and @samp{#pragma interface} also have an
11937 effect on function inlining.
11939 If you define a class in a header file marked with @samp{#pragma
11940 interface}, the effect on an inline function defined in that class is
11941 similar to an explicit @code{extern} declaration---the compiler emits
11942 no code at all to define an independent version of the function. Its
11943 definition is used only for inlining with its callers.
11945 @opindex fno-implement-inlines
11946 Conversely, when you include the same header file in a main source file
11947 that declares it as @samp{#pragma implementation}, the compiler emits
11948 code for the function itself; this defines a version of the function
11949 that can be found via pointers (or by callers compiled without
11950 inlining). If all calls to the function can be inlined, you can avoid
11951 emitting the function by compiling with @option{-fno-implement-inlines}.
11952 If any calls were not inlined, you will get linker errors.
11954 @node Template Instantiation
11955 @section Where's the Template?
11956 @cindex template instantiation
11958 C++ templates are the first language feature to require more
11959 intelligence from the environment than one usually finds on a UNIX
11960 system. Somehow the compiler and linker have to make sure that each
11961 template instance occurs exactly once in the executable if it is needed,
11962 and not at all otherwise. There are two basic approaches to this
11963 problem, which are referred to as the Borland model and the Cfront model.
11966 @item Borland model
11967 Borland C++ solved the template instantiation problem by adding the code
11968 equivalent of common blocks to their linker; the compiler emits template
11969 instances in each translation unit that uses them, and the linker
11970 collapses them together. The advantage of this model is that the linker
11971 only has to consider the object files themselves; there is no external
11972 complexity to worry about. This disadvantage is that compilation time
11973 is increased because the template code is being compiled repeatedly.
11974 Code written for this model tends to include definitions of all
11975 templates in the header file, since they must be seen to be
11979 The AT&T C++ translator, Cfront, solved the template instantiation
11980 problem by creating the notion of a template repository, an
11981 automatically maintained place where template instances are stored. A
11982 more modern version of the repository works as follows: As individual
11983 object files are built, the compiler places any template definitions and
11984 instantiations encountered in the repository. At link time, the link
11985 wrapper adds in the objects in the repository and compiles any needed
11986 instances that were not previously emitted. The advantages of this
11987 model are more optimal compilation speed and the ability to use the
11988 system linker; to implement the Borland model a compiler vendor also
11989 needs to replace the linker. The disadvantages are vastly increased
11990 complexity, and thus potential for error; for some code this can be
11991 just as transparent, but in practice it can been very difficult to build
11992 multiple programs in one directory and one program in multiple
11993 directories. Code written for this model tends to separate definitions
11994 of non-inline member templates into a separate file, which should be
11995 compiled separately.
11998 When used with GNU ld version 2.8 or later on an ELF system such as
11999 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
12000 Borland model. On other systems, G++ implements neither automatic
12003 A future version of G++ will support a hybrid model whereby the compiler
12004 will emit any instantiations for which the template definition is
12005 included in the compile, and store template definitions and
12006 instantiation context information into the object file for the rest.
12007 The link wrapper will extract that information as necessary and invoke
12008 the compiler to produce the remaining instantiations. The linker will
12009 then combine duplicate instantiations.
12011 In the mean time, you have the following options for dealing with
12012 template instantiations:
12017 Compile your template-using code with @option{-frepo}. The compiler will
12018 generate files with the extension @samp{.rpo} listing all of the
12019 template instantiations used in the corresponding object files which
12020 could be instantiated there; the link wrapper, @samp{collect2}, will
12021 then update the @samp{.rpo} files to tell the compiler where to place
12022 those instantiations and rebuild any affected object files. The
12023 link-time overhead is negligible after the first pass, as the compiler
12024 will continue to place the instantiations in the same files.
12026 This is your best option for application code written for the Borland
12027 model, as it will just work. Code written for the Cfront model will
12028 need to be modified so that the template definitions are available at
12029 one or more points of instantiation; usually this is as simple as adding
12030 @code{#include <tmethods.cc>} to the end of each template header.
12032 For library code, if you want the library to provide all of the template
12033 instantiations it needs, just try to link all of its object files
12034 together; the link will fail, but cause the instantiations to be
12035 generated as a side effect. Be warned, however, that this may cause
12036 conflicts if multiple libraries try to provide the same instantiations.
12037 For greater control, use explicit instantiation as described in the next
12041 @opindex fno-implicit-templates
12042 Compile your code with @option{-fno-implicit-templates} to disable the
12043 implicit generation of template instances, and explicitly instantiate
12044 all the ones you use. This approach requires more knowledge of exactly
12045 which instances you need than do the others, but it's less
12046 mysterious and allows greater control. You can scatter the explicit
12047 instantiations throughout your program, perhaps putting them in the
12048 translation units where the instances are used or the translation units
12049 that define the templates themselves; you can put all of the explicit
12050 instantiations you need into one big file; or you can create small files
12057 template class Foo<int>;
12058 template ostream& operator <<
12059 (ostream&, const Foo<int>&);
12062 for each of the instances you need, and create a template instantiation
12063 library from those.
12065 If you are using Cfront-model code, you can probably get away with not
12066 using @option{-fno-implicit-templates} when compiling files that don't
12067 @samp{#include} the member template definitions.
12069 If you use one big file to do the instantiations, you may want to
12070 compile it without @option{-fno-implicit-templates} so you get all of the
12071 instances required by your explicit instantiations (but not by any
12072 other files) without having to specify them as well.
12074 G++ has extended the template instantiation syntax given in the ISO
12075 standard to allow forward declaration of explicit instantiations
12076 (with @code{extern}), instantiation of the compiler support data for a
12077 template class (i.e.@: the vtable) without instantiating any of its
12078 members (with @code{inline}), and instantiation of only the static data
12079 members of a template class, without the support data or member
12080 functions (with (@code{static}):
12083 extern template int max (int, int);
12084 inline template class Foo<int>;
12085 static template class Foo<int>;
12089 Do nothing. Pretend G++ does implement automatic instantiation
12090 management. Code written for the Borland model will work fine, but
12091 each translation unit will contain instances of each of the templates it
12092 uses. In a large program, this can lead to an unacceptable amount of code
12096 @node Bound member functions
12097 @section Extracting the function pointer from a bound pointer to member function
12099 @cindex pointer to member function
12100 @cindex bound pointer to member function
12102 In C++, pointer to member functions (PMFs) are implemented using a wide
12103 pointer of sorts to handle all the possible call mechanisms; the PMF
12104 needs to store information about how to adjust the @samp{this} pointer,
12105 and if the function pointed to is virtual, where to find the vtable, and
12106 where in the vtable to look for the member function. If you are using
12107 PMFs in an inner loop, you should really reconsider that decision. If
12108 that is not an option, you can extract the pointer to the function that
12109 would be called for a given object/PMF pair and call it directly inside
12110 the inner loop, to save a bit of time.
12112 Note that you will still be paying the penalty for the call through a
12113 function pointer; on most modern architectures, such a call defeats the
12114 branch prediction features of the CPU@. This is also true of normal
12115 virtual function calls.
12117 The syntax for this extension is
12121 extern int (A::*fp)();
12122 typedef int (*fptr)(A *);
12124 fptr p = (fptr)(a.*fp);
12127 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
12128 no object is needed to obtain the address of the function. They can be
12129 converted to function pointers directly:
12132 fptr p1 = (fptr)(&A::foo);
12135 @opindex Wno-pmf-conversions
12136 You must specify @option{-Wno-pmf-conversions} to use this extension.
12138 @node C++ Attributes
12139 @section C++-Specific Variable, Function, and Type Attributes
12141 Some attributes only make sense for C++ programs.
12144 @item init_priority (@var{priority})
12145 @cindex init_priority attribute
12148 In Standard C++, objects defined at namespace scope are guaranteed to be
12149 initialized in an order in strict accordance with that of their definitions
12150 @emph{in a given translation unit}. No guarantee is made for initializations
12151 across translation units. However, GNU C++ allows users to control the
12152 order of initialization of objects defined at namespace scope with the
12153 @code{init_priority} attribute by specifying a relative @var{priority},
12154 a constant integral expression currently bounded between 101 and 65535
12155 inclusive. Lower numbers indicate a higher priority.
12157 In the following example, @code{A} would normally be created before
12158 @code{B}, but the @code{init_priority} attribute has reversed that order:
12161 Some_Class A __attribute__ ((init_priority (2000)));
12162 Some_Class B __attribute__ ((init_priority (543)));
12166 Note that the particular values of @var{priority} do not matter; only their
12169 @item java_interface
12170 @cindex java_interface attribute
12172 This type attribute informs C++ that the class is a Java interface. It may
12173 only be applied to classes declared within an @code{extern "Java"} block.
12174 Calls to methods declared in this interface will be dispatched using GCJ's
12175 interface table mechanism, instead of regular virtual table dispatch.
12179 See also @xref{Namespace Association}.
12181 @node Namespace Association
12182 @section Namespace Association
12184 @strong{Caution:} The semantics of this extension are not fully
12185 defined. Users should refrain from using this extension as its
12186 semantics may change subtly over time. It is possible that this
12187 extension will be removed in future versions of G++.
12189 A using-directive with @code{__attribute ((strong))} is stronger
12190 than a normal using-directive in two ways:
12194 Templates from the used namespace can be specialized and explicitly
12195 instantiated as though they were members of the using namespace.
12198 The using namespace is considered an associated namespace of all
12199 templates in the used namespace for purposes of argument-dependent
12203 The used namespace must be nested within the using namespace so that
12204 normal unqualified lookup works properly.
12206 This is useful for composing a namespace transparently from
12207 implementation namespaces. For example:
12212 template <class T> struct A @{ @};
12214 using namespace debug __attribute ((__strong__));
12215 template <> struct A<int> @{ @}; // @r{ok to specialize}
12217 template <class T> void f (A<T>);
12222 f (std::A<float>()); // @r{lookup finds} std::f
12228 @section Type Traits
12230 The C++ front-end implements syntactic extensions that allow to
12231 determine at compile time various characteristics of a type (or of a
12235 @item __has_nothrow_assign (type)
12236 If @code{type} is const qualified or is a reference type then the trait is
12237 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
12238 is true, else if @code{type} is a cv class or union type with copy assignment
12239 operators that are known not to throw an exception then the trait is true,
12240 else it is false. Requires: @code{type} shall be a complete type, an array
12241 type of unknown bound, or is a @code{void} type.
12243 @item __has_nothrow_copy (type)
12244 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
12245 @code{type} is a cv class or union type with copy constructors that
12246 are known not to throw an exception then the trait is true, else it is false.
12247 Requires: @code{type} shall be a complete type, an array type of
12248 unknown bound, or is a @code{void} type.
12250 @item __has_nothrow_constructor (type)
12251 If @code{__has_trivial_constructor (type)} is true then the trait is
12252 true, else if @code{type} is a cv class or union type (or array
12253 thereof) with a default constructor that is known not to throw an
12254 exception then the trait is true, else it is false. Requires:
12255 @code{type} shall be a complete type, an array type of unknown bound,
12256 or is a @code{void} type.
12258 @item __has_trivial_assign (type)
12259 If @code{type} is const qualified or is a reference type then the trait is
12260 false. Otherwise if @code{__is_pod (type)} is true then the trait is
12261 true, else if @code{type} is a cv class or union type with a trivial
12262 copy assignment ([class.copy]) then the trait is true, else it is
12263 false. Requires: @code{type} shall be a complete type, an array type
12264 of unknown bound, or is a @code{void} type.
12266 @item __has_trivial_copy (type)
12267 If @code{__is_pod (type)} is true or @code{type} is a reference type
12268 then the trait is true, else if @code{type} is a cv class or union type
12269 with a trivial copy constructor ([class.copy]) then the trait
12270 is true, else it is false. Requires: @code{type} shall be a complete
12271 type, an array type of unknown bound, or is a @code{void} type.
12273 @item __has_trivial_constructor (type)
12274 If @code{__is_pod (type)} is true then the trait is true, else if
12275 @code{type} is a cv class or union type (or array thereof) with a
12276 trivial default constructor ([class.ctor]) then the trait is true,
12277 else it is false. Requires: @code{type} shall be a complete type, an
12278 array type of unknown bound, or is a @code{void} type.
12280 @item __has_trivial_destructor (type)
12281 If @code{__is_pod (type)} is true or @code{type} is a reference type then
12282 the trait is true, else if @code{type} is a cv class or union type (or
12283 array thereof) with a trivial destructor ([class.dtor]) then the trait
12284 is true, else it is false. Requires: @code{type} shall be a complete
12285 type, an array type of unknown bound, or is a @code{void} type.
12287 @item __has_virtual_destructor (type)
12288 If @code{type} is a class type with a virtual destructor
12289 ([class.dtor]) then the trait is true, else it is false. Requires:
12290 @code{type} shall be a complete type, an array type of unknown bound,
12291 or is a @code{void} type.
12293 @item __is_abstract (type)
12294 If @code{type} is an abstract class ([class.abstract]) then the trait
12295 is true, else it is false. Requires: @code{type} shall be a complete
12296 type, an array type of unknown bound, or is a @code{void} type.
12298 @item __is_base_of (base_type, derived_type)
12299 If @code{base_type} is a base class of @code{derived_type}
12300 ([class.derived]) then the trait is true, otherwise it is false.
12301 Top-level cv qualifications of @code{base_type} and
12302 @code{derived_type} are ignored. For the purposes of this trait, a
12303 class type is considered is own base. Requires: if @code{__is_class
12304 (base_type)} and @code{__is_class (derived_type)} are true and
12305 @code{base_type} and @code{derived_type} are not the same type
12306 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
12307 type. Diagnostic is produced if this requirement is not met.
12309 @item __is_class (type)
12310 If @code{type} is a cv class type, and not a union type
12311 ([basic.compound]) the the trait is true, else it is false.
12313 @item __is_empty (type)
12314 If @code{__is_class (type)} is false then the trait is false.
12315 Otherwise @code{type} is considered empty if and only if: @code{type}
12316 has no non-static data members, or all non-static data members, if
12317 any, are bit-fields of lenght 0, and @code{type} has no virtual
12318 members, and @code{type} has no virtual base classes, and @code{type}
12319 has no base classes @code{base_type} for which
12320 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
12321 be a complete type, an array type of unknown bound, or is a
12324 @item __is_enum (type)
12325 If @code{type} is a cv enumeration type ([basic.compound]) the the trait is
12326 true, else it is false.
12328 @item __is_pod (type)
12329 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
12330 else it is false. Requires: @code{type} shall be a complete type,
12331 an array type of unknown bound, or is a @code{void} type.
12333 @item __is_polymorphic (type)
12334 If @code{type} is a polymorphic class ([class.virtual]) then the trait
12335 is true, else it is false. Requires: @code{type} shall be a complete
12336 type, an array type of unknown bound, or is a @code{void} type.
12338 @item __is_union (type)
12339 If @code{type} is a cv union type ([basic.compound]) the the trait is
12340 true, else it is false.
12344 @node Java Exceptions
12345 @section Java Exceptions
12347 The Java language uses a slightly different exception handling model
12348 from C++. Normally, GNU C++ will automatically detect when you are
12349 writing C++ code that uses Java exceptions, and handle them
12350 appropriately. However, if C++ code only needs to execute destructors
12351 when Java exceptions are thrown through it, GCC will guess incorrectly.
12352 Sample problematic code is:
12355 struct S @{ ~S(); @};
12356 extern void bar(); // @r{is written in Java, and may throw exceptions}
12365 The usual effect of an incorrect guess is a link failure, complaining of
12366 a missing routine called @samp{__gxx_personality_v0}.
12368 You can inform the compiler that Java exceptions are to be used in a
12369 translation unit, irrespective of what it might think, by writing
12370 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
12371 @samp{#pragma} must appear before any functions that throw or catch
12372 exceptions, or run destructors when exceptions are thrown through them.
12374 You cannot mix Java and C++ exceptions in the same translation unit. It
12375 is believed to be safe to throw a C++ exception from one file through
12376 another file compiled for the Java exception model, or vice versa, but
12377 there may be bugs in this area.
12379 @node Deprecated Features
12380 @section Deprecated Features
12382 In the past, the GNU C++ compiler was extended to experiment with new
12383 features, at a time when the C++ language was still evolving. Now that
12384 the C++ standard is complete, some of those features are superseded by
12385 superior alternatives. Using the old features might cause a warning in
12386 some cases that the feature will be dropped in the future. In other
12387 cases, the feature might be gone already.
12389 While the list below is not exhaustive, it documents some of the options
12390 that are now deprecated:
12393 @item -fexternal-templates
12394 @itemx -falt-external-templates
12395 These are two of the many ways for G++ to implement template
12396 instantiation. @xref{Template Instantiation}. The C++ standard clearly
12397 defines how template definitions have to be organized across
12398 implementation units. G++ has an implicit instantiation mechanism that
12399 should work just fine for standard-conforming code.
12401 @item -fstrict-prototype
12402 @itemx -fno-strict-prototype
12403 Previously it was possible to use an empty prototype parameter list to
12404 indicate an unspecified number of parameters (like C), rather than no
12405 parameters, as C++ demands. This feature has been removed, except where
12406 it is required for backwards compatibility @xref{Backwards Compatibility}.
12409 G++ allows a virtual function returning @samp{void *} to be overridden
12410 by one returning a different pointer type. This extension to the
12411 covariant return type rules is now deprecated and will be removed from a
12414 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
12415 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
12416 and are now removed from G++. Code using these operators should be
12417 modified to use @code{std::min} and @code{std::max} instead.
12419 The named return value extension has been deprecated, and is now
12422 The use of initializer lists with new expressions has been deprecated,
12423 and is now removed from G++.
12425 Floating and complex non-type template parameters have been deprecated,
12426 and are now removed from G++.
12428 The implicit typename extension has been deprecated and is now
12431 The use of default arguments in function pointers, function typedefs
12432 and other places where they are not permitted by the standard is
12433 deprecated and will be removed from a future version of G++.
12435 G++ allows floating-point literals to appear in integral constant expressions,
12436 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
12437 This extension is deprecated and will be removed from a future version.
12439 G++ allows static data members of const floating-point type to be declared
12440 with an initializer in a class definition. The standard only allows
12441 initializers for static members of const integral types and const
12442 enumeration types so this extension has been deprecated and will be removed
12443 from a future version.
12445 @node Backwards Compatibility
12446 @section Backwards Compatibility
12447 @cindex Backwards Compatibility
12448 @cindex ARM [Annotated C++ Reference Manual]
12450 Now that there is a definitive ISO standard C++, G++ has a specification
12451 to adhere to. The C++ language evolved over time, and features that
12452 used to be acceptable in previous drafts of the standard, such as the ARM
12453 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
12454 compilation of C++ written to such drafts, G++ contains some backwards
12455 compatibilities. @emph{All such backwards compatibility features are
12456 liable to disappear in future versions of G++.} They should be considered
12457 deprecated @xref{Deprecated Features}.
12461 If a variable is declared at for scope, it used to remain in scope until
12462 the end of the scope which contained the for statement (rather than just
12463 within the for scope). G++ retains this, but issues a warning, if such a
12464 variable is accessed outside the for scope.
12466 @item Implicit C language
12467 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
12468 scope to set the language. On such systems, all header files are
12469 implicitly scoped inside a C language scope. Also, an empty prototype
12470 @code{()} will be treated as an unspecified number of arguments, rather
12471 than no arguments, as C++ demands.