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.
2518 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2519 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2520 use the normal calling convention based on @code{jsr} and @code{rts}.
2521 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2525 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2526 Use this attribute together with @code{interrupt_handler},
2527 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2528 entry code should enable nested interrupts or exceptions.
2531 @cindex NMI handler functions on the Blackfin processor
2532 Use this attribute on the Blackfin to indicate that the specified function
2533 is an NMI handler. The compiler will generate function entry and
2534 exit sequences suitable for use in an NMI handler when this
2535 attribute is present.
2537 @item no_instrument_function
2538 @cindex @code{no_instrument_function} function attribute
2539 @opindex finstrument-functions
2540 If @option{-finstrument-functions} is given, profiling function calls will
2541 be generated at entry and exit of most user-compiled functions.
2542 Functions with this attribute will not be so instrumented.
2545 @cindex @code{noinline} function attribute
2546 This function attribute prevents a function from being considered for
2548 @c Don't enumerate the optimizations by name here; we try to be
2549 @c future-compatible with this mechanism.
2550 If the function does not have side-effects, there are optimizations
2551 other than inlining that causes function calls to be optimized away,
2552 although the function call is live. To keep such calls from being
2557 (@pxref{Extended Asm}) in the called function, to serve as a special
2560 @item nonnull (@var{arg-index}, @dots{})
2561 @cindex @code{nonnull} function attribute
2562 The @code{nonnull} attribute specifies that some function parameters should
2563 be non-null pointers. For instance, the declaration:
2567 my_memcpy (void *dest, const void *src, size_t len)
2568 __attribute__((nonnull (1, 2)));
2572 causes the compiler to check that, in calls to @code{my_memcpy},
2573 arguments @var{dest} and @var{src} are non-null. If the compiler
2574 determines that a null pointer is passed in an argument slot marked
2575 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2576 is issued. The compiler may also choose to make optimizations based
2577 on the knowledge that certain function arguments will not be null.
2579 If no argument index list is given to the @code{nonnull} attribute,
2580 all pointer arguments are marked as non-null. To illustrate, the
2581 following declaration is equivalent to the previous example:
2585 my_memcpy (void *dest, const void *src, size_t len)
2586 __attribute__((nonnull));
2590 @cindex @code{noreturn} function attribute
2591 A few standard library functions, such as @code{abort} and @code{exit},
2592 cannot return. GCC knows this automatically. Some programs define
2593 their own functions that never return. You can declare them
2594 @code{noreturn} to tell the compiler this fact. For example,
2598 void fatal () __attribute__ ((noreturn));
2601 fatal (/* @r{@dots{}} */)
2603 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2609 The @code{noreturn} keyword tells the compiler to assume that
2610 @code{fatal} cannot return. It can then optimize without regard to what
2611 would happen if @code{fatal} ever did return. This makes slightly
2612 better code. More importantly, it helps avoid spurious warnings of
2613 uninitialized variables.
2615 The @code{noreturn} keyword does not affect the exceptional path when that
2616 applies: a @code{noreturn}-marked function may still return to the caller
2617 by throwing an exception or calling @code{longjmp}.
2619 Do not assume that registers saved by the calling function are
2620 restored before calling the @code{noreturn} function.
2622 It does not make sense for a @code{noreturn} function to have a return
2623 type other than @code{void}.
2625 The attribute @code{noreturn} is not implemented in GCC versions
2626 earlier than 2.5. An alternative way to declare that a function does
2627 not return, which works in the current version and in some older
2628 versions, is as follows:
2631 typedef void voidfn ();
2633 volatile voidfn fatal;
2636 This approach does not work in GNU C++.
2639 @cindex @code{nothrow} function attribute
2640 The @code{nothrow} attribute is used to inform the compiler that a
2641 function cannot throw an exception. For example, most functions in
2642 the standard C library can be guaranteed not to throw an exception
2643 with the notable exceptions of @code{qsort} and @code{bsearch} that
2644 take function pointer arguments. The @code{nothrow} attribute is not
2645 implemented in GCC versions earlier than 3.3.
2648 @cindex @code{pure} function attribute
2649 Many functions have no effects except the return value and their
2650 return value depends only on the parameters and/or global variables.
2651 Such a function can be subject
2652 to common subexpression elimination and loop optimization just as an
2653 arithmetic operator would be. These functions should be declared
2654 with the attribute @code{pure}. For example,
2657 int square (int) __attribute__ ((pure));
2661 says that the hypothetical function @code{square} is safe to call
2662 fewer times than the program says.
2664 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2665 Interesting non-pure functions are functions with infinite loops or those
2666 depending on volatile memory or other system resource, that may change between
2667 two consecutive calls (such as @code{feof} in a multithreading environment).
2669 The attribute @code{pure} is not implemented in GCC versions earlier
2673 @cindex @code{hot} function attribute
2674 The @code{hot} attribute is used to inform the compiler that a function is a
2675 hot spot of the compiled program. The function is optimized more aggressively
2676 and on many target it is placed into special subsection of the text section so
2677 all hot functions appears close together improving locality.
2679 When profile feedback is available, via @option{-fprofile-use}, hot functions
2680 are automatically detected and this attribute is ignored.
2682 The @code{hot} attribute is not implemented in GCC versions earlier than 4.3.
2685 @cindex @code{cold} function attribute
2686 The @code{cold} attribute is used to inform the compiler that a function is
2687 unlikely executed. The function is optimized for size rather than speed and on
2688 many targets it is placed into special subsection of the text section so all
2689 cold functions appears close together improving code locality of non-cold parts
2690 of program. The paths leading to call of cold functions within code are marked
2691 as unlikely by the branch prediction mechanism. It is thus useful to mark
2692 functions used to handle unlikely conditions, such as @code{perror}, as cold to
2693 improve optimization of hot functions that do call marked functions in rare
2696 When profile feedback is available, via @option{-fprofile-use}, hot functions
2697 are automatically detected and this attribute is ignored.
2699 The @code{hot} attribute is not implemented in GCC versions earlier than 4.3.
2701 @item regparm (@var{number})
2702 @cindex @code{regparm} attribute
2703 @cindex functions that are passed arguments in registers on the 386
2704 On the Intel 386, the @code{regparm} attribute causes the compiler to
2705 pass arguments number one to @var{number} if they are of integral type
2706 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2707 take a variable number of arguments will continue to be passed all of their
2708 arguments on the stack.
2710 Beware that on some ELF systems this attribute is unsuitable for
2711 global functions in shared libraries with lazy binding (which is the
2712 default). Lazy binding will send the first call via resolving code in
2713 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2714 per the standard calling conventions. Solaris 8 is affected by this.
2715 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2716 safe since the loaders there save all registers. (Lazy binding can be
2717 disabled with the linker or the loader if desired, to avoid the
2721 @cindex @code{sseregparm} attribute
2722 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2723 causes the compiler to pass up to 3 floating point arguments in
2724 SSE registers instead of on the stack. Functions that take a
2725 variable number of arguments will continue to pass all of their
2726 floating point arguments on the stack.
2728 @item force_align_arg_pointer
2729 @cindex @code{force_align_arg_pointer} attribute
2730 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
2731 applied to individual function definitions, generating an alternate
2732 prologue and epilogue that realigns the runtime stack. This supports
2733 mixing legacy codes that run with a 4-byte aligned stack with modern
2734 codes that keep a 16-byte stack for SSE compatibility. The alternate
2735 prologue and epilogue are slower and bigger than the regular ones, and
2736 the alternate prologue requires a scratch register; this lowers the
2737 number of registers available if used in conjunction with the
2738 @code{regparm} attribute. The @code{force_align_arg_pointer}
2739 attribute is incompatible with nested functions; this is considered a
2743 @cindex @code{resbank} attribute
2744 On the SH2A target, this attribute enables the high-speed register
2745 saving and restoration using a register bank for @code{interrupt_handler}
2746 routines. Saving to the bank is performed automatcially after the CPU
2747 accepts an interrupt that uses a register bank.
2749 The nineteen 32-bit registers comprising general register R0 to R14,
2750 control register GBR, and system registers MACH, MACL, and PR and the
2751 vector table address offset are saved into a register bank. Register
2752 banks are stacked in first-in last-out (FILO) sequence. Restoration
2753 from the bank is executed by issuing a RESBANK instruction.
2756 @cindex @code{returns_twice} attribute
2757 The @code{returns_twice} attribute tells the compiler that a function may
2758 return more than one time. The compiler will ensure that all registers
2759 are dead before calling such a function and will emit a warning about
2760 the variables that may be clobbered after the second return from the
2761 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2762 The @code{longjmp}-like counterpart of such function, if any, might need
2763 to be marked with the @code{noreturn} attribute.
2766 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2767 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2768 all registers except the stack pointer should be saved in the prologue
2769 regardless of whether they are used or not.
2771 @item section ("@var{section-name}")
2772 @cindex @code{section} function attribute
2773 Normally, the compiler places the code it generates in the @code{text} section.
2774 Sometimes, however, you need additional sections, or you need certain
2775 particular functions to appear in special sections. The @code{section}
2776 attribute specifies that a function lives in a particular section.
2777 For example, the declaration:
2780 extern void foobar (void) __attribute__ ((section ("bar")));
2784 puts the function @code{foobar} in the @code{bar} section.
2786 Some file formats do not support arbitrary sections so the @code{section}
2787 attribute is not available on all platforms.
2788 If you need to map the entire contents of a module to a particular
2789 section, consider using the facilities of the linker instead.
2792 @cindex @code{sentinel} function attribute
2793 This function attribute ensures that a parameter in a function call is
2794 an explicit @code{NULL}. The attribute is only valid on variadic
2795 functions. By default, the sentinel is located at position zero, the
2796 last parameter of the function call. If an optional integer position
2797 argument P is supplied to the attribute, the sentinel must be located at
2798 position P counting backwards from the end of the argument list.
2801 __attribute__ ((sentinel))
2803 __attribute__ ((sentinel(0)))
2806 The attribute is automatically set with a position of 0 for the built-in
2807 functions @code{execl} and @code{execlp}. The built-in function
2808 @code{execle} has the attribute set with a position of 1.
2810 A valid @code{NULL} in this context is defined as zero with any pointer
2811 type. If your system defines the @code{NULL} macro with an integer type
2812 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2813 with a copy that redefines NULL appropriately.
2815 The warnings for missing or incorrect sentinels are enabled with
2819 See long_call/short_call.
2822 See longcall/shortcall.
2825 @cindex signal handler functions on the AVR processors
2826 Use this attribute on the AVR to indicate that the specified
2827 function is a signal handler. The compiler will generate function
2828 entry and exit sequences suitable for use in a signal handler when this
2829 attribute is present. Interrupts will be disabled inside the function.
2832 Use this attribute on the SH to indicate an @code{interrupt_handler}
2833 function should switch to an alternate stack. It expects a string
2834 argument that names a global variable holding the address of the
2839 void f () __attribute__ ((interrupt_handler,
2840 sp_switch ("alt_stack")));
2844 @cindex functions that pop the argument stack on the 386
2845 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2846 assume that the called function will pop off the stack space used to
2847 pass arguments, unless it takes a variable number of arguments.
2850 @cindex tiny data section on the H8/300H and H8S
2851 Use this attribute on the H8/300H and H8S to indicate that the specified
2852 variable should be placed into the tiny data section.
2853 The compiler will generate more efficient code for loads and stores
2854 on data in the tiny data section. Note the tiny data area is limited to
2855 slightly under 32kbytes of data.
2858 Use this attribute on the SH for an @code{interrupt_handler} to return using
2859 @code{trapa} instead of @code{rte}. This attribute expects an integer
2860 argument specifying the trap number to be used.
2863 @cindex @code{unused} attribute.
2864 This attribute, attached to a function, means that the function is meant
2865 to be possibly unused. GCC will not produce a warning for this
2869 @cindex @code{used} attribute.
2870 This attribute, attached to a function, means that code must be emitted
2871 for the function even if it appears that the function is not referenced.
2872 This is useful, for example, when the function is referenced only in
2876 @cindex @code{version_id} attribute on IA64 HP-UX
2877 This attribute, attached to a global variable or function, renames a
2878 symbol to contain a version string, thus allowing for function level
2879 versioning. HP-UX system header files may use version level functioning
2880 for some system calls.
2883 extern int foo () __attribute__((version_id ("20040821")));
2886 Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}.
2888 @item visibility ("@var{visibility_type}")
2889 @cindex @code{visibility} attribute
2890 This attribute affects the linkage of the declaration to which it is attached.
2891 There are four supported @var{visibility_type} values: default,
2892 hidden, protected or internal visibility.
2895 void __attribute__ ((visibility ("protected")))
2896 f () @{ /* @r{Do something.} */; @}
2897 int i __attribute__ ((visibility ("hidden")));
2900 The possible values of @var{visibility_type} correspond to the
2901 visibility settings in the ELF gABI.
2904 @c keep this list of visibilities in alphabetical order.
2907 Default visibility is the normal case for the object file format.
2908 This value is available for the visibility attribute to override other
2909 options that may change the assumed visibility of entities.
2911 On ELF, default visibility means that the declaration is visible to other
2912 modules and, in shared libraries, means that the declared entity may be
2915 On Darwin, default visibility means that the declaration is visible to
2918 Default visibility corresponds to ``external linkage'' in the language.
2921 Hidden visibility indicates that the entity declared will have a new
2922 form of linkage, which we'll call ``hidden linkage''. Two
2923 declarations of an object with hidden linkage refer to the same object
2924 if they are in the same shared object.
2927 Internal visibility is like hidden visibility, but with additional
2928 processor specific semantics. Unless otherwise specified by the
2929 psABI, GCC defines internal visibility to mean that a function is
2930 @emph{never} called from another module. Compare this with hidden
2931 functions which, while they cannot be referenced directly by other
2932 modules, can be referenced indirectly via function pointers. By
2933 indicating that a function cannot be called from outside the module,
2934 GCC may for instance omit the load of a PIC register since it is known
2935 that the calling function loaded the correct value.
2938 Protected visibility is like default visibility except that it
2939 indicates that references within the defining module will bind to the
2940 definition in that module. That is, the declared entity cannot be
2941 overridden by another module.
2945 All visibilities are supported on many, but not all, ELF targets
2946 (supported when the assembler supports the @samp{.visibility}
2947 pseudo-op). Default visibility is supported everywhere. Hidden
2948 visibility is supported on Darwin targets.
2950 The visibility attribute should be applied only to declarations which
2951 would otherwise have external linkage. The attribute should be applied
2952 consistently, so that the same entity should not be declared with
2953 different settings of the attribute.
2955 In C++, the visibility attribute applies to types as well as functions
2956 and objects, because in C++ types have linkage. A class must not have
2957 greater visibility than its non-static data member types and bases,
2958 and class members default to the visibility of their class. Also, a
2959 declaration without explicit visibility is limited to the visibility
2962 In C++, you can mark member functions and static member variables of a
2963 class with the visibility attribute. This is useful if if you know a
2964 particular method or static member variable should only be used from
2965 one shared object; then you can mark it hidden while the rest of the
2966 class has default visibility. Care must be taken to avoid breaking
2967 the One Definition Rule; for example, it is usually not useful to mark
2968 an inline method as hidden without marking the whole class as hidden.
2970 A C++ namespace declaration can also have the visibility attribute.
2971 This attribute applies only to the particular namespace body, not to
2972 other definitions of the same namespace; it is equivalent to using
2973 @samp{#pragma GCC visibility} before and after the namespace
2974 definition (@pxref{Visibility Pragmas}).
2976 In C++, if a template argument has limited visibility, this
2977 restriction is implicitly propagated to the template instantiation.
2978 Otherwise, template instantiations and specializations default to the
2979 visibility of their template.
2981 If both the template and enclosing class have explicit visibility, the
2982 visibility from the template is used.
2984 @item warn_unused_result
2985 @cindex @code{warn_unused_result} attribute
2986 The @code{warn_unused_result} attribute causes a warning to be emitted
2987 if a caller of the function with this attribute does not use its
2988 return value. This is useful for functions where not checking
2989 the result is either a security problem or always a bug, such as
2993 int fn () __attribute__ ((warn_unused_result));
2996 if (fn () < 0) return -1;
3002 results in warning on line 5.
3005 @cindex @code{weak} attribute
3006 The @code{weak} attribute causes the declaration to be emitted as a weak
3007 symbol rather than a global. This is primarily useful in defining
3008 library functions which can be overridden in user code, though it can
3009 also be used with non-function declarations. Weak symbols are supported
3010 for ELF targets, and also for a.out targets when using the GNU assembler
3014 @itemx weakref ("@var{target}")
3015 @cindex @code{weakref} attribute
3016 The @code{weakref} attribute marks a declaration as a weak reference.
3017 Without arguments, it should be accompanied by an @code{alias} attribute
3018 naming the target symbol. Optionally, the @var{target} may be given as
3019 an argument to @code{weakref} itself. In either case, @code{weakref}
3020 implicitly marks the declaration as @code{weak}. Without a
3021 @var{target}, given as an argument to @code{weakref} or to @code{alias},
3022 @code{weakref} is equivalent to @code{weak}.
3025 static int x() __attribute__ ((weakref ("y")));
3026 /* is equivalent to... */
3027 static int x() __attribute__ ((weak, weakref, alias ("y")));
3029 static int x() __attribute__ ((weakref));
3030 static int x() __attribute__ ((alias ("y")));
3033 A weak reference is an alias that does not by itself require a
3034 definition to be given for the target symbol. If the target symbol is
3035 only referenced through weak references, then the becomes a @code{weak}
3036 undefined symbol. If it is directly referenced, however, then such
3037 strong references prevail, and a definition will be required for the
3038 symbol, not necessarily in the same translation unit.
3040 The effect is equivalent to moving all references to the alias to a
3041 separate translation unit, renaming the alias to the aliased symbol,
3042 declaring it as weak, compiling the two separate translation units and
3043 performing a reloadable link on them.
3045 At present, a declaration to which @code{weakref} is attached can
3046 only be @code{static}.
3048 @item externally_visible
3049 @cindex @code{externally_visible} attribute.
3050 This attribute, attached to a global variable or function nullify
3051 effect of @option{-fwhole-program} command line option, so the object
3052 remain visible outside the current compilation unit
3056 You can specify multiple attributes in a declaration by separating them
3057 by commas within the double parentheses or by immediately following an
3058 attribute declaration with another attribute declaration.
3060 @cindex @code{#pragma}, reason for not using
3061 @cindex pragma, reason for not using
3062 Some people object to the @code{__attribute__} feature, suggesting that
3063 ISO C's @code{#pragma} should be used instead. At the time
3064 @code{__attribute__} was designed, there were two reasons for not doing
3069 It is impossible to generate @code{#pragma} commands from a macro.
3072 There is no telling what the same @code{#pragma} might mean in another
3076 These two reasons applied to almost any application that might have been
3077 proposed for @code{#pragma}. It was basically a mistake to use
3078 @code{#pragma} for @emph{anything}.
3080 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
3081 to be generated from macros. In addition, a @code{#pragma GCC}
3082 namespace is now in use for GCC-specific pragmas. However, it has been
3083 found convenient to use @code{__attribute__} to achieve a natural
3084 attachment of attributes to their corresponding declarations, whereas
3085 @code{#pragma GCC} is of use for constructs that do not naturally form
3086 part of the grammar. @xref{Other Directives,,Miscellaneous
3087 Preprocessing Directives, cpp, The GNU C Preprocessor}.
3089 @node Attribute Syntax
3090 @section Attribute Syntax
3091 @cindex attribute syntax
3093 This section describes the syntax with which @code{__attribute__} may be
3094 used, and the constructs to which attribute specifiers bind, for the C
3095 language. Some details may vary for C++ and Objective-C@. Because of
3096 infelicities in the grammar for attributes, some forms described here
3097 may not be successfully parsed in all cases.
3099 There are some problems with the semantics of attributes in C++. For
3100 example, there are no manglings for attributes, although they may affect
3101 code generation, so problems may arise when attributed types are used in
3102 conjunction with templates or overloading. Similarly, @code{typeid}
3103 does not distinguish between types with different attributes. Support
3104 for attributes in C++ may be restricted in future to attributes on
3105 declarations only, but not on nested declarators.
3107 @xref{Function Attributes}, for details of the semantics of attributes
3108 applying to functions. @xref{Variable Attributes}, for details of the
3109 semantics of attributes applying to variables. @xref{Type Attributes},
3110 for details of the semantics of attributes applying to structure, union
3111 and enumerated types.
3113 An @dfn{attribute specifier} is of the form
3114 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
3115 is a possibly empty comma-separated sequence of @dfn{attributes}, where
3116 each attribute is one of the following:
3120 Empty. Empty attributes are ignored.
3123 A word (which may be an identifier such as @code{unused}, or a reserved
3124 word such as @code{const}).
3127 A word, followed by, in parentheses, parameters for the attribute.
3128 These parameters take one of the following forms:
3132 An identifier. For example, @code{mode} attributes use this form.
3135 An identifier followed by a comma and a non-empty comma-separated list
3136 of expressions. For example, @code{format} attributes use this form.
3139 A possibly empty comma-separated list of expressions. For example,
3140 @code{format_arg} attributes use this form with the list being a single
3141 integer constant expression, and @code{alias} attributes use this form
3142 with the list being a single string constant.
3146 An @dfn{attribute specifier list} is a sequence of one or more attribute
3147 specifiers, not separated by any other tokens.
3149 In GNU C, an attribute specifier list may appear after the colon following a
3150 label, other than a @code{case} or @code{default} label. The only
3151 attribute it makes sense to use after a label is @code{unused}. This
3152 feature is intended for code generated by programs which contains labels
3153 that may be unused but which is compiled with @option{-Wall}. It would
3154 not normally be appropriate to use in it human-written code, though it
3155 could be useful in cases where the code that jumps to the label is
3156 contained within an @code{#ifdef} conditional. GNU C++ does not permit
3157 such placement of attribute lists, as it is permissible for a
3158 declaration, which could begin with an attribute list, to be labelled in
3159 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
3160 does not arise there.
3162 An attribute specifier list may appear as part of a @code{struct},
3163 @code{union} or @code{enum} specifier. It may go either immediately
3164 after the @code{struct}, @code{union} or @code{enum} keyword, or after
3165 the closing brace. The former syntax is preferred.
3166 Where attribute specifiers follow the closing brace, they are considered
3167 to relate to the structure, union or enumerated type defined, not to any
3168 enclosing declaration the type specifier appears in, and the type
3169 defined is not complete until after the attribute specifiers.
3170 @c Otherwise, there would be the following problems: a shift/reduce
3171 @c conflict between attributes binding the struct/union/enum and
3172 @c binding to the list of specifiers/qualifiers; and "aligned"
3173 @c attributes could use sizeof for the structure, but the size could be
3174 @c changed later by "packed" attributes.
3176 Otherwise, an attribute specifier appears as part of a declaration,
3177 counting declarations of unnamed parameters and type names, and relates
3178 to that declaration (which may be nested in another declaration, for
3179 example in the case of a parameter declaration), or to a particular declarator
3180 within a declaration. Where an
3181 attribute specifier is applied to a parameter declared as a function or
3182 an array, it should apply to the function or array rather than the
3183 pointer to which the parameter is implicitly converted, but this is not
3184 yet correctly implemented.
3186 Any list of specifiers and qualifiers at the start of a declaration may
3187 contain attribute specifiers, whether or not such a list may in that
3188 context contain storage class specifiers. (Some attributes, however,
3189 are essentially in the nature of storage class specifiers, and only make
3190 sense where storage class specifiers may be used; for example,
3191 @code{section}.) There is one necessary limitation to this syntax: the
3192 first old-style parameter declaration in a function definition cannot
3193 begin with an attribute specifier, because such an attribute applies to
3194 the function instead by syntax described below (which, however, is not
3195 yet implemented in this case). In some other cases, attribute
3196 specifiers are permitted by this grammar but not yet supported by the
3197 compiler. All attribute specifiers in this place relate to the
3198 declaration as a whole. In the obsolescent usage where a type of
3199 @code{int} is implied by the absence of type specifiers, such a list of
3200 specifiers and qualifiers may be an attribute specifier list with no
3201 other specifiers or qualifiers.
3203 At present, the first parameter in a function prototype must have some
3204 type specifier which is not an attribute specifier; this resolves an
3205 ambiguity in the interpretation of @code{void f(int
3206 (__attribute__((foo)) x))}, but is subject to change. At present, if
3207 the parentheses of a function declarator contain only attributes then
3208 those attributes are ignored, rather than yielding an error or warning
3209 or implying a single parameter of type int, but this is subject to
3212 An attribute specifier list may appear immediately before a declarator
3213 (other than the first) in a comma-separated list of declarators in a
3214 declaration of more than one identifier using a single list of
3215 specifiers and qualifiers. Such attribute specifiers apply
3216 only to the identifier before whose declarator they appear. For
3220 __attribute__((noreturn)) void d0 (void),
3221 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
3226 the @code{noreturn} attribute applies to all the functions
3227 declared; the @code{format} attribute only applies to @code{d1}.
3229 An attribute specifier list may appear immediately before the comma,
3230 @code{=} or semicolon terminating the declaration of an identifier other
3231 than a function definition. Such attribute specifiers apply
3232 to the declared object or function. Where an
3233 assembler name for an object or function is specified (@pxref{Asm
3234 Labels}), the attribute must follow the @code{asm}
3237 An attribute specifier list may, in future, be permitted to appear after
3238 the declarator in a function definition (before any old-style parameter
3239 declarations or the function body).
3241 Attribute specifiers may be mixed with type qualifiers appearing inside
3242 the @code{[]} of a parameter array declarator, in the C99 construct by
3243 which such qualifiers are applied to the pointer to which the array is
3244 implicitly converted. Such attribute specifiers apply to the pointer,
3245 not to the array, but at present this is not implemented and they are
3248 An attribute specifier list may appear at the start of a nested
3249 declarator. At present, there are some limitations in this usage: the
3250 attributes correctly apply to the declarator, but for most individual
3251 attributes the semantics this implies are not implemented.
3252 When attribute specifiers follow the @code{*} of a pointer
3253 declarator, they may be mixed with any type qualifiers present.
3254 The following describes the formal semantics of this syntax. It will make the
3255 most sense if you are familiar with the formal specification of
3256 declarators in the ISO C standard.
3258 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
3259 D1}, where @code{T} contains declaration specifiers that specify a type
3260 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
3261 contains an identifier @var{ident}. The type specified for @var{ident}
3262 for derived declarators whose type does not include an attribute
3263 specifier is as in the ISO C standard.
3265 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
3266 and the declaration @code{T D} specifies the type
3267 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3268 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3269 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
3271 If @code{D1} has the form @code{*
3272 @var{type-qualifier-and-attribute-specifier-list} D}, and the
3273 declaration @code{T D} specifies the type
3274 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3275 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3276 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
3282 void (__attribute__((noreturn)) ****f) (void);
3286 specifies the type ``pointer to pointer to pointer to pointer to
3287 non-returning function returning @code{void}''. As another example,
3290 char *__attribute__((aligned(8))) *f;
3294 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
3295 Note again that this does not work with most attributes; for example,
3296 the usage of @samp{aligned} and @samp{noreturn} attributes given above
3297 is not yet supported.
3299 For compatibility with existing code written for compiler versions that
3300 did not implement attributes on nested declarators, some laxity is
3301 allowed in the placing of attributes. If an attribute that only applies
3302 to types is applied to a declaration, it will be treated as applying to
3303 the type of that declaration. If an attribute that only applies to
3304 declarations is applied to the type of a declaration, it will be treated
3305 as applying to that declaration; and, for compatibility with code
3306 placing the attributes immediately before the identifier declared, such
3307 an attribute applied to a function return type will be treated as
3308 applying to the function type, and such an attribute applied to an array
3309 element type will be treated as applying to the array type. If an
3310 attribute that only applies to function types is applied to a
3311 pointer-to-function type, it will be treated as applying to the pointer
3312 target type; if such an attribute is applied to a function return type
3313 that is not a pointer-to-function type, it will be treated as applying
3314 to the function type.
3316 @node Function Prototypes
3317 @section Prototypes and Old-Style Function Definitions
3318 @cindex function prototype declarations
3319 @cindex old-style function definitions
3320 @cindex promotion of formal parameters
3322 GNU C extends ISO C to allow a function prototype to override a later
3323 old-style non-prototype definition. Consider the following example:
3326 /* @r{Use prototypes unless the compiler is old-fashioned.} */
3333 /* @r{Prototype function declaration.} */
3334 int isroot P((uid_t));
3336 /* @r{Old-style function definition.} */
3338 isroot (x) /* @r{??? lossage here ???} */
3345 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
3346 not allow this example, because subword arguments in old-style
3347 non-prototype definitions are promoted. Therefore in this example the
3348 function definition's argument is really an @code{int}, which does not
3349 match the prototype argument type of @code{short}.
3351 This restriction of ISO C makes it hard to write code that is portable
3352 to traditional C compilers, because the programmer does not know
3353 whether the @code{uid_t} type is @code{short}, @code{int}, or
3354 @code{long}. Therefore, in cases like these GNU C allows a prototype
3355 to override a later old-style definition. More precisely, in GNU C, a
3356 function prototype argument type overrides the argument type specified
3357 by a later old-style definition if the former type is the same as the
3358 latter type before promotion. Thus in GNU C the above example is
3359 equivalent to the following:
3372 GNU C++ does not support old-style function definitions, so this
3373 extension is irrelevant.
3376 @section C++ Style Comments
3378 @cindex C++ comments
3379 @cindex comments, C++ style
3381 In GNU C, you may use C++ style comments, which start with @samp{//} and
3382 continue until the end of the line. Many other C implementations allow
3383 such comments, and they are included in the 1999 C standard. However,
3384 C++ style comments are not recognized if you specify an @option{-std}
3385 option specifying a version of ISO C before C99, or @option{-ansi}
3386 (equivalent to @option{-std=c89}).
3389 @section Dollar Signs in Identifier Names
3391 @cindex dollar signs in identifier names
3392 @cindex identifier names, dollar signs in
3394 In GNU C, you may normally use dollar signs in identifier names.
3395 This is because many traditional C implementations allow such identifiers.
3396 However, dollar signs in identifiers are not supported on a few target
3397 machines, typically because the target assembler does not allow them.
3399 @node Character Escapes
3400 @section The Character @key{ESC} in Constants
3402 You can use the sequence @samp{\e} in a string or character constant to
3403 stand for the ASCII character @key{ESC}.
3406 @section Inquiring on Alignment of Types or Variables
3408 @cindex type alignment
3409 @cindex variable alignment
3411 The keyword @code{__alignof__} allows you to inquire about how an object
3412 is aligned, or the minimum alignment usually required by a type. Its
3413 syntax is just like @code{sizeof}.
3415 For example, if the target machine requires a @code{double} value to be
3416 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
3417 This is true on many RISC machines. On more traditional machine
3418 designs, @code{__alignof__ (double)} is 4 or even 2.
3420 Some machines never actually require alignment; they allow reference to any
3421 data type even at an odd address. For these machines, @code{__alignof__}
3422 reports the smallest alignment that GCC will give the data type, usually as
3423 mandated by the target ABI.
3425 If the operand of @code{__alignof__} is an lvalue rather than a type,
3426 its value is the required alignment for its type, taking into account
3427 any minimum alignment specified with GCC's @code{__attribute__}
3428 extension (@pxref{Variable Attributes}). For example, after this
3432 struct foo @{ int x; char y; @} foo1;
3436 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
3437 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
3439 It is an error to ask for the alignment of an incomplete type.
3441 @node Variable Attributes
3442 @section Specifying Attributes of Variables
3443 @cindex attribute of variables
3444 @cindex variable attributes
3446 The keyword @code{__attribute__} allows you to specify special
3447 attributes of variables or structure fields. This keyword is followed
3448 by an attribute specification inside double parentheses. Some
3449 attributes are currently defined generically for variables.
3450 Other attributes are defined for variables on particular target
3451 systems. Other attributes are available for functions
3452 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
3453 Other front ends might define more attributes
3454 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
3456 You may also specify attributes with @samp{__} preceding and following
3457 each keyword. This allows you to use them in header files without
3458 being concerned about a possible macro of the same name. For example,
3459 you may use @code{__aligned__} instead of @code{aligned}.
3461 @xref{Attribute Syntax}, for details of the exact syntax for using
3465 @cindex @code{aligned} attribute
3466 @item aligned (@var{alignment})
3467 This attribute specifies a minimum alignment for the variable or
3468 structure field, measured in bytes. For example, the declaration:
3471 int x __attribute__ ((aligned (16))) = 0;
3475 causes the compiler to allocate the global variable @code{x} on a
3476 16-byte boundary. On a 68040, this could be used in conjunction with
3477 an @code{asm} expression to access the @code{move16} instruction which
3478 requires 16-byte aligned operands.
3480 You can also specify the alignment of structure fields. For example, to
3481 create a double-word aligned @code{int} pair, you could write:
3484 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
3488 This is an alternative to creating a union with a @code{double} member
3489 that forces the union to be double-word aligned.
3491 As in the preceding examples, you can explicitly specify the alignment
3492 (in bytes) that you wish the compiler to use for a given variable or
3493 structure field. Alternatively, you can leave out the alignment factor
3494 and just ask the compiler to align a variable or field to the maximum
3495 useful alignment for the target machine you are compiling for. For
3496 example, you could write:
3499 short array[3] __attribute__ ((aligned));
3502 Whenever you leave out the alignment factor in an @code{aligned} attribute
3503 specification, the compiler automatically sets the alignment for the declared
3504 variable or field to the largest alignment which is ever used for any data
3505 type on the target machine you are compiling for. Doing this can often make
3506 copy operations more efficient, because the compiler can use whatever
3507 instructions copy the biggest chunks of memory when performing copies to
3508 or from the variables or fields that you have aligned this way.
3510 When used on a struct, or struct member, the @code{aligned} attribute can
3511 only increase the alignment; in order to decrease it, the @code{packed}
3512 attribute must be specified as well. When used as part of a typedef, the
3513 @code{aligned} attribute can both increase and decrease alignment, and
3514 specifying the @code{packed} attribute will generate a warning.
3516 Note that the effectiveness of @code{aligned} attributes may be limited
3517 by inherent limitations in your linker. On many systems, the linker is
3518 only able to arrange for variables to be aligned up to a certain maximum
3519 alignment. (For some linkers, the maximum supported alignment may
3520 be very very small.) If your linker is only able to align variables
3521 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3522 in an @code{__attribute__} will still only provide you with 8 byte
3523 alignment. See your linker documentation for further information.
3525 The @code{aligned} attribute can also be used for functions
3526 (@pxref{Function Attributes}.)
3528 @item cleanup (@var{cleanup_function})
3529 @cindex @code{cleanup} attribute
3530 The @code{cleanup} attribute runs a function when the variable goes
3531 out of scope. This attribute can only be applied to auto function
3532 scope variables; it may not be applied to parameters or variables
3533 with static storage duration. The function must take one parameter,
3534 a pointer to a type compatible with the variable. The return value
3535 of the function (if any) is ignored.
3537 If @option{-fexceptions} is enabled, then @var{cleanup_function}
3538 will be run during the stack unwinding that happens during the
3539 processing of the exception. Note that the @code{cleanup} attribute
3540 does not allow the exception to be caught, only to perform an action.
3541 It is undefined what happens if @var{cleanup_function} does not
3546 @cindex @code{common} attribute
3547 @cindex @code{nocommon} attribute
3550 The @code{common} attribute requests GCC to place a variable in
3551 ``common'' storage. The @code{nocommon} attribute requests the
3552 opposite---to allocate space for it directly.
3554 These attributes override the default chosen by the
3555 @option{-fno-common} and @option{-fcommon} flags respectively.
3558 @cindex @code{deprecated} attribute
3559 The @code{deprecated} attribute results in a warning if the variable
3560 is used anywhere in the source file. This is useful when identifying
3561 variables that are expected to be removed in a future version of a
3562 program. The warning also includes the location of the declaration
3563 of the deprecated variable, to enable users to easily find further
3564 information about why the variable is deprecated, or what they should
3565 do instead. Note that the warning only occurs for uses:
3568 extern int old_var __attribute__ ((deprecated));
3570 int new_fn () @{ return old_var; @}
3573 results in a warning on line 3 but not line 2.
3575 The @code{deprecated} attribute can also be used for functions and
3576 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3578 @item mode (@var{mode})
3579 @cindex @code{mode} attribute
3580 This attribute specifies the data type for the declaration---whichever
3581 type corresponds to the mode @var{mode}. This in effect lets you
3582 request an integer or floating point type according to its width.
3584 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3585 indicate the mode corresponding to a one-byte integer, @samp{word} or
3586 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3587 or @samp{__pointer__} for the mode used to represent pointers.
3590 @cindex @code{packed} attribute
3591 The @code{packed} attribute specifies that a variable or structure field
3592 should have the smallest possible alignment---one byte for a variable,
3593 and one bit for a field, unless you specify a larger value with the
3594 @code{aligned} attribute.
3596 Here is a structure in which the field @code{x} is packed, so that it
3597 immediately follows @code{a}:
3603 int x[2] __attribute__ ((packed));
3607 @item section ("@var{section-name}")
3608 @cindex @code{section} variable attribute
3609 Normally, the compiler places the objects it generates in sections like
3610 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3611 or you need certain particular variables to appear in special sections,
3612 for example to map to special hardware. The @code{section}
3613 attribute specifies that a variable (or function) lives in a particular
3614 section. For example, this small program uses several specific section names:
3617 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3618 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3619 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3620 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3624 /* @r{Initialize stack pointer} */
3625 init_sp (stack + sizeof (stack));
3627 /* @r{Initialize initialized data} */
3628 memcpy (&init_data, &data, &edata - &data);
3630 /* @r{Turn on the serial ports} */
3637 Use the @code{section} attribute with an @emph{initialized} definition
3638 of a @emph{global} variable, as shown in the example. GCC issues
3639 a warning and otherwise ignores the @code{section} attribute in
3640 uninitialized variable declarations.
3642 You may only use the @code{section} attribute with a fully initialized
3643 global definition because of the way linkers work. The linker requires
3644 each object be defined once, with the exception that uninitialized
3645 variables tentatively go in the @code{common} (or @code{bss}) section
3646 and can be multiply ``defined''. You can force a variable to be
3647 initialized with the @option{-fno-common} flag or the @code{nocommon}
3650 Some file formats do not support arbitrary sections so the @code{section}
3651 attribute is not available on all platforms.
3652 If you need to map the entire contents of a module to a particular
3653 section, consider using the facilities of the linker instead.
3656 @cindex @code{shared} variable attribute
3657 On Microsoft Windows, in addition to putting variable definitions in a named
3658 section, the section can also be shared among all running copies of an
3659 executable or DLL@. For example, this small program defines shared data
3660 by putting it in a named section @code{shared} and marking the section
3664 int foo __attribute__((section ("shared"), shared)) = 0;
3669 /* @r{Read and write foo. All running
3670 copies see the same value.} */
3676 You may only use the @code{shared} attribute along with @code{section}
3677 attribute with a fully initialized global definition because of the way
3678 linkers work. See @code{section} attribute for more information.
3680 The @code{shared} attribute is only available on Microsoft Windows@.
3682 @item tls_model ("@var{tls_model}")
3683 @cindex @code{tls_model} attribute
3684 The @code{tls_model} attribute sets thread-local storage model
3685 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3686 overriding @option{-ftls-model=} command line switch on a per-variable
3688 The @var{tls_model} argument should be one of @code{global-dynamic},
3689 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3691 Not all targets support this attribute.
3694 This attribute, attached to a variable, means that the variable is meant
3695 to be possibly unused. GCC will not produce a warning for this
3699 This attribute, attached to a variable, means that the variable must be
3700 emitted even if it appears that the variable is not referenced.
3702 @item vector_size (@var{bytes})
3703 This attribute specifies the vector size for the variable, measured in
3704 bytes. For example, the declaration:
3707 int foo __attribute__ ((vector_size (16)));
3711 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3712 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3713 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3715 This attribute is only applicable to integral and float scalars,
3716 although arrays, pointers, and function return values are allowed in
3717 conjunction with this construct.
3719 Aggregates with this attribute are invalid, even if they are of the same
3720 size as a corresponding scalar. For example, the declaration:
3723 struct S @{ int a; @};
3724 struct S __attribute__ ((vector_size (16))) foo;
3728 is invalid even if the size of the structure is the same as the size of
3732 The @code{selectany} attribute causes an initialized global variable to
3733 have link-once semantics. When multiple definitions of the variable are
3734 encountered by the linker, the first is selected and the remainder are
3735 discarded. Following usage by the Microsoft compiler, the linker is told
3736 @emph{not} to warn about size or content differences of the multiple
3739 Although the primary usage of this attribute is for POD types, the
3740 attribute can also be applied to global C++ objects that are initialized
3741 by a constructor. In this case, the static initialization and destruction
3742 code for the object is emitted in each translation defining the object,
3743 but the calls to the constructor and destructor are protected by a
3744 link-once guard variable.
3746 The @code{selectany} attribute is only available on Microsoft Windows
3747 targets. You can use @code{__declspec (selectany)} as a synonym for
3748 @code{__attribute__ ((selectany))} for compatibility with other
3752 The @code{weak} attribute is described in @xref{Function Attributes}.
3755 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3758 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3762 @subsection Blackfin Variable Attributes
3764 Three attributes are currently defined for the Blackfin.
3770 @cindex @code{l1_data} variable attribute
3771 @cindex @code{l1_data_A} variable attribute
3772 @cindex @code{l1_data_B} variable attribute
3773 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
3774 Variables with @code{l1_data} attribute will be put into the specific section
3775 named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into
3776 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
3777 attribute will be put into the specific section named @code{.l1.data.B}.
3780 @subsection M32R/D Variable Attributes
3782 One attribute is currently defined for the M32R/D@.
3785 @item model (@var{model-name})
3786 @cindex variable addressability on the M32R/D
3787 Use this attribute on the M32R/D to set the addressability of an object.
3788 The identifier @var{model-name} is one of @code{small}, @code{medium},
3789 or @code{large}, representing each of the code models.
3791 Small model objects live in the lower 16MB of memory (so that their
3792 addresses can be loaded with the @code{ld24} instruction).
3794 Medium and large model objects may live anywhere in the 32-bit address space
3795 (the compiler will generate @code{seth/add3} instructions to load their
3799 @anchor{i386 Variable Attributes}
3800 @subsection i386 Variable Attributes
3802 Two attributes are currently defined for i386 configurations:
3803 @code{ms_struct} and @code{gcc_struct}
3808 @cindex @code{ms_struct} attribute
3809 @cindex @code{gcc_struct} attribute
3811 If @code{packed} is used on a structure, or if bit-fields are used
3812 it may be that the Microsoft ABI packs them differently
3813 than GCC would normally pack them. Particularly when moving packed
3814 data between functions compiled with GCC and the native Microsoft compiler
3815 (either via function call or as data in a file), it may be necessary to access
3818 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3819 compilers to match the native Microsoft compiler.
3821 The Microsoft structure layout algorithm is fairly simple with the exception
3822 of the bitfield packing:
3824 The padding and alignment of members of structures and whether a bit field
3825 can straddle a storage-unit boundary
3828 @item Structure members are stored sequentially in the order in which they are
3829 declared: the first member has the lowest memory address and the last member
3832 @item Every data object has an alignment-requirement. The alignment-requirement
3833 for all data except structures, unions, and arrays is either the size of the
3834 object or the current packing size (specified with either the aligned attribute
3835 or the pack pragma), whichever is less. For structures, unions, and arrays,
3836 the alignment-requirement is the largest alignment-requirement of its members.
3837 Every object is allocated an offset so that:
3839 offset % alignment-requirement == 0
3841 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
3842 unit if the integral types are the same size and if the next bit field fits
3843 into the current allocation unit without crossing the boundary imposed by the
3844 common alignment requirements of the bit fields.
3847 Handling of zero-length bitfields:
3849 MSVC interprets zero-length bitfields in the following ways:
3852 @item If a zero-length bitfield is inserted between two bitfields that would
3853 normally be coalesced, the bitfields will not be coalesced.
3860 unsigned long bf_1 : 12;
3862 unsigned long bf_2 : 12;
3866 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
3867 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
3869 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
3870 alignment of the zero-length bitfield is greater than the member that follows it,
3871 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
3891 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
3892 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
3893 bitfield will not affect the alignment of @code{bar} or, as a result, the size
3896 Taking this into account, it is important to note the following:
3899 @item If a zero-length bitfield follows a normal bitfield, the type of the
3900 zero-length bitfield may affect the alignment of the structure as whole. For
3901 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
3902 normal bitfield, and is of type short.
3904 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
3905 still affect the alignment of the structure:
3915 Here, @code{t4} will take up 4 bytes.
3918 @item Zero-length bitfields following non-bitfield members are ignored:
3929 Here, @code{t5} will take up 2 bytes.
3933 @subsection PowerPC Variable Attributes
3935 Three attributes currently are defined for PowerPC configurations:
3936 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
3938 For full documentation of the struct attributes please see the
3939 documentation in the @xref{i386 Variable Attributes}, section.
3941 For documentation of @code{altivec} attribute please see the
3942 documentation in the @xref{PowerPC Type Attributes}, section.
3944 @subsection SPU Variable Attributes
3946 The SPU supports the @code{spu_vector} attribute for variables. For
3947 documentation of this attribute please see the documentation in the
3948 @xref{SPU Type Attributes}, section.
3950 @subsection Xstormy16 Variable Attributes
3952 One attribute is currently defined for xstormy16 configurations:
3957 @cindex @code{below100} attribute
3959 If a variable has the @code{below100} attribute (@code{BELOW100} is
3960 allowed also), GCC will place the variable in the first 0x100 bytes of
3961 memory and use special opcodes to access it. Such variables will be
3962 placed in either the @code{.bss_below100} section or the
3963 @code{.data_below100} section.
3967 @subsection AVR Variable Attributes
3971 @cindex @code{progmem} variable attribute
3972 The @code{progmem} attribute is used on the AVR to place data in the Program
3973 Memory address space. The AVR is a Harvard Architecture processor and data
3974 normally resides in the Data Memory address space.
3977 @node Type Attributes
3978 @section Specifying Attributes of Types
3979 @cindex attribute of types
3980 @cindex type attributes
3982 The keyword @code{__attribute__} allows you to specify special
3983 attributes of @code{struct} and @code{union} types when you define
3984 such types. This keyword is followed by an attribute specification
3985 inside double parentheses. Seven attributes are currently defined for
3986 types: @code{aligned}, @code{packed}, @code{transparent_union},
3987 @code{unused}, @code{deprecated}, @code{visibility}, and
3988 @code{may_alias}. Other attributes are defined for functions
3989 (@pxref{Function Attributes}) and for variables (@pxref{Variable
3992 You may also specify any one of these attributes with @samp{__}
3993 preceding and following its keyword. This allows you to use these
3994 attributes in header files without being concerned about a possible
3995 macro of the same name. For example, you may use @code{__aligned__}
3996 instead of @code{aligned}.
3998 You may specify type attributes in an enum, struct or union type
3999 declaration or definition, or for other types in a @code{typedef}
4002 For an enum, struct or union type, you may specify attributes either
4003 between the enum, struct or union tag and the name of the type, or
4004 just past the closing curly brace of the @emph{definition}. The
4005 former syntax is preferred.
4007 @xref{Attribute Syntax}, for details of the exact syntax for using
4011 @cindex @code{aligned} attribute
4012 @item aligned (@var{alignment})
4013 This attribute specifies a minimum alignment (in bytes) for variables
4014 of the specified type. For example, the declarations:
4017 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
4018 typedef int more_aligned_int __attribute__ ((aligned (8)));
4022 force the compiler to insure (as far as it can) that each variable whose
4023 type is @code{struct S} or @code{more_aligned_int} will be allocated and
4024 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
4025 variables of type @code{struct S} aligned to 8-byte boundaries allows
4026 the compiler to use the @code{ldd} and @code{std} (doubleword load and
4027 store) instructions when copying one variable of type @code{struct S} to
4028 another, thus improving run-time efficiency.
4030 Note that the alignment of any given @code{struct} or @code{union} type
4031 is required by the ISO C standard to be at least a perfect multiple of
4032 the lowest common multiple of the alignments of all of the members of
4033 the @code{struct} or @code{union} in question. This means that you @emph{can}
4034 effectively adjust the alignment of a @code{struct} or @code{union}
4035 type by attaching an @code{aligned} attribute to any one of the members
4036 of such a type, but the notation illustrated in the example above is a
4037 more obvious, intuitive, and readable way to request the compiler to
4038 adjust the alignment of an entire @code{struct} or @code{union} type.
4040 As in the preceding example, you can explicitly specify the alignment
4041 (in bytes) that you wish the compiler to use for a given @code{struct}
4042 or @code{union} type. Alternatively, you can leave out the alignment factor
4043 and just ask the compiler to align a type to the maximum
4044 useful alignment for the target machine you are compiling for. For
4045 example, you could write:
4048 struct S @{ short f[3]; @} __attribute__ ((aligned));
4051 Whenever you leave out the alignment factor in an @code{aligned}
4052 attribute specification, the compiler automatically sets the alignment
4053 for the type to the largest alignment which is ever used for any data
4054 type on the target machine you are compiling for. Doing this can often
4055 make copy operations more efficient, because the compiler can use
4056 whatever instructions copy the biggest chunks of memory when performing
4057 copies to or from the variables which have types that you have aligned
4060 In the example above, if the size of each @code{short} is 2 bytes, then
4061 the size of the entire @code{struct S} type is 6 bytes. The smallest
4062 power of two which is greater than or equal to that is 8, so the
4063 compiler sets the alignment for the entire @code{struct S} type to 8
4066 Note that although you can ask the compiler to select a time-efficient
4067 alignment for a given type and then declare only individual stand-alone
4068 objects of that type, the compiler's ability to select a time-efficient
4069 alignment is primarily useful only when you plan to create arrays of
4070 variables having the relevant (efficiently aligned) type. If you
4071 declare or use arrays of variables of an efficiently-aligned type, then
4072 it is likely that your program will also be doing pointer arithmetic (or
4073 subscripting, which amounts to the same thing) on pointers to the
4074 relevant type, and the code that the compiler generates for these
4075 pointer arithmetic operations will often be more efficient for
4076 efficiently-aligned types than for other types.
4078 The @code{aligned} attribute can only increase the alignment; but you
4079 can decrease it by specifying @code{packed} as well. See below.
4081 Note that the effectiveness of @code{aligned} attributes may be limited
4082 by inherent limitations in your linker. On many systems, the linker is
4083 only able to arrange for variables to be aligned up to a certain maximum
4084 alignment. (For some linkers, the maximum supported alignment may
4085 be very very small.) If your linker is only able to align variables
4086 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4087 in an @code{__attribute__} will still only provide you with 8 byte
4088 alignment. See your linker documentation for further information.
4091 This attribute, attached to @code{struct} or @code{union} type
4092 definition, specifies that each member (other than zero-width bitfields)
4093 of the structure or union is placed to minimize the memory required. When
4094 attached to an @code{enum} definition, it indicates that the smallest
4095 integral type should be used.
4097 @opindex fshort-enums
4098 Specifying this attribute for @code{struct} and @code{union} types is
4099 equivalent to specifying the @code{packed} attribute on each of the
4100 structure or union members. Specifying the @option{-fshort-enums}
4101 flag on the line is equivalent to specifying the @code{packed}
4102 attribute on all @code{enum} definitions.
4104 In the following example @code{struct my_packed_struct}'s members are
4105 packed closely together, but the internal layout of its @code{s} member
4106 is not packed---to do that, @code{struct my_unpacked_struct} would need to
4110 struct my_unpacked_struct
4116 struct __attribute__ ((__packed__)) my_packed_struct
4120 struct my_unpacked_struct s;
4124 You may only specify this attribute on the definition of a @code{enum},
4125 @code{struct} or @code{union}, not on a @code{typedef} which does not
4126 also define the enumerated type, structure or union.
4128 @item transparent_union
4129 This attribute, attached to a @code{union} type definition, indicates
4130 that any function parameter having that union type causes calls to that
4131 function to be treated in a special way.
4133 First, the argument corresponding to a transparent union type can be of
4134 any type in the union; no cast is required. Also, if the union contains
4135 a pointer type, the corresponding argument can be a null pointer
4136 constant or a void pointer expression; and if the union contains a void
4137 pointer type, the corresponding argument can be any pointer expression.
4138 If the union member type is a pointer, qualifiers like @code{const} on
4139 the referenced type must be respected, just as with normal pointer
4142 Second, the argument is passed to the function using the calling
4143 conventions of the first member of the transparent union, not the calling
4144 conventions of the union itself. All members of the union must have the
4145 same machine representation; this is necessary for this argument passing
4148 Transparent unions are designed for library functions that have multiple
4149 interfaces for compatibility reasons. For example, suppose the
4150 @code{wait} function must accept either a value of type @code{int *} to
4151 comply with Posix, or a value of type @code{union wait *} to comply with
4152 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
4153 @code{wait} would accept both kinds of arguments, but it would also
4154 accept any other pointer type and this would make argument type checking
4155 less useful. Instead, @code{<sys/wait.h>} might define the interface
4159 typedef union __attribute__ ((__transparent_union__))
4163 @} wait_status_ptr_t;
4165 pid_t wait (wait_status_ptr_t);
4168 This interface allows either @code{int *} or @code{union wait *}
4169 arguments to be passed, using the @code{int *} calling convention.
4170 The program can call @code{wait} with arguments of either type:
4173 int w1 () @{ int w; return wait (&w); @}
4174 int w2 () @{ union wait w; return wait (&w); @}
4177 With this interface, @code{wait}'s implementation might look like this:
4180 pid_t wait (wait_status_ptr_t p)
4182 return waitpid (-1, p.__ip, 0);
4187 When attached to a type (including a @code{union} or a @code{struct}),
4188 this attribute means that variables of that type are meant to appear
4189 possibly unused. GCC will not produce a warning for any variables of
4190 that type, even if the variable appears to do nothing. This is often
4191 the case with lock or thread classes, which are usually defined and then
4192 not referenced, but contain constructors and destructors that have
4193 nontrivial bookkeeping functions.
4196 The @code{deprecated} attribute results in a warning if the type
4197 is used anywhere in the source file. This is useful when identifying
4198 types that are expected to be removed in a future version of a program.
4199 If possible, the warning also includes the location of the declaration
4200 of the deprecated type, to enable users to easily find further
4201 information about why the type is deprecated, or what they should do
4202 instead. Note that the warnings only occur for uses and then only
4203 if the type is being applied to an identifier that itself is not being
4204 declared as deprecated.
4207 typedef int T1 __attribute__ ((deprecated));
4211 typedef T1 T3 __attribute__ ((deprecated));
4212 T3 z __attribute__ ((deprecated));
4215 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
4216 warning is issued for line 4 because T2 is not explicitly
4217 deprecated. Line 5 has no warning because T3 is explicitly
4218 deprecated. Similarly for line 6.
4220 The @code{deprecated} attribute can also be used for functions and
4221 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
4224 Accesses to objects with types with this attribute are not subjected to
4225 type-based alias analysis, but are instead assumed to be able to alias
4226 any other type of objects, just like the @code{char} type. See
4227 @option{-fstrict-aliasing} for more information on aliasing issues.
4232 typedef short __attribute__((__may_alias__)) short_a;
4238 short_a *b = (short_a *) &a;
4242 if (a == 0x12345678)
4249 If you replaced @code{short_a} with @code{short} in the variable
4250 declaration, the above program would abort when compiled with
4251 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
4252 above in recent GCC versions.
4255 In C++, attribute visibility (@pxref{Function Attributes}) can also be
4256 applied to class, struct, union and enum types. Unlike other type
4257 attributes, the attribute must appear between the initial keyword and
4258 the name of the type; it cannot appear after the body of the type.
4260 Note that the type visibility is applied to vague linkage entities
4261 associated with the class (vtable, typeinfo node, etc.). In
4262 particular, if a class is thrown as an exception in one shared object
4263 and caught in another, the class must have default visibility.
4264 Otherwise the two shared objects will be unable to use the same
4265 typeinfo node and exception handling will break.
4267 @subsection ARM Type Attributes
4269 On those ARM targets that support @code{dllimport} (such as Symbian
4270 OS), you can use the @code{notshared} attribute to indicate that the
4271 virtual table and other similar data for a class should not be
4272 exported from a DLL@. For example:
4275 class __declspec(notshared) C @{
4277 __declspec(dllimport) C();
4281 __declspec(dllexport)
4285 In this code, @code{C::C} is exported from the current DLL, but the
4286 virtual table for @code{C} is not exported. (You can use
4287 @code{__attribute__} instead of @code{__declspec} if you prefer, but
4288 most Symbian OS code uses @code{__declspec}.)
4290 @anchor{i386 Type Attributes}
4291 @subsection i386 Type Attributes
4293 Two attributes are currently defined for i386 configurations:
4294 @code{ms_struct} and @code{gcc_struct}
4298 @cindex @code{ms_struct}
4299 @cindex @code{gcc_struct}
4301 If @code{packed} is used on a structure, or if bit-fields are used
4302 it may be that the Microsoft ABI packs them differently
4303 than GCC would normally pack them. Particularly when moving packed
4304 data between functions compiled with GCC and the native Microsoft compiler
4305 (either via function call or as data in a file), it may be necessary to access
4308 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4309 compilers to match the native Microsoft compiler.
4312 To specify multiple attributes, separate them by commas within the
4313 double parentheses: for example, @samp{__attribute__ ((aligned (16),
4316 @anchor{PowerPC Type Attributes}
4317 @subsection PowerPC Type Attributes
4319 Three attributes currently are defined for PowerPC configurations:
4320 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4322 For full documentation of the struct attributes please see the
4323 documentation in the @xref{i386 Type Attributes}, section.
4325 The @code{altivec} attribute allows one to declare AltiVec vector data
4326 types supported by the AltiVec Programming Interface Manual. The
4327 attribute requires an argument to specify one of three vector types:
4328 @code{vector__}, @code{pixel__} (always followed by unsigned short),
4329 and @code{bool__} (always followed by unsigned).
4332 __attribute__((altivec(vector__)))
4333 __attribute__((altivec(pixel__))) unsigned short
4334 __attribute__((altivec(bool__))) unsigned
4337 These attributes mainly are intended to support the @code{__vector},
4338 @code{__pixel}, and @code{__bool} AltiVec keywords.
4340 @anchor{SPU Type Attributes}
4341 @subsection SPU Type Attributes
4343 The SPU supports the @code{spu_vector} attribute for types. This attribute
4344 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
4345 Language Extensions Specification. It is intended to support the
4346 @code{__vector} keyword.
4350 @section An Inline Function is As Fast As a Macro
4351 @cindex inline functions
4352 @cindex integrating function code
4354 @cindex macros, inline alternative
4356 By declaring a function inline, you can direct GCC to make
4357 calls to that function faster. One way GCC can achieve this is to
4358 integrate that function's code into the code for its callers. This
4359 makes execution faster by eliminating the function-call overhead; in
4360 addition, if any of the actual argument values are constant, their
4361 known values may permit simplifications at compile time so that not
4362 all of the inline function's code needs to be included. The effect on
4363 code size is less predictable; object code may be larger or smaller
4364 with function inlining, depending on the particular case. You can
4365 also direct GCC to try to integrate all ``simple enough'' functions
4366 into their callers with the option @option{-finline-functions}.
4368 GCC implements three different semantics of declaring a function
4369 inline. One is available with @option{-std=gnu89} or
4370 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
4371 on all inline declarations, another when @option{-std=c99} or
4372 @option{-std=gnu99} (without @option{-fgnu89-inline}), and the third
4373 is used when compiling C++.
4375 To declare a function inline, use the @code{inline} keyword in its
4376 declaration, like this:
4386 If you are writing a header file to be included in ISO C89 programs, write
4387 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
4389 The three types of inlining behave similarly in two important cases:
4390 when the @code{inline} keyword is used on a @code{static} function,
4391 like the example above, and when a function is first declared without
4392 using the @code{inline} keyword and then is defined with
4393 @code{inline}, like this:
4396 extern int inc (int *a);
4404 In both of these common cases, the program behaves the same as if you
4405 had not used the @code{inline} keyword, except for its speed.
4407 @cindex inline functions, omission of
4408 @opindex fkeep-inline-functions
4409 When a function is both inline and @code{static}, if all calls to the
4410 function are integrated into the caller, and the function's address is
4411 never used, then the function's own assembler code is never referenced.
4412 In this case, GCC does not actually output assembler code for the
4413 function, unless you specify the option @option{-fkeep-inline-functions}.
4414 Some calls cannot be integrated for various reasons (in particular,
4415 calls that precede the function's definition cannot be integrated, and
4416 neither can recursive calls within the definition). If there is a
4417 nonintegrated call, then the function is compiled to assembler code as
4418 usual. The function must also be compiled as usual if the program
4419 refers to its address, because that can't be inlined.
4422 Note that certain usages in a function definition can make it unsuitable
4423 for inline substitution. Among these usages are: use of varargs, use of
4424 alloca, use of variable sized data types (@pxref{Variable Length}),
4425 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
4426 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
4427 will warn when a function marked @code{inline} could not be substituted,
4428 and will give the reason for the failure.
4430 @cindex automatic @code{inline} for C++ member fns
4431 @cindex @code{inline} automatic for C++ member fns
4432 @cindex member fns, automatically @code{inline}
4433 @cindex C++ member fns, automatically @code{inline}
4434 @opindex fno-default-inline
4435 As required by ISO C++, GCC considers member functions defined within
4436 the body of a class to be marked inline even if they are
4437 not explicitly declared with the @code{inline} keyword. You can
4438 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
4439 Options,,Options Controlling C++ Dialect}.
4441 GCC does not inline any functions when not optimizing unless you specify
4442 the @samp{always_inline} attribute for the function, like this:
4445 /* @r{Prototype.} */
4446 inline void foo (const char) __attribute__((always_inline));
4449 The remainder of this section is specific to GNU C89 inlining.
4451 @cindex non-static inline function
4452 When an inline function is not @code{static}, then the compiler must assume
4453 that there may be calls from other source files; since a global symbol can
4454 be defined only once in any program, the function must not be defined in
4455 the other source files, so the calls therein cannot be integrated.
4456 Therefore, a non-@code{static} inline function is always compiled on its
4457 own in the usual fashion.
4459 If you specify both @code{inline} and @code{extern} in the function
4460 definition, then the definition is used only for inlining. In no case
4461 is the function compiled on its own, not even if you refer to its
4462 address explicitly. Such an address becomes an external reference, as
4463 if you had only declared the function, and had not defined it.
4465 This combination of @code{inline} and @code{extern} has almost the
4466 effect of a macro. The way to use it is to put a function definition in
4467 a header file with these keywords, and put another copy of the
4468 definition (lacking @code{inline} and @code{extern}) in a library file.
4469 The definition in the header file will cause most calls to the function
4470 to be inlined. If any uses of the function remain, they will refer to
4471 the single copy in the library.
4474 @section Assembler Instructions with C Expression Operands
4475 @cindex extended @code{asm}
4476 @cindex @code{asm} expressions
4477 @cindex assembler instructions
4480 In an assembler instruction using @code{asm}, you can specify the
4481 operands of the instruction using C expressions. This means you need not
4482 guess which registers or memory locations will contain the data you want
4485 You must specify an assembler instruction template much like what
4486 appears in a machine description, plus an operand constraint string for
4489 For example, here is how to use the 68881's @code{fsinx} instruction:
4492 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
4496 Here @code{angle} is the C expression for the input operand while
4497 @code{result} is that of the output operand. Each has @samp{"f"} as its
4498 operand constraint, saying that a floating point register is required.
4499 The @samp{=} in @samp{=f} indicates that the operand is an output; all
4500 output operands' constraints must use @samp{=}. The constraints use the
4501 same language used in the machine description (@pxref{Constraints}).
4503 Each operand is described by an operand-constraint string followed by
4504 the C expression in parentheses. A colon separates the assembler
4505 template from the first output operand and another separates the last
4506 output operand from the first input, if any. Commas separate the
4507 operands within each group. The total number of operands is currently
4508 limited to 30; this limitation may be lifted in some future version of
4511 If there are no output operands but there are input operands, you must
4512 place two consecutive colons surrounding the place where the output
4515 As of GCC version 3.1, it is also possible to specify input and output
4516 operands using symbolic names which can be referenced within the
4517 assembler code. These names are specified inside square brackets
4518 preceding the constraint string, and can be referenced inside the
4519 assembler code using @code{%[@var{name}]} instead of a percentage sign
4520 followed by the operand number. Using named operands the above example
4524 asm ("fsinx %[angle],%[output]"
4525 : [output] "=f" (result)
4526 : [angle] "f" (angle));
4530 Note that the symbolic operand names have no relation whatsoever to
4531 other C identifiers. You may use any name you like, even those of
4532 existing C symbols, but you must ensure that no two operands within the same
4533 assembler construct use the same symbolic name.
4535 Output operand expressions must be lvalues; the compiler can check this.
4536 The input operands need not be lvalues. The compiler cannot check
4537 whether the operands have data types that are reasonable for the
4538 instruction being executed. It does not parse the assembler instruction
4539 template and does not know what it means or even whether it is valid
4540 assembler input. The extended @code{asm} feature is most often used for
4541 machine instructions the compiler itself does not know exist. If
4542 the output expression cannot be directly addressed (for example, it is a
4543 bit-field), your constraint must allow a register. In that case, GCC
4544 will use the register as the output of the @code{asm}, and then store
4545 that register into the output.
4547 The ordinary output operands must be write-only; GCC will assume that
4548 the values in these operands before the instruction are dead and need
4549 not be generated. Extended asm supports input-output or read-write
4550 operands. Use the constraint character @samp{+} to indicate such an
4551 operand and list it with the output operands. You should only use
4552 read-write operands when the constraints for the operand (or the
4553 operand in which only some of the bits are to be changed) allow a
4556 You may, as an alternative, logically split its function into two
4557 separate operands, one input operand and one write-only output
4558 operand. The connection between them is expressed by constraints
4559 which say they need to be in the same location when the instruction
4560 executes. You can use the same C expression for both operands, or
4561 different expressions. For example, here we write the (fictitious)
4562 @samp{combine} instruction with @code{bar} as its read-only source
4563 operand and @code{foo} as its read-write destination:
4566 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
4570 The constraint @samp{"0"} for operand 1 says that it must occupy the
4571 same location as operand 0. A number in constraint is allowed only in
4572 an input operand and it must refer to an output operand.
4574 Only a number in the constraint can guarantee that one operand will be in
4575 the same place as another. The mere fact that @code{foo} is the value
4576 of both operands is not enough to guarantee that they will be in the
4577 same place in the generated assembler code. The following would not
4581 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
4584 Various optimizations or reloading could cause operands 0 and 1 to be in
4585 different registers; GCC knows no reason not to do so. For example, the
4586 compiler might find a copy of the value of @code{foo} in one register and
4587 use it for operand 1, but generate the output operand 0 in a different
4588 register (copying it afterward to @code{foo}'s own address). Of course,
4589 since the register for operand 1 is not even mentioned in the assembler
4590 code, the result will not work, but GCC can't tell that.
4592 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
4593 the operand number for a matching constraint. For example:
4596 asm ("cmoveq %1,%2,%[result]"
4597 : [result] "=r"(result)
4598 : "r" (test), "r"(new), "[result]"(old));
4601 Sometimes you need to make an @code{asm} operand be a specific register,
4602 but there's no matching constraint letter for that register @emph{by
4603 itself}. To force the operand into that register, use a local variable
4604 for the operand and specify the register in the variable declaration.
4605 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
4606 register constraint letter that matches the register:
4609 register int *p1 asm ("r0") = @dots{};
4610 register int *p2 asm ("r1") = @dots{};
4611 register int *result asm ("r0");
4612 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4615 @anchor{Example of asm with clobbered asm reg}
4616 In the above example, beware that a register that is call-clobbered by
4617 the target ABI will be overwritten by any function call in the
4618 assignment, including library calls for arithmetic operators.
4619 Assuming it is a call-clobbered register, this may happen to @code{r0}
4620 above by the assignment to @code{p2}. If you have to use such a
4621 register, use temporary variables for expressions between the register
4626 register int *p1 asm ("r0") = @dots{};
4627 register int *p2 asm ("r1") = t1;
4628 register int *result asm ("r0");
4629 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4632 Some instructions clobber specific hard registers. To describe this,
4633 write a third colon after the input operands, followed by the names of
4634 the clobbered hard registers (given as strings). Here is a realistic
4635 example for the VAX:
4638 asm volatile ("movc3 %0,%1,%2"
4639 : /* @r{no outputs} */
4640 : "g" (from), "g" (to), "g" (count)
4641 : "r0", "r1", "r2", "r3", "r4", "r5");
4644 You may not write a clobber description in a way that overlaps with an
4645 input or output operand. For example, you may not have an operand
4646 describing a register class with one member if you mention that register
4647 in the clobber list. Variables declared to live in specific registers
4648 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
4649 have no part mentioned in the clobber description.
4650 There is no way for you to specify that an input
4651 operand is modified without also specifying it as an output
4652 operand. Note that if all the output operands you specify are for this
4653 purpose (and hence unused), you will then also need to specify
4654 @code{volatile} for the @code{asm} construct, as described below, to
4655 prevent GCC from deleting the @code{asm} statement as unused.
4657 If you refer to a particular hardware register from the assembler code,
4658 you will probably have to list the register after the third colon to
4659 tell the compiler the register's value is modified. In some assemblers,
4660 the register names begin with @samp{%}; to produce one @samp{%} in the
4661 assembler code, you must write @samp{%%} in the input.
4663 If your assembler instruction can alter the condition code register, add
4664 @samp{cc} to the list of clobbered registers. GCC on some machines
4665 represents the condition codes as a specific hardware register;
4666 @samp{cc} serves to name this register. On other machines, the
4667 condition code is handled differently, and specifying @samp{cc} has no
4668 effect. But it is valid no matter what the machine.
4670 If your assembler instructions access memory in an unpredictable
4671 fashion, add @samp{memory} to the list of clobbered registers. This
4672 will cause GCC to not keep memory values cached in registers across the
4673 assembler instruction and not optimize stores or loads to that memory.
4674 You will also want to add the @code{volatile} keyword if the memory
4675 affected is not listed in the inputs or outputs of the @code{asm}, as
4676 the @samp{memory} clobber does not count as a side-effect of the
4677 @code{asm}. If you know how large the accessed memory is, you can add
4678 it as input or output but if this is not known, you should add
4679 @samp{memory}. As an example, if you access ten bytes of a string, you
4680 can use a memory input like:
4683 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
4686 Note that in the following example the memory input is necessary,
4687 otherwise GCC might optimize the store to @code{x} away:
4694 asm ("magic stuff accessing an 'int' pointed to by '%1'"
4695 "=&d" (r) : "a" (y), "m" (*y));
4700 You can put multiple assembler instructions together in a single
4701 @code{asm} template, separated by the characters normally used in assembly
4702 code for the system. A combination that works in most places is a newline
4703 to break the line, plus a tab character to move to the instruction field
4704 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
4705 assembler allows semicolons as a line-breaking character. Note that some
4706 assembler dialects use semicolons to start a comment.
4707 The input operands are guaranteed not to use any of the clobbered
4708 registers, and neither will the output operands' addresses, so you can
4709 read and write the clobbered registers as many times as you like. Here
4710 is an example of multiple instructions in a template; it assumes the
4711 subroutine @code{_foo} accepts arguments in registers 9 and 10:
4714 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
4716 : "g" (from), "g" (to)
4720 Unless an output operand has the @samp{&} constraint modifier, GCC
4721 may allocate it in the same register as an unrelated input operand, on
4722 the assumption the inputs are consumed before the outputs are produced.
4723 This assumption may be false if the assembler code actually consists of
4724 more than one instruction. In such a case, use @samp{&} for each output
4725 operand that may not overlap an input. @xref{Modifiers}.
4727 If you want to test the condition code produced by an assembler
4728 instruction, you must include a branch and a label in the @code{asm}
4729 construct, as follows:
4732 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
4738 This assumes your assembler supports local labels, as the GNU assembler
4739 and most Unix assemblers do.
4741 Speaking of labels, jumps from one @code{asm} to another are not
4742 supported. The compiler's optimizers do not know about these jumps, and
4743 therefore they cannot take account of them when deciding how to
4746 @cindex macros containing @code{asm}
4747 Usually the most convenient way to use these @code{asm} instructions is to
4748 encapsulate them in macros that look like functions. For example,
4752 (@{ double __value, __arg = (x); \
4753 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
4758 Here the variable @code{__arg} is used to make sure that the instruction
4759 operates on a proper @code{double} value, and to accept only those
4760 arguments @code{x} which can convert automatically to a @code{double}.
4762 Another way to make sure the instruction operates on the correct data
4763 type is to use a cast in the @code{asm}. This is different from using a
4764 variable @code{__arg} in that it converts more different types. For
4765 example, if the desired type were @code{int}, casting the argument to
4766 @code{int} would accept a pointer with no complaint, while assigning the
4767 argument to an @code{int} variable named @code{__arg} would warn about
4768 using a pointer unless the caller explicitly casts it.
4770 If an @code{asm} has output operands, GCC assumes for optimization
4771 purposes the instruction has no side effects except to change the output
4772 operands. This does not mean instructions with a side effect cannot be
4773 used, but you must be careful, because the compiler may eliminate them
4774 if the output operands aren't used, or move them out of loops, or
4775 replace two with one if they constitute a common subexpression. Also,
4776 if your instruction does have a side effect on a variable that otherwise
4777 appears not to change, the old value of the variable may be reused later
4778 if it happens to be found in a register.
4780 You can prevent an @code{asm} instruction from being deleted
4781 by writing the keyword @code{volatile} after
4782 the @code{asm}. For example:
4785 #define get_and_set_priority(new) \
4787 asm volatile ("get_and_set_priority %0, %1" \
4788 : "=g" (__old) : "g" (new)); \
4793 The @code{volatile} keyword indicates that the instruction has
4794 important side-effects. GCC will not delete a volatile @code{asm} if
4795 it is reachable. (The instruction can still be deleted if GCC can
4796 prove that control-flow will never reach the location of the
4797 instruction.) Note that even a volatile @code{asm} instruction
4798 can be moved relative to other code, including across jump
4799 instructions. For example, on many targets there is a system
4800 register which can be set to control the rounding mode of
4801 floating point operations. You might try
4802 setting it with a volatile @code{asm}, like this PowerPC example:
4805 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
4810 This will not work reliably, as the compiler may move the addition back
4811 before the volatile @code{asm}. To make it work you need to add an
4812 artificial dependency to the @code{asm} referencing a variable in the code
4813 you don't want moved, for example:
4816 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
4820 Similarly, you can't expect a
4821 sequence of volatile @code{asm} instructions to remain perfectly
4822 consecutive. If you want consecutive output, use a single @code{asm}.
4823 Also, GCC will perform some optimizations across a volatile @code{asm}
4824 instruction; GCC does not ``forget everything'' when it encounters
4825 a volatile @code{asm} instruction the way some other compilers do.
4827 An @code{asm} instruction without any output operands will be treated
4828 identically to a volatile @code{asm} instruction.
4830 It is a natural idea to look for a way to give access to the condition
4831 code left by the assembler instruction. However, when we attempted to
4832 implement this, we found no way to make it work reliably. The problem
4833 is that output operands might need reloading, which would result in
4834 additional following ``store'' instructions. On most machines, these
4835 instructions would alter the condition code before there was time to
4836 test it. This problem doesn't arise for ordinary ``test'' and
4837 ``compare'' instructions because they don't have any output operands.
4839 For reasons similar to those described above, it is not possible to give
4840 an assembler instruction access to the condition code left by previous
4843 If you are writing a header file that should be includable in ISO C
4844 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
4847 @subsection Size of an @code{asm}
4849 Some targets require that GCC track the size of each instruction used in
4850 order to generate correct code. Because the final length of an
4851 @code{asm} is only known by the assembler, GCC must make an estimate as
4852 to how big it will be. The estimate is formed by counting the number of
4853 statements in the pattern of the @code{asm} and multiplying that by the
4854 length of the longest instruction on that processor. Statements in the
4855 @code{asm} are identified by newline characters and whatever statement
4856 separator characters are supported by the assembler; on most processors
4857 this is the `@code{;}' character.
4859 Normally, GCC's estimate is perfectly adequate to ensure that correct
4860 code is generated, but it is possible to confuse the compiler if you use
4861 pseudo instructions or assembler macros that expand into multiple real
4862 instructions or if you use assembler directives that expand to more
4863 space in the object file than would be needed for a single instruction.
4864 If this happens then the assembler will produce a diagnostic saying that
4865 a label is unreachable.
4867 @subsection i386 floating point asm operands
4869 There are several rules on the usage of stack-like regs in
4870 asm_operands insns. These rules apply only to the operands that are
4875 Given a set of input regs that die in an asm_operands, it is
4876 necessary to know which are implicitly popped by the asm, and
4877 which must be explicitly popped by gcc.
4879 An input reg that is implicitly popped by the asm must be
4880 explicitly clobbered, unless it is constrained to match an
4884 For any input reg that is implicitly popped by an asm, it is
4885 necessary to know how to adjust the stack to compensate for the pop.
4886 If any non-popped input is closer to the top of the reg-stack than
4887 the implicitly popped reg, it would not be possible to know what the
4888 stack looked like---it's not clear how the rest of the stack ``slides
4891 All implicitly popped input regs must be closer to the top of
4892 the reg-stack than any input that is not implicitly popped.
4894 It is possible that if an input dies in an insn, reload might
4895 use the input reg for an output reload. Consider this example:
4898 asm ("foo" : "=t" (a) : "f" (b));
4901 This asm says that input B is not popped by the asm, and that
4902 the asm pushes a result onto the reg-stack, i.e., the stack is one
4903 deeper after the asm than it was before. But, it is possible that
4904 reload will think that it can use the same reg for both the input and
4905 the output, if input B dies in this insn.
4907 If any input operand uses the @code{f} constraint, all output reg
4908 constraints must use the @code{&} earlyclobber.
4910 The asm above would be written as
4913 asm ("foo" : "=&t" (a) : "f" (b));
4917 Some operands need to be in particular places on the stack. All
4918 output operands fall in this category---there is no other way to
4919 know which regs the outputs appear in unless the user indicates
4920 this in the constraints.
4922 Output operands must specifically indicate which reg an output
4923 appears in after an asm. @code{=f} is not allowed: the operand
4924 constraints must select a class with a single reg.
4927 Output operands may not be ``inserted'' between existing stack regs.
4928 Since no 387 opcode uses a read/write operand, all output operands
4929 are dead before the asm_operands, and are pushed by the asm_operands.
4930 It makes no sense to push anywhere but the top of the reg-stack.
4932 Output operands must start at the top of the reg-stack: output
4933 operands may not ``skip'' a reg.
4936 Some asm statements may need extra stack space for internal
4937 calculations. This can be guaranteed by clobbering stack registers
4938 unrelated to the inputs and outputs.
4942 Here are a couple of reasonable asms to want to write. This asm
4943 takes one input, which is internally popped, and produces two outputs.
4946 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4949 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4950 and replaces them with one output. The user must code the @code{st(1)}
4951 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4954 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4960 @section Controlling Names Used in Assembler Code
4961 @cindex assembler names for identifiers
4962 @cindex names used in assembler code
4963 @cindex identifiers, names in assembler code
4965 You can specify the name to be used in the assembler code for a C
4966 function or variable by writing the @code{asm} (or @code{__asm__})
4967 keyword after the declarator as follows:
4970 int foo asm ("myfoo") = 2;
4974 This specifies that the name to be used for the variable @code{foo} in
4975 the assembler code should be @samp{myfoo} rather than the usual
4978 On systems where an underscore is normally prepended to the name of a C
4979 function or variable, this feature allows you to define names for the
4980 linker that do not start with an underscore.
4982 It does not make sense to use this feature with a non-static local
4983 variable since such variables do not have assembler names. If you are
4984 trying to put the variable in a particular register, see @ref{Explicit
4985 Reg Vars}. GCC presently accepts such code with a warning, but will
4986 probably be changed to issue an error, rather than a warning, in the
4989 You cannot use @code{asm} in this way in a function @emph{definition}; but
4990 you can get the same effect by writing a declaration for the function
4991 before its definition and putting @code{asm} there, like this:
4994 extern func () asm ("FUNC");
5001 It is up to you to make sure that the assembler names you choose do not
5002 conflict with any other assembler symbols. Also, you must not use a
5003 register name; that would produce completely invalid assembler code. GCC
5004 does not as yet have the ability to store static variables in registers.
5005 Perhaps that will be added.
5007 @node Explicit Reg Vars
5008 @section Variables in Specified Registers
5009 @cindex explicit register variables
5010 @cindex variables in specified registers
5011 @cindex specified registers
5012 @cindex registers, global allocation
5014 GNU C allows you to put a few global variables into specified hardware
5015 registers. You can also specify the register in which an ordinary
5016 register variable should be allocated.
5020 Global register variables reserve registers throughout the program.
5021 This may be useful in programs such as programming language
5022 interpreters which have a couple of global variables that are accessed
5026 Local register variables in specific registers do not reserve the
5027 registers, except at the point where they are used as input or output
5028 operands in an @code{asm} statement and the @code{asm} statement itself is
5029 not deleted. The compiler's data flow analysis is capable of determining
5030 where the specified registers contain live values, and where they are
5031 available for other uses. Stores into local register variables may be deleted
5032 when they appear to be dead according to dataflow analysis. References
5033 to local register variables may be deleted or moved or simplified.
5035 These local variables are sometimes convenient for use with the extended
5036 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
5037 output of the assembler instruction directly into a particular register.
5038 (This will work provided the register you specify fits the constraints
5039 specified for that operand in the @code{asm}.)
5047 @node Global Reg Vars
5048 @subsection Defining Global Register Variables
5049 @cindex global register variables
5050 @cindex registers, global variables in
5052 You can define a global register variable in GNU C like this:
5055 register int *foo asm ("a5");
5059 Here @code{a5} is the name of the register which should be used. Choose a
5060 register which is normally saved and restored by function calls on your
5061 machine, so that library routines will not clobber it.
5063 Naturally the register name is cpu-dependent, so you would need to
5064 conditionalize your program according to cpu type. The register
5065 @code{a5} would be a good choice on a 68000 for a variable of pointer
5066 type. On machines with register windows, be sure to choose a ``global''
5067 register that is not affected magically by the function call mechanism.
5069 In addition, operating systems on one type of cpu may differ in how they
5070 name the registers; then you would need additional conditionals. For
5071 example, some 68000 operating systems call this register @code{%a5}.
5073 Eventually there may be a way of asking the compiler to choose a register
5074 automatically, but first we need to figure out how it should choose and
5075 how to enable you to guide the choice. No solution is evident.
5077 Defining a global register variable in a certain register reserves that
5078 register entirely for this use, at least within the current compilation.
5079 The register will not be allocated for any other purpose in the functions
5080 in the current compilation. The register will not be saved and restored by
5081 these functions. Stores into this register are never deleted even if they
5082 would appear to be dead, but references may be deleted or moved or
5085 It is not safe to access the global register variables from signal
5086 handlers, or from more than one thread of control, because the system
5087 library routines may temporarily use the register for other things (unless
5088 you recompile them specially for the task at hand).
5090 @cindex @code{qsort}, and global register variables
5091 It is not safe for one function that uses a global register variable to
5092 call another such function @code{foo} by way of a third function
5093 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
5094 different source file in which the variable wasn't declared). This is
5095 because @code{lose} might save the register and put some other value there.
5096 For example, you can't expect a global register variable to be available in
5097 the comparison-function that you pass to @code{qsort}, since @code{qsort}
5098 might have put something else in that register. (If you are prepared to
5099 recompile @code{qsort} with the same global register variable, you can
5100 solve this problem.)
5102 If you want to recompile @code{qsort} or other source files which do not
5103 actually use your global register variable, so that they will not use that
5104 register for any other purpose, then it suffices to specify the compiler
5105 option @option{-ffixed-@var{reg}}. You need not actually add a global
5106 register declaration to their source code.
5108 A function which can alter the value of a global register variable cannot
5109 safely be called from a function compiled without this variable, because it
5110 could clobber the value the caller expects to find there on return.
5111 Therefore, the function which is the entry point into the part of the
5112 program that uses the global register variable must explicitly save and
5113 restore the value which belongs to its caller.
5115 @cindex register variable after @code{longjmp}
5116 @cindex global register after @code{longjmp}
5117 @cindex value after @code{longjmp}
5120 On most machines, @code{longjmp} will restore to each global register
5121 variable the value it had at the time of the @code{setjmp}. On some
5122 machines, however, @code{longjmp} will not change the value of global
5123 register variables. To be portable, the function that called @code{setjmp}
5124 should make other arrangements to save the values of the global register
5125 variables, and to restore them in a @code{longjmp}. This way, the same
5126 thing will happen regardless of what @code{longjmp} does.
5128 All global register variable declarations must precede all function
5129 definitions. If such a declaration could appear after function
5130 definitions, the declaration would be too late to prevent the register from
5131 being used for other purposes in the preceding functions.
5133 Global register variables may not have initial values, because an
5134 executable file has no means to supply initial contents for a register.
5136 On the SPARC, there are reports that g3 @dots{} g7 are suitable
5137 registers, but certain library functions, such as @code{getwd}, as well
5138 as the subroutines for division and remainder, modify g3 and g4. g1 and
5139 g2 are local temporaries.
5141 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
5142 Of course, it will not do to use more than a few of those.
5144 @node Local Reg Vars
5145 @subsection Specifying Registers for Local Variables
5146 @cindex local variables, specifying registers
5147 @cindex specifying registers for local variables
5148 @cindex registers for local variables
5150 You can define a local register variable with a specified register
5154 register int *foo asm ("a5");
5158 Here @code{a5} is the name of the register which should be used. Note
5159 that this is the same syntax used for defining global register
5160 variables, but for a local variable it would appear within a function.
5162 Naturally the register name is cpu-dependent, but this is not a
5163 problem, since specific registers are most often useful with explicit
5164 assembler instructions (@pxref{Extended Asm}). Both of these things
5165 generally require that you conditionalize your program according to
5168 In addition, operating systems on one type of cpu may differ in how they
5169 name the registers; then you would need additional conditionals. For
5170 example, some 68000 operating systems call this register @code{%a5}.
5172 Defining such a register variable does not reserve the register; it
5173 remains available for other uses in places where flow control determines
5174 the variable's value is not live.
5176 This option does not guarantee that GCC will generate code that has
5177 this variable in the register you specify at all times. You may not
5178 code an explicit reference to this register in the @emph{assembler
5179 instruction template} part of an @code{asm} statement and assume it will
5180 always refer to this variable. However, using the variable as an
5181 @code{asm} @emph{operand} guarantees that the specified register is used
5184 Stores into local register variables may be deleted when they appear to be dead
5185 according to dataflow analysis. References to local register variables may
5186 be deleted or moved or simplified.
5188 As for global register variables, it's recommended that you choose a
5189 register which is normally saved and restored by function calls on
5190 your machine, so that library routines will not clobber it. A common
5191 pitfall is to initialize multiple call-clobbered registers with
5192 arbitrary expressions, where a function call or library call for an
5193 arithmetic operator will overwrite a register value from a previous
5194 assignment, for example @code{r0} below:
5196 register int *p1 asm ("r0") = @dots{};
5197 register int *p2 asm ("r1") = @dots{};
5199 In those cases, a solution is to use a temporary variable for
5200 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
5202 @node Alternate Keywords
5203 @section Alternate Keywords
5204 @cindex alternate keywords
5205 @cindex keywords, alternate
5207 @option{-ansi} and the various @option{-std} options disable certain
5208 keywords. This causes trouble when you want to use GNU C extensions, or
5209 a general-purpose header file that should be usable by all programs,
5210 including ISO C programs. The keywords @code{asm}, @code{typeof} and
5211 @code{inline} are not available in programs compiled with
5212 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
5213 program compiled with @option{-std=c99}). The ISO C99 keyword
5214 @code{restrict} is only available when @option{-std=gnu99} (which will
5215 eventually be the default) or @option{-std=c99} (or the equivalent
5216 @option{-std=iso9899:1999}) is used.
5218 The way to solve these problems is to put @samp{__} at the beginning and
5219 end of each problematical keyword. For example, use @code{__asm__}
5220 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
5222 Other C compilers won't accept these alternative keywords; if you want to
5223 compile with another compiler, you can define the alternate keywords as
5224 macros to replace them with the customary keywords. It looks like this:
5232 @findex __extension__
5234 @option{-pedantic} and other options cause warnings for many GNU C extensions.
5236 prevent such warnings within one expression by writing
5237 @code{__extension__} before the expression. @code{__extension__} has no
5238 effect aside from this.
5240 @node Incomplete Enums
5241 @section Incomplete @code{enum} Types
5243 You can define an @code{enum} tag without specifying its possible values.
5244 This results in an incomplete type, much like what you get if you write
5245 @code{struct foo} without describing the elements. A later declaration
5246 which does specify the possible values completes the type.
5248 You can't allocate variables or storage using the type while it is
5249 incomplete. However, you can work with pointers to that type.
5251 This extension may not be very useful, but it makes the handling of
5252 @code{enum} more consistent with the way @code{struct} and @code{union}
5255 This extension is not supported by GNU C++.
5257 @node Function Names
5258 @section Function Names as Strings
5259 @cindex @code{__func__} identifier
5260 @cindex @code{__FUNCTION__} identifier
5261 @cindex @code{__PRETTY_FUNCTION__} identifier
5263 GCC provides three magic variables which hold the name of the current
5264 function, as a string. The first of these is @code{__func__}, which
5265 is part of the C99 standard:
5268 The identifier @code{__func__} is implicitly declared by the translator
5269 as if, immediately following the opening brace of each function
5270 definition, the declaration
5273 static const char __func__[] = "function-name";
5276 appeared, where function-name is the name of the lexically-enclosing
5277 function. This name is the unadorned name of the function.
5280 @code{__FUNCTION__} is another name for @code{__func__}. Older
5281 versions of GCC recognize only this name. However, it is not
5282 standardized. For maximum portability, we recommend you use
5283 @code{__func__}, but provide a fallback definition with the
5287 #if __STDC_VERSION__ < 199901L
5289 # define __func__ __FUNCTION__
5291 # define __func__ "<unknown>"
5296 In C, @code{__PRETTY_FUNCTION__} is yet another name for
5297 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
5298 the type signature of the function as well as its bare name. For
5299 example, this program:
5303 extern int printf (char *, ...);
5310 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
5311 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
5329 __PRETTY_FUNCTION__ = void a::sub(int)
5332 These identifiers are not preprocessor macros. In GCC 3.3 and
5333 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
5334 were treated as string literals; they could be used to initialize
5335 @code{char} arrays, and they could be concatenated with other string
5336 literals. GCC 3.4 and later treat them as variables, like
5337 @code{__func__}. In C++, @code{__FUNCTION__} and
5338 @code{__PRETTY_FUNCTION__} have always been variables.
5340 @node Return Address
5341 @section Getting the Return or Frame Address of a Function
5343 These functions may be used to get information about the callers of a
5346 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
5347 This function returns the return address of the current function, or of
5348 one of its callers. The @var{level} argument is number of frames to
5349 scan up the call stack. A value of @code{0} yields the return address
5350 of the current function, a value of @code{1} yields the return address
5351 of the caller of the current function, and so forth. When inlining
5352 the expected behavior is that the function will return the address of
5353 the function that will be returned to. To work around this behavior use
5354 the @code{noinline} function attribute.
5356 The @var{level} argument must be a constant integer.
5358 On some machines it may be impossible to determine the return address of
5359 any function other than the current one; in such cases, or when the top
5360 of the stack has been reached, this function will return @code{0} or a
5361 random value. In addition, @code{__builtin_frame_address} may be used
5362 to determine if the top of the stack has been reached.
5364 This function should only be used with a nonzero argument for debugging
5368 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
5369 This function is similar to @code{__builtin_return_address}, but it
5370 returns the address of the function frame rather than the return address
5371 of the function. Calling @code{__builtin_frame_address} with a value of
5372 @code{0} yields the frame address of the current function, a value of
5373 @code{1} yields the frame address of the caller of the current function,
5376 The frame is the area on the stack which holds local variables and saved
5377 registers. The frame address is normally the address of the first word
5378 pushed on to the stack by the function. However, the exact definition
5379 depends upon the processor and the calling convention. If the processor
5380 has a dedicated frame pointer register, and the function has a frame,
5381 then @code{__builtin_frame_address} will return the value of the frame
5384 On some machines it may be impossible to determine the frame address of
5385 any function other than the current one; in such cases, or when the top
5386 of the stack has been reached, this function will return @code{0} if
5387 the first frame pointer is properly initialized by the startup code.
5389 This function should only be used with a nonzero argument for debugging
5393 @node Vector Extensions
5394 @section Using vector instructions through built-in functions
5396 On some targets, the instruction set contains SIMD vector instructions that
5397 operate on multiple values contained in one large register at the same time.
5398 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
5401 The first step in using these extensions is to provide the necessary data
5402 types. This should be done using an appropriate @code{typedef}:
5405 typedef int v4si __attribute__ ((vector_size (16)));
5408 The @code{int} type specifies the base type, while the attribute specifies
5409 the vector size for the variable, measured in bytes. For example, the
5410 declaration above causes the compiler to set the mode for the @code{v4si}
5411 type to be 16 bytes wide and divided into @code{int} sized units. For
5412 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
5413 corresponding mode of @code{foo} will be @acronym{V4SI}.
5415 The @code{vector_size} attribute is only applicable to integral and
5416 float scalars, although arrays, pointers, and function return values
5417 are allowed in conjunction with this construct.
5419 All the basic integer types can be used as base types, both as signed
5420 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
5421 @code{long long}. In addition, @code{float} and @code{double} can be
5422 used to build floating-point vector types.
5424 Specifying a combination that is not valid for the current architecture
5425 will cause GCC to synthesize the instructions using a narrower mode.
5426 For example, if you specify a variable of type @code{V4SI} and your
5427 architecture does not allow for this specific SIMD type, GCC will
5428 produce code that uses 4 @code{SIs}.
5430 The types defined in this manner can be used with a subset of normal C
5431 operations. Currently, GCC will allow using the following operators
5432 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
5434 The operations behave like C++ @code{valarrays}. Addition is defined as
5435 the addition of the corresponding elements of the operands. For
5436 example, in the code below, each of the 4 elements in @var{a} will be
5437 added to the corresponding 4 elements in @var{b} and the resulting
5438 vector will be stored in @var{c}.
5441 typedef int v4si __attribute__ ((vector_size (16)));
5448 Subtraction, multiplication, division, and the logical operations
5449 operate in a similar manner. Likewise, the result of using the unary
5450 minus or complement operators on a vector type is a vector whose
5451 elements are the negative or complemented values of the corresponding
5452 elements in the operand.
5454 You can declare variables and use them in function calls and returns, as
5455 well as in assignments and some casts. You can specify a vector type as
5456 a return type for a function. Vector types can also be used as function
5457 arguments. It is possible to cast from one vector type to another,
5458 provided they are of the same size (in fact, you can also cast vectors
5459 to and from other datatypes of the same size).
5461 You cannot operate between vectors of different lengths or different
5462 signedness without a cast.
5464 A port that supports hardware vector operations, usually provides a set
5465 of built-in functions that can be used to operate on vectors. For
5466 example, a function to add two vectors and multiply the result by a
5467 third could look like this:
5470 v4si f (v4si a, v4si b, v4si c)
5472 v4si tmp = __builtin_addv4si (a, b);
5473 return __builtin_mulv4si (tmp, c);
5480 @findex __builtin_offsetof
5482 GCC implements for both C and C++ a syntactic extension to implement
5483 the @code{offsetof} macro.
5487 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
5489 offsetof_member_designator:
5491 | offsetof_member_designator "." @code{identifier}
5492 | offsetof_member_designator "[" @code{expr} "]"
5495 This extension is sufficient such that
5498 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
5501 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
5502 may be dependent. In either case, @var{member} may consist of a single
5503 identifier, or a sequence of member accesses and array references.
5505 @node Atomic Builtins
5506 @section Built-in functions for atomic memory access
5508 The following builtins are intended to be compatible with those described
5509 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
5510 section 7.4. As such, they depart from the normal GCC practice of using
5511 the ``__builtin_'' prefix, and further that they are overloaded such that
5512 they work on multiple types.
5514 The definition given in the Intel documentation allows only for the use of
5515 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
5516 counterparts. GCC will allow any integral scalar or pointer type that is
5517 1, 2, 4 or 8 bytes in length.
5519 Not all operations are supported by all target processors. If a particular
5520 operation cannot be implemented on the target processor, a warning will be
5521 generated and a call an external function will be generated. The external
5522 function will carry the same name as the builtin, with an additional suffix
5523 @samp{_@var{n}} where @var{n} is the size of the data type.
5525 @c ??? Should we have a mechanism to suppress this warning? This is almost
5526 @c useful for implementing the operation under the control of an external
5529 In most cases, these builtins are considered a @dfn{full barrier}. That is,
5530 no memory operand will be moved across the operation, either forward or
5531 backward. Further, instructions will be issued as necessary to prevent the
5532 processor from speculating loads across the operation and from queuing stores
5533 after the operation.
5535 All of the routines are are described in the Intel documentation to take
5536 ``an optional list of variables protected by the memory barrier''. It's
5537 not clear what is meant by that; it could mean that @emph{only} the
5538 following variables are protected, or it could mean that these variables
5539 should in addition be protected. At present GCC ignores this list and
5540 protects all variables which are globally accessible. If in the future
5541 we make some use of this list, an empty list will continue to mean all
5542 globally accessible variables.
5545 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
5546 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
5547 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
5548 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
5549 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
5550 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
5551 @findex __sync_fetch_and_add
5552 @findex __sync_fetch_and_sub
5553 @findex __sync_fetch_and_or
5554 @findex __sync_fetch_and_and
5555 @findex __sync_fetch_and_xor
5556 @findex __sync_fetch_and_nand
5557 These builtins perform the operation suggested by the name, and
5558 returns the value that had previously been in memory. That is,
5561 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
5562 @{ tmp = *ptr; *ptr = ~tmp & value; return tmp; @} // nand
5565 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
5566 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
5567 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
5568 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
5569 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
5570 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
5571 @findex __sync_add_and_fetch
5572 @findex __sync_sub_and_fetch
5573 @findex __sync_or_and_fetch
5574 @findex __sync_and_and_fetch
5575 @findex __sync_xor_and_fetch
5576 @findex __sync_nand_and_fetch
5577 These builtins perform the operation suggested by the name, and
5578 return the new value. That is,
5581 @{ *ptr @var{op}= value; return *ptr; @}
5582 @{ *ptr = ~*ptr & value; return *ptr; @} // nand
5585 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5586 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5587 @findex __sync_bool_compare_and_swap
5588 @findex __sync_val_compare_and_swap
5589 These builtins perform an atomic compare and swap. That is, if the current
5590 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
5593 The ``bool'' version returns true if the comparison is successful and
5594 @var{newval} was written. The ``val'' version returns the contents
5595 of @code{*@var{ptr}} before the operation.
5597 @item __sync_synchronize (...)
5598 @findex __sync_synchronize
5599 This builtin issues a full memory barrier.
5601 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
5602 @findex __sync_lock_test_and_set
5603 This builtin, as described by Intel, is not a traditional test-and-set
5604 operation, but rather an atomic exchange operation. It writes @var{value}
5605 into @code{*@var{ptr}}, and returns the previous contents of
5608 Many targets have only minimal support for such locks, and do not support
5609 a full exchange operation. In this case, a target may support reduced
5610 functionality here by which the @emph{only} valid value to store is the
5611 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
5612 is implementation defined.
5614 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
5615 This means that references after the builtin cannot move to (or be
5616 speculated to) before the builtin, but previous memory stores may not
5617 be globally visible yet, and previous memory loads may not yet be
5620 @item void __sync_lock_release (@var{type} *ptr, ...)
5621 @findex __sync_lock_release
5622 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
5623 Normally this means writing the constant 0 to @code{*@var{ptr}}.
5625 This builtin is not a full barrier, but rather a @dfn{release barrier}.
5626 This means that all previous memory stores are globally visible, and all
5627 previous memory loads have been satisfied, but following memory reads
5628 are not prevented from being speculated to before the barrier.
5631 @node Object Size Checking
5632 @section Object Size Checking Builtins
5633 @findex __builtin_object_size
5634 @findex __builtin___memcpy_chk
5635 @findex __builtin___mempcpy_chk
5636 @findex __builtin___memmove_chk
5637 @findex __builtin___memset_chk
5638 @findex __builtin___strcpy_chk
5639 @findex __builtin___stpcpy_chk
5640 @findex __builtin___strncpy_chk
5641 @findex __builtin___strcat_chk
5642 @findex __builtin___strncat_chk
5643 @findex __builtin___sprintf_chk
5644 @findex __builtin___snprintf_chk
5645 @findex __builtin___vsprintf_chk
5646 @findex __builtin___vsnprintf_chk
5647 @findex __builtin___printf_chk
5648 @findex __builtin___vprintf_chk
5649 @findex __builtin___fprintf_chk
5650 @findex __builtin___vfprintf_chk
5652 GCC implements a limited buffer overflow protection mechanism
5653 that can prevent some buffer overflow attacks.
5655 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
5656 is a built-in construct that returns a constant number of bytes from
5657 @var{ptr} to the end of the object @var{ptr} pointer points to
5658 (if known at compile time). @code{__builtin_object_size} never evaluates
5659 its arguments for side-effects. If there are any side-effects in them, it
5660 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5661 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
5662 point to and all of them are known at compile time, the returned number
5663 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
5664 0 and minimum if nonzero. If it is not possible to determine which objects
5665 @var{ptr} points to at compile time, @code{__builtin_object_size} should
5666 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5667 for @var{type} 2 or 3.
5669 @var{type} is an integer constant from 0 to 3. If the least significant
5670 bit is clear, objects are whole variables, if it is set, a closest
5671 surrounding subobject is considered the object a pointer points to.
5672 The second bit determines if maximum or minimum of remaining bytes
5676 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
5677 char *p = &var.buf1[1], *q = &var.b;
5679 /* Here the object p points to is var. */
5680 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
5681 /* The subobject p points to is var.buf1. */
5682 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
5683 /* The object q points to is var. */
5684 assert (__builtin_object_size (q, 0)
5685 == (char *) (&var + 1) - (char *) &var.b);
5686 /* The subobject q points to is var.b. */
5687 assert (__builtin_object_size (q, 1) == sizeof (var.b));
5691 There are built-in functions added for many common string operation
5692 functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk}
5693 built-in is provided. This built-in has an additional last argument,
5694 which is the number of bytes remaining in object the @var{dest}
5695 argument points to or @code{(size_t) -1} if the size is not known.
5697 The built-in functions are optimized into the normal string functions
5698 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
5699 it is known at compile time that the destination object will not
5700 be overflown. If the compiler can determine at compile time the
5701 object will be always overflown, it issues a warning.
5703 The intended use can be e.g.
5707 #define bos0(dest) __builtin_object_size (dest, 0)
5708 #define memcpy(dest, src, n) \
5709 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
5713 /* It is unknown what object p points to, so this is optimized
5714 into plain memcpy - no checking is possible. */
5715 memcpy (p, "abcde", n);
5716 /* Destination is known and length too. It is known at compile
5717 time there will be no overflow. */
5718 memcpy (&buf[5], "abcde", 5);
5719 /* Destination is known, but the length is not known at compile time.
5720 This will result in __memcpy_chk call that can check for overflow
5722 memcpy (&buf[5], "abcde", n);
5723 /* Destination is known and it is known at compile time there will
5724 be overflow. There will be a warning and __memcpy_chk call that
5725 will abort the program at runtime. */
5726 memcpy (&buf[6], "abcde", 5);
5729 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
5730 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
5731 @code{strcat} and @code{strncat}.
5733 There are also checking built-in functions for formatted output functions.
5735 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
5736 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5737 const char *fmt, ...);
5738 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
5740 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5741 const char *fmt, va_list ap);
5744 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
5745 etc.@: functions and can contain implementation specific flags on what
5746 additional security measures the checking function might take, such as
5747 handling @code{%n} differently.
5749 The @var{os} argument is the object size @var{s} points to, like in the
5750 other built-in functions. There is a small difference in the behavior
5751 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
5752 optimized into the non-checking functions only if @var{flag} is 0, otherwise
5753 the checking function is called with @var{os} argument set to
5756 In addition to this, there are checking built-in functions
5757 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
5758 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
5759 These have just one additional argument, @var{flag}, right before
5760 format string @var{fmt}. If the compiler is able to optimize them to
5761 @code{fputc} etc.@: functions, it will, otherwise the checking function
5762 should be called and the @var{flag} argument passed to it.
5764 @node Other Builtins
5765 @section Other built-in functions provided by GCC
5766 @cindex built-in functions
5767 @findex __builtin_isfinite
5768 @findex __builtin_isnormal
5769 @findex __builtin_isgreater
5770 @findex __builtin_isgreaterequal
5771 @findex __builtin_isless
5772 @findex __builtin_islessequal
5773 @findex __builtin_islessgreater
5774 @findex __builtin_isunordered
5775 @findex __builtin_powi
5776 @findex __builtin_powif
5777 @findex __builtin_powil
5935 @findex fprintf_unlocked
5937 @findex fputs_unlocked
6054 @findex printf_unlocked
6086 @findex significandf
6087 @findex significandl
6158 GCC provides a large number of built-in functions other than the ones
6159 mentioned above. Some of these are for internal use in the processing
6160 of exceptions or variable-length argument lists and will not be
6161 documented here because they may change from time to time; we do not
6162 recommend general use of these functions.
6164 The remaining functions are provided for optimization purposes.
6166 @opindex fno-builtin
6167 GCC includes built-in versions of many of the functions in the standard
6168 C library. The versions prefixed with @code{__builtin_} will always be
6169 treated as having the same meaning as the C library function even if you
6170 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
6171 Many of these functions are only optimized in certain cases; if they are
6172 not optimized in a particular case, a call to the library function will
6177 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
6178 @option{-std=c99}), the functions
6179 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
6180 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
6181 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
6182 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
6183 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
6184 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
6185 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
6186 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
6187 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
6188 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
6189 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
6190 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
6191 @code{signbitd64}, @code{signbitd128}, @code{significandf},
6192 @code{significandl}, @code{significand}, @code{sincosf},
6193 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
6194 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
6195 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
6196 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
6198 may be handled as built-in functions.
6199 All these functions have corresponding versions
6200 prefixed with @code{__builtin_}, which may be used even in strict C89
6203 The ISO C99 functions
6204 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
6205 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
6206 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
6207 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
6208 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
6209 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
6210 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
6211 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
6212 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
6213 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
6214 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
6215 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
6216 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
6217 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
6218 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
6219 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
6220 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
6221 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
6222 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
6223 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
6224 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
6225 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
6226 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
6227 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
6228 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
6229 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
6230 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
6231 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
6232 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
6233 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
6234 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
6235 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
6236 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
6237 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
6238 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
6239 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
6240 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
6241 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
6242 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
6243 are handled as built-in functions
6244 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
6246 There are also built-in versions of the ISO C99 functions
6247 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
6248 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
6249 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
6250 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
6251 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
6252 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
6253 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
6254 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
6255 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
6256 that are recognized in any mode since ISO C90 reserves these names for
6257 the purpose to which ISO C99 puts them. All these functions have
6258 corresponding versions prefixed with @code{__builtin_}.
6260 The ISO C94 functions
6261 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
6262 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
6263 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
6265 are handled as built-in functions
6266 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
6268 The ISO C90 functions
6269 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
6270 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
6271 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
6272 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
6273 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
6274 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
6275 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
6276 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
6277 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
6278 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
6279 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
6280 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
6281 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
6282 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
6283 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
6284 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
6285 are all recognized as built-in functions unless
6286 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
6287 is specified for an individual function). All of these functions have
6288 corresponding versions prefixed with @code{__builtin_}.
6290 GCC provides built-in versions of the ISO C99 floating point comparison
6291 macros that avoid raising exceptions for unordered operands. They have
6292 the same names as the standard macros ( @code{isgreater},
6293 @code{isgreaterequal}, @code{isless}, @code{islessequal},
6294 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
6295 prefixed. We intend for a library implementor to be able to simply
6296 @code{#define} each standard macro to its built-in equivalent.
6297 In the same fashion, GCC provides @code{isfinite} and @code{isnormal}
6298 built-ins used with @code{__builtin_} prefixed.
6300 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
6302 You can use the built-in function @code{__builtin_types_compatible_p} to
6303 determine whether two types are the same.
6305 This built-in function returns 1 if the unqualified versions of the
6306 types @var{type1} and @var{type2} (which are types, not expressions) are
6307 compatible, 0 otherwise. The result of this built-in function can be
6308 used in integer constant expressions.
6310 This built-in function ignores top level qualifiers (e.g., @code{const},
6311 @code{volatile}). For example, @code{int} is equivalent to @code{const
6314 The type @code{int[]} and @code{int[5]} are compatible. On the other
6315 hand, @code{int} and @code{char *} are not compatible, even if the size
6316 of their types, on the particular architecture are the same. Also, the
6317 amount of pointer indirection is taken into account when determining
6318 similarity. Consequently, @code{short *} is not similar to
6319 @code{short **}. Furthermore, two types that are typedefed are
6320 considered compatible if their underlying types are compatible.
6322 An @code{enum} type is not considered to be compatible with another
6323 @code{enum} type even if both are compatible with the same integer
6324 type; this is what the C standard specifies.
6325 For example, @code{enum @{foo, bar@}} is not similar to
6326 @code{enum @{hot, dog@}}.
6328 You would typically use this function in code whose execution varies
6329 depending on the arguments' types. For example:
6334 typeof (x) tmp = (x); \
6335 if (__builtin_types_compatible_p (typeof (x), long double)) \
6336 tmp = foo_long_double (tmp); \
6337 else if (__builtin_types_compatible_p (typeof (x), double)) \
6338 tmp = foo_double (tmp); \
6339 else if (__builtin_types_compatible_p (typeof (x), float)) \
6340 tmp = foo_float (tmp); \
6347 @emph{Note:} This construct is only available for C@.
6351 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
6353 You can use the built-in function @code{__builtin_choose_expr} to
6354 evaluate code depending on the value of a constant expression. This
6355 built-in function returns @var{exp1} if @var{const_exp}, which is a
6356 constant expression that must be able to be determined at compile time,
6357 is nonzero. Otherwise it returns 0.
6359 This built-in function is analogous to the @samp{? :} operator in C,
6360 except that the expression returned has its type unaltered by promotion
6361 rules. Also, the built-in function does not evaluate the expression
6362 that was not chosen. For example, if @var{const_exp} evaluates to true,
6363 @var{exp2} is not evaluated even if it has side-effects.
6365 This built-in function can return an lvalue if the chosen argument is an
6368 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
6369 type. Similarly, if @var{exp2} is returned, its return type is the same
6376 __builtin_choose_expr ( \
6377 __builtin_types_compatible_p (typeof (x), double), \
6379 __builtin_choose_expr ( \
6380 __builtin_types_compatible_p (typeof (x), float), \
6382 /* @r{The void expression results in a compile-time error} \
6383 @r{when assigning the result to something.} */ \
6387 @emph{Note:} This construct is only available for C@. Furthermore, the
6388 unused expression (@var{exp1} or @var{exp2} depending on the value of
6389 @var{const_exp}) may still generate syntax errors. This may change in
6394 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
6395 You can use the built-in function @code{__builtin_constant_p} to
6396 determine if a value is known to be constant at compile-time and hence
6397 that GCC can perform constant-folding on expressions involving that
6398 value. The argument of the function is the value to test. The function
6399 returns the integer 1 if the argument is known to be a compile-time
6400 constant and 0 if it is not known to be a compile-time constant. A
6401 return of 0 does not indicate that the value is @emph{not} a constant,
6402 but merely that GCC cannot prove it is a constant with the specified
6403 value of the @option{-O} option.
6405 You would typically use this function in an embedded application where
6406 memory was a critical resource. If you have some complex calculation,
6407 you may want it to be folded if it involves constants, but need to call
6408 a function if it does not. For example:
6411 #define Scale_Value(X) \
6412 (__builtin_constant_p (X) \
6413 ? ((X) * SCALE + OFFSET) : Scale (X))
6416 You may use this built-in function in either a macro or an inline
6417 function. However, if you use it in an inlined function and pass an
6418 argument of the function as the argument to the built-in, GCC will
6419 never return 1 when you call the inline function with a string constant
6420 or compound literal (@pxref{Compound Literals}) and will not return 1
6421 when you pass a constant numeric value to the inline function unless you
6422 specify the @option{-O} option.
6424 You may also use @code{__builtin_constant_p} in initializers for static
6425 data. For instance, you can write
6428 static const int table[] = @{
6429 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
6435 This is an acceptable initializer even if @var{EXPRESSION} is not a
6436 constant expression. GCC must be more conservative about evaluating the
6437 built-in in this case, because it has no opportunity to perform
6440 Previous versions of GCC did not accept this built-in in data
6441 initializers. The earliest version where it is completely safe is
6445 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
6446 @opindex fprofile-arcs
6447 You may use @code{__builtin_expect} to provide the compiler with
6448 branch prediction information. In general, you should prefer to
6449 use actual profile feedback for this (@option{-fprofile-arcs}), as
6450 programmers are notoriously bad at predicting how their programs
6451 actually perform. However, there are applications in which this
6452 data is hard to collect.
6454 The return value is the value of @var{exp}, which should be an integral
6455 expression. The semantics of the built-in are that it is expected that
6456 @var{exp} == @var{c}. For example:
6459 if (__builtin_expect (x, 0))
6464 would indicate that we do not expect to call @code{foo}, since
6465 we expect @code{x} to be zero. Since you are limited to integral
6466 expressions for @var{exp}, you should use constructions such as
6469 if (__builtin_expect (ptr != NULL, 1))
6474 when testing pointer or floating-point values.
6477 @deftypefn {Built-in Function} void __builtin_trap (void)
6478 This function causes the program to exit abnormally. GCC implements
6479 this function by using a target-dependent mechanism (such as
6480 intentionally executing an illegal instruction) or by calling
6481 @code{abort}. The mechanism used may vary from release to release so
6482 you should not rely on any particular implementation.
6485 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
6486 This function is used to flush the processor's instruction cache for
6487 the region of memory between @var{begin} inclusive and @var{end}
6488 exclusive. Some targets require that the instruction cache be
6489 flushed, after modifying memory containing code, in order to obtain
6490 deterministic behavior.
6492 If the target does not require instruction cache flushes,
6493 @code{__builtin___clear_cache} has no effect. Otherwise either
6494 instructions are emitted in-line to clear the instruction cache or a
6495 call to the @code{__clear_cache} function in libgcc is made.
6498 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
6499 This function is used to minimize cache-miss latency by moving data into
6500 a cache before it is accessed.
6501 You can insert calls to @code{__builtin_prefetch} into code for which
6502 you know addresses of data in memory that is likely to be accessed soon.
6503 If the target supports them, data prefetch instructions will be generated.
6504 If the prefetch is done early enough before the access then the data will
6505 be in the cache by the time it is accessed.
6507 The value of @var{addr} is the address of the memory to prefetch.
6508 There are two optional arguments, @var{rw} and @var{locality}.
6509 The value of @var{rw} is a compile-time constant one or zero; one
6510 means that the prefetch is preparing for a write to the memory address
6511 and zero, the default, means that the prefetch is preparing for a read.
6512 The value @var{locality} must be a compile-time constant integer between
6513 zero and three. A value of zero means that the data has no temporal
6514 locality, so it need not be left in the cache after the access. A value
6515 of three means that the data has a high degree of temporal locality and
6516 should be left in all levels of cache possible. Values of one and two
6517 mean, respectively, a low or moderate degree of temporal locality. The
6521 for (i = 0; i < n; i++)
6524 __builtin_prefetch (&a[i+j], 1, 1);
6525 __builtin_prefetch (&b[i+j], 0, 1);
6530 Data prefetch does not generate faults if @var{addr} is invalid, but
6531 the address expression itself must be valid. For example, a prefetch
6532 of @code{p->next} will not fault if @code{p->next} is not a valid
6533 address, but evaluation will fault if @code{p} is not a valid address.
6535 If the target does not support data prefetch, the address expression
6536 is evaluated if it includes side effects but no other code is generated
6537 and GCC does not issue a warning.
6540 @deftypefn {Built-in Function} double __builtin_huge_val (void)
6541 Returns a positive infinity, if supported by the floating-point format,
6542 else @code{DBL_MAX}. This function is suitable for implementing the
6543 ISO C macro @code{HUGE_VAL}.
6546 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
6547 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
6550 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
6551 Similar to @code{__builtin_huge_val}, except the return
6552 type is @code{long double}.
6555 @deftypefn {Built-in Function} double __builtin_inf (void)
6556 Similar to @code{__builtin_huge_val}, except a warning is generated
6557 if the target floating-point format does not support infinities.
6560 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
6561 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
6564 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
6565 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
6568 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
6569 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
6572 @deftypefn {Built-in Function} float __builtin_inff (void)
6573 Similar to @code{__builtin_inf}, except the return type is @code{float}.
6574 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
6577 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
6578 Similar to @code{__builtin_inf}, except the return
6579 type is @code{long double}.
6582 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
6583 This is an implementation of the ISO C99 function @code{nan}.
6585 Since ISO C99 defines this function in terms of @code{strtod}, which we
6586 do not implement, a description of the parsing is in order. The string
6587 is parsed as by @code{strtol}; that is, the base is recognized by
6588 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
6589 in the significand such that the least significant bit of the number
6590 is at the least significant bit of the significand. The number is
6591 truncated to fit the significand field provided. The significand is
6592 forced to be a quiet NaN@.
6594 This function, if given a string literal all of which would have been
6595 consumed by strtol, is evaluated early enough that it is considered a
6596 compile-time constant.
6599 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
6600 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
6603 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
6604 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
6607 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
6608 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
6611 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
6612 Similar to @code{__builtin_nan}, except the return type is @code{float}.
6615 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
6616 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
6619 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
6620 Similar to @code{__builtin_nan}, except the significand is forced
6621 to be a signaling NaN@. The @code{nans} function is proposed by
6622 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
6625 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
6626 Similar to @code{__builtin_nans}, except the return type is @code{float}.
6629 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
6630 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
6633 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
6634 Returns one plus the index of the least significant 1-bit of @var{x}, or
6635 if @var{x} is zero, returns zero.
6638 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
6639 Returns the number of leading 0-bits in @var{x}, starting at the most
6640 significant bit position. If @var{x} is 0, the result is undefined.
6643 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
6644 Returns the number of trailing 0-bits in @var{x}, starting at the least
6645 significant bit position. If @var{x} is 0, the result is undefined.
6648 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
6649 Returns the number of 1-bits in @var{x}.
6652 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
6653 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
6657 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
6658 Similar to @code{__builtin_ffs}, except the argument type is
6659 @code{unsigned long}.
6662 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
6663 Similar to @code{__builtin_clz}, except the argument type is
6664 @code{unsigned long}.
6667 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
6668 Similar to @code{__builtin_ctz}, except the argument type is
6669 @code{unsigned long}.
6672 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
6673 Similar to @code{__builtin_popcount}, except the argument type is
6674 @code{unsigned long}.
6677 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
6678 Similar to @code{__builtin_parity}, except the argument type is
6679 @code{unsigned long}.
6682 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
6683 Similar to @code{__builtin_ffs}, except the argument type is
6684 @code{unsigned long long}.
6687 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
6688 Similar to @code{__builtin_clz}, except the argument type is
6689 @code{unsigned long long}.
6692 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
6693 Similar to @code{__builtin_ctz}, except the argument type is
6694 @code{unsigned long long}.
6697 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
6698 Similar to @code{__builtin_popcount}, except the argument type is
6699 @code{unsigned long long}.
6702 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
6703 Similar to @code{__builtin_parity}, except the argument type is
6704 @code{unsigned long long}.
6707 @deftypefn {Built-in Function} double __builtin_powi (double, int)
6708 Returns the first argument raised to the power of the second. Unlike the
6709 @code{pow} function no guarantees about precision and rounding are made.
6712 @deftypefn {Built-in Function} float __builtin_powif (float, int)
6713 Similar to @code{__builtin_powi}, except the argument and return types
6717 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
6718 Similar to @code{__builtin_powi}, except the argument and return types
6719 are @code{long double}.
6722 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
6723 Returns @var{x} with the order of the bytes reversed; for example,
6724 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
6728 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
6729 Similar to @code{__builtin_bswap32}, except the argument and return types
6733 @node Target Builtins
6734 @section Built-in Functions Specific to Particular Target Machines
6736 On some target machines, GCC supports many built-in functions specific
6737 to those machines. Generally these generate calls to specific machine
6738 instructions, but allow the compiler to schedule those calls.
6741 * Alpha Built-in Functions::
6742 * ARM iWMMXt Built-in Functions::
6743 * ARM NEON Intrinsics::
6744 * Blackfin Built-in Functions::
6745 * FR-V Built-in Functions::
6746 * X86 Built-in Functions::
6747 * MIPS DSP Built-in Functions::
6748 * MIPS Paired-Single Support::
6749 * PowerPC AltiVec Built-in Functions::
6750 * SPARC VIS Built-in Functions::
6751 * SPU Built-in Functions::
6754 @node Alpha Built-in Functions
6755 @subsection Alpha Built-in Functions
6757 These built-in functions are available for the Alpha family of
6758 processors, depending on the command-line switches used.
6760 The following built-in functions are always available. They
6761 all generate the machine instruction that is part of the name.
6764 long __builtin_alpha_implver (void)
6765 long __builtin_alpha_rpcc (void)
6766 long __builtin_alpha_amask (long)
6767 long __builtin_alpha_cmpbge (long, long)
6768 long __builtin_alpha_extbl (long, long)
6769 long __builtin_alpha_extwl (long, long)
6770 long __builtin_alpha_extll (long, long)
6771 long __builtin_alpha_extql (long, long)
6772 long __builtin_alpha_extwh (long, long)
6773 long __builtin_alpha_extlh (long, long)
6774 long __builtin_alpha_extqh (long, long)
6775 long __builtin_alpha_insbl (long, long)
6776 long __builtin_alpha_inswl (long, long)
6777 long __builtin_alpha_insll (long, long)
6778 long __builtin_alpha_insql (long, long)
6779 long __builtin_alpha_inswh (long, long)
6780 long __builtin_alpha_inslh (long, long)
6781 long __builtin_alpha_insqh (long, long)
6782 long __builtin_alpha_mskbl (long, long)
6783 long __builtin_alpha_mskwl (long, long)
6784 long __builtin_alpha_mskll (long, long)
6785 long __builtin_alpha_mskql (long, long)
6786 long __builtin_alpha_mskwh (long, long)
6787 long __builtin_alpha_msklh (long, long)
6788 long __builtin_alpha_mskqh (long, long)
6789 long __builtin_alpha_umulh (long, long)
6790 long __builtin_alpha_zap (long, long)
6791 long __builtin_alpha_zapnot (long, long)
6794 The following built-in functions are always with @option{-mmax}
6795 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
6796 later. They all generate the machine instruction that is part
6800 long __builtin_alpha_pklb (long)
6801 long __builtin_alpha_pkwb (long)
6802 long __builtin_alpha_unpkbl (long)
6803 long __builtin_alpha_unpkbw (long)
6804 long __builtin_alpha_minub8 (long, long)
6805 long __builtin_alpha_minsb8 (long, long)
6806 long __builtin_alpha_minuw4 (long, long)
6807 long __builtin_alpha_minsw4 (long, long)
6808 long __builtin_alpha_maxub8 (long, long)
6809 long __builtin_alpha_maxsb8 (long, long)
6810 long __builtin_alpha_maxuw4 (long, long)
6811 long __builtin_alpha_maxsw4 (long, long)
6812 long __builtin_alpha_perr (long, long)
6815 The following built-in functions are always with @option{-mcix}
6816 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
6817 later. They all generate the machine instruction that is part
6821 long __builtin_alpha_cttz (long)
6822 long __builtin_alpha_ctlz (long)
6823 long __builtin_alpha_ctpop (long)
6826 The following builtins are available on systems that use the OSF/1
6827 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
6828 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
6829 @code{rdval} and @code{wrval}.
6832 void *__builtin_thread_pointer (void)
6833 void __builtin_set_thread_pointer (void *)
6836 @node ARM iWMMXt Built-in Functions
6837 @subsection ARM iWMMXt Built-in Functions
6839 These built-in functions are available for the ARM family of
6840 processors when the @option{-mcpu=iwmmxt} switch is used:
6843 typedef int v2si __attribute__ ((vector_size (8)));
6844 typedef short v4hi __attribute__ ((vector_size (8)));
6845 typedef char v8qi __attribute__ ((vector_size (8)));
6847 int __builtin_arm_getwcx (int)
6848 void __builtin_arm_setwcx (int, int)
6849 int __builtin_arm_textrmsb (v8qi, int)
6850 int __builtin_arm_textrmsh (v4hi, int)
6851 int __builtin_arm_textrmsw (v2si, int)
6852 int __builtin_arm_textrmub (v8qi, int)
6853 int __builtin_arm_textrmuh (v4hi, int)
6854 int __builtin_arm_textrmuw (v2si, int)
6855 v8qi __builtin_arm_tinsrb (v8qi, int)
6856 v4hi __builtin_arm_tinsrh (v4hi, int)
6857 v2si __builtin_arm_tinsrw (v2si, int)
6858 long long __builtin_arm_tmia (long long, int, int)
6859 long long __builtin_arm_tmiabb (long long, int, int)
6860 long long __builtin_arm_tmiabt (long long, int, int)
6861 long long __builtin_arm_tmiaph (long long, int, int)
6862 long long __builtin_arm_tmiatb (long long, int, int)
6863 long long __builtin_arm_tmiatt (long long, int, int)
6864 int __builtin_arm_tmovmskb (v8qi)
6865 int __builtin_arm_tmovmskh (v4hi)
6866 int __builtin_arm_tmovmskw (v2si)
6867 long long __builtin_arm_waccb (v8qi)
6868 long long __builtin_arm_wacch (v4hi)
6869 long long __builtin_arm_waccw (v2si)
6870 v8qi __builtin_arm_waddb (v8qi, v8qi)
6871 v8qi __builtin_arm_waddbss (v8qi, v8qi)
6872 v8qi __builtin_arm_waddbus (v8qi, v8qi)
6873 v4hi __builtin_arm_waddh (v4hi, v4hi)
6874 v4hi __builtin_arm_waddhss (v4hi, v4hi)
6875 v4hi __builtin_arm_waddhus (v4hi, v4hi)
6876 v2si __builtin_arm_waddw (v2si, v2si)
6877 v2si __builtin_arm_waddwss (v2si, v2si)
6878 v2si __builtin_arm_waddwus (v2si, v2si)
6879 v8qi __builtin_arm_walign (v8qi, v8qi, int)
6880 long long __builtin_arm_wand(long long, long long)
6881 long long __builtin_arm_wandn (long long, long long)
6882 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
6883 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
6884 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
6885 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
6886 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
6887 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
6888 v2si __builtin_arm_wcmpeqw (v2si, v2si)
6889 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
6890 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
6891 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
6892 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
6893 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
6894 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
6895 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
6896 long long __builtin_arm_wmacsz (v4hi, v4hi)
6897 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
6898 long long __builtin_arm_wmacuz (v4hi, v4hi)
6899 v4hi __builtin_arm_wmadds (v4hi, v4hi)
6900 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
6901 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
6902 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
6903 v2si __builtin_arm_wmaxsw (v2si, v2si)
6904 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
6905 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
6906 v2si __builtin_arm_wmaxuw (v2si, v2si)
6907 v8qi __builtin_arm_wminsb (v8qi, v8qi)
6908 v4hi __builtin_arm_wminsh (v4hi, v4hi)
6909 v2si __builtin_arm_wminsw (v2si, v2si)
6910 v8qi __builtin_arm_wminub (v8qi, v8qi)
6911 v4hi __builtin_arm_wminuh (v4hi, v4hi)
6912 v2si __builtin_arm_wminuw (v2si, v2si)
6913 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
6914 v4hi __builtin_arm_wmulul (v4hi, v4hi)
6915 v4hi __builtin_arm_wmulum (v4hi, v4hi)
6916 long long __builtin_arm_wor (long long, long long)
6917 v2si __builtin_arm_wpackdss (long long, long long)
6918 v2si __builtin_arm_wpackdus (long long, long long)
6919 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
6920 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
6921 v4hi __builtin_arm_wpackwss (v2si, v2si)
6922 v4hi __builtin_arm_wpackwus (v2si, v2si)
6923 long long __builtin_arm_wrord (long long, long long)
6924 long long __builtin_arm_wrordi (long long, int)
6925 v4hi __builtin_arm_wrorh (v4hi, long long)
6926 v4hi __builtin_arm_wrorhi (v4hi, int)
6927 v2si __builtin_arm_wrorw (v2si, long long)
6928 v2si __builtin_arm_wrorwi (v2si, int)
6929 v2si __builtin_arm_wsadb (v8qi, v8qi)
6930 v2si __builtin_arm_wsadbz (v8qi, v8qi)
6931 v2si __builtin_arm_wsadh (v4hi, v4hi)
6932 v2si __builtin_arm_wsadhz (v4hi, v4hi)
6933 v4hi __builtin_arm_wshufh (v4hi, int)
6934 long long __builtin_arm_wslld (long long, long long)
6935 long long __builtin_arm_wslldi (long long, int)
6936 v4hi __builtin_arm_wsllh (v4hi, long long)
6937 v4hi __builtin_arm_wsllhi (v4hi, int)
6938 v2si __builtin_arm_wsllw (v2si, long long)
6939 v2si __builtin_arm_wsllwi (v2si, int)
6940 long long __builtin_arm_wsrad (long long, long long)
6941 long long __builtin_arm_wsradi (long long, int)
6942 v4hi __builtin_arm_wsrah (v4hi, long long)
6943 v4hi __builtin_arm_wsrahi (v4hi, int)
6944 v2si __builtin_arm_wsraw (v2si, long long)
6945 v2si __builtin_arm_wsrawi (v2si, int)
6946 long long __builtin_arm_wsrld (long long, long long)
6947 long long __builtin_arm_wsrldi (long long, int)
6948 v4hi __builtin_arm_wsrlh (v4hi, long long)
6949 v4hi __builtin_arm_wsrlhi (v4hi, int)
6950 v2si __builtin_arm_wsrlw (v2si, long long)
6951 v2si __builtin_arm_wsrlwi (v2si, int)
6952 v8qi __builtin_arm_wsubb (v8qi, v8qi)
6953 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
6954 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
6955 v4hi __builtin_arm_wsubh (v4hi, v4hi)
6956 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
6957 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
6958 v2si __builtin_arm_wsubw (v2si, v2si)
6959 v2si __builtin_arm_wsubwss (v2si, v2si)
6960 v2si __builtin_arm_wsubwus (v2si, v2si)
6961 v4hi __builtin_arm_wunpckehsb (v8qi)
6962 v2si __builtin_arm_wunpckehsh (v4hi)
6963 long long __builtin_arm_wunpckehsw (v2si)
6964 v4hi __builtin_arm_wunpckehub (v8qi)
6965 v2si __builtin_arm_wunpckehuh (v4hi)
6966 long long __builtin_arm_wunpckehuw (v2si)
6967 v4hi __builtin_arm_wunpckelsb (v8qi)
6968 v2si __builtin_arm_wunpckelsh (v4hi)
6969 long long __builtin_arm_wunpckelsw (v2si)
6970 v4hi __builtin_arm_wunpckelub (v8qi)
6971 v2si __builtin_arm_wunpckeluh (v4hi)
6972 long long __builtin_arm_wunpckeluw (v2si)
6973 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
6974 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
6975 v2si __builtin_arm_wunpckihw (v2si, v2si)
6976 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
6977 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
6978 v2si __builtin_arm_wunpckilw (v2si, v2si)
6979 long long __builtin_arm_wxor (long long, long long)
6980 long long __builtin_arm_wzero ()
6983 @node ARM NEON Intrinsics
6984 @subsection ARM NEON Intrinsics
6986 These built-in intrinsics for the ARM Advanced SIMD extension are available
6987 when the @option{-mfpu=neon} switch is used:
6989 @include arm-neon-intrinsics.texi
6991 @node Blackfin Built-in Functions
6992 @subsection Blackfin Built-in Functions
6994 Currently, there are two Blackfin-specific built-in functions. These are
6995 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
6996 using inline assembly; by using these built-in functions the compiler can
6997 automatically add workarounds for hardware errata involving these
6998 instructions. These functions are named as follows:
7001 void __builtin_bfin_csync (void)
7002 void __builtin_bfin_ssync (void)
7005 @node FR-V Built-in Functions
7006 @subsection FR-V Built-in Functions
7008 GCC provides many FR-V-specific built-in functions. In general,
7009 these functions are intended to be compatible with those described
7010 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
7011 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
7012 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
7013 pointer rather than by value.
7015 Most of the functions are named after specific FR-V instructions.
7016 Such functions are said to be ``directly mapped'' and are summarized
7017 here in tabular form.
7021 * Directly-mapped Integer Functions::
7022 * Directly-mapped Media Functions::
7023 * Raw read/write Functions::
7024 * Other Built-in Functions::
7027 @node Argument Types
7028 @subsubsection Argument Types
7030 The arguments to the built-in functions can be divided into three groups:
7031 register numbers, compile-time constants and run-time values. In order
7032 to make this classification clear at a glance, the arguments and return
7033 values are given the following pseudo types:
7035 @multitable @columnfractions .20 .30 .15 .35
7036 @item Pseudo type @tab Real C type @tab Constant? @tab Description
7037 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
7038 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
7039 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
7040 @item @code{uw2} @tab @code{unsigned long long} @tab No
7041 @tab an unsigned doubleword
7042 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
7043 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
7044 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
7045 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
7048 These pseudo types are not defined by GCC, they are simply a notational
7049 convenience used in this manual.
7051 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
7052 and @code{sw2} are evaluated at run time. They correspond to
7053 register operands in the underlying FR-V instructions.
7055 @code{const} arguments represent immediate operands in the underlying
7056 FR-V instructions. They must be compile-time constants.
7058 @code{acc} arguments are evaluated at compile time and specify the number
7059 of an accumulator register. For example, an @code{acc} argument of 2
7060 will select the ACC2 register.
7062 @code{iacc} arguments are similar to @code{acc} arguments but specify the
7063 number of an IACC register. See @pxref{Other Built-in Functions}
7066 @node Directly-mapped Integer Functions
7067 @subsubsection Directly-mapped Integer Functions
7069 The functions listed below map directly to FR-V I-type instructions.
7071 @multitable @columnfractions .45 .32 .23
7072 @item Function prototype @tab Example usage @tab Assembly output
7073 @item @code{sw1 __ADDSS (sw1, sw1)}
7074 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
7075 @tab @code{ADDSS @var{a},@var{b},@var{c}}
7076 @item @code{sw1 __SCAN (sw1, sw1)}
7077 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
7078 @tab @code{SCAN @var{a},@var{b},@var{c}}
7079 @item @code{sw1 __SCUTSS (sw1)}
7080 @tab @code{@var{b} = __SCUTSS (@var{a})}
7081 @tab @code{SCUTSS @var{a},@var{b}}
7082 @item @code{sw1 __SLASS (sw1, sw1)}
7083 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
7084 @tab @code{SLASS @var{a},@var{b},@var{c}}
7085 @item @code{void __SMASS (sw1, sw1)}
7086 @tab @code{__SMASS (@var{a}, @var{b})}
7087 @tab @code{SMASS @var{a},@var{b}}
7088 @item @code{void __SMSSS (sw1, sw1)}
7089 @tab @code{__SMSSS (@var{a}, @var{b})}
7090 @tab @code{SMSSS @var{a},@var{b}}
7091 @item @code{void __SMU (sw1, sw1)}
7092 @tab @code{__SMU (@var{a}, @var{b})}
7093 @tab @code{SMU @var{a},@var{b}}
7094 @item @code{sw2 __SMUL (sw1, sw1)}
7095 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
7096 @tab @code{SMUL @var{a},@var{b},@var{c}}
7097 @item @code{sw1 __SUBSS (sw1, sw1)}
7098 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
7099 @tab @code{SUBSS @var{a},@var{b},@var{c}}
7100 @item @code{uw2 __UMUL (uw1, uw1)}
7101 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
7102 @tab @code{UMUL @var{a},@var{b},@var{c}}
7105 @node Directly-mapped Media Functions
7106 @subsubsection Directly-mapped Media Functions
7108 The functions listed below map directly to FR-V M-type instructions.
7110 @multitable @columnfractions .45 .32 .23
7111 @item Function prototype @tab Example usage @tab Assembly output
7112 @item @code{uw1 __MABSHS (sw1)}
7113 @tab @code{@var{b} = __MABSHS (@var{a})}
7114 @tab @code{MABSHS @var{a},@var{b}}
7115 @item @code{void __MADDACCS (acc, acc)}
7116 @tab @code{__MADDACCS (@var{b}, @var{a})}
7117 @tab @code{MADDACCS @var{a},@var{b}}
7118 @item @code{sw1 __MADDHSS (sw1, sw1)}
7119 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
7120 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
7121 @item @code{uw1 __MADDHUS (uw1, uw1)}
7122 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
7123 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
7124 @item @code{uw1 __MAND (uw1, uw1)}
7125 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
7126 @tab @code{MAND @var{a},@var{b},@var{c}}
7127 @item @code{void __MASACCS (acc, acc)}
7128 @tab @code{__MASACCS (@var{b}, @var{a})}
7129 @tab @code{MASACCS @var{a},@var{b}}
7130 @item @code{uw1 __MAVEH (uw1, uw1)}
7131 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
7132 @tab @code{MAVEH @var{a},@var{b},@var{c}}
7133 @item @code{uw2 __MBTOH (uw1)}
7134 @tab @code{@var{b} = __MBTOH (@var{a})}
7135 @tab @code{MBTOH @var{a},@var{b}}
7136 @item @code{void __MBTOHE (uw1 *, uw1)}
7137 @tab @code{__MBTOHE (&@var{b}, @var{a})}
7138 @tab @code{MBTOHE @var{a},@var{b}}
7139 @item @code{void __MCLRACC (acc)}
7140 @tab @code{__MCLRACC (@var{a})}
7141 @tab @code{MCLRACC @var{a}}
7142 @item @code{void __MCLRACCA (void)}
7143 @tab @code{__MCLRACCA ()}
7144 @tab @code{MCLRACCA}
7145 @item @code{uw1 __Mcop1 (uw1, uw1)}
7146 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
7147 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
7148 @item @code{uw1 __Mcop2 (uw1, uw1)}
7149 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
7150 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
7151 @item @code{uw1 __MCPLHI (uw2, const)}
7152 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
7153 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
7154 @item @code{uw1 __MCPLI (uw2, const)}
7155 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
7156 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
7157 @item @code{void __MCPXIS (acc, sw1, sw1)}
7158 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
7159 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
7160 @item @code{void __MCPXIU (acc, uw1, uw1)}
7161 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
7162 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
7163 @item @code{void __MCPXRS (acc, sw1, sw1)}
7164 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
7165 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
7166 @item @code{void __MCPXRU (acc, uw1, uw1)}
7167 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
7168 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
7169 @item @code{uw1 __MCUT (acc, uw1)}
7170 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
7171 @tab @code{MCUT @var{a},@var{b},@var{c}}
7172 @item @code{uw1 __MCUTSS (acc, sw1)}
7173 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
7174 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
7175 @item @code{void __MDADDACCS (acc, acc)}
7176 @tab @code{__MDADDACCS (@var{b}, @var{a})}
7177 @tab @code{MDADDACCS @var{a},@var{b}}
7178 @item @code{void __MDASACCS (acc, acc)}
7179 @tab @code{__MDASACCS (@var{b}, @var{a})}
7180 @tab @code{MDASACCS @var{a},@var{b}}
7181 @item @code{uw2 __MDCUTSSI (acc, const)}
7182 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
7183 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
7184 @item @code{uw2 __MDPACKH (uw2, uw2)}
7185 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
7186 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
7187 @item @code{uw2 __MDROTLI (uw2, const)}
7188 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
7189 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
7190 @item @code{void __MDSUBACCS (acc, acc)}
7191 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
7192 @tab @code{MDSUBACCS @var{a},@var{b}}
7193 @item @code{void __MDUNPACKH (uw1 *, uw2)}
7194 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
7195 @tab @code{MDUNPACKH @var{a},@var{b}}
7196 @item @code{uw2 __MEXPDHD (uw1, const)}
7197 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
7198 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
7199 @item @code{uw1 __MEXPDHW (uw1, const)}
7200 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
7201 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
7202 @item @code{uw1 __MHDSETH (uw1, const)}
7203 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
7204 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
7205 @item @code{sw1 __MHDSETS (const)}
7206 @tab @code{@var{b} = __MHDSETS (@var{a})}
7207 @tab @code{MHDSETS #@var{a},@var{b}}
7208 @item @code{uw1 __MHSETHIH (uw1, const)}
7209 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
7210 @tab @code{MHSETHIH #@var{a},@var{b}}
7211 @item @code{sw1 __MHSETHIS (sw1, const)}
7212 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
7213 @tab @code{MHSETHIS #@var{a},@var{b}}
7214 @item @code{uw1 __MHSETLOH (uw1, const)}
7215 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
7216 @tab @code{MHSETLOH #@var{a},@var{b}}
7217 @item @code{sw1 __MHSETLOS (sw1, const)}
7218 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
7219 @tab @code{MHSETLOS #@var{a},@var{b}}
7220 @item @code{uw1 __MHTOB (uw2)}
7221 @tab @code{@var{b} = __MHTOB (@var{a})}
7222 @tab @code{MHTOB @var{a},@var{b}}
7223 @item @code{void __MMACHS (acc, sw1, sw1)}
7224 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
7225 @tab @code{MMACHS @var{a},@var{b},@var{c}}
7226 @item @code{void __MMACHU (acc, uw1, uw1)}
7227 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
7228 @tab @code{MMACHU @var{a},@var{b},@var{c}}
7229 @item @code{void __MMRDHS (acc, sw1, sw1)}
7230 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
7231 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
7232 @item @code{void __MMRDHU (acc, uw1, uw1)}
7233 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
7234 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
7235 @item @code{void __MMULHS (acc, sw1, sw1)}
7236 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
7237 @tab @code{MMULHS @var{a},@var{b},@var{c}}
7238 @item @code{void __MMULHU (acc, uw1, uw1)}
7239 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
7240 @tab @code{MMULHU @var{a},@var{b},@var{c}}
7241 @item @code{void __MMULXHS (acc, sw1, sw1)}
7242 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
7243 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
7244 @item @code{void __MMULXHU (acc, uw1, uw1)}
7245 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
7246 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
7247 @item @code{uw1 __MNOT (uw1)}
7248 @tab @code{@var{b} = __MNOT (@var{a})}
7249 @tab @code{MNOT @var{a},@var{b}}
7250 @item @code{uw1 __MOR (uw1, uw1)}
7251 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
7252 @tab @code{MOR @var{a},@var{b},@var{c}}
7253 @item @code{uw1 __MPACKH (uh, uh)}
7254 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
7255 @tab @code{MPACKH @var{a},@var{b},@var{c}}
7256 @item @code{sw2 __MQADDHSS (sw2, sw2)}
7257 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
7258 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
7259 @item @code{uw2 __MQADDHUS (uw2, uw2)}
7260 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
7261 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
7262 @item @code{void __MQCPXIS (acc, sw2, sw2)}
7263 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
7264 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
7265 @item @code{void __MQCPXIU (acc, uw2, uw2)}
7266 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
7267 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
7268 @item @code{void __MQCPXRS (acc, sw2, sw2)}
7269 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
7270 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
7271 @item @code{void __MQCPXRU (acc, uw2, uw2)}
7272 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
7273 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
7274 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
7275 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
7276 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
7277 @item @code{sw2 __MQLMTHS (sw2, sw2)}
7278 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
7279 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
7280 @item @code{void __MQMACHS (acc, sw2, sw2)}
7281 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
7282 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
7283 @item @code{void __MQMACHU (acc, uw2, uw2)}
7284 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
7285 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
7286 @item @code{void __MQMACXHS (acc, sw2, sw2)}
7287 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
7288 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
7289 @item @code{void __MQMULHS (acc, sw2, sw2)}
7290 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
7291 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
7292 @item @code{void __MQMULHU (acc, uw2, uw2)}
7293 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
7294 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
7295 @item @code{void __MQMULXHS (acc, sw2, sw2)}
7296 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
7297 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
7298 @item @code{void __MQMULXHU (acc, uw2, uw2)}
7299 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
7300 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
7301 @item @code{sw2 __MQSATHS (sw2, sw2)}
7302 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
7303 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
7304 @item @code{uw2 __MQSLLHI (uw2, int)}
7305 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
7306 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
7307 @item @code{sw2 __MQSRAHI (sw2, int)}
7308 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
7309 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
7310 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
7311 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
7312 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
7313 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
7314 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
7315 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
7316 @item @code{void __MQXMACHS (acc, sw2, sw2)}
7317 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
7318 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
7319 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
7320 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
7321 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
7322 @item @code{uw1 __MRDACC (acc)}
7323 @tab @code{@var{b} = __MRDACC (@var{a})}
7324 @tab @code{MRDACC @var{a},@var{b}}
7325 @item @code{uw1 __MRDACCG (acc)}
7326 @tab @code{@var{b} = __MRDACCG (@var{a})}
7327 @tab @code{MRDACCG @var{a},@var{b}}
7328 @item @code{uw1 __MROTLI (uw1, const)}
7329 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
7330 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
7331 @item @code{uw1 __MROTRI (uw1, const)}
7332 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
7333 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
7334 @item @code{sw1 __MSATHS (sw1, sw1)}
7335 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
7336 @tab @code{MSATHS @var{a},@var{b},@var{c}}
7337 @item @code{uw1 __MSATHU (uw1, uw1)}
7338 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
7339 @tab @code{MSATHU @var{a},@var{b},@var{c}}
7340 @item @code{uw1 __MSLLHI (uw1, const)}
7341 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
7342 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
7343 @item @code{sw1 __MSRAHI (sw1, const)}
7344 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
7345 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
7346 @item @code{uw1 __MSRLHI (uw1, const)}
7347 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
7348 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
7349 @item @code{void __MSUBACCS (acc, acc)}
7350 @tab @code{__MSUBACCS (@var{b}, @var{a})}
7351 @tab @code{MSUBACCS @var{a},@var{b}}
7352 @item @code{sw1 __MSUBHSS (sw1, sw1)}
7353 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
7354 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
7355 @item @code{uw1 __MSUBHUS (uw1, uw1)}
7356 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
7357 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
7358 @item @code{void __MTRAP (void)}
7359 @tab @code{__MTRAP ()}
7361 @item @code{uw2 __MUNPACKH (uw1)}
7362 @tab @code{@var{b} = __MUNPACKH (@var{a})}
7363 @tab @code{MUNPACKH @var{a},@var{b}}
7364 @item @code{uw1 __MWCUT (uw2, uw1)}
7365 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
7366 @tab @code{MWCUT @var{a},@var{b},@var{c}}
7367 @item @code{void __MWTACC (acc, uw1)}
7368 @tab @code{__MWTACC (@var{b}, @var{a})}
7369 @tab @code{MWTACC @var{a},@var{b}}
7370 @item @code{void __MWTACCG (acc, uw1)}
7371 @tab @code{__MWTACCG (@var{b}, @var{a})}
7372 @tab @code{MWTACCG @var{a},@var{b}}
7373 @item @code{uw1 __MXOR (uw1, uw1)}
7374 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
7375 @tab @code{MXOR @var{a},@var{b},@var{c}}
7378 @node Raw read/write Functions
7379 @subsubsection Raw read/write Functions
7381 This sections describes built-in functions related to read and write
7382 instructions to access memory. These functions generate
7383 @code{membar} instructions to flush the I/O load and stores where
7384 appropriate, as described in Fujitsu's manual described above.
7388 @item unsigned char __builtin_read8 (void *@var{data})
7389 @item unsigned short __builtin_read16 (void *@var{data})
7390 @item unsigned long __builtin_read32 (void *@var{data})
7391 @item unsigned long long __builtin_read64 (void *@var{data})
7393 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
7394 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
7395 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
7396 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
7399 @node Other Built-in Functions
7400 @subsubsection Other Built-in Functions
7402 This section describes built-in functions that are not named after
7403 a specific FR-V instruction.
7406 @item sw2 __IACCreadll (iacc @var{reg})
7407 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
7408 for future expansion and must be 0.
7410 @item sw1 __IACCreadl (iacc @var{reg})
7411 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
7412 Other values of @var{reg} are rejected as invalid.
7414 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
7415 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
7416 is reserved for future expansion and must be 0.
7418 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
7419 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
7420 is 1. Other values of @var{reg} are rejected as invalid.
7422 @item void __data_prefetch0 (const void *@var{x})
7423 Use the @code{dcpl} instruction to load the contents of address @var{x}
7424 into the data cache.
7426 @item void __data_prefetch (const void *@var{x})
7427 Use the @code{nldub} instruction to load the contents of address @var{x}
7428 into the data cache. The instruction will be issued in slot I1@.
7431 @node X86 Built-in Functions
7432 @subsection X86 Built-in Functions
7434 These built-in functions are available for the i386 and x86-64 family
7435 of computers, depending on the command-line switches used.
7437 Note that, if you specify command-line switches such as @option{-msse},
7438 the compiler could use the extended instruction sets even if the built-ins
7439 are not used explicitly in the program. For this reason, applications
7440 which perform runtime CPU detection must compile separate files for each
7441 supported architecture, using the appropriate flags. In particular,
7442 the file containing the CPU detection code should be compiled without
7445 The following machine modes are available for use with MMX built-in functions
7446 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
7447 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
7448 vector of eight 8-bit integers. Some of the built-in functions operate on
7449 MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode.
7451 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
7452 of two 32-bit floating point values.
7454 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
7455 floating point values. Some instructions use a vector of four 32-bit
7456 integers, these use @code{V4SI}. Finally, some instructions operate on an
7457 entire vector register, interpreting it as a 128-bit integer, these use mode
7460 In 64-bit mode, the x86-64 family of processors uses additional built-in
7461 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
7462 floating point and @code{TC} 128-bit complex floating point values.
7464 The following floating point built-in functions are available in 64-bit
7465 mode. All of them implement the function that is part of the name.
7468 __float128 __builtin_fabsq (__float128)
7469 __float128 __builtin_copysignq (__float128, __float128)
7472 The following floating point built-in functions are made available in the
7476 @item __float128 __builtin_infq (void)
7477 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
7480 The following built-in functions are made available by @option{-mmmx}.
7481 All of them generate the machine instruction that is part of the name.
7484 v8qi __builtin_ia32_paddb (v8qi, v8qi)
7485 v4hi __builtin_ia32_paddw (v4hi, v4hi)
7486 v2si __builtin_ia32_paddd (v2si, v2si)
7487 v8qi __builtin_ia32_psubb (v8qi, v8qi)
7488 v4hi __builtin_ia32_psubw (v4hi, v4hi)
7489 v2si __builtin_ia32_psubd (v2si, v2si)
7490 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
7491 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
7492 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
7493 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
7494 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
7495 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
7496 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
7497 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
7498 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
7499 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
7500 di __builtin_ia32_pand (di, di)
7501 di __builtin_ia32_pandn (di,di)
7502 di __builtin_ia32_por (di, di)
7503 di __builtin_ia32_pxor (di, di)
7504 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7505 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7506 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7507 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7508 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7509 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7510 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7511 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7512 v2si __builtin_ia32_punpckhdq (v2si, v2si)
7513 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7514 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7515 v2si __builtin_ia32_punpckldq (v2si, v2si)
7516 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7517 v4hi __builtin_ia32_packssdw (v2si, v2si)
7518 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7520 v4hi __builtin_ia32_psllw (v4hi, v4hi)
7521 v2si __builtin_ia32_pslld (v2si, v2si)
7522 v1di __builtin_ia32_psllq (v1di, v1di)
7523 v4hi __builtin_ia32_psrlw (v4hi, v4hi)
7524 v2si __builtin_ia32_psrld (v2si, v2si)
7525 v1di __builtin_ia32_psrlq (v1di, v1di)
7526 v4hi __builtin_ia32_psraw (v4hi, v4hi)
7527 v2si __builtin_ia32_psrad (v2si, v2si)
7528 v4hi __builtin_ia32_psllwi (v4hi, int)
7529 v2si __builtin_ia32_pslldi (v2si, int)
7530 v1di __builtin_ia32_psllqi (v1di, int)
7531 v4hi __builtin_ia32_psrlwi (v4hi, int)
7532 v2si __builtin_ia32_psrldi (v2si, int)
7533 v1di __builtin_ia32_psrlqi (v1di, int)
7534 v4hi __builtin_ia32_psrawi (v4hi, int)
7535 v2si __builtin_ia32_psradi (v2si, int)
7539 The following built-in functions are made available either with
7540 @option{-msse}, or with a combination of @option{-m3dnow} and
7541 @option{-march=athlon}. All of them generate the machine
7542 instruction that is part of the name.
7545 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7546 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7547 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7548 v1di __builtin_ia32_psadbw (v8qi, v8qi)
7549 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7550 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7551 v8qi __builtin_ia32_pminub (v8qi, v8qi)
7552 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7553 int __builtin_ia32_pextrw (v4hi, int)
7554 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7555 int __builtin_ia32_pmovmskb (v8qi)
7556 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7557 void __builtin_ia32_movntq (di *, di)
7558 void __builtin_ia32_sfence (void)
7561 The following built-in functions are available when @option{-msse} is used.
7562 All of them generate the machine instruction that is part of the name.
7565 int __builtin_ia32_comieq (v4sf, v4sf)
7566 int __builtin_ia32_comineq (v4sf, v4sf)
7567 int __builtin_ia32_comilt (v4sf, v4sf)
7568 int __builtin_ia32_comile (v4sf, v4sf)
7569 int __builtin_ia32_comigt (v4sf, v4sf)
7570 int __builtin_ia32_comige (v4sf, v4sf)
7571 int __builtin_ia32_ucomieq (v4sf, v4sf)
7572 int __builtin_ia32_ucomineq (v4sf, v4sf)
7573 int __builtin_ia32_ucomilt (v4sf, v4sf)
7574 int __builtin_ia32_ucomile (v4sf, v4sf)
7575 int __builtin_ia32_ucomigt (v4sf, v4sf)
7576 int __builtin_ia32_ucomige (v4sf, v4sf)
7577 v4sf __builtin_ia32_addps (v4sf, v4sf)
7578 v4sf __builtin_ia32_subps (v4sf, v4sf)
7579 v4sf __builtin_ia32_mulps (v4sf, v4sf)
7580 v4sf __builtin_ia32_divps (v4sf, v4sf)
7581 v4sf __builtin_ia32_addss (v4sf, v4sf)
7582 v4sf __builtin_ia32_subss (v4sf, v4sf)
7583 v4sf __builtin_ia32_mulss (v4sf, v4sf)
7584 v4sf __builtin_ia32_divss (v4sf, v4sf)
7585 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
7586 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
7587 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
7588 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
7589 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
7590 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
7591 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
7592 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
7593 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
7594 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
7595 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
7596 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
7597 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
7598 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
7599 v4si __builtin_ia32_cmpless (v4sf, v4sf)
7600 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
7601 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
7602 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
7603 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
7604 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
7605 v4sf __builtin_ia32_maxps (v4sf, v4sf)
7606 v4sf __builtin_ia32_maxss (v4sf, v4sf)
7607 v4sf __builtin_ia32_minps (v4sf, v4sf)
7608 v4sf __builtin_ia32_minss (v4sf, v4sf)
7609 v4sf __builtin_ia32_andps (v4sf, v4sf)
7610 v4sf __builtin_ia32_andnps (v4sf, v4sf)
7611 v4sf __builtin_ia32_orps (v4sf, v4sf)
7612 v4sf __builtin_ia32_xorps (v4sf, v4sf)
7613 v4sf __builtin_ia32_movss (v4sf, v4sf)
7614 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
7615 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
7616 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
7617 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
7618 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
7619 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
7620 v2si __builtin_ia32_cvtps2pi (v4sf)
7621 int __builtin_ia32_cvtss2si (v4sf)
7622 v2si __builtin_ia32_cvttps2pi (v4sf)
7623 int __builtin_ia32_cvttss2si (v4sf)
7624 v4sf __builtin_ia32_rcpps (v4sf)
7625 v4sf __builtin_ia32_rsqrtps (v4sf)
7626 v4sf __builtin_ia32_sqrtps (v4sf)
7627 v4sf __builtin_ia32_rcpss (v4sf)
7628 v4sf __builtin_ia32_rsqrtss (v4sf)
7629 v4sf __builtin_ia32_sqrtss (v4sf)
7630 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
7631 void __builtin_ia32_movntps (float *, v4sf)
7632 int __builtin_ia32_movmskps (v4sf)
7635 The following built-in functions are available when @option{-msse} is used.
7638 @item v4sf __builtin_ia32_loadaps (float *)
7639 Generates the @code{movaps} machine instruction as a load from memory.
7640 @item void __builtin_ia32_storeaps (float *, v4sf)
7641 Generates the @code{movaps} machine instruction as a store to memory.
7642 @item v4sf __builtin_ia32_loadups (float *)
7643 Generates the @code{movups} machine instruction as a load from memory.
7644 @item void __builtin_ia32_storeups (float *, v4sf)
7645 Generates the @code{movups} machine instruction as a store to memory.
7646 @item v4sf __builtin_ia32_loadsss (float *)
7647 Generates the @code{movss} machine instruction as a load from memory.
7648 @item void __builtin_ia32_storess (float *, v4sf)
7649 Generates the @code{movss} machine instruction as a store to memory.
7650 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
7651 Generates the @code{movhps} machine instruction as a load from memory.
7652 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
7653 Generates the @code{movlps} machine instruction as a load from memory
7654 @item void __builtin_ia32_storehps (v4sf, v2si *)
7655 Generates the @code{movhps} machine instruction as a store to memory.
7656 @item void __builtin_ia32_storelps (v4sf, v2si *)
7657 Generates the @code{movlps} machine instruction as a store to memory.
7660 The following built-in functions are available when @option{-msse2} is used.
7661 All of them generate the machine instruction that is part of the name.
7664 int __builtin_ia32_comisdeq (v2df, v2df)
7665 int __builtin_ia32_comisdlt (v2df, v2df)
7666 int __builtin_ia32_comisdle (v2df, v2df)
7667 int __builtin_ia32_comisdgt (v2df, v2df)
7668 int __builtin_ia32_comisdge (v2df, v2df)
7669 int __builtin_ia32_comisdneq (v2df, v2df)
7670 int __builtin_ia32_ucomisdeq (v2df, v2df)
7671 int __builtin_ia32_ucomisdlt (v2df, v2df)
7672 int __builtin_ia32_ucomisdle (v2df, v2df)
7673 int __builtin_ia32_ucomisdgt (v2df, v2df)
7674 int __builtin_ia32_ucomisdge (v2df, v2df)
7675 int __builtin_ia32_ucomisdneq (v2df, v2df)
7676 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
7677 v2df __builtin_ia32_cmpltpd (v2df, v2df)
7678 v2df __builtin_ia32_cmplepd (v2df, v2df)
7679 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
7680 v2df __builtin_ia32_cmpgepd (v2df, v2df)
7681 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
7682 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
7683 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
7684 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
7685 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
7686 v2df __builtin_ia32_cmpngepd (v2df, v2df)
7687 v2df __builtin_ia32_cmpordpd (v2df, v2df)
7688 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
7689 v2df __builtin_ia32_cmpltsd (v2df, v2df)
7690 v2df __builtin_ia32_cmplesd (v2df, v2df)
7691 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
7692 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
7693 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
7694 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
7695 v2df __builtin_ia32_cmpordsd (v2df, v2df)
7696 v2di __builtin_ia32_paddq (v2di, v2di)
7697 v2di __builtin_ia32_psubq (v2di, v2di)
7698 v2df __builtin_ia32_addpd (v2df, v2df)
7699 v2df __builtin_ia32_subpd (v2df, v2df)
7700 v2df __builtin_ia32_mulpd (v2df, v2df)
7701 v2df __builtin_ia32_divpd (v2df, v2df)
7702 v2df __builtin_ia32_addsd (v2df, v2df)
7703 v2df __builtin_ia32_subsd (v2df, v2df)
7704 v2df __builtin_ia32_mulsd (v2df, v2df)
7705 v2df __builtin_ia32_divsd (v2df, v2df)
7706 v2df __builtin_ia32_minpd (v2df, v2df)
7707 v2df __builtin_ia32_maxpd (v2df, v2df)
7708 v2df __builtin_ia32_minsd (v2df, v2df)
7709 v2df __builtin_ia32_maxsd (v2df, v2df)
7710 v2df __builtin_ia32_andpd (v2df, v2df)
7711 v2df __builtin_ia32_andnpd (v2df, v2df)
7712 v2df __builtin_ia32_orpd (v2df, v2df)
7713 v2df __builtin_ia32_xorpd (v2df, v2df)
7714 v2df __builtin_ia32_movsd (v2df, v2df)
7715 v2df __builtin_ia32_unpckhpd (v2df, v2df)
7716 v2df __builtin_ia32_unpcklpd (v2df, v2df)
7717 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
7718 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
7719 v4si __builtin_ia32_paddd128 (v4si, v4si)
7720 v2di __builtin_ia32_paddq128 (v2di, v2di)
7721 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
7722 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
7723 v4si __builtin_ia32_psubd128 (v4si, v4si)
7724 v2di __builtin_ia32_psubq128 (v2di, v2di)
7725 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
7726 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
7727 v2di __builtin_ia32_pand128 (v2di, v2di)
7728 v2di __builtin_ia32_pandn128 (v2di, v2di)
7729 v2di __builtin_ia32_por128 (v2di, v2di)
7730 v2di __builtin_ia32_pxor128 (v2di, v2di)
7731 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
7732 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
7733 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
7734 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
7735 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
7736 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
7737 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
7738 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
7739 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
7740 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
7741 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
7742 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
7743 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
7744 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
7745 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
7746 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
7747 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
7748 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
7749 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
7750 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
7751 v16qi __builtin_ia32_packsswb128 (v8hi, v8hi)
7752 v8hi __builtin_ia32_packssdw128 (v4si, v4si)
7753 v16qi __builtin_ia32_packuswb128 (v8hi, v8hi)
7754 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
7755 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
7756 v2df __builtin_ia32_loadupd (double *)
7757 void __builtin_ia32_storeupd (double *, v2df)
7758 v2df __builtin_ia32_loadhpd (v2df, double *)
7759 v2df __builtin_ia32_loadlpd (v2df, double *)
7760 int __builtin_ia32_movmskpd (v2df)
7761 int __builtin_ia32_pmovmskb128 (v16qi)
7762 void __builtin_ia32_movnti (int *, int)
7763 void __builtin_ia32_movntpd (double *, v2df)
7764 void __builtin_ia32_movntdq (v2df *, v2df)
7765 v4si __builtin_ia32_pshufd (v4si, int)
7766 v8hi __builtin_ia32_pshuflw (v8hi, int)
7767 v8hi __builtin_ia32_pshufhw (v8hi, int)
7768 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
7769 v2df __builtin_ia32_sqrtpd (v2df)
7770 v2df __builtin_ia32_sqrtsd (v2df)
7771 v2df __builtin_ia32_shufpd (v2df, v2df, int)
7772 v2df __builtin_ia32_cvtdq2pd (v4si)
7773 v4sf __builtin_ia32_cvtdq2ps (v4si)
7774 v4si __builtin_ia32_cvtpd2dq (v2df)
7775 v2si __builtin_ia32_cvtpd2pi (v2df)
7776 v4sf __builtin_ia32_cvtpd2ps (v2df)
7777 v4si __builtin_ia32_cvttpd2dq (v2df)
7778 v2si __builtin_ia32_cvttpd2pi (v2df)
7779 v2df __builtin_ia32_cvtpi2pd (v2si)
7780 int __builtin_ia32_cvtsd2si (v2df)
7781 int __builtin_ia32_cvttsd2si (v2df)
7782 long long __builtin_ia32_cvtsd2si64 (v2df)
7783 long long __builtin_ia32_cvttsd2si64 (v2df)
7784 v4si __builtin_ia32_cvtps2dq (v4sf)
7785 v2df __builtin_ia32_cvtps2pd (v4sf)
7786 v4si __builtin_ia32_cvttps2dq (v4sf)
7787 v2df __builtin_ia32_cvtsi2sd (v2df, int)
7788 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
7789 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
7790 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
7791 void __builtin_ia32_clflush (const void *)
7792 void __builtin_ia32_lfence (void)
7793 void __builtin_ia32_mfence (void)
7794 v16qi __builtin_ia32_loaddqu (const char *)
7795 void __builtin_ia32_storedqu (char *, v16qi)
7796 v1di __builtin_ia32_pmuludq (v2si, v2si)
7797 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
7798 v8hi __builtin_ia32_psllw128 (v8hi, v8hi)
7799 v4si __builtin_ia32_pslld128 (v4si, v4si)
7800 v2di __builtin_ia32_psllq128 (v2di, v2di)
7801 v8hi __builtin_ia32_psrlw128 (v8hi, v8hi)
7802 v4si __builtin_ia32_psrld128 (v4si, v4si)
7803 v2di __builtin_ia32_psrlq128 (v2di, v2di)
7804 v8hi __builtin_ia32_psraw128 (v8hi, v8hi)
7805 v4si __builtin_ia32_psrad128 (v4si, v4si)
7806 v2di __builtin_ia32_pslldqi128 (v2di, int)
7807 v8hi __builtin_ia32_psllwi128 (v8hi, int)
7808 v4si __builtin_ia32_pslldi128 (v4si, int)
7809 v2di __builtin_ia32_psllqi128 (v2di, int)
7810 v2di __builtin_ia32_psrldqi128 (v2di, int)
7811 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
7812 v4si __builtin_ia32_psrldi128 (v4si, int)
7813 v2di __builtin_ia32_psrlqi128 (v2di, int)
7814 v8hi __builtin_ia32_psrawi128 (v8hi, int)
7815 v4si __builtin_ia32_psradi128 (v4si, int)
7816 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
7819 The following built-in functions are available when @option{-msse3} is used.
7820 All of them generate the machine instruction that is part of the name.
7823 v2df __builtin_ia32_addsubpd (v2df, v2df)
7824 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
7825 v2df __builtin_ia32_haddpd (v2df, v2df)
7826 v4sf __builtin_ia32_haddps (v4sf, v4sf)
7827 v2df __builtin_ia32_hsubpd (v2df, v2df)
7828 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
7829 v16qi __builtin_ia32_lddqu (char const *)
7830 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
7831 v2df __builtin_ia32_movddup (v2df)
7832 v4sf __builtin_ia32_movshdup (v4sf)
7833 v4sf __builtin_ia32_movsldup (v4sf)
7834 void __builtin_ia32_mwait (unsigned int, unsigned int)
7837 The following built-in functions are available when @option{-msse3} is used.
7840 @item v2df __builtin_ia32_loadddup (double const *)
7841 Generates the @code{movddup} machine instruction as a load from memory.
7844 The following built-in functions are available when @option{-mssse3} is used.
7845 All of them generate the machine instruction that is part of the name
7849 v2si __builtin_ia32_phaddd (v2si, v2si)
7850 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
7851 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
7852 v2si __builtin_ia32_phsubd (v2si, v2si)
7853 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
7854 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
7855 v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi)
7856 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
7857 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
7858 v8qi __builtin_ia32_psignb (v8qi, v8qi)
7859 v2si __builtin_ia32_psignd (v2si, v2si)
7860 v4hi __builtin_ia32_psignw (v4hi, v4hi)
7861 long long __builtin_ia32_palignr (long long, long long, int)
7862 v8qi __builtin_ia32_pabsb (v8qi)
7863 v2si __builtin_ia32_pabsd (v2si)
7864 v4hi __builtin_ia32_pabsw (v4hi)
7867 The following built-in functions are available when @option{-mssse3} is used.
7868 All of them generate the machine instruction that is part of the name
7872 v4si __builtin_ia32_phaddd128 (v4si, v4si)
7873 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
7874 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
7875 v4si __builtin_ia32_phsubd128 (v4si, v4si)
7876 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
7877 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
7878 v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
7879 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
7880 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
7881 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
7882 v4si __builtin_ia32_psignd128 (v4si, v4si)
7883 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
7884 v2di __builtin_ia32_palignr128 (v2di, v2di, int)
7885 v16qi __builtin_ia32_pabsb128 (v16qi)
7886 v4si __builtin_ia32_pabsd128 (v4si)
7887 v8hi __builtin_ia32_pabsw128 (v8hi)
7890 The following built-in functions are available when @option{-msse4.1} is
7891 used. All of them generate the machine instruction that is part of the
7895 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
7896 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
7897 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
7898 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
7899 v2df __builtin_ia32_dppd (v2df, v2df, const int)
7900 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
7901 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
7902 v2di __builtin_ia32_movntdqa (v2di *);
7903 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
7904 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
7905 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
7906 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
7907 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
7908 v8hi __builtin_ia32_phminposuw128 (v8hi)
7909 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
7910 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
7911 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
7912 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
7913 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
7914 v4si __builtin_ia32_pminsd128 (v4si, v4si)
7915 v4si __builtin_ia32_pminud128 (v4si, v4si)
7916 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
7917 v4si __builtin_ia32_pmovsxbd128 (v16qi)
7918 v2di __builtin_ia32_pmovsxbq128 (v16qi)
7919 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
7920 v2di __builtin_ia32_pmovsxdq128 (v4si)
7921 v4si __builtin_ia32_pmovsxwd128 (v8hi)
7922 v2di __builtin_ia32_pmovsxwq128 (v8hi)
7923 v4si __builtin_ia32_pmovzxbd128 (v16qi)
7924 v2di __builtin_ia32_pmovzxbq128 (v16qi)
7925 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
7926 v2di __builtin_ia32_pmovzxdq128 (v4si)
7927 v4si __builtin_ia32_pmovzxwd128 (v8hi)
7928 v2di __builtin_ia32_pmovzxwq128 (v8hi)
7929 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
7930 v4si __builtin_ia32_pmulld128 (v4si, v4si)
7931 int __builtin_ia32_ptestc128 (v2di, v2di)
7932 int __builtin_ia32_ptestnzc128 (v2di, v2di)
7933 int __builtin_ia32_ptestz128 (v2di, v2di)
7934 v2df __builtin_ia32_roundpd (v2df, const int)
7935 v4sf __builtin_ia32_roundps (v4sf, const int)
7936 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
7937 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
7940 The following built-in functions are available when @option{-msse4.1} is
7944 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
7945 Generates the @code{insertps} machine instruction.
7946 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
7947 Generates the @code{pextrb} machine instruction.
7948 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
7949 Generates the @code{pinsrb} machine instruction.
7950 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
7951 Generates the @code{pinsrd} machine instruction.
7952 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
7953 Generates the @code{pinsrq} machine instruction in 64bit mode.
7956 The following built-in functions are changed to generate new SSE4.1
7957 instructions when @option{-msse4.1} is used.
7960 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
7961 Generates the @code{extractps} machine instruction.
7962 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
7963 Generates the @code{pextrd} machine instruction.
7964 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
7965 Generates the @code{pextrq} machine instruction in 64bit mode.
7968 The following built-in functions are available when @option{-msse4.2} is
7969 used. All of them generate the machine instruction that is part of the
7973 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
7974 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
7975 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
7976 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
7977 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
7978 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
7979 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
7980 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
7981 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
7982 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
7983 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
7984 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
7985 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
7986 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
7987 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
7990 The following built-in functions are available when @option{-msse4.2} is
7994 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
7995 Generates the @code{crc32b} machine instruction.
7996 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
7997 Generates the @code{crc32w} machine instruction.
7998 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
7999 Generates the @code{crc32l} machine instruction.
8000 @item unsigned long long __builtin_ia32_crc32di (unsigned int, unsigned long long)
8003 The following built-in functions are changed to generate new SSE4.2
8004 instructions when @option{-msse4.2} is used.
8007 @item int __builtin_popcount (unsigned int)
8008 Generates the @code{popcntl} machine instruction.
8009 @item int __builtin_popcountl (unsigned long)
8010 Generates the @code{popcntl} or @code{popcntq} machine instruction,
8011 depending on the size of @code{unsigned long}.
8012 @item int __builtin_popcountll (unsigned long long)
8013 Generates the @code{popcntq} machine instruction.
8016 The following built-in functions are available when @option{-maes} is
8017 used. All of them generate the machine instruction that is part of the
8021 v2di __builtin_ia32_aesenc128 (v2di, v2di)
8022 v2di __builtin_ia32_aesenclast128 (v2di, v2di)
8023 v2di __builtin_ia32_aesdec128 (v2di, v2di)
8024 v2di __builtin_ia32_aesdeclast128 (v2di, v2di)
8025 v2di __builtin_ia32_aeskeygenassist128 (v2di, const int)
8026 v2di __builtin_ia32_aesimc128 (v2di)
8029 The following built-in function is available when @option{-mpclmul} is
8033 @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int)
8034 Generates the @code{pclmulqdq} machine instruction.
8037 The following built-in functions are available when @option{-msse4a} is used.
8038 All of them generate the machine instruction that is part of the name.
8041 void __builtin_ia32_movntsd (double *, v2df)
8042 void __builtin_ia32_movntss (float *, v4sf)
8043 v2di __builtin_ia32_extrq (v2di, v16qi)
8044 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
8045 v2di __builtin_ia32_insertq (v2di, v2di)
8046 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
8049 The following built-in functions are available when @option{-msse5} is used.
8050 All of them generate the machine instruction that is part of the name
8054 v2df __builtin_ia32_comeqpd (v2df, v2df)
8055 v2df __builtin_ia32_comeqps (v2df, v2df)
8056 v4sf __builtin_ia32_comeqsd (v4sf, v4sf)
8057 v4sf __builtin_ia32_comeqss (v4sf, v4sf)
8058 v2df __builtin_ia32_comfalsepd (v2df, v2df)
8059 v2df __builtin_ia32_comfalseps (v2df, v2df)
8060 v4sf __builtin_ia32_comfalsesd (v4sf, v4sf)
8061 v4sf __builtin_ia32_comfalsess (v4sf, v4sf)
8062 v2df __builtin_ia32_comgepd (v2df, v2df)
8063 v2df __builtin_ia32_comgeps (v2df, v2df)
8064 v4sf __builtin_ia32_comgesd (v4sf, v4sf)
8065 v4sf __builtin_ia32_comgess (v4sf, v4sf)
8066 v2df __builtin_ia32_comgtpd (v2df, v2df)
8067 v2df __builtin_ia32_comgtps (v2df, v2df)
8068 v4sf __builtin_ia32_comgtsd (v4sf, v4sf)
8069 v4sf __builtin_ia32_comgtss (v4sf, v4sf)
8070 v2df __builtin_ia32_comlepd (v2df, v2df)
8071 v2df __builtin_ia32_comleps (v2df, v2df)
8072 v4sf __builtin_ia32_comlesd (v4sf, v4sf)
8073 v4sf __builtin_ia32_comless (v4sf, v4sf)
8074 v2df __builtin_ia32_comltpd (v2df, v2df)
8075 v2df __builtin_ia32_comltps (v2df, v2df)
8076 v4sf __builtin_ia32_comltsd (v4sf, v4sf)
8077 v4sf __builtin_ia32_comltss (v4sf, v4sf)
8078 v2df __builtin_ia32_comnepd (v2df, v2df)
8079 v2df __builtin_ia32_comneps (v2df, v2df)
8080 v4sf __builtin_ia32_comnesd (v4sf, v4sf)
8081 v4sf __builtin_ia32_comness (v4sf, v4sf)
8082 v2df __builtin_ia32_comordpd (v2df, v2df)
8083 v2df __builtin_ia32_comordps (v2df, v2df)
8084 v4sf __builtin_ia32_comordsd (v4sf, v4sf)
8085 v4sf __builtin_ia32_comordss (v4sf, v4sf)
8086 v2df __builtin_ia32_comtruepd (v2df, v2df)
8087 v2df __builtin_ia32_comtrueps (v2df, v2df)
8088 v4sf __builtin_ia32_comtruesd (v4sf, v4sf)
8089 v4sf __builtin_ia32_comtruess (v4sf, v4sf)
8090 v2df __builtin_ia32_comueqpd (v2df, v2df)
8091 v2df __builtin_ia32_comueqps (v2df, v2df)
8092 v4sf __builtin_ia32_comueqsd (v4sf, v4sf)
8093 v4sf __builtin_ia32_comueqss (v4sf, v4sf)
8094 v2df __builtin_ia32_comugepd (v2df, v2df)
8095 v2df __builtin_ia32_comugeps (v2df, v2df)
8096 v4sf __builtin_ia32_comugesd (v4sf, v4sf)
8097 v4sf __builtin_ia32_comugess (v4sf, v4sf)
8098 v2df __builtin_ia32_comugtpd (v2df, v2df)
8099 v2df __builtin_ia32_comugtps (v2df, v2df)
8100 v4sf __builtin_ia32_comugtsd (v4sf, v4sf)
8101 v4sf __builtin_ia32_comugtss (v4sf, v4sf)
8102 v2df __builtin_ia32_comulepd (v2df, v2df)
8103 v2df __builtin_ia32_comuleps (v2df, v2df)
8104 v4sf __builtin_ia32_comulesd (v4sf, v4sf)
8105 v4sf __builtin_ia32_comuless (v4sf, v4sf)
8106 v2df __builtin_ia32_comultpd (v2df, v2df)
8107 v2df __builtin_ia32_comultps (v2df, v2df)
8108 v4sf __builtin_ia32_comultsd (v4sf, v4sf)
8109 v4sf __builtin_ia32_comultss (v4sf, v4sf)
8110 v2df __builtin_ia32_comunepd (v2df, v2df)
8111 v2df __builtin_ia32_comuneps (v2df, v2df)
8112 v4sf __builtin_ia32_comunesd (v4sf, v4sf)
8113 v4sf __builtin_ia32_comuness (v4sf, v4sf)
8114 v2df __builtin_ia32_comunordpd (v2df, v2df)
8115 v2df __builtin_ia32_comunordps (v2df, v2df)
8116 v4sf __builtin_ia32_comunordsd (v4sf, v4sf)
8117 v4sf __builtin_ia32_comunordss (v4sf, v4sf)
8118 v2df __builtin_ia32_fmaddpd (v2df, v2df, v2df)
8119 v4sf __builtin_ia32_fmaddps (v4sf, v4sf, v4sf)
8120 v2df __builtin_ia32_fmaddsd (v2df, v2df, v2df)
8121 v4sf __builtin_ia32_fmaddss (v4sf, v4sf, v4sf)
8122 v2df __builtin_ia32_fmsubpd (v2df, v2df, v2df)
8123 v4sf __builtin_ia32_fmsubps (v4sf, v4sf, v4sf)
8124 v2df __builtin_ia32_fmsubsd (v2df, v2df, v2df)
8125 v4sf __builtin_ia32_fmsubss (v4sf, v4sf, v4sf)
8126 v2df __builtin_ia32_fnmaddpd (v2df, v2df, v2df)
8127 v4sf __builtin_ia32_fnmaddps (v4sf, v4sf, v4sf)
8128 v2df __builtin_ia32_fnmaddsd (v2df, v2df, v2df)
8129 v4sf __builtin_ia32_fnmaddss (v4sf, v4sf, v4sf)
8130 v2df __builtin_ia32_fnmsubpd (v2df, v2df, v2df)
8131 v4sf __builtin_ia32_fnmsubps (v4sf, v4sf, v4sf)
8132 v2df __builtin_ia32_fnmsubsd (v2df, v2df, v2df)
8133 v4sf __builtin_ia32_fnmsubss (v4sf, v4sf, v4sf)
8134 v2df __builtin_ia32_frczpd (v2df)
8135 v4sf __builtin_ia32_frczps (v4sf)
8136 v2df __builtin_ia32_frczsd (v2df, v2df)
8137 v4sf __builtin_ia32_frczss (v4sf, v4sf)
8138 v2di __builtin_ia32_pcmov (v2di, v2di, v2di)
8139 v2di __builtin_ia32_pcmov_v2di (v2di, v2di, v2di)
8140 v4si __builtin_ia32_pcmov_v4si (v4si, v4si, v4si)
8141 v8hi __builtin_ia32_pcmov_v8hi (v8hi, v8hi, v8hi)
8142 v16qi __builtin_ia32_pcmov_v16qi (v16qi, v16qi, v16qi)
8143 v2df __builtin_ia32_pcmov_v2df (v2df, v2df, v2df)
8144 v4sf __builtin_ia32_pcmov_v4sf (v4sf, v4sf, v4sf)
8145 v16qi __builtin_ia32_pcomeqb (v16qi, v16qi)
8146 v8hi __builtin_ia32_pcomeqw (v8hi, v8hi)
8147 v4si __builtin_ia32_pcomeqd (v4si, v4si)
8148 v2di __builtin_ia32_pcomeqq (v2di, v2di)
8149 v16qi __builtin_ia32_pcomequb (v16qi, v16qi)
8150 v4si __builtin_ia32_pcomequd (v4si, v4si)
8151 v2di __builtin_ia32_pcomequq (v2di, v2di)
8152 v8hi __builtin_ia32_pcomequw (v8hi, v8hi)
8153 v8hi __builtin_ia32_pcomeqw (v8hi, v8hi)
8154 v16qi __builtin_ia32_pcomfalseb (v16qi, v16qi)
8155 v4si __builtin_ia32_pcomfalsed (v4si, v4si)
8156 v2di __builtin_ia32_pcomfalseq (v2di, v2di)
8157 v16qi __builtin_ia32_pcomfalseub (v16qi, v16qi)
8158 v4si __builtin_ia32_pcomfalseud (v4si, v4si)
8159 v2di __builtin_ia32_pcomfalseuq (v2di, v2di)
8160 v8hi __builtin_ia32_pcomfalseuw (v8hi, v8hi)
8161 v8hi __builtin_ia32_pcomfalsew (v8hi, v8hi)
8162 v16qi __builtin_ia32_pcomgeb (v16qi, v16qi)
8163 v4si __builtin_ia32_pcomged (v4si, v4si)
8164 v2di __builtin_ia32_pcomgeq (v2di, v2di)
8165 v16qi __builtin_ia32_pcomgeub (v16qi, v16qi)
8166 v4si __builtin_ia32_pcomgeud (v4si, v4si)
8167 v2di __builtin_ia32_pcomgeuq (v2di, v2di)
8168 v8hi __builtin_ia32_pcomgeuw (v8hi, v8hi)
8169 v8hi __builtin_ia32_pcomgew (v8hi, v8hi)
8170 v16qi __builtin_ia32_pcomgtb (v16qi, v16qi)
8171 v4si __builtin_ia32_pcomgtd (v4si, v4si)
8172 v2di __builtin_ia32_pcomgtq (v2di, v2di)
8173 v16qi __builtin_ia32_pcomgtub (v16qi, v16qi)
8174 v4si __builtin_ia32_pcomgtud (v4si, v4si)
8175 v2di __builtin_ia32_pcomgtuq (v2di, v2di)
8176 v8hi __builtin_ia32_pcomgtuw (v8hi, v8hi)
8177 v8hi __builtin_ia32_pcomgtw (v8hi, v8hi)
8178 v16qi __builtin_ia32_pcomleb (v16qi, v16qi)
8179 v4si __builtin_ia32_pcomled (v4si, v4si)
8180 v2di __builtin_ia32_pcomleq (v2di, v2di)
8181 v16qi __builtin_ia32_pcomleub (v16qi, v16qi)
8182 v4si __builtin_ia32_pcomleud (v4si, v4si)
8183 v2di __builtin_ia32_pcomleuq (v2di, v2di)
8184 v8hi __builtin_ia32_pcomleuw (v8hi, v8hi)
8185 v8hi __builtin_ia32_pcomlew (v8hi, v8hi)
8186 v16qi __builtin_ia32_pcomltb (v16qi, v16qi)
8187 v4si __builtin_ia32_pcomltd (v4si, v4si)
8188 v2di __builtin_ia32_pcomltq (v2di, v2di)
8189 v16qi __builtin_ia32_pcomltub (v16qi, v16qi)
8190 v4si __builtin_ia32_pcomltud (v4si, v4si)
8191 v2di __builtin_ia32_pcomltuq (v2di, v2di)
8192 v8hi __builtin_ia32_pcomltuw (v8hi, v8hi)
8193 v8hi __builtin_ia32_pcomltw (v8hi, v8hi)
8194 v16qi __builtin_ia32_pcomneb (v16qi, v16qi)
8195 v4si __builtin_ia32_pcomned (v4si, v4si)
8196 v2di __builtin_ia32_pcomneq (v2di, v2di)
8197 v16qi __builtin_ia32_pcomneub (v16qi, v16qi)
8198 v4si __builtin_ia32_pcomneud (v4si, v4si)
8199 v2di __builtin_ia32_pcomneuq (v2di, v2di)
8200 v8hi __builtin_ia32_pcomneuw (v8hi, v8hi)
8201 v8hi __builtin_ia32_pcomnew (v8hi, v8hi)
8202 v16qi __builtin_ia32_pcomtrueb (v16qi, v16qi)
8203 v4si __builtin_ia32_pcomtrued (v4si, v4si)
8204 v2di __builtin_ia32_pcomtrueq (v2di, v2di)
8205 v16qi __builtin_ia32_pcomtrueub (v16qi, v16qi)
8206 v4si __builtin_ia32_pcomtrueud (v4si, v4si)
8207 v2di __builtin_ia32_pcomtrueuq (v2di, v2di)
8208 v8hi __builtin_ia32_pcomtrueuw (v8hi, v8hi)
8209 v8hi __builtin_ia32_pcomtruew (v8hi, v8hi)
8210 v4df __builtin_ia32_permpd (v2df, v2df, v16qi)
8211 v4sf __builtin_ia32_permps (v4sf, v4sf, v16qi)
8212 v4si __builtin_ia32_phaddbd (v16qi)
8213 v2di __builtin_ia32_phaddbq (v16qi)
8214 v8hi __builtin_ia32_phaddbw (v16qi)
8215 v2di __builtin_ia32_phadddq (v4si)
8216 v4si __builtin_ia32_phaddubd (v16qi)
8217 v2di __builtin_ia32_phaddubq (v16qi)
8218 v8hi __builtin_ia32_phaddubw (v16qi)
8219 v2di __builtin_ia32_phaddudq (v4si)
8220 v4si __builtin_ia32_phadduwd (v8hi)
8221 v2di __builtin_ia32_phadduwq (v8hi)
8222 v4si __builtin_ia32_phaddwd (v8hi)
8223 v2di __builtin_ia32_phaddwq (v8hi)
8224 v8hi __builtin_ia32_phsubbw (v16qi)
8225 v2di __builtin_ia32_phsubdq (v4si)
8226 v4si __builtin_ia32_phsubwd (v8hi)
8227 v4si __builtin_ia32_pmacsdd (v4si, v4si, v4si)
8228 v2di __builtin_ia32_pmacsdqh (v4si, v4si, v2di)
8229 v2di __builtin_ia32_pmacsdql (v4si, v4si, v2di)
8230 v4si __builtin_ia32_pmacssdd (v4si, v4si, v4si)
8231 v2di __builtin_ia32_pmacssdqh (v4si, v4si, v2di)
8232 v2di __builtin_ia32_pmacssdql (v4si, v4si, v2di)
8233 v4si __builtin_ia32_pmacsswd (v8hi, v8hi, v4si)
8234 v8hi __builtin_ia32_pmacssww (v8hi, v8hi, v8hi)
8235 v4si __builtin_ia32_pmacswd (v8hi, v8hi, v4si)
8236 v8hi __builtin_ia32_pmacsww (v8hi, v8hi, v8hi)
8237 v4si __builtin_ia32_pmadcsswd (v8hi, v8hi, v4si)
8238 v4si __builtin_ia32_pmadcswd (v8hi, v8hi, v4si)
8239 v16qi __builtin_ia32_pperm (v16qi, v16qi, v16qi)
8240 v16qi __builtin_ia32_protb (v16qi, v16qi)
8241 v4si __builtin_ia32_protd (v4si, v4si)
8242 v2di __builtin_ia32_protq (v2di, v2di)
8243 v8hi __builtin_ia32_protw (v8hi, v8hi)
8244 v16qi __builtin_ia32_pshab (v16qi, v16qi)
8245 v4si __builtin_ia32_pshad (v4si, v4si)
8246 v2di __builtin_ia32_pshaq (v2di, v2di)
8247 v8hi __builtin_ia32_pshaw (v8hi, v8hi)
8248 v16qi __builtin_ia32_pshlb (v16qi, v16qi)
8249 v4si __builtin_ia32_pshld (v4si, v4si)
8250 v2di __builtin_ia32_pshlq (v2di, v2di)
8251 v8hi __builtin_ia32_pshlw (v8hi, v8hi)
8254 The following builtin-in functions are available when @option{-msse5}
8255 is used. The second argument must be an integer constant and generate
8256 the machine instruction that is part of the name with the @samp{_imm}
8260 v16qi __builtin_ia32_protb_imm (v16qi, int)
8261 v4si __builtin_ia32_protd_imm (v4si, int)
8262 v2di __builtin_ia32_protq_imm (v2di, int)
8263 v8hi __builtin_ia32_protw_imm (v8hi, int)
8266 The following built-in functions are available when @option{-m3dnow} is used.
8267 All of them generate the machine instruction that is part of the name.
8270 void __builtin_ia32_femms (void)
8271 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
8272 v2si __builtin_ia32_pf2id (v2sf)
8273 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
8274 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
8275 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
8276 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
8277 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
8278 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
8279 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
8280 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
8281 v2sf __builtin_ia32_pfrcp (v2sf)
8282 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
8283 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
8284 v2sf __builtin_ia32_pfrsqrt (v2sf)
8285 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
8286 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
8287 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
8288 v2sf __builtin_ia32_pi2fd (v2si)
8289 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
8292 The following built-in functions are available when both @option{-m3dnow}
8293 and @option{-march=athlon} are used. All of them generate the machine
8294 instruction that is part of the name.
8297 v2si __builtin_ia32_pf2iw (v2sf)
8298 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
8299 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
8300 v2sf __builtin_ia32_pi2fw (v2si)
8301 v2sf __builtin_ia32_pswapdsf (v2sf)
8302 v2si __builtin_ia32_pswapdsi (v2si)
8305 @node MIPS DSP Built-in Functions
8306 @subsection MIPS DSP Built-in Functions
8308 The MIPS DSP Application-Specific Extension (ASE) includes new
8309 instructions that are designed to improve the performance of DSP and
8310 media applications. It provides instructions that operate on packed
8311 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
8313 GCC supports MIPS DSP operations using both the generic
8314 vector extensions (@pxref{Vector Extensions}) and a collection of
8315 MIPS-specific built-in functions. Both kinds of support are
8316 enabled by the @option{-mdsp} command-line option.
8318 Revision 2 of the ASE was introduced in the second half of 2006.
8319 This revision adds extra instructions to the original ASE, but is
8320 otherwise backwards-compatible with it. You can select revision 2
8321 using the command-line option @option{-mdspr2}; this option implies
8324 At present, GCC only provides support for operations on 32-bit
8325 vectors. The vector type associated with 8-bit integer data is
8326 usually called @code{v4i8}, the vector type associated with Q7
8327 is usually called @code{v4q7}, the vector type associated with 16-bit
8328 integer data is usually called @code{v2i16}, and the vector type
8329 associated with Q15 is usually called @code{v2q15}. They can be
8330 defined in C as follows:
8333 typedef signed char v4i8 __attribute__ ((vector_size(4)));
8334 typedef signed char v4q7 __attribute__ ((vector_size(4)));
8335 typedef short v2i16 __attribute__ ((vector_size(4)));
8336 typedef short v2q15 __attribute__ ((vector_size(4)));
8339 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
8340 initialized in the same way as aggregates. For example:
8343 v4i8 a = @{1, 2, 3, 4@};
8345 b = (v4i8) @{5, 6, 7, 8@};
8347 v2q15 c = @{0x0fcb, 0x3a75@};
8349 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
8352 @emph{Note:} The CPU's endianness determines the order in which values
8353 are packed. On little-endian targets, the first value is the least
8354 significant and the last value is the most significant. The opposite
8355 order applies to big-endian targets. For example, the code above will
8356 set the lowest byte of @code{a} to @code{1} on little-endian targets
8357 and @code{4} on big-endian targets.
8359 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
8360 representation. As shown in this example, the integer representation
8361 of a Q7 value can be obtained by multiplying the fractional value by
8362 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
8363 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
8366 The table below lists the @code{v4i8} and @code{v2q15} operations for which
8367 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
8368 and @code{c} and @code{d} are @code{v2q15} values.
8370 @multitable @columnfractions .50 .50
8371 @item C code @tab MIPS instruction
8372 @item @code{a + b} @tab @code{addu.qb}
8373 @item @code{c + d} @tab @code{addq.ph}
8374 @item @code{a - b} @tab @code{subu.qb}
8375 @item @code{c - d} @tab @code{subq.ph}
8378 The table below lists the @code{v2i16} operation for which
8379 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
8380 @code{v2i16} values.
8382 @multitable @columnfractions .50 .50
8383 @item C code @tab MIPS instruction
8384 @item @code{e * f} @tab @code{mul.ph}
8387 It is easier to describe the DSP built-in functions if we first define
8388 the following types:
8393 typedef unsigned int ui32;
8394 typedef long long a64;
8397 @code{q31} and @code{i32} are actually the same as @code{int}, but we
8398 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
8399 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
8400 @code{long long}, but we use @code{a64} to indicate values that will
8401 be placed in one of the four DSP accumulators (@code{$ac0},
8402 @code{$ac1}, @code{$ac2} or @code{$ac3}).
8404 Also, some built-in functions prefer or require immediate numbers as
8405 parameters, because the corresponding DSP instructions accept both immediate
8406 numbers and register operands, or accept immediate numbers only. The
8407 immediate parameters are listed as follows.
8416 imm_n32_31: -32 to 31.
8417 imm_n512_511: -512 to 511.
8420 The following built-in functions map directly to a particular MIPS DSP
8421 instruction. Please refer to the architecture specification
8422 for details on what each instruction does.
8425 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
8426 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
8427 q31 __builtin_mips_addq_s_w (q31, q31)
8428 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
8429 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
8430 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
8431 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
8432 q31 __builtin_mips_subq_s_w (q31, q31)
8433 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
8434 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
8435 i32 __builtin_mips_addsc (i32, i32)
8436 i32 __builtin_mips_addwc (i32, i32)
8437 i32 __builtin_mips_modsub (i32, i32)
8438 i32 __builtin_mips_raddu_w_qb (v4i8)
8439 v2q15 __builtin_mips_absq_s_ph (v2q15)
8440 q31 __builtin_mips_absq_s_w (q31)
8441 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
8442 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
8443 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
8444 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
8445 q31 __builtin_mips_preceq_w_phl (v2q15)
8446 q31 __builtin_mips_preceq_w_phr (v2q15)
8447 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
8448 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
8449 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
8450 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
8451 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
8452 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
8453 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
8454 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
8455 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
8456 v4i8 __builtin_mips_shll_qb (v4i8, i32)
8457 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
8458 v2q15 __builtin_mips_shll_ph (v2q15, i32)
8459 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
8460 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
8461 q31 __builtin_mips_shll_s_w (q31, imm0_31)
8462 q31 __builtin_mips_shll_s_w (q31, i32)
8463 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
8464 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
8465 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
8466 v2q15 __builtin_mips_shra_ph (v2q15, i32)
8467 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
8468 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
8469 q31 __builtin_mips_shra_r_w (q31, imm0_31)
8470 q31 __builtin_mips_shra_r_w (q31, i32)
8471 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
8472 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
8473 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
8474 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
8475 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
8476 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
8477 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
8478 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
8479 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
8480 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
8481 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
8482 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
8483 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
8484 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
8485 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
8486 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
8487 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
8488 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
8489 i32 __builtin_mips_bitrev (i32)
8490 i32 __builtin_mips_insv (i32, i32)
8491 v4i8 __builtin_mips_repl_qb (imm0_255)
8492 v4i8 __builtin_mips_repl_qb (i32)
8493 v2q15 __builtin_mips_repl_ph (imm_n512_511)
8494 v2q15 __builtin_mips_repl_ph (i32)
8495 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
8496 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
8497 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
8498 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
8499 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
8500 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
8501 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
8502 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
8503 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
8504 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
8505 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
8506 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
8507 i32 __builtin_mips_extr_w (a64, imm0_31)
8508 i32 __builtin_mips_extr_w (a64, i32)
8509 i32 __builtin_mips_extr_r_w (a64, imm0_31)
8510 i32 __builtin_mips_extr_s_h (a64, i32)
8511 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
8512 i32 __builtin_mips_extr_rs_w (a64, i32)
8513 i32 __builtin_mips_extr_s_h (a64, imm0_31)
8514 i32 __builtin_mips_extr_r_w (a64, i32)
8515 i32 __builtin_mips_extp (a64, imm0_31)
8516 i32 __builtin_mips_extp (a64, i32)
8517 i32 __builtin_mips_extpdp (a64, imm0_31)
8518 i32 __builtin_mips_extpdp (a64, i32)
8519 a64 __builtin_mips_shilo (a64, imm_n32_31)
8520 a64 __builtin_mips_shilo (a64, i32)
8521 a64 __builtin_mips_mthlip (a64, i32)
8522 void __builtin_mips_wrdsp (i32, imm0_63)
8523 i32 __builtin_mips_rddsp (imm0_63)
8524 i32 __builtin_mips_lbux (void *, i32)
8525 i32 __builtin_mips_lhx (void *, i32)
8526 i32 __builtin_mips_lwx (void *, i32)
8527 i32 __builtin_mips_bposge32 (void)
8530 The following built-in functions map directly to a particular MIPS DSP REV 2
8531 instruction. Please refer to the architecture specification
8532 for details on what each instruction does.
8535 v4q7 __builtin_mips_absq_s_qb (v4q7);
8536 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
8537 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
8538 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
8539 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
8540 i32 __builtin_mips_append (i32, i32, imm0_31);
8541 i32 __builtin_mips_balign (i32, i32, imm0_3);
8542 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
8543 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
8544 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
8545 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
8546 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
8547 a64 __builtin_mips_madd (a64, i32, i32);
8548 a64 __builtin_mips_maddu (a64, ui32, ui32);
8549 a64 __builtin_mips_msub (a64, i32, i32);
8550 a64 __builtin_mips_msubu (a64, ui32, ui32);
8551 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
8552 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
8553 q31 __builtin_mips_mulq_rs_w (q31, q31);
8554 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
8555 q31 __builtin_mips_mulq_s_w (q31, q31);
8556 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
8557 a64 __builtin_mips_mult (i32, i32);
8558 a64 __builtin_mips_multu (ui32, ui32);
8559 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
8560 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
8561 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
8562 i32 __builtin_mips_prepend (i32, i32, imm0_31);
8563 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
8564 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
8565 v4i8 __builtin_mips_shra_qb (v4i8, i32);
8566 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
8567 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
8568 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
8569 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
8570 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
8571 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
8572 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
8573 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
8574 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
8575 q31 __builtin_mips_addqh_w (q31, q31);
8576 q31 __builtin_mips_addqh_r_w (q31, q31);
8577 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
8578 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
8579 q31 __builtin_mips_subqh_w (q31, q31);
8580 q31 __builtin_mips_subqh_r_w (q31, q31);
8581 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
8582 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
8583 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
8584 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
8585 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
8586 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
8590 @node MIPS Paired-Single Support
8591 @subsection MIPS Paired-Single Support
8593 The MIPS64 architecture includes a number of instructions that
8594 operate on pairs of single-precision floating-point values.
8595 Each pair is packed into a 64-bit floating-point register,
8596 with one element being designated the ``upper half'' and
8597 the other being designated the ``lower half''.
8599 GCC supports paired-single operations using both the generic
8600 vector extensions (@pxref{Vector Extensions}) and a collection of
8601 MIPS-specific built-in functions. Both kinds of support are
8602 enabled by the @option{-mpaired-single} command-line option.
8604 The vector type associated with paired-single values is usually
8605 called @code{v2sf}. It can be defined in C as follows:
8608 typedef float v2sf __attribute__ ((vector_size (8)));
8611 @code{v2sf} values are initialized in the same way as aggregates.
8615 v2sf a = @{1.5, 9.1@};
8618 b = (v2sf) @{e, f@};
8621 @emph{Note:} The CPU's endianness determines which value is stored in
8622 the upper half of a register and which value is stored in the lower half.
8623 On little-endian targets, the first value is the lower one and the second
8624 value is the upper one. The opposite order applies to big-endian targets.
8625 For example, the code above will set the lower half of @code{a} to
8626 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
8629 * Paired-Single Arithmetic::
8630 * Paired-Single Built-in Functions::
8631 * MIPS-3D Built-in Functions::
8634 @node Paired-Single Arithmetic
8635 @subsubsection Paired-Single Arithmetic
8637 The table below lists the @code{v2sf} operations for which hardware
8638 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
8639 values and @code{x} is an integral value.
8641 @multitable @columnfractions .50 .50
8642 @item C code @tab MIPS instruction
8643 @item @code{a + b} @tab @code{add.ps}
8644 @item @code{a - b} @tab @code{sub.ps}
8645 @item @code{-a} @tab @code{neg.ps}
8646 @item @code{a * b} @tab @code{mul.ps}
8647 @item @code{a * b + c} @tab @code{madd.ps}
8648 @item @code{a * b - c} @tab @code{msub.ps}
8649 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
8650 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
8651 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
8654 Note that the multiply-accumulate instructions can be disabled
8655 using the command-line option @code{-mno-fused-madd}.
8657 @node Paired-Single Built-in Functions
8658 @subsubsection Paired-Single Built-in Functions
8660 The following paired-single functions map directly to a particular
8661 MIPS instruction. Please refer to the architecture specification
8662 for details on what each instruction does.
8665 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
8666 Pair lower lower (@code{pll.ps}).
8668 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
8669 Pair upper lower (@code{pul.ps}).
8671 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
8672 Pair lower upper (@code{plu.ps}).
8674 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
8675 Pair upper upper (@code{puu.ps}).
8677 @item v2sf __builtin_mips_cvt_ps_s (float, float)
8678 Convert pair to paired single (@code{cvt.ps.s}).
8680 @item float __builtin_mips_cvt_s_pl (v2sf)
8681 Convert pair lower to single (@code{cvt.s.pl}).
8683 @item float __builtin_mips_cvt_s_pu (v2sf)
8684 Convert pair upper to single (@code{cvt.s.pu}).
8686 @item v2sf __builtin_mips_abs_ps (v2sf)
8687 Absolute value (@code{abs.ps}).
8689 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
8690 Align variable (@code{alnv.ps}).
8692 @emph{Note:} The value of the third parameter must be 0 or 4
8693 modulo 8, otherwise the result will be unpredictable. Please read the
8694 instruction description for details.
8697 The following multi-instruction functions are also available.
8698 In each case, @var{cond} can be any of the 16 floating-point conditions:
8699 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
8700 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
8701 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
8704 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8705 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8706 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
8707 @code{movt.ps}/@code{movf.ps}).
8709 The @code{movt} functions return the value @var{x} computed by:
8712 c.@var{cond}.ps @var{cc},@var{a},@var{b}
8713 mov.ps @var{x},@var{c}
8714 movt.ps @var{x},@var{d},@var{cc}
8717 The @code{movf} functions are similar but use @code{movf.ps} instead
8720 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8721 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8722 Comparison of two paired-single values (@code{c.@var{cond}.ps},
8723 @code{bc1t}/@code{bc1f}).
8725 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
8726 and return either the upper or lower half of the result. For example:
8730 if (__builtin_mips_upper_c_eq_ps (a, b))
8731 upper_halves_are_equal ();
8733 upper_halves_are_unequal ();
8735 if (__builtin_mips_lower_c_eq_ps (a, b))
8736 lower_halves_are_equal ();
8738 lower_halves_are_unequal ();
8742 @node MIPS-3D Built-in Functions
8743 @subsubsection MIPS-3D Built-in Functions
8745 The MIPS-3D Application-Specific Extension (ASE) includes additional
8746 paired-single instructions that are designed to improve the performance
8747 of 3D graphics operations. Support for these instructions is controlled
8748 by the @option{-mips3d} command-line option.
8750 The functions listed below map directly to a particular MIPS-3D
8751 instruction. Please refer to the architecture specification for
8752 more details on what each instruction does.
8755 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
8756 Reduction add (@code{addr.ps}).
8758 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
8759 Reduction multiply (@code{mulr.ps}).
8761 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
8762 Convert paired single to paired word (@code{cvt.pw.ps}).
8764 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
8765 Convert paired word to paired single (@code{cvt.ps.pw}).
8767 @item float __builtin_mips_recip1_s (float)
8768 @itemx double __builtin_mips_recip1_d (double)
8769 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
8770 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
8772 @item float __builtin_mips_recip2_s (float, float)
8773 @itemx double __builtin_mips_recip2_d (double, double)
8774 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
8775 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
8777 @item float __builtin_mips_rsqrt1_s (float)
8778 @itemx double __builtin_mips_rsqrt1_d (double)
8779 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
8780 Reduced precision reciprocal square root (sequence step 1)
8781 (@code{rsqrt1.@var{fmt}}).
8783 @item float __builtin_mips_rsqrt2_s (float, float)
8784 @itemx double __builtin_mips_rsqrt2_d (double, double)
8785 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
8786 Reduced precision reciprocal square root (sequence step 2)
8787 (@code{rsqrt2.@var{fmt}}).
8790 The following multi-instruction functions are also available.
8791 In each case, @var{cond} can be any of the 16 floating-point conditions:
8792 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
8793 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
8794 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
8797 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
8798 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
8799 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
8800 @code{bc1t}/@code{bc1f}).
8802 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
8803 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
8808 if (__builtin_mips_cabs_eq_s (a, b))
8814 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8815 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8816 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
8817 @code{bc1t}/@code{bc1f}).
8819 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
8820 and return either the upper or lower half of the result. For example:
8824 if (__builtin_mips_upper_cabs_eq_ps (a, b))
8825 upper_halves_are_equal ();
8827 upper_halves_are_unequal ();
8829 if (__builtin_mips_lower_cabs_eq_ps (a, b))
8830 lower_halves_are_equal ();
8832 lower_halves_are_unequal ();
8835 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8836 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8837 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
8838 @code{movt.ps}/@code{movf.ps}).
8840 The @code{movt} functions return the value @var{x} computed by:
8843 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
8844 mov.ps @var{x},@var{c}
8845 movt.ps @var{x},@var{d},@var{cc}
8848 The @code{movf} functions are similar but use @code{movf.ps} instead
8851 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8852 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8853 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8854 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8855 Comparison of two paired-single values
8856 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
8857 @code{bc1any2t}/@code{bc1any2f}).
8859 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
8860 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
8861 result is true and the @code{all} forms return true if both results are true.
8866 if (__builtin_mips_any_c_eq_ps (a, b))
8871 if (__builtin_mips_all_c_eq_ps (a, b))
8877 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8878 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8879 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8880 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8881 Comparison of four paired-single values
8882 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
8883 @code{bc1any4t}/@code{bc1any4f}).
8885 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
8886 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
8887 The @code{any} forms return true if any of the four results are true
8888 and the @code{all} forms return true if all four results are true.
8893 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
8898 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
8905 @node PowerPC AltiVec Built-in Functions
8906 @subsection PowerPC AltiVec Built-in Functions
8908 GCC provides an interface for the PowerPC family of processors to access
8909 the AltiVec operations described in Motorola's AltiVec Programming
8910 Interface Manual. The interface is made available by including
8911 @code{<altivec.h>} and using @option{-maltivec} and
8912 @option{-mabi=altivec}. The interface supports the following vector
8916 vector unsigned char
8920 vector unsigned short
8931 GCC's implementation of the high-level language interface available from
8932 C and C++ code differs from Motorola's documentation in several ways.
8937 A vector constant is a list of constant expressions within curly braces.
8940 A vector initializer requires no cast if the vector constant is of the
8941 same type as the variable it is initializing.
8944 If @code{signed} or @code{unsigned} is omitted, the signedness of the
8945 vector type is the default signedness of the base type. The default
8946 varies depending on the operating system, so a portable program should
8947 always specify the signedness.
8950 Compiling with @option{-maltivec} adds keywords @code{__vector},
8951 @code{__pixel}, and @code{__bool}. Macros @option{vector},
8952 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
8956 GCC allows using a @code{typedef} name as the type specifier for a
8960 For C, overloaded functions are implemented with macros so the following
8964 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
8967 Since @code{vec_add} is a macro, the vector constant in the example
8968 is treated as four separate arguments. Wrap the entire argument in
8969 parentheses for this to work.
8972 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
8973 Internally, GCC uses built-in functions to achieve the functionality in
8974 the aforementioned header file, but they are not supported and are
8975 subject to change without notice.
8977 The following interfaces are supported for the generic and specific
8978 AltiVec operations and the AltiVec predicates. In cases where there
8979 is a direct mapping between generic and specific operations, only the
8980 generic names are shown here, although the specific operations can also
8983 Arguments that are documented as @code{const int} require literal
8984 integral values within the range required for that operation.
8987 vector signed char vec_abs (vector signed char);
8988 vector signed short vec_abs (vector signed short);
8989 vector signed int vec_abs (vector signed int);
8990 vector float vec_abs (vector float);
8992 vector signed char vec_abss (vector signed char);
8993 vector signed short vec_abss (vector signed short);
8994 vector signed int vec_abss (vector signed int);
8996 vector signed char vec_add (vector bool char, vector signed char);
8997 vector signed char vec_add (vector signed char, vector bool char);
8998 vector signed char vec_add (vector signed char, vector signed char);
8999 vector unsigned char vec_add (vector bool char, vector unsigned char);
9000 vector unsigned char vec_add (vector unsigned char, vector bool char);
9001 vector unsigned char vec_add (vector unsigned char,
9002 vector unsigned char);
9003 vector signed short vec_add (vector bool short, vector signed short);
9004 vector signed short vec_add (vector signed short, vector bool short);
9005 vector signed short vec_add (vector signed short, vector signed short);
9006 vector unsigned short vec_add (vector bool short,
9007 vector unsigned short);
9008 vector unsigned short vec_add (vector unsigned short,
9010 vector unsigned short vec_add (vector unsigned short,
9011 vector unsigned short);
9012 vector signed int vec_add (vector bool int, vector signed int);
9013 vector signed int vec_add (vector signed int, vector bool int);
9014 vector signed int vec_add (vector signed int, vector signed int);
9015 vector unsigned int vec_add (vector bool int, vector unsigned int);
9016 vector unsigned int vec_add (vector unsigned int, vector bool int);
9017 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
9018 vector float vec_add (vector float, vector float);
9020 vector float vec_vaddfp (vector float, vector float);
9022 vector signed int vec_vadduwm (vector bool int, vector signed int);
9023 vector signed int vec_vadduwm (vector signed int, vector bool int);
9024 vector signed int vec_vadduwm (vector signed int, vector signed int);
9025 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
9026 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
9027 vector unsigned int vec_vadduwm (vector unsigned int,
9028 vector unsigned int);
9030 vector signed short vec_vadduhm (vector bool short,
9031 vector signed short);
9032 vector signed short vec_vadduhm (vector signed short,
9034 vector signed short vec_vadduhm (vector signed short,
9035 vector signed short);
9036 vector unsigned short vec_vadduhm (vector bool short,
9037 vector unsigned short);
9038 vector unsigned short vec_vadduhm (vector unsigned short,
9040 vector unsigned short vec_vadduhm (vector unsigned short,
9041 vector unsigned short);
9043 vector signed char vec_vaddubm (vector bool char, vector signed char);
9044 vector signed char vec_vaddubm (vector signed char, vector bool char);
9045 vector signed char vec_vaddubm (vector signed char, vector signed char);
9046 vector unsigned char vec_vaddubm (vector bool char,
9047 vector unsigned char);
9048 vector unsigned char vec_vaddubm (vector unsigned char,
9050 vector unsigned char vec_vaddubm (vector unsigned char,
9051 vector unsigned char);
9053 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
9055 vector unsigned char vec_adds (vector bool char, vector unsigned char);
9056 vector unsigned char vec_adds (vector unsigned char, vector bool char);
9057 vector unsigned char vec_adds (vector unsigned char,
9058 vector unsigned char);
9059 vector signed char vec_adds (vector bool char, vector signed char);
9060 vector signed char vec_adds (vector signed char, vector bool char);
9061 vector signed char vec_adds (vector signed char, vector signed char);
9062 vector unsigned short vec_adds (vector bool short,
9063 vector unsigned short);
9064 vector unsigned short vec_adds (vector unsigned short,
9066 vector unsigned short vec_adds (vector unsigned short,
9067 vector unsigned short);
9068 vector signed short vec_adds (vector bool short, vector signed short);
9069 vector signed short vec_adds (vector signed short, vector bool short);
9070 vector signed short vec_adds (vector signed short, vector signed short);
9071 vector unsigned int vec_adds (vector bool int, vector unsigned int);
9072 vector unsigned int vec_adds (vector unsigned int, vector bool int);
9073 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
9074 vector signed int vec_adds (vector bool int, vector signed int);
9075 vector signed int vec_adds (vector signed int, vector bool int);
9076 vector signed int vec_adds (vector signed int, vector signed int);
9078 vector signed int vec_vaddsws (vector bool int, vector signed int);
9079 vector signed int vec_vaddsws (vector signed int, vector bool int);
9080 vector signed int vec_vaddsws (vector signed int, vector signed int);
9082 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
9083 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
9084 vector unsigned int vec_vadduws (vector unsigned int,
9085 vector unsigned int);
9087 vector signed short vec_vaddshs (vector bool short,
9088 vector signed short);
9089 vector signed short vec_vaddshs (vector signed short,
9091 vector signed short vec_vaddshs (vector signed short,
9092 vector signed short);
9094 vector unsigned short vec_vadduhs (vector bool short,
9095 vector unsigned short);
9096 vector unsigned short vec_vadduhs (vector unsigned short,
9098 vector unsigned short vec_vadduhs (vector unsigned short,
9099 vector unsigned short);
9101 vector signed char vec_vaddsbs (vector bool char, vector signed char);
9102 vector signed char vec_vaddsbs (vector signed char, vector bool char);
9103 vector signed char vec_vaddsbs (vector signed char, vector signed char);
9105 vector unsigned char vec_vaddubs (vector bool char,
9106 vector unsigned char);
9107 vector unsigned char vec_vaddubs (vector unsigned char,
9109 vector unsigned char vec_vaddubs (vector unsigned char,
9110 vector unsigned char);
9112 vector float vec_and (vector float, vector float);
9113 vector float vec_and (vector float, vector bool int);
9114 vector float vec_and (vector bool int, vector float);
9115 vector bool int vec_and (vector bool int, vector bool int);
9116 vector signed int vec_and (vector bool int, vector signed int);
9117 vector signed int vec_and (vector signed int, vector bool int);
9118 vector signed int vec_and (vector signed int, vector signed int);
9119 vector unsigned int vec_and (vector bool int, vector unsigned int);
9120 vector unsigned int vec_and (vector unsigned int, vector bool int);
9121 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
9122 vector bool short vec_and (vector bool short, vector bool short);
9123 vector signed short vec_and (vector bool short, vector signed short);
9124 vector signed short vec_and (vector signed short, vector bool short);
9125 vector signed short vec_and (vector signed short, vector signed short);
9126 vector unsigned short vec_and (vector bool short,
9127 vector unsigned short);
9128 vector unsigned short vec_and (vector unsigned short,
9130 vector unsigned short vec_and (vector unsigned short,
9131 vector unsigned short);
9132 vector signed char vec_and (vector bool char, vector signed char);
9133 vector bool char vec_and (vector bool char, vector bool char);
9134 vector signed char vec_and (vector signed char, vector bool char);
9135 vector signed char vec_and (vector signed char, vector signed char);
9136 vector unsigned char vec_and (vector bool char, vector unsigned char);
9137 vector unsigned char vec_and (vector unsigned char, vector bool char);
9138 vector unsigned char vec_and (vector unsigned char,
9139 vector unsigned char);
9141 vector float vec_andc (vector float, vector float);
9142 vector float vec_andc (vector float, vector bool int);
9143 vector float vec_andc (vector bool int, vector float);
9144 vector bool int vec_andc (vector bool int, vector bool int);
9145 vector signed int vec_andc (vector bool int, vector signed int);
9146 vector signed int vec_andc (vector signed int, vector bool int);
9147 vector signed int vec_andc (vector signed int, vector signed int);
9148 vector unsigned int vec_andc (vector bool int, vector unsigned int);
9149 vector unsigned int vec_andc (vector unsigned int, vector bool int);
9150 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
9151 vector bool short vec_andc (vector bool short, vector bool short);
9152 vector signed short vec_andc (vector bool short, vector signed short);
9153 vector signed short vec_andc (vector signed short, vector bool short);
9154 vector signed short vec_andc (vector signed short, vector signed short);
9155 vector unsigned short vec_andc (vector bool short,
9156 vector unsigned short);
9157 vector unsigned short vec_andc (vector unsigned short,
9159 vector unsigned short vec_andc (vector unsigned short,
9160 vector unsigned short);
9161 vector signed char vec_andc (vector bool char, vector signed char);
9162 vector bool char vec_andc (vector bool char, vector bool char);
9163 vector signed char vec_andc (vector signed char, vector bool char);
9164 vector signed char vec_andc (vector signed char, vector signed char);
9165 vector unsigned char vec_andc (vector bool char, vector unsigned char);
9166 vector unsigned char vec_andc (vector unsigned char, vector bool char);
9167 vector unsigned char vec_andc (vector unsigned char,
9168 vector unsigned char);
9170 vector unsigned char vec_avg (vector unsigned char,
9171 vector unsigned char);
9172 vector signed char vec_avg (vector signed char, vector signed char);
9173 vector unsigned short vec_avg (vector unsigned short,
9174 vector unsigned short);
9175 vector signed short vec_avg (vector signed short, vector signed short);
9176 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
9177 vector signed int vec_avg (vector signed int, vector signed int);
9179 vector signed int vec_vavgsw (vector signed int, vector signed int);
9181 vector unsigned int vec_vavguw (vector unsigned int,
9182 vector unsigned int);
9184 vector signed short vec_vavgsh (vector signed short,
9185 vector signed short);
9187 vector unsigned short vec_vavguh (vector unsigned short,
9188 vector unsigned short);
9190 vector signed char vec_vavgsb (vector signed char, vector signed char);
9192 vector unsigned char vec_vavgub (vector unsigned char,
9193 vector unsigned char);
9195 vector float vec_ceil (vector float);
9197 vector signed int vec_cmpb (vector float, vector float);
9199 vector bool char vec_cmpeq (vector signed char, vector signed char);
9200 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
9201 vector bool short vec_cmpeq (vector signed short, vector signed short);
9202 vector bool short vec_cmpeq (vector unsigned short,
9203 vector unsigned short);
9204 vector bool int vec_cmpeq (vector signed int, vector signed int);
9205 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
9206 vector bool int vec_cmpeq (vector float, vector float);
9208 vector bool int vec_vcmpeqfp (vector float, vector float);
9210 vector bool int vec_vcmpequw (vector signed int, vector signed int);
9211 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
9213 vector bool short vec_vcmpequh (vector signed short,
9214 vector signed short);
9215 vector bool short vec_vcmpequh (vector unsigned short,
9216 vector unsigned short);
9218 vector bool char vec_vcmpequb (vector signed char, vector signed char);
9219 vector bool char vec_vcmpequb (vector unsigned char,
9220 vector unsigned char);
9222 vector bool int vec_cmpge (vector float, vector float);
9224 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
9225 vector bool char vec_cmpgt (vector signed char, vector signed char);
9226 vector bool short vec_cmpgt (vector unsigned short,
9227 vector unsigned short);
9228 vector bool short vec_cmpgt (vector signed short, vector signed short);
9229 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
9230 vector bool int vec_cmpgt (vector signed int, vector signed int);
9231 vector bool int vec_cmpgt (vector float, vector float);
9233 vector bool int vec_vcmpgtfp (vector float, vector float);
9235 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
9237 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
9239 vector bool short vec_vcmpgtsh (vector signed short,
9240 vector signed short);
9242 vector bool short vec_vcmpgtuh (vector unsigned short,
9243 vector unsigned short);
9245 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
9247 vector bool char vec_vcmpgtub (vector unsigned char,
9248 vector unsigned char);
9250 vector bool int vec_cmple (vector float, vector float);
9252 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
9253 vector bool char vec_cmplt (vector signed char, vector signed char);
9254 vector bool short vec_cmplt (vector unsigned short,
9255 vector unsigned short);
9256 vector bool short vec_cmplt (vector signed short, vector signed short);
9257 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
9258 vector bool int vec_cmplt (vector signed int, vector signed int);
9259 vector bool int vec_cmplt (vector float, vector float);
9261 vector float vec_ctf (vector unsigned int, const int);
9262 vector float vec_ctf (vector signed int, const int);
9264 vector float vec_vcfsx (vector signed int, const int);
9266 vector float vec_vcfux (vector unsigned int, const int);
9268 vector signed int vec_cts (vector float, const int);
9270 vector unsigned int vec_ctu (vector float, const int);
9272 void vec_dss (const int);
9274 void vec_dssall (void);
9276 void vec_dst (const vector unsigned char *, int, const int);
9277 void vec_dst (const vector signed char *, int, const int);
9278 void vec_dst (const vector bool char *, int, const int);
9279 void vec_dst (const vector unsigned short *, int, const int);
9280 void vec_dst (const vector signed short *, int, const int);
9281 void vec_dst (const vector bool short *, int, const int);
9282 void vec_dst (const vector pixel *, int, const int);
9283 void vec_dst (const vector unsigned int *, int, const int);
9284 void vec_dst (const vector signed int *, int, const int);
9285 void vec_dst (const vector bool int *, int, const int);
9286 void vec_dst (const vector float *, int, const int);
9287 void vec_dst (const unsigned char *, int, const int);
9288 void vec_dst (const signed char *, int, const int);
9289 void vec_dst (const unsigned short *, int, const int);
9290 void vec_dst (const short *, int, const int);
9291 void vec_dst (const unsigned int *, int, const int);
9292 void vec_dst (const int *, int, const int);
9293 void vec_dst (const unsigned long *, int, const int);
9294 void vec_dst (const long *, int, const int);
9295 void vec_dst (const float *, int, const int);
9297 void vec_dstst (const vector unsigned char *, int, const int);
9298 void vec_dstst (const vector signed char *, int, const int);
9299 void vec_dstst (const vector bool char *, int, const int);
9300 void vec_dstst (const vector unsigned short *, int, const int);
9301 void vec_dstst (const vector signed short *, int, const int);
9302 void vec_dstst (const vector bool short *, int, const int);
9303 void vec_dstst (const vector pixel *, int, const int);
9304 void vec_dstst (const vector unsigned int *, int, const int);
9305 void vec_dstst (const vector signed int *, int, const int);
9306 void vec_dstst (const vector bool int *, int, const int);
9307 void vec_dstst (const vector float *, int, const int);
9308 void vec_dstst (const unsigned char *, int, const int);
9309 void vec_dstst (const signed char *, int, const int);
9310 void vec_dstst (const unsigned short *, int, const int);
9311 void vec_dstst (const short *, int, const int);
9312 void vec_dstst (const unsigned int *, int, const int);
9313 void vec_dstst (const int *, int, const int);
9314 void vec_dstst (const unsigned long *, int, const int);
9315 void vec_dstst (const long *, int, const int);
9316 void vec_dstst (const float *, int, const int);
9318 void vec_dststt (const vector unsigned char *, int, const int);
9319 void vec_dststt (const vector signed char *, int, const int);
9320 void vec_dststt (const vector bool char *, int, const int);
9321 void vec_dststt (const vector unsigned short *, int, const int);
9322 void vec_dststt (const vector signed short *, int, const int);
9323 void vec_dststt (const vector bool short *, int, const int);
9324 void vec_dststt (const vector pixel *, int, const int);
9325 void vec_dststt (const vector unsigned int *, int, const int);
9326 void vec_dststt (const vector signed int *, int, const int);
9327 void vec_dststt (const vector bool int *, int, const int);
9328 void vec_dststt (const vector float *, int, const int);
9329 void vec_dststt (const unsigned char *, int, const int);
9330 void vec_dststt (const signed char *, int, const int);
9331 void vec_dststt (const unsigned short *, int, const int);
9332 void vec_dststt (const short *, int, const int);
9333 void vec_dststt (const unsigned int *, int, const int);
9334 void vec_dststt (const int *, int, const int);
9335 void vec_dststt (const unsigned long *, int, const int);
9336 void vec_dststt (const long *, int, const int);
9337 void vec_dststt (const float *, int, const int);
9339 void vec_dstt (const vector unsigned char *, int, const int);
9340 void vec_dstt (const vector signed char *, int, const int);
9341 void vec_dstt (const vector bool char *, int, const int);
9342 void vec_dstt (const vector unsigned short *, int, const int);
9343 void vec_dstt (const vector signed short *, int, const int);
9344 void vec_dstt (const vector bool short *, int, const int);
9345 void vec_dstt (const vector pixel *, int, const int);
9346 void vec_dstt (const vector unsigned int *, int, const int);
9347 void vec_dstt (const vector signed int *, int, const int);
9348 void vec_dstt (const vector bool int *, int, const int);
9349 void vec_dstt (const vector float *, int, const int);
9350 void vec_dstt (const unsigned char *, int, const int);
9351 void vec_dstt (const signed char *, int, const int);
9352 void vec_dstt (const unsigned short *, int, const int);
9353 void vec_dstt (const short *, int, const int);
9354 void vec_dstt (const unsigned int *, int, const int);
9355 void vec_dstt (const int *, int, const int);
9356 void vec_dstt (const unsigned long *, int, const int);
9357 void vec_dstt (const long *, int, const int);
9358 void vec_dstt (const float *, int, const int);
9360 vector float vec_expte (vector float);
9362 vector float vec_floor (vector float);
9364 vector float vec_ld (int, const vector float *);
9365 vector float vec_ld (int, const float *);
9366 vector bool int vec_ld (int, const vector bool int *);
9367 vector signed int vec_ld (int, const vector signed int *);
9368 vector signed int vec_ld (int, const int *);
9369 vector signed int vec_ld (int, const long *);
9370 vector unsigned int vec_ld (int, const vector unsigned int *);
9371 vector unsigned int vec_ld (int, const unsigned int *);
9372 vector unsigned int vec_ld (int, const unsigned long *);
9373 vector bool short vec_ld (int, const vector bool short *);
9374 vector pixel vec_ld (int, const vector pixel *);
9375 vector signed short vec_ld (int, const vector signed short *);
9376 vector signed short vec_ld (int, const short *);
9377 vector unsigned short vec_ld (int, const vector unsigned short *);
9378 vector unsigned short vec_ld (int, const unsigned short *);
9379 vector bool char vec_ld (int, const vector bool char *);
9380 vector signed char vec_ld (int, const vector signed char *);
9381 vector signed char vec_ld (int, const signed char *);
9382 vector unsigned char vec_ld (int, const vector unsigned char *);
9383 vector unsigned char vec_ld (int, const unsigned char *);
9385 vector signed char vec_lde (int, const signed char *);
9386 vector unsigned char vec_lde (int, const unsigned char *);
9387 vector signed short vec_lde (int, const short *);
9388 vector unsigned short vec_lde (int, const unsigned short *);
9389 vector float vec_lde (int, const float *);
9390 vector signed int vec_lde (int, const int *);
9391 vector unsigned int vec_lde (int, const unsigned int *);
9392 vector signed int vec_lde (int, const long *);
9393 vector unsigned int vec_lde (int, const unsigned long *);
9395 vector float vec_lvewx (int, float *);
9396 vector signed int vec_lvewx (int, int *);
9397 vector unsigned int vec_lvewx (int, unsigned int *);
9398 vector signed int vec_lvewx (int, long *);
9399 vector unsigned int vec_lvewx (int, unsigned long *);
9401 vector signed short vec_lvehx (int, short *);
9402 vector unsigned short vec_lvehx (int, unsigned short *);
9404 vector signed char vec_lvebx (int, char *);
9405 vector unsigned char vec_lvebx (int, unsigned char *);
9407 vector float vec_ldl (int, const vector float *);
9408 vector float vec_ldl (int, const float *);
9409 vector bool int vec_ldl (int, const vector bool int *);
9410 vector signed int vec_ldl (int, const vector signed int *);
9411 vector signed int vec_ldl (int, const int *);
9412 vector signed int vec_ldl (int, const long *);
9413 vector unsigned int vec_ldl (int, const vector unsigned int *);
9414 vector unsigned int vec_ldl (int, const unsigned int *);
9415 vector unsigned int vec_ldl (int, const unsigned long *);
9416 vector bool short vec_ldl (int, const vector bool short *);
9417 vector pixel vec_ldl (int, const vector pixel *);
9418 vector signed short vec_ldl (int, const vector signed short *);
9419 vector signed short vec_ldl (int, const short *);
9420 vector unsigned short vec_ldl (int, const vector unsigned short *);
9421 vector unsigned short vec_ldl (int, const unsigned short *);
9422 vector bool char vec_ldl (int, const vector bool char *);
9423 vector signed char vec_ldl (int, const vector signed char *);
9424 vector signed char vec_ldl (int, const signed char *);
9425 vector unsigned char vec_ldl (int, const vector unsigned char *);
9426 vector unsigned char vec_ldl (int, const unsigned char *);
9428 vector float vec_loge (vector float);
9430 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
9431 vector unsigned char vec_lvsl (int, const volatile signed char *);
9432 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
9433 vector unsigned char vec_lvsl (int, const volatile short *);
9434 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
9435 vector unsigned char vec_lvsl (int, const volatile int *);
9436 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
9437 vector unsigned char vec_lvsl (int, const volatile long *);
9438 vector unsigned char vec_lvsl (int, const volatile float *);
9440 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
9441 vector unsigned char vec_lvsr (int, const volatile signed char *);
9442 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
9443 vector unsigned char vec_lvsr (int, const volatile short *);
9444 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
9445 vector unsigned char vec_lvsr (int, const volatile int *);
9446 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
9447 vector unsigned char vec_lvsr (int, const volatile long *);
9448 vector unsigned char vec_lvsr (int, const volatile float *);
9450 vector float vec_madd (vector float, vector float, vector float);
9452 vector signed short vec_madds (vector signed short,
9453 vector signed short,
9454 vector signed short);
9456 vector unsigned char vec_max (vector bool char, vector unsigned char);
9457 vector unsigned char vec_max (vector unsigned char, vector bool char);
9458 vector unsigned char vec_max (vector unsigned char,
9459 vector unsigned char);
9460 vector signed char vec_max (vector bool char, vector signed char);
9461 vector signed char vec_max (vector signed char, vector bool char);
9462 vector signed char vec_max (vector signed char, vector signed char);
9463 vector unsigned short vec_max (vector bool short,
9464 vector unsigned short);
9465 vector unsigned short vec_max (vector unsigned short,
9467 vector unsigned short vec_max (vector unsigned short,
9468 vector unsigned short);
9469 vector signed short vec_max (vector bool short, vector signed short);
9470 vector signed short vec_max (vector signed short, vector bool short);
9471 vector signed short vec_max (vector signed short, vector signed short);
9472 vector unsigned int vec_max (vector bool int, vector unsigned int);
9473 vector unsigned int vec_max (vector unsigned int, vector bool int);
9474 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
9475 vector signed int vec_max (vector bool int, vector signed int);
9476 vector signed int vec_max (vector signed int, vector bool int);
9477 vector signed int vec_max (vector signed int, vector signed int);
9478 vector float vec_max (vector float, vector float);
9480 vector float vec_vmaxfp (vector float, vector float);
9482 vector signed int vec_vmaxsw (vector bool int, vector signed int);
9483 vector signed int vec_vmaxsw (vector signed int, vector bool int);
9484 vector signed int vec_vmaxsw (vector signed int, vector signed int);
9486 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
9487 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
9488 vector unsigned int vec_vmaxuw (vector unsigned int,
9489 vector unsigned int);
9491 vector signed short vec_vmaxsh (vector bool short, vector signed short);
9492 vector signed short vec_vmaxsh (vector signed short, vector bool short);
9493 vector signed short vec_vmaxsh (vector signed short,
9494 vector signed short);
9496 vector unsigned short vec_vmaxuh (vector bool short,
9497 vector unsigned short);
9498 vector unsigned short vec_vmaxuh (vector unsigned short,
9500 vector unsigned short vec_vmaxuh (vector unsigned short,
9501 vector unsigned short);
9503 vector signed char vec_vmaxsb (vector bool char, vector signed char);
9504 vector signed char vec_vmaxsb (vector signed char, vector bool char);
9505 vector signed char vec_vmaxsb (vector signed char, vector signed char);
9507 vector unsigned char vec_vmaxub (vector bool char,
9508 vector unsigned char);
9509 vector unsigned char vec_vmaxub (vector unsigned char,
9511 vector unsigned char vec_vmaxub (vector unsigned char,
9512 vector unsigned char);
9514 vector bool char vec_mergeh (vector bool char, vector bool char);
9515 vector signed char vec_mergeh (vector signed char, vector signed char);
9516 vector unsigned char vec_mergeh (vector unsigned char,
9517 vector unsigned char);
9518 vector bool short vec_mergeh (vector bool short, vector bool short);
9519 vector pixel vec_mergeh (vector pixel, vector pixel);
9520 vector signed short vec_mergeh (vector signed short,
9521 vector signed short);
9522 vector unsigned short vec_mergeh (vector unsigned short,
9523 vector unsigned short);
9524 vector float vec_mergeh (vector float, vector float);
9525 vector bool int vec_mergeh (vector bool int, vector bool int);
9526 vector signed int vec_mergeh (vector signed int, vector signed int);
9527 vector unsigned int vec_mergeh (vector unsigned int,
9528 vector unsigned int);
9530 vector float vec_vmrghw (vector float, vector float);
9531 vector bool int vec_vmrghw (vector bool int, vector bool int);
9532 vector signed int vec_vmrghw (vector signed int, vector signed int);
9533 vector unsigned int vec_vmrghw (vector unsigned int,
9534 vector unsigned int);
9536 vector bool short vec_vmrghh (vector bool short, vector bool short);
9537 vector signed short vec_vmrghh (vector signed short,
9538 vector signed short);
9539 vector unsigned short vec_vmrghh (vector unsigned short,
9540 vector unsigned short);
9541 vector pixel vec_vmrghh (vector pixel, vector pixel);
9543 vector bool char vec_vmrghb (vector bool char, vector bool char);
9544 vector signed char vec_vmrghb (vector signed char, vector signed char);
9545 vector unsigned char vec_vmrghb (vector unsigned char,
9546 vector unsigned char);
9548 vector bool char vec_mergel (vector bool char, vector bool char);
9549 vector signed char vec_mergel (vector signed char, vector signed char);
9550 vector unsigned char vec_mergel (vector unsigned char,
9551 vector unsigned char);
9552 vector bool short vec_mergel (vector bool short, vector bool short);
9553 vector pixel vec_mergel (vector pixel, vector pixel);
9554 vector signed short vec_mergel (vector signed short,
9555 vector signed short);
9556 vector unsigned short vec_mergel (vector unsigned short,
9557 vector unsigned short);
9558 vector float vec_mergel (vector float, vector float);
9559 vector bool int vec_mergel (vector bool int, vector bool int);
9560 vector signed int vec_mergel (vector signed int, vector signed int);
9561 vector unsigned int vec_mergel (vector unsigned int,
9562 vector unsigned int);
9564 vector float vec_vmrglw (vector float, vector float);
9565 vector signed int vec_vmrglw (vector signed int, vector signed int);
9566 vector unsigned int vec_vmrglw (vector unsigned int,
9567 vector unsigned int);
9568 vector bool int vec_vmrglw (vector bool int, vector bool int);
9570 vector bool short vec_vmrglh (vector bool short, vector bool short);
9571 vector signed short vec_vmrglh (vector signed short,
9572 vector signed short);
9573 vector unsigned short vec_vmrglh (vector unsigned short,
9574 vector unsigned short);
9575 vector pixel vec_vmrglh (vector pixel, vector pixel);
9577 vector bool char vec_vmrglb (vector bool char, vector bool char);
9578 vector signed char vec_vmrglb (vector signed char, vector signed char);
9579 vector unsigned char vec_vmrglb (vector unsigned char,
9580 vector unsigned char);
9582 vector unsigned short vec_mfvscr (void);
9584 vector unsigned char vec_min (vector bool char, vector unsigned char);
9585 vector unsigned char vec_min (vector unsigned char, vector bool char);
9586 vector unsigned char vec_min (vector unsigned char,
9587 vector unsigned char);
9588 vector signed char vec_min (vector bool char, vector signed char);
9589 vector signed char vec_min (vector signed char, vector bool char);
9590 vector signed char vec_min (vector signed char, vector signed char);
9591 vector unsigned short vec_min (vector bool short,
9592 vector unsigned short);
9593 vector unsigned short vec_min (vector unsigned short,
9595 vector unsigned short vec_min (vector unsigned short,
9596 vector unsigned short);
9597 vector signed short vec_min (vector bool short, vector signed short);
9598 vector signed short vec_min (vector signed short, vector bool short);
9599 vector signed short vec_min (vector signed short, vector signed short);
9600 vector unsigned int vec_min (vector bool int, vector unsigned int);
9601 vector unsigned int vec_min (vector unsigned int, vector bool int);
9602 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
9603 vector signed int vec_min (vector bool int, vector signed int);
9604 vector signed int vec_min (vector signed int, vector bool int);
9605 vector signed int vec_min (vector signed int, vector signed int);
9606 vector float vec_min (vector float, vector float);
9608 vector float vec_vminfp (vector float, vector float);
9610 vector signed int vec_vminsw (vector bool int, vector signed int);
9611 vector signed int vec_vminsw (vector signed int, vector bool int);
9612 vector signed int vec_vminsw (vector signed int, vector signed int);
9614 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
9615 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
9616 vector unsigned int vec_vminuw (vector unsigned int,
9617 vector unsigned int);
9619 vector signed short vec_vminsh (vector bool short, vector signed short);
9620 vector signed short vec_vminsh (vector signed short, vector bool short);
9621 vector signed short vec_vminsh (vector signed short,
9622 vector signed short);
9624 vector unsigned short vec_vminuh (vector bool short,
9625 vector unsigned short);
9626 vector unsigned short vec_vminuh (vector unsigned short,
9628 vector unsigned short vec_vminuh (vector unsigned short,
9629 vector unsigned short);
9631 vector signed char vec_vminsb (vector bool char, vector signed char);
9632 vector signed char vec_vminsb (vector signed char, vector bool char);
9633 vector signed char vec_vminsb (vector signed char, vector signed char);
9635 vector unsigned char vec_vminub (vector bool char,
9636 vector unsigned char);
9637 vector unsigned char vec_vminub (vector unsigned char,
9639 vector unsigned char vec_vminub (vector unsigned char,
9640 vector unsigned char);
9642 vector signed short vec_mladd (vector signed short,
9643 vector signed short,
9644 vector signed short);
9645 vector signed short vec_mladd (vector signed short,
9646 vector unsigned short,
9647 vector unsigned short);
9648 vector signed short vec_mladd (vector unsigned short,
9649 vector signed short,
9650 vector signed short);
9651 vector unsigned short vec_mladd (vector unsigned short,
9652 vector unsigned short,
9653 vector unsigned short);
9655 vector signed short vec_mradds (vector signed short,
9656 vector signed short,
9657 vector signed short);
9659 vector unsigned int vec_msum (vector unsigned char,
9660 vector unsigned char,
9661 vector unsigned int);
9662 vector signed int vec_msum (vector signed char,
9663 vector unsigned char,
9665 vector unsigned int vec_msum (vector unsigned short,
9666 vector unsigned short,
9667 vector unsigned int);
9668 vector signed int vec_msum (vector signed short,
9669 vector signed short,
9672 vector signed int vec_vmsumshm (vector signed short,
9673 vector signed short,
9676 vector unsigned int vec_vmsumuhm (vector unsigned short,
9677 vector unsigned short,
9678 vector unsigned int);
9680 vector signed int vec_vmsummbm (vector signed char,
9681 vector unsigned char,
9684 vector unsigned int vec_vmsumubm (vector unsigned char,
9685 vector unsigned char,
9686 vector unsigned int);
9688 vector unsigned int vec_msums (vector unsigned short,
9689 vector unsigned short,
9690 vector unsigned int);
9691 vector signed int vec_msums (vector signed short,
9692 vector signed short,
9695 vector signed int vec_vmsumshs (vector signed short,
9696 vector signed short,
9699 vector unsigned int vec_vmsumuhs (vector unsigned short,
9700 vector unsigned short,
9701 vector unsigned int);
9703 void vec_mtvscr (vector signed int);
9704 void vec_mtvscr (vector unsigned int);
9705 void vec_mtvscr (vector bool int);
9706 void vec_mtvscr (vector signed short);
9707 void vec_mtvscr (vector unsigned short);
9708 void vec_mtvscr (vector bool short);
9709 void vec_mtvscr (vector pixel);
9710 void vec_mtvscr (vector signed char);
9711 void vec_mtvscr (vector unsigned char);
9712 void vec_mtvscr (vector bool char);
9714 vector unsigned short vec_mule (vector unsigned char,
9715 vector unsigned char);
9716 vector signed short vec_mule (vector signed char,
9717 vector signed char);
9718 vector unsigned int vec_mule (vector unsigned short,
9719 vector unsigned short);
9720 vector signed int vec_mule (vector signed short, vector signed short);
9722 vector signed int vec_vmulesh (vector signed short,
9723 vector signed short);
9725 vector unsigned int vec_vmuleuh (vector unsigned short,
9726 vector unsigned short);
9728 vector signed short vec_vmulesb (vector signed char,
9729 vector signed char);
9731 vector unsigned short vec_vmuleub (vector unsigned char,
9732 vector unsigned char);
9734 vector unsigned short vec_mulo (vector unsigned char,
9735 vector unsigned char);
9736 vector signed short vec_mulo (vector signed char, vector signed char);
9737 vector unsigned int vec_mulo (vector unsigned short,
9738 vector unsigned short);
9739 vector signed int vec_mulo (vector signed short, vector signed short);
9741 vector signed int vec_vmulosh (vector signed short,
9742 vector signed short);
9744 vector unsigned int vec_vmulouh (vector unsigned short,
9745 vector unsigned short);
9747 vector signed short vec_vmulosb (vector signed char,
9748 vector signed char);
9750 vector unsigned short vec_vmuloub (vector unsigned char,
9751 vector unsigned char);
9753 vector float vec_nmsub (vector float, vector float, vector float);
9755 vector float vec_nor (vector float, vector float);
9756 vector signed int vec_nor (vector signed int, vector signed int);
9757 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
9758 vector bool int vec_nor (vector bool int, vector bool int);
9759 vector signed short vec_nor (vector signed short, vector signed short);
9760 vector unsigned short vec_nor (vector unsigned short,
9761 vector unsigned short);
9762 vector bool short vec_nor (vector bool short, vector bool short);
9763 vector signed char vec_nor (vector signed char, vector signed char);
9764 vector unsigned char vec_nor (vector unsigned char,
9765 vector unsigned char);
9766 vector bool char vec_nor (vector bool char, vector bool char);
9768 vector float vec_or (vector float, vector float);
9769 vector float vec_or (vector float, vector bool int);
9770 vector float vec_or (vector bool int, vector float);
9771 vector bool int vec_or (vector bool int, vector bool int);
9772 vector signed int vec_or (vector bool int, vector signed int);
9773 vector signed int vec_or (vector signed int, vector bool int);
9774 vector signed int vec_or (vector signed int, vector signed int);
9775 vector unsigned int vec_or (vector bool int, vector unsigned int);
9776 vector unsigned int vec_or (vector unsigned int, vector bool int);
9777 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
9778 vector bool short vec_or (vector bool short, vector bool short);
9779 vector signed short vec_or (vector bool short, vector signed short);
9780 vector signed short vec_or (vector signed short, vector bool short);
9781 vector signed short vec_or (vector signed short, vector signed short);
9782 vector unsigned short vec_or (vector bool short, vector unsigned short);
9783 vector unsigned short vec_or (vector unsigned short, vector bool short);
9784 vector unsigned short vec_or (vector unsigned short,
9785 vector unsigned short);
9786 vector signed char vec_or (vector bool char, vector signed char);
9787 vector bool char vec_or (vector bool char, vector bool char);
9788 vector signed char vec_or (vector signed char, vector bool char);
9789 vector signed char vec_or (vector signed char, vector signed char);
9790 vector unsigned char vec_or (vector bool char, vector unsigned char);
9791 vector unsigned char vec_or (vector unsigned char, vector bool char);
9792 vector unsigned char vec_or (vector unsigned char,
9793 vector unsigned char);
9795 vector signed char vec_pack (vector signed short, vector signed short);
9796 vector unsigned char vec_pack (vector unsigned short,
9797 vector unsigned short);
9798 vector bool char vec_pack (vector bool short, vector bool short);
9799 vector signed short vec_pack (vector signed int, vector signed int);
9800 vector unsigned short vec_pack (vector unsigned int,
9801 vector unsigned int);
9802 vector bool short vec_pack (vector bool int, vector bool int);
9804 vector bool short vec_vpkuwum (vector bool int, vector bool int);
9805 vector signed short vec_vpkuwum (vector signed int, vector signed int);
9806 vector unsigned short vec_vpkuwum (vector unsigned int,
9807 vector unsigned int);
9809 vector bool char vec_vpkuhum (vector bool short, vector bool short);
9810 vector signed char vec_vpkuhum (vector signed short,
9811 vector signed short);
9812 vector unsigned char vec_vpkuhum (vector unsigned short,
9813 vector unsigned short);
9815 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
9817 vector unsigned char vec_packs (vector unsigned short,
9818 vector unsigned short);
9819 vector signed char vec_packs (vector signed short, vector signed short);
9820 vector unsigned short vec_packs (vector unsigned int,
9821 vector unsigned int);
9822 vector signed short vec_packs (vector signed int, vector signed int);
9824 vector signed short vec_vpkswss (vector signed int, vector signed int);
9826 vector unsigned short vec_vpkuwus (vector unsigned int,
9827 vector unsigned int);
9829 vector signed char vec_vpkshss (vector signed short,
9830 vector signed short);
9832 vector unsigned char vec_vpkuhus (vector unsigned short,
9833 vector unsigned short);
9835 vector unsigned char vec_packsu (vector unsigned short,
9836 vector unsigned short);
9837 vector unsigned char vec_packsu (vector signed short,
9838 vector signed short);
9839 vector unsigned short vec_packsu (vector unsigned int,
9840 vector unsigned int);
9841 vector unsigned short vec_packsu (vector signed int, vector signed int);
9843 vector unsigned short vec_vpkswus (vector signed int,
9846 vector unsigned char vec_vpkshus (vector signed short,
9847 vector signed short);
9849 vector float vec_perm (vector float,
9851 vector unsigned char);
9852 vector signed int vec_perm (vector signed int,
9854 vector unsigned char);
9855 vector unsigned int vec_perm (vector unsigned int,
9856 vector unsigned int,
9857 vector unsigned char);
9858 vector bool int vec_perm (vector bool int,
9860 vector unsigned char);
9861 vector signed short vec_perm (vector signed short,
9862 vector signed short,
9863 vector unsigned char);
9864 vector unsigned short vec_perm (vector unsigned short,
9865 vector unsigned short,
9866 vector unsigned char);
9867 vector bool short vec_perm (vector bool short,
9869 vector unsigned char);
9870 vector pixel vec_perm (vector pixel,
9872 vector unsigned char);
9873 vector signed char vec_perm (vector signed char,
9875 vector unsigned char);
9876 vector unsigned char vec_perm (vector unsigned char,
9877 vector unsigned char,
9878 vector unsigned char);
9879 vector bool char vec_perm (vector bool char,
9881 vector unsigned char);
9883 vector float vec_re (vector float);
9885 vector signed char vec_rl (vector signed char,
9886 vector unsigned char);
9887 vector unsigned char vec_rl (vector unsigned char,
9888 vector unsigned char);
9889 vector signed short vec_rl (vector signed short, vector unsigned short);
9890 vector unsigned short vec_rl (vector unsigned short,
9891 vector unsigned short);
9892 vector signed int vec_rl (vector signed int, vector unsigned int);
9893 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
9895 vector signed int vec_vrlw (vector signed int, vector unsigned int);
9896 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
9898 vector signed short vec_vrlh (vector signed short,
9899 vector unsigned short);
9900 vector unsigned short vec_vrlh (vector unsigned short,
9901 vector unsigned short);
9903 vector signed char vec_vrlb (vector signed char, vector unsigned char);
9904 vector unsigned char vec_vrlb (vector unsigned char,
9905 vector unsigned char);
9907 vector float vec_round (vector float);
9909 vector float vec_rsqrte (vector float);
9911 vector float vec_sel (vector float, vector float, vector bool int);
9912 vector float vec_sel (vector float, vector float, vector unsigned int);
9913 vector signed int vec_sel (vector signed int,
9916 vector signed int vec_sel (vector signed int,
9918 vector unsigned int);
9919 vector unsigned int vec_sel (vector unsigned int,
9920 vector unsigned int,
9922 vector unsigned int vec_sel (vector unsigned int,
9923 vector unsigned int,
9924 vector unsigned int);
9925 vector bool int vec_sel (vector bool int,
9928 vector bool int vec_sel (vector bool int,
9930 vector unsigned int);
9931 vector signed short vec_sel (vector signed short,
9932 vector signed short,
9934 vector signed short vec_sel (vector signed short,
9935 vector signed short,
9936 vector unsigned short);
9937 vector unsigned short vec_sel (vector unsigned short,
9938 vector unsigned short,
9940 vector unsigned short vec_sel (vector unsigned short,
9941 vector unsigned short,
9942 vector unsigned short);
9943 vector bool short vec_sel (vector bool short,
9946 vector bool short vec_sel (vector bool short,
9948 vector unsigned short);
9949 vector signed char vec_sel (vector signed char,
9952 vector signed char vec_sel (vector signed char,
9954 vector unsigned char);
9955 vector unsigned char vec_sel (vector unsigned char,
9956 vector unsigned char,
9958 vector unsigned char vec_sel (vector unsigned char,
9959 vector unsigned char,
9960 vector unsigned char);
9961 vector bool char vec_sel (vector bool char,
9964 vector bool char vec_sel (vector bool char,
9966 vector unsigned char);
9968 vector signed char vec_sl (vector signed char,
9969 vector unsigned char);
9970 vector unsigned char vec_sl (vector unsigned char,
9971 vector unsigned char);
9972 vector signed short vec_sl (vector signed short, vector unsigned short);
9973 vector unsigned short vec_sl (vector unsigned short,
9974 vector unsigned short);
9975 vector signed int vec_sl (vector signed int, vector unsigned int);
9976 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
9978 vector signed int vec_vslw (vector signed int, vector unsigned int);
9979 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
9981 vector signed short vec_vslh (vector signed short,
9982 vector unsigned short);
9983 vector unsigned short vec_vslh (vector unsigned short,
9984 vector unsigned short);
9986 vector signed char vec_vslb (vector signed char, vector unsigned char);
9987 vector unsigned char vec_vslb (vector unsigned char,
9988 vector unsigned char);
9990 vector float vec_sld (vector float, vector float, const int);
9991 vector signed int vec_sld (vector signed int,
9994 vector unsigned int vec_sld (vector unsigned int,
9995 vector unsigned int,
9997 vector bool int vec_sld (vector bool int,
10000 vector signed short vec_sld (vector signed short,
10001 vector signed short,
10003 vector unsigned short vec_sld (vector unsigned short,
10004 vector unsigned short,
10006 vector bool short vec_sld (vector bool short,
10009 vector pixel vec_sld (vector pixel,
10012 vector signed char vec_sld (vector signed char,
10013 vector signed char,
10015 vector unsigned char vec_sld (vector unsigned char,
10016 vector unsigned char,
10018 vector bool char vec_sld (vector bool char,
10022 vector signed int vec_sll (vector signed int,
10023 vector unsigned int);
10024 vector signed int vec_sll (vector signed int,
10025 vector unsigned short);
10026 vector signed int vec_sll (vector signed int,
10027 vector unsigned char);
10028 vector unsigned int vec_sll (vector unsigned int,
10029 vector unsigned int);
10030 vector unsigned int vec_sll (vector unsigned int,
10031 vector unsigned short);
10032 vector unsigned int vec_sll (vector unsigned int,
10033 vector unsigned char);
10034 vector bool int vec_sll (vector bool int,
10035 vector unsigned int);
10036 vector bool int vec_sll (vector bool int,
10037 vector unsigned short);
10038 vector bool int vec_sll (vector bool int,
10039 vector unsigned char);
10040 vector signed short vec_sll (vector signed short,
10041 vector unsigned int);
10042 vector signed short vec_sll (vector signed short,
10043 vector unsigned short);
10044 vector signed short vec_sll (vector signed short,
10045 vector unsigned char);
10046 vector unsigned short vec_sll (vector unsigned short,
10047 vector unsigned int);
10048 vector unsigned short vec_sll (vector unsigned short,
10049 vector unsigned short);
10050 vector unsigned short vec_sll (vector unsigned short,
10051 vector unsigned char);
10052 vector bool short vec_sll (vector bool short, vector unsigned int);
10053 vector bool short vec_sll (vector bool short, vector unsigned short);
10054 vector bool short vec_sll (vector bool short, vector unsigned char);
10055 vector pixel vec_sll (vector pixel, vector unsigned int);
10056 vector pixel vec_sll (vector pixel, vector unsigned short);
10057 vector pixel vec_sll (vector pixel, vector unsigned char);
10058 vector signed char vec_sll (vector signed char, vector unsigned int);
10059 vector signed char vec_sll (vector signed char, vector unsigned short);
10060 vector signed char vec_sll (vector signed char, vector unsigned char);
10061 vector unsigned char vec_sll (vector unsigned char,
10062 vector unsigned int);
10063 vector unsigned char vec_sll (vector unsigned char,
10064 vector unsigned short);
10065 vector unsigned char vec_sll (vector unsigned char,
10066 vector unsigned char);
10067 vector bool char vec_sll (vector bool char, vector unsigned int);
10068 vector bool char vec_sll (vector bool char, vector unsigned short);
10069 vector bool char vec_sll (vector bool char, vector unsigned char);
10071 vector float vec_slo (vector float, vector signed char);
10072 vector float vec_slo (vector float, vector unsigned char);
10073 vector signed int vec_slo (vector signed int, vector signed char);
10074 vector signed int vec_slo (vector signed int, vector unsigned char);
10075 vector unsigned int vec_slo (vector unsigned int, vector signed char);
10076 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
10077 vector signed short vec_slo (vector signed short, vector signed char);
10078 vector signed short vec_slo (vector signed short, vector unsigned char);
10079 vector unsigned short vec_slo (vector unsigned short,
10080 vector signed char);
10081 vector unsigned short vec_slo (vector unsigned short,
10082 vector unsigned char);
10083 vector pixel vec_slo (vector pixel, vector signed char);
10084 vector pixel vec_slo (vector pixel, vector unsigned char);
10085 vector signed char vec_slo (vector signed char, vector signed char);
10086 vector signed char vec_slo (vector signed char, vector unsigned char);
10087 vector unsigned char vec_slo (vector unsigned char, vector signed char);
10088 vector unsigned char vec_slo (vector unsigned char,
10089 vector unsigned char);
10091 vector signed char vec_splat (vector signed char, const int);
10092 vector unsigned char vec_splat (vector unsigned char, const int);
10093 vector bool char vec_splat (vector bool char, const int);
10094 vector signed short vec_splat (vector signed short, const int);
10095 vector unsigned short vec_splat (vector unsigned short, const int);
10096 vector bool short vec_splat (vector bool short, const int);
10097 vector pixel vec_splat (vector pixel, const int);
10098 vector float vec_splat (vector float, const int);
10099 vector signed int vec_splat (vector signed int, const int);
10100 vector unsigned int vec_splat (vector unsigned int, const int);
10101 vector bool int vec_splat (vector bool int, const int);
10103 vector float vec_vspltw (vector float, const int);
10104 vector signed int vec_vspltw (vector signed int, const int);
10105 vector unsigned int vec_vspltw (vector unsigned int, const int);
10106 vector bool int vec_vspltw (vector bool int, const int);
10108 vector bool short vec_vsplth (vector bool short, const int);
10109 vector signed short vec_vsplth (vector signed short, const int);
10110 vector unsigned short vec_vsplth (vector unsigned short, const int);
10111 vector pixel vec_vsplth (vector pixel, const int);
10113 vector signed char vec_vspltb (vector signed char, const int);
10114 vector unsigned char vec_vspltb (vector unsigned char, const int);
10115 vector bool char vec_vspltb (vector bool char, const int);
10117 vector signed char vec_splat_s8 (const int);
10119 vector signed short vec_splat_s16 (const int);
10121 vector signed int vec_splat_s32 (const int);
10123 vector unsigned char vec_splat_u8 (const int);
10125 vector unsigned short vec_splat_u16 (const int);
10127 vector unsigned int vec_splat_u32 (const int);
10129 vector signed char vec_sr (vector signed char, vector unsigned char);
10130 vector unsigned char vec_sr (vector unsigned char,
10131 vector unsigned char);
10132 vector signed short vec_sr (vector signed short,
10133 vector unsigned short);
10134 vector unsigned short vec_sr (vector unsigned short,
10135 vector unsigned short);
10136 vector signed int vec_sr (vector signed int, vector unsigned int);
10137 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
10139 vector signed int vec_vsrw (vector signed int, vector unsigned int);
10140 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
10142 vector signed short vec_vsrh (vector signed short,
10143 vector unsigned short);
10144 vector unsigned short vec_vsrh (vector unsigned short,
10145 vector unsigned short);
10147 vector signed char vec_vsrb (vector signed char, vector unsigned char);
10148 vector unsigned char vec_vsrb (vector unsigned char,
10149 vector unsigned char);
10151 vector signed char vec_sra (vector signed char, vector unsigned char);
10152 vector unsigned char vec_sra (vector unsigned char,
10153 vector unsigned char);
10154 vector signed short vec_sra (vector signed short,
10155 vector unsigned short);
10156 vector unsigned short vec_sra (vector unsigned short,
10157 vector unsigned short);
10158 vector signed int vec_sra (vector signed int, vector unsigned int);
10159 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
10161 vector signed int vec_vsraw (vector signed int, vector unsigned int);
10162 vector unsigned int vec_vsraw (vector unsigned int,
10163 vector unsigned int);
10165 vector signed short vec_vsrah (vector signed short,
10166 vector unsigned short);
10167 vector unsigned short vec_vsrah (vector unsigned short,
10168 vector unsigned short);
10170 vector signed char vec_vsrab (vector signed char, vector unsigned char);
10171 vector unsigned char vec_vsrab (vector unsigned char,
10172 vector unsigned char);
10174 vector signed int vec_srl (vector signed int, vector unsigned int);
10175 vector signed int vec_srl (vector signed int, vector unsigned short);
10176 vector signed int vec_srl (vector signed int, vector unsigned char);
10177 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
10178 vector unsigned int vec_srl (vector unsigned int,
10179 vector unsigned short);
10180 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
10181 vector bool int vec_srl (vector bool int, vector unsigned int);
10182 vector bool int vec_srl (vector bool int, vector unsigned short);
10183 vector bool int vec_srl (vector bool int, vector unsigned char);
10184 vector signed short vec_srl (vector signed short, vector unsigned int);
10185 vector signed short vec_srl (vector signed short,
10186 vector unsigned short);
10187 vector signed short vec_srl (vector signed short, vector unsigned char);
10188 vector unsigned short vec_srl (vector unsigned short,
10189 vector unsigned int);
10190 vector unsigned short vec_srl (vector unsigned short,
10191 vector unsigned short);
10192 vector unsigned short vec_srl (vector unsigned short,
10193 vector unsigned char);
10194 vector bool short vec_srl (vector bool short, vector unsigned int);
10195 vector bool short vec_srl (vector bool short, vector unsigned short);
10196 vector bool short vec_srl (vector bool short, vector unsigned char);
10197 vector pixel vec_srl (vector pixel, vector unsigned int);
10198 vector pixel vec_srl (vector pixel, vector unsigned short);
10199 vector pixel vec_srl (vector pixel, vector unsigned char);
10200 vector signed char vec_srl (vector signed char, vector unsigned int);
10201 vector signed char vec_srl (vector signed char, vector unsigned short);
10202 vector signed char vec_srl (vector signed char, vector unsigned char);
10203 vector unsigned char vec_srl (vector unsigned char,
10204 vector unsigned int);
10205 vector unsigned char vec_srl (vector unsigned char,
10206 vector unsigned short);
10207 vector unsigned char vec_srl (vector unsigned char,
10208 vector unsigned char);
10209 vector bool char vec_srl (vector bool char, vector unsigned int);
10210 vector bool char vec_srl (vector bool char, vector unsigned short);
10211 vector bool char vec_srl (vector bool char, vector unsigned char);
10213 vector float vec_sro (vector float, vector signed char);
10214 vector float vec_sro (vector float, vector unsigned char);
10215 vector signed int vec_sro (vector signed int, vector signed char);
10216 vector signed int vec_sro (vector signed int, vector unsigned char);
10217 vector unsigned int vec_sro (vector unsigned int, vector signed char);
10218 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
10219 vector signed short vec_sro (vector signed short, vector signed char);
10220 vector signed short vec_sro (vector signed short, vector unsigned char);
10221 vector unsigned short vec_sro (vector unsigned short,
10222 vector signed char);
10223 vector unsigned short vec_sro (vector unsigned short,
10224 vector unsigned char);
10225 vector pixel vec_sro (vector pixel, vector signed char);
10226 vector pixel vec_sro (vector pixel, vector unsigned char);
10227 vector signed char vec_sro (vector signed char, vector signed char);
10228 vector signed char vec_sro (vector signed char, vector unsigned char);
10229 vector unsigned char vec_sro (vector unsigned char, vector signed char);
10230 vector unsigned char vec_sro (vector unsigned char,
10231 vector unsigned char);
10233 void vec_st (vector float, int, vector float *);
10234 void vec_st (vector float, int, float *);
10235 void vec_st (vector signed int, int, vector signed int *);
10236 void vec_st (vector signed int, int, int *);
10237 void vec_st (vector unsigned int, int, vector unsigned int *);
10238 void vec_st (vector unsigned int, int, unsigned int *);
10239 void vec_st (vector bool int, int, vector bool int *);
10240 void vec_st (vector bool int, int, unsigned int *);
10241 void vec_st (vector bool int, int, int *);
10242 void vec_st (vector signed short, int, vector signed short *);
10243 void vec_st (vector signed short, int, short *);
10244 void vec_st (vector unsigned short, int, vector unsigned short *);
10245 void vec_st (vector unsigned short, int, unsigned short *);
10246 void vec_st (vector bool short, int, vector bool short *);
10247 void vec_st (vector bool short, int, unsigned short *);
10248 void vec_st (vector pixel, int, vector pixel *);
10249 void vec_st (vector pixel, int, unsigned short *);
10250 void vec_st (vector pixel, int, short *);
10251 void vec_st (vector bool short, int, short *);
10252 void vec_st (vector signed char, int, vector signed char *);
10253 void vec_st (vector signed char, int, signed char *);
10254 void vec_st (vector unsigned char, int, vector unsigned char *);
10255 void vec_st (vector unsigned char, int, unsigned char *);
10256 void vec_st (vector bool char, int, vector bool char *);
10257 void vec_st (vector bool char, int, unsigned char *);
10258 void vec_st (vector bool char, int, signed char *);
10260 void vec_ste (vector signed char, int, signed char *);
10261 void vec_ste (vector unsigned char, int, unsigned char *);
10262 void vec_ste (vector bool char, int, signed char *);
10263 void vec_ste (vector bool char, int, unsigned char *);
10264 void vec_ste (vector signed short, int, short *);
10265 void vec_ste (vector unsigned short, int, unsigned short *);
10266 void vec_ste (vector bool short, int, short *);
10267 void vec_ste (vector bool short, int, unsigned short *);
10268 void vec_ste (vector pixel, int, short *);
10269 void vec_ste (vector pixel, int, unsigned short *);
10270 void vec_ste (vector float, int, float *);
10271 void vec_ste (vector signed int, int, int *);
10272 void vec_ste (vector unsigned int, int, unsigned int *);
10273 void vec_ste (vector bool int, int, int *);
10274 void vec_ste (vector bool int, int, unsigned int *);
10276 void vec_stvewx (vector float, int, float *);
10277 void vec_stvewx (vector signed int, int, int *);
10278 void vec_stvewx (vector unsigned int, int, unsigned int *);
10279 void vec_stvewx (vector bool int, int, int *);
10280 void vec_stvewx (vector bool int, int, unsigned int *);
10282 void vec_stvehx (vector signed short, int, short *);
10283 void vec_stvehx (vector unsigned short, int, unsigned short *);
10284 void vec_stvehx (vector bool short, int, short *);
10285 void vec_stvehx (vector bool short, int, unsigned short *);
10286 void vec_stvehx (vector pixel, int, short *);
10287 void vec_stvehx (vector pixel, int, unsigned short *);
10289 void vec_stvebx (vector signed char, int, signed char *);
10290 void vec_stvebx (vector unsigned char, int, unsigned char *);
10291 void vec_stvebx (vector bool char, int, signed char *);
10292 void vec_stvebx (vector bool char, int, unsigned char *);
10294 void vec_stl (vector float, int, vector float *);
10295 void vec_stl (vector float, int, float *);
10296 void vec_stl (vector signed int, int, vector signed int *);
10297 void vec_stl (vector signed int, int, int *);
10298 void vec_stl (vector unsigned int, int, vector unsigned int *);
10299 void vec_stl (vector unsigned int, int, unsigned int *);
10300 void vec_stl (vector bool int, int, vector bool int *);
10301 void vec_stl (vector bool int, int, unsigned int *);
10302 void vec_stl (vector bool int, int, int *);
10303 void vec_stl (vector signed short, int, vector signed short *);
10304 void vec_stl (vector signed short, int, short *);
10305 void vec_stl (vector unsigned short, int, vector unsigned short *);
10306 void vec_stl (vector unsigned short, int, unsigned short *);
10307 void vec_stl (vector bool short, int, vector bool short *);
10308 void vec_stl (vector bool short, int, unsigned short *);
10309 void vec_stl (vector bool short, int, short *);
10310 void vec_stl (vector pixel, int, vector pixel *);
10311 void vec_stl (vector pixel, int, unsigned short *);
10312 void vec_stl (vector pixel, int, short *);
10313 void vec_stl (vector signed char, int, vector signed char *);
10314 void vec_stl (vector signed char, int, signed char *);
10315 void vec_stl (vector unsigned char, int, vector unsigned char *);
10316 void vec_stl (vector unsigned char, int, unsigned char *);
10317 void vec_stl (vector bool char, int, vector bool char *);
10318 void vec_stl (vector bool char, int, unsigned char *);
10319 void vec_stl (vector bool char, int, signed char *);
10321 vector signed char vec_sub (vector bool char, vector signed char);
10322 vector signed char vec_sub (vector signed char, vector bool char);
10323 vector signed char vec_sub (vector signed char, vector signed char);
10324 vector unsigned char vec_sub (vector bool char, vector unsigned char);
10325 vector unsigned char vec_sub (vector unsigned char, vector bool char);
10326 vector unsigned char vec_sub (vector unsigned char,
10327 vector unsigned char);
10328 vector signed short vec_sub (vector bool short, vector signed short);
10329 vector signed short vec_sub (vector signed short, vector bool short);
10330 vector signed short vec_sub (vector signed short, vector signed short);
10331 vector unsigned short vec_sub (vector bool short,
10332 vector unsigned short);
10333 vector unsigned short vec_sub (vector unsigned short,
10334 vector bool short);
10335 vector unsigned short vec_sub (vector unsigned short,
10336 vector unsigned short);
10337 vector signed int vec_sub (vector bool int, vector signed int);
10338 vector signed int vec_sub (vector signed int, vector bool int);
10339 vector signed int vec_sub (vector signed int, vector signed int);
10340 vector unsigned int vec_sub (vector bool int, vector unsigned int);
10341 vector unsigned int vec_sub (vector unsigned int, vector bool int);
10342 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
10343 vector float vec_sub (vector float, vector float);
10345 vector float vec_vsubfp (vector float, vector float);
10347 vector signed int vec_vsubuwm (vector bool int, vector signed int);
10348 vector signed int vec_vsubuwm (vector signed int, vector bool int);
10349 vector signed int vec_vsubuwm (vector signed int, vector signed int);
10350 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
10351 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
10352 vector unsigned int vec_vsubuwm (vector unsigned int,
10353 vector unsigned int);
10355 vector signed short vec_vsubuhm (vector bool short,
10356 vector signed short);
10357 vector signed short vec_vsubuhm (vector signed short,
10358 vector bool short);
10359 vector signed short vec_vsubuhm (vector signed short,
10360 vector signed short);
10361 vector unsigned short vec_vsubuhm (vector bool short,
10362 vector unsigned short);
10363 vector unsigned short vec_vsubuhm (vector unsigned short,
10364 vector bool short);
10365 vector unsigned short vec_vsubuhm (vector unsigned short,
10366 vector unsigned short);
10368 vector signed char vec_vsububm (vector bool char, vector signed char);
10369 vector signed char vec_vsububm (vector signed char, vector bool char);
10370 vector signed char vec_vsububm (vector signed char, vector signed char);
10371 vector unsigned char vec_vsububm (vector bool char,
10372 vector unsigned char);
10373 vector unsigned char vec_vsububm (vector unsigned char,
10375 vector unsigned char vec_vsububm (vector unsigned char,
10376 vector unsigned char);
10378 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
10380 vector unsigned char vec_subs (vector bool char, vector unsigned char);
10381 vector unsigned char vec_subs (vector unsigned char, vector bool char);
10382 vector unsigned char vec_subs (vector unsigned char,
10383 vector unsigned char);
10384 vector signed char vec_subs (vector bool char, vector signed char);
10385 vector signed char vec_subs (vector signed char, vector bool char);
10386 vector signed char vec_subs (vector signed char, vector signed char);
10387 vector unsigned short vec_subs (vector bool short,
10388 vector unsigned short);
10389 vector unsigned short vec_subs (vector unsigned short,
10390 vector bool short);
10391 vector unsigned short vec_subs (vector unsigned short,
10392 vector unsigned short);
10393 vector signed short vec_subs (vector bool short, vector signed short);
10394 vector signed short vec_subs (vector signed short, vector bool short);
10395 vector signed short vec_subs (vector signed short, vector signed short);
10396 vector unsigned int vec_subs (vector bool int, vector unsigned int);
10397 vector unsigned int vec_subs (vector unsigned int, vector bool int);
10398 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
10399 vector signed int vec_subs (vector bool int, vector signed int);
10400 vector signed int vec_subs (vector signed int, vector bool int);
10401 vector signed int vec_subs (vector signed int, vector signed int);
10403 vector signed int vec_vsubsws (vector bool int, vector signed int);
10404 vector signed int vec_vsubsws (vector signed int, vector bool int);
10405 vector signed int vec_vsubsws (vector signed int, vector signed int);
10407 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
10408 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
10409 vector unsigned int vec_vsubuws (vector unsigned int,
10410 vector unsigned int);
10412 vector signed short vec_vsubshs (vector bool short,
10413 vector signed short);
10414 vector signed short vec_vsubshs (vector signed short,
10415 vector bool short);
10416 vector signed short vec_vsubshs (vector signed short,
10417 vector signed short);
10419 vector unsigned short vec_vsubuhs (vector bool short,
10420 vector unsigned short);
10421 vector unsigned short vec_vsubuhs (vector unsigned short,
10422 vector bool short);
10423 vector unsigned short vec_vsubuhs (vector unsigned short,
10424 vector unsigned short);
10426 vector signed char vec_vsubsbs (vector bool char, vector signed char);
10427 vector signed char vec_vsubsbs (vector signed char, vector bool char);
10428 vector signed char vec_vsubsbs (vector signed char, vector signed char);
10430 vector unsigned char vec_vsububs (vector bool char,
10431 vector unsigned char);
10432 vector unsigned char vec_vsububs (vector unsigned char,
10434 vector unsigned char vec_vsububs (vector unsigned char,
10435 vector unsigned char);
10437 vector unsigned int vec_sum4s (vector unsigned char,
10438 vector unsigned int);
10439 vector signed int vec_sum4s (vector signed char, vector signed int);
10440 vector signed int vec_sum4s (vector signed short, vector signed int);
10442 vector signed int vec_vsum4shs (vector signed short, vector signed int);
10444 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
10446 vector unsigned int vec_vsum4ubs (vector unsigned char,
10447 vector unsigned int);
10449 vector signed int vec_sum2s (vector signed int, vector signed int);
10451 vector signed int vec_sums (vector signed int, vector signed int);
10453 vector float vec_trunc (vector float);
10455 vector signed short vec_unpackh (vector signed char);
10456 vector bool short vec_unpackh (vector bool char);
10457 vector signed int vec_unpackh (vector signed short);
10458 vector bool int vec_unpackh (vector bool short);
10459 vector unsigned int vec_unpackh (vector pixel);
10461 vector bool int vec_vupkhsh (vector bool short);
10462 vector signed int vec_vupkhsh (vector signed short);
10464 vector unsigned int vec_vupkhpx (vector pixel);
10466 vector bool short vec_vupkhsb (vector bool char);
10467 vector signed short vec_vupkhsb (vector signed char);
10469 vector signed short vec_unpackl (vector signed char);
10470 vector bool short vec_unpackl (vector bool char);
10471 vector unsigned int vec_unpackl (vector pixel);
10472 vector signed int vec_unpackl (vector signed short);
10473 vector bool int vec_unpackl (vector bool short);
10475 vector unsigned int vec_vupklpx (vector pixel);
10477 vector bool int vec_vupklsh (vector bool short);
10478 vector signed int vec_vupklsh (vector signed short);
10480 vector bool short vec_vupklsb (vector bool char);
10481 vector signed short vec_vupklsb (vector signed char);
10483 vector float vec_xor (vector float, vector float);
10484 vector float vec_xor (vector float, vector bool int);
10485 vector float vec_xor (vector bool int, vector float);
10486 vector bool int vec_xor (vector bool int, vector bool int);
10487 vector signed int vec_xor (vector bool int, vector signed int);
10488 vector signed int vec_xor (vector signed int, vector bool int);
10489 vector signed int vec_xor (vector signed int, vector signed int);
10490 vector unsigned int vec_xor (vector bool int, vector unsigned int);
10491 vector unsigned int vec_xor (vector unsigned int, vector bool int);
10492 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
10493 vector bool short vec_xor (vector bool short, vector bool short);
10494 vector signed short vec_xor (vector bool short, vector signed short);
10495 vector signed short vec_xor (vector signed short, vector bool short);
10496 vector signed short vec_xor (vector signed short, vector signed short);
10497 vector unsigned short vec_xor (vector bool short,
10498 vector unsigned short);
10499 vector unsigned short vec_xor (vector unsigned short,
10500 vector bool short);
10501 vector unsigned short vec_xor (vector unsigned short,
10502 vector unsigned short);
10503 vector signed char vec_xor (vector bool char, vector signed char);
10504 vector bool char vec_xor (vector bool char, vector bool char);
10505 vector signed char vec_xor (vector signed char, vector bool char);
10506 vector signed char vec_xor (vector signed char, vector signed char);
10507 vector unsigned char vec_xor (vector bool char, vector unsigned char);
10508 vector unsigned char vec_xor (vector unsigned char, vector bool char);
10509 vector unsigned char vec_xor (vector unsigned char,
10510 vector unsigned char);
10512 int vec_all_eq (vector signed char, vector bool char);
10513 int vec_all_eq (vector signed char, vector signed char);
10514 int vec_all_eq (vector unsigned char, vector bool char);
10515 int vec_all_eq (vector unsigned char, vector unsigned char);
10516 int vec_all_eq (vector bool char, vector bool char);
10517 int vec_all_eq (vector bool char, vector unsigned char);
10518 int vec_all_eq (vector bool char, vector signed char);
10519 int vec_all_eq (vector signed short, vector bool short);
10520 int vec_all_eq (vector signed short, vector signed short);
10521 int vec_all_eq (vector unsigned short, vector bool short);
10522 int vec_all_eq (vector unsigned short, vector unsigned short);
10523 int vec_all_eq (vector bool short, vector bool short);
10524 int vec_all_eq (vector bool short, vector unsigned short);
10525 int vec_all_eq (vector bool short, vector signed short);
10526 int vec_all_eq (vector pixel, vector pixel);
10527 int vec_all_eq (vector signed int, vector bool int);
10528 int vec_all_eq (vector signed int, vector signed int);
10529 int vec_all_eq (vector unsigned int, vector bool int);
10530 int vec_all_eq (vector unsigned int, vector unsigned int);
10531 int vec_all_eq (vector bool int, vector bool int);
10532 int vec_all_eq (vector bool int, vector unsigned int);
10533 int vec_all_eq (vector bool int, vector signed int);
10534 int vec_all_eq (vector float, vector float);
10536 int vec_all_ge (vector bool char, vector unsigned char);
10537 int vec_all_ge (vector unsigned char, vector bool char);
10538 int vec_all_ge (vector unsigned char, vector unsigned char);
10539 int vec_all_ge (vector bool char, vector signed char);
10540 int vec_all_ge (vector signed char, vector bool char);
10541 int vec_all_ge (vector signed char, vector signed char);
10542 int vec_all_ge (vector bool short, vector unsigned short);
10543 int vec_all_ge (vector unsigned short, vector bool short);
10544 int vec_all_ge (vector unsigned short, vector unsigned short);
10545 int vec_all_ge (vector signed short, vector signed short);
10546 int vec_all_ge (vector bool short, vector signed short);
10547 int vec_all_ge (vector signed short, vector bool short);
10548 int vec_all_ge (vector bool int, vector unsigned int);
10549 int vec_all_ge (vector unsigned int, vector bool int);
10550 int vec_all_ge (vector unsigned int, vector unsigned int);
10551 int vec_all_ge (vector bool int, vector signed int);
10552 int vec_all_ge (vector signed int, vector bool int);
10553 int vec_all_ge (vector signed int, vector signed int);
10554 int vec_all_ge (vector float, vector float);
10556 int vec_all_gt (vector bool char, vector unsigned char);
10557 int vec_all_gt (vector unsigned char, vector bool char);
10558 int vec_all_gt (vector unsigned char, vector unsigned char);
10559 int vec_all_gt (vector bool char, vector signed char);
10560 int vec_all_gt (vector signed char, vector bool char);
10561 int vec_all_gt (vector signed char, vector signed char);
10562 int vec_all_gt (vector bool short, vector unsigned short);
10563 int vec_all_gt (vector unsigned short, vector bool short);
10564 int vec_all_gt (vector unsigned short, vector unsigned short);
10565 int vec_all_gt (vector bool short, vector signed short);
10566 int vec_all_gt (vector signed short, vector bool short);
10567 int vec_all_gt (vector signed short, vector signed short);
10568 int vec_all_gt (vector bool int, vector unsigned int);
10569 int vec_all_gt (vector unsigned int, vector bool int);
10570 int vec_all_gt (vector unsigned int, vector unsigned int);
10571 int vec_all_gt (vector bool int, vector signed int);
10572 int vec_all_gt (vector signed int, vector bool int);
10573 int vec_all_gt (vector signed int, vector signed int);
10574 int vec_all_gt (vector float, vector float);
10576 int vec_all_in (vector float, vector float);
10578 int vec_all_le (vector bool char, vector unsigned char);
10579 int vec_all_le (vector unsigned char, vector bool char);
10580 int vec_all_le (vector unsigned char, vector unsigned char);
10581 int vec_all_le (vector bool char, vector signed char);
10582 int vec_all_le (vector signed char, vector bool char);
10583 int vec_all_le (vector signed char, vector signed char);
10584 int vec_all_le (vector bool short, vector unsigned short);
10585 int vec_all_le (vector unsigned short, vector bool short);
10586 int vec_all_le (vector unsigned short, vector unsigned short);
10587 int vec_all_le (vector bool short, vector signed short);
10588 int vec_all_le (vector signed short, vector bool short);
10589 int vec_all_le (vector signed short, vector signed short);
10590 int vec_all_le (vector bool int, vector unsigned int);
10591 int vec_all_le (vector unsigned int, vector bool int);
10592 int vec_all_le (vector unsigned int, vector unsigned int);
10593 int vec_all_le (vector bool int, vector signed int);
10594 int vec_all_le (vector signed int, vector bool int);
10595 int vec_all_le (vector signed int, vector signed int);
10596 int vec_all_le (vector float, vector float);
10598 int vec_all_lt (vector bool char, vector unsigned char);
10599 int vec_all_lt (vector unsigned char, vector bool char);
10600 int vec_all_lt (vector unsigned char, vector unsigned char);
10601 int vec_all_lt (vector bool char, vector signed char);
10602 int vec_all_lt (vector signed char, vector bool char);
10603 int vec_all_lt (vector signed char, vector signed char);
10604 int vec_all_lt (vector bool short, vector unsigned short);
10605 int vec_all_lt (vector unsigned short, vector bool short);
10606 int vec_all_lt (vector unsigned short, vector unsigned short);
10607 int vec_all_lt (vector bool short, vector signed short);
10608 int vec_all_lt (vector signed short, vector bool short);
10609 int vec_all_lt (vector signed short, vector signed short);
10610 int vec_all_lt (vector bool int, vector unsigned int);
10611 int vec_all_lt (vector unsigned int, vector bool int);
10612 int vec_all_lt (vector unsigned int, vector unsigned int);
10613 int vec_all_lt (vector bool int, vector signed int);
10614 int vec_all_lt (vector signed int, vector bool int);
10615 int vec_all_lt (vector signed int, vector signed int);
10616 int vec_all_lt (vector float, vector float);
10618 int vec_all_nan (vector float);
10620 int vec_all_ne (vector signed char, vector bool char);
10621 int vec_all_ne (vector signed char, vector signed char);
10622 int vec_all_ne (vector unsigned char, vector bool char);
10623 int vec_all_ne (vector unsigned char, vector unsigned char);
10624 int vec_all_ne (vector bool char, vector bool char);
10625 int vec_all_ne (vector bool char, vector unsigned char);
10626 int vec_all_ne (vector bool char, vector signed char);
10627 int vec_all_ne (vector signed short, vector bool short);
10628 int vec_all_ne (vector signed short, vector signed short);
10629 int vec_all_ne (vector unsigned short, vector bool short);
10630 int vec_all_ne (vector unsigned short, vector unsigned short);
10631 int vec_all_ne (vector bool short, vector bool short);
10632 int vec_all_ne (vector bool short, vector unsigned short);
10633 int vec_all_ne (vector bool short, vector signed short);
10634 int vec_all_ne (vector pixel, vector pixel);
10635 int vec_all_ne (vector signed int, vector bool int);
10636 int vec_all_ne (vector signed int, vector signed int);
10637 int vec_all_ne (vector unsigned int, vector bool int);
10638 int vec_all_ne (vector unsigned int, vector unsigned int);
10639 int vec_all_ne (vector bool int, vector bool int);
10640 int vec_all_ne (vector bool int, vector unsigned int);
10641 int vec_all_ne (vector bool int, vector signed int);
10642 int vec_all_ne (vector float, vector float);
10644 int vec_all_nge (vector float, vector float);
10646 int vec_all_ngt (vector float, vector float);
10648 int vec_all_nle (vector float, vector float);
10650 int vec_all_nlt (vector float, vector float);
10652 int vec_all_numeric (vector float);
10654 int vec_any_eq (vector signed char, vector bool char);
10655 int vec_any_eq (vector signed char, vector signed char);
10656 int vec_any_eq (vector unsigned char, vector bool char);
10657 int vec_any_eq (vector unsigned char, vector unsigned char);
10658 int vec_any_eq (vector bool char, vector bool char);
10659 int vec_any_eq (vector bool char, vector unsigned char);
10660 int vec_any_eq (vector bool char, vector signed char);
10661 int vec_any_eq (vector signed short, vector bool short);
10662 int vec_any_eq (vector signed short, vector signed short);
10663 int vec_any_eq (vector unsigned short, vector bool short);
10664 int vec_any_eq (vector unsigned short, vector unsigned short);
10665 int vec_any_eq (vector bool short, vector bool short);
10666 int vec_any_eq (vector bool short, vector unsigned short);
10667 int vec_any_eq (vector bool short, vector signed short);
10668 int vec_any_eq (vector pixel, vector pixel);
10669 int vec_any_eq (vector signed int, vector bool int);
10670 int vec_any_eq (vector signed int, vector signed int);
10671 int vec_any_eq (vector unsigned int, vector bool int);
10672 int vec_any_eq (vector unsigned int, vector unsigned int);
10673 int vec_any_eq (vector bool int, vector bool int);
10674 int vec_any_eq (vector bool int, vector unsigned int);
10675 int vec_any_eq (vector bool int, vector signed int);
10676 int vec_any_eq (vector float, vector float);
10678 int vec_any_ge (vector signed char, vector bool char);
10679 int vec_any_ge (vector unsigned char, vector bool char);
10680 int vec_any_ge (vector unsigned char, vector unsigned char);
10681 int vec_any_ge (vector signed char, vector signed char);
10682 int vec_any_ge (vector bool char, vector unsigned char);
10683 int vec_any_ge (vector bool char, vector signed char);
10684 int vec_any_ge (vector unsigned short, vector bool short);
10685 int vec_any_ge (vector unsigned short, vector unsigned short);
10686 int vec_any_ge (vector signed short, vector signed short);
10687 int vec_any_ge (vector signed short, vector bool short);
10688 int vec_any_ge (vector bool short, vector unsigned short);
10689 int vec_any_ge (vector bool short, vector signed short);
10690 int vec_any_ge (vector signed int, vector bool int);
10691 int vec_any_ge (vector unsigned int, vector bool int);
10692 int vec_any_ge (vector unsigned int, vector unsigned int);
10693 int vec_any_ge (vector signed int, vector signed int);
10694 int vec_any_ge (vector bool int, vector unsigned int);
10695 int vec_any_ge (vector bool int, vector signed int);
10696 int vec_any_ge (vector float, vector float);
10698 int vec_any_gt (vector bool char, vector unsigned char);
10699 int vec_any_gt (vector unsigned char, vector bool char);
10700 int vec_any_gt (vector unsigned char, vector unsigned char);
10701 int vec_any_gt (vector bool char, vector signed char);
10702 int vec_any_gt (vector signed char, vector bool char);
10703 int vec_any_gt (vector signed char, vector signed char);
10704 int vec_any_gt (vector bool short, vector unsigned short);
10705 int vec_any_gt (vector unsigned short, vector bool short);
10706 int vec_any_gt (vector unsigned short, vector unsigned short);
10707 int vec_any_gt (vector bool short, vector signed short);
10708 int vec_any_gt (vector signed short, vector bool short);
10709 int vec_any_gt (vector signed short, vector signed short);
10710 int vec_any_gt (vector bool int, vector unsigned int);
10711 int vec_any_gt (vector unsigned int, vector bool int);
10712 int vec_any_gt (vector unsigned int, vector unsigned int);
10713 int vec_any_gt (vector bool int, vector signed int);
10714 int vec_any_gt (vector signed int, vector bool int);
10715 int vec_any_gt (vector signed int, vector signed int);
10716 int vec_any_gt (vector float, vector float);
10718 int vec_any_le (vector bool char, vector unsigned char);
10719 int vec_any_le (vector unsigned char, vector bool char);
10720 int vec_any_le (vector unsigned char, vector unsigned char);
10721 int vec_any_le (vector bool char, vector signed char);
10722 int vec_any_le (vector signed char, vector bool char);
10723 int vec_any_le (vector signed char, vector signed char);
10724 int vec_any_le (vector bool short, vector unsigned short);
10725 int vec_any_le (vector unsigned short, vector bool short);
10726 int vec_any_le (vector unsigned short, vector unsigned short);
10727 int vec_any_le (vector bool short, vector signed short);
10728 int vec_any_le (vector signed short, vector bool short);
10729 int vec_any_le (vector signed short, vector signed short);
10730 int vec_any_le (vector bool int, vector unsigned int);
10731 int vec_any_le (vector unsigned int, vector bool int);
10732 int vec_any_le (vector unsigned int, vector unsigned int);
10733 int vec_any_le (vector bool int, vector signed int);
10734 int vec_any_le (vector signed int, vector bool int);
10735 int vec_any_le (vector signed int, vector signed int);
10736 int vec_any_le (vector float, vector float);
10738 int vec_any_lt (vector bool char, vector unsigned char);
10739 int vec_any_lt (vector unsigned char, vector bool char);
10740 int vec_any_lt (vector unsigned char, vector unsigned char);
10741 int vec_any_lt (vector bool char, vector signed char);
10742 int vec_any_lt (vector signed char, vector bool char);
10743 int vec_any_lt (vector signed char, vector signed char);
10744 int vec_any_lt (vector bool short, vector unsigned short);
10745 int vec_any_lt (vector unsigned short, vector bool short);
10746 int vec_any_lt (vector unsigned short, vector unsigned short);
10747 int vec_any_lt (vector bool short, vector signed short);
10748 int vec_any_lt (vector signed short, vector bool short);
10749 int vec_any_lt (vector signed short, vector signed short);
10750 int vec_any_lt (vector bool int, vector unsigned int);
10751 int vec_any_lt (vector unsigned int, vector bool int);
10752 int vec_any_lt (vector unsigned int, vector unsigned int);
10753 int vec_any_lt (vector bool int, vector signed int);
10754 int vec_any_lt (vector signed int, vector bool int);
10755 int vec_any_lt (vector signed int, vector signed int);
10756 int vec_any_lt (vector float, vector float);
10758 int vec_any_nan (vector float);
10760 int vec_any_ne (vector signed char, vector bool char);
10761 int vec_any_ne (vector signed char, vector signed char);
10762 int vec_any_ne (vector unsigned char, vector bool char);
10763 int vec_any_ne (vector unsigned char, vector unsigned char);
10764 int vec_any_ne (vector bool char, vector bool char);
10765 int vec_any_ne (vector bool char, vector unsigned char);
10766 int vec_any_ne (vector bool char, vector signed char);
10767 int vec_any_ne (vector signed short, vector bool short);
10768 int vec_any_ne (vector signed short, vector signed short);
10769 int vec_any_ne (vector unsigned short, vector bool short);
10770 int vec_any_ne (vector unsigned short, vector unsigned short);
10771 int vec_any_ne (vector bool short, vector bool short);
10772 int vec_any_ne (vector bool short, vector unsigned short);
10773 int vec_any_ne (vector bool short, vector signed short);
10774 int vec_any_ne (vector pixel, vector pixel);
10775 int vec_any_ne (vector signed int, vector bool int);
10776 int vec_any_ne (vector signed int, vector signed int);
10777 int vec_any_ne (vector unsigned int, vector bool int);
10778 int vec_any_ne (vector unsigned int, vector unsigned int);
10779 int vec_any_ne (vector bool int, vector bool int);
10780 int vec_any_ne (vector bool int, vector unsigned int);
10781 int vec_any_ne (vector bool int, vector signed int);
10782 int vec_any_ne (vector float, vector float);
10784 int vec_any_nge (vector float, vector float);
10786 int vec_any_ngt (vector float, vector float);
10788 int vec_any_nle (vector float, vector float);
10790 int vec_any_nlt (vector float, vector float);
10792 int vec_any_numeric (vector float);
10794 int vec_any_out (vector float, vector float);
10797 @node SPARC VIS Built-in Functions
10798 @subsection SPARC VIS Built-in Functions
10800 GCC supports SIMD operations on the SPARC using both the generic vector
10801 extensions (@pxref{Vector Extensions}) as well as built-in functions for
10802 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
10803 switch, the VIS extension is exposed as the following built-in functions:
10806 typedef int v2si __attribute__ ((vector_size (8)));
10807 typedef short v4hi __attribute__ ((vector_size (8)));
10808 typedef short v2hi __attribute__ ((vector_size (4)));
10809 typedef char v8qi __attribute__ ((vector_size (8)));
10810 typedef char v4qi __attribute__ ((vector_size (4)));
10812 void * __builtin_vis_alignaddr (void *, long);
10813 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
10814 v2si __builtin_vis_faligndatav2si (v2si, v2si);
10815 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
10816 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
10818 v4hi __builtin_vis_fexpand (v4qi);
10820 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
10821 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
10822 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
10823 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
10824 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
10825 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
10826 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
10828 v4qi __builtin_vis_fpack16 (v4hi);
10829 v8qi __builtin_vis_fpack32 (v2si, v2si);
10830 v2hi __builtin_vis_fpackfix (v2si);
10831 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
10833 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
10836 @node SPU Built-in Functions
10837 @subsection SPU Built-in Functions
10839 GCC provides extensions for the SPU processor as described in the
10840 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
10841 found at @uref{http://cell.scei.co.jp/} or
10842 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
10843 implementation differs in several ways.
10848 The optional extension of specifying vector constants in parentheses is
10852 A vector initializer requires no cast if the vector constant is of the
10853 same type as the variable it is initializing.
10856 If @code{signed} or @code{unsigned} is omitted, the signedness of the
10857 vector type is the default signedness of the base type. The default
10858 varies depending on the operating system, so a portable program should
10859 always specify the signedness.
10862 By default, the keyword @code{__vector} is added. The macro
10863 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
10867 GCC allows using a @code{typedef} name as the type specifier for a
10871 For C, overloaded functions are implemented with macros so the following
10875 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
10878 Since @code{spu_add} is a macro, the vector constant in the example
10879 is treated as four separate arguments. Wrap the entire argument in
10880 parentheses for this to work.
10883 The extended version of @code{__builtin_expect} is not supported.
10887 @emph{Note:} Only the interface described in the aforementioned
10888 specification is supported. Internally, GCC uses built-in functions to
10889 implement the required functionality, but these are not supported and
10890 are subject to change without notice.
10892 @node Target Format Checks
10893 @section Format Checks Specific to Particular Target Machines
10895 For some target machines, GCC supports additional options to the
10897 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
10900 * Solaris Format Checks::
10903 @node Solaris Format Checks
10904 @subsection Solaris Format Checks
10906 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
10907 check. @code{cmn_err} accepts a subset of the standard @code{printf}
10908 conversions, and the two-argument @code{%b} conversion for displaying
10909 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
10912 @section Pragmas Accepted by GCC
10916 GCC supports several types of pragmas, primarily in order to compile
10917 code originally written for other compilers. Note that in general
10918 we do not recommend the use of pragmas; @xref{Function Attributes},
10919 for further explanation.
10924 * RS/6000 and PowerPC Pragmas::
10926 * Solaris Pragmas::
10927 * Symbol-Renaming Pragmas::
10928 * Structure-Packing Pragmas::
10930 * Diagnostic Pragmas::
10931 * Visibility Pragmas::
10932 * Push/Pop Macro Pragmas::
10936 @subsection ARM Pragmas
10938 The ARM target defines pragmas for controlling the default addition of
10939 @code{long_call} and @code{short_call} attributes to functions.
10940 @xref{Function Attributes}, for information about the effects of these
10945 @cindex pragma, long_calls
10946 Set all subsequent functions to have the @code{long_call} attribute.
10948 @item no_long_calls
10949 @cindex pragma, no_long_calls
10950 Set all subsequent functions to have the @code{short_call} attribute.
10952 @item long_calls_off
10953 @cindex pragma, long_calls_off
10954 Do not affect the @code{long_call} or @code{short_call} attributes of
10955 subsequent functions.
10959 @subsection M32C Pragmas
10962 @item memregs @var{number}
10963 @cindex pragma, memregs
10964 Overrides the command line option @code{-memregs=} for the current
10965 file. Use with care! This pragma must be before any function in the
10966 file, and mixing different memregs values in different objects may
10967 make them incompatible. This pragma is useful when a
10968 performance-critical function uses a memreg for temporary values,
10969 as it may allow you to reduce the number of memregs used.
10973 @node RS/6000 and PowerPC Pragmas
10974 @subsection RS/6000 and PowerPC Pragmas
10976 The RS/6000 and PowerPC targets define one pragma for controlling
10977 whether or not the @code{longcall} attribute is added to function
10978 declarations by default. This pragma overrides the @option{-mlongcall}
10979 option, but not the @code{longcall} and @code{shortcall} attributes.
10980 @xref{RS/6000 and PowerPC Options}, for more information about when long
10981 calls are and are not necessary.
10985 @cindex pragma, longcall
10986 Apply the @code{longcall} attribute to all subsequent function
10990 Do not apply the @code{longcall} attribute to subsequent function
10994 @c Describe h8300 pragmas here.
10995 @c Describe sh pragmas here.
10996 @c Describe v850 pragmas here.
10998 @node Darwin Pragmas
10999 @subsection Darwin Pragmas
11001 The following pragmas are available for all architectures running the
11002 Darwin operating system. These are useful for compatibility with other
11006 @item mark @var{tokens}@dots{}
11007 @cindex pragma, mark
11008 This pragma is accepted, but has no effect.
11010 @item options align=@var{alignment}
11011 @cindex pragma, options align
11012 This pragma sets the alignment of fields in structures. The values of
11013 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
11014 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
11015 properly; to restore the previous setting, use @code{reset} for the
11018 @item segment @var{tokens}@dots{}
11019 @cindex pragma, segment
11020 This pragma is accepted, but has no effect.
11022 @item unused (@var{var} [, @var{var}]@dots{})
11023 @cindex pragma, unused
11024 This pragma declares variables to be possibly unused. GCC will not
11025 produce warnings for the listed variables. The effect is similar to
11026 that of the @code{unused} attribute, except that this pragma may appear
11027 anywhere within the variables' scopes.
11030 @node Solaris Pragmas
11031 @subsection Solaris Pragmas
11033 The Solaris target supports @code{#pragma redefine_extname}
11034 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
11035 @code{#pragma} directives for compatibility with the system compiler.
11038 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
11039 @cindex pragma, align
11041 Increase the minimum alignment of each @var{variable} to @var{alignment}.
11042 This is the same as GCC's @code{aligned} attribute @pxref{Variable
11043 Attributes}). Macro expansion occurs on the arguments to this pragma
11044 when compiling C and Objective-C@. It does not currently occur when
11045 compiling C++, but this is a bug which may be fixed in a future
11048 @item fini (@var{function} [, @var{function}]...)
11049 @cindex pragma, fini
11051 This pragma causes each listed @var{function} to be called after
11052 main, or during shared module unloading, by adding a call to the
11053 @code{.fini} section.
11055 @item init (@var{function} [, @var{function}]...)
11056 @cindex pragma, init
11058 This pragma causes each listed @var{function} to be called during
11059 initialization (before @code{main}) or during shared module loading, by
11060 adding a call to the @code{.init} section.
11064 @node Symbol-Renaming Pragmas
11065 @subsection Symbol-Renaming Pragmas
11067 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
11068 supports two @code{#pragma} directives which change the name used in
11069 assembly for a given declaration. These pragmas are only available on
11070 platforms whose system headers need them. To get this effect on all
11071 platforms supported by GCC, use the asm labels extension (@pxref{Asm
11075 @item redefine_extname @var{oldname} @var{newname}
11076 @cindex pragma, redefine_extname
11078 This pragma gives the C function @var{oldname} the assembly symbol
11079 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
11080 will be defined if this pragma is available (currently only on
11083 @item extern_prefix @var{string}
11084 @cindex pragma, extern_prefix
11086 This pragma causes all subsequent external function and variable
11087 declarations to have @var{string} prepended to their assembly symbols.
11088 This effect may be terminated with another @code{extern_prefix} pragma
11089 whose argument is an empty string. The preprocessor macro
11090 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
11091 available (currently only on Tru64 UNIX)@.
11094 These pragmas and the asm labels extension interact in a complicated
11095 manner. Here are some corner cases you may want to be aware of.
11098 @item Both pragmas silently apply only to declarations with external
11099 linkage. Asm labels do not have this restriction.
11101 @item In C++, both pragmas silently apply only to declarations with
11102 ``C'' linkage. Again, asm labels do not have this restriction.
11104 @item If any of the three ways of changing the assembly name of a
11105 declaration is applied to a declaration whose assembly name has
11106 already been determined (either by a previous use of one of these
11107 features, or because the compiler needed the assembly name in order to
11108 generate code), and the new name is different, a warning issues and
11109 the name does not change.
11111 @item The @var{oldname} used by @code{#pragma redefine_extname} is
11112 always the C-language name.
11114 @item If @code{#pragma extern_prefix} is in effect, and a declaration
11115 occurs with an asm label attached, the prefix is silently ignored for
11118 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
11119 apply to the same declaration, whichever triggered first wins, and a
11120 warning issues if they contradict each other. (We would like to have
11121 @code{#pragma redefine_extname} always win, for consistency with asm
11122 labels, but if @code{#pragma extern_prefix} triggers first we have no
11123 way of knowing that that happened.)
11126 @node Structure-Packing Pragmas
11127 @subsection Structure-Packing Pragmas
11129 For compatibility with Microsoft Windows compilers, GCC supports a
11130 set of @code{#pragma} directives which change the maximum alignment of
11131 members of structures (other than zero-width bitfields), unions, and
11132 classes subsequently defined. The @var{n} value below always is required
11133 to be a small power of two and specifies the new alignment in bytes.
11136 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
11137 @item @code{#pragma pack()} sets the alignment to the one that was in
11138 effect when compilation started (see also command line option
11139 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
11140 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
11141 setting on an internal stack and then optionally sets the new alignment.
11142 @item @code{#pragma pack(pop)} restores the alignment setting to the one
11143 saved at the top of the internal stack (and removes that stack entry).
11144 Note that @code{#pragma pack([@var{n}])} does not influence this internal
11145 stack; thus it is possible to have @code{#pragma pack(push)} followed by
11146 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
11147 @code{#pragma pack(pop)}.
11150 Some targets, e.g.@: i386 and powerpc, support the @code{ms_struct}
11151 @code{#pragma} which lays out a structure as the documented
11152 @code{__attribute__ ((ms_struct))}.
11154 @item @code{#pragma ms_struct on} turns on the layout for structures
11156 @item @code{#pragma ms_struct off} turns off the layout for structures
11158 @item @code{#pragma ms_struct reset} goes back to the default layout.
11162 @subsection Weak Pragmas
11164 For compatibility with SVR4, GCC supports a set of @code{#pragma}
11165 directives for declaring symbols to be weak, and defining weak
11169 @item #pragma weak @var{symbol}
11170 @cindex pragma, weak
11171 This pragma declares @var{symbol} to be weak, as if the declaration
11172 had the attribute of the same name. The pragma may appear before
11173 or after the declaration of @var{symbol}, but must appear before
11174 either its first use or its definition. It is not an error for
11175 @var{symbol} to never be defined at all.
11177 @item #pragma weak @var{symbol1} = @var{symbol2}
11178 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
11179 It is an error if @var{symbol2} is not defined in the current
11183 @node Diagnostic Pragmas
11184 @subsection Diagnostic Pragmas
11186 GCC allows the user to selectively enable or disable certain types of
11187 diagnostics, and change the kind of the diagnostic. For example, a
11188 project's policy might require that all sources compile with
11189 @option{-Werror} but certain files might have exceptions allowing
11190 specific types of warnings. Or, a project might selectively enable
11191 diagnostics and treat them as errors depending on which preprocessor
11192 macros are defined.
11195 @item #pragma GCC diagnostic @var{kind} @var{option}
11196 @cindex pragma, diagnostic
11198 Modifies the disposition of a diagnostic. Note that not all
11199 diagnostics are modifiable; at the moment only warnings (normally
11200 controlled by @samp{-W@dots{}}) can be controlled, and not all of them.
11201 Use @option{-fdiagnostics-show-option} to determine which diagnostics
11202 are controllable and which option controls them.
11204 @var{kind} is @samp{error} to treat this diagnostic as an error,
11205 @samp{warning} to treat it like a warning (even if @option{-Werror} is
11206 in effect), or @samp{ignored} if the diagnostic is to be ignored.
11207 @var{option} is a double quoted string which matches the command line
11211 #pragma GCC diagnostic warning "-Wformat"
11212 #pragma GCC diagnostic error "-Wformat"
11213 #pragma GCC diagnostic ignored "-Wformat"
11216 Note that these pragmas override any command line options. Also,
11217 while it is syntactically valid to put these pragmas anywhere in your
11218 sources, the only supported location for them is before any data or
11219 functions are defined. Doing otherwise may result in unpredictable
11220 results depending on how the optimizer manages your sources. If the
11221 same option is listed multiple times, the last one specified is the
11222 one that is in effect. This pragma is not intended to be a general
11223 purpose replacement for command line options, but for implementing
11224 strict control over project policies.
11228 @node Visibility Pragmas
11229 @subsection Visibility Pragmas
11232 @item #pragma GCC visibility push(@var{visibility})
11233 @itemx #pragma GCC visibility pop
11234 @cindex pragma, visibility
11236 This pragma allows the user to set the visibility for multiple
11237 declarations without having to give each a visibility attribute
11238 @xref{Function Attributes}, for more information about visibility and
11239 the attribute syntax.
11241 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
11242 declarations. Class members and template specializations are not
11243 affected; if you want to override the visibility for a particular
11244 member or instantiation, you must use an attribute.
11249 @node Push/Pop Macro Pragmas
11250 @subsection Push/Pop Macro Pragmas
11252 For compatibility with Microsoft Windows compilers, GCC supports
11253 @samp{#pragma push_macro(@var{"macro_name"})}
11254 and @samp{#pragma pop_macro(@var{"macro_name"})}.
11257 @item #pragma push_macro(@var{"macro_name"})
11258 @cindex pragma, push_macro
11259 This pragma saves the value of the macro named as @var{macro_name} to
11260 the top of the stack for this macro.
11262 @item #pragma pop_macro(@var{"macro_name"})
11263 @cindex pragma, pop_macro
11264 This pragma sets the value of the macro named as @var{macro_name} to
11265 the value on top of the stack for this macro. If the stack for
11266 @var{macro_name} is empty, the value of the macro remains unchanged.
11273 #pragma push_macro("X")
11276 #pragma pop_macro("X")
11280 In this example, the definition of X as 1 is saved by @code{#pragma
11281 push_macro} and restored by @code{#pragma pop_macro}.
11283 @node Unnamed Fields
11284 @section Unnamed struct/union fields within structs/unions
11288 For compatibility with other compilers, GCC allows you to define
11289 a structure or union that contains, as fields, structures and unions
11290 without names. For example:
11303 In this example, the user would be able to access members of the unnamed
11304 union with code like @samp{foo.b}. Note that only unnamed structs and
11305 unions are allowed, you may not have, for example, an unnamed
11308 You must never create such structures that cause ambiguous field definitions.
11309 For example, this structure:
11320 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
11321 Such constructs are not supported and must be avoided. In the future,
11322 such constructs may be detected and treated as compilation errors.
11324 @opindex fms-extensions
11325 Unless @option{-fms-extensions} is used, the unnamed field must be a
11326 structure or union definition without a tag (for example, @samp{struct
11327 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
11328 also be a definition with a tag such as @samp{struct foo @{ int a;
11329 @};}, a reference to a previously defined structure or union such as
11330 @samp{struct foo;}, or a reference to a @code{typedef} name for a
11331 previously defined structure or union type.
11334 @section Thread-Local Storage
11335 @cindex Thread-Local Storage
11336 @cindex @acronym{TLS}
11339 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
11340 are allocated such that there is one instance of the variable per extant
11341 thread. The run-time model GCC uses to implement this originates
11342 in the IA-64 processor-specific ABI, but has since been migrated
11343 to other processors as well. It requires significant support from
11344 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
11345 system libraries (@file{libc.so} and @file{libpthread.so}), so it
11346 is not available everywhere.
11348 At the user level, the extension is visible with a new storage
11349 class keyword: @code{__thread}. For example:
11353 extern __thread struct state s;
11354 static __thread char *p;
11357 The @code{__thread} specifier may be used alone, with the @code{extern}
11358 or @code{static} specifiers, but with no other storage class specifier.
11359 When used with @code{extern} or @code{static}, @code{__thread} must appear
11360 immediately after the other storage class specifier.
11362 The @code{__thread} specifier may be applied to any global, file-scoped
11363 static, function-scoped static, or static data member of a class. It may
11364 not be applied to block-scoped automatic or non-static data member.
11366 When the address-of operator is applied to a thread-local variable, it is
11367 evaluated at run-time and returns the address of the current thread's
11368 instance of that variable. An address so obtained may be used by any
11369 thread. When a thread terminates, any pointers to thread-local variables
11370 in that thread become invalid.
11372 No static initialization may refer to the address of a thread-local variable.
11374 In C++, if an initializer is present for a thread-local variable, it must
11375 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
11378 See @uref{http://people.redhat.com/drepper/tls.pdf,
11379 ELF Handling For Thread-Local Storage} for a detailed explanation of
11380 the four thread-local storage addressing models, and how the run-time
11381 is expected to function.
11384 * C99 Thread-Local Edits::
11385 * C++98 Thread-Local Edits::
11388 @node C99 Thread-Local Edits
11389 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
11391 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
11392 that document the exact semantics of the language extension.
11396 @cite{5.1.2 Execution environments}
11398 Add new text after paragraph 1
11401 Within either execution environment, a @dfn{thread} is a flow of
11402 control within a program. It is implementation defined whether
11403 or not there may be more than one thread associated with a program.
11404 It is implementation defined how threads beyond the first are
11405 created, the name and type of the function called at thread
11406 startup, and how threads may be terminated. However, objects
11407 with thread storage duration shall be initialized before thread
11412 @cite{6.2.4 Storage durations of objects}
11414 Add new text before paragraph 3
11417 An object whose identifier is declared with the storage-class
11418 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
11419 Its lifetime is the entire execution of the thread, and its
11420 stored value is initialized only once, prior to thread startup.
11424 @cite{6.4.1 Keywords}
11426 Add @code{__thread}.
11429 @cite{6.7.1 Storage-class specifiers}
11431 Add @code{__thread} to the list of storage class specifiers in
11434 Change paragraph 2 to
11437 With the exception of @code{__thread}, at most one storage-class
11438 specifier may be given [@dots{}]. The @code{__thread} specifier may
11439 be used alone, or immediately following @code{extern} or
11443 Add new text after paragraph 6
11446 The declaration of an identifier for a variable that has
11447 block scope that specifies @code{__thread} shall also
11448 specify either @code{extern} or @code{static}.
11450 The @code{__thread} specifier shall be used only with
11455 @node C++98 Thread-Local Edits
11456 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
11458 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
11459 that document the exact semantics of the language extension.
11463 @b{[intro.execution]}
11465 New text after paragraph 4
11468 A @dfn{thread} is a flow of control within the abstract machine.
11469 It is implementation defined whether or not there may be more than
11473 New text after paragraph 7
11476 It is unspecified whether additional action must be taken to
11477 ensure when and whether side effects are visible to other threads.
11483 Add @code{__thread}.
11486 @b{[basic.start.main]}
11488 Add after paragraph 5
11491 The thread that begins execution at the @code{main} function is called
11492 the @dfn{main thread}. It is implementation defined how functions
11493 beginning threads other than the main thread are designated or typed.
11494 A function so designated, as well as the @code{main} function, is called
11495 a @dfn{thread startup function}. It is implementation defined what
11496 happens if a thread startup function returns. It is implementation
11497 defined what happens to other threads when any thread calls @code{exit}.
11501 @b{[basic.start.init]}
11503 Add after paragraph 4
11506 The storage for an object of thread storage duration shall be
11507 statically initialized before the first statement of the thread startup
11508 function. An object of thread storage duration shall not require
11509 dynamic initialization.
11513 @b{[basic.start.term]}
11515 Add after paragraph 3
11518 The type of an object with thread storage duration shall not have a
11519 non-trivial destructor, nor shall it be an array type whose elements
11520 (directly or indirectly) have non-trivial destructors.
11526 Add ``thread storage duration'' to the list in paragraph 1.
11531 Thread, static, and automatic storage durations are associated with
11532 objects introduced by declarations [@dots{}].
11535 Add @code{__thread} to the list of specifiers in paragraph 3.
11538 @b{[basic.stc.thread]}
11540 New section before @b{[basic.stc.static]}
11543 The keyword @code{__thread} applied to a non-local object gives the
11544 object thread storage duration.
11546 A local variable or class data member declared both @code{static}
11547 and @code{__thread} gives the variable or member thread storage
11552 @b{[basic.stc.static]}
11557 All objects which have neither thread storage duration, dynamic
11558 storage duration nor are local [@dots{}].
11564 Add @code{__thread} to the list in paragraph 1.
11569 With the exception of @code{__thread}, at most one
11570 @var{storage-class-specifier} shall appear in a given
11571 @var{decl-specifier-seq}. The @code{__thread} specifier may
11572 be used alone, or immediately following the @code{extern} or
11573 @code{static} specifiers. [@dots{}]
11576 Add after paragraph 5
11579 The @code{__thread} specifier can be applied only to the names of objects
11580 and to anonymous unions.
11586 Add after paragraph 6
11589 Non-@code{static} members shall not be @code{__thread}.
11593 @node Binary constants
11594 @section Binary constants using the @samp{0b} prefix
11595 @cindex Binary constants using the @samp{0b} prefix
11597 Integer constants can be written as binary constants, consisting of a
11598 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
11599 @samp{0B}. This is particularly useful in environments that operate a
11600 lot on the bit-level (like microcontrollers).
11602 The following statements are identical:
11611 The type of these constants follows the same rules as for octal or
11612 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
11615 @node C++ Extensions
11616 @chapter Extensions to the C++ Language
11617 @cindex extensions, C++ language
11618 @cindex C++ language extensions
11620 The GNU compiler provides these extensions to the C++ language (and you
11621 can also use most of the C language extensions in your C++ programs). If you
11622 want to write code that checks whether these features are available, you can
11623 test for the GNU compiler the same way as for C programs: check for a
11624 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
11625 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
11626 Predefined Macros,cpp,The GNU C Preprocessor}).
11629 * Volatiles:: What constitutes an access to a volatile object.
11630 * Restricted Pointers:: C99 restricted pointers and references.
11631 * Vague Linkage:: Where G++ puts inlines, vtables and such.
11632 * C++ Interface:: You can use a single C++ header file for both
11633 declarations and definitions.
11634 * Template Instantiation:: Methods for ensuring that exactly one copy of
11635 each needed template instantiation is emitted.
11636 * Bound member functions:: You can extract a function pointer to the
11637 method denoted by a @samp{->*} or @samp{.*} expression.
11638 * C++ Attributes:: Variable, function, and type attributes for C++ only.
11639 * Namespace Association:: Strong using-directives for namespace association.
11640 * Type Traits:: Compiler support for type traits
11641 * Java Exceptions:: Tweaking exception handling to work with Java.
11642 * Deprecated Features:: Things will disappear from g++.
11643 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
11647 @section When is a Volatile Object Accessed?
11648 @cindex accessing volatiles
11649 @cindex volatile read
11650 @cindex volatile write
11651 @cindex volatile access
11653 Both the C and C++ standard have the concept of volatile objects. These
11654 are normally accessed by pointers and used for accessing hardware. The
11655 standards encourage compilers to refrain from optimizations concerning
11656 accesses to volatile objects. The C standard leaves it implementation
11657 defined as to what constitutes a volatile access. The C++ standard omits
11658 to specify this, except to say that C++ should behave in a similar manner
11659 to C with respect to volatiles, where possible. The minimum either
11660 standard specifies is that at a sequence point all previous accesses to
11661 volatile objects have stabilized and no subsequent accesses have
11662 occurred. Thus an implementation is free to reorder and combine
11663 volatile accesses which occur between sequence points, but cannot do so
11664 for accesses across a sequence point. The use of volatiles does not
11665 allow you to violate the restriction on updating objects multiple times
11666 within a sequence point.
11668 @xref{Qualifiers implementation, , Volatile qualifier and the C compiler}.
11670 The behavior differs slightly between C and C++ in the non-obvious cases:
11673 volatile int *src = @var{somevalue};
11677 With C, such expressions are rvalues, and GCC interprets this either as a
11678 read of the volatile object being pointed to or only as request to evaluate
11679 the side-effects. The C++ standard specifies that such expressions do not
11680 undergo lvalue to rvalue conversion, and that the type of the dereferenced
11681 object may be incomplete. The C++ standard does not specify explicitly
11682 that it is this lvalue to rvalue conversion which may be responsible for
11683 causing an access. However, there is reason to believe that it is,
11684 because otherwise certain simple expressions become undefined. However,
11685 because it would surprise most programmers, G++ treats dereferencing a
11686 pointer to volatile object of complete type when the value is unused as
11687 GCC would do for an equivalent type in C@. When the object has incomplete
11688 type, G++ issues a warning; if you wish to force an error, you must
11689 force a conversion to rvalue with, for instance, a static cast.
11691 When using a reference to volatile, G++ does not treat equivalent
11692 expressions as accesses to volatiles, but instead issues a warning that
11693 no volatile is accessed. The rationale for this is that otherwise it
11694 becomes difficult to determine where volatile access occur, and not
11695 possible to ignore the return value from functions returning volatile
11696 references. Again, if you wish to force a read, cast the reference to
11699 @node Restricted Pointers
11700 @section Restricting Pointer Aliasing
11701 @cindex restricted pointers
11702 @cindex restricted references
11703 @cindex restricted this pointer
11705 As with the C front end, G++ understands the C99 feature of restricted pointers,
11706 specified with the @code{__restrict__}, or @code{__restrict} type
11707 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
11708 language flag, @code{restrict} is not a keyword in C++.
11710 In addition to allowing restricted pointers, you can specify restricted
11711 references, which indicate that the reference is not aliased in the local
11715 void fn (int *__restrict__ rptr, int &__restrict__ rref)
11722 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
11723 @var{rref} refers to a (different) unaliased integer.
11725 You may also specify whether a member function's @var{this} pointer is
11726 unaliased by using @code{__restrict__} as a member function qualifier.
11729 void T::fn () __restrict__
11736 Within the body of @code{T::fn}, @var{this} will have the effective
11737 definition @code{T *__restrict__ const this}. Notice that the
11738 interpretation of a @code{__restrict__} member function qualifier is
11739 different to that of @code{const} or @code{volatile} qualifier, in that it
11740 is applied to the pointer rather than the object. This is consistent with
11741 other compilers which implement restricted pointers.
11743 As with all outermost parameter qualifiers, @code{__restrict__} is
11744 ignored in function definition matching. This means you only need to
11745 specify @code{__restrict__} in a function definition, rather than
11746 in a function prototype as well.
11748 @node Vague Linkage
11749 @section Vague Linkage
11750 @cindex vague linkage
11752 There are several constructs in C++ which require space in the object
11753 file but are not clearly tied to a single translation unit. We say that
11754 these constructs have ``vague linkage''. Typically such constructs are
11755 emitted wherever they are needed, though sometimes we can be more
11759 @item Inline Functions
11760 Inline functions are typically defined in a header file which can be
11761 included in many different compilations. Hopefully they can usually be
11762 inlined, but sometimes an out-of-line copy is necessary, if the address
11763 of the function is taken or if inlining fails. In general, we emit an
11764 out-of-line copy in all translation units where one is needed. As an
11765 exception, we only emit inline virtual functions with the vtable, since
11766 it will always require a copy.
11768 Local static variables and string constants used in an inline function
11769 are also considered to have vague linkage, since they must be shared
11770 between all inlined and out-of-line instances of the function.
11774 C++ virtual functions are implemented in most compilers using a lookup
11775 table, known as a vtable. The vtable contains pointers to the virtual
11776 functions provided by a class, and each object of the class contains a
11777 pointer to its vtable (or vtables, in some multiple-inheritance
11778 situations). If the class declares any non-inline, non-pure virtual
11779 functions, the first one is chosen as the ``key method'' for the class,
11780 and the vtable is only emitted in the translation unit where the key
11783 @emph{Note:} If the chosen key method is later defined as inline, the
11784 vtable will still be emitted in every translation unit which defines it.
11785 Make sure that any inline virtuals are declared inline in the class
11786 body, even if they are not defined there.
11788 @item type_info objects
11791 C++ requires information about types to be written out in order to
11792 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
11793 For polymorphic classes (classes with virtual functions), the type_info
11794 object is written out along with the vtable so that @samp{dynamic_cast}
11795 can determine the dynamic type of a class object at runtime. For all
11796 other types, we write out the type_info object when it is used: when
11797 applying @samp{typeid} to an expression, throwing an object, or
11798 referring to a type in a catch clause or exception specification.
11800 @item Template Instantiations
11801 Most everything in this section also applies to template instantiations,
11802 but there are other options as well.
11803 @xref{Template Instantiation,,Where's the Template?}.
11807 When used with GNU ld version 2.8 or later on an ELF system such as
11808 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
11809 these constructs will be discarded at link time. This is known as
11812 On targets that don't support COMDAT, but do support weak symbols, GCC
11813 will use them. This way one copy will override all the others, but
11814 the unused copies will still take up space in the executable.
11816 For targets which do not support either COMDAT or weak symbols,
11817 most entities with vague linkage will be emitted as local symbols to
11818 avoid duplicate definition errors from the linker. This will not happen
11819 for local statics in inlines, however, as having multiple copies will
11820 almost certainly break things.
11822 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
11823 another way to control placement of these constructs.
11825 @node C++ Interface
11826 @section #pragma interface and implementation
11828 @cindex interface and implementation headers, C++
11829 @cindex C++ interface and implementation headers
11830 @cindex pragmas, interface and implementation
11832 @code{#pragma interface} and @code{#pragma implementation} provide the
11833 user with a way of explicitly directing the compiler to emit entities
11834 with vague linkage (and debugging information) in a particular
11837 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
11838 most cases, because of COMDAT support and the ``key method'' heuristic
11839 mentioned in @ref{Vague Linkage}. Using them can actually cause your
11840 program to grow due to unnecessary out-of-line copies of inline
11841 functions. Currently (3.4) the only benefit of these
11842 @code{#pragma}s is reduced duplication of debugging information, and
11843 that should be addressed soon on DWARF 2 targets with the use of
11847 @item #pragma interface
11848 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
11849 @kindex #pragma interface
11850 Use this directive in @emph{header files} that define object classes, to save
11851 space in most of the object files that use those classes. Normally,
11852 local copies of certain information (backup copies of inline member
11853 functions, debugging information, and the internal tables that implement
11854 virtual functions) must be kept in each object file that includes class
11855 definitions. You can use this pragma to avoid such duplication. When a
11856 header file containing @samp{#pragma interface} is included in a
11857 compilation, this auxiliary information will not be generated (unless
11858 the main input source file itself uses @samp{#pragma implementation}).
11859 Instead, the object files will contain references to be resolved at link
11862 The second form of this directive is useful for the case where you have
11863 multiple headers with the same name in different directories. If you
11864 use this form, you must specify the same string to @samp{#pragma
11867 @item #pragma implementation
11868 @itemx #pragma implementation "@var{objects}.h"
11869 @kindex #pragma implementation
11870 Use this pragma in a @emph{main input file}, when you want full output from
11871 included header files to be generated (and made globally visible). The
11872 included header file, in turn, should use @samp{#pragma interface}.
11873 Backup copies of inline member functions, debugging information, and the
11874 internal tables used to implement virtual functions are all generated in
11875 implementation files.
11877 @cindex implied @code{#pragma implementation}
11878 @cindex @code{#pragma implementation}, implied
11879 @cindex naming convention, implementation headers
11880 If you use @samp{#pragma implementation} with no argument, it applies to
11881 an include file with the same basename@footnote{A file's @dfn{basename}
11882 was the name stripped of all leading path information and of trailing
11883 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
11884 file. For example, in @file{allclass.cc}, giving just
11885 @samp{#pragma implementation}
11886 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
11888 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
11889 an implementation file whenever you would include it from
11890 @file{allclass.cc} even if you never specified @samp{#pragma
11891 implementation}. This was deemed to be more trouble than it was worth,
11892 however, and disabled.
11894 Use the string argument if you want a single implementation file to
11895 include code from multiple header files. (You must also use
11896 @samp{#include} to include the header file; @samp{#pragma
11897 implementation} only specifies how to use the file---it doesn't actually
11900 There is no way to split up the contents of a single header file into
11901 multiple implementation files.
11904 @cindex inlining and C++ pragmas
11905 @cindex C++ pragmas, effect on inlining
11906 @cindex pragmas in C++, effect on inlining
11907 @samp{#pragma implementation} and @samp{#pragma interface} also have an
11908 effect on function inlining.
11910 If you define a class in a header file marked with @samp{#pragma
11911 interface}, the effect on an inline function defined in that class is
11912 similar to an explicit @code{extern} declaration---the compiler emits
11913 no code at all to define an independent version of the function. Its
11914 definition is used only for inlining with its callers.
11916 @opindex fno-implement-inlines
11917 Conversely, when you include the same header file in a main source file
11918 that declares it as @samp{#pragma implementation}, the compiler emits
11919 code for the function itself; this defines a version of the function
11920 that can be found via pointers (or by callers compiled without
11921 inlining). If all calls to the function can be inlined, you can avoid
11922 emitting the function by compiling with @option{-fno-implement-inlines}.
11923 If any calls were not inlined, you will get linker errors.
11925 @node Template Instantiation
11926 @section Where's the Template?
11927 @cindex template instantiation
11929 C++ templates are the first language feature to require more
11930 intelligence from the environment than one usually finds on a UNIX
11931 system. Somehow the compiler and linker have to make sure that each
11932 template instance occurs exactly once in the executable if it is needed,
11933 and not at all otherwise. There are two basic approaches to this
11934 problem, which are referred to as the Borland model and the Cfront model.
11937 @item Borland model
11938 Borland C++ solved the template instantiation problem by adding the code
11939 equivalent of common blocks to their linker; the compiler emits template
11940 instances in each translation unit that uses them, and the linker
11941 collapses them together. The advantage of this model is that the linker
11942 only has to consider the object files themselves; there is no external
11943 complexity to worry about. This disadvantage is that compilation time
11944 is increased because the template code is being compiled repeatedly.
11945 Code written for this model tends to include definitions of all
11946 templates in the header file, since they must be seen to be
11950 The AT&T C++ translator, Cfront, solved the template instantiation
11951 problem by creating the notion of a template repository, an
11952 automatically maintained place where template instances are stored. A
11953 more modern version of the repository works as follows: As individual
11954 object files are built, the compiler places any template definitions and
11955 instantiations encountered in the repository. At link time, the link
11956 wrapper adds in the objects in the repository and compiles any needed
11957 instances that were not previously emitted. The advantages of this
11958 model are more optimal compilation speed and the ability to use the
11959 system linker; to implement the Borland model a compiler vendor also
11960 needs to replace the linker. The disadvantages are vastly increased
11961 complexity, and thus potential for error; for some code this can be
11962 just as transparent, but in practice it can been very difficult to build
11963 multiple programs in one directory and one program in multiple
11964 directories. Code written for this model tends to separate definitions
11965 of non-inline member templates into a separate file, which should be
11966 compiled separately.
11969 When used with GNU ld version 2.8 or later on an ELF system such as
11970 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
11971 Borland model. On other systems, G++ implements neither automatic
11974 A future version of G++ will support a hybrid model whereby the compiler
11975 will emit any instantiations for which the template definition is
11976 included in the compile, and store template definitions and
11977 instantiation context information into the object file for the rest.
11978 The link wrapper will extract that information as necessary and invoke
11979 the compiler to produce the remaining instantiations. The linker will
11980 then combine duplicate instantiations.
11982 In the mean time, you have the following options for dealing with
11983 template instantiations:
11988 Compile your template-using code with @option{-frepo}. The compiler will
11989 generate files with the extension @samp{.rpo} listing all of the
11990 template instantiations used in the corresponding object files which
11991 could be instantiated there; the link wrapper, @samp{collect2}, will
11992 then update the @samp{.rpo} files to tell the compiler where to place
11993 those instantiations and rebuild any affected object files. The
11994 link-time overhead is negligible after the first pass, as the compiler
11995 will continue to place the instantiations in the same files.
11997 This is your best option for application code written for the Borland
11998 model, as it will just work. Code written for the Cfront model will
11999 need to be modified so that the template definitions are available at
12000 one or more points of instantiation; usually this is as simple as adding
12001 @code{#include <tmethods.cc>} to the end of each template header.
12003 For library code, if you want the library to provide all of the template
12004 instantiations it needs, just try to link all of its object files
12005 together; the link will fail, but cause the instantiations to be
12006 generated as a side effect. Be warned, however, that this may cause
12007 conflicts if multiple libraries try to provide the same instantiations.
12008 For greater control, use explicit instantiation as described in the next
12012 @opindex fno-implicit-templates
12013 Compile your code with @option{-fno-implicit-templates} to disable the
12014 implicit generation of template instances, and explicitly instantiate
12015 all the ones you use. This approach requires more knowledge of exactly
12016 which instances you need than do the others, but it's less
12017 mysterious and allows greater control. You can scatter the explicit
12018 instantiations throughout your program, perhaps putting them in the
12019 translation units where the instances are used or the translation units
12020 that define the templates themselves; you can put all of the explicit
12021 instantiations you need into one big file; or you can create small files
12028 template class Foo<int>;
12029 template ostream& operator <<
12030 (ostream&, const Foo<int>&);
12033 for each of the instances you need, and create a template instantiation
12034 library from those.
12036 If you are using Cfront-model code, you can probably get away with not
12037 using @option{-fno-implicit-templates} when compiling files that don't
12038 @samp{#include} the member template definitions.
12040 If you use one big file to do the instantiations, you may want to
12041 compile it without @option{-fno-implicit-templates} so you get all of the
12042 instances required by your explicit instantiations (but not by any
12043 other files) without having to specify them as well.
12045 G++ has extended the template instantiation syntax given in the ISO
12046 standard to allow forward declaration of explicit instantiations
12047 (with @code{extern}), instantiation of the compiler support data for a
12048 template class (i.e.@: the vtable) without instantiating any of its
12049 members (with @code{inline}), and instantiation of only the static data
12050 members of a template class, without the support data or member
12051 functions (with (@code{static}):
12054 extern template int max (int, int);
12055 inline template class Foo<int>;
12056 static template class Foo<int>;
12060 Do nothing. Pretend G++ does implement automatic instantiation
12061 management. Code written for the Borland model will work fine, but
12062 each translation unit will contain instances of each of the templates it
12063 uses. In a large program, this can lead to an unacceptable amount of code
12067 @node Bound member functions
12068 @section Extracting the function pointer from a bound pointer to member function
12070 @cindex pointer to member function
12071 @cindex bound pointer to member function
12073 In C++, pointer to member functions (PMFs) are implemented using a wide
12074 pointer of sorts to handle all the possible call mechanisms; the PMF
12075 needs to store information about how to adjust the @samp{this} pointer,
12076 and if the function pointed to is virtual, where to find the vtable, and
12077 where in the vtable to look for the member function. If you are using
12078 PMFs in an inner loop, you should really reconsider that decision. If
12079 that is not an option, you can extract the pointer to the function that
12080 would be called for a given object/PMF pair and call it directly inside
12081 the inner loop, to save a bit of time.
12083 Note that you will still be paying the penalty for the call through a
12084 function pointer; on most modern architectures, such a call defeats the
12085 branch prediction features of the CPU@. This is also true of normal
12086 virtual function calls.
12088 The syntax for this extension is
12092 extern int (A::*fp)();
12093 typedef int (*fptr)(A *);
12095 fptr p = (fptr)(a.*fp);
12098 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
12099 no object is needed to obtain the address of the function. They can be
12100 converted to function pointers directly:
12103 fptr p1 = (fptr)(&A::foo);
12106 @opindex Wno-pmf-conversions
12107 You must specify @option{-Wno-pmf-conversions} to use this extension.
12109 @node C++ Attributes
12110 @section C++-Specific Variable, Function, and Type Attributes
12112 Some attributes only make sense for C++ programs.
12115 @item init_priority (@var{priority})
12116 @cindex init_priority attribute
12119 In Standard C++, objects defined at namespace scope are guaranteed to be
12120 initialized in an order in strict accordance with that of their definitions
12121 @emph{in a given translation unit}. No guarantee is made for initializations
12122 across translation units. However, GNU C++ allows users to control the
12123 order of initialization of objects defined at namespace scope with the
12124 @code{init_priority} attribute by specifying a relative @var{priority},
12125 a constant integral expression currently bounded between 101 and 65535
12126 inclusive. Lower numbers indicate a higher priority.
12128 In the following example, @code{A} would normally be created before
12129 @code{B}, but the @code{init_priority} attribute has reversed that order:
12132 Some_Class A __attribute__ ((init_priority (2000)));
12133 Some_Class B __attribute__ ((init_priority (543)));
12137 Note that the particular values of @var{priority} do not matter; only their
12140 @item java_interface
12141 @cindex java_interface attribute
12143 This type attribute informs C++ that the class is a Java interface. It may
12144 only be applied to classes declared within an @code{extern "Java"} block.
12145 Calls to methods declared in this interface will be dispatched using GCJ's
12146 interface table mechanism, instead of regular virtual table dispatch.
12150 See also @xref{Namespace Association}.
12152 @node Namespace Association
12153 @section Namespace Association
12155 @strong{Caution:} The semantics of this extension are not fully
12156 defined. Users should refrain from using this extension as its
12157 semantics may change subtly over time. It is possible that this
12158 extension will be removed in future versions of G++.
12160 A using-directive with @code{__attribute ((strong))} is stronger
12161 than a normal using-directive in two ways:
12165 Templates from the used namespace can be specialized and explicitly
12166 instantiated as though they were members of the using namespace.
12169 The using namespace is considered an associated namespace of all
12170 templates in the used namespace for purposes of argument-dependent
12174 The used namespace must be nested within the using namespace so that
12175 normal unqualified lookup works properly.
12177 This is useful for composing a namespace transparently from
12178 implementation namespaces. For example:
12183 template <class T> struct A @{ @};
12185 using namespace debug __attribute ((__strong__));
12186 template <> struct A<int> @{ @}; // @r{ok to specialize}
12188 template <class T> void f (A<T>);
12193 f (std::A<float>()); // @r{lookup finds} std::f
12199 @section Type Traits
12201 The C++ front-end implements syntactic extensions that allow to
12202 determine at compile time various characteristics of a type (or of a
12206 @item __has_nothrow_assign (type)
12207 If @code{type} is const qualified or is a reference type then the trait is
12208 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
12209 is true, else if @code{type} is a cv class or union type with copy assignment
12210 operators that are known not to throw an exception then the trait is true,
12211 else it is false. Requires: @code{type} shall be a complete type, an array
12212 type of unknown bound, or is a @code{void} type.
12214 @item __has_nothrow_copy (type)
12215 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
12216 @code{type} is a cv class or union type with copy constructors that
12217 are known not to throw an exception then the trait is true, else it is false.
12218 Requires: @code{type} shall be a complete type, an array type of
12219 unknown bound, or is a @code{void} type.
12221 @item __has_nothrow_constructor (type)
12222 If @code{__has_trivial_constructor (type)} is true then the trait is
12223 true, else if @code{type} is a cv class or union type (or array
12224 thereof) with a default constructor that is known not to throw an
12225 exception then the trait is true, else it is false. Requires:
12226 @code{type} shall be a complete type, an array type of unknown bound,
12227 or is a @code{void} type.
12229 @item __has_trivial_assign (type)
12230 If @code{type} is const qualified or is a reference type then the trait is
12231 false. Otherwise if @code{__is_pod (type)} is true then the trait is
12232 true, else if @code{type} is a cv class or union type with a trivial
12233 copy assignment ([class.copy]) then the trait is true, else it is
12234 false. Requires: @code{type} shall be a complete type, an array type
12235 of unknown bound, or is a @code{void} type.
12237 @item __has_trivial_copy (type)
12238 If @code{__is_pod (type)} is true or @code{type} is a reference type
12239 then the trait is true, else if @code{type} is a cv class or union type
12240 with a trivial copy constructor ([class.copy]) then the trait
12241 is true, else it is false. Requires: @code{type} shall be a complete
12242 type, an array type of unknown bound, or is a @code{void} type.
12244 @item __has_trivial_constructor (type)
12245 If @code{__is_pod (type)} is true then the trait is true, else if
12246 @code{type} is a cv class or union type (or array thereof) with a
12247 trivial default constructor ([class.ctor]) then the trait is true,
12248 else it is false. Requires: @code{type} shall be a complete type, an
12249 array type of unknown bound, or is a @code{void} type.
12251 @item __has_trivial_destructor (type)
12252 If @code{__is_pod (type)} is true or @code{type} is a reference type then
12253 the trait is true, else if @code{type} is a cv class or union type (or
12254 array thereof) with a trivial destructor ([class.dtor]) then the trait
12255 is true, else it is false. Requires: @code{type} shall be a complete
12256 type, an array type of unknown bound, or is a @code{void} type.
12258 @item __has_virtual_destructor (type)
12259 If @code{type} is a class type with a virtual destructor
12260 ([class.dtor]) then the trait is true, else it is false. Requires:
12261 @code{type} shall be a complete type, an array type of unknown bound,
12262 or is a @code{void} type.
12264 @item __is_abstract (type)
12265 If @code{type} is an abstract class ([class.abstract]) then the trait
12266 is true, else it is false. Requires: @code{type} shall be a complete
12267 type, an array type of unknown bound, or is a @code{void} type.
12269 @item __is_base_of (base_type, derived_type)
12270 If @code{base_type} is a base class of @code{derived_type}
12271 ([class.derived]) then the trait is true, otherwise it is false.
12272 Top-level cv qualifications of @code{base_type} and
12273 @code{derived_type} are ignored. For the purposes of this trait, a
12274 class type is considered is own base. Requires: if @code{__is_class
12275 (base_type)} and @code{__is_class (derived_type)} are true and
12276 @code{base_type} and @code{derived_type} are not the same type
12277 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
12278 type. Diagnostic is produced if this requirement is not met.
12280 @item __is_class (type)
12281 If @code{type} is a cv class type, and not a union type
12282 ([basic.compound]) the the trait is true, else it is false.
12284 @item __is_empty (type)
12285 If @code{__is_class (type)} is false then the trait is false.
12286 Otherwise @code{type} is considered empty if and only if: @code{type}
12287 has no non-static data members, or all non-static data members, if
12288 any, are bit-fields of lenght 0, and @code{type} has no virtual
12289 members, and @code{type} has no virtual base classes, and @code{type}
12290 has no base classes @code{base_type} for which
12291 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
12292 be a complete type, an array type of unknown bound, or is a
12295 @item __is_enum (type)
12296 If @code{type} is a cv enumeration type ([basic.compound]) the the trait is
12297 true, else it is false.
12299 @item __is_pod (type)
12300 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
12301 else it is false. Requires: @code{type} shall be a complete type,
12302 an array type of unknown bound, or is a @code{void} type.
12304 @item __is_polymorphic (type)
12305 If @code{type} is a polymorphic class ([class.virtual]) then the trait
12306 is true, else it is false. Requires: @code{type} shall be a complete
12307 type, an array type of unknown bound, or is a @code{void} type.
12309 @item __is_union (type)
12310 If @code{type} is a cv union type ([basic.compound]) the the trait is
12311 true, else it is false.
12315 @node Java Exceptions
12316 @section Java Exceptions
12318 The Java language uses a slightly different exception handling model
12319 from C++. Normally, GNU C++ will automatically detect when you are
12320 writing C++ code that uses Java exceptions, and handle them
12321 appropriately. However, if C++ code only needs to execute destructors
12322 when Java exceptions are thrown through it, GCC will guess incorrectly.
12323 Sample problematic code is:
12326 struct S @{ ~S(); @};
12327 extern void bar(); // @r{is written in Java, and may throw exceptions}
12336 The usual effect of an incorrect guess is a link failure, complaining of
12337 a missing routine called @samp{__gxx_personality_v0}.
12339 You can inform the compiler that Java exceptions are to be used in a
12340 translation unit, irrespective of what it might think, by writing
12341 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
12342 @samp{#pragma} must appear before any functions that throw or catch
12343 exceptions, or run destructors when exceptions are thrown through them.
12345 You cannot mix Java and C++ exceptions in the same translation unit. It
12346 is believed to be safe to throw a C++ exception from one file through
12347 another file compiled for the Java exception model, or vice versa, but
12348 there may be bugs in this area.
12350 @node Deprecated Features
12351 @section Deprecated Features
12353 In the past, the GNU C++ compiler was extended to experiment with new
12354 features, at a time when the C++ language was still evolving. Now that
12355 the C++ standard is complete, some of those features are superseded by
12356 superior alternatives. Using the old features might cause a warning in
12357 some cases that the feature will be dropped in the future. In other
12358 cases, the feature might be gone already.
12360 While the list below is not exhaustive, it documents some of the options
12361 that are now deprecated:
12364 @item -fexternal-templates
12365 @itemx -falt-external-templates
12366 These are two of the many ways for G++ to implement template
12367 instantiation. @xref{Template Instantiation}. The C++ standard clearly
12368 defines how template definitions have to be organized across
12369 implementation units. G++ has an implicit instantiation mechanism that
12370 should work just fine for standard-conforming code.
12372 @item -fstrict-prototype
12373 @itemx -fno-strict-prototype
12374 Previously it was possible to use an empty prototype parameter list to
12375 indicate an unspecified number of parameters (like C), rather than no
12376 parameters, as C++ demands. This feature has been removed, except where
12377 it is required for backwards compatibility @xref{Backwards Compatibility}.
12380 G++ allows a virtual function returning @samp{void *} to be overridden
12381 by one returning a different pointer type. This extension to the
12382 covariant return type rules is now deprecated and will be removed from a
12385 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
12386 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
12387 and are now removed from G++. Code using these operators should be
12388 modified to use @code{std::min} and @code{std::max} instead.
12390 The named return value extension has been deprecated, and is now
12393 The use of initializer lists with new expressions has been deprecated,
12394 and is now removed from G++.
12396 Floating and complex non-type template parameters have been deprecated,
12397 and are now removed from G++.
12399 The implicit typename extension has been deprecated and is now
12402 The use of default arguments in function pointers, function typedefs
12403 and other places where they are not permitted by the standard is
12404 deprecated and will be removed from a future version of G++.
12406 G++ allows floating-point literals to appear in integral constant expressions,
12407 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
12408 This extension is deprecated and will be removed from a future version.
12410 G++ allows static data members of const floating-point type to be declared
12411 with an initializer in a class definition. The standard only allows
12412 initializers for static members of const integral types and const
12413 enumeration types so this extension has been deprecated and will be removed
12414 from a future version.
12416 @node Backwards Compatibility
12417 @section Backwards Compatibility
12418 @cindex Backwards Compatibility
12419 @cindex ARM [Annotated C++ Reference Manual]
12421 Now that there is a definitive ISO standard C++, G++ has a specification
12422 to adhere to. The C++ language evolved over time, and features that
12423 used to be acceptable in previous drafts of the standard, such as the ARM
12424 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
12425 compilation of C++ written to such drafts, G++ contains some backwards
12426 compatibilities. @emph{All such backwards compatibility features are
12427 liable to disappear in future versions of G++.} They should be considered
12428 deprecated @xref{Deprecated Features}.
12432 If a variable is declared at for scope, it used to remain in scope until
12433 the end of the scope which contained the for statement (rather than just
12434 within the for scope). G++ retains this, but issues a warning, if such a
12435 variable is accessed outside the for scope.
12437 @item Implicit C language
12438 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
12439 scope to set the language. On such systems, all header files are
12440 implicitly scoped inside a C language scope. Also, an empty prototype
12441 @code{()} will be treated as an unspecified number of arguments, rather
12442 than no arguments, as C++ demands.