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
1795 @cindex functions that have different compilation options on the 386
1796 @cindex functions that have different optimization options
1798 In GNU C, you declare certain things about functions called in your program
1799 which help the compiler optimize function calls and check your code more
1802 The keyword @code{__attribute__} allows you to specify special
1803 attributes when making a declaration. This keyword is followed by an
1804 attribute specification inside double parentheses. The following
1805 attributes are currently defined for functions on all targets:
1806 @code{aligned}, @code{alloc_size}, @code{noreturn},
1807 @code{returns_twice}, @code{noinline}, @code{always_inline},
1808 @code{flatten}, @code{pure}, @code{const}, @code{nothrow},
1809 @code{sentinel}, @code{format}, @code{format_arg},
1810 @code{no_instrument_function}, @code{section}, @code{constructor},
1811 @code{destructor}, @code{used}, @code{unused}, @code{deprecated},
1812 @code{weak}, @code{malloc}, @code{alias}, @code{warn_unused_result},
1813 @code{nonnull}, @code{gnu_inline}, @code{externally_visible},
1814 @code{hot}, @code{cold}, @code{artificial}, @code{error}
1816 Several other attributes are defined for functions on particular
1817 target systems. Other attributes, including @code{section} are
1818 supported for variables declarations (@pxref{Variable Attributes}) and
1819 for types (@pxref{Type Attributes}).
1821 You may also specify attributes with @samp{__} preceding and following
1822 each keyword. This allows you to use them in header files without
1823 being concerned about a possible macro of the same name. For example,
1824 you may use @code{__noreturn__} instead of @code{noreturn}.
1826 @xref{Attribute Syntax}, for details of the exact syntax for using
1830 @c Keep this table alphabetized by attribute name. Treat _ as space.
1832 @item alias ("@var{target}")
1833 @cindex @code{alias} attribute
1834 The @code{alias} attribute causes the declaration to be emitted as an
1835 alias for another symbol, which must be specified. For instance,
1838 void __f () @{ /* @r{Do something.} */; @}
1839 void f () __attribute__ ((weak, alias ("__f")));
1842 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
1843 mangled name for the target must be used. It is an error if @samp{__f}
1844 is not defined in the same translation unit.
1846 Not all target machines support this attribute.
1848 @item aligned (@var{alignment})
1849 @cindex @code{aligned} attribute
1850 This attribute specifies a minimum alignment for the function,
1853 You cannot use this attribute to decrease the alignment of a function,
1854 only to increase it. However, when you explicitly specify a function
1855 alignment this will override the effect of the
1856 @option{-falign-functions} (@pxref{Optimize Options}) option for this
1859 Note that the effectiveness of @code{aligned} attributes may be
1860 limited by inherent limitations in your linker. On many systems, the
1861 linker is only able to arrange for functions to be aligned up to a
1862 certain maximum alignment. (For some linkers, the maximum supported
1863 alignment may be very very small.) See your linker documentation for
1864 further information.
1866 The @code{aligned} attribute can also be used for variables and fields
1867 (@pxref{Variable Attributes}.)
1870 @cindex @code{alloc_size} attribute
1871 The @code{alloc_size} attribute is used to tell the compiler that the
1872 function return value points to memory, where the size is given by
1873 one or two of the functions parameters. GCC uses this
1874 information to improve the correctness of @code{__builtin_object_size}.
1876 The function parameter(s) denoting the allocated size are specified by
1877 one or two integer arguments supplied to the attribute. The allocated size
1878 is either the value of the single function argument specified or the product
1879 of the two function arguments specified. Argument numbering starts at
1885 void* my_calloc(size_t, size_t) __attribute__((alloc_size(1,2)))
1886 void my_realloc(void*, size_t) __attribute__((alloc_size(2)))
1889 declares that my_calloc will return memory of the size given by
1890 the product of parameter 1 and 2 and that my_realloc will return memory
1891 of the size given by parameter 2.
1894 @cindex @code{always_inline} function attribute
1895 Generally, functions are not inlined unless optimization is specified.
1896 For functions declared inline, this attribute inlines the function even
1897 if no optimization level was specified.
1900 @cindex @code{gnu_inline} function attribute
1901 This attribute should be used with a function which is also declared
1902 with the @code{inline} keyword. It directs GCC to treat the function
1903 as if it were defined in gnu89 mode even when compiling in C99 or
1906 If the function is declared @code{extern}, then this definition of the
1907 function is used only for inlining. In no case is the function
1908 compiled as a standalone function, not even if you take its address
1909 explicitly. Such an address becomes an external reference, as if you
1910 had only declared the function, and had not defined it. This has
1911 almost the effect of a macro. The way to use this is to put a
1912 function definition in a header file with this attribute, and put
1913 another copy of the function, without @code{extern}, in a library
1914 file. The definition in the header file will cause most calls to the
1915 function to be inlined. If any uses of the function remain, they will
1916 refer to the single copy in the library. Note that the two
1917 definitions of the functions need not be precisely the same, although
1918 if they do not have the same effect your program may behave oddly.
1920 In C, if the function is neither @code{extern} nor @code{static}, then
1921 the function is compiled as a standalone function, as well as being
1922 inlined where possible.
1924 This is how GCC traditionally handled functions declared
1925 @code{inline}. Since ISO C99 specifies a different semantics for
1926 @code{inline}, this function attribute is provided as a transition
1927 measure and as a useful feature in its own right. This attribute is
1928 available in GCC 4.1.3 and later. It is available if either of the
1929 preprocessor macros @code{__GNUC_GNU_INLINE__} or
1930 @code{__GNUC_STDC_INLINE__} are defined. @xref{Inline,,An Inline
1931 Function is As Fast As a Macro}.
1933 In C++, this attribute does not depend on @code{extern} in any way,
1934 but it still requires the @code{inline} keyword to enable its special
1938 @cindex @code{artificial} function attribute
1939 This attribute is useful for small inline wrappers which if possible
1940 should appear during debugging as a unit, depending on the debug
1941 info format it will either mean marking the function as artificial
1942 or using the caller location for all instructions within the inlined
1946 @cindex @code{flatten} function attribute
1947 Generally, inlining into a function is limited. For a function marked with
1948 this attribute, every call inside this function will be inlined, if possible.
1949 Whether the function itself is considered for inlining depends on its size and
1950 the current inlining parameters.
1952 @item error ("@var{message}")
1953 @cindex @code{error} function attribute
1954 If this attribute is used on a function declaration and a call to such a function
1955 is not eliminated through dead code elimination or other optimizations, an error
1956 which will include @var{message} will be diagnosed. This is useful
1957 for compile time checking, especially together with @code{__builtin_constant_p}
1958 and inline functions where checking the inline function arguments is not
1959 possible through @code{extern char [(condition) ? 1 : -1];} tricks.
1960 While it is possible to leave the function undefined and thus invoke
1961 a link failure, when using this attribute the problem will be diagnosed
1962 earlier and with exact location of the call even in presence of inline
1963 functions or when not emitting debugging information.
1965 @item warning ("@var{message}")
1966 @cindex @code{warning} function attribute
1967 If this attribute is used on a function declaration and a call to such a function
1968 is not eliminated through dead code elimination or other optimizations, a warning
1969 which will include @var{message} will be diagnosed. This is useful
1970 for compile time checking, especially together with @code{__builtin_constant_p}
1971 and inline functions. While it is possible to define the function with
1972 a message in @code{.gnu.warning*} section, when using this attribute the problem
1973 will be diagnosed earlier and with exact location of the call even in presence
1974 of inline functions or when not emitting debugging information.
1977 @cindex functions that do pop the argument stack on the 386
1979 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1980 assume that the calling function will pop off the stack space used to
1981 pass arguments. This is
1982 useful to override the effects of the @option{-mrtd} switch.
1985 @cindex @code{const} function attribute
1986 Many functions do not examine any values except their arguments, and
1987 have no effects except the return value. Basically this is just slightly
1988 more strict class than the @code{pure} attribute below, since function is not
1989 allowed to read global memory.
1991 @cindex pointer arguments
1992 Note that a function that has pointer arguments and examines the data
1993 pointed to must @emph{not} be declared @code{const}. Likewise, a
1994 function that calls a non-@code{const} function usually must not be
1995 @code{const}. It does not make sense for a @code{const} function to
1998 The attribute @code{const} is not implemented in GCC versions earlier
1999 than 2.5. An alternative way to declare that a function has no side
2000 effects, which works in the current version and in some older versions,
2004 typedef int intfn ();
2006 extern const intfn square;
2009 This approach does not work in GNU C++ from 2.6.0 on, since the language
2010 specifies that the @samp{const} must be attached to the return value.
2014 @itemx constructor (@var{priority})
2015 @itemx destructor (@var{priority})
2016 @cindex @code{constructor} function attribute
2017 @cindex @code{destructor} function attribute
2018 The @code{constructor} attribute causes the function to be called
2019 automatically before execution enters @code{main ()}. Similarly, the
2020 @code{destructor} attribute causes the function to be called
2021 automatically after @code{main ()} has completed or @code{exit ()} has
2022 been called. Functions with these attributes are useful for
2023 initializing data that will be used implicitly during the execution of
2026 You may provide an optional integer priority to control the order in
2027 which constructor and destructor functions are run. A constructor
2028 with a smaller priority number runs before a constructor with a larger
2029 priority number; the opposite relationship holds for destructors. So,
2030 if you have a constructor that allocates a resource and a destructor
2031 that deallocates the same resource, both functions typically have the
2032 same priority. The priorities for constructor and destructor
2033 functions are the same as those specified for namespace-scope C++
2034 objects (@pxref{C++ Attributes}).
2036 These attributes are not currently implemented for Objective-C@.
2039 @cindex @code{deprecated} attribute.
2040 The @code{deprecated} attribute results in a warning if the function
2041 is used anywhere in the source file. This is useful when identifying
2042 functions that are expected to be removed in a future version of a
2043 program. The warning also includes the location of the declaration
2044 of the deprecated function, to enable users to easily find further
2045 information about why the function is deprecated, or what they should
2046 do instead. Note that the warnings only occurs for uses:
2049 int old_fn () __attribute__ ((deprecated));
2051 int (*fn_ptr)() = old_fn;
2054 results in a warning on line 3 but not line 2.
2056 The @code{deprecated} attribute can also be used for variables and
2057 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2060 @cindex @code{__declspec(dllexport)}
2061 On Microsoft Windows targets and Symbian OS targets the
2062 @code{dllexport} attribute causes the compiler to provide a global
2063 pointer to a pointer in a DLL, so that it can be referenced with the
2064 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
2065 name is formed by combining @code{_imp__} and the function or variable
2068 You can use @code{__declspec(dllexport)} as a synonym for
2069 @code{__attribute__ ((dllexport))} for compatibility with other
2072 On systems that support the @code{visibility} attribute, this
2073 attribute also implies ``default'' visibility. It is an error to
2074 explicitly specify any other visibility.
2076 Currently, the @code{dllexport} attribute is ignored for inlined
2077 functions, unless the @option{-fkeep-inline-functions} flag has been
2078 used. The attribute is also ignored for undefined symbols.
2080 When applied to C++ classes, the attribute marks defined non-inlined
2081 member functions and static data members as exports. Static consts
2082 initialized in-class are not marked unless they are also defined
2085 For Microsoft Windows targets there are alternative methods for
2086 including the symbol in the DLL's export table such as using a
2087 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
2088 the @option{--export-all} linker flag.
2091 @cindex @code{__declspec(dllimport)}
2092 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
2093 attribute causes the compiler to reference a function or variable via
2094 a global pointer to a pointer that is set up by the DLL exporting the
2095 symbol. The attribute implies @code{extern}. On Microsoft Windows
2096 targets, the pointer name is formed by combining @code{_imp__} and the
2097 function or variable name.
2099 You can use @code{__declspec(dllimport)} as a synonym for
2100 @code{__attribute__ ((dllimport))} for compatibility with other
2103 On systems that support the @code{visibility} attribute, this
2104 attribute also implies ``default'' visibility. It is an error to
2105 explicitly specify any other visibility.
2107 Currently, the attribute is ignored for inlined functions. If the
2108 attribute is applied to a symbol @emph{definition}, an error is reported.
2109 If a symbol previously declared @code{dllimport} is later defined, the
2110 attribute is ignored in subsequent references, and a warning is emitted.
2111 The attribute is also overridden by a subsequent declaration as
2114 When applied to C++ classes, the attribute marks non-inlined
2115 member functions and static data members as imports. However, the
2116 attribute is ignored for virtual methods to allow creation of vtables
2119 On the SH Symbian OS target the @code{dllimport} attribute also has
2120 another affect---it can cause the vtable and run-time type information
2121 for a class to be exported. This happens when the class has a
2122 dllimport'ed constructor or a non-inline, non-pure virtual function
2123 and, for either of those two conditions, the class also has a inline
2124 constructor or destructor and has a key function that is defined in
2125 the current translation unit.
2127 For Microsoft Windows based targets the use of the @code{dllimport}
2128 attribute on functions is not necessary, but provides a small
2129 performance benefit by eliminating a thunk in the DLL@. The use of the
2130 @code{dllimport} attribute on imported variables was required on older
2131 versions of the GNU linker, but can now be avoided by passing the
2132 @option{--enable-auto-import} switch to the GNU linker. As with
2133 functions, using the attribute for a variable eliminates a thunk in
2136 One drawback to using this attribute is that a pointer to a
2137 @emph{variable} marked as @code{dllimport} cannot be used as a constant
2138 address. However, a pointer to a @emph{function} with the
2139 @code{dllimport} attribute can be used as a constant initializer; in
2140 this case, the address of a stub function in the import lib is
2141 referenced. On Microsoft Windows targets, the attribute can be disabled
2142 for functions by setting the @option{-mnop-fun-dllimport} flag.
2145 @cindex eight bit data on the H8/300, H8/300H, and H8S
2146 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2147 variable should be placed into the eight bit data section.
2148 The compiler will generate more efficient code for certain operations
2149 on data in the eight bit data area. Note the eight bit data area is limited to
2152 You must use GAS and GLD from GNU binutils version 2.7 or later for
2153 this attribute to work correctly.
2155 @item exception_handler
2156 @cindex exception handler functions on the Blackfin processor
2157 Use this attribute on the Blackfin to indicate that the specified function
2158 is an exception handler. The compiler will generate function entry and
2159 exit sequences suitable for use in an exception handler when this
2160 attribute is present.
2162 @item externally_visible
2163 @cindex @code{externally_visible} attribute.
2164 This attribute, attached to a global variable or function, nullifies
2165 the effect of the @option{-fwhole-program} command-line option, so the
2166 object remains visible outside the current compilation unit.
2169 @cindex functions which handle memory bank switching
2170 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
2171 use a calling convention that takes care of switching memory banks when
2172 entering and leaving a function. This calling convention is also the
2173 default when using the @option{-mlong-calls} option.
2175 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
2176 to call and return from a function.
2178 On 68HC11 the compiler will generate a sequence of instructions
2179 to invoke a board-specific routine to switch the memory bank and call the
2180 real function. The board-specific routine simulates a @code{call}.
2181 At the end of a function, it will jump to a board-specific routine
2182 instead of using @code{rts}. The board-specific return routine simulates
2186 @cindex functions that pop the argument stack on the 386
2187 On the Intel 386, the @code{fastcall} attribute causes the compiler to
2188 pass the first argument (if of integral type) in the register ECX and
2189 the second argument (if of integral type) in the register EDX@. Subsequent
2190 and other typed arguments are passed on the stack. The called function will
2191 pop the arguments off the stack. If the number of arguments is variable all
2192 arguments are pushed on the stack.
2194 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2195 @cindex @code{format} function attribute
2197 The @code{format} attribute specifies that a function takes @code{printf},
2198 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
2199 should be type-checked against a format string. For example, the
2204 my_printf (void *my_object, const char *my_format, ...)
2205 __attribute__ ((format (printf, 2, 3)));
2209 causes the compiler to check the arguments in calls to @code{my_printf}
2210 for consistency with the @code{printf} style format string argument
2213 The parameter @var{archetype} determines how the format string is
2214 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime},
2215 @code{gnu_printf}, @code{gnu_scanf}, @code{gnu_strftime} or
2216 @code{strfmon}. (You can also use @code{__printf__},
2217 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) On
2218 MinGW targets, @code{ms_printf}, @code{ms_scanf}, and
2219 @code{ms_strftime} are also present.
2220 @var{archtype} values such as @code{printf} refer to the formats accepted
2221 by the system's C run-time library, while @code{gnu_} values always refer
2222 to the formats accepted by the GNU C Library. On Microsoft Windows
2223 targets, @code{ms_} values refer to the formats accepted by the
2224 @file{msvcrt.dll} library.
2225 The parameter @var{string-index}
2226 specifies which argument is the format string argument (starting
2227 from 1), while @var{first-to-check} is the number of the first
2228 argument to check against the format string. For functions
2229 where the arguments are not available to be checked (such as
2230 @code{vprintf}), specify the third parameter as zero. In this case the
2231 compiler only checks the format string for consistency. For
2232 @code{strftime} formats, the third parameter is required to be zero.
2233 Since non-static C++ methods have an implicit @code{this} argument, the
2234 arguments of such methods should be counted from two, not one, when
2235 giving values for @var{string-index} and @var{first-to-check}.
2237 In the example above, the format string (@code{my_format}) is the second
2238 argument of the function @code{my_print}, and the arguments to check
2239 start with the third argument, so the correct parameters for the format
2240 attribute are 2 and 3.
2242 @opindex ffreestanding
2243 @opindex fno-builtin
2244 The @code{format} attribute allows you to identify your own functions
2245 which take format strings as arguments, so that GCC can check the
2246 calls to these functions for errors. The compiler always (unless
2247 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
2248 for the standard library functions @code{printf}, @code{fprintf},
2249 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2250 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2251 warnings are requested (using @option{-Wformat}), so there is no need to
2252 modify the header file @file{stdio.h}. In C99 mode, the functions
2253 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2254 @code{vsscanf} are also checked. Except in strictly conforming C
2255 standard modes, the X/Open function @code{strfmon} is also checked as
2256 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2257 @xref{C Dialect Options,,Options Controlling C Dialect}.
2259 The target may provide additional types of format checks.
2260 @xref{Target Format Checks,,Format Checks Specific to Particular
2263 @item format_arg (@var{string-index})
2264 @cindex @code{format_arg} function attribute
2265 @opindex Wformat-nonliteral
2266 The @code{format_arg} attribute specifies that a function takes a format
2267 string for a @code{printf}, @code{scanf}, @code{strftime} or
2268 @code{strfmon} style function and modifies it (for example, to translate
2269 it into another language), so the result can be passed to a
2270 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2271 function (with the remaining arguments to the format function the same
2272 as they would have been for the unmodified string). For example, the
2277 my_dgettext (char *my_domain, const char *my_format)
2278 __attribute__ ((format_arg (2)));
2282 causes the compiler to check the arguments in calls to a @code{printf},
2283 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2284 format string argument is a call to the @code{my_dgettext} function, for
2285 consistency with the format string argument @code{my_format}. If the
2286 @code{format_arg} attribute had not been specified, all the compiler
2287 could tell in such calls to format functions would be that the format
2288 string argument is not constant; this would generate a warning when
2289 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2290 without the attribute.
2292 The parameter @var{string-index} specifies which argument is the format
2293 string argument (starting from one). Since non-static C++ methods have
2294 an implicit @code{this} argument, the arguments of such methods should
2295 be counted from two.
2297 The @code{format-arg} attribute allows you to identify your own
2298 functions which modify format strings, so that GCC can check the
2299 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2300 type function whose operands are a call to one of your own function.
2301 The compiler always treats @code{gettext}, @code{dgettext}, and
2302 @code{dcgettext} in this manner except when strict ISO C support is
2303 requested by @option{-ansi} or an appropriate @option{-std} option, or
2304 @option{-ffreestanding} or @option{-fno-builtin}
2305 is used. @xref{C Dialect Options,,Options
2306 Controlling C Dialect}.
2308 @item function_vector
2309 @cindex calling functions through the function vector on H8/300, M16C, M32C and SH2A processors
2310 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2311 function should be called through the function vector. Calling a
2312 function through the function vector will reduce code size, however;
2313 the function vector has a limited size (maximum 128 entries on the H8/300
2314 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2316 In SH2A target, this attribute declares a function to be called using the
2317 TBR relative addressing mode. The argument to this attribute is the entry
2318 number of the same function in a vector table containing all the TBR
2319 relative addressable functions. For the successful jump, register TBR
2320 should contain the start address of this TBR relative vector table.
2321 In the startup routine of the user application, user needs to care of this
2322 TBR register initialization. The TBR relative vector table can have at
2323 max 256 function entries. The jumps to these functions will be generated
2324 using a SH2A specific, non delayed branch instruction JSR/N @@(disp8,TBR).
2325 You must use GAS and GLD from GNU binutils version 2.7 or later for
2326 this attribute to work correctly.
2328 Please refer the example of M16C target, to see the use of this
2329 attribute while declaring a function,
2331 In an application, for a function being called once, this attribute will
2332 save at least 8 bytes of code; and if other successive calls are being
2333 made to the same function, it will save 2 bytes of code per each of these
2336 On M16C/M32C targets, the @code{function_vector} attribute declares a
2337 special page subroutine call function. Use of this attribute reduces
2338 the code size by 2 bytes for each call generated to the
2339 subroutine. The argument to the attribute is the vector number entry
2340 from the special page vector table which contains the 16 low-order
2341 bits of the subroutine's entry address. Each vector table has special
2342 page number (18 to 255) which are used in @code{jsrs} instruction.
2343 Jump addresses of the routines are generated by adding 0x0F0000 (in
2344 case of M16C targets) or 0xFF0000 (in case of M32C targets), to the 2
2345 byte addresses set in the vector table. Therefore you need to ensure
2346 that all the special page vector routines should get mapped within the
2347 address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF
2350 In the following example 2 bytes will be saved for each call to
2351 function @code{foo}.
2354 void foo (void) __attribute__((function_vector(0x18)));
2365 If functions are defined in one file and are called in another file,
2366 then be sure to write this declaration in both files.
2368 This attribute is ignored for R8C target.
2371 @cindex interrupt handler functions
2372 Use this attribute on the ARM, AVR, CRX, M32C, M32R/D, m68k,
2373 and Xstormy16 ports to indicate that the specified function is an
2374 interrupt handler. The compiler will generate function entry and exit
2375 sequences suitable for use in an interrupt handler when this attribute
2378 Note, interrupt handlers for the Blackfin, H8/300, H8/300H, H8S, and
2379 SH processors can be specified via the @code{interrupt_handler} attribute.
2381 Note, on the AVR, interrupts will be enabled inside the function.
2383 Note, for the ARM, you can specify the kind of interrupt to be handled by
2384 adding an optional parameter to the interrupt attribute like this:
2387 void f () __attribute__ ((interrupt ("IRQ")));
2390 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2392 On ARMv7-M the interrupt type is ignored, and the attribute means the function
2393 may be called with a word aligned stack pointer.
2395 @item interrupt_handler
2396 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
2397 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
2398 indicate that the specified function is an interrupt handler. The compiler
2399 will generate function entry and exit sequences suitable for use in an
2400 interrupt handler when this attribute is present.
2402 @item interrupt_thread
2403 @cindex interrupt thread functions on fido
2404 Use this attribute on fido, a subarchitecture of the m68k, to indicate
2405 that the specified function is an interrupt handler that is designed
2406 to run as a thread. The compiler omits generate prologue/epilogue
2407 sequences and replaces the return instruction with a @code{sleep}
2408 instruction. This attribute is available only on fido.
2411 @cindex User stack pointer in interrupts on the Blackfin
2412 When used together with @code{interrupt_handler}, @code{exception_handler}
2413 or @code{nmi_handler}, code will be generated to load the stack pointer
2414 from the USP register in the function prologue.
2417 @cindex @code{l1_text} function attribute
2418 This attribute specifies a function to be placed into L1 Instruction
2419 SRAM@. The function will be put into a specific section named @code{.l1.text}.
2420 With @option{-mfdpic}, function calls with a such function as the callee
2421 or caller will use inlined PLT.
2423 @item long_call/short_call
2424 @cindex indirect calls on ARM
2425 This attribute specifies how a particular function is called on
2426 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2427 command line switch and @code{#pragma long_calls} settings. The
2428 @code{long_call} attribute indicates that the function might be far
2429 away from the call site and require a different (more expensive)
2430 calling sequence. The @code{short_call} attribute always places
2431 the offset to the function from the call site into the @samp{BL}
2432 instruction directly.
2434 @item longcall/shortcall
2435 @cindex functions called via pointer on the RS/6000 and PowerPC
2436 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
2437 indicates that the function might be far away from the call site and
2438 require a different (more expensive) calling sequence. The
2439 @code{shortcall} attribute indicates that the function is always close
2440 enough for the shorter calling sequence to be used. These attributes
2441 override both the @option{-mlongcall} switch and, on the RS/6000 and
2442 PowerPC, the @code{#pragma longcall} setting.
2444 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2445 calls are necessary.
2447 @item long_call/near/far
2448 @cindex indirect calls on MIPS
2449 These attributes specify how a particular function is called on MIPS@.
2450 The attributes override the @option{-mlong-calls} (@pxref{MIPS Options})
2451 command-line switch. The @code{long_call} and @code{far} attributes are
2452 synonyms, and cause the compiler to always call
2453 the function by first loading its address into a register, and then using
2454 the contents of that register. The @code{near} attribute has the opposite
2455 effect; it specifies that non-PIC calls should be made using the more
2456 efficient @code{jal} instruction.
2459 @cindex @code{malloc} attribute
2460 The @code{malloc} attribute is used to tell the compiler that a function
2461 may be treated as if any non-@code{NULL} pointer it returns cannot
2462 alias any other pointer valid when the function returns.
2463 This will often improve optimization.
2464 Standard functions with this property include @code{malloc} and
2465 @code{calloc}. @code{realloc}-like functions have this property as
2466 long as the old pointer is never referred to (including comparing it
2467 to the new pointer) after the function returns a non-@code{NULL}
2470 @item mips16/nomips16
2471 @cindex @code{mips16} attribute
2472 @cindex @code{nomips16} attribute
2474 On MIPS targets, you can use the @code{mips16} and @code{nomips16}
2475 function attributes to locally select or turn off MIPS16 code generation.
2476 A function with the @code{mips16} attribute is emitted as MIPS16 code,
2477 while MIPS16 code generation is disabled for functions with the
2478 @code{nomips16} attribute. These attributes override the
2479 @option{-mips16} and @option{-mno-mips16} options on the command line
2480 (@pxref{MIPS Options}).
2482 When compiling files containing mixed MIPS16 and non-MIPS16 code, the
2483 preprocessor symbol @code{__mips16} reflects the setting on the command line,
2484 not that within individual functions. Mixed MIPS16 and non-MIPS16 code
2485 may interact badly with some GCC extensions such as @code{__builtin_apply}
2486 (@pxref{Constructing Calls}).
2488 @item model (@var{model-name})
2489 @cindex function addressability on the M32R/D
2490 @cindex variable addressability on the IA-64
2492 On the M32R/D, use this attribute to set the addressability of an
2493 object, and of the code generated for a function. The identifier
2494 @var{model-name} is one of @code{small}, @code{medium}, or
2495 @code{large}, representing each of the code models.
2497 Small model objects live in the lower 16MB of memory (so that their
2498 addresses can be loaded with the @code{ld24} instruction), and are
2499 callable with the @code{bl} instruction.
2501 Medium model objects may live anywhere in the 32-bit address space (the
2502 compiler will generate @code{seth/add3} instructions to load their addresses),
2503 and are callable with the @code{bl} instruction.
2505 Large model objects may live anywhere in the 32-bit address space (the
2506 compiler will generate @code{seth/add3} instructions to load their addresses),
2507 and may not be reachable with the @code{bl} instruction (the compiler will
2508 generate the much slower @code{seth/add3/jl} instruction sequence).
2510 On IA-64, use this attribute to set the addressability of an object.
2511 At present, the only supported identifier for @var{model-name} is
2512 @code{small}, indicating addressability via ``small'' (22-bit)
2513 addresses (so that their addresses can be loaded with the @code{addl}
2514 instruction). Caveat: such addressing is by definition not position
2515 independent and hence this attribute must not be used for objects
2516 defined by shared libraries.
2518 @item ms_abi/sysv_abi
2519 @cindex @code{ms_abi} attribute
2520 @cindex @code{sysv_abi} attribute
2522 On 64-bit x86_65-*-* targets, you can use an ABI attribute to indicate
2523 which calling convention should be used for a function. The @code{ms_abi}
2524 attribute tells the compiler to use the Microsoft ABI, while the
2525 @code{sysv_abi} attribute tells the compiler to use the ABI used on
2526 GNU/Linux and other systems. The default is to use the Microsoft ABI
2527 when targeting Windows. On all other systems, the default is the AMD ABI.
2529 Note, This feature is currently sorried out for Windows targets trying to
2532 @cindex function without a prologue/epilogue code
2533 Use this attribute on the ARM, AVR, IP2K and SPU ports to indicate that
2534 the specified function does not need prologue/epilogue sequences generated by
2535 the compiler. It is up to the programmer to provide these sequences. The
2536 only statements that can be safely included in naked functions are
2537 @code{asm} statements that do not have operands. All other statements,
2538 including declarations of local variables, @code{if} statements, and so
2539 forth, should be avoided. Naked functions should be used to implement the
2540 body of an assembly function, while allowing the compiler to construct
2541 the requisite function declaration for the assembler.
2544 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2545 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2546 use the normal calling convention based on @code{jsr} and @code{rts}.
2547 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2551 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2552 Use this attribute together with @code{interrupt_handler},
2553 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2554 entry code should enable nested interrupts or exceptions.
2557 @cindex NMI handler functions on the Blackfin processor
2558 Use this attribute on the Blackfin to indicate that the specified function
2559 is an NMI handler. The compiler will generate function entry and
2560 exit sequences suitable for use in an NMI handler when this
2561 attribute is present.
2563 @item no_instrument_function
2564 @cindex @code{no_instrument_function} function attribute
2565 @opindex finstrument-functions
2566 If @option{-finstrument-functions} is given, profiling function calls will
2567 be generated at entry and exit of most user-compiled functions.
2568 Functions with this attribute will not be so instrumented.
2571 @cindex @code{noinline} function attribute
2572 This function attribute prevents a function from being considered for
2574 @c Don't enumerate the optimizations by name here; we try to be
2575 @c future-compatible with this mechanism.
2576 If the function does not have side-effects, there are optimizations
2577 other than inlining that causes function calls to be optimized away,
2578 although the function call is live. To keep such calls from being
2583 (@pxref{Extended Asm}) in the called function, to serve as a special
2586 @item nonnull (@var{arg-index}, @dots{})
2587 @cindex @code{nonnull} function attribute
2588 The @code{nonnull} attribute specifies that some function parameters should
2589 be non-null pointers. For instance, the declaration:
2593 my_memcpy (void *dest, const void *src, size_t len)
2594 __attribute__((nonnull (1, 2)));
2598 causes the compiler to check that, in calls to @code{my_memcpy},
2599 arguments @var{dest} and @var{src} are non-null. If the compiler
2600 determines that a null pointer is passed in an argument slot marked
2601 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2602 is issued. The compiler may also choose to make optimizations based
2603 on the knowledge that certain function arguments will not be null.
2605 If no argument index list is given to the @code{nonnull} attribute,
2606 all pointer arguments are marked as non-null. To illustrate, the
2607 following declaration is equivalent to the previous example:
2611 my_memcpy (void *dest, const void *src, size_t len)
2612 __attribute__((nonnull));
2616 @cindex @code{noreturn} function attribute
2617 A few standard library functions, such as @code{abort} and @code{exit},
2618 cannot return. GCC knows this automatically. Some programs define
2619 their own functions that never return. You can declare them
2620 @code{noreturn} to tell the compiler this fact. For example,
2624 void fatal () __attribute__ ((noreturn));
2627 fatal (/* @r{@dots{}} */)
2629 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2635 The @code{noreturn} keyword tells the compiler to assume that
2636 @code{fatal} cannot return. It can then optimize without regard to what
2637 would happen if @code{fatal} ever did return. This makes slightly
2638 better code. More importantly, it helps avoid spurious warnings of
2639 uninitialized variables.
2641 The @code{noreturn} keyword does not affect the exceptional path when that
2642 applies: a @code{noreturn}-marked function may still return to the caller
2643 by throwing an exception or calling @code{longjmp}.
2645 Do not assume that registers saved by the calling function are
2646 restored before calling the @code{noreturn} function.
2648 It does not make sense for a @code{noreturn} function to have a return
2649 type other than @code{void}.
2651 The attribute @code{noreturn} is not implemented in GCC versions
2652 earlier than 2.5. An alternative way to declare that a function does
2653 not return, which works in the current version and in some older
2654 versions, is as follows:
2657 typedef void voidfn ();
2659 volatile voidfn fatal;
2662 This approach does not work in GNU C++.
2665 @cindex @code{nothrow} function attribute
2666 The @code{nothrow} attribute is used to inform the compiler that a
2667 function cannot throw an exception. For example, most functions in
2668 the standard C library can be guaranteed not to throw an exception
2669 with the notable exceptions of @code{qsort} and @code{bsearch} that
2670 take function pointer arguments. The @code{nothrow} attribute is not
2671 implemented in GCC versions earlier than 3.3.
2674 @cindex @code{optimize} function attribute
2675 The @code{optimize} attribute is used to specify that a function is to
2676 be compiled with different optimization options than specified on the
2677 command line. Arguments can either be numbers or strings. Numbers
2678 are assumed to be an optimization level. Strings that begin with
2679 @code{O} are assumed to be an optimization option, while other options
2680 are assumed to be used with a @code{-f} prefix. You can also use the
2681 @samp{#pragma GCC optimize} pragma to set the optimization options
2682 that affect more than one function.
2683 @xref{Function Specific Option Pragmas}, for details about the
2684 @samp{#pragma GCC optimize} pragma.
2686 This can be used for instance to have frequently executed functions
2687 compiled with more aggressive optimization options that produce faster
2688 and larger code, while other functions can be called with less
2692 @cindex @code{pure} function attribute
2693 Many functions have no effects except the return value and their
2694 return value depends only on the parameters and/or global variables.
2695 Such a function can be subject
2696 to common subexpression elimination and loop optimization just as an
2697 arithmetic operator would be. These functions should be declared
2698 with the attribute @code{pure}. For example,
2701 int square (int) __attribute__ ((pure));
2705 says that the hypothetical function @code{square} is safe to call
2706 fewer times than the program says.
2708 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2709 Interesting non-pure functions are functions with infinite loops or those
2710 depending on volatile memory or other system resource, that may change between
2711 two consecutive calls (such as @code{feof} in a multithreading environment).
2713 The attribute @code{pure} is not implemented in GCC versions earlier
2717 @cindex @code{hot} function attribute
2718 The @code{hot} attribute is used to inform the compiler that a function is a
2719 hot spot of the compiled program. The function is optimized more aggressively
2720 and on many target it is placed into special subsection of the text section so
2721 all hot functions appears close together improving locality.
2723 When profile feedback is available, via @option{-fprofile-use}, hot functions
2724 are automatically detected and this attribute is ignored.
2726 The @code{hot} attribute is not implemented in GCC versions earlier
2730 @cindex @code{cold} function attribute
2731 The @code{cold} attribute is used to inform the compiler that a function is
2732 unlikely executed. The function is optimized for size rather than speed and on
2733 many targets it is placed into special subsection of the text section so all
2734 cold functions appears close together improving code locality of non-cold parts
2735 of program. The paths leading to call of cold functions within code are marked
2736 as unlikely by the branch prediction mechanism. It is thus useful to mark
2737 functions used to handle unlikely conditions, such as @code{perror}, as cold to
2738 improve optimization of hot functions that do call marked functions in rare
2741 When profile feedback is available, via @option{-fprofile-use}, hot functions
2742 are automatically detected and this attribute is ignored.
2744 The @code{cold} attribute is not implemented in GCC versions earlier than 4.3.
2746 @item regparm (@var{number})
2747 @cindex @code{regparm} attribute
2748 @cindex functions that are passed arguments in registers on the 386
2749 On the Intel 386, the @code{regparm} attribute causes the compiler to
2750 pass arguments number one to @var{number} if they are of integral type
2751 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2752 take a variable number of arguments will continue to be passed all of their
2753 arguments on the stack.
2755 Beware that on some ELF systems this attribute is unsuitable for
2756 global functions in shared libraries with lazy binding (which is the
2757 default). Lazy binding will send the first call via resolving code in
2758 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2759 per the standard calling conventions. Solaris 8 is affected by this.
2760 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2761 safe since the loaders there save all registers. (Lazy binding can be
2762 disabled with the linker or the loader if desired, to avoid the
2766 @cindex @code{sseregparm} attribute
2767 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2768 causes the compiler to pass up to 3 floating point arguments in
2769 SSE registers instead of on the stack. Functions that take a
2770 variable number of arguments will continue to pass all of their
2771 floating point arguments on the stack.
2773 @item force_align_arg_pointer
2774 @cindex @code{force_align_arg_pointer} attribute
2775 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
2776 applied to individual function definitions, generating an alternate
2777 prologue and epilogue that realigns the runtime stack if necessary.
2778 This supports mixing legacy codes that run with a 4-byte aligned stack
2779 with modern codes that keep a 16-byte stack for SSE compatibility.
2782 @cindex @code{resbank} attribute
2783 On the SH2A target, this attribute enables the high-speed register
2784 saving and restoration using a register bank for @code{interrupt_handler}
2785 routines. Saving to the bank is performed automatcially after the CPU
2786 accepts an interrupt that uses a register bank.
2788 The nineteen 32-bit registers comprising general register R0 to R14,
2789 control register GBR, and system registers MACH, MACL, and PR and the
2790 vector table address offset are saved into a register bank. Register
2791 banks are stacked in first-in last-out (FILO) sequence. Restoration
2792 from the bank is executed by issuing a RESBANK instruction.
2795 @cindex @code{returns_twice} attribute
2796 The @code{returns_twice} attribute tells the compiler that a function may
2797 return more than one time. The compiler will ensure that all registers
2798 are dead before calling such a function and will emit a warning about
2799 the variables that may be clobbered after the second return from the
2800 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2801 The @code{longjmp}-like counterpart of such function, if any, might need
2802 to be marked with the @code{noreturn} attribute.
2805 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2806 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2807 all registers except the stack pointer should be saved in the prologue
2808 regardless of whether they are used or not.
2810 @item section ("@var{section-name}")
2811 @cindex @code{section} function attribute
2812 Normally, the compiler places the code it generates in the @code{text} section.
2813 Sometimes, however, you need additional sections, or you need certain
2814 particular functions to appear in special sections. The @code{section}
2815 attribute specifies that a function lives in a particular section.
2816 For example, the declaration:
2819 extern void foobar (void) __attribute__ ((section ("bar")));
2823 puts the function @code{foobar} in the @code{bar} section.
2825 Some file formats do not support arbitrary sections so the @code{section}
2826 attribute is not available on all platforms.
2827 If you need to map the entire contents of a module to a particular
2828 section, consider using the facilities of the linker instead.
2831 @cindex @code{sentinel} function attribute
2832 This function attribute ensures that a parameter in a function call is
2833 an explicit @code{NULL}. The attribute is only valid on variadic
2834 functions. By default, the sentinel is located at position zero, the
2835 last parameter of the function call. If an optional integer position
2836 argument P is supplied to the attribute, the sentinel must be located at
2837 position P counting backwards from the end of the argument list.
2840 __attribute__ ((sentinel))
2842 __attribute__ ((sentinel(0)))
2845 The attribute is automatically set with a position of 0 for the built-in
2846 functions @code{execl} and @code{execlp}. The built-in function
2847 @code{execle} has the attribute set with a position of 1.
2849 A valid @code{NULL} in this context is defined as zero with any pointer
2850 type. If your system defines the @code{NULL} macro with an integer type
2851 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2852 with a copy that redefines NULL appropriately.
2854 The warnings for missing or incorrect sentinels are enabled with
2858 See long_call/short_call.
2861 See longcall/shortcall.
2864 @cindex signal handler functions on the AVR processors
2865 Use this attribute on the AVR to indicate that the specified
2866 function is a signal handler. The compiler will generate function
2867 entry and exit sequences suitable for use in a signal handler when this
2868 attribute is present. Interrupts will be disabled inside the function.
2871 Use this attribute on the SH to indicate an @code{interrupt_handler}
2872 function should switch to an alternate stack. It expects a string
2873 argument that names a global variable holding the address of the
2878 void f () __attribute__ ((interrupt_handler,
2879 sp_switch ("alt_stack")));
2883 @cindex functions that pop the argument stack on the 386
2884 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2885 assume that the called function will pop off the stack space used to
2886 pass arguments, unless it takes a variable number of arguments.
2889 @cindex @code{target} function attribute
2890 The @code{target} attribute is used to specify that a function is to
2891 be compiled with different target options than specified on the
2892 command line. This can be used for instance to have functions
2893 compiled with a different ISA (instruction set architecture) than the
2894 default. You can also use the @samp{#pragma GCC target} pragma to set
2895 more than one function to be compiled with specific target options.
2896 @xref{Function Specific Option Pragmas}, for details about the
2897 @samp{#pragma GCC target} pragma.
2899 For instance on a 386, you could compile one function with
2900 @code{target("sse4.1,arch=core2")} and another with
2901 @code{target("sse4a,arch=amdfam10")} that would be equivalent to
2902 compiling the first function with @option{-msse4.1} and
2903 @option{-march=core2} options, and the second function with
2904 @option{-msse4a} and @option{-march=amdfam10} options. It is up to the
2905 user to make sure that a function is only invoked on a machine that
2906 supports the particular ISA it was compiled for (for example by using
2907 @code{cpuid} on 386 to determine what feature bits and architecture
2911 int core2_func (void) __attribute__ ((__target__ ("arch=core2")));
2912 int sse3_func (void) __attribute__ ((__target__ ("sse3")));
2915 On the 386, the following options are allowed:
2920 @cindex @code{target("abm")} attribute
2921 Enable/disable the generation of the advanced bit instructions.
2925 @cindex @code{target("aes")} attribute
2926 Enable/disable the generation of the AES instructions.
2930 @cindex @code{target("mmx")} attribute
2931 Enable/disable the generation of the MMX instructions.
2935 @cindex @code{target("pclmul")} attribute
2936 Enable/disable the generation of the PCLMUL instructions.
2940 @cindex @code{target("popcnt")} attribute
2941 Enable/disable the generation of the POPCNT instruction.
2945 @cindex @code{target("sse")} attribute
2946 Enable/disable the generation of the SSE instructions.
2950 @cindex @code{target("sse2")} attribute
2951 Enable/disable the generation of the SSE2 instructions.
2955 @cindex @code{target("sse3")} attribute
2956 Enable/disable the generation of the SSE3 instructions.
2960 @cindex @code{target("sse4")} attribute
2961 Enable/disable the generation of the SSE4 instructions (both SSE4.1
2966 @cindex @code{target("sse4.1")} attribute
2967 Enable/disable the generation of the sse4.1 instructions.
2971 @cindex @code{target("sse4.2")} attribute
2972 Enable/disable the generation of the sse4.2 instructions.
2976 @cindex @code{target("sse4a")} attribute
2977 Enable/disable the generation of the SSE4A instructions.
2981 @cindex @code{target("sse5")} attribute
2982 Enable/disable the generation of the SSE5 instructions.
2986 @cindex @code{target("ssse3")} attribute
2987 Enable/disable the generation of the SSSE3 instructions.
2991 @cindex @code{target("cld")} attribute
2992 Enable/disable the generation of the CLD before string moves.
2994 @item fancy-math-387
2995 @itemx no-fancy-math-387
2996 @cindex @code{target("fancy-math-387")} attribute
2997 Enable/disable the generation of the @code{sin}, @code{cos}, and
2998 @code{sqrt} instructions on the 387 floating point unit.
3001 @itemx no-fused-madd
3002 @cindex @code{target("fused-madd")} attribute
3003 Enable/disable the generation of the fused multiply/add instructions.
3007 @cindex @code{target("ieee-fp")} attribute
3008 Enable/disable the generation of floating point that depends on IEEE arithmetic.
3010 @item inline-all-stringops
3011 @itemx no-inline-all-stringops
3012 @cindex @code{target("inline-all-stringops")} attribute
3013 Enable/disable inlining of string operations.
3015 @item inline-stringops-dynamically
3016 @itemx no-inline-stringops-dynamically
3017 @cindex @code{target("inline-stringops-dynamically")} attribute
3018 Enable/disable the generation of the inline code to do small string
3019 operations and calling the library routines for large operations.
3021 @item align-stringops
3022 @itemx no-align-stringops
3023 @cindex @code{target("align-stringops")} attribute
3024 Do/do not align destination of inlined string operations.
3028 @cindex @code{target("recip")} attribute
3029 Enable/disable the generation of RCPSS, RCPPS, RSQRTSS and RSQRTPS
3030 instructions followed an additional Newton-Rhapson step instead of
3031 doing a floating point division.
3033 @item arch=@var{ARCH}
3034 @cindex @code{target("arch=@var{ARCH}")} attribute
3035 Specify the architecture to generate code for in compiling the function.
3037 @item tune=@var{TUNE}
3038 @cindex @code{target("tune=@var{TUNE}")} attribute
3039 Specify the architecture to tune for in compiling the function.
3041 @item fpmath=@var{FPMATH}
3042 @cindex @code{target("fpmath=@var{FPMATH}")} attribute
3043 Specify which floating point unit to use. The
3044 @code{target("fpmath=sse,387")} option must be specified as
3045 @code{target("fpmath=sse+387")} because the comma would separate
3049 On the 386, you can use either multiple strings to specify multiple
3050 options, or you can separate the option with a comma (@code{,}).
3052 On the 386, the inliner will not inline a function that has different
3053 target options than the caller, unless the callee has a subset of the
3054 target options of the caller. For example a function declared with
3055 @code{target("sse5")} can inline a function with
3056 @code{target("sse2")}, since @code{-msse5} implies @code{-msse2}.
3058 The @code{target} attribute is not implemented in GCC versions earlier
3059 than 4.4, and at present only the 386 uses it.
3062 @cindex tiny data section on the H8/300H and H8S
3063 Use this attribute on the H8/300H and H8S to indicate that the specified
3064 variable should be placed into the tiny data section.
3065 The compiler will generate more efficient code for loads and stores
3066 on data in the tiny data section. Note the tiny data area is limited to
3067 slightly under 32kbytes of data.
3070 Use this attribute on the SH for an @code{interrupt_handler} to return using
3071 @code{trapa} instead of @code{rte}. This attribute expects an integer
3072 argument specifying the trap number to be used.
3075 @cindex @code{unused} attribute.
3076 This attribute, attached to a function, means that the function is meant
3077 to be possibly unused. GCC will not produce a warning for this
3081 @cindex @code{used} attribute.
3082 This attribute, attached to a function, means that code must be emitted
3083 for the function even if it appears that the function is not referenced.
3084 This is useful, for example, when the function is referenced only in
3088 @cindex @code{version_id} attribute on IA64 HP-UX
3089 This attribute, attached to a global variable or function, renames a
3090 symbol to contain a version string, thus allowing for function level
3091 versioning. HP-UX system header files may use version level functioning
3092 for some system calls.
3095 extern int foo () __attribute__((version_id ("20040821")));
3098 Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}.
3100 @item visibility ("@var{visibility_type}")
3101 @cindex @code{visibility} attribute
3102 This attribute affects the linkage of the declaration to which it is attached.
3103 There are four supported @var{visibility_type} values: default,
3104 hidden, protected or internal visibility.
3107 void __attribute__ ((visibility ("protected")))
3108 f () @{ /* @r{Do something.} */; @}
3109 int i __attribute__ ((visibility ("hidden")));
3112 The possible values of @var{visibility_type} correspond to the
3113 visibility settings in the ELF gABI.
3116 @c keep this list of visibilities in alphabetical order.
3119 Default visibility is the normal case for the object file format.
3120 This value is available for the visibility attribute to override other
3121 options that may change the assumed visibility of entities.
3123 On ELF, default visibility means that the declaration is visible to other
3124 modules and, in shared libraries, means that the declared entity may be
3127 On Darwin, default visibility means that the declaration is visible to
3130 Default visibility corresponds to ``external linkage'' in the language.
3133 Hidden visibility indicates that the entity declared will have a new
3134 form of linkage, which we'll call ``hidden linkage''. Two
3135 declarations of an object with hidden linkage refer to the same object
3136 if they are in the same shared object.
3139 Internal visibility is like hidden visibility, but with additional
3140 processor specific semantics. Unless otherwise specified by the
3141 psABI, GCC defines internal visibility to mean that a function is
3142 @emph{never} called from another module. Compare this with hidden
3143 functions which, while they cannot be referenced directly by other
3144 modules, can be referenced indirectly via function pointers. By
3145 indicating that a function cannot be called from outside the module,
3146 GCC may for instance omit the load of a PIC register since it is known
3147 that the calling function loaded the correct value.
3150 Protected visibility is like default visibility except that it
3151 indicates that references within the defining module will bind to the
3152 definition in that module. That is, the declared entity cannot be
3153 overridden by another module.
3157 All visibilities are supported on many, but not all, ELF targets
3158 (supported when the assembler supports the @samp{.visibility}
3159 pseudo-op). Default visibility is supported everywhere. Hidden
3160 visibility is supported on Darwin targets.
3162 The visibility attribute should be applied only to declarations which
3163 would otherwise have external linkage. The attribute should be applied
3164 consistently, so that the same entity should not be declared with
3165 different settings of the attribute.
3167 In C++, the visibility attribute applies to types as well as functions
3168 and objects, because in C++ types have linkage. A class must not have
3169 greater visibility than its non-static data member types and bases,
3170 and class members default to the visibility of their class. Also, a
3171 declaration without explicit visibility is limited to the visibility
3174 In C++, you can mark member functions and static member variables of a
3175 class with the visibility attribute. This is useful if you know a
3176 particular method or static member variable should only be used from
3177 one shared object; then you can mark it hidden while the rest of the
3178 class has default visibility. Care must be taken to avoid breaking
3179 the One Definition Rule; for example, it is usually not useful to mark
3180 an inline method as hidden without marking the whole class as hidden.
3182 A C++ namespace declaration can also have the visibility attribute.
3183 This attribute applies only to the particular namespace body, not to
3184 other definitions of the same namespace; it is equivalent to using
3185 @samp{#pragma GCC visibility} before and after the namespace
3186 definition (@pxref{Visibility Pragmas}).
3188 In C++, if a template argument has limited visibility, this
3189 restriction is implicitly propagated to the template instantiation.
3190 Otherwise, template instantiations and specializations default to the
3191 visibility of their template.
3193 If both the template and enclosing class have explicit visibility, the
3194 visibility from the template is used.
3196 @item warn_unused_result
3197 @cindex @code{warn_unused_result} attribute
3198 The @code{warn_unused_result} attribute causes a warning to be emitted
3199 if a caller of the function with this attribute does not use its
3200 return value. This is useful for functions where not checking
3201 the result is either a security problem or always a bug, such as
3205 int fn () __attribute__ ((warn_unused_result));
3208 if (fn () < 0) return -1;
3214 results in warning on line 5.
3217 @cindex @code{weak} attribute
3218 The @code{weak} attribute causes the declaration to be emitted as a weak
3219 symbol rather than a global. This is primarily useful in defining
3220 library functions which can be overridden in user code, though it can
3221 also be used with non-function declarations. Weak symbols are supported
3222 for ELF targets, and also for a.out targets when using the GNU assembler
3226 @itemx weakref ("@var{target}")
3227 @cindex @code{weakref} attribute
3228 The @code{weakref} attribute marks a declaration as a weak reference.
3229 Without arguments, it should be accompanied by an @code{alias} attribute
3230 naming the target symbol. Optionally, the @var{target} may be given as
3231 an argument to @code{weakref} itself. In either case, @code{weakref}
3232 implicitly marks the declaration as @code{weak}. Without a
3233 @var{target}, given as an argument to @code{weakref} or to @code{alias},
3234 @code{weakref} is equivalent to @code{weak}.
3237 static int x() __attribute__ ((weakref ("y")));
3238 /* is equivalent to... */
3239 static int x() __attribute__ ((weak, weakref, alias ("y")));
3241 static int x() __attribute__ ((weakref));
3242 static int x() __attribute__ ((alias ("y")));
3245 A weak reference is an alias that does not by itself require a
3246 definition to be given for the target symbol. If the target symbol is
3247 only referenced through weak references, then the becomes a @code{weak}
3248 undefined symbol. If it is directly referenced, however, then such
3249 strong references prevail, and a definition will be required for the
3250 symbol, not necessarily in the same translation unit.
3252 The effect is equivalent to moving all references to the alias to a
3253 separate translation unit, renaming the alias to the aliased symbol,
3254 declaring it as weak, compiling the two separate translation units and
3255 performing a reloadable link on them.
3257 At present, a declaration to which @code{weakref} is attached can
3258 only be @code{static}.
3262 You can specify multiple attributes in a declaration by separating them
3263 by commas within the double parentheses or by immediately following an
3264 attribute declaration with another attribute declaration.
3266 @cindex @code{#pragma}, reason for not using
3267 @cindex pragma, reason for not using
3268 Some people object to the @code{__attribute__} feature, suggesting that
3269 ISO C's @code{#pragma} should be used instead. At the time
3270 @code{__attribute__} was designed, there were two reasons for not doing
3275 It is impossible to generate @code{#pragma} commands from a macro.
3278 There is no telling what the same @code{#pragma} might mean in another
3282 These two reasons applied to almost any application that might have been
3283 proposed for @code{#pragma}. It was basically a mistake to use
3284 @code{#pragma} for @emph{anything}.
3286 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
3287 to be generated from macros. In addition, a @code{#pragma GCC}
3288 namespace is now in use for GCC-specific pragmas. However, it has been
3289 found convenient to use @code{__attribute__} to achieve a natural
3290 attachment of attributes to their corresponding declarations, whereas
3291 @code{#pragma GCC} is of use for constructs that do not naturally form
3292 part of the grammar. @xref{Other Directives,,Miscellaneous
3293 Preprocessing Directives, cpp, The GNU C Preprocessor}.
3295 @node Attribute Syntax
3296 @section Attribute Syntax
3297 @cindex attribute syntax
3299 This section describes the syntax with which @code{__attribute__} may be
3300 used, and the constructs to which attribute specifiers bind, for the C
3301 language. Some details may vary for C++ and Objective-C@. Because of
3302 infelicities in the grammar for attributes, some forms described here
3303 may not be successfully parsed in all cases.
3305 There are some problems with the semantics of attributes in C++. For
3306 example, there are no manglings for attributes, although they may affect
3307 code generation, so problems may arise when attributed types are used in
3308 conjunction with templates or overloading. Similarly, @code{typeid}
3309 does not distinguish between types with different attributes. Support
3310 for attributes in C++ may be restricted in future to attributes on
3311 declarations only, but not on nested declarators.
3313 @xref{Function Attributes}, for details of the semantics of attributes
3314 applying to functions. @xref{Variable Attributes}, for details of the
3315 semantics of attributes applying to variables. @xref{Type Attributes},
3316 for details of the semantics of attributes applying to structure, union
3317 and enumerated types.
3319 An @dfn{attribute specifier} is of the form
3320 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
3321 is a possibly empty comma-separated sequence of @dfn{attributes}, where
3322 each attribute is one of the following:
3326 Empty. Empty attributes are ignored.
3329 A word (which may be an identifier such as @code{unused}, or a reserved
3330 word such as @code{const}).
3333 A word, followed by, in parentheses, parameters for the attribute.
3334 These parameters take one of the following forms:
3338 An identifier. For example, @code{mode} attributes use this form.
3341 An identifier followed by a comma and a non-empty comma-separated list
3342 of expressions. For example, @code{format} attributes use this form.
3345 A possibly empty comma-separated list of expressions. For example,
3346 @code{format_arg} attributes use this form with the list being a single
3347 integer constant expression, and @code{alias} attributes use this form
3348 with the list being a single string constant.
3352 An @dfn{attribute specifier list} is a sequence of one or more attribute
3353 specifiers, not separated by any other tokens.
3355 In GNU C, an attribute specifier list may appear after the colon following a
3356 label, other than a @code{case} or @code{default} label. The only
3357 attribute it makes sense to use after a label is @code{unused}. This
3358 feature is intended for code generated by programs which contains labels
3359 that may be unused but which is compiled with @option{-Wall}. It would
3360 not normally be appropriate to use in it human-written code, though it
3361 could be useful in cases where the code that jumps to the label is
3362 contained within an @code{#ifdef} conditional. GNU C++ does not permit
3363 such placement of attribute lists, as it is permissible for a
3364 declaration, which could begin with an attribute list, to be labelled in
3365 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
3366 does not arise there.
3368 An attribute specifier list may appear as part of a @code{struct},
3369 @code{union} or @code{enum} specifier. It may go either immediately
3370 after the @code{struct}, @code{union} or @code{enum} keyword, or after
3371 the closing brace. The former syntax is preferred.
3372 Where attribute specifiers follow the closing brace, they are considered
3373 to relate to the structure, union or enumerated type defined, not to any
3374 enclosing declaration the type specifier appears in, and the type
3375 defined is not complete until after the attribute specifiers.
3376 @c Otherwise, there would be the following problems: a shift/reduce
3377 @c conflict between attributes binding the struct/union/enum and
3378 @c binding to the list of specifiers/qualifiers; and "aligned"
3379 @c attributes could use sizeof for the structure, but the size could be
3380 @c changed later by "packed" attributes.
3382 Otherwise, an attribute specifier appears as part of a declaration,
3383 counting declarations of unnamed parameters and type names, and relates
3384 to that declaration (which may be nested in another declaration, for
3385 example in the case of a parameter declaration), or to a particular declarator
3386 within a declaration. Where an
3387 attribute specifier is applied to a parameter declared as a function or
3388 an array, it should apply to the function or array rather than the
3389 pointer to which the parameter is implicitly converted, but this is not
3390 yet correctly implemented.
3392 Any list of specifiers and qualifiers at the start of a declaration may
3393 contain attribute specifiers, whether or not such a list may in that
3394 context contain storage class specifiers. (Some attributes, however,
3395 are essentially in the nature of storage class specifiers, and only make
3396 sense where storage class specifiers may be used; for example,
3397 @code{section}.) There is one necessary limitation to this syntax: the
3398 first old-style parameter declaration in a function definition cannot
3399 begin with an attribute specifier, because such an attribute applies to
3400 the function instead by syntax described below (which, however, is not
3401 yet implemented in this case). In some other cases, attribute
3402 specifiers are permitted by this grammar but not yet supported by the
3403 compiler. All attribute specifiers in this place relate to the
3404 declaration as a whole. In the obsolescent usage where a type of
3405 @code{int} is implied by the absence of type specifiers, such a list of
3406 specifiers and qualifiers may be an attribute specifier list with no
3407 other specifiers or qualifiers.
3409 At present, the first parameter in a function prototype must have some
3410 type specifier which is not an attribute specifier; this resolves an
3411 ambiguity in the interpretation of @code{void f(int
3412 (__attribute__((foo)) x))}, but is subject to change. At present, if
3413 the parentheses of a function declarator contain only attributes then
3414 those attributes are ignored, rather than yielding an error or warning
3415 or implying a single parameter of type int, but this is subject to
3418 An attribute specifier list may appear immediately before a declarator
3419 (other than the first) in a comma-separated list of declarators in a
3420 declaration of more than one identifier using a single list of
3421 specifiers and qualifiers. Such attribute specifiers apply
3422 only to the identifier before whose declarator they appear. For
3426 __attribute__((noreturn)) void d0 (void),
3427 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
3432 the @code{noreturn} attribute applies to all the functions
3433 declared; the @code{format} attribute only applies to @code{d1}.
3435 An attribute specifier list may appear immediately before the comma,
3436 @code{=} or semicolon terminating the declaration of an identifier other
3437 than a function definition. Such attribute specifiers apply
3438 to the declared object or function. Where an
3439 assembler name for an object or function is specified (@pxref{Asm
3440 Labels}), the attribute must follow the @code{asm}
3443 An attribute specifier list may, in future, be permitted to appear after
3444 the declarator in a function definition (before any old-style parameter
3445 declarations or the function body).
3447 Attribute specifiers may be mixed with type qualifiers appearing inside
3448 the @code{[]} of a parameter array declarator, in the C99 construct by
3449 which such qualifiers are applied to the pointer to which the array is
3450 implicitly converted. Such attribute specifiers apply to the pointer,
3451 not to the array, but at present this is not implemented and they are
3454 An attribute specifier list may appear at the start of a nested
3455 declarator. At present, there are some limitations in this usage: the
3456 attributes correctly apply to the declarator, but for most individual
3457 attributes the semantics this implies are not implemented.
3458 When attribute specifiers follow the @code{*} of a pointer
3459 declarator, they may be mixed with any type qualifiers present.
3460 The following describes the formal semantics of this syntax. It will make the
3461 most sense if you are familiar with the formal specification of
3462 declarators in the ISO C standard.
3464 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
3465 D1}, where @code{T} contains declaration specifiers that specify a type
3466 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
3467 contains an identifier @var{ident}. The type specified for @var{ident}
3468 for derived declarators whose type does not include an attribute
3469 specifier is as in the ISO C standard.
3471 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
3472 and the declaration @code{T D} specifies the type
3473 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3474 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3475 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
3477 If @code{D1} has the form @code{*
3478 @var{type-qualifier-and-attribute-specifier-list} D}, and the
3479 declaration @code{T D} specifies the type
3480 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3481 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3482 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
3488 void (__attribute__((noreturn)) ****f) (void);
3492 specifies the type ``pointer to pointer to pointer to pointer to
3493 non-returning function returning @code{void}''. As another example,
3496 char *__attribute__((aligned(8))) *f;
3500 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
3501 Note again that this does not work with most attributes; for example,
3502 the usage of @samp{aligned} and @samp{noreturn} attributes given above
3503 is not yet supported.
3505 For compatibility with existing code written for compiler versions that
3506 did not implement attributes on nested declarators, some laxity is
3507 allowed in the placing of attributes. If an attribute that only applies
3508 to types is applied to a declaration, it will be treated as applying to
3509 the type of that declaration. If an attribute that only applies to
3510 declarations is applied to the type of a declaration, it will be treated
3511 as applying to that declaration; and, for compatibility with code
3512 placing the attributes immediately before the identifier declared, such
3513 an attribute applied to a function return type will be treated as
3514 applying to the function type, and such an attribute applied to an array
3515 element type will be treated as applying to the array type. If an
3516 attribute that only applies to function types is applied to a
3517 pointer-to-function type, it will be treated as applying to the pointer
3518 target type; if such an attribute is applied to a function return type
3519 that is not a pointer-to-function type, it will be treated as applying
3520 to the function type.
3522 @node Function Prototypes
3523 @section Prototypes and Old-Style Function Definitions
3524 @cindex function prototype declarations
3525 @cindex old-style function definitions
3526 @cindex promotion of formal parameters
3528 GNU C extends ISO C to allow a function prototype to override a later
3529 old-style non-prototype definition. Consider the following example:
3532 /* @r{Use prototypes unless the compiler is old-fashioned.} */
3539 /* @r{Prototype function declaration.} */
3540 int isroot P((uid_t));
3542 /* @r{Old-style function definition.} */
3544 isroot (x) /* @r{??? lossage here ???} */
3551 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
3552 not allow this example, because subword arguments in old-style
3553 non-prototype definitions are promoted. Therefore in this example the
3554 function definition's argument is really an @code{int}, which does not
3555 match the prototype argument type of @code{short}.
3557 This restriction of ISO C makes it hard to write code that is portable
3558 to traditional C compilers, because the programmer does not know
3559 whether the @code{uid_t} type is @code{short}, @code{int}, or
3560 @code{long}. Therefore, in cases like these GNU C allows a prototype
3561 to override a later old-style definition. More precisely, in GNU C, a
3562 function prototype argument type overrides the argument type specified
3563 by a later old-style definition if the former type is the same as the
3564 latter type before promotion. Thus in GNU C the above example is
3565 equivalent to the following:
3578 GNU C++ does not support old-style function definitions, so this
3579 extension is irrelevant.
3582 @section C++ Style Comments
3584 @cindex C++ comments
3585 @cindex comments, C++ style
3587 In GNU C, you may use C++ style comments, which start with @samp{//} and
3588 continue until the end of the line. Many other C implementations allow
3589 such comments, and they are included in the 1999 C standard. However,
3590 C++ style comments are not recognized if you specify an @option{-std}
3591 option specifying a version of ISO C before C99, or @option{-ansi}
3592 (equivalent to @option{-std=c89}).
3595 @section Dollar Signs in Identifier Names
3597 @cindex dollar signs in identifier names
3598 @cindex identifier names, dollar signs in
3600 In GNU C, you may normally use dollar signs in identifier names.
3601 This is because many traditional C implementations allow such identifiers.
3602 However, dollar signs in identifiers are not supported on a few target
3603 machines, typically because the target assembler does not allow them.
3605 @node Character Escapes
3606 @section The Character @key{ESC} in Constants
3608 You can use the sequence @samp{\e} in a string or character constant to
3609 stand for the ASCII character @key{ESC}.
3612 @section Inquiring on Alignment of Types or Variables
3614 @cindex type alignment
3615 @cindex variable alignment
3617 The keyword @code{__alignof__} allows you to inquire about how an object
3618 is aligned, or the minimum alignment usually required by a type. Its
3619 syntax is just like @code{sizeof}.
3621 For example, if the target machine requires a @code{double} value to be
3622 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
3623 This is true on many RISC machines. On more traditional machine
3624 designs, @code{__alignof__ (double)} is 4 or even 2.
3626 Some machines never actually require alignment; they allow reference to any
3627 data type even at an odd address. For these machines, @code{__alignof__}
3628 reports the smallest alignment that GCC will give the data type, usually as
3629 mandated by the target ABI.
3631 If the operand of @code{__alignof__} is an lvalue rather than a type,
3632 its value is the required alignment for its type, taking into account
3633 any minimum alignment specified with GCC's @code{__attribute__}
3634 extension (@pxref{Variable Attributes}). For example, after this
3638 struct foo @{ int x; char y; @} foo1;
3642 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
3643 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
3645 It is an error to ask for the alignment of an incomplete type.
3647 @node Variable Attributes
3648 @section Specifying Attributes of Variables
3649 @cindex attribute of variables
3650 @cindex variable attributes
3652 The keyword @code{__attribute__} allows you to specify special
3653 attributes of variables or structure fields. This keyword is followed
3654 by an attribute specification inside double parentheses. Some
3655 attributes are currently defined generically for variables.
3656 Other attributes are defined for variables on particular target
3657 systems. Other attributes are available for functions
3658 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
3659 Other front ends might define more attributes
3660 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
3662 You may also specify attributes with @samp{__} preceding and following
3663 each keyword. This allows you to use them in header files without
3664 being concerned about a possible macro of the same name. For example,
3665 you may use @code{__aligned__} instead of @code{aligned}.
3667 @xref{Attribute Syntax}, for details of the exact syntax for using
3671 @cindex @code{aligned} attribute
3672 @item aligned (@var{alignment})
3673 This attribute specifies a minimum alignment for the variable or
3674 structure field, measured in bytes. For example, the declaration:
3677 int x __attribute__ ((aligned (16))) = 0;
3681 causes the compiler to allocate the global variable @code{x} on a
3682 16-byte boundary. On a 68040, this could be used in conjunction with
3683 an @code{asm} expression to access the @code{move16} instruction which
3684 requires 16-byte aligned operands.
3686 You can also specify the alignment of structure fields. For example, to
3687 create a double-word aligned @code{int} pair, you could write:
3690 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
3694 This is an alternative to creating a union with a @code{double} member
3695 that forces the union to be double-word aligned.
3697 As in the preceding examples, you can explicitly specify the alignment
3698 (in bytes) that you wish the compiler to use for a given variable or
3699 structure field. Alternatively, you can leave out the alignment factor
3700 and just ask the compiler to align a variable or field to the maximum
3701 useful alignment for the target machine you are compiling for. For
3702 example, you could write:
3705 short array[3] __attribute__ ((aligned));
3708 Whenever you leave out the alignment factor in an @code{aligned} attribute
3709 specification, the compiler automatically sets the alignment for the declared
3710 variable or field to the largest alignment which is ever used for any data
3711 type on the target machine you are compiling for. Doing this can often make
3712 copy operations more efficient, because the compiler can use whatever
3713 instructions copy the biggest chunks of memory when performing copies to
3714 or from the variables or fields that you have aligned this way.
3716 When used on a struct, or struct member, the @code{aligned} attribute can
3717 only increase the alignment; in order to decrease it, the @code{packed}
3718 attribute must be specified as well. When used as part of a typedef, the
3719 @code{aligned} attribute can both increase and decrease alignment, and
3720 specifying the @code{packed} attribute will generate a warning.
3722 Note that the effectiveness of @code{aligned} attributes may be limited
3723 by inherent limitations in your linker. On many systems, the linker is
3724 only able to arrange for variables to be aligned up to a certain maximum
3725 alignment. (For some linkers, the maximum supported alignment may
3726 be very very small.) If your linker is only able to align variables
3727 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3728 in an @code{__attribute__} will still only provide you with 8 byte
3729 alignment. See your linker documentation for further information.
3731 The @code{aligned} attribute can also be used for functions
3732 (@pxref{Function Attributes}.)
3734 @item cleanup (@var{cleanup_function})
3735 @cindex @code{cleanup} attribute
3736 The @code{cleanup} attribute runs a function when the variable goes
3737 out of scope. This attribute can only be applied to auto function
3738 scope variables; it may not be applied to parameters or variables
3739 with static storage duration. The function must take one parameter,
3740 a pointer to a type compatible with the variable. The return value
3741 of the function (if any) is ignored.
3743 If @option{-fexceptions} is enabled, then @var{cleanup_function}
3744 will be run during the stack unwinding that happens during the
3745 processing of the exception. Note that the @code{cleanup} attribute
3746 does not allow the exception to be caught, only to perform an action.
3747 It is undefined what happens if @var{cleanup_function} does not
3752 @cindex @code{common} attribute
3753 @cindex @code{nocommon} attribute
3756 The @code{common} attribute requests GCC to place a variable in
3757 ``common'' storage. The @code{nocommon} attribute requests the
3758 opposite---to allocate space for it directly.
3760 These attributes override the default chosen by the
3761 @option{-fno-common} and @option{-fcommon} flags respectively.
3764 @cindex @code{deprecated} attribute
3765 The @code{deprecated} attribute results in a warning if the variable
3766 is used anywhere in the source file. This is useful when identifying
3767 variables that are expected to be removed in a future version of a
3768 program. The warning also includes the location of the declaration
3769 of the deprecated variable, to enable users to easily find further
3770 information about why the variable is deprecated, or what they should
3771 do instead. Note that the warning only occurs for uses:
3774 extern int old_var __attribute__ ((deprecated));
3776 int new_fn () @{ return old_var; @}
3779 results in a warning on line 3 but not line 2.
3781 The @code{deprecated} attribute can also be used for functions and
3782 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3784 @item mode (@var{mode})
3785 @cindex @code{mode} attribute
3786 This attribute specifies the data type for the declaration---whichever
3787 type corresponds to the mode @var{mode}. This in effect lets you
3788 request an integer or floating point type according to its width.
3790 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3791 indicate the mode corresponding to a one-byte integer, @samp{word} or
3792 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3793 or @samp{__pointer__} for the mode used to represent pointers.
3796 @cindex @code{packed} attribute
3797 The @code{packed} attribute specifies that a variable or structure field
3798 should have the smallest possible alignment---one byte for a variable,
3799 and one bit for a field, unless you specify a larger value with the
3800 @code{aligned} attribute.
3802 Here is a structure in which the field @code{x} is packed, so that it
3803 immediately follows @code{a}:
3809 int x[2] __attribute__ ((packed));
3813 @item section ("@var{section-name}")
3814 @cindex @code{section} variable attribute
3815 Normally, the compiler places the objects it generates in sections like
3816 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3817 or you need certain particular variables to appear in special sections,
3818 for example to map to special hardware. The @code{section}
3819 attribute specifies that a variable (or function) lives in a particular
3820 section. For example, this small program uses several specific section names:
3823 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3824 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3825 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3826 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3830 /* @r{Initialize stack pointer} */
3831 init_sp (stack + sizeof (stack));
3833 /* @r{Initialize initialized data} */
3834 memcpy (&init_data, &data, &edata - &data);
3836 /* @r{Turn on the serial ports} */
3843 Use the @code{section} attribute with an @emph{initialized} definition
3844 of a @emph{global} variable, as shown in the example. GCC issues
3845 a warning and otherwise ignores the @code{section} attribute in
3846 uninitialized variable declarations.
3848 You may only use the @code{section} attribute with a fully initialized
3849 global definition because of the way linkers work. The linker requires
3850 each object be defined once, with the exception that uninitialized
3851 variables tentatively go in the @code{common} (or @code{bss}) section
3852 and can be multiply ``defined''. You can force a variable to be
3853 initialized with the @option{-fno-common} flag or the @code{nocommon}
3856 Some file formats do not support arbitrary sections so the @code{section}
3857 attribute is not available on all platforms.
3858 If you need to map the entire contents of a module to a particular
3859 section, consider using the facilities of the linker instead.
3862 @cindex @code{shared} variable attribute
3863 On Microsoft Windows, in addition to putting variable definitions in a named
3864 section, the section can also be shared among all running copies of an
3865 executable or DLL@. For example, this small program defines shared data
3866 by putting it in a named section @code{shared} and marking the section
3870 int foo __attribute__((section ("shared"), shared)) = 0;
3875 /* @r{Read and write foo. All running
3876 copies see the same value.} */
3882 You may only use the @code{shared} attribute along with @code{section}
3883 attribute with a fully initialized global definition because of the way
3884 linkers work. See @code{section} attribute for more information.
3886 The @code{shared} attribute is only available on Microsoft Windows@.
3888 @item tls_model ("@var{tls_model}")
3889 @cindex @code{tls_model} attribute
3890 The @code{tls_model} attribute sets thread-local storage model
3891 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3892 overriding @option{-ftls-model=} command line switch on a per-variable
3894 The @var{tls_model} argument should be one of @code{global-dynamic},
3895 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3897 Not all targets support this attribute.
3900 This attribute, attached to a variable, means that the variable is meant
3901 to be possibly unused. GCC will not produce a warning for this
3905 This attribute, attached to a variable, means that the variable must be
3906 emitted even if it appears that the variable is not referenced.
3908 @item vector_size (@var{bytes})
3909 This attribute specifies the vector size for the variable, measured in
3910 bytes. For example, the declaration:
3913 int foo __attribute__ ((vector_size (16)));
3917 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3918 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3919 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3921 This attribute is only applicable to integral and float scalars,
3922 although arrays, pointers, and function return values are allowed in
3923 conjunction with this construct.
3925 Aggregates with this attribute are invalid, even if they are of the same
3926 size as a corresponding scalar. For example, the declaration:
3929 struct S @{ int a; @};
3930 struct S __attribute__ ((vector_size (16))) foo;
3934 is invalid even if the size of the structure is the same as the size of
3938 The @code{selectany} attribute causes an initialized global variable to
3939 have link-once semantics. When multiple definitions of the variable are
3940 encountered by the linker, the first is selected and the remainder are
3941 discarded. Following usage by the Microsoft compiler, the linker is told
3942 @emph{not} to warn about size or content differences of the multiple
3945 Although the primary usage of this attribute is for POD types, the
3946 attribute can also be applied to global C++ objects that are initialized
3947 by a constructor. In this case, the static initialization and destruction
3948 code for the object is emitted in each translation defining the object,
3949 but the calls to the constructor and destructor are protected by a
3950 link-once guard variable.
3952 The @code{selectany} attribute is only available on Microsoft Windows
3953 targets. You can use @code{__declspec (selectany)} as a synonym for
3954 @code{__attribute__ ((selectany))} for compatibility with other
3958 The @code{weak} attribute is described in @ref{Function Attributes}.
3961 The @code{dllimport} attribute is described in @ref{Function Attributes}.
3964 The @code{dllexport} attribute is described in @ref{Function Attributes}.
3968 @subsection Blackfin Variable Attributes
3970 Three attributes are currently defined for the Blackfin.
3976 @cindex @code{l1_data} variable attribute
3977 @cindex @code{l1_data_A} variable attribute
3978 @cindex @code{l1_data_B} variable attribute
3979 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
3980 Variables with @code{l1_data} attribute will be put into the specific section
3981 named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into
3982 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
3983 attribute will be put into the specific section named @code{.l1.data.B}.
3986 @subsection M32R/D Variable Attributes
3988 One attribute is currently defined for the M32R/D@.
3991 @item model (@var{model-name})
3992 @cindex variable addressability on the M32R/D
3993 Use this attribute on the M32R/D to set the addressability of an object.
3994 The identifier @var{model-name} is one of @code{small}, @code{medium},
3995 or @code{large}, representing each of the code models.
3997 Small model objects live in the lower 16MB of memory (so that their
3998 addresses can be loaded with the @code{ld24} instruction).
4000 Medium and large model objects may live anywhere in the 32-bit address space
4001 (the compiler will generate @code{seth/add3} instructions to load their
4005 @anchor{i386 Variable Attributes}
4006 @subsection i386 Variable Attributes
4008 Two attributes are currently defined for i386 configurations:
4009 @code{ms_struct} and @code{gcc_struct}
4014 @cindex @code{ms_struct} attribute
4015 @cindex @code{gcc_struct} attribute
4017 If @code{packed} is used on a structure, or if bit-fields are used
4018 it may be that the Microsoft ABI packs them differently
4019 than GCC would normally pack them. Particularly when moving packed
4020 data between functions compiled with GCC and the native Microsoft compiler
4021 (either via function call or as data in a file), it may be necessary to access
4024 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4025 compilers to match the native Microsoft compiler.
4027 The Microsoft structure layout algorithm is fairly simple with the exception
4028 of the bitfield packing:
4030 The padding and alignment of members of structures and whether a bit field
4031 can straddle a storage-unit boundary
4034 @item Structure members are stored sequentially in the order in which they are
4035 declared: the first member has the lowest memory address and the last member
4038 @item Every data object has an alignment-requirement. The alignment-requirement
4039 for all data except structures, unions, and arrays is either the size of the
4040 object or the current packing size (specified with either the aligned attribute
4041 or the pack pragma), whichever is less. For structures, unions, and arrays,
4042 the alignment-requirement is the largest alignment-requirement of its members.
4043 Every object is allocated an offset so that:
4045 offset % alignment-requirement == 0
4047 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
4048 unit if the integral types are the same size and if the next bit field fits
4049 into the current allocation unit without crossing the boundary imposed by the
4050 common alignment requirements of the bit fields.
4053 Handling of zero-length bitfields:
4055 MSVC interprets zero-length bitfields in the following ways:
4058 @item If a zero-length bitfield is inserted between two bitfields that would
4059 normally be coalesced, the bitfields will not be coalesced.
4066 unsigned long bf_1 : 12;
4068 unsigned long bf_2 : 12;
4072 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
4073 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
4075 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
4076 alignment of the zero-length bitfield is greater than the member that follows it,
4077 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
4097 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
4098 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
4099 bitfield will not affect the alignment of @code{bar} or, as a result, the size
4102 Taking this into account, it is important to note the following:
4105 @item If a zero-length bitfield follows a normal bitfield, the type of the
4106 zero-length bitfield may affect the alignment of the structure as whole. For
4107 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
4108 normal bitfield, and is of type short.
4110 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
4111 still affect the alignment of the structure:
4121 Here, @code{t4} will take up 4 bytes.
4124 @item Zero-length bitfields following non-bitfield members are ignored:
4135 Here, @code{t5} will take up 2 bytes.
4139 @subsection PowerPC Variable Attributes
4141 Three attributes currently are defined for PowerPC configurations:
4142 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4144 For full documentation of the struct attributes please see the
4145 documentation in @ref{i386 Variable Attributes}.
4147 For documentation of @code{altivec} attribute please see the
4148 documentation in @ref{PowerPC Type Attributes}.
4150 @subsection SPU Variable Attributes
4152 The SPU supports the @code{spu_vector} attribute for variables. For
4153 documentation of this attribute please see the documentation in
4154 @ref{SPU Type Attributes}.
4156 @subsection Xstormy16 Variable Attributes
4158 One attribute is currently defined for xstormy16 configurations:
4163 @cindex @code{below100} attribute
4165 If a variable has the @code{below100} attribute (@code{BELOW100} is
4166 allowed also), GCC will place the variable in the first 0x100 bytes of
4167 memory and use special opcodes to access it. Such variables will be
4168 placed in either the @code{.bss_below100} section or the
4169 @code{.data_below100} section.
4173 @subsection AVR Variable Attributes
4177 @cindex @code{progmem} variable attribute
4178 The @code{progmem} attribute is used on the AVR to place data in the Program
4179 Memory address space. The AVR is a Harvard Architecture processor and data
4180 normally resides in the Data Memory address space.
4183 @node Type Attributes
4184 @section Specifying Attributes of Types
4185 @cindex attribute of types
4186 @cindex type attributes
4188 The keyword @code{__attribute__} allows you to specify special
4189 attributes of @code{struct} and @code{union} types when you define
4190 such types. This keyword is followed by an attribute specification
4191 inside double parentheses. Seven attributes are currently defined for
4192 types: @code{aligned}, @code{packed}, @code{transparent_union},
4193 @code{unused}, @code{deprecated}, @code{visibility}, and
4194 @code{may_alias}. Other attributes are defined for functions
4195 (@pxref{Function Attributes}) and for variables (@pxref{Variable
4198 You may also specify any one of these attributes with @samp{__}
4199 preceding and following its keyword. This allows you to use these
4200 attributes in header files without being concerned about a possible
4201 macro of the same name. For example, you may use @code{__aligned__}
4202 instead of @code{aligned}.
4204 You may specify type attributes in an enum, struct or union type
4205 declaration or definition, or for other types in a @code{typedef}
4208 For an enum, struct or union type, you may specify attributes either
4209 between the enum, struct or union tag and the name of the type, or
4210 just past the closing curly brace of the @emph{definition}. The
4211 former syntax is preferred.
4213 @xref{Attribute Syntax}, for details of the exact syntax for using
4217 @cindex @code{aligned} attribute
4218 @item aligned (@var{alignment})
4219 This attribute specifies a minimum alignment (in bytes) for variables
4220 of the specified type. For example, the declarations:
4223 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
4224 typedef int more_aligned_int __attribute__ ((aligned (8)));
4228 force the compiler to insure (as far as it can) that each variable whose
4229 type is @code{struct S} or @code{more_aligned_int} will be allocated and
4230 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
4231 variables of type @code{struct S} aligned to 8-byte boundaries allows
4232 the compiler to use the @code{ldd} and @code{std} (doubleword load and
4233 store) instructions when copying one variable of type @code{struct S} to
4234 another, thus improving run-time efficiency.
4236 Note that the alignment of any given @code{struct} or @code{union} type
4237 is required by the ISO C standard to be at least a perfect multiple of
4238 the lowest common multiple of the alignments of all of the members of
4239 the @code{struct} or @code{union} in question. This means that you @emph{can}
4240 effectively adjust the alignment of a @code{struct} or @code{union}
4241 type by attaching an @code{aligned} attribute to any one of the members
4242 of such a type, but the notation illustrated in the example above is a
4243 more obvious, intuitive, and readable way to request the compiler to
4244 adjust the alignment of an entire @code{struct} or @code{union} type.
4246 As in the preceding example, you can explicitly specify the alignment
4247 (in bytes) that you wish the compiler to use for a given @code{struct}
4248 or @code{union} type. Alternatively, you can leave out the alignment factor
4249 and just ask the compiler to align a type to the maximum
4250 useful alignment for the target machine you are compiling for. For
4251 example, you could write:
4254 struct S @{ short f[3]; @} __attribute__ ((aligned));
4257 Whenever you leave out the alignment factor in an @code{aligned}
4258 attribute specification, the compiler automatically sets the alignment
4259 for the type to the largest alignment which is ever used for any data
4260 type on the target machine you are compiling for. Doing this can often
4261 make copy operations more efficient, because the compiler can use
4262 whatever instructions copy the biggest chunks of memory when performing
4263 copies to or from the variables which have types that you have aligned
4266 In the example above, if the size of each @code{short} is 2 bytes, then
4267 the size of the entire @code{struct S} type is 6 bytes. The smallest
4268 power of two which is greater than or equal to that is 8, so the
4269 compiler sets the alignment for the entire @code{struct S} type to 8
4272 Note that although you can ask the compiler to select a time-efficient
4273 alignment for a given type and then declare only individual stand-alone
4274 objects of that type, the compiler's ability to select a time-efficient
4275 alignment is primarily useful only when you plan to create arrays of
4276 variables having the relevant (efficiently aligned) type. If you
4277 declare or use arrays of variables of an efficiently-aligned type, then
4278 it is likely that your program will also be doing pointer arithmetic (or
4279 subscripting, which amounts to the same thing) on pointers to the
4280 relevant type, and the code that the compiler generates for these
4281 pointer arithmetic operations will often be more efficient for
4282 efficiently-aligned types than for other types.
4284 The @code{aligned} attribute can only increase the alignment; but you
4285 can decrease it by specifying @code{packed} as well. See below.
4287 Note that the effectiveness of @code{aligned} attributes may be limited
4288 by inherent limitations in your linker. On many systems, the linker is
4289 only able to arrange for variables to be aligned up to a certain maximum
4290 alignment. (For some linkers, the maximum supported alignment may
4291 be very very small.) If your linker is only able to align variables
4292 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4293 in an @code{__attribute__} will still only provide you with 8 byte
4294 alignment. See your linker documentation for further information.
4297 This attribute, attached to @code{struct} or @code{union} type
4298 definition, specifies that each member (other than zero-width bitfields)
4299 of the structure or union is placed to minimize the memory required. When
4300 attached to an @code{enum} definition, it indicates that the smallest
4301 integral type should be used.
4303 @opindex fshort-enums
4304 Specifying this attribute for @code{struct} and @code{union} types is
4305 equivalent to specifying the @code{packed} attribute on each of the
4306 structure or union members. Specifying the @option{-fshort-enums}
4307 flag on the line is equivalent to specifying the @code{packed}
4308 attribute on all @code{enum} definitions.
4310 In the following example @code{struct my_packed_struct}'s members are
4311 packed closely together, but the internal layout of its @code{s} member
4312 is not packed---to do that, @code{struct my_unpacked_struct} would need to
4316 struct my_unpacked_struct
4322 struct __attribute__ ((__packed__)) my_packed_struct
4326 struct my_unpacked_struct s;
4330 You may only specify this attribute on the definition of a @code{enum},
4331 @code{struct} or @code{union}, not on a @code{typedef} which does not
4332 also define the enumerated type, structure or union.
4334 @item transparent_union
4335 This attribute, attached to a @code{union} type definition, indicates
4336 that any function parameter having that union type causes calls to that
4337 function to be treated in a special way.
4339 First, the argument corresponding to a transparent union type can be of
4340 any type in the union; no cast is required. Also, if the union contains
4341 a pointer type, the corresponding argument can be a null pointer
4342 constant or a void pointer expression; and if the union contains a void
4343 pointer type, the corresponding argument can be any pointer expression.
4344 If the union member type is a pointer, qualifiers like @code{const} on
4345 the referenced type must be respected, just as with normal pointer
4348 Second, the argument is passed to the function using the calling
4349 conventions of the first member of the transparent union, not the calling
4350 conventions of the union itself. All members of the union must have the
4351 same machine representation; this is necessary for this argument passing
4354 Transparent unions are designed for library functions that have multiple
4355 interfaces for compatibility reasons. For example, suppose the
4356 @code{wait} function must accept either a value of type @code{int *} to
4357 comply with Posix, or a value of type @code{union wait *} to comply with
4358 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
4359 @code{wait} would accept both kinds of arguments, but it would also
4360 accept any other pointer type and this would make argument type checking
4361 less useful. Instead, @code{<sys/wait.h>} might define the interface
4365 typedef union __attribute__ ((__transparent_union__))
4369 @} wait_status_ptr_t;
4371 pid_t wait (wait_status_ptr_t);
4374 This interface allows either @code{int *} or @code{union wait *}
4375 arguments to be passed, using the @code{int *} calling convention.
4376 The program can call @code{wait} with arguments of either type:
4379 int w1 () @{ int w; return wait (&w); @}
4380 int w2 () @{ union wait w; return wait (&w); @}
4383 With this interface, @code{wait}'s implementation might look like this:
4386 pid_t wait (wait_status_ptr_t p)
4388 return waitpid (-1, p.__ip, 0);
4393 When attached to a type (including a @code{union} or a @code{struct}),
4394 this attribute means that variables of that type are meant to appear
4395 possibly unused. GCC will not produce a warning for any variables of
4396 that type, even if the variable appears to do nothing. This is often
4397 the case with lock or thread classes, which are usually defined and then
4398 not referenced, but contain constructors and destructors that have
4399 nontrivial bookkeeping functions.
4402 The @code{deprecated} attribute results in a warning if the type
4403 is used anywhere in the source file. This is useful when identifying
4404 types that are expected to be removed in a future version of a program.
4405 If possible, the warning also includes the location of the declaration
4406 of the deprecated type, to enable users to easily find further
4407 information about why the type is deprecated, or what they should do
4408 instead. Note that the warnings only occur for uses and then only
4409 if the type is being applied to an identifier that itself is not being
4410 declared as deprecated.
4413 typedef int T1 __attribute__ ((deprecated));
4417 typedef T1 T3 __attribute__ ((deprecated));
4418 T3 z __attribute__ ((deprecated));
4421 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
4422 warning is issued for line 4 because T2 is not explicitly
4423 deprecated. Line 5 has no warning because T3 is explicitly
4424 deprecated. Similarly for line 6.
4426 The @code{deprecated} attribute can also be used for functions and
4427 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
4430 Accesses through pointers to types with this attribute are not subject
4431 to type-based alias analysis, but are instead assumed to be able to alias
4432 any other type of objects. In the context of 6.5/7 an lvalue expression
4433 dereferencing such a pointer is treated like having a character type.
4434 See @option{-fstrict-aliasing} for more information on aliasing issues.
4435 This extension exists to support some vector APIs, in which pointers to
4436 one vector type are permitted to alias pointers to a different vector type.
4438 Note that an object of a type with this attribute does not have any
4444 typedef short __attribute__((__may_alias__)) short_a;
4450 short_a *b = (short_a *) &a;
4454 if (a == 0x12345678)
4461 If you replaced @code{short_a} with @code{short} in the variable
4462 declaration, the above program would abort when compiled with
4463 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
4464 above in recent GCC versions.
4467 In C++, attribute visibility (@pxref{Function Attributes}) can also be
4468 applied to class, struct, union and enum types. Unlike other type
4469 attributes, the attribute must appear between the initial keyword and
4470 the name of the type; it cannot appear after the body of the type.
4472 Note that the type visibility is applied to vague linkage entities
4473 associated with the class (vtable, typeinfo node, etc.). In
4474 particular, if a class is thrown as an exception in one shared object
4475 and caught in another, the class must have default visibility.
4476 Otherwise the two shared objects will be unable to use the same
4477 typeinfo node and exception handling will break.
4481 @subsection ARM Type Attributes
4483 On those ARM targets that support @code{dllimport} (such as Symbian
4484 OS), you can use the @code{notshared} attribute to indicate that the
4485 virtual table and other similar data for a class should not be
4486 exported from a DLL@. For example:
4489 class __declspec(notshared) C @{
4491 __declspec(dllimport) C();
4495 __declspec(dllexport)
4499 In this code, @code{C::C} is exported from the current DLL, but the
4500 virtual table for @code{C} is not exported. (You can use
4501 @code{__attribute__} instead of @code{__declspec} if you prefer, but
4502 most Symbian OS code uses @code{__declspec}.)
4504 @anchor{i386 Type Attributes}
4505 @subsection i386 Type Attributes
4507 Two attributes are currently defined for i386 configurations:
4508 @code{ms_struct} and @code{gcc_struct}.
4514 @cindex @code{ms_struct}
4515 @cindex @code{gcc_struct}
4517 If @code{packed} is used on a structure, or if bit-fields are used
4518 it may be that the Microsoft ABI packs them differently
4519 than GCC would normally pack them. Particularly when moving packed
4520 data between functions compiled with GCC and the native Microsoft compiler
4521 (either via function call or as data in a file), it may be necessary to access
4524 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4525 compilers to match the native Microsoft compiler.
4528 To specify multiple attributes, separate them by commas within the
4529 double parentheses: for example, @samp{__attribute__ ((aligned (16),
4532 @anchor{PowerPC Type Attributes}
4533 @subsection PowerPC Type Attributes
4535 Three attributes currently are defined for PowerPC configurations:
4536 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4538 For full documentation of the @code{ms_struct} and @code{gcc_struct}
4539 attributes please see the documentation in @ref{i386 Type Attributes}.
4541 The @code{altivec} attribute allows one to declare AltiVec vector data
4542 types supported by the AltiVec Programming Interface Manual. The
4543 attribute requires an argument to specify one of three vector types:
4544 @code{vector__}, @code{pixel__} (always followed by unsigned short),
4545 and @code{bool__} (always followed by unsigned).
4548 __attribute__((altivec(vector__)))
4549 __attribute__((altivec(pixel__))) unsigned short
4550 __attribute__((altivec(bool__))) unsigned
4553 These attributes mainly are intended to support the @code{__vector},
4554 @code{__pixel}, and @code{__bool} AltiVec keywords.
4556 @anchor{SPU Type Attributes}
4557 @subsection SPU Type Attributes
4559 The SPU supports the @code{spu_vector} attribute for types. This attribute
4560 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
4561 Language Extensions Specification. It is intended to support the
4562 @code{__vector} keyword.
4566 @section An Inline Function is As Fast As a Macro
4567 @cindex inline functions
4568 @cindex integrating function code
4570 @cindex macros, inline alternative
4572 By declaring a function inline, you can direct GCC to make
4573 calls to that function faster. One way GCC can achieve this is to
4574 integrate that function's code into the code for its callers. This
4575 makes execution faster by eliminating the function-call overhead; in
4576 addition, if any of the actual argument values are constant, their
4577 known values may permit simplifications at compile time so that not
4578 all of the inline function's code needs to be included. The effect on
4579 code size is less predictable; object code may be larger or smaller
4580 with function inlining, depending on the particular case. You can
4581 also direct GCC to try to integrate all ``simple enough'' functions
4582 into their callers with the option @option{-finline-functions}.
4584 GCC implements three different semantics of declaring a function
4585 inline. One is available with @option{-std=gnu89} or
4586 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
4587 on all inline declarations, another when @option{-std=c99} or
4588 @option{-std=gnu99} (without @option{-fgnu89-inline}), and the third
4589 is used when compiling C++.
4591 To declare a function inline, use the @code{inline} keyword in its
4592 declaration, like this:
4602 If you are writing a header file to be included in ISO C89 programs, write
4603 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
4605 The three types of inlining behave similarly in two important cases:
4606 when the @code{inline} keyword is used on a @code{static} function,
4607 like the example above, and when a function is first declared without
4608 using the @code{inline} keyword and then is defined with
4609 @code{inline}, like this:
4612 extern int inc (int *a);
4620 In both of these common cases, the program behaves the same as if you
4621 had not used the @code{inline} keyword, except for its speed.
4623 @cindex inline functions, omission of
4624 @opindex fkeep-inline-functions
4625 When a function is both inline and @code{static}, if all calls to the
4626 function are integrated into the caller, and the function's address is
4627 never used, then the function's own assembler code is never referenced.
4628 In this case, GCC does not actually output assembler code for the
4629 function, unless you specify the option @option{-fkeep-inline-functions}.
4630 Some calls cannot be integrated for various reasons (in particular,
4631 calls that precede the function's definition cannot be integrated, and
4632 neither can recursive calls within the definition). If there is a
4633 nonintegrated call, then the function is compiled to assembler code as
4634 usual. The function must also be compiled as usual if the program
4635 refers to its address, because that can't be inlined.
4638 Note that certain usages in a function definition can make it unsuitable
4639 for inline substitution. Among these usages are: use of varargs, use of
4640 alloca, use of variable sized data types (@pxref{Variable Length}),
4641 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
4642 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
4643 will warn when a function marked @code{inline} could not be substituted,
4644 and will give the reason for the failure.
4646 @cindex automatic @code{inline} for C++ member fns
4647 @cindex @code{inline} automatic for C++ member fns
4648 @cindex member fns, automatically @code{inline}
4649 @cindex C++ member fns, automatically @code{inline}
4650 @opindex fno-default-inline
4651 As required by ISO C++, GCC considers member functions defined within
4652 the body of a class to be marked inline even if they are
4653 not explicitly declared with the @code{inline} keyword. You can
4654 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
4655 Options,,Options Controlling C++ Dialect}.
4657 GCC does not inline any functions when not optimizing unless you specify
4658 the @samp{always_inline} attribute for the function, like this:
4661 /* @r{Prototype.} */
4662 inline void foo (const char) __attribute__((always_inline));
4665 The remainder of this section is specific to GNU C89 inlining.
4667 @cindex non-static inline function
4668 When an inline function is not @code{static}, then the compiler must assume
4669 that there may be calls from other source files; since a global symbol can
4670 be defined only once in any program, the function must not be defined in
4671 the other source files, so the calls therein cannot be integrated.
4672 Therefore, a non-@code{static} inline function is always compiled on its
4673 own in the usual fashion.
4675 If you specify both @code{inline} and @code{extern} in the function
4676 definition, then the definition is used only for inlining. In no case
4677 is the function compiled on its own, not even if you refer to its
4678 address explicitly. Such an address becomes an external reference, as
4679 if you had only declared the function, and had not defined it.
4681 This combination of @code{inline} and @code{extern} has almost the
4682 effect of a macro. The way to use it is to put a function definition in
4683 a header file with these keywords, and put another copy of the
4684 definition (lacking @code{inline} and @code{extern}) in a library file.
4685 The definition in the header file will cause most calls to the function
4686 to be inlined. If any uses of the function remain, they will refer to
4687 the single copy in the library.
4690 @section Assembler Instructions with C Expression Operands
4691 @cindex extended @code{asm}
4692 @cindex @code{asm} expressions
4693 @cindex assembler instructions
4696 In an assembler instruction using @code{asm}, you can specify the
4697 operands of the instruction using C expressions. This means you need not
4698 guess which registers or memory locations will contain the data you want
4701 You must specify an assembler instruction template much like what
4702 appears in a machine description, plus an operand constraint string for
4705 For example, here is how to use the 68881's @code{fsinx} instruction:
4708 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
4712 Here @code{angle} is the C expression for the input operand while
4713 @code{result} is that of the output operand. Each has @samp{"f"} as its
4714 operand constraint, saying that a floating point register is required.
4715 The @samp{=} in @samp{=f} indicates that the operand is an output; all
4716 output operands' constraints must use @samp{=}. The constraints use the
4717 same language used in the machine description (@pxref{Constraints}).
4719 Each operand is described by an operand-constraint string followed by
4720 the C expression in parentheses. A colon separates the assembler
4721 template from the first output operand and another separates the last
4722 output operand from the first input, if any. Commas separate the
4723 operands within each group. The total number of operands is currently
4724 limited to 30; this limitation may be lifted in some future version of
4727 If there are no output operands but there are input operands, you must
4728 place two consecutive colons surrounding the place where the output
4731 As of GCC version 3.1, it is also possible to specify input and output
4732 operands using symbolic names which can be referenced within the
4733 assembler code. These names are specified inside square brackets
4734 preceding the constraint string, and can be referenced inside the
4735 assembler code using @code{%[@var{name}]} instead of a percentage sign
4736 followed by the operand number. Using named operands the above example
4740 asm ("fsinx %[angle],%[output]"
4741 : [output] "=f" (result)
4742 : [angle] "f" (angle));
4746 Note that the symbolic operand names have no relation whatsoever to
4747 other C identifiers. You may use any name you like, even those of
4748 existing C symbols, but you must ensure that no two operands within the same
4749 assembler construct use the same symbolic name.
4751 Output operand expressions must be lvalues; the compiler can check this.
4752 The input operands need not be lvalues. The compiler cannot check
4753 whether the operands have data types that are reasonable for the
4754 instruction being executed. It does not parse the assembler instruction
4755 template and does not know what it means or even whether it is valid
4756 assembler input. The extended @code{asm} feature is most often used for
4757 machine instructions the compiler itself does not know exist. If
4758 the output expression cannot be directly addressed (for example, it is a
4759 bit-field), your constraint must allow a register. In that case, GCC
4760 will use the register as the output of the @code{asm}, and then store
4761 that register into the output.
4763 The ordinary output operands must be write-only; GCC will assume that
4764 the values in these operands before the instruction are dead and need
4765 not be generated. Extended asm supports input-output or read-write
4766 operands. Use the constraint character @samp{+} to indicate such an
4767 operand and list it with the output operands. You should only use
4768 read-write operands when the constraints for the operand (or the
4769 operand in which only some of the bits are to be changed) allow a
4772 You may, as an alternative, logically split its function into two
4773 separate operands, one input operand and one write-only output
4774 operand. The connection between them is expressed by constraints
4775 which say they need to be in the same location when the instruction
4776 executes. You can use the same C expression for both operands, or
4777 different expressions. For example, here we write the (fictitious)
4778 @samp{combine} instruction with @code{bar} as its read-only source
4779 operand and @code{foo} as its read-write destination:
4782 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
4786 The constraint @samp{"0"} for operand 1 says that it must occupy the
4787 same location as operand 0. A number in constraint is allowed only in
4788 an input operand and it must refer to an output operand.
4790 Only a number in the constraint can guarantee that one operand will be in
4791 the same place as another. The mere fact that @code{foo} is the value
4792 of both operands is not enough to guarantee that they will be in the
4793 same place in the generated assembler code. The following would not
4797 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
4800 Various optimizations or reloading could cause operands 0 and 1 to be in
4801 different registers; GCC knows no reason not to do so. For example, the
4802 compiler might find a copy of the value of @code{foo} in one register and
4803 use it for operand 1, but generate the output operand 0 in a different
4804 register (copying it afterward to @code{foo}'s own address). Of course,
4805 since the register for operand 1 is not even mentioned in the assembler
4806 code, the result will not work, but GCC can't tell that.
4808 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
4809 the operand number for a matching constraint. For example:
4812 asm ("cmoveq %1,%2,%[result]"
4813 : [result] "=r"(result)
4814 : "r" (test), "r"(new), "[result]"(old));
4817 Sometimes you need to make an @code{asm} operand be a specific register,
4818 but there's no matching constraint letter for that register @emph{by
4819 itself}. To force the operand into that register, use a local variable
4820 for the operand and specify the register in the variable declaration.
4821 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
4822 register constraint letter that matches the register:
4825 register int *p1 asm ("r0") = @dots{};
4826 register int *p2 asm ("r1") = @dots{};
4827 register int *result asm ("r0");
4828 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4831 @anchor{Example of asm with clobbered asm reg}
4832 In the above example, beware that a register that is call-clobbered by
4833 the target ABI will be overwritten by any function call in the
4834 assignment, including library calls for arithmetic operators.
4835 Assuming it is a call-clobbered register, this may happen to @code{r0}
4836 above by the assignment to @code{p2}. If you have to use such a
4837 register, use temporary variables for expressions between the register
4842 register int *p1 asm ("r0") = @dots{};
4843 register int *p2 asm ("r1") = t1;
4844 register int *result asm ("r0");
4845 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4848 Some instructions clobber specific hard registers. To describe this,
4849 write a third colon after the input operands, followed by the names of
4850 the clobbered hard registers (given as strings). Here is a realistic
4851 example for the VAX:
4854 asm volatile ("movc3 %0,%1,%2"
4855 : /* @r{no outputs} */
4856 : "g" (from), "g" (to), "g" (count)
4857 : "r0", "r1", "r2", "r3", "r4", "r5");
4860 You may not write a clobber description in a way that overlaps with an
4861 input or output operand. For example, you may not have an operand
4862 describing a register class with one member if you mention that register
4863 in the clobber list. Variables declared to live in specific registers
4864 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
4865 have no part mentioned in the clobber description.
4866 There is no way for you to specify that an input
4867 operand is modified without also specifying it as an output
4868 operand. Note that if all the output operands you specify are for this
4869 purpose (and hence unused), you will then also need to specify
4870 @code{volatile} for the @code{asm} construct, as described below, to
4871 prevent GCC from deleting the @code{asm} statement as unused.
4873 If you refer to a particular hardware register from the assembler code,
4874 you will probably have to list the register after the third colon to
4875 tell the compiler the register's value is modified. In some assemblers,
4876 the register names begin with @samp{%}; to produce one @samp{%} in the
4877 assembler code, you must write @samp{%%} in the input.
4879 If your assembler instruction can alter the condition code register, add
4880 @samp{cc} to the list of clobbered registers. GCC on some machines
4881 represents the condition codes as a specific hardware register;
4882 @samp{cc} serves to name this register. On other machines, the
4883 condition code is handled differently, and specifying @samp{cc} has no
4884 effect. But it is valid no matter what the machine.
4886 If your assembler instructions access memory in an unpredictable
4887 fashion, add @samp{memory} to the list of clobbered registers. This
4888 will cause GCC to not keep memory values cached in registers across the
4889 assembler instruction and not optimize stores or loads to that memory.
4890 You will also want to add the @code{volatile} keyword if the memory
4891 affected is not listed in the inputs or outputs of the @code{asm}, as
4892 the @samp{memory} clobber does not count as a side-effect of the
4893 @code{asm}. If you know how large the accessed memory is, you can add
4894 it as input or output but if this is not known, you should add
4895 @samp{memory}. As an example, if you access ten bytes of a string, you
4896 can use a memory input like:
4899 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
4902 Note that in the following example the memory input is necessary,
4903 otherwise GCC might optimize the store to @code{x} away:
4910 asm ("magic stuff accessing an 'int' pointed to by '%1'"
4911 "=&d" (r) : "a" (y), "m" (*y));
4916 You can put multiple assembler instructions together in a single
4917 @code{asm} template, separated by the characters normally used in assembly
4918 code for the system. A combination that works in most places is a newline
4919 to break the line, plus a tab character to move to the instruction field
4920 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
4921 assembler allows semicolons as a line-breaking character. Note that some
4922 assembler dialects use semicolons to start a comment.
4923 The input operands are guaranteed not to use any of the clobbered
4924 registers, and neither will the output operands' addresses, so you can
4925 read and write the clobbered registers as many times as you like. Here
4926 is an example of multiple instructions in a template; it assumes the
4927 subroutine @code{_foo} accepts arguments in registers 9 and 10:
4930 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
4932 : "g" (from), "g" (to)
4936 Unless an output operand has the @samp{&} constraint modifier, GCC
4937 may allocate it in the same register as an unrelated input operand, on
4938 the assumption the inputs are consumed before the outputs are produced.
4939 This assumption may be false if the assembler code actually consists of
4940 more than one instruction. In such a case, use @samp{&} for each output
4941 operand that may not overlap an input. @xref{Modifiers}.
4943 If you want to test the condition code produced by an assembler
4944 instruction, you must include a branch and a label in the @code{asm}
4945 construct, as follows:
4948 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
4954 This assumes your assembler supports local labels, as the GNU assembler
4955 and most Unix assemblers do.
4957 Speaking of labels, jumps from one @code{asm} to another are not
4958 supported. The compiler's optimizers do not know about these jumps, and
4959 therefore they cannot take account of them when deciding how to
4962 @cindex macros containing @code{asm}
4963 Usually the most convenient way to use these @code{asm} instructions is to
4964 encapsulate them in macros that look like functions. For example,
4968 (@{ double __value, __arg = (x); \
4969 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
4974 Here the variable @code{__arg} is used to make sure that the instruction
4975 operates on a proper @code{double} value, and to accept only those
4976 arguments @code{x} which can convert automatically to a @code{double}.
4978 Another way to make sure the instruction operates on the correct data
4979 type is to use a cast in the @code{asm}. This is different from using a
4980 variable @code{__arg} in that it converts more different types. For
4981 example, if the desired type were @code{int}, casting the argument to
4982 @code{int} would accept a pointer with no complaint, while assigning the
4983 argument to an @code{int} variable named @code{__arg} would warn about
4984 using a pointer unless the caller explicitly casts it.
4986 If an @code{asm} has output operands, GCC assumes for optimization
4987 purposes the instruction has no side effects except to change the output
4988 operands. This does not mean instructions with a side effect cannot be
4989 used, but you must be careful, because the compiler may eliminate them
4990 if the output operands aren't used, or move them out of loops, or
4991 replace two with one if they constitute a common subexpression. Also,
4992 if your instruction does have a side effect on a variable that otherwise
4993 appears not to change, the old value of the variable may be reused later
4994 if it happens to be found in a register.
4996 You can prevent an @code{asm} instruction from being deleted
4997 by writing the keyword @code{volatile} after
4998 the @code{asm}. For example:
5001 #define get_and_set_priority(new) \
5003 asm volatile ("get_and_set_priority %0, %1" \
5004 : "=g" (__old) : "g" (new)); \
5009 The @code{volatile} keyword indicates that the instruction has
5010 important side-effects. GCC will not delete a volatile @code{asm} if
5011 it is reachable. (The instruction can still be deleted if GCC can
5012 prove that control-flow will never reach the location of the
5013 instruction.) Note that even a volatile @code{asm} instruction
5014 can be moved relative to other code, including across jump
5015 instructions. For example, on many targets there is a system
5016 register which can be set to control the rounding mode of
5017 floating point operations. You might try
5018 setting it with a volatile @code{asm}, like this PowerPC example:
5021 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
5026 This will not work reliably, as the compiler may move the addition back
5027 before the volatile @code{asm}. To make it work you need to add an
5028 artificial dependency to the @code{asm} referencing a variable in the code
5029 you don't want moved, for example:
5032 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
5036 Similarly, you can't expect a
5037 sequence of volatile @code{asm} instructions to remain perfectly
5038 consecutive. If you want consecutive output, use a single @code{asm}.
5039 Also, GCC will perform some optimizations across a volatile @code{asm}
5040 instruction; GCC does not ``forget everything'' when it encounters
5041 a volatile @code{asm} instruction the way some other compilers do.
5043 An @code{asm} instruction without any output operands will be treated
5044 identically to a volatile @code{asm} instruction.
5046 It is a natural idea to look for a way to give access to the condition
5047 code left by the assembler instruction. However, when we attempted to
5048 implement this, we found no way to make it work reliably. The problem
5049 is that output operands might need reloading, which would result in
5050 additional following ``store'' instructions. On most machines, these
5051 instructions would alter the condition code before there was time to
5052 test it. This problem doesn't arise for ordinary ``test'' and
5053 ``compare'' instructions because they don't have any output operands.
5055 For reasons similar to those described above, it is not possible to give
5056 an assembler instruction access to the condition code left by previous
5059 If you are writing a header file that should be includable in ISO C
5060 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
5063 @subsection Size of an @code{asm}
5065 Some targets require that GCC track the size of each instruction used in
5066 order to generate correct code. Because the final length of an
5067 @code{asm} is only known by the assembler, GCC must make an estimate as
5068 to how big it will be. The estimate is formed by counting the number of
5069 statements in the pattern of the @code{asm} and multiplying that by the
5070 length of the longest instruction on that processor. Statements in the
5071 @code{asm} are identified by newline characters and whatever statement
5072 separator characters are supported by the assembler; on most processors
5073 this is the `@code{;}' character.
5075 Normally, GCC's estimate is perfectly adequate to ensure that correct
5076 code is generated, but it is possible to confuse the compiler if you use
5077 pseudo instructions or assembler macros that expand into multiple real
5078 instructions or if you use assembler directives that expand to more
5079 space in the object file than would be needed for a single instruction.
5080 If this happens then the assembler will produce a diagnostic saying that
5081 a label is unreachable.
5083 @subsection i386 floating point asm operands
5085 There are several rules on the usage of stack-like regs in
5086 asm_operands insns. These rules apply only to the operands that are
5091 Given a set of input regs that die in an asm_operands, it is
5092 necessary to know which are implicitly popped by the asm, and
5093 which must be explicitly popped by gcc.
5095 An input reg that is implicitly popped by the asm must be
5096 explicitly clobbered, unless it is constrained to match an
5100 For any input reg that is implicitly popped by an asm, it is
5101 necessary to know how to adjust the stack to compensate for the pop.
5102 If any non-popped input is closer to the top of the reg-stack than
5103 the implicitly popped reg, it would not be possible to know what the
5104 stack looked like---it's not clear how the rest of the stack ``slides
5107 All implicitly popped input regs must be closer to the top of
5108 the reg-stack than any input that is not implicitly popped.
5110 It is possible that if an input dies in an insn, reload might
5111 use the input reg for an output reload. Consider this example:
5114 asm ("foo" : "=t" (a) : "f" (b));
5117 This asm says that input B is not popped by the asm, and that
5118 the asm pushes a result onto the reg-stack, i.e., the stack is one
5119 deeper after the asm than it was before. But, it is possible that
5120 reload will think that it can use the same reg for both the input and
5121 the output, if input B dies in this insn.
5123 If any input operand uses the @code{f} constraint, all output reg
5124 constraints must use the @code{&} earlyclobber.
5126 The asm above would be written as
5129 asm ("foo" : "=&t" (a) : "f" (b));
5133 Some operands need to be in particular places on the stack. All
5134 output operands fall in this category---there is no other way to
5135 know which regs the outputs appear in unless the user indicates
5136 this in the constraints.
5138 Output operands must specifically indicate which reg an output
5139 appears in after an asm. @code{=f} is not allowed: the operand
5140 constraints must select a class with a single reg.
5143 Output operands may not be ``inserted'' between existing stack regs.
5144 Since no 387 opcode uses a read/write operand, all output operands
5145 are dead before the asm_operands, and are pushed by the asm_operands.
5146 It makes no sense to push anywhere but the top of the reg-stack.
5148 Output operands must start at the top of the reg-stack: output
5149 operands may not ``skip'' a reg.
5152 Some asm statements may need extra stack space for internal
5153 calculations. This can be guaranteed by clobbering stack registers
5154 unrelated to the inputs and outputs.
5158 Here are a couple of reasonable asms to want to write. This asm
5159 takes one input, which is internally popped, and produces two outputs.
5162 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
5165 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
5166 and replaces them with one output. The user must code the @code{st(1)}
5167 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
5170 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
5176 @section Controlling Names Used in Assembler Code
5177 @cindex assembler names for identifiers
5178 @cindex names used in assembler code
5179 @cindex identifiers, names in assembler code
5181 You can specify the name to be used in the assembler code for a C
5182 function or variable by writing the @code{asm} (or @code{__asm__})
5183 keyword after the declarator as follows:
5186 int foo asm ("myfoo") = 2;
5190 This specifies that the name to be used for the variable @code{foo} in
5191 the assembler code should be @samp{myfoo} rather than the usual
5194 On systems where an underscore is normally prepended to the name of a C
5195 function or variable, this feature allows you to define names for the
5196 linker that do not start with an underscore.
5198 It does not make sense to use this feature with a non-static local
5199 variable since such variables do not have assembler names. If you are
5200 trying to put the variable in a particular register, see @ref{Explicit
5201 Reg Vars}. GCC presently accepts such code with a warning, but will
5202 probably be changed to issue an error, rather than a warning, in the
5205 You cannot use @code{asm} in this way in a function @emph{definition}; but
5206 you can get the same effect by writing a declaration for the function
5207 before its definition and putting @code{asm} there, like this:
5210 extern func () asm ("FUNC");
5217 It is up to you to make sure that the assembler names you choose do not
5218 conflict with any other assembler symbols. Also, you must not use a
5219 register name; that would produce completely invalid assembler code. GCC
5220 does not as yet have the ability to store static variables in registers.
5221 Perhaps that will be added.
5223 @node Explicit Reg Vars
5224 @section Variables in Specified Registers
5225 @cindex explicit register variables
5226 @cindex variables in specified registers
5227 @cindex specified registers
5228 @cindex registers, global allocation
5230 GNU C allows you to put a few global variables into specified hardware
5231 registers. You can also specify the register in which an ordinary
5232 register variable should be allocated.
5236 Global register variables reserve registers throughout the program.
5237 This may be useful in programs such as programming language
5238 interpreters which have a couple of global variables that are accessed
5242 Local register variables in specific registers do not reserve the
5243 registers, except at the point where they are used as input or output
5244 operands in an @code{asm} statement and the @code{asm} statement itself is
5245 not deleted. The compiler's data flow analysis is capable of determining
5246 where the specified registers contain live values, and where they are
5247 available for other uses. Stores into local register variables may be deleted
5248 when they appear to be dead according to dataflow analysis. References
5249 to local register variables may be deleted or moved or simplified.
5251 These local variables are sometimes convenient for use with the extended
5252 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
5253 output of the assembler instruction directly into a particular register.
5254 (This will work provided the register you specify fits the constraints
5255 specified for that operand in the @code{asm}.)
5263 @node Global Reg Vars
5264 @subsection Defining Global Register Variables
5265 @cindex global register variables
5266 @cindex registers, global variables in
5268 You can define a global register variable in GNU C like this:
5271 register int *foo asm ("a5");
5275 Here @code{a5} is the name of the register which should be used. Choose a
5276 register which is normally saved and restored by function calls on your
5277 machine, so that library routines will not clobber it.
5279 Naturally the register name is cpu-dependent, so you would need to
5280 conditionalize your program according to cpu type. The register
5281 @code{a5} would be a good choice on a 68000 for a variable of pointer
5282 type. On machines with register windows, be sure to choose a ``global''
5283 register that is not affected magically by the function call mechanism.
5285 In addition, operating systems on one type of cpu may differ in how they
5286 name the registers; then you would need additional conditionals. For
5287 example, some 68000 operating systems call this register @code{%a5}.
5289 Eventually there may be a way of asking the compiler to choose a register
5290 automatically, but first we need to figure out how it should choose and
5291 how to enable you to guide the choice. No solution is evident.
5293 Defining a global register variable in a certain register reserves that
5294 register entirely for this use, at least within the current compilation.
5295 The register will not be allocated for any other purpose in the functions
5296 in the current compilation. The register will not be saved and restored by
5297 these functions. Stores into this register are never deleted even if they
5298 would appear to be dead, but references may be deleted or moved or
5301 It is not safe to access the global register variables from signal
5302 handlers, or from more than one thread of control, because the system
5303 library routines may temporarily use the register for other things (unless
5304 you recompile them specially for the task at hand).
5306 @cindex @code{qsort}, and global register variables
5307 It is not safe for one function that uses a global register variable to
5308 call another such function @code{foo} by way of a third function
5309 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
5310 different source file in which the variable wasn't declared). This is
5311 because @code{lose} might save the register and put some other value there.
5312 For example, you can't expect a global register variable to be available in
5313 the comparison-function that you pass to @code{qsort}, since @code{qsort}
5314 might have put something else in that register. (If you are prepared to
5315 recompile @code{qsort} with the same global register variable, you can
5316 solve this problem.)
5318 If you want to recompile @code{qsort} or other source files which do not
5319 actually use your global register variable, so that they will not use that
5320 register for any other purpose, then it suffices to specify the compiler
5321 option @option{-ffixed-@var{reg}}. You need not actually add a global
5322 register declaration to their source code.
5324 A function which can alter the value of a global register variable cannot
5325 safely be called from a function compiled without this variable, because it
5326 could clobber the value the caller expects to find there on return.
5327 Therefore, the function which is the entry point into the part of the
5328 program that uses the global register variable must explicitly save and
5329 restore the value which belongs to its caller.
5331 @cindex register variable after @code{longjmp}
5332 @cindex global register after @code{longjmp}
5333 @cindex value after @code{longjmp}
5336 On most machines, @code{longjmp} will restore to each global register
5337 variable the value it had at the time of the @code{setjmp}. On some
5338 machines, however, @code{longjmp} will not change the value of global
5339 register variables. To be portable, the function that called @code{setjmp}
5340 should make other arrangements to save the values of the global register
5341 variables, and to restore them in a @code{longjmp}. This way, the same
5342 thing will happen regardless of what @code{longjmp} does.
5344 All global register variable declarations must precede all function
5345 definitions. If such a declaration could appear after function
5346 definitions, the declaration would be too late to prevent the register from
5347 being used for other purposes in the preceding functions.
5349 Global register variables may not have initial values, because an
5350 executable file has no means to supply initial contents for a register.
5352 On the SPARC, there are reports that g3 @dots{} g7 are suitable
5353 registers, but certain library functions, such as @code{getwd}, as well
5354 as the subroutines for division and remainder, modify g3 and g4. g1 and
5355 g2 are local temporaries.
5357 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
5358 Of course, it will not do to use more than a few of those.
5360 @node Local Reg Vars
5361 @subsection Specifying Registers for Local Variables
5362 @cindex local variables, specifying registers
5363 @cindex specifying registers for local variables
5364 @cindex registers for local variables
5366 You can define a local register variable with a specified register
5370 register int *foo asm ("a5");
5374 Here @code{a5} is the name of the register which should be used. Note
5375 that this is the same syntax used for defining global register
5376 variables, but for a local variable it would appear within a function.
5378 Naturally the register name is cpu-dependent, but this is not a
5379 problem, since specific registers are most often useful with explicit
5380 assembler instructions (@pxref{Extended Asm}). Both of these things
5381 generally require that you conditionalize your program according to
5384 In addition, operating systems on one type of cpu may differ in how they
5385 name the registers; then you would need additional conditionals. For
5386 example, some 68000 operating systems call this register @code{%a5}.
5388 Defining such a register variable does not reserve the register; it
5389 remains available for other uses in places where flow control determines
5390 the variable's value is not live.
5392 This option does not guarantee that GCC will generate code that has
5393 this variable in the register you specify at all times. You may not
5394 code an explicit reference to this register in the @emph{assembler
5395 instruction template} part of an @code{asm} statement and assume it will
5396 always refer to this variable. However, using the variable as an
5397 @code{asm} @emph{operand} guarantees that the specified register is used
5400 Stores into local register variables may be deleted when they appear to be dead
5401 according to dataflow analysis. References to local register variables may
5402 be deleted or moved or simplified.
5404 As for global register variables, it's recommended that you choose a
5405 register which is normally saved and restored by function calls on
5406 your machine, so that library routines will not clobber it. A common
5407 pitfall is to initialize multiple call-clobbered registers with
5408 arbitrary expressions, where a function call or library call for an
5409 arithmetic operator will overwrite a register value from a previous
5410 assignment, for example @code{r0} below:
5412 register int *p1 asm ("r0") = @dots{};
5413 register int *p2 asm ("r1") = @dots{};
5415 In those cases, a solution is to use a temporary variable for
5416 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
5418 @node Alternate Keywords
5419 @section Alternate Keywords
5420 @cindex alternate keywords
5421 @cindex keywords, alternate
5423 @option{-ansi} and the various @option{-std} options disable certain
5424 keywords. This causes trouble when you want to use GNU C extensions, or
5425 a general-purpose header file that should be usable by all programs,
5426 including ISO C programs. The keywords @code{asm}, @code{typeof} and
5427 @code{inline} are not available in programs compiled with
5428 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
5429 program compiled with @option{-std=c99}). The ISO C99 keyword
5430 @code{restrict} is only available when @option{-std=gnu99} (which will
5431 eventually be the default) or @option{-std=c99} (or the equivalent
5432 @option{-std=iso9899:1999}) is used.
5434 The way to solve these problems is to put @samp{__} at the beginning and
5435 end of each problematical keyword. For example, use @code{__asm__}
5436 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
5438 Other C compilers won't accept these alternative keywords; if you want to
5439 compile with another compiler, you can define the alternate keywords as
5440 macros to replace them with the customary keywords. It looks like this:
5448 @findex __extension__
5450 @option{-pedantic} and other options cause warnings for many GNU C extensions.
5452 prevent such warnings within one expression by writing
5453 @code{__extension__} before the expression. @code{__extension__} has no
5454 effect aside from this.
5456 @node Incomplete Enums
5457 @section Incomplete @code{enum} Types
5459 You can define an @code{enum} tag without specifying its possible values.
5460 This results in an incomplete type, much like what you get if you write
5461 @code{struct foo} without describing the elements. A later declaration
5462 which does specify the possible values completes the type.
5464 You can't allocate variables or storage using the type while it is
5465 incomplete. However, you can work with pointers to that type.
5467 This extension may not be very useful, but it makes the handling of
5468 @code{enum} more consistent with the way @code{struct} and @code{union}
5471 This extension is not supported by GNU C++.
5473 @node Function Names
5474 @section Function Names as Strings
5475 @cindex @code{__func__} identifier
5476 @cindex @code{__FUNCTION__} identifier
5477 @cindex @code{__PRETTY_FUNCTION__} identifier
5479 GCC provides three magic variables which hold the name of the current
5480 function, as a string. The first of these is @code{__func__}, which
5481 is part of the C99 standard:
5483 The identifier @code{__func__} is implicitly declared by the translator
5484 as if, immediately following the opening brace of each function
5485 definition, the declaration
5488 static const char __func__[] = "function-name";
5492 appeared, where function-name is the name of the lexically-enclosing
5493 function. This name is the unadorned name of the function.
5495 @code{__FUNCTION__} is another name for @code{__func__}. Older
5496 versions of GCC recognize only this name. However, it is not
5497 standardized. For maximum portability, we recommend you use
5498 @code{__func__}, but provide a fallback definition with the
5502 #if __STDC_VERSION__ < 199901L
5504 # define __func__ __FUNCTION__
5506 # define __func__ "<unknown>"
5511 In C, @code{__PRETTY_FUNCTION__} is yet another name for
5512 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
5513 the type signature of the function as well as its bare name. For
5514 example, this program:
5518 extern int printf (char *, ...);
5525 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
5526 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
5544 __PRETTY_FUNCTION__ = void a::sub(int)
5547 These identifiers are not preprocessor macros. In GCC 3.3 and
5548 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
5549 were treated as string literals; they could be used to initialize
5550 @code{char} arrays, and they could be concatenated with other string
5551 literals. GCC 3.4 and later treat them as variables, like
5552 @code{__func__}. In C++, @code{__FUNCTION__} and
5553 @code{__PRETTY_FUNCTION__} have always been variables.
5555 @node Return Address
5556 @section Getting the Return or Frame Address of a Function
5558 These functions may be used to get information about the callers of a
5561 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
5562 This function returns the return address of the current function, or of
5563 one of its callers. The @var{level} argument is number of frames to
5564 scan up the call stack. A value of @code{0} yields the return address
5565 of the current function, a value of @code{1} yields the return address
5566 of the caller of the current function, and so forth. When inlining
5567 the expected behavior is that the function will return the address of
5568 the function that will be returned to. To work around this behavior use
5569 the @code{noinline} function attribute.
5571 The @var{level} argument must be a constant integer.
5573 On some machines it may be impossible to determine the return address of
5574 any function other than the current one; in such cases, or when the top
5575 of the stack has been reached, this function will return @code{0} or a
5576 random value. In addition, @code{__builtin_frame_address} may be used
5577 to determine if the top of the stack has been reached.
5579 This function should only be used with a nonzero argument for debugging
5583 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
5584 This function is similar to @code{__builtin_return_address}, but it
5585 returns the address of the function frame rather than the return address
5586 of the function. Calling @code{__builtin_frame_address} with a value of
5587 @code{0} yields the frame address of the current function, a value of
5588 @code{1} yields the frame address of the caller of the current function,
5591 The frame is the area on the stack which holds local variables and saved
5592 registers. The frame address is normally the address of the first word
5593 pushed on to the stack by the function. However, the exact definition
5594 depends upon the processor and the calling convention. If the processor
5595 has a dedicated frame pointer register, and the function has a frame,
5596 then @code{__builtin_frame_address} will return the value of the frame
5599 On some machines it may be impossible to determine the frame address of
5600 any function other than the current one; in such cases, or when the top
5601 of the stack has been reached, this function will return @code{0} if
5602 the first frame pointer is properly initialized by the startup code.
5604 This function should only be used with a nonzero argument for debugging
5608 @node Vector Extensions
5609 @section Using vector instructions through built-in functions
5611 On some targets, the instruction set contains SIMD vector instructions that
5612 operate on multiple values contained in one large register at the same time.
5613 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
5616 The first step in using these extensions is to provide the necessary data
5617 types. This should be done using an appropriate @code{typedef}:
5620 typedef int v4si __attribute__ ((vector_size (16)));
5623 The @code{int} type specifies the base type, while the attribute specifies
5624 the vector size for the variable, measured in bytes. For example, the
5625 declaration above causes the compiler to set the mode for the @code{v4si}
5626 type to be 16 bytes wide and divided into @code{int} sized units. For
5627 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
5628 corresponding mode of @code{foo} will be @acronym{V4SI}.
5630 The @code{vector_size} attribute is only applicable to integral and
5631 float scalars, although arrays, pointers, and function return values
5632 are allowed in conjunction with this construct.
5634 All the basic integer types can be used as base types, both as signed
5635 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
5636 @code{long long}. In addition, @code{float} and @code{double} can be
5637 used to build floating-point vector types.
5639 Specifying a combination that is not valid for the current architecture
5640 will cause GCC to synthesize the instructions using a narrower mode.
5641 For example, if you specify a variable of type @code{V4SI} and your
5642 architecture does not allow for this specific SIMD type, GCC will
5643 produce code that uses 4 @code{SIs}.
5645 The types defined in this manner can be used with a subset of normal C
5646 operations. Currently, GCC will allow using the following operators
5647 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
5649 The operations behave like C++ @code{valarrays}. Addition is defined as
5650 the addition of the corresponding elements of the operands. For
5651 example, in the code below, each of the 4 elements in @var{a} will be
5652 added to the corresponding 4 elements in @var{b} and the resulting
5653 vector will be stored in @var{c}.
5656 typedef int v4si __attribute__ ((vector_size (16)));
5663 Subtraction, multiplication, division, and the logical operations
5664 operate in a similar manner. Likewise, the result of using the unary
5665 minus or complement operators on a vector type is a vector whose
5666 elements are the negative or complemented values of the corresponding
5667 elements in the operand.
5669 You can declare variables and use them in function calls and returns, as
5670 well as in assignments and some casts. You can specify a vector type as
5671 a return type for a function. Vector types can also be used as function
5672 arguments. It is possible to cast from one vector type to another,
5673 provided they are of the same size (in fact, you can also cast vectors
5674 to and from other datatypes of the same size).
5676 You cannot operate between vectors of different lengths or different
5677 signedness without a cast.
5679 A port that supports hardware vector operations, usually provides a set
5680 of built-in functions that can be used to operate on vectors. For
5681 example, a function to add two vectors and multiply the result by a
5682 third could look like this:
5685 v4si f (v4si a, v4si b, v4si c)
5687 v4si tmp = __builtin_addv4si (a, b);
5688 return __builtin_mulv4si (tmp, c);
5695 @findex __builtin_offsetof
5697 GCC implements for both C and C++ a syntactic extension to implement
5698 the @code{offsetof} macro.
5702 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
5704 offsetof_member_designator:
5706 | offsetof_member_designator "." @code{identifier}
5707 | offsetof_member_designator "[" @code{expr} "]"
5710 This extension is sufficient such that
5713 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
5716 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
5717 may be dependent. In either case, @var{member} may consist of a single
5718 identifier, or a sequence of member accesses and array references.
5720 @node Atomic Builtins
5721 @section Built-in functions for atomic memory access
5723 The following builtins are intended to be compatible with those described
5724 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
5725 section 7.4. As such, they depart from the normal GCC practice of using
5726 the ``__builtin_'' prefix, and further that they are overloaded such that
5727 they work on multiple types.
5729 The definition given in the Intel documentation allows only for the use of
5730 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
5731 counterparts. GCC will allow any integral scalar or pointer type that is
5732 1, 2, 4 or 8 bytes in length.
5734 Not all operations are supported by all target processors. If a particular
5735 operation cannot be implemented on the target processor, a warning will be
5736 generated and a call an external function will be generated. The external
5737 function will carry the same name as the builtin, with an additional suffix
5738 @samp{_@var{n}} where @var{n} is the size of the data type.
5740 @c ??? Should we have a mechanism to suppress this warning? This is almost
5741 @c useful for implementing the operation under the control of an external
5744 In most cases, these builtins are considered a @dfn{full barrier}. That is,
5745 no memory operand will be moved across the operation, either forward or
5746 backward. Further, instructions will be issued as necessary to prevent the
5747 processor from speculating loads across the operation and from queuing stores
5748 after the operation.
5750 All of the routines are described in the Intel documentation to take
5751 ``an optional list of variables protected by the memory barrier''. It's
5752 not clear what is meant by that; it could mean that @emph{only} the
5753 following variables are protected, or it could mean that these variables
5754 should in addition be protected. At present GCC ignores this list and
5755 protects all variables which are globally accessible. If in the future
5756 we make some use of this list, an empty list will continue to mean all
5757 globally accessible variables.
5760 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
5761 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
5762 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
5763 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
5764 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
5765 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
5766 @findex __sync_fetch_and_add
5767 @findex __sync_fetch_and_sub
5768 @findex __sync_fetch_and_or
5769 @findex __sync_fetch_and_and
5770 @findex __sync_fetch_and_xor
5771 @findex __sync_fetch_and_nand
5772 These builtins perform the operation suggested by the name, and
5773 returns the value that had previously been in memory. That is,
5776 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
5777 @{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; @} // nand
5780 @emph{Note:} GCC 4.4 and later implement @code{__sync_fetch_and_nand}
5781 builtin as @code{*ptr = ~(tmp & value)} instead of @code{*ptr = ~tmp & value}.
5783 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
5784 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
5785 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
5786 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
5787 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
5788 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
5789 @findex __sync_add_and_fetch
5790 @findex __sync_sub_and_fetch
5791 @findex __sync_or_and_fetch
5792 @findex __sync_and_and_fetch
5793 @findex __sync_xor_and_fetch
5794 @findex __sync_nand_and_fetch
5795 These builtins perform the operation suggested by the name, and
5796 return the new value. That is,
5799 @{ *ptr @var{op}= value; return *ptr; @}
5800 @{ *ptr = ~(*ptr & value); return *ptr; @} // nand
5803 @emph{Note:} GCC 4.4 and later implement @code{__sync_nand_and_fetch}
5804 builtin as @code{*ptr = ~(*ptr & value)} instead of
5805 @code{*ptr = ~*ptr & value}.
5807 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5808 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5809 @findex __sync_bool_compare_and_swap
5810 @findex __sync_val_compare_and_swap
5811 These builtins perform an atomic compare and swap. That is, if the current
5812 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
5815 The ``bool'' version returns true if the comparison is successful and
5816 @var{newval} was written. The ``val'' version returns the contents
5817 of @code{*@var{ptr}} before the operation.
5819 @item __sync_synchronize (...)
5820 @findex __sync_synchronize
5821 This builtin issues a full memory barrier.
5823 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
5824 @findex __sync_lock_test_and_set
5825 This builtin, as described by Intel, is not a traditional test-and-set
5826 operation, but rather an atomic exchange operation. It writes @var{value}
5827 into @code{*@var{ptr}}, and returns the previous contents of
5830 Many targets have only minimal support for such locks, and do not support
5831 a full exchange operation. In this case, a target may support reduced
5832 functionality here by which the @emph{only} valid value to store is the
5833 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
5834 is implementation defined.
5836 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
5837 This means that references after the builtin cannot move to (or be
5838 speculated to) before the builtin, but previous memory stores may not
5839 be globally visible yet, and previous memory loads may not yet be
5842 @item void __sync_lock_release (@var{type} *ptr, ...)
5843 @findex __sync_lock_release
5844 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
5845 Normally this means writing the constant 0 to @code{*@var{ptr}}.
5847 This builtin is not a full barrier, but rather a @dfn{release barrier}.
5848 This means that all previous memory stores are globally visible, and all
5849 previous memory loads have been satisfied, but following memory reads
5850 are not prevented from being speculated to before the barrier.
5853 @node Object Size Checking
5854 @section Object Size Checking Builtins
5855 @findex __builtin_object_size
5856 @findex __builtin___memcpy_chk
5857 @findex __builtin___mempcpy_chk
5858 @findex __builtin___memmove_chk
5859 @findex __builtin___memset_chk
5860 @findex __builtin___strcpy_chk
5861 @findex __builtin___stpcpy_chk
5862 @findex __builtin___strncpy_chk
5863 @findex __builtin___strcat_chk
5864 @findex __builtin___strncat_chk
5865 @findex __builtin___sprintf_chk
5866 @findex __builtin___snprintf_chk
5867 @findex __builtin___vsprintf_chk
5868 @findex __builtin___vsnprintf_chk
5869 @findex __builtin___printf_chk
5870 @findex __builtin___vprintf_chk
5871 @findex __builtin___fprintf_chk
5872 @findex __builtin___vfprintf_chk
5874 GCC implements a limited buffer overflow protection mechanism
5875 that can prevent some buffer overflow attacks.
5877 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
5878 is a built-in construct that returns a constant number of bytes from
5879 @var{ptr} to the end of the object @var{ptr} pointer points to
5880 (if known at compile time). @code{__builtin_object_size} never evaluates
5881 its arguments for side-effects. If there are any side-effects in them, it
5882 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5883 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
5884 point to and all of them are known at compile time, the returned number
5885 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
5886 0 and minimum if nonzero. If it is not possible to determine which objects
5887 @var{ptr} points to at compile time, @code{__builtin_object_size} should
5888 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5889 for @var{type} 2 or 3.
5891 @var{type} is an integer constant from 0 to 3. If the least significant
5892 bit is clear, objects are whole variables, if it is set, a closest
5893 surrounding subobject is considered the object a pointer points to.
5894 The second bit determines if maximum or minimum of remaining bytes
5898 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
5899 char *p = &var.buf1[1], *q = &var.b;
5901 /* Here the object p points to is var. */
5902 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
5903 /* The subobject p points to is var.buf1. */
5904 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
5905 /* The object q points to is var. */
5906 assert (__builtin_object_size (q, 0)
5907 == (char *) (&var + 1) - (char *) &var.b);
5908 /* The subobject q points to is var.b. */
5909 assert (__builtin_object_size (q, 1) == sizeof (var.b));
5913 There are built-in functions added for many common string operation
5914 functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk}
5915 built-in is provided. This built-in has an additional last argument,
5916 which is the number of bytes remaining in object the @var{dest}
5917 argument points to or @code{(size_t) -1} if the size is not known.
5919 The built-in functions are optimized into the normal string functions
5920 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
5921 it is known at compile time that the destination object will not
5922 be overflown. If the compiler can determine at compile time the
5923 object will be always overflown, it issues a warning.
5925 The intended use can be e.g.
5929 #define bos0(dest) __builtin_object_size (dest, 0)
5930 #define memcpy(dest, src, n) \
5931 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
5935 /* It is unknown what object p points to, so this is optimized
5936 into plain memcpy - no checking is possible. */
5937 memcpy (p, "abcde", n);
5938 /* Destination is known and length too. It is known at compile
5939 time there will be no overflow. */
5940 memcpy (&buf[5], "abcde", 5);
5941 /* Destination is known, but the length is not known at compile time.
5942 This will result in __memcpy_chk call that can check for overflow
5944 memcpy (&buf[5], "abcde", n);
5945 /* Destination is known and it is known at compile time there will
5946 be overflow. There will be a warning and __memcpy_chk call that
5947 will abort the program at runtime. */
5948 memcpy (&buf[6], "abcde", 5);
5951 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
5952 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
5953 @code{strcat} and @code{strncat}.
5955 There are also checking built-in functions for formatted output functions.
5957 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
5958 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5959 const char *fmt, ...);
5960 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
5962 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5963 const char *fmt, va_list ap);
5966 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
5967 etc.@: functions and can contain implementation specific flags on what
5968 additional security measures the checking function might take, such as
5969 handling @code{%n} differently.
5971 The @var{os} argument is the object size @var{s} points to, like in the
5972 other built-in functions. There is a small difference in the behavior
5973 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
5974 optimized into the non-checking functions only if @var{flag} is 0, otherwise
5975 the checking function is called with @var{os} argument set to
5978 In addition to this, there are checking built-in functions
5979 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
5980 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
5981 These have just one additional argument, @var{flag}, right before
5982 format string @var{fmt}. If the compiler is able to optimize them to
5983 @code{fputc} etc.@: functions, it will, otherwise the checking function
5984 should be called and the @var{flag} argument passed to it.
5986 @node Other Builtins
5987 @section Other built-in functions provided by GCC
5988 @cindex built-in functions
5989 @findex __builtin_fpclassify
5990 @findex __builtin_isfinite
5991 @findex __builtin_isnormal
5992 @findex __builtin_isgreater
5993 @findex __builtin_isgreaterequal
5994 @findex __builtin_isinf_sign
5995 @findex __builtin_isless
5996 @findex __builtin_islessequal
5997 @findex __builtin_islessgreater
5998 @findex __builtin_isunordered
5999 @findex __builtin_powi
6000 @findex __builtin_powif
6001 @findex __builtin_powil
6159 @findex fprintf_unlocked
6161 @findex fputs_unlocked
6278 @findex printf_unlocked
6310 @findex significandf
6311 @findex significandl
6382 GCC provides a large number of built-in functions other than the ones
6383 mentioned above. Some of these are for internal use in the processing
6384 of exceptions or variable-length argument lists and will not be
6385 documented here because they may change from time to time; we do not
6386 recommend general use of these functions.
6388 The remaining functions are provided for optimization purposes.
6390 @opindex fno-builtin
6391 GCC includes built-in versions of many of the functions in the standard
6392 C library. The versions prefixed with @code{__builtin_} will always be
6393 treated as having the same meaning as the C library function even if you
6394 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
6395 Many of these functions are only optimized in certain cases; if they are
6396 not optimized in a particular case, a call to the library function will
6401 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
6402 @option{-std=c99}), the functions
6403 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
6404 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
6405 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
6406 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
6407 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
6408 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
6409 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
6410 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
6411 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
6412 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
6413 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
6414 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
6415 @code{signbitd64}, @code{signbitd128}, @code{significandf},
6416 @code{significandl}, @code{significand}, @code{sincosf},
6417 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
6418 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
6419 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
6420 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
6422 may be handled as built-in functions.
6423 All these functions have corresponding versions
6424 prefixed with @code{__builtin_}, which may be used even in strict C89
6427 The ISO C99 functions
6428 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
6429 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
6430 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
6431 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
6432 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
6433 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
6434 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
6435 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
6436 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
6437 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
6438 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
6439 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
6440 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
6441 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
6442 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
6443 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
6444 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
6445 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
6446 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
6447 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
6448 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
6449 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
6450 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
6451 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
6452 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
6453 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
6454 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
6455 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
6456 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
6457 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
6458 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
6459 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
6460 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
6461 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
6462 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
6463 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
6464 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
6465 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
6466 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
6467 are handled as built-in functions
6468 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
6470 There are also built-in versions of the ISO C99 functions
6471 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
6472 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
6473 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
6474 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
6475 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
6476 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
6477 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
6478 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
6479 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
6480 that are recognized in any mode since ISO C90 reserves these names for
6481 the purpose to which ISO C99 puts them. All these functions have
6482 corresponding versions prefixed with @code{__builtin_}.
6484 The ISO C94 functions
6485 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
6486 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
6487 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
6489 are handled as built-in functions
6490 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
6492 The ISO C90 functions
6493 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
6494 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
6495 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
6496 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
6497 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
6498 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
6499 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
6500 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
6501 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
6502 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
6503 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
6504 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
6505 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
6506 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
6507 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
6508 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
6509 are all recognized as built-in functions unless
6510 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
6511 is specified for an individual function). All of these functions have
6512 corresponding versions prefixed with @code{__builtin_}.
6514 GCC provides built-in versions of the ISO C99 floating point comparison
6515 macros that avoid raising exceptions for unordered operands. They have
6516 the same names as the standard macros ( @code{isgreater},
6517 @code{isgreaterequal}, @code{isless}, @code{islessequal},
6518 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
6519 prefixed. We intend for a library implementor to be able to simply
6520 @code{#define} each standard macro to its built-in equivalent.
6521 In the same fashion, GCC provides @code{fpclassify}, @code{isfinite},
6522 @code{isinf_sign} and @code{isnormal} built-ins used with
6523 @code{__builtin_} prefixed. The @code{isinf} and @code{isnan}
6524 builtins appear both with and without the @code{__builtin_} prefix.
6526 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
6528 You can use the built-in function @code{__builtin_types_compatible_p} to
6529 determine whether two types are the same.
6531 This built-in function returns 1 if the unqualified versions of the
6532 types @var{type1} and @var{type2} (which are types, not expressions) are
6533 compatible, 0 otherwise. The result of this built-in function can be
6534 used in integer constant expressions.
6536 This built-in function ignores top level qualifiers (e.g., @code{const},
6537 @code{volatile}). For example, @code{int} is equivalent to @code{const
6540 The type @code{int[]} and @code{int[5]} are compatible. On the other
6541 hand, @code{int} and @code{char *} are not compatible, even if the size
6542 of their types, on the particular architecture are the same. Also, the
6543 amount of pointer indirection is taken into account when determining
6544 similarity. Consequently, @code{short *} is not similar to
6545 @code{short **}. Furthermore, two types that are typedefed are
6546 considered compatible if their underlying types are compatible.
6548 An @code{enum} type is not considered to be compatible with another
6549 @code{enum} type even if both are compatible with the same integer
6550 type; this is what the C standard specifies.
6551 For example, @code{enum @{foo, bar@}} is not similar to
6552 @code{enum @{hot, dog@}}.
6554 You would typically use this function in code whose execution varies
6555 depending on the arguments' types. For example:
6560 typeof (x) tmp = (x); \
6561 if (__builtin_types_compatible_p (typeof (x), long double)) \
6562 tmp = foo_long_double (tmp); \
6563 else if (__builtin_types_compatible_p (typeof (x), double)) \
6564 tmp = foo_double (tmp); \
6565 else if (__builtin_types_compatible_p (typeof (x), float)) \
6566 tmp = foo_float (tmp); \
6573 @emph{Note:} This construct is only available for C@.
6577 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
6579 You can use the built-in function @code{__builtin_choose_expr} to
6580 evaluate code depending on the value of a constant expression. This
6581 built-in function returns @var{exp1} if @var{const_exp}, which is a
6582 constant expression that must be able to be determined at compile time,
6583 is nonzero. Otherwise it returns 0.
6585 This built-in function is analogous to the @samp{? :} operator in C,
6586 except that the expression returned has its type unaltered by promotion
6587 rules. Also, the built-in function does not evaluate the expression
6588 that was not chosen. For example, if @var{const_exp} evaluates to true,
6589 @var{exp2} is not evaluated even if it has side-effects.
6591 This built-in function can return an lvalue if the chosen argument is an
6594 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
6595 type. Similarly, if @var{exp2} is returned, its return type is the same
6602 __builtin_choose_expr ( \
6603 __builtin_types_compatible_p (typeof (x), double), \
6605 __builtin_choose_expr ( \
6606 __builtin_types_compatible_p (typeof (x), float), \
6608 /* @r{The void expression results in a compile-time error} \
6609 @r{when assigning the result to something.} */ \
6613 @emph{Note:} This construct is only available for C@. Furthermore, the
6614 unused expression (@var{exp1} or @var{exp2} depending on the value of
6615 @var{const_exp}) may still generate syntax errors. This may change in
6620 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
6621 You can use the built-in function @code{__builtin_constant_p} to
6622 determine if a value is known to be constant at compile-time and hence
6623 that GCC can perform constant-folding on expressions involving that
6624 value. The argument of the function is the value to test. The function
6625 returns the integer 1 if the argument is known to be a compile-time
6626 constant and 0 if it is not known to be a compile-time constant. A
6627 return of 0 does not indicate that the value is @emph{not} a constant,
6628 but merely that GCC cannot prove it is a constant with the specified
6629 value of the @option{-O} option.
6631 You would typically use this function in an embedded application where
6632 memory was a critical resource. If you have some complex calculation,
6633 you may want it to be folded if it involves constants, but need to call
6634 a function if it does not. For example:
6637 #define Scale_Value(X) \
6638 (__builtin_constant_p (X) \
6639 ? ((X) * SCALE + OFFSET) : Scale (X))
6642 You may use this built-in function in either a macro or an inline
6643 function. However, if you use it in an inlined function and pass an
6644 argument of the function as the argument to the built-in, GCC will
6645 never return 1 when you call the inline function with a string constant
6646 or compound literal (@pxref{Compound Literals}) and will not return 1
6647 when you pass a constant numeric value to the inline function unless you
6648 specify the @option{-O} option.
6650 You may also use @code{__builtin_constant_p} in initializers for static
6651 data. For instance, you can write
6654 static const int table[] = @{
6655 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
6661 This is an acceptable initializer even if @var{EXPRESSION} is not a
6662 constant expression. GCC must be more conservative about evaluating the
6663 built-in in this case, because it has no opportunity to perform
6666 Previous versions of GCC did not accept this built-in in data
6667 initializers. The earliest version where it is completely safe is
6671 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
6672 @opindex fprofile-arcs
6673 You may use @code{__builtin_expect} to provide the compiler with
6674 branch prediction information. In general, you should prefer to
6675 use actual profile feedback for this (@option{-fprofile-arcs}), as
6676 programmers are notoriously bad at predicting how their programs
6677 actually perform. However, there are applications in which this
6678 data is hard to collect.
6680 The return value is the value of @var{exp}, which should be an integral
6681 expression. The semantics of the built-in are that it is expected that
6682 @var{exp} == @var{c}. For example:
6685 if (__builtin_expect (x, 0))
6690 would indicate that we do not expect to call @code{foo}, since
6691 we expect @code{x} to be zero. Since you are limited to integral
6692 expressions for @var{exp}, you should use constructions such as
6695 if (__builtin_expect (ptr != NULL, 1))
6700 when testing pointer or floating-point values.
6703 @deftypefn {Built-in Function} void __builtin_trap (void)
6704 This function causes the program to exit abnormally. GCC implements
6705 this function by using a target-dependent mechanism (such as
6706 intentionally executing an illegal instruction) or by calling
6707 @code{abort}. The mechanism used may vary from release to release so
6708 you should not rely on any particular implementation.
6711 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
6712 This function is used to flush the processor's instruction cache for
6713 the region of memory between @var{begin} inclusive and @var{end}
6714 exclusive. Some targets require that the instruction cache be
6715 flushed, after modifying memory containing code, in order to obtain
6716 deterministic behavior.
6718 If the target does not require instruction cache flushes,
6719 @code{__builtin___clear_cache} has no effect. Otherwise either
6720 instructions are emitted in-line to clear the instruction cache or a
6721 call to the @code{__clear_cache} function in libgcc is made.
6724 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
6725 This function is used to minimize cache-miss latency by moving data into
6726 a cache before it is accessed.
6727 You can insert calls to @code{__builtin_prefetch} into code for which
6728 you know addresses of data in memory that is likely to be accessed soon.
6729 If the target supports them, data prefetch instructions will be generated.
6730 If the prefetch is done early enough before the access then the data will
6731 be in the cache by the time it is accessed.
6733 The value of @var{addr} is the address of the memory to prefetch.
6734 There are two optional arguments, @var{rw} and @var{locality}.
6735 The value of @var{rw} is a compile-time constant one or zero; one
6736 means that the prefetch is preparing for a write to the memory address
6737 and zero, the default, means that the prefetch is preparing for a read.
6738 The value @var{locality} must be a compile-time constant integer between
6739 zero and three. A value of zero means that the data has no temporal
6740 locality, so it need not be left in the cache after the access. A value
6741 of three means that the data has a high degree of temporal locality and
6742 should be left in all levels of cache possible. Values of one and two
6743 mean, respectively, a low or moderate degree of temporal locality. The
6747 for (i = 0; i < n; i++)
6750 __builtin_prefetch (&a[i+j], 1, 1);
6751 __builtin_prefetch (&b[i+j], 0, 1);
6756 Data prefetch does not generate faults if @var{addr} is invalid, but
6757 the address expression itself must be valid. For example, a prefetch
6758 of @code{p->next} will not fault if @code{p->next} is not a valid
6759 address, but evaluation will fault if @code{p} is not a valid address.
6761 If the target does not support data prefetch, the address expression
6762 is evaluated if it includes side effects but no other code is generated
6763 and GCC does not issue a warning.
6766 @deftypefn {Built-in Function} double __builtin_huge_val (void)
6767 Returns a positive infinity, if supported by the floating-point format,
6768 else @code{DBL_MAX}. This function is suitable for implementing the
6769 ISO C macro @code{HUGE_VAL}.
6772 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
6773 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
6776 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
6777 Similar to @code{__builtin_huge_val}, except the return
6778 type is @code{long double}.
6781 @deftypefn {Built-in Function} int __builtin_fpclassify (int, int, int, int, int, ...)
6782 This built-in implements the C99 fpclassify functionality. The first
6783 five int arguments should be the target library's notion of the
6784 possible FP classes and are used for return values. They must be
6785 constant values and they must appear in this order: @code{FP_NAN},
6786 @code{FP_INFINITE}, @code{FP_NORMAL}, @code{FP_SUBNORMAL} and
6787 @code{FP_ZERO}. The ellipsis is for exactly one floating point value
6788 to classify. GCC treats the last argument as type-generic, which
6789 means it does not do default promotion from float to double.
6792 @deftypefn {Built-in Function} double __builtin_inf (void)
6793 Similar to @code{__builtin_huge_val}, except a warning is generated
6794 if the target floating-point format does not support infinities.
6797 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
6798 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
6801 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
6802 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
6805 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
6806 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
6809 @deftypefn {Built-in Function} float __builtin_inff (void)
6810 Similar to @code{__builtin_inf}, except the return type is @code{float}.
6811 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
6814 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
6815 Similar to @code{__builtin_inf}, except the return
6816 type is @code{long double}.
6819 @deftypefn {Built-in Function} int __builtin_isinf_sign (...)
6820 Similar to @code{isinf}, except the return value will be negative for
6821 an argument of @code{-Inf}. Note while the parameter list is an
6822 ellipsis, this function only accepts exactly one floating point
6823 argument. GCC treats this parameter as type-generic, which means it
6824 does not do default promotion from float to double.
6827 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
6828 This is an implementation of the ISO C99 function @code{nan}.
6830 Since ISO C99 defines this function in terms of @code{strtod}, which we
6831 do not implement, a description of the parsing is in order. The string
6832 is parsed as by @code{strtol}; that is, the base is recognized by
6833 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
6834 in the significand such that the least significant bit of the number
6835 is at the least significant bit of the significand. The number is
6836 truncated to fit the significand field provided. The significand is
6837 forced to be a quiet NaN@.
6839 This function, if given a string literal all of which would have been
6840 consumed by strtol, is evaluated early enough that it is considered a
6841 compile-time constant.
6844 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
6845 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
6848 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
6849 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
6852 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
6853 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
6856 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
6857 Similar to @code{__builtin_nan}, except the return type is @code{float}.
6860 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
6861 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
6864 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
6865 Similar to @code{__builtin_nan}, except the significand is forced
6866 to be a signaling NaN@. The @code{nans} function is proposed by
6867 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
6870 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
6871 Similar to @code{__builtin_nans}, except the return type is @code{float}.
6874 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
6875 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
6878 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
6879 Returns one plus the index of the least significant 1-bit of @var{x}, or
6880 if @var{x} is zero, returns zero.
6883 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
6884 Returns the number of leading 0-bits in @var{x}, starting at the most
6885 significant bit position. If @var{x} is 0, the result is undefined.
6888 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
6889 Returns the number of trailing 0-bits in @var{x}, starting at the least
6890 significant bit position. If @var{x} is 0, the result is undefined.
6893 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
6894 Returns the number of 1-bits in @var{x}.
6897 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
6898 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
6902 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
6903 Similar to @code{__builtin_ffs}, except the argument type is
6904 @code{unsigned long}.
6907 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
6908 Similar to @code{__builtin_clz}, except the argument type is
6909 @code{unsigned long}.
6912 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
6913 Similar to @code{__builtin_ctz}, except the argument type is
6914 @code{unsigned long}.
6917 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
6918 Similar to @code{__builtin_popcount}, except the argument type is
6919 @code{unsigned long}.
6922 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
6923 Similar to @code{__builtin_parity}, except the argument type is
6924 @code{unsigned long}.
6927 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
6928 Similar to @code{__builtin_ffs}, except the argument type is
6929 @code{unsigned long long}.
6932 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
6933 Similar to @code{__builtin_clz}, except the argument type is
6934 @code{unsigned long long}.
6937 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
6938 Similar to @code{__builtin_ctz}, except the argument type is
6939 @code{unsigned long long}.
6942 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
6943 Similar to @code{__builtin_popcount}, except the argument type is
6944 @code{unsigned long long}.
6947 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
6948 Similar to @code{__builtin_parity}, except the argument type is
6949 @code{unsigned long long}.
6952 @deftypefn {Built-in Function} double __builtin_powi (double, int)
6953 Returns the first argument raised to the power of the second. Unlike the
6954 @code{pow} function no guarantees about precision and rounding are made.
6957 @deftypefn {Built-in Function} float __builtin_powif (float, int)
6958 Similar to @code{__builtin_powi}, except the argument and return types
6962 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
6963 Similar to @code{__builtin_powi}, except the argument and return types
6964 are @code{long double}.
6967 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
6968 Returns @var{x} with the order of the bytes reversed; for example,
6969 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
6973 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
6974 Similar to @code{__builtin_bswap32}, except the argument and return types
6978 @node Target Builtins
6979 @section Built-in Functions Specific to Particular Target Machines
6981 On some target machines, GCC supports many built-in functions specific
6982 to those machines. Generally these generate calls to specific machine
6983 instructions, but allow the compiler to schedule those calls.
6986 * Alpha Built-in Functions::
6987 * ARM iWMMXt Built-in Functions::
6988 * ARM NEON Intrinsics::
6989 * Blackfin Built-in Functions::
6990 * FR-V Built-in Functions::
6991 * X86 Built-in Functions::
6992 * MIPS DSP Built-in Functions::
6993 * MIPS Paired-Single Support::
6994 * MIPS Loongson Built-in Functions::
6995 * Other MIPS Built-in Functions::
6996 * picoChip Built-in Functions::
6997 * PowerPC AltiVec Built-in Functions::
6998 * SPARC VIS Built-in Functions::
6999 * SPU Built-in Functions::
7002 @node Alpha Built-in Functions
7003 @subsection Alpha Built-in Functions
7005 These built-in functions are available for the Alpha family of
7006 processors, depending on the command-line switches used.
7008 The following built-in functions are always available. They
7009 all generate the machine instruction that is part of the name.
7012 long __builtin_alpha_implver (void)
7013 long __builtin_alpha_rpcc (void)
7014 long __builtin_alpha_amask (long)
7015 long __builtin_alpha_cmpbge (long, long)
7016 long __builtin_alpha_extbl (long, long)
7017 long __builtin_alpha_extwl (long, long)
7018 long __builtin_alpha_extll (long, long)
7019 long __builtin_alpha_extql (long, long)
7020 long __builtin_alpha_extwh (long, long)
7021 long __builtin_alpha_extlh (long, long)
7022 long __builtin_alpha_extqh (long, long)
7023 long __builtin_alpha_insbl (long, long)
7024 long __builtin_alpha_inswl (long, long)
7025 long __builtin_alpha_insll (long, long)
7026 long __builtin_alpha_insql (long, long)
7027 long __builtin_alpha_inswh (long, long)
7028 long __builtin_alpha_inslh (long, long)
7029 long __builtin_alpha_insqh (long, long)
7030 long __builtin_alpha_mskbl (long, long)
7031 long __builtin_alpha_mskwl (long, long)
7032 long __builtin_alpha_mskll (long, long)
7033 long __builtin_alpha_mskql (long, long)
7034 long __builtin_alpha_mskwh (long, long)
7035 long __builtin_alpha_msklh (long, long)
7036 long __builtin_alpha_mskqh (long, long)
7037 long __builtin_alpha_umulh (long, long)
7038 long __builtin_alpha_zap (long, long)
7039 long __builtin_alpha_zapnot (long, long)
7042 The following built-in functions are always with @option{-mmax}
7043 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
7044 later. They all generate the machine instruction that is part
7048 long __builtin_alpha_pklb (long)
7049 long __builtin_alpha_pkwb (long)
7050 long __builtin_alpha_unpkbl (long)
7051 long __builtin_alpha_unpkbw (long)
7052 long __builtin_alpha_minub8 (long, long)
7053 long __builtin_alpha_minsb8 (long, long)
7054 long __builtin_alpha_minuw4 (long, long)
7055 long __builtin_alpha_minsw4 (long, long)
7056 long __builtin_alpha_maxub8 (long, long)
7057 long __builtin_alpha_maxsb8 (long, long)
7058 long __builtin_alpha_maxuw4 (long, long)
7059 long __builtin_alpha_maxsw4 (long, long)
7060 long __builtin_alpha_perr (long, long)
7063 The following built-in functions are always with @option{-mcix}
7064 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
7065 later. They all generate the machine instruction that is part
7069 long __builtin_alpha_cttz (long)
7070 long __builtin_alpha_ctlz (long)
7071 long __builtin_alpha_ctpop (long)
7074 The following builtins are available on systems that use the OSF/1
7075 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
7076 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
7077 @code{rdval} and @code{wrval}.
7080 void *__builtin_thread_pointer (void)
7081 void __builtin_set_thread_pointer (void *)
7084 @node ARM iWMMXt Built-in Functions
7085 @subsection ARM iWMMXt Built-in Functions
7087 These built-in functions are available for the ARM family of
7088 processors when the @option{-mcpu=iwmmxt} switch is used:
7091 typedef int v2si __attribute__ ((vector_size (8)));
7092 typedef short v4hi __attribute__ ((vector_size (8)));
7093 typedef char v8qi __attribute__ ((vector_size (8)));
7095 int __builtin_arm_getwcx (int)
7096 void __builtin_arm_setwcx (int, int)
7097 int __builtin_arm_textrmsb (v8qi, int)
7098 int __builtin_arm_textrmsh (v4hi, int)
7099 int __builtin_arm_textrmsw (v2si, int)
7100 int __builtin_arm_textrmub (v8qi, int)
7101 int __builtin_arm_textrmuh (v4hi, int)
7102 int __builtin_arm_textrmuw (v2si, int)
7103 v8qi __builtin_arm_tinsrb (v8qi, int)
7104 v4hi __builtin_arm_tinsrh (v4hi, int)
7105 v2si __builtin_arm_tinsrw (v2si, int)
7106 long long __builtin_arm_tmia (long long, int, int)
7107 long long __builtin_arm_tmiabb (long long, int, int)
7108 long long __builtin_arm_tmiabt (long long, int, int)
7109 long long __builtin_arm_tmiaph (long long, int, int)
7110 long long __builtin_arm_tmiatb (long long, int, int)
7111 long long __builtin_arm_tmiatt (long long, int, int)
7112 int __builtin_arm_tmovmskb (v8qi)
7113 int __builtin_arm_tmovmskh (v4hi)
7114 int __builtin_arm_tmovmskw (v2si)
7115 long long __builtin_arm_waccb (v8qi)
7116 long long __builtin_arm_wacch (v4hi)
7117 long long __builtin_arm_waccw (v2si)
7118 v8qi __builtin_arm_waddb (v8qi, v8qi)
7119 v8qi __builtin_arm_waddbss (v8qi, v8qi)
7120 v8qi __builtin_arm_waddbus (v8qi, v8qi)
7121 v4hi __builtin_arm_waddh (v4hi, v4hi)
7122 v4hi __builtin_arm_waddhss (v4hi, v4hi)
7123 v4hi __builtin_arm_waddhus (v4hi, v4hi)
7124 v2si __builtin_arm_waddw (v2si, v2si)
7125 v2si __builtin_arm_waddwss (v2si, v2si)
7126 v2si __builtin_arm_waddwus (v2si, v2si)
7127 v8qi __builtin_arm_walign (v8qi, v8qi, int)
7128 long long __builtin_arm_wand(long long, long long)
7129 long long __builtin_arm_wandn (long long, long long)
7130 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
7131 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
7132 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
7133 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
7134 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
7135 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
7136 v2si __builtin_arm_wcmpeqw (v2si, v2si)
7137 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
7138 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
7139 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
7140 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
7141 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
7142 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
7143 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
7144 long long __builtin_arm_wmacsz (v4hi, v4hi)
7145 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
7146 long long __builtin_arm_wmacuz (v4hi, v4hi)
7147 v4hi __builtin_arm_wmadds (v4hi, v4hi)
7148 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
7149 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
7150 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
7151 v2si __builtin_arm_wmaxsw (v2si, v2si)
7152 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
7153 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
7154 v2si __builtin_arm_wmaxuw (v2si, v2si)
7155 v8qi __builtin_arm_wminsb (v8qi, v8qi)
7156 v4hi __builtin_arm_wminsh (v4hi, v4hi)
7157 v2si __builtin_arm_wminsw (v2si, v2si)
7158 v8qi __builtin_arm_wminub (v8qi, v8qi)
7159 v4hi __builtin_arm_wminuh (v4hi, v4hi)
7160 v2si __builtin_arm_wminuw (v2si, v2si)
7161 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
7162 v4hi __builtin_arm_wmulul (v4hi, v4hi)
7163 v4hi __builtin_arm_wmulum (v4hi, v4hi)
7164 long long __builtin_arm_wor (long long, long long)
7165 v2si __builtin_arm_wpackdss (long long, long long)
7166 v2si __builtin_arm_wpackdus (long long, long long)
7167 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
7168 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
7169 v4hi __builtin_arm_wpackwss (v2si, v2si)
7170 v4hi __builtin_arm_wpackwus (v2si, v2si)
7171 long long __builtin_arm_wrord (long long, long long)
7172 long long __builtin_arm_wrordi (long long, int)
7173 v4hi __builtin_arm_wrorh (v4hi, long long)
7174 v4hi __builtin_arm_wrorhi (v4hi, int)
7175 v2si __builtin_arm_wrorw (v2si, long long)
7176 v2si __builtin_arm_wrorwi (v2si, int)
7177 v2si __builtin_arm_wsadb (v8qi, v8qi)
7178 v2si __builtin_arm_wsadbz (v8qi, v8qi)
7179 v2si __builtin_arm_wsadh (v4hi, v4hi)
7180 v2si __builtin_arm_wsadhz (v4hi, v4hi)
7181 v4hi __builtin_arm_wshufh (v4hi, int)
7182 long long __builtin_arm_wslld (long long, long long)
7183 long long __builtin_arm_wslldi (long long, int)
7184 v4hi __builtin_arm_wsllh (v4hi, long long)
7185 v4hi __builtin_arm_wsllhi (v4hi, int)
7186 v2si __builtin_arm_wsllw (v2si, long long)
7187 v2si __builtin_arm_wsllwi (v2si, int)
7188 long long __builtin_arm_wsrad (long long, long long)
7189 long long __builtin_arm_wsradi (long long, int)
7190 v4hi __builtin_arm_wsrah (v4hi, long long)
7191 v4hi __builtin_arm_wsrahi (v4hi, int)
7192 v2si __builtin_arm_wsraw (v2si, long long)
7193 v2si __builtin_arm_wsrawi (v2si, int)
7194 long long __builtin_arm_wsrld (long long, long long)
7195 long long __builtin_arm_wsrldi (long long, int)
7196 v4hi __builtin_arm_wsrlh (v4hi, long long)
7197 v4hi __builtin_arm_wsrlhi (v4hi, int)
7198 v2si __builtin_arm_wsrlw (v2si, long long)
7199 v2si __builtin_arm_wsrlwi (v2si, int)
7200 v8qi __builtin_arm_wsubb (v8qi, v8qi)
7201 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
7202 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
7203 v4hi __builtin_arm_wsubh (v4hi, v4hi)
7204 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
7205 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
7206 v2si __builtin_arm_wsubw (v2si, v2si)
7207 v2si __builtin_arm_wsubwss (v2si, v2si)
7208 v2si __builtin_arm_wsubwus (v2si, v2si)
7209 v4hi __builtin_arm_wunpckehsb (v8qi)
7210 v2si __builtin_arm_wunpckehsh (v4hi)
7211 long long __builtin_arm_wunpckehsw (v2si)
7212 v4hi __builtin_arm_wunpckehub (v8qi)
7213 v2si __builtin_arm_wunpckehuh (v4hi)
7214 long long __builtin_arm_wunpckehuw (v2si)
7215 v4hi __builtin_arm_wunpckelsb (v8qi)
7216 v2si __builtin_arm_wunpckelsh (v4hi)
7217 long long __builtin_arm_wunpckelsw (v2si)
7218 v4hi __builtin_arm_wunpckelub (v8qi)
7219 v2si __builtin_arm_wunpckeluh (v4hi)
7220 long long __builtin_arm_wunpckeluw (v2si)
7221 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
7222 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
7223 v2si __builtin_arm_wunpckihw (v2si, v2si)
7224 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
7225 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
7226 v2si __builtin_arm_wunpckilw (v2si, v2si)
7227 long long __builtin_arm_wxor (long long, long long)
7228 long long __builtin_arm_wzero ()
7231 @node ARM NEON Intrinsics
7232 @subsection ARM NEON Intrinsics
7234 These built-in intrinsics for the ARM Advanced SIMD extension are available
7235 when the @option{-mfpu=neon} switch is used:
7237 @include arm-neon-intrinsics.texi
7239 @node Blackfin Built-in Functions
7240 @subsection Blackfin Built-in Functions
7242 Currently, there are two Blackfin-specific built-in functions. These are
7243 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
7244 using inline assembly; by using these built-in functions the compiler can
7245 automatically add workarounds for hardware errata involving these
7246 instructions. These functions are named as follows:
7249 void __builtin_bfin_csync (void)
7250 void __builtin_bfin_ssync (void)
7253 @node FR-V Built-in Functions
7254 @subsection FR-V Built-in Functions
7256 GCC provides many FR-V-specific built-in functions. In general,
7257 these functions are intended to be compatible with those described
7258 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
7259 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
7260 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
7261 pointer rather than by value.
7263 Most of the functions are named after specific FR-V instructions.
7264 Such functions are said to be ``directly mapped'' and are summarized
7265 here in tabular form.
7269 * Directly-mapped Integer Functions::
7270 * Directly-mapped Media Functions::
7271 * Raw read/write Functions::
7272 * Other Built-in Functions::
7275 @node Argument Types
7276 @subsubsection Argument Types
7278 The arguments to the built-in functions can be divided into three groups:
7279 register numbers, compile-time constants and run-time values. In order
7280 to make this classification clear at a glance, the arguments and return
7281 values are given the following pseudo types:
7283 @multitable @columnfractions .20 .30 .15 .35
7284 @item Pseudo type @tab Real C type @tab Constant? @tab Description
7285 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
7286 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
7287 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
7288 @item @code{uw2} @tab @code{unsigned long long} @tab No
7289 @tab an unsigned doubleword
7290 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
7291 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
7292 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
7293 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
7296 These pseudo types are not defined by GCC, they are simply a notational
7297 convenience used in this manual.
7299 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
7300 and @code{sw2} are evaluated at run time. They correspond to
7301 register operands in the underlying FR-V instructions.
7303 @code{const} arguments represent immediate operands in the underlying
7304 FR-V instructions. They must be compile-time constants.
7306 @code{acc} arguments are evaluated at compile time and specify the number
7307 of an accumulator register. For example, an @code{acc} argument of 2
7308 will select the ACC2 register.
7310 @code{iacc} arguments are similar to @code{acc} arguments but specify the
7311 number of an IACC register. See @pxref{Other Built-in Functions}
7314 @node Directly-mapped Integer Functions
7315 @subsubsection Directly-mapped Integer Functions
7317 The functions listed below map directly to FR-V I-type instructions.
7319 @multitable @columnfractions .45 .32 .23
7320 @item Function prototype @tab Example usage @tab Assembly output
7321 @item @code{sw1 __ADDSS (sw1, sw1)}
7322 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
7323 @tab @code{ADDSS @var{a},@var{b},@var{c}}
7324 @item @code{sw1 __SCAN (sw1, sw1)}
7325 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
7326 @tab @code{SCAN @var{a},@var{b},@var{c}}
7327 @item @code{sw1 __SCUTSS (sw1)}
7328 @tab @code{@var{b} = __SCUTSS (@var{a})}
7329 @tab @code{SCUTSS @var{a},@var{b}}
7330 @item @code{sw1 __SLASS (sw1, sw1)}
7331 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
7332 @tab @code{SLASS @var{a},@var{b},@var{c}}
7333 @item @code{void __SMASS (sw1, sw1)}
7334 @tab @code{__SMASS (@var{a}, @var{b})}
7335 @tab @code{SMASS @var{a},@var{b}}
7336 @item @code{void __SMSSS (sw1, sw1)}
7337 @tab @code{__SMSSS (@var{a}, @var{b})}
7338 @tab @code{SMSSS @var{a},@var{b}}
7339 @item @code{void __SMU (sw1, sw1)}
7340 @tab @code{__SMU (@var{a}, @var{b})}
7341 @tab @code{SMU @var{a},@var{b}}
7342 @item @code{sw2 __SMUL (sw1, sw1)}
7343 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
7344 @tab @code{SMUL @var{a},@var{b},@var{c}}
7345 @item @code{sw1 __SUBSS (sw1, sw1)}
7346 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
7347 @tab @code{SUBSS @var{a},@var{b},@var{c}}
7348 @item @code{uw2 __UMUL (uw1, uw1)}
7349 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
7350 @tab @code{UMUL @var{a},@var{b},@var{c}}
7353 @node Directly-mapped Media Functions
7354 @subsubsection Directly-mapped Media Functions
7356 The functions listed below map directly to FR-V M-type instructions.
7358 @multitable @columnfractions .45 .32 .23
7359 @item Function prototype @tab Example usage @tab Assembly output
7360 @item @code{uw1 __MABSHS (sw1)}
7361 @tab @code{@var{b} = __MABSHS (@var{a})}
7362 @tab @code{MABSHS @var{a},@var{b}}
7363 @item @code{void __MADDACCS (acc, acc)}
7364 @tab @code{__MADDACCS (@var{b}, @var{a})}
7365 @tab @code{MADDACCS @var{a},@var{b}}
7366 @item @code{sw1 __MADDHSS (sw1, sw1)}
7367 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
7368 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
7369 @item @code{uw1 __MADDHUS (uw1, uw1)}
7370 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
7371 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
7372 @item @code{uw1 __MAND (uw1, uw1)}
7373 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
7374 @tab @code{MAND @var{a},@var{b},@var{c}}
7375 @item @code{void __MASACCS (acc, acc)}
7376 @tab @code{__MASACCS (@var{b}, @var{a})}
7377 @tab @code{MASACCS @var{a},@var{b}}
7378 @item @code{uw1 __MAVEH (uw1, uw1)}
7379 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
7380 @tab @code{MAVEH @var{a},@var{b},@var{c}}
7381 @item @code{uw2 __MBTOH (uw1)}
7382 @tab @code{@var{b} = __MBTOH (@var{a})}
7383 @tab @code{MBTOH @var{a},@var{b}}
7384 @item @code{void __MBTOHE (uw1 *, uw1)}
7385 @tab @code{__MBTOHE (&@var{b}, @var{a})}
7386 @tab @code{MBTOHE @var{a},@var{b}}
7387 @item @code{void __MCLRACC (acc)}
7388 @tab @code{__MCLRACC (@var{a})}
7389 @tab @code{MCLRACC @var{a}}
7390 @item @code{void __MCLRACCA (void)}
7391 @tab @code{__MCLRACCA ()}
7392 @tab @code{MCLRACCA}
7393 @item @code{uw1 __Mcop1 (uw1, uw1)}
7394 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
7395 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
7396 @item @code{uw1 __Mcop2 (uw1, uw1)}
7397 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
7398 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
7399 @item @code{uw1 __MCPLHI (uw2, const)}
7400 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
7401 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
7402 @item @code{uw1 __MCPLI (uw2, const)}
7403 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
7404 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
7405 @item @code{void __MCPXIS (acc, sw1, sw1)}
7406 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
7407 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
7408 @item @code{void __MCPXIU (acc, uw1, uw1)}
7409 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
7410 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
7411 @item @code{void __MCPXRS (acc, sw1, sw1)}
7412 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
7413 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
7414 @item @code{void __MCPXRU (acc, uw1, uw1)}
7415 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
7416 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
7417 @item @code{uw1 __MCUT (acc, uw1)}
7418 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
7419 @tab @code{MCUT @var{a},@var{b},@var{c}}
7420 @item @code{uw1 __MCUTSS (acc, sw1)}
7421 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
7422 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
7423 @item @code{void __MDADDACCS (acc, acc)}
7424 @tab @code{__MDADDACCS (@var{b}, @var{a})}
7425 @tab @code{MDADDACCS @var{a},@var{b}}
7426 @item @code{void __MDASACCS (acc, acc)}
7427 @tab @code{__MDASACCS (@var{b}, @var{a})}
7428 @tab @code{MDASACCS @var{a},@var{b}}
7429 @item @code{uw2 __MDCUTSSI (acc, const)}
7430 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
7431 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
7432 @item @code{uw2 __MDPACKH (uw2, uw2)}
7433 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
7434 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
7435 @item @code{uw2 __MDROTLI (uw2, const)}
7436 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
7437 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
7438 @item @code{void __MDSUBACCS (acc, acc)}
7439 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
7440 @tab @code{MDSUBACCS @var{a},@var{b}}
7441 @item @code{void __MDUNPACKH (uw1 *, uw2)}
7442 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
7443 @tab @code{MDUNPACKH @var{a},@var{b}}
7444 @item @code{uw2 __MEXPDHD (uw1, const)}
7445 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
7446 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
7447 @item @code{uw1 __MEXPDHW (uw1, const)}
7448 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
7449 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
7450 @item @code{uw1 __MHDSETH (uw1, const)}
7451 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
7452 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
7453 @item @code{sw1 __MHDSETS (const)}
7454 @tab @code{@var{b} = __MHDSETS (@var{a})}
7455 @tab @code{MHDSETS #@var{a},@var{b}}
7456 @item @code{uw1 __MHSETHIH (uw1, const)}
7457 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
7458 @tab @code{MHSETHIH #@var{a},@var{b}}
7459 @item @code{sw1 __MHSETHIS (sw1, const)}
7460 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
7461 @tab @code{MHSETHIS #@var{a},@var{b}}
7462 @item @code{uw1 __MHSETLOH (uw1, const)}
7463 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
7464 @tab @code{MHSETLOH #@var{a},@var{b}}
7465 @item @code{sw1 __MHSETLOS (sw1, const)}
7466 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
7467 @tab @code{MHSETLOS #@var{a},@var{b}}
7468 @item @code{uw1 __MHTOB (uw2)}
7469 @tab @code{@var{b} = __MHTOB (@var{a})}
7470 @tab @code{MHTOB @var{a},@var{b}}
7471 @item @code{void __MMACHS (acc, sw1, sw1)}
7472 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
7473 @tab @code{MMACHS @var{a},@var{b},@var{c}}
7474 @item @code{void __MMACHU (acc, uw1, uw1)}
7475 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
7476 @tab @code{MMACHU @var{a},@var{b},@var{c}}
7477 @item @code{void __MMRDHS (acc, sw1, sw1)}
7478 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
7479 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
7480 @item @code{void __MMRDHU (acc, uw1, uw1)}
7481 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
7482 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
7483 @item @code{void __MMULHS (acc, sw1, sw1)}
7484 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
7485 @tab @code{MMULHS @var{a},@var{b},@var{c}}
7486 @item @code{void __MMULHU (acc, uw1, uw1)}
7487 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
7488 @tab @code{MMULHU @var{a},@var{b},@var{c}}
7489 @item @code{void __MMULXHS (acc, sw1, sw1)}
7490 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
7491 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
7492 @item @code{void __MMULXHU (acc, uw1, uw1)}
7493 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
7494 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
7495 @item @code{uw1 __MNOT (uw1)}
7496 @tab @code{@var{b} = __MNOT (@var{a})}
7497 @tab @code{MNOT @var{a},@var{b}}
7498 @item @code{uw1 __MOR (uw1, uw1)}
7499 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
7500 @tab @code{MOR @var{a},@var{b},@var{c}}
7501 @item @code{uw1 __MPACKH (uh, uh)}
7502 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
7503 @tab @code{MPACKH @var{a},@var{b},@var{c}}
7504 @item @code{sw2 __MQADDHSS (sw2, sw2)}
7505 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
7506 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
7507 @item @code{uw2 __MQADDHUS (uw2, uw2)}
7508 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
7509 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
7510 @item @code{void __MQCPXIS (acc, sw2, sw2)}
7511 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
7512 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
7513 @item @code{void __MQCPXIU (acc, uw2, uw2)}
7514 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
7515 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
7516 @item @code{void __MQCPXRS (acc, sw2, sw2)}
7517 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
7518 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
7519 @item @code{void __MQCPXRU (acc, uw2, uw2)}
7520 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
7521 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
7522 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
7523 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
7524 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
7525 @item @code{sw2 __MQLMTHS (sw2, sw2)}
7526 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
7527 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
7528 @item @code{void __MQMACHS (acc, sw2, sw2)}
7529 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
7530 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
7531 @item @code{void __MQMACHU (acc, uw2, uw2)}
7532 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
7533 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
7534 @item @code{void __MQMACXHS (acc, sw2, sw2)}
7535 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
7536 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
7537 @item @code{void __MQMULHS (acc, sw2, sw2)}
7538 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
7539 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
7540 @item @code{void __MQMULHU (acc, uw2, uw2)}
7541 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
7542 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
7543 @item @code{void __MQMULXHS (acc, sw2, sw2)}
7544 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
7545 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
7546 @item @code{void __MQMULXHU (acc, uw2, uw2)}
7547 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
7548 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
7549 @item @code{sw2 __MQSATHS (sw2, sw2)}
7550 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
7551 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
7552 @item @code{uw2 __MQSLLHI (uw2, int)}
7553 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
7554 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
7555 @item @code{sw2 __MQSRAHI (sw2, int)}
7556 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
7557 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
7558 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
7559 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
7560 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
7561 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
7562 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
7563 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
7564 @item @code{void __MQXMACHS (acc, sw2, sw2)}
7565 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
7566 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
7567 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
7568 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
7569 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
7570 @item @code{uw1 __MRDACC (acc)}
7571 @tab @code{@var{b} = __MRDACC (@var{a})}
7572 @tab @code{MRDACC @var{a},@var{b}}
7573 @item @code{uw1 __MRDACCG (acc)}
7574 @tab @code{@var{b} = __MRDACCG (@var{a})}
7575 @tab @code{MRDACCG @var{a},@var{b}}
7576 @item @code{uw1 __MROTLI (uw1, const)}
7577 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
7578 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
7579 @item @code{uw1 __MROTRI (uw1, const)}
7580 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
7581 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
7582 @item @code{sw1 __MSATHS (sw1, sw1)}
7583 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
7584 @tab @code{MSATHS @var{a},@var{b},@var{c}}
7585 @item @code{uw1 __MSATHU (uw1, uw1)}
7586 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
7587 @tab @code{MSATHU @var{a},@var{b},@var{c}}
7588 @item @code{uw1 __MSLLHI (uw1, const)}
7589 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
7590 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
7591 @item @code{sw1 __MSRAHI (sw1, const)}
7592 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
7593 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
7594 @item @code{uw1 __MSRLHI (uw1, const)}
7595 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
7596 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
7597 @item @code{void __MSUBACCS (acc, acc)}
7598 @tab @code{__MSUBACCS (@var{b}, @var{a})}
7599 @tab @code{MSUBACCS @var{a},@var{b}}
7600 @item @code{sw1 __MSUBHSS (sw1, sw1)}
7601 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
7602 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
7603 @item @code{uw1 __MSUBHUS (uw1, uw1)}
7604 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
7605 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
7606 @item @code{void __MTRAP (void)}
7607 @tab @code{__MTRAP ()}
7609 @item @code{uw2 __MUNPACKH (uw1)}
7610 @tab @code{@var{b} = __MUNPACKH (@var{a})}
7611 @tab @code{MUNPACKH @var{a},@var{b}}
7612 @item @code{uw1 __MWCUT (uw2, uw1)}
7613 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
7614 @tab @code{MWCUT @var{a},@var{b},@var{c}}
7615 @item @code{void __MWTACC (acc, uw1)}
7616 @tab @code{__MWTACC (@var{b}, @var{a})}
7617 @tab @code{MWTACC @var{a},@var{b}}
7618 @item @code{void __MWTACCG (acc, uw1)}
7619 @tab @code{__MWTACCG (@var{b}, @var{a})}
7620 @tab @code{MWTACCG @var{a},@var{b}}
7621 @item @code{uw1 __MXOR (uw1, uw1)}
7622 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
7623 @tab @code{MXOR @var{a},@var{b},@var{c}}
7626 @node Raw read/write Functions
7627 @subsubsection Raw read/write Functions
7629 This sections describes built-in functions related to read and write
7630 instructions to access memory. These functions generate
7631 @code{membar} instructions to flush the I/O load and stores where
7632 appropriate, as described in Fujitsu's manual described above.
7636 @item unsigned char __builtin_read8 (void *@var{data})
7637 @item unsigned short __builtin_read16 (void *@var{data})
7638 @item unsigned long __builtin_read32 (void *@var{data})
7639 @item unsigned long long __builtin_read64 (void *@var{data})
7641 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
7642 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
7643 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
7644 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
7647 @node Other Built-in Functions
7648 @subsubsection Other Built-in Functions
7650 This section describes built-in functions that are not named after
7651 a specific FR-V instruction.
7654 @item sw2 __IACCreadll (iacc @var{reg})
7655 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
7656 for future expansion and must be 0.
7658 @item sw1 __IACCreadl (iacc @var{reg})
7659 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
7660 Other values of @var{reg} are rejected as invalid.
7662 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
7663 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
7664 is reserved for future expansion and must be 0.
7666 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
7667 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
7668 is 1. Other values of @var{reg} are rejected as invalid.
7670 @item void __data_prefetch0 (const void *@var{x})
7671 Use the @code{dcpl} instruction to load the contents of address @var{x}
7672 into the data cache.
7674 @item void __data_prefetch (const void *@var{x})
7675 Use the @code{nldub} instruction to load the contents of address @var{x}
7676 into the data cache. The instruction will be issued in slot I1@.
7679 @node X86 Built-in Functions
7680 @subsection X86 Built-in Functions
7682 These built-in functions are available for the i386 and x86-64 family
7683 of computers, depending on the command-line switches used.
7685 Note that, if you specify command-line switches such as @option{-msse},
7686 the compiler could use the extended instruction sets even if the built-ins
7687 are not used explicitly in the program. For this reason, applications
7688 which perform runtime CPU detection must compile separate files for each
7689 supported architecture, using the appropriate flags. In particular,
7690 the file containing the CPU detection code should be compiled without
7693 The following machine modes are available for use with MMX built-in functions
7694 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
7695 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
7696 vector of eight 8-bit integers. Some of the built-in functions operate on
7697 MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode.
7699 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
7700 of two 32-bit floating point values.
7702 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
7703 floating point values. Some instructions use a vector of four 32-bit
7704 integers, these use @code{V4SI}. Finally, some instructions operate on an
7705 entire vector register, interpreting it as a 128-bit integer, these use mode
7708 In 64-bit mode, the x86-64 family of processors uses additional built-in
7709 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
7710 floating point and @code{TC} 128-bit complex floating point values.
7712 The following floating point built-in functions are available in 64-bit
7713 mode. All of them implement the function that is part of the name.
7716 __float128 __builtin_fabsq (__float128)
7717 __float128 __builtin_copysignq (__float128, __float128)
7720 The following floating point built-in functions are made available in the
7724 @item __float128 __builtin_infq (void)
7725 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
7728 The following built-in functions are made available by @option{-mmmx}.
7729 All of them generate the machine instruction that is part of the name.
7732 v8qi __builtin_ia32_paddb (v8qi, v8qi)
7733 v4hi __builtin_ia32_paddw (v4hi, v4hi)
7734 v2si __builtin_ia32_paddd (v2si, v2si)
7735 v8qi __builtin_ia32_psubb (v8qi, v8qi)
7736 v4hi __builtin_ia32_psubw (v4hi, v4hi)
7737 v2si __builtin_ia32_psubd (v2si, v2si)
7738 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
7739 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
7740 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
7741 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
7742 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
7743 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
7744 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
7745 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
7746 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
7747 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
7748 di __builtin_ia32_pand (di, di)
7749 di __builtin_ia32_pandn (di,di)
7750 di __builtin_ia32_por (di, di)
7751 di __builtin_ia32_pxor (di, di)
7752 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7753 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7754 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7755 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7756 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7757 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7758 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7759 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7760 v2si __builtin_ia32_punpckhdq (v2si, v2si)
7761 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7762 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7763 v2si __builtin_ia32_punpckldq (v2si, v2si)
7764 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7765 v4hi __builtin_ia32_packssdw (v2si, v2si)
7766 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7768 v4hi __builtin_ia32_psllw (v4hi, v4hi)
7769 v2si __builtin_ia32_pslld (v2si, v2si)
7770 v1di __builtin_ia32_psllq (v1di, v1di)
7771 v4hi __builtin_ia32_psrlw (v4hi, v4hi)
7772 v2si __builtin_ia32_psrld (v2si, v2si)
7773 v1di __builtin_ia32_psrlq (v1di, v1di)
7774 v4hi __builtin_ia32_psraw (v4hi, v4hi)
7775 v2si __builtin_ia32_psrad (v2si, v2si)
7776 v4hi __builtin_ia32_psllwi (v4hi, int)
7777 v2si __builtin_ia32_pslldi (v2si, int)
7778 v1di __builtin_ia32_psllqi (v1di, int)
7779 v4hi __builtin_ia32_psrlwi (v4hi, int)
7780 v2si __builtin_ia32_psrldi (v2si, int)
7781 v1di __builtin_ia32_psrlqi (v1di, int)
7782 v4hi __builtin_ia32_psrawi (v4hi, int)
7783 v2si __builtin_ia32_psradi (v2si, int)
7787 The following built-in functions are made available either with
7788 @option{-msse}, or with a combination of @option{-m3dnow} and
7789 @option{-march=athlon}. All of them generate the machine
7790 instruction that is part of the name.
7793 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7794 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7795 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7796 v1di __builtin_ia32_psadbw (v8qi, v8qi)
7797 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7798 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7799 v8qi __builtin_ia32_pminub (v8qi, v8qi)
7800 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7801 int __builtin_ia32_pextrw (v4hi, int)
7802 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7803 int __builtin_ia32_pmovmskb (v8qi)
7804 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7805 void __builtin_ia32_movntq (di *, di)
7806 void __builtin_ia32_sfence (void)
7809 The following built-in functions are available when @option{-msse} is used.
7810 All of them generate the machine instruction that is part of the name.
7813 int __builtin_ia32_comieq (v4sf, v4sf)
7814 int __builtin_ia32_comineq (v4sf, v4sf)
7815 int __builtin_ia32_comilt (v4sf, v4sf)
7816 int __builtin_ia32_comile (v4sf, v4sf)
7817 int __builtin_ia32_comigt (v4sf, v4sf)
7818 int __builtin_ia32_comige (v4sf, v4sf)
7819 int __builtin_ia32_ucomieq (v4sf, v4sf)
7820 int __builtin_ia32_ucomineq (v4sf, v4sf)
7821 int __builtin_ia32_ucomilt (v4sf, v4sf)
7822 int __builtin_ia32_ucomile (v4sf, v4sf)
7823 int __builtin_ia32_ucomigt (v4sf, v4sf)
7824 int __builtin_ia32_ucomige (v4sf, v4sf)
7825 v4sf __builtin_ia32_addps (v4sf, v4sf)
7826 v4sf __builtin_ia32_subps (v4sf, v4sf)
7827 v4sf __builtin_ia32_mulps (v4sf, v4sf)
7828 v4sf __builtin_ia32_divps (v4sf, v4sf)
7829 v4sf __builtin_ia32_addss (v4sf, v4sf)
7830 v4sf __builtin_ia32_subss (v4sf, v4sf)
7831 v4sf __builtin_ia32_mulss (v4sf, v4sf)
7832 v4sf __builtin_ia32_divss (v4sf, v4sf)
7833 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
7834 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
7835 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
7836 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
7837 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
7838 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
7839 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
7840 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
7841 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
7842 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
7843 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
7844 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
7845 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
7846 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
7847 v4si __builtin_ia32_cmpless (v4sf, v4sf)
7848 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
7849 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
7850 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
7851 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
7852 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
7853 v4sf __builtin_ia32_maxps (v4sf, v4sf)
7854 v4sf __builtin_ia32_maxss (v4sf, v4sf)
7855 v4sf __builtin_ia32_minps (v4sf, v4sf)
7856 v4sf __builtin_ia32_minss (v4sf, v4sf)
7857 v4sf __builtin_ia32_andps (v4sf, v4sf)
7858 v4sf __builtin_ia32_andnps (v4sf, v4sf)
7859 v4sf __builtin_ia32_orps (v4sf, v4sf)
7860 v4sf __builtin_ia32_xorps (v4sf, v4sf)
7861 v4sf __builtin_ia32_movss (v4sf, v4sf)
7862 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
7863 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
7864 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
7865 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
7866 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
7867 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
7868 v2si __builtin_ia32_cvtps2pi (v4sf)
7869 int __builtin_ia32_cvtss2si (v4sf)
7870 v2si __builtin_ia32_cvttps2pi (v4sf)
7871 int __builtin_ia32_cvttss2si (v4sf)
7872 v4sf __builtin_ia32_rcpps (v4sf)
7873 v4sf __builtin_ia32_rsqrtps (v4sf)
7874 v4sf __builtin_ia32_sqrtps (v4sf)
7875 v4sf __builtin_ia32_rcpss (v4sf)
7876 v4sf __builtin_ia32_rsqrtss (v4sf)
7877 v4sf __builtin_ia32_sqrtss (v4sf)
7878 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
7879 void __builtin_ia32_movntps (float *, v4sf)
7880 int __builtin_ia32_movmskps (v4sf)
7883 The following built-in functions are available when @option{-msse} is used.
7886 @item v4sf __builtin_ia32_loadaps (float *)
7887 Generates the @code{movaps} machine instruction as a load from memory.
7888 @item void __builtin_ia32_storeaps (float *, v4sf)
7889 Generates the @code{movaps} machine instruction as a store to memory.
7890 @item v4sf __builtin_ia32_loadups (float *)
7891 Generates the @code{movups} machine instruction as a load from memory.
7892 @item void __builtin_ia32_storeups (float *, v4sf)
7893 Generates the @code{movups} machine instruction as a store to memory.
7894 @item v4sf __builtin_ia32_loadsss (float *)
7895 Generates the @code{movss} machine instruction as a load from memory.
7896 @item void __builtin_ia32_storess (float *, v4sf)
7897 Generates the @code{movss} machine instruction as a store to memory.
7898 @item v4sf __builtin_ia32_loadhps (v4sf, const v2sf *)
7899 Generates the @code{movhps} machine instruction as a load from memory.
7900 @item v4sf __builtin_ia32_loadlps (v4sf, const v2sf *)
7901 Generates the @code{movlps} machine instruction as a load from memory
7902 @item void __builtin_ia32_storehps (v2sf *, v4sf)
7903 Generates the @code{movhps} machine instruction as a store to memory.
7904 @item void __builtin_ia32_storelps (v2sf *, v4sf)
7905 Generates the @code{movlps} machine instruction as a store to memory.
7908 The following built-in functions are available when @option{-msse2} is used.
7909 All of them generate the machine instruction that is part of the name.
7912 int __builtin_ia32_comisdeq (v2df, v2df)
7913 int __builtin_ia32_comisdlt (v2df, v2df)
7914 int __builtin_ia32_comisdle (v2df, v2df)
7915 int __builtin_ia32_comisdgt (v2df, v2df)
7916 int __builtin_ia32_comisdge (v2df, v2df)
7917 int __builtin_ia32_comisdneq (v2df, v2df)
7918 int __builtin_ia32_ucomisdeq (v2df, v2df)
7919 int __builtin_ia32_ucomisdlt (v2df, v2df)
7920 int __builtin_ia32_ucomisdle (v2df, v2df)
7921 int __builtin_ia32_ucomisdgt (v2df, v2df)
7922 int __builtin_ia32_ucomisdge (v2df, v2df)
7923 int __builtin_ia32_ucomisdneq (v2df, v2df)
7924 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
7925 v2df __builtin_ia32_cmpltpd (v2df, v2df)
7926 v2df __builtin_ia32_cmplepd (v2df, v2df)
7927 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
7928 v2df __builtin_ia32_cmpgepd (v2df, v2df)
7929 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
7930 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
7931 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
7932 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
7933 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
7934 v2df __builtin_ia32_cmpngepd (v2df, v2df)
7935 v2df __builtin_ia32_cmpordpd (v2df, v2df)
7936 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
7937 v2df __builtin_ia32_cmpltsd (v2df, v2df)
7938 v2df __builtin_ia32_cmplesd (v2df, v2df)
7939 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
7940 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
7941 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
7942 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
7943 v2df __builtin_ia32_cmpordsd (v2df, v2df)
7944 v2di __builtin_ia32_paddq (v2di, v2di)
7945 v2di __builtin_ia32_psubq (v2di, v2di)
7946 v2df __builtin_ia32_addpd (v2df, v2df)
7947 v2df __builtin_ia32_subpd (v2df, v2df)
7948 v2df __builtin_ia32_mulpd (v2df, v2df)
7949 v2df __builtin_ia32_divpd (v2df, v2df)
7950 v2df __builtin_ia32_addsd (v2df, v2df)
7951 v2df __builtin_ia32_subsd (v2df, v2df)
7952 v2df __builtin_ia32_mulsd (v2df, v2df)
7953 v2df __builtin_ia32_divsd (v2df, v2df)
7954 v2df __builtin_ia32_minpd (v2df, v2df)
7955 v2df __builtin_ia32_maxpd (v2df, v2df)
7956 v2df __builtin_ia32_minsd (v2df, v2df)
7957 v2df __builtin_ia32_maxsd (v2df, v2df)
7958 v2df __builtin_ia32_andpd (v2df, v2df)
7959 v2df __builtin_ia32_andnpd (v2df, v2df)
7960 v2df __builtin_ia32_orpd (v2df, v2df)
7961 v2df __builtin_ia32_xorpd (v2df, v2df)
7962 v2df __builtin_ia32_movsd (v2df, v2df)
7963 v2df __builtin_ia32_unpckhpd (v2df, v2df)
7964 v2df __builtin_ia32_unpcklpd (v2df, v2df)
7965 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
7966 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
7967 v4si __builtin_ia32_paddd128 (v4si, v4si)
7968 v2di __builtin_ia32_paddq128 (v2di, v2di)
7969 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
7970 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
7971 v4si __builtin_ia32_psubd128 (v4si, v4si)
7972 v2di __builtin_ia32_psubq128 (v2di, v2di)
7973 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
7974 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
7975 v2di __builtin_ia32_pand128 (v2di, v2di)
7976 v2di __builtin_ia32_pandn128 (v2di, v2di)
7977 v2di __builtin_ia32_por128 (v2di, v2di)
7978 v2di __builtin_ia32_pxor128 (v2di, v2di)
7979 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
7980 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
7981 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
7982 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
7983 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
7984 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
7985 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
7986 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
7987 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
7988 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
7989 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
7990 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
7991 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
7992 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
7993 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
7994 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
7995 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
7996 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
7997 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
7998 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
7999 v16qi __builtin_ia32_packsswb128 (v8hi, v8hi)
8000 v8hi __builtin_ia32_packssdw128 (v4si, v4si)
8001 v16qi __builtin_ia32_packuswb128 (v8hi, v8hi)
8002 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
8003 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
8004 v2df __builtin_ia32_loadupd (double *)
8005 void __builtin_ia32_storeupd (double *, v2df)
8006 v2df __builtin_ia32_loadhpd (v2df, double const *)
8007 v2df __builtin_ia32_loadlpd (v2df, double const *)
8008 int __builtin_ia32_movmskpd (v2df)
8009 int __builtin_ia32_pmovmskb128 (v16qi)
8010 void __builtin_ia32_movnti (int *, int)
8011 void __builtin_ia32_movntpd (double *, v2df)
8012 void __builtin_ia32_movntdq (v2df *, v2df)
8013 v4si __builtin_ia32_pshufd (v4si, int)
8014 v8hi __builtin_ia32_pshuflw (v8hi, int)
8015 v8hi __builtin_ia32_pshufhw (v8hi, int)
8016 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
8017 v2df __builtin_ia32_sqrtpd (v2df)
8018 v2df __builtin_ia32_sqrtsd (v2df)
8019 v2df __builtin_ia32_shufpd (v2df, v2df, int)
8020 v2df __builtin_ia32_cvtdq2pd (v4si)
8021 v4sf __builtin_ia32_cvtdq2ps (v4si)
8022 v4si __builtin_ia32_cvtpd2dq (v2df)
8023 v2si __builtin_ia32_cvtpd2pi (v2df)
8024 v4sf __builtin_ia32_cvtpd2ps (v2df)
8025 v4si __builtin_ia32_cvttpd2dq (v2df)
8026 v2si __builtin_ia32_cvttpd2pi (v2df)
8027 v2df __builtin_ia32_cvtpi2pd (v2si)
8028 int __builtin_ia32_cvtsd2si (v2df)
8029 int __builtin_ia32_cvttsd2si (v2df)
8030 long long __builtin_ia32_cvtsd2si64 (v2df)
8031 long long __builtin_ia32_cvttsd2si64 (v2df)
8032 v4si __builtin_ia32_cvtps2dq (v4sf)
8033 v2df __builtin_ia32_cvtps2pd (v4sf)
8034 v4si __builtin_ia32_cvttps2dq (v4sf)
8035 v2df __builtin_ia32_cvtsi2sd (v2df, int)
8036 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
8037 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
8038 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
8039 void __builtin_ia32_clflush (const void *)
8040 void __builtin_ia32_lfence (void)
8041 void __builtin_ia32_mfence (void)
8042 v16qi __builtin_ia32_loaddqu (const char *)
8043 void __builtin_ia32_storedqu (char *, v16qi)
8044 v1di __builtin_ia32_pmuludq (v2si, v2si)
8045 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
8046 v8hi __builtin_ia32_psllw128 (v8hi, v8hi)
8047 v4si __builtin_ia32_pslld128 (v4si, v4si)
8048 v2di __builtin_ia32_psllq128 (v2di, v2di)
8049 v8hi __builtin_ia32_psrlw128 (v8hi, v8hi)
8050 v4si __builtin_ia32_psrld128 (v4si, v4si)
8051 v2di __builtin_ia32_psrlq128 (v2di, v2di)
8052 v8hi __builtin_ia32_psraw128 (v8hi, v8hi)
8053 v4si __builtin_ia32_psrad128 (v4si, v4si)
8054 v2di __builtin_ia32_pslldqi128 (v2di, int)
8055 v8hi __builtin_ia32_psllwi128 (v8hi, int)
8056 v4si __builtin_ia32_pslldi128 (v4si, int)
8057 v2di __builtin_ia32_psllqi128 (v2di, int)
8058 v2di __builtin_ia32_psrldqi128 (v2di, int)
8059 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
8060 v4si __builtin_ia32_psrldi128 (v4si, int)
8061 v2di __builtin_ia32_psrlqi128 (v2di, int)
8062 v8hi __builtin_ia32_psrawi128 (v8hi, int)
8063 v4si __builtin_ia32_psradi128 (v4si, int)
8064 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
8065 v2di __builtin_ia32_movq128 (v2di)
8068 The following built-in functions are available when @option{-msse3} is used.
8069 All of them generate the machine instruction that is part of the name.
8072 v2df __builtin_ia32_addsubpd (v2df, v2df)
8073 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
8074 v2df __builtin_ia32_haddpd (v2df, v2df)
8075 v4sf __builtin_ia32_haddps (v4sf, v4sf)
8076 v2df __builtin_ia32_hsubpd (v2df, v2df)
8077 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
8078 v16qi __builtin_ia32_lddqu (char const *)
8079 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
8080 v2df __builtin_ia32_movddup (v2df)
8081 v4sf __builtin_ia32_movshdup (v4sf)
8082 v4sf __builtin_ia32_movsldup (v4sf)
8083 void __builtin_ia32_mwait (unsigned int, unsigned int)
8086 The following built-in functions are available when @option{-msse3} is used.
8089 @item v2df __builtin_ia32_loadddup (double const *)
8090 Generates the @code{movddup} machine instruction as a load from memory.
8093 The following built-in functions are available when @option{-mssse3} is used.
8094 All of them generate the machine instruction that is part of the name
8098 v2si __builtin_ia32_phaddd (v2si, v2si)
8099 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
8100 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
8101 v2si __builtin_ia32_phsubd (v2si, v2si)
8102 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
8103 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
8104 v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi)
8105 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
8106 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
8107 v8qi __builtin_ia32_psignb (v8qi, v8qi)
8108 v2si __builtin_ia32_psignd (v2si, v2si)
8109 v4hi __builtin_ia32_psignw (v4hi, v4hi)
8110 v1di __builtin_ia32_palignr (v1di, v1di, int)
8111 v8qi __builtin_ia32_pabsb (v8qi)
8112 v2si __builtin_ia32_pabsd (v2si)
8113 v4hi __builtin_ia32_pabsw (v4hi)
8116 The following built-in functions are available when @option{-mssse3} is used.
8117 All of them generate the machine instruction that is part of the name
8121 v4si __builtin_ia32_phaddd128 (v4si, v4si)
8122 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
8123 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
8124 v4si __builtin_ia32_phsubd128 (v4si, v4si)
8125 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
8126 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
8127 v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
8128 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
8129 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
8130 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
8131 v4si __builtin_ia32_psignd128 (v4si, v4si)
8132 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
8133 v2di __builtin_ia32_palignr128 (v2di, v2di, int)
8134 v16qi __builtin_ia32_pabsb128 (v16qi)
8135 v4si __builtin_ia32_pabsd128 (v4si)
8136 v8hi __builtin_ia32_pabsw128 (v8hi)
8139 The following built-in functions are available when @option{-msse4.1} is
8140 used. All of them generate the machine instruction that is part of the
8144 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
8145 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
8146 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
8147 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
8148 v2df __builtin_ia32_dppd (v2df, v2df, const int)
8149 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
8150 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
8151 v2di __builtin_ia32_movntdqa (v2di *);
8152 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
8153 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
8154 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
8155 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
8156 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
8157 v8hi __builtin_ia32_phminposuw128 (v8hi)
8158 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
8159 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
8160 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
8161 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
8162 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
8163 v4si __builtin_ia32_pminsd128 (v4si, v4si)
8164 v4si __builtin_ia32_pminud128 (v4si, v4si)
8165 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
8166 v4si __builtin_ia32_pmovsxbd128 (v16qi)
8167 v2di __builtin_ia32_pmovsxbq128 (v16qi)
8168 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
8169 v2di __builtin_ia32_pmovsxdq128 (v4si)
8170 v4si __builtin_ia32_pmovsxwd128 (v8hi)
8171 v2di __builtin_ia32_pmovsxwq128 (v8hi)
8172 v4si __builtin_ia32_pmovzxbd128 (v16qi)
8173 v2di __builtin_ia32_pmovzxbq128 (v16qi)
8174 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
8175 v2di __builtin_ia32_pmovzxdq128 (v4si)
8176 v4si __builtin_ia32_pmovzxwd128 (v8hi)
8177 v2di __builtin_ia32_pmovzxwq128 (v8hi)
8178 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
8179 v4si __builtin_ia32_pmulld128 (v4si, v4si)
8180 int __builtin_ia32_ptestc128 (v2di, v2di)
8181 int __builtin_ia32_ptestnzc128 (v2di, v2di)
8182 int __builtin_ia32_ptestz128 (v2di, v2di)
8183 v2df __builtin_ia32_roundpd (v2df, const int)
8184 v4sf __builtin_ia32_roundps (v4sf, const int)
8185 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
8186 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
8189 The following built-in functions are available when @option{-msse4.1} is
8193 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
8194 Generates the @code{insertps} machine instruction.
8195 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
8196 Generates the @code{pextrb} machine instruction.
8197 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
8198 Generates the @code{pinsrb} machine instruction.
8199 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
8200 Generates the @code{pinsrd} machine instruction.
8201 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
8202 Generates the @code{pinsrq} machine instruction in 64bit mode.
8205 The following built-in functions are changed to generate new SSE4.1
8206 instructions when @option{-msse4.1} is used.
8209 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
8210 Generates the @code{extractps} machine instruction.
8211 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
8212 Generates the @code{pextrd} machine instruction.
8213 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
8214 Generates the @code{pextrq} machine instruction in 64bit mode.
8217 The following built-in functions are available when @option{-msse4.2} is
8218 used. All of them generate the machine instruction that is part of the
8222 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
8223 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
8224 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
8225 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
8226 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
8227 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
8228 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
8229 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
8230 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
8231 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
8232 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
8233 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
8234 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
8235 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
8236 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
8239 The following built-in functions are available when @option{-msse4.2} is
8243 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
8244 Generates the @code{crc32b} machine instruction.
8245 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
8246 Generates the @code{crc32w} machine instruction.
8247 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
8248 Generates the @code{crc32l} machine instruction.
8249 @item unsigned long long __builtin_ia32_crc32di (unsigned long long, unsigned long long)
8252 The following built-in functions are changed to generate new SSE4.2
8253 instructions when @option{-msse4.2} is used.
8256 @item int __builtin_popcount (unsigned int)
8257 Generates the @code{popcntl} machine instruction.
8258 @item int __builtin_popcountl (unsigned long)
8259 Generates the @code{popcntl} or @code{popcntq} machine instruction,
8260 depending on the size of @code{unsigned long}.
8261 @item int __builtin_popcountll (unsigned long long)
8262 Generates the @code{popcntq} machine instruction.
8265 The following built-in functions are available when @option{-mavx} is
8266 used. All of them generate the machine instruction that is part of the
8270 v4df __builtin_ia32_addpd256 (v4df,v4df)
8271 v8sf __builtin_ia32_addps256 (v8sf,v8sf)
8272 v4df __builtin_ia32_addsubpd256 (v4df,v4df)
8273 v8sf __builtin_ia32_addsubps256 (v8sf,v8sf)
8274 v4df __builtin_ia32_andnpd256 (v4df,v4df)
8275 v8sf __builtin_ia32_andnps256 (v8sf,v8sf)
8276 v4df __builtin_ia32_andpd256 (v4df,v4df)
8277 v8sf __builtin_ia32_andps256 (v8sf,v8sf)
8278 v4df __builtin_ia32_blendpd256 (v4df,v4df,int)
8279 v8sf __builtin_ia32_blendps256 (v8sf,v8sf,int)
8280 v4df __builtin_ia32_blendvpd256 (v4df,v4df,v4df)
8281 v8sf __builtin_ia32_blendvps256 (v8sf,v8sf,v8sf)
8282 v2df __builtin_ia32_cmppd (v2df,v2df,int)
8283 v4df __builtin_ia32_cmppd256 (v4df,v4df,int)
8284 v4sf __builtin_ia32_cmpps (v4sf,v4sf,int)
8285 v8sf __builtin_ia32_cmpps256 (v8sf,v8sf,int)
8286 v2df __builtin_ia32_cmpsd (v2df,v2df,int)
8287 v4sf __builtin_ia32_cmpss (v4sf,v4sf,int)
8288 v4df __builtin_ia32_cvtdq2pd256 (v4si)
8289 v8sf __builtin_ia32_cvtdq2ps256 (v8si)
8290 v4si __builtin_ia32_cvtpd2dq256 (v4df)
8291 v4sf __builtin_ia32_cvtpd2ps256 (v4df)
8292 v8si __builtin_ia32_cvtps2dq256 (v8sf)
8293 v4df __builtin_ia32_cvtps2pd256 (v4sf)
8294 v4si __builtin_ia32_cvttpd2dq256 (v4df)
8295 v8si __builtin_ia32_cvttps2dq256 (v8sf)
8296 v4df __builtin_ia32_divpd256 (v4df,v4df)
8297 v8sf __builtin_ia32_divps256 (v8sf,v8sf)
8298 v8sf __builtin_ia32_dpps256 (v8sf,v8sf,int)
8299 v4df __builtin_ia32_haddpd256 (v4df,v4df)
8300 v8sf __builtin_ia32_haddps256 (v8sf,v8sf)
8301 v4df __builtin_ia32_hsubpd256 (v4df,v4df)
8302 v8sf __builtin_ia32_hsubps256 (v8sf,v8sf)
8303 v32qi __builtin_ia32_lddqu256 (pcchar)
8304 v32qi __builtin_ia32_loaddqu256 (pcchar)
8305 v4df __builtin_ia32_loadupd256 (pcdouble)
8306 v8sf __builtin_ia32_loadups256 (pcfloat)
8307 v2df __builtin_ia32_maskloadpd (pcv2df,v2df)
8308 v4df __builtin_ia32_maskloadpd256 (pcv4df,v4df)
8309 v4sf __builtin_ia32_maskloadps (pcv4sf,v4sf)
8310 v8sf __builtin_ia32_maskloadps256 (pcv8sf,v8sf)
8311 void __builtin_ia32_maskstorepd (pv2df,v2df,v2df)
8312 void __builtin_ia32_maskstorepd256 (pv4df,v4df,v4df)
8313 void __builtin_ia32_maskstoreps (pv4sf,v4sf,v4sf)
8314 void __builtin_ia32_maskstoreps256 (pv8sf,v8sf,v8sf)
8315 v4df __builtin_ia32_maxpd256 (v4df,v4df)
8316 v8sf __builtin_ia32_maxps256 (v8sf,v8sf)
8317 v4df __builtin_ia32_minpd256 (v4df,v4df)
8318 v8sf __builtin_ia32_minps256 (v8sf,v8sf)
8319 v4df __builtin_ia32_movddup256 (v4df)
8320 int __builtin_ia32_movmskpd256 (v4df)
8321 int __builtin_ia32_movmskps256 (v8sf)
8322 v8sf __builtin_ia32_movshdup256 (v8sf)
8323 v8sf __builtin_ia32_movsldup256 (v8sf)
8324 v4df __builtin_ia32_mulpd256 (v4df,v4df)
8325 v8sf __builtin_ia32_mulps256 (v8sf,v8sf)
8326 v4df __builtin_ia32_orpd256 (v4df,v4df)
8327 v8sf __builtin_ia32_orps256 (v8sf,v8sf)
8328 v2df __builtin_ia32_pd_pd256 (v4df)
8329 v4df __builtin_ia32_pd256_pd (v2df)
8330 v4sf __builtin_ia32_ps_ps256 (v8sf)
8331 v8sf __builtin_ia32_ps256_ps (v4sf)
8332 int __builtin_ia32_ptestc256 (v4di,v4di,ptest)
8333 int __builtin_ia32_ptestnzc256 (v4di,v4di,ptest)
8334 int __builtin_ia32_ptestz256 (v4di,v4di,ptest)
8335 v8sf __builtin_ia32_rcpps256 (v8sf)
8336 v4df __builtin_ia32_roundpd256 (v4df,int)
8337 v8sf __builtin_ia32_roundps256 (v8sf,int)
8338 v8sf __builtin_ia32_rsqrtps_nr256 (v8sf)
8339 v8sf __builtin_ia32_rsqrtps256 (v8sf)
8340 v4df __builtin_ia32_shufpd256 (v4df,v4df,int)
8341 v8sf __builtin_ia32_shufps256 (v8sf,v8sf,int)
8342 v4si __builtin_ia32_si_si256 (v8si)
8343 v8si __builtin_ia32_si256_si (v4si)
8344 v4df __builtin_ia32_sqrtpd256 (v4df)
8345 v8sf __builtin_ia32_sqrtps_nr256 (v8sf)
8346 v8sf __builtin_ia32_sqrtps256 (v8sf)
8347 void __builtin_ia32_storedqu256 (pchar,v32qi)
8348 void __builtin_ia32_storeupd256 (pdouble,v4df)
8349 void __builtin_ia32_storeups256 (pfloat,v8sf)
8350 v4df __builtin_ia32_subpd256 (v4df,v4df)
8351 v8sf __builtin_ia32_subps256 (v8sf,v8sf)
8352 v4df __builtin_ia32_unpckhpd256 (v4df,v4df)
8353 v8sf __builtin_ia32_unpckhps256 (v8sf,v8sf)
8354 v4df __builtin_ia32_unpcklpd256 (v4df,v4df)
8355 v8sf __builtin_ia32_unpcklps256 (v8sf,v8sf)
8356 v4df __builtin_ia32_vbroadcastf128_pd256 (pcv2df)
8357 v8sf __builtin_ia32_vbroadcastf128_ps256 (pcv4sf)
8358 v4df __builtin_ia32_vbroadcastsd256 (pcdouble)
8359 v4sf __builtin_ia32_vbroadcastss (pcfloat)
8360 v8sf __builtin_ia32_vbroadcastss256 (pcfloat)
8361 v2df __builtin_ia32_vextractf128_pd256 (v4df,int)
8362 v4sf __builtin_ia32_vextractf128_ps256 (v8sf,int)
8363 v4si __builtin_ia32_vextractf128_si256 (v8si,int)
8364 v4df __builtin_ia32_vinsertf128_pd256 (v4df,v2df,int)
8365 v8sf __builtin_ia32_vinsertf128_ps256 (v8sf,v4sf,int)
8366 v8si __builtin_ia32_vinsertf128_si256 (v8si,v4si,int)
8367 v4df __builtin_ia32_vperm2f128_pd256 (v4df,v4df,int)
8368 v8sf __builtin_ia32_vperm2f128_ps256 (v8sf,v8sf,int)
8369 v8si __builtin_ia32_vperm2f128_si256 (v8si,v8si,int)
8370 v2df __builtin_ia32_vpermil2pd (v2df,v2df,v2di,int)
8371 v4df __builtin_ia32_vpermil2pd256 (v4df,v4df,v4di,int)
8372 v4sf __builtin_ia32_vpermil2ps (v4sf,v4sf,v4si,int)
8373 v8sf __builtin_ia32_vpermil2ps256 (v8sf,v8sf,v8si,int)
8374 v2df __builtin_ia32_vpermilpd (v2df,int)
8375 v4df __builtin_ia32_vpermilpd256 (v4df,int)
8376 v4sf __builtin_ia32_vpermilps (v4sf,int)
8377 v8sf __builtin_ia32_vpermilps256 (v8sf,int)
8378 v2df __builtin_ia32_vpermilvarpd (v2df,v2di)
8379 v4df __builtin_ia32_vpermilvarpd256 (v4df,v4di)
8380 v4sf __builtin_ia32_vpermilvarps (v4sf,v4si)
8381 v8sf __builtin_ia32_vpermilvarps256 (v8sf,v8si)
8382 int __builtin_ia32_vtestcpd (v2df,v2df,ptest)
8383 int __builtin_ia32_vtestcpd256 (v4df,v4df,ptest)
8384 int __builtin_ia32_vtestcps (v4sf,v4sf,ptest)
8385 int __builtin_ia32_vtestcps256 (v8sf,v8sf,ptest)
8386 int __builtin_ia32_vtestnzcpd (v2df,v2df,ptest)
8387 int __builtin_ia32_vtestnzcpd256 (v4df,v4df,ptest)
8388 int __builtin_ia32_vtestnzcps (v4sf,v4sf,ptest)
8389 int __builtin_ia32_vtestnzcps256 (v8sf,v8sf,ptest)
8390 int __builtin_ia32_vtestzpd (v2df,v2df,ptest)
8391 int __builtin_ia32_vtestzpd256 (v4df,v4df,ptest)
8392 int __builtin_ia32_vtestzps (v4sf,v4sf,ptest)
8393 int __builtin_ia32_vtestzps256 (v8sf,v8sf,ptest)
8394 void __builtin_ia32_vzeroall (void)
8395 void __builtin_ia32_vzeroupper (void)
8396 v4df __builtin_ia32_xorpd256 (v4df,v4df)
8397 v8sf __builtin_ia32_xorps256 (v8sf,v8sf)
8400 The following built-in functions are available when @option{-maes} is
8401 used. All of them generate the machine instruction that is part of the
8405 v2di __builtin_ia32_aesenc128 (v2di, v2di)
8406 v2di __builtin_ia32_aesenclast128 (v2di, v2di)
8407 v2di __builtin_ia32_aesdec128 (v2di, v2di)
8408 v2di __builtin_ia32_aesdeclast128 (v2di, v2di)
8409 v2di __builtin_ia32_aeskeygenassist128 (v2di, const int)
8410 v2di __builtin_ia32_aesimc128 (v2di)
8413 The following built-in function is available when @option{-mpclmul} is
8417 @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int)
8418 Generates the @code{pclmulqdq} machine instruction.
8421 The following built-in functions are available when @option{-msse4a} is used.
8422 All of them generate the machine instruction that is part of the name.
8425 void __builtin_ia32_movntsd (double *, v2df)
8426 void __builtin_ia32_movntss (float *, v4sf)
8427 v2di __builtin_ia32_extrq (v2di, v16qi)
8428 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
8429 v2di __builtin_ia32_insertq (v2di, v2di)
8430 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
8433 The following built-in functions are available when @option{-msse5} is used.
8434 All of them generate the machine instruction that is part of the name
8438 v2df __builtin_ia32_comeqpd (v2df, v2df)
8439 v2df __builtin_ia32_comeqps (v2df, v2df)
8440 v4sf __builtin_ia32_comeqsd (v4sf, v4sf)
8441 v4sf __builtin_ia32_comeqss (v4sf, v4sf)
8442 v2df __builtin_ia32_comfalsepd (v2df, v2df)
8443 v2df __builtin_ia32_comfalseps (v2df, v2df)
8444 v4sf __builtin_ia32_comfalsesd (v4sf, v4sf)
8445 v4sf __builtin_ia32_comfalsess (v4sf, v4sf)
8446 v2df __builtin_ia32_comgepd (v2df, v2df)
8447 v2df __builtin_ia32_comgeps (v2df, v2df)
8448 v4sf __builtin_ia32_comgesd (v4sf, v4sf)
8449 v4sf __builtin_ia32_comgess (v4sf, v4sf)
8450 v2df __builtin_ia32_comgtpd (v2df, v2df)
8451 v2df __builtin_ia32_comgtps (v2df, v2df)
8452 v4sf __builtin_ia32_comgtsd (v4sf, v4sf)
8453 v4sf __builtin_ia32_comgtss (v4sf, v4sf)
8454 v2df __builtin_ia32_comlepd (v2df, v2df)
8455 v2df __builtin_ia32_comleps (v2df, v2df)
8456 v4sf __builtin_ia32_comlesd (v4sf, v4sf)
8457 v4sf __builtin_ia32_comless (v4sf, v4sf)
8458 v2df __builtin_ia32_comltpd (v2df, v2df)
8459 v2df __builtin_ia32_comltps (v2df, v2df)
8460 v4sf __builtin_ia32_comltsd (v4sf, v4sf)
8461 v4sf __builtin_ia32_comltss (v4sf, v4sf)
8462 v2df __builtin_ia32_comnepd (v2df, v2df)
8463 v2df __builtin_ia32_comneps (v2df, v2df)
8464 v4sf __builtin_ia32_comnesd (v4sf, v4sf)
8465 v4sf __builtin_ia32_comness (v4sf, v4sf)
8466 v2df __builtin_ia32_comordpd (v2df, v2df)
8467 v2df __builtin_ia32_comordps (v2df, v2df)
8468 v4sf __builtin_ia32_comordsd (v4sf, v4sf)
8469 v4sf __builtin_ia32_comordss (v4sf, v4sf)
8470 v2df __builtin_ia32_comtruepd (v2df, v2df)
8471 v2df __builtin_ia32_comtrueps (v2df, v2df)
8472 v4sf __builtin_ia32_comtruesd (v4sf, v4sf)
8473 v4sf __builtin_ia32_comtruess (v4sf, v4sf)
8474 v2df __builtin_ia32_comueqpd (v2df, v2df)
8475 v2df __builtin_ia32_comueqps (v2df, v2df)
8476 v4sf __builtin_ia32_comueqsd (v4sf, v4sf)
8477 v4sf __builtin_ia32_comueqss (v4sf, v4sf)
8478 v2df __builtin_ia32_comugepd (v2df, v2df)
8479 v2df __builtin_ia32_comugeps (v2df, v2df)
8480 v4sf __builtin_ia32_comugesd (v4sf, v4sf)
8481 v4sf __builtin_ia32_comugess (v4sf, v4sf)
8482 v2df __builtin_ia32_comugtpd (v2df, v2df)
8483 v2df __builtin_ia32_comugtps (v2df, v2df)
8484 v4sf __builtin_ia32_comugtsd (v4sf, v4sf)
8485 v4sf __builtin_ia32_comugtss (v4sf, v4sf)
8486 v2df __builtin_ia32_comulepd (v2df, v2df)
8487 v2df __builtin_ia32_comuleps (v2df, v2df)
8488 v4sf __builtin_ia32_comulesd (v4sf, v4sf)
8489 v4sf __builtin_ia32_comuless (v4sf, v4sf)
8490 v2df __builtin_ia32_comultpd (v2df, v2df)
8491 v2df __builtin_ia32_comultps (v2df, v2df)
8492 v4sf __builtin_ia32_comultsd (v4sf, v4sf)
8493 v4sf __builtin_ia32_comultss (v4sf, v4sf)
8494 v2df __builtin_ia32_comunepd (v2df, v2df)
8495 v2df __builtin_ia32_comuneps (v2df, v2df)
8496 v4sf __builtin_ia32_comunesd (v4sf, v4sf)
8497 v4sf __builtin_ia32_comuness (v4sf, v4sf)
8498 v2df __builtin_ia32_comunordpd (v2df, v2df)
8499 v2df __builtin_ia32_comunordps (v2df, v2df)
8500 v4sf __builtin_ia32_comunordsd (v4sf, v4sf)
8501 v4sf __builtin_ia32_comunordss (v4sf, v4sf)
8502 v2df __builtin_ia32_fmaddpd (v2df, v2df, v2df)
8503 v4sf __builtin_ia32_fmaddps (v4sf, v4sf, v4sf)
8504 v2df __builtin_ia32_fmaddsd (v2df, v2df, v2df)
8505 v4sf __builtin_ia32_fmaddss (v4sf, v4sf, v4sf)
8506 v2df __builtin_ia32_fmsubpd (v2df, v2df, v2df)
8507 v4sf __builtin_ia32_fmsubps (v4sf, v4sf, v4sf)
8508 v2df __builtin_ia32_fmsubsd (v2df, v2df, v2df)
8509 v4sf __builtin_ia32_fmsubss (v4sf, v4sf, v4sf)
8510 v2df __builtin_ia32_fnmaddpd (v2df, v2df, v2df)
8511 v4sf __builtin_ia32_fnmaddps (v4sf, v4sf, v4sf)
8512 v2df __builtin_ia32_fnmaddsd (v2df, v2df, v2df)
8513 v4sf __builtin_ia32_fnmaddss (v4sf, v4sf, v4sf)
8514 v2df __builtin_ia32_fnmsubpd (v2df, v2df, v2df)
8515 v4sf __builtin_ia32_fnmsubps (v4sf, v4sf, v4sf)
8516 v2df __builtin_ia32_fnmsubsd (v2df, v2df, v2df)
8517 v4sf __builtin_ia32_fnmsubss (v4sf, v4sf, v4sf)
8518 v2df __builtin_ia32_frczpd (v2df)
8519 v4sf __builtin_ia32_frczps (v4sf)
8520 v2df __builtin_ia32_frczsd (v2df, v2df)
8521 v4sf __builtin_ia32_frczss (v4sf, v4sf)
8522 v2di __builtin_ia32_pcmov (v2di, v2di, v2di)
8523 v2di __builtin_ia32_pcmov_v2di (v2di, v2di, v2di)
8524 v4si __builtin_ia32_pcmov_v4si (v4si, v4si, v4si)
8525 v8hi __builtin_ia32_pcmov_v8hi (v8hi, v8hi, v8hi)
8526 v16qi __builtin_ia32_pcmov_v16qi (v16qi, v16qi, v16qi)
8527 v2df __builtin_ia32_pcmov_v2df (v2df, v2df, v2df)
8528 v4sf __builtin_ia32_pcmov_v4sf (v4sf, v4sf, v4sf)
8529 v16qi __builtin_ia32_pcomeqb (v16qi, v16qi)
8530 v8hi __builtin_ia32_pcomeqw (v8hi, v8hi)
8531 v4si __builtin_ia32_pcomeqd (v4si, v4si)
8532 v2di __builtin_ia32_pcomeqq (v2di, v2di)
8533 v16qi __builtin_ia32_pcomequb (v16qi, v16qi)
8534 v4si __builtin_ia32_pcomequd (v4si, v4si)
8535 v2di __builtin_ia32_pcomequq (v2di, v2di)
8536 v8hi __builtin_ia32_pcomequw (v8hi, v8hi)
8537 v8hi __builtin_ia32_pcomeqw (v8hi, v8hi)
8538 v16qi __builtin_ia32_pcomfalseb (v16qi, v16qi)
8539 v4si __builtin_ia32_pcomfalsed (v4si, v4si)
8540 v2di __builtin_ia32_pcomfalseq (v2di, v2di)
8541 v16qi __builtin_ia32_pcomfalseub (v16qi, v16qi)
8542 v4si __builtin_ia32_pcomfalseud (v4si, v4si)
8543 v2di __builtin_ia32_pcomfalseuq (v2di, v2di)
8544 v8hi __builtin_ia32_pcomfalseuw (v8hi, v8hi)
8545 v8hi __builtin_ia32_pcomfalsew (v8hi, v8hi)
8546 v16qi __builtin_ia32_pcomgeb (v16qi, v16qi)
8547 v4si __builtin_ia32_pcomged (v4si, v4si)
8548 v2di __builtin_ia32_pcomgeq (v2di, v2di)
8549 v16qi __builtin_ia32_pcomgeub (v16qi, v16qi)
8550 v4si __builtin_ia32_pcomgeud (v4si, v4si)
8551 v2di __builtin_ia32_pcomgeuq (v2di, v2di)
8552 v8hi __builtin_ia32_pcomgeuw (v8hi, v8hi)
8553 v8hi __builtin_ia32_pcomgew (v8hi, v8hi)
8554 v16qi __builtin_ia32_pcomgtb (v16qi, v16qi)
8555 v4si __builtin_ia32_pcomgtd (v4si, v4si)
8556 v2di __builtin_ia32_pcomgtq (v2di, v2di)
8557 v16qi __builtin_ia32_pcomgtub (v16qi, v16qi)
8558 v4si __builtin_ia32_pcomgtud (v4si, v4si)
8559 v2di __builtin_ia32_pcomgtuq (v2di, v2di)
8560 v8hi __builtin_ia32_pcomgtuw (v8hi, v8hi)
8561 v8hi __builtin_ia32_pcomgtw (v8hi, v8hi)
8562 v16qi __builtin_ia32_pcomleb (v16qi, v16qi)
8563 v4si __builtin_ia32_pcomled (v4si, v4si)
8564 v2di __builtin_ia32_pcomleq (v2di, v2di)
8565 v16qi __builtin_ia32_pcomleub (v16qi, v16qi)
8566 v4si __builtin_ia32_pcomleud (v4si, v4si)
8567 v2di __builtin_ia32_pcomleuq (v2di, v2di)
8568 v8hi __builtin_ia32_pcomleuw (v8hi, v8hi)
8569 v8hi __builtin_ia32_pcomlew (v8hi, v8hi)
8570 v16qi __builtin_ia32_pcomltb (v16qi, v16qi)
8571 v4si __builtin_ia32_pcomltd (v4si, v4si)
8572 v2di __builtin_ia32_pcomltq (v2di, v2di)
8573 v16qi __builtin_ia32_pcomltub (v16qi, v16qi)
8574 v4si __builtin_ia32_pcomltud (v4si, v4si)
8575 v2di __builtin_ia32_pcomltuq (v2di, v2di)
8576 v8hi __builtin_ia32_pcomltuw (v8hi, v8hi)
8577 v8hi __builtin_ia32_pcomltw (v8hi, v8hi)
8578 v16qi __builtin_ia32_pcomneb (v16qi, v16qi)
8579 v4si __builtin_ia32_pcomned (v4si, v4si)
8580 v2di __builtin_ia32_pcomneq (v2di, v2di)
8581 v16qi __builtin_ia32_pcomneub (v16qi, v16qi)
8582 v4si __builtin_ia32_pcomneud (v4si, v4si)
8583 v2di __builtin_ia32_pcomneuq (v2di, v2di)
8584 v8hi __builtin_ia32_pcomneuw (v8hi, v8hi)
8585 v8hi __builtin_ia32_pcomnew (v8hi, v8hi)
8586 v16qi __builtin_ia32_pcomtrueb (v16qi, v16qi)
8587 v4si __builtin_ia32_pcomtrued (v4si, v4si)
8588 v2di __builtin_ia32_pcomtrueq (v2di, v2di)
8589 v16qi __builtin_ia32_pcomtrueub (v16qi, v16qi)
8590 v4si __builtin_ia32_pcomtrueud (v4si, v4si)
8591 v2di __builtin_ia32_pcomtrueuq (v2di, v2di)
8592 v8hi __builtin_ia32_pcomtrueuw (v8hi, v8hi)
8593 v8hi __builtin_ia32_pcomtruew (v8hi, v8hi)
8594 v4df __builtin_ia32_permpd (v2df, v2df, v16qi)
8595 v4sf __builtin_ia32_permps (v4sf, v4sf, v16qi)
8596 v4si __builtin_ia32_phaddbd (v16qi)
8597 v2di __builtin_ia32_phaddbq (v16qi)
8598 v8hi __builtin_ia32_phaddbw (v16qi)
8599 v2di __builtin_ia32_phadddq (v4si)
8600 v4si __builtin_ia32_phaddubd (v16qi)
8601 v2di __builtin_ia32_phaddubq (v16qi)
8602 v8hi __builtin_ia32_phaddubw (v16qi)
8603 v2di __builtin_ia32_phaddudq (v4si)
8604 v4si __builtin_ia32_phadduwd (v8hi)
8605 v2di __builtin_ia32_phadduwq (v8hi)
8606 v4si __builtin_ia32_phaddwd (v8hi)
8607 v2di __builtin_ia32_phaddwq (v8hi)
8608 v8hi __builtin_ia32_phsubbw (v16qi)
8609 v2di __builtin_ia32_phsubdq (v4si)
8610 v4si __builtin_ia32_phsubwd (v8hi)
8611 v4si __builtin_ia32_pmacsdd (v4si, v4si, v4si)
8612 v2di __builtin_ia32_pmacsdqh (v4si, v4si, v2di)
8613 v2di __builtin_ia32_pmacsdql (v4si, v4si, v2di)
8614 v4si __builtin_ia32_pmacssdd (v4si, v4si, v4si)
8615 v2di __builtin_ia32_pmacssdqh (v4si, v4si, v2di)
8616 v2di __builtin_ia32_pmacssdql (v4si, v4si, v2di)
8617 v4si __builtin_ia32_pmacsswd (v8hi, v8hi, v4si)
8618 v8hi __builtin_ia32_pmacssww (v8hi, v8hi, v8hi)
8619 v4si __builtin_ia32_pmacswd (v8hi, v8hi, v4si)
8620 v8hi __builtin_ia32_pmacsww (v8hi, v8hi, v8hi)
8621 v4si __builtin_ia32_pmadcsswd (v8hi, v8hi, v4si)
8622 v4si __builtin_ia32_pmadcswd (v8hi, v8hi, v4si)
8623 v16qi __builtin_ia32_pperm (v16qi, v16qi, v16qi)
8624 v16qi __builtin_ia32_protb (v16qi, v16qi)
8625 v4si __builtin_ia32_protd (v4si, v4si)
8626 v2di __builtin_ia32_protq (v2di, v2di)
8627 v8hi __builtin_ia32_protw (v8hi, v8hi)
8628 v16qi __builtin_ia32_pshab (v16qi, v16qi)
8629 v4si __builtin_ia32_pshad (v4si, v4si)
8630 v2di __builtin_ia32_pshaq (v2di, v2di)
8631 v8hi __builtin_ia32_pshaw (v8hi, v8hi)
8632 v16qi __builtin_ia32_pshlb (v16qi, v16qi)
8633 v4si __builtin_ia32_pshld (v4si, v4si)
8634 v2di __builtin_ia32_pshlq (v2di, v2di)
8635 v8hi __builtin_ia32_pshlw (v8hi, v8hi)
8638 The following builtin-in functions are available when @option{-msse5}
8639 is used. The second argument must be an integer constant and generate
8640 the machine instruction that is part of the name with the @samp{_imm}
8644 v16qi __builtin_ia32_protb_imm (v16qi, int)
8645 v4si __builtin_ia32_protd_imm (v4si, int)
8646 v2di __builtin_ia32_protq_imm (v2di, int)
8647 v8hi __builtin_ia32_protw_imm (v8hi, int)
8650 The following built-in functions are available when @option{-m3dnow} is used.
8651 All of them generate the machine instruction that is part of the name.
8654 void __builtin_ia32_femms (void)
8655 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
8656 v2si __builtin_ia32_pf2id (v2sf)
8657 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
8658 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
8659 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
8660 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
8661 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
8662 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
8663 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
8664 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
8665 v2sf __builtin_ia32_pfrcp (v2sf)
8666 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
8667 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
8668 v2sf __builtin_ia32_pfrsqrt (v2sf)
8669 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
8670 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
8671 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
8672 v2sf __builtin_ia32_pi2fd (v2si)
8673 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
8676 The following built-in functions are available when both @option{-m3dnow}
8677 and @option{-march=athlon} are used. All of them generate the machine
8678 instruction that is part of the name.
8681 v2si __builtin_ia32_pf2iw (v2sf)
8682 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
8683 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
8684 v2sf __builtin_ia32_pi2fw (v2si)
8685 v2sf __builtin_ia32_pswapdsf (v2sf)
8686 v2si __builtin_ia32_pswapdsi (v2si)
8689 @node MIPS DSP Built-in Functions
8690 @subsection MIPS DSP Built-in Functions
8692 The MIPS DSP Application-Specific Extension (ASE) includes new
8693 instructions that are designed to improve the performance of DSP and
8694 media applications. It provides instructions that operate on packed
8695 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
8697 GCC supports MIPS DSP operations using both the generic
8698 vector extensions (@pxref{Vector Extensions}) and a collection of
8699 MIPS-specific built-in functions. Both kinds of support are
8700 enabled by the @option{-mdsp} command-line option.
8702 Revision 2 of the ASE was introduced in the second half of 2006.
8703 This revision adds extra instructions to the original ASE, but is
8704 otherwise backwards-compatible with it. You can select revision 2
8705 using the command-line option @option{-mdspr2}; this option implies
8708 The SCOUNT and POS bits of the DSP control register are global. The
8709 WRDSP, EXTPDP, EXTPDPV and MTHLIP instructions modify the SCOUNT and
8710 POS bits. During optimization, the compiler will not delete these
8711 instructions and it will not delete calls to functions containing
8714 At present, GCC only provides support for operations on 32-bit
8715 vectors. The vector type associated with 8-bit integer data is
8716 usually called @code{v4i8}, the vector type associated with Q7
8717 is usually called @code{v4q7}, the vector type associated with 16-bit
8718 integer data is usually called @code{v2i16}, and the vector type
8719 associated with Q15 is usually called @code{v2q15}. They can be
8720 defined in C as follows:
8723 typedef signed char v4i8 __attribute__ ((vector_size(4)));
8724 typedef signed char v4q7 __attribute__ ((vector_size(4)));
8725 typedef short v2i16 __attribute__ ((vector_size(4)));
8726 typedef short v2q15 __attribute__ ((vector_size(4)));
8729 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
8730 initialized in the same way as aggregates. For example:
8733 v4i8 a = @{1, 2, 3, 4@};
8735 b = (v4i8) @{5, 6, 7, 8@};
8737 v2q15 c = @{0x0fcb, 0x3a75@};
8739 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
8742 @emph{Note:} The CPU's endianness determines the order in which values
8743 are packed. On little-endian targets, the first value is the least
8744 significant and the last value is the most significant. The opposite
8745 order applies to big-endian targets. For example, the code above will
8746 set the lowest byte of @code{a} to @code{1} on little-endian targets
8747 and @code{4} on big-endian targets.
8749 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
8750 representation. As shown in this example, the integer representation
8751 of a Q7 value can be obtained by multiplying the fractional value by
8752 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
8753 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
8756 The table below lists the @code{v4i8} and @code{v2q15} operations for which
8757 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
8758 and @code{c} and @code{d} are @code{v2q15} values.
8760 @multitable @columnfractions .50 .50
8761 @item C code @tab MIPS instruction
8762 @item @code{a + b} @tab @code{addu.qb}
8763 @item @code{c + d} @tab @code{addq.ph}
8764 @item @code{a - b} @tab @code{subu.qb}
8765 @item @code{c - d} @tab @code{subq.ph}
8768 The table below lists the @code{v2i16} operation for which
8769 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
8770 @code{v2i16} values.
8772 @multitable @columnfractions .50 .50
8773 @item C code @tab MIPS instruction
8774 @item @code{e * f} @tab @code{mul.ph}
8777 It is easier to describe the DSP built-in functions if we first define
8778 the following types:
8783 typedef unsigned int ui32;
8784 typedef long long a64;
8787 @code{q31} and @code{i32} are actually the same as @code{int}, but we
8788 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
8789 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
8790 @code{long long}, but we use @code{a64} to indicate values that will
8791 be placed in one of the four DSP accumulators (@code{$ac0},
8792 @code{$ac1}, @code{$ac2} or @code{$ac3}).
8794 Also, some built-in functions prefer or require immediate numbers as
8795 parameters, because the corresponding DSP instructions accept both immediate
8796 numbers and register operands, or accept immediate numbers only. The
8797 immediate parameters are listed as follows.
8806 imm_n32_31: -32 to 31.
8807 imm_n512_511: -512 to 511.
8810 The following built-in functions map directly to a particular MIPS DSP
8811 instruction. Please refer to the architecture specification
8812 for details on what each instruction does.
8815 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
8816 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
8817 q31 __builtin_mips_addq_s_w (q31, q31)
8818 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
8819 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
8820 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
8821 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
8822 q31 __builtin_mips_subq_s_w (q31, q31)
8823 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
8824 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
8825 i32 __builtin_mips_addsc (i32, i32)
8826 i32 __builtin_mips_addwc (i32, i32)
8827 i32 __builtin_mips_modsub (i32, i32)
8828 i32 __builtin_mips_raddu_w_qb (v4i8)
8829 v2q15 __builtin_mips_absq_s_ph (v2q15)
8830 q31 __builtin_mips_absq_s_w (q31)
8831 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
8832 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
8833 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
8834 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
8835 q31 __builtin_mips_preceq_w_phl (v2q15)
8836 q31 __builtin_mips_preceq_w_phr (v2q15)
8837 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
8838 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
8839 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
8840 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
8841 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
8842 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
8843 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
8844 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
8845 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
8846 v4i8 __builtin_mips_shll_qb (v4i8, i32)
8847 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
8848 v2q15 __builtin_mips_shll_ph (v2q15, i32)
8849 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
8850 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
8851 q31 __builtin_mips_shll_s_w (q31, imm0_31)
8852 q31 __builtin_mips_shll_s_w (q31, i32)
8853 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
8854 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
8855 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
8856 v2q15 __builtin_mips_shra_ph (v2q15, i32)
8857 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
8858 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
8859 q31 __builtin_mips_shra_r_w (q31, imm0_31)
8860 q31 __builtin_mips_shra_r_w (q31, i32)
8861 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
8862 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
8863 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
8864 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
8865 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
8866 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
8867 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
8868 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
8869 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
8870 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
8871 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
8872 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
8873 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
8874 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
8875 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
8876 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
8877 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
8878 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
8879 i32 __builtin_mips_bitrev (i32)
8880 i32 __builtin_mips_insv (i32, i32)
8881 v4i8 __builtin_mips_repl_qb (imm0_255)
8882 v4i8 __builtin_mips_repl_qb (i32)
8883 v2q15 __builtin_mips_repl_ph (imm_n512_511)
8884 v2q15 __builtin_mips_repl_ph (i32)
8885 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
8886 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
8887 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
8888 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
8889 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
8890 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
8891 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
8892 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
8893 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
8894 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
8895 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
8896 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
8897 i32 __builtin_mips_extr_w (a64, imm0_31)
8898 i32 __builtin_mips_extr_w (a64, i32)
8899 i32 __builtin_mips_extr_r_w (a64, imm0_31)
8900 i32 __builtin_mips_extr_s_h (a64, i32)
8901 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
8902 i32 __builtin_mips_extr_rs_w (a64, i32)
8903 i32 __builtin_mips_extr_s_h (a64, imm0_31)
8904 i32 __builtin_mips_extr_r_w (a64, i32)
8905 i32 __builtin_mips_extp (a64, imm0_31)
8906 i32 __builtin_mips_extp (a64, i32)
8907 i32 __builtin_mips_extpdp (a64, imm0_31)
8908 i32 __builtin_mips_extpdp (a64, i32)
8909 a64 __builtin_mips_shilo (a64, imm_n32_31)
8910 a64 __builtin_mips_shilo (a64, i32)
8911 a64 __builtin_mips_mthlip (a64, i32)
8912 void __builtin_mips_wrdsp (i32, imm0_63)
8913 i32 __builtin_mips_rddsp (imm0_63)
8914 i32 __builtin_mips_lbux (void *, i32)
8915 i32 __builtin_mips_lhx (void *, i32)
8916 i32 __builtin_mips_lwx (void *, i32)
8917 i32 __builtin_mips_bposge32 (void)
8920 The following built-in functions map directly to a particular MIPS DSP REV 2
8921 instruction. Please refer to the architecture specification
8922 for details on what each instruction does.
8925 v4q7 __builtin_mips_absq_s_qb (v4q7);
8926 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
8927 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
8928 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
8929 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
8930 i32 __builtin_mips_append (i32, i32, imm0_31);
8931 i32 __builtin_mips_balign (i32, i32, imm0_3);
8932 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
8933 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
8934 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
8935 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
8936 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
8937 a64 __builtin_mips_madd (a64, i32, i32);
8938 a64 __builtin_mips_maddu (a64, ui32, ui32);
8939 a64 __builtin_mips_msub (a64, i32, i32);
8940 a64 __builtin_mips_msubu (a64, ui32, ui32);
8941 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
8942 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
8943 q31 __builtin_mips_mulq_rs_w (q31, q31);
8944 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
8945 q31 __builtin_mips_mulq_s_w (q31, q31);
8946 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
8947 a64 __builtin_mips_mult (i32, i32);
8948 a64 __builtin_mips_multu (ui32, ui32);
8949 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
8950 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
8951 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
8952 i32 __builtin_mips_prepend (i32, i32, imm0_31);
8953 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
8954 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
8955 v4i8 __builtin_mips_shra_qb (v4i8, i32);
8956 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
8957 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
8958 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
8959 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
8960 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
8961 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
8962 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
8963 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
8964 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
8965 q31 __builtin_mips_addqh_w (q31, q31);
8966 q31 __builtin_mips_addqh_r_w (q31, q31);
8967 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
8968 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
8969 q31 __builtin_mips_subqh_w (q31, q31);
8970 q31 __builtin_mips_subqh_r_w (q31, q31);
8971 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
8972 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
8973 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
8974 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
8975 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
8976 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
8980 @node MIPS Paired-Single Support
8981 @subsection MIPS Paired-Single Support
8983 The MIPS64 architecture includes a number of instructions that
8984 operate on pairs of single-precision floating-point values.
8985 Each pair is packed into a 64-bit floating-point register,
8986 with one element being designated the ``upper half'' and
8987 the other being designated the ``lower half''.
8989 GCC supports paired-single operations using both the generic
8990 vector extensions (@pxref{Vector Extensions}) and a collection of
8991 MIPS-specific built-in functions. Both kinds of support are
8992 enabled by the @option{-mpaired-single} command-line option.
8994 The vector type associated with paired-single values is usually
8995 called @code{v2sf}. It can be defined in C as follows:
8998 typedef float v2sf __attribute__ ((vector_size (8)));
9001 @code{v2sf} values are initialized in the same way as aggregates.
9005 v2sf a = @{1.5, 9.1@};
9008 b = (v2sf) @{e, f@};
9011 @emph{Note:} The CPU's endianness determines which value is stored in
9012 the upper half of a register and which value is stored in the lower half.
9013 On little-endian targets, the first value is the lower one and the second
9014 value is the upper one. The opposite order applies to big-endian targets.
9015 For example, the code above will set the lower half of @code{a} to
9016 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
9018 @node MIPS Loongson Built-in Functions
9019 @subsection MIPS Loongson Built-in Functions
9021 GCC provides intrinsics to access the SIMD instructions provided by the
9022 ST Microelectronics Loongson-2E and -2F processors. These intrinsics,
9023 available after inclusion of the @code{loongson.h} header file,
9024 operate on the following 64-bit vector types:
9027 @item @code{uint8x8_t}, a vector of eight unsigned 8-bit integers;
9028 @item @code{uint16x4_t}, a vector of four unsigned 16-bit integers;
9029 @item @code{uint32x2_t}, a vector of two unsigned 32-bit integers;
9030 @item @code{int8x8_t}, a vector of eight signed 8-bit integers;
9031 @item @code{int16x4_t}, a vector of four signed 16-bit integers;
9032 @item @code{int32x2_t}, a vector of two signed 32-bit integers.
9035 The intrinsics provided are listed below; each is named after the
9036 machine instruction to which it corresponds, with suffixes added as
9037 appropriate to distinguish intrinsics that expand to the same machine
9038 instruction yet have different argument types. Refer to the architecture
9039 documentation for a description of the functionality of each
9043 int16x4_t packsswh (int32x2_t s, int32x2_t t);
9044 int8x8_t packsshb (int16x4_t s, int16x4_t t);
9045 uint8x8_t packushb (uint16x4_t s, uint16x4_t t);
9046 uint32x2_t paddw_u (uint32x2_t s, uint32x2_t t);
9047 uint16x4_t paddh_u (uint16x4_t s, uint16x4_t t);
9048 uint8x8_t paddb_u (uint8x8_t s, uint8x8_t t);
9049 int32x2_t paddw_s (int32x2_t s, int32x2_t t);
9050 int16x4_t paddh_s (int16x4_t s, int16x4_t t);
9051 int8x8_t paddb_s (int8x8_t s, int8x8_t t);
9052 uint64_t paddd_u (uint64_t s, uint64_t t);
9053 int64_t paddd_s (int64_t s, int64_t t);
9054 int16x4_t paddsh (int16x4_t s, int16x4_t t);
9055 int8x8_t paddsb (int8x8_t s, int8x8_t t);
9056 uint16x4_t paddush (uint16x4_t s, uint16x4_t t);
9057 uint8x8_t paddusb (uint8x8_t s, uint8x8_t t);
9058 uint64_t pandn_ud (uint64_t s, uint64_t t);
9059 uint32x2_t pandn_uw (uint32x2_t s, uint32x2_t t);
9060 uint16x4_t pandn_uh (uint16x4_t s, uint16x4_t t);
9061 uint8x8_t pandn_ub (uint8x8_t s, uint8x8_t t);
9062 int64_t pandn_sd (int64_t s, int64_t t);
9063 int32x2_t pandn_sw (int32x2_t s, int32x2_t t);
9064 int16x4_t pandn_sh (int16x4_t s, int16x4_t t);
9065 int8x8_t pandn_sb (int8x8_t s, int8x8_t t);
9066 uint16x4_t pavgh (uint16x4_t s, uint16x4_t t);
9067 uint8x8_t pavgb (uint8x8_t s, uint8x8_t t);
9068 uint32x2_t pcmpeqw_u (uint32x2_t s, uint32x2_t t);
9069 uint16x4_t pcmpeqh_u (uint16x4_t s, uint16x4_t t);
9070 uint8x8_t pcmpeqb_u (uint8x8_t s, uint8x8_t t);
9071 int32x2_t pcmpeqw_s (int32x2_t s, int32x2_t t);
9072 int16x4_t pcmpeqh_s (int16x4_t s, int16x4_t t);
9073 int8x8_t pcmpeqb_s (int8x8_t s, int8x8_t t);
9074 uint32x2_t pcmpgtw_u (uint32x2_t s, uint32x2_t t);
9075 uint16x4_t pcmpgth_u (uint16x4_t s, uint16x4_t t);
9076 uint8x8_t pcmpgtb_u (uint8x8_t s, uint8x8_t t);
9077 int32x2_t pcmpgtw_s (int32x2_t s, int32x2_t t);
9078 int16x4_t pcmpgth_s (int16x4_t s, int16x4_t t);
9079 int8x8_t pcmpgtb_s (int8x8_t s, int8x8_t t);
9080 uint16x4_t pextrh_u (uint16x4_t s, int field);
9081 int16x4_t pextrh_s (int16x4_t s, int field);
9082 uint16x4_t pinsrh_0_u (uint16x4_t s, uint16x4_t t);
9083 uint16x4_t pinsrh_1_u (uint16x4_t s, uint16x4_t t);
9084 uint16x4_t pinsrh_2_u (uint16x4_t s, uint16x4_t t);
9085 uint16x4_t pinsrh_3_u (uint16x4_t s, uint16x4_t t);
9086 int16x4_t pinsrh_0_s (int16x4_t s, int16x4_t t);
9087 int16x4_t pinsrh_1_s (int16x4_t s, int16x4_t t);
9088 int16x4_t pinsrh_2_s (int16x4_t s, int16x4_t t);
9089 int16x4_t pinsrh_3_s (int16x4_t s, int16x4_t t);
9090 int32x2_t pmaddhw (int16x4_t s, int16x4_t t);
9091 int16x4_t pmaxsh (int16x4_t s, int16x4_t t);
9092 uint8x8_t pmaxub (uint8x8_t s, uint8x8_t t);
9093 int16x4_t pminsh (int16x4_t s, int16x4_t t);
9094 uint8x8_t pminub (uint8x8_t s, uint8x8_t t);
9095 uint8x8_t pmovmskb_u (uint8x8_t s);
9096 int8x8_t pmovmskb_s (int8x8_t s);
9097 uint16x4_t pmulhuh (uint16x4_t s, uint16x4_t t);
9098 int16x4_t pmulhh (int16x4_t s, int16x4_t t);
9099 int16x4_t pmullh (int16x4_t s, int16x4_t t);
9100 int64_t pmuluw (uint32x2_t s, uint32x2_t t);
9101 uint8x8_t pasubub (uint8x8_t s, uint8x8_t t);
9102 uint16x4_t biadd (uint8x8_t s);
9103 uint16x4_t psadbh (uint8x8_t s, uint8x8_t t);
9104 uint16x4_t pshufh_u (uint16x4_t dest, uint16x4_t s, uint8_t order);
9105 int16x4_t pshufh_s (int16x4_t dest, int16x4_t s, uint8_t order);
9106 uint16x4_t psllh_u (uint16x4_t s, uint8_t amount);
9107 int16x4_t psllh_s (int16x4_t s, uint8_t amount);
9108 uint32x2_t psllw_u (uint32x2_t s, uint8_t amount);
9109 int32x2_t psllw_s (int32x2_t s, uint8_t amount);
9110 uint16x4_t psrlh_u (uint16x4_t s, uint8_t amount);
9111 int16x4_t psrlh_s (int16x4_t s, uint8_t amount);
9112 uint32x2_t psrlw_u (uint32x2_t s, uint8_t amount);
9113 int32x2_t psrlw_s (int32x2_t s, uint8_t amount);
9114 uint16x4_t psrah_u (uint16x4_t s, uint8_t amount);
9115 int16x4_t psrah_s (int16x4_t s, uint8_t amount);
9116 uint32x2_t psraw_u (uint32x2_t s, uint8_t amount);
9117 int32x2_t psraw_s (int32x2_t s, uint8_t amount);
9118 uint32x2_t psubw_u (uint32x2_t s, uint32x2_t t);
9119 uint16x4_t psubh_u (uint16x4_t s, uint16x4_t t);
9120 uint8x8_t psubb_u (uint8x8_t s, uint8x8_t t);
9121 int32x2_t psubw_s (int32x2_t s, int32x2_t t);
9122 int16x4_t psubh_s (int16x4_t s, int16x4_t t);
9123 int8x8_t psubb_s (int8x8_t s, int8x8_t t);
9124 uint64_t psubd_u (uint64_t s, uint64_t t);
9125 int64_t psubd_s (int64_t s, int64_t t);
9126 int16x4_t psubsh (int16x4_t s, int16x4_t t);
9127 int8x8_t psubsb (int8x8_t s, int8x8_t t);
9128 uint16x4_t psubush (uint16x4_t s, uint16x4_t t);
9129 uint8x8_t psubusb (uint8x8_t s, uint8x8_t t);
9130 uint32x2_t punpckhwd_u (uint32x2_t s, uint32x2_t t);
9131 uint16x4_t punpckhhw_u (uint16x4_t s, uint16x4_t t);
9132 uint8x8_t punpckhbh_u (uint8x8_t s, uint8x8_t t);
9133 int32x2_t punpckhwd_s (int32x2_t s, int32x2_t t);
9134 int16x4_t punpckhhw_s (int16x4_t s, int16x4_t t);
9135 int8x8_t punpckhbh_s (int8x8_t s, int8x8_t t);
9136 uint32x2_t punpcklwd_u (uint32x2_t s, uint32x2_t t);
9137 uint16x4_t punpcklhw_u (uint16x4_t s, uint16x4_t t);
9138 uint8x8_t punpcklbh_u (uint8x8_t s, uint8x8_t t);
9139 int32x2_t punpcklwd_s (int32x2_t s, int32x2_t t);
9140 int16x4_t punpcklhw_s (int16x4_t s, int16x4_t t);
9141 int8x8_t punpcklbh_s (int8x8_t s, int8x8_t t);
9145 * Paired-Single Arithmetic::
9146 * Paired-Single Built-in Functions::
9147 * MIPS-3D Built-in Functions::
9150 @node Paired-Single Arithmetic
9151 @subsubsection Paired-Single Arithmetic
9153 The table below lists the @code{v2sf} operations for which hardware
9154 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
9155 values and @code{x} is an integral value.
9157 @multitable @columnfractions .50 .50
9158 @item C code @tab MIPS instruction
9159 @item @code{a + b} @tab @code{add.ps}
9160 @item @code{a - b} @tab @code{sub.ps}
9161 @item @code{-a} @tab @code{neg.ps}
9162 @item @code{a * b} @tab @code{mul.ps}
9163 @item @code{a * b + c} @tab @code{madd.ps}
9164 @item @code{a * b - c} @tab @code{msub.ps}
9165 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
9166 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
9167 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
9170 Note that the multiply-accumulate instructions can be disabled
9171 using the command-line option @code{-mno-fused-madd}.
9173 @node Paired-Single Built-in Functions
9174 @subsubsection Paired-Single Built-in Functions
9176 The following paired-single functions map directly to a particular
9177 MIPS instruction. Please refer to the architecture specification
9178 for details on what each instruction does.
9181 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
9182 Pair lower lower (@code{pll.ps}).
9184 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
9185 Pair upper lower (@code{pul.ps}).
9187 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
9188 Pair lower upper (@code{plu.ps}).
9190 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
9191 Pair upper upper (@code{puu.ps}).
9193 @item v2sf __builtin_mips_cvt_ps_s (float, float)
9194 Convert pair to paired single (@code{cvt.ps.s}).
9196 @item float __builtin_mips_cvt_s_pl (v2sf)
9197 Convert pair lower to single (@code{cvt.s.pl}).
9199 @item float __builtin_mips_cvt_s_pu (v2sf)
9200 Convert pair upper to single (@code{cvt.s.pu}).
9202 @item v2sf __builtin_mips_abs_ps (v2sf)
9203 Absolute value (@code{abs.ps}).
9205 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
9206 Align variable (@code{alnv.ps}).
9208 @emph{Note:} The value of the third parameter must be 0 or 4
9209 modulo 8, otherwise the result will be unpredictable. Please read the
9210 instruction description for details.
9213 The following multi-instruction functions are also available.
9214 In each case, @var{cond} can be any of the 16 floating-point conditions:
9215 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
9216 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
9217 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
9220 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9221 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9222 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
9223 @code{movt.ps}/@code{movf.ps}).
9225 The @code{movt} functions return the value @var{x} computed by:
9228 c.@var{cond}.ps @var{cc},@var{a},@var{b}
9229 mov.ps @var{x},@var{c}
9230 movt.ps @var{x},@var{d},@var{cc}
9233 The @code{movf} functions are similar but use @code{movf.ps} instead
9236 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9237 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9238 Comparison of two paired-single values (@code{c.@var{cond}.ps},
9239 @code{bc1t}/@code{bc1f}).
9241 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
9242 and return either the upper or lower half of the result. For example:
9246 if (__builtin_mips_upper_c_eq_ps (a, b))
9247 upper_halves_are_equal ();
9249 upper_halves_are_unequal ();
9251 if (__builtin_mips_lower_c_eq_ps (a, b))
9252 lower_halves_are_equal ();
9254 lower_halves_are_unequal ();
9258 @node MIPS-3D Built-in Functions
9259 @subsubsection MIPS-3D Built-in Functions
9261 The MIPS-3D Application-Specific Extension (ASE) includes additional
9262 paired-single instructions that are designed to improve the performance
9263 of 3D graphics operations. Support for these instructions is controlled
9264 by the @option{-mips3d} command-line option.
9266 The functions listed below map directly to a particular MIPS-3D
9267 instruction. Please refer to the architecture specification for
9268 more details on what each instruction does.
9271 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
9272 Reduction add (@code{addr.ps}).
9274 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
9275 Reduction multiply (@code{mulr.ps}).
9277 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
9278 Convert paired single to paired word (@code{cvt.pw.ps}).
9280 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
9281 Convert paired word to paired single (@code{cvt.ps.pw}).
9283 @item float __builtin_mips_recip1_s (float)
9284 @itemx double __builtin_mips_recip1_d (double)
9285 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
9286 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
9288 @item float __builtin_mips_recip2_s (float, float)
9289 @itemx double __builtin_mips_recip2_d (double, double)
9290 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
9291 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
9293 @item float __builtin_mips_rsqrt1_s (float)
9294 @itemx double __builtin_mips_rsqrt1_d (double)
9295 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
9296 Reduced precision reciprocal square root (sequence step 1)
9297 (@code{rsqrt1.@var{fmt}}).
9299 @item float __builtin_mips_rsqrt2_s (float, float)
9300 @itemx double __builtin_mips_rsqrt2_d (double, double)
9301 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
9302 Reduced precision reciprocal square root (sequence step 2)
9303 (@code{rsqrt2.@var{fmt}}).
9306 The following multi-instruction functions are also available.
9307 In each case, @var{cond} can be any of the 16 floating-point conditions:
9308 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
9309 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
9310 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
9313 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
9314 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
9315 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
9316 @code{bc1t}/@code{bc1f}).
9318 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
9319 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
9324 if (__builtin_mips_cabs_eq_s (a, b))
9330 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9331 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9332 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
9333 @code{bc1t}/@code{bc1f}).
9335 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
9336 and return either the upper or lower half of the result. For example:
9340 if (__builtin_mips_upper_cabs_eq_ps (a, b))
9341 upper_halves_are_equal ();
9343 upper_halves_are_unequal ();
9345 if (__builtin_mips_lower_cabs_eq_ps (a, b))
9346 lower_halves_are_equal ();
9348 lower_halves_are_unequal ();
9351 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9352 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9353 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
9354 @code{movt.ps}/@code{movf.ps}).
9356 The @code{movt} functions return the value @var{x} computed by:
9359 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
9360 mov.ps @var{x},@var{c}
9361 movt.ps @var{x},@var{d},@var{cc}
9364 The @code{movf} functions are similar but use @code{movf.ps} instead
9367 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9368 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9369 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9370 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9371 Comparison of two paired-single values
9372 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
9373 @code{bc1any2t}/@code{bc1any2f}).
9375 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
9376 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
9377 result is true and the @code{all} forms return true if both results are true.
9382 if (__builtin_mips_any_c_eq_ps (a, b))
9387 if (__builtin_mips_all_c_eq_ps (a, b))
9393 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9394 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9395 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9396 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9397 Comparison of four paired-single values
9398 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
9399 @code{bc1any4t}/@code{bc1any4f}).
9401 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
9402 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
9403 The @code{any} forms return true if any of the four results are true
9404 and the @code{all} forms return true if all four results are true.
9409 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
9414 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
9421 @node picoChip Built-in Functions
9422 @subsection picoChip Built-in Functions
9424 GCC provides an interface to selected machine instructions from the
9425 picoChip instruction set.
9428 @item int __builtin_sbc (int @var{value})
9429 Sign bit count. Return the number of consecutive bits in @var{value}
9430 which have the same value as the sign-bit. The result is the number of
9431 leading sign bits minus one, giving the number of redundant sign bits in
9434 @item int __builtin_byteswap (int @var{value})
9435 Byte swap. Return the result of swapping the upper and lower bytes of
9438 @item int __builtin_brev (int @var{value})
9439 Bit reversal. Return the result of reversing the bits in
9440 @var{value}. Bit 15 is swapped with bit 0, bit 14 is swapped with bit 1,
9443 @item int __builtin_adds (int @var{x}, int @var{y})
9444 Saturating addition. Return the result of adding @var{x} and @var{y},
9445 storing the value 32767 if the result overflows.
9447 @item int __builtin_subs (int @var{x}, int @var{y})
9448 Saturating subtraction. Return the result of subtracting @var{y} from
9449 @var{x}, storing the value -32768 if the result overflows.
9451 @item void __builtin_halt (void)
9452 Halt. The processor will stop execution. This built-in is useful for
9453 implementing assertions.
9457 @node Other MIPS Built-in Functions
9458 @subsection Other MIPS Built-in Functions
9460 GCC provides other MIPS-specific built-in functions:
9463 @item void __builtin_mips_cache (int @var{op}, const volatile void *@var{addr})
9464 Insert a @samp{cache} instruction with operands @var{op} and @var{addr}.
9465 GCC defines the preprocessor macro @code{___GCC_HAVE_BUILTIN_MIPS_CACHE}
9466 when this function is available.
9469 @node PowerPC AltiVec Built-in Functions
9470 @subsection PowerPC AltiVec Built-in Functions
9472 GCC provides an interface for the PowerPC family of processors to access
9473 the AltiVec operations described in Motorola's AltiVec Programming
9474 Interface Manual. The interface is made available by including
9475 @code{<altivec.h>} and using @option{-maltivec} and
9476 @option{-mabi=altivec}. The interface supports the following vector
9480 vector unsigned char
9484 vector unsigned short
9495 GCC's implementation of the high-level language interface available from
9496 C and C++ code differs from Motorola's documentation in several ways.
9501 A vector constant is a list of constant expressions within curly braces.
9504 A vector initializer requires no cast if the vector constant is of the
9505 same type as the variable it is initializing.
9508 If @code{signed} or @code{unsigned} is omitted, the signedness of the
9509 vector type is the default signedness of the base type. The default
9510 varies depending on the operating system, so a portable program should
9511 always specify the signedness.
9514 Compiling with @option{-maltivec} adds keywords @code{__vector},
9515 @code{vector}, @code{__pixel}, @code{pixel}, @code{__bool} and
9516 @code{bool}. When compiling ISO C, the context-sensitive substitution
9517 of the keywords @code{vector}, @code{pixel} and @code{bool} is
9518 disabled. To use them, you must include @code{<altivec.h>} instead.
9521 GCC allows using a @code{typedef} name as the type specifier for a
9525 For C, overloaded functions are implemented with macros so the following
9529 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
9532 Since @code{vec_add} is a macro, the vector constant in the example
9533 is treated as four separate arguments. Wrap the entire argument in
9534 parentheses for this to work.
9537 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
9538 Internally, GCC uses built-in functions to achieve the functionality in
9539 the aforementioned header file, but they are not supported and are
9540 subject to change without notice.
9542 The following interfaces are supported for the generic and specific
9543 AltiVec operations and the AltiVec predicates. In cases where there
9544 is a direct mapping between generic and specific operations, only the
9545 generic names are shown here, although the specific operations can also
9548 Arguments that are documented as @code{const int} require literal
9549 integral values within the range required for that operation.
9552 vector signed char vec_abs (vector signed char);
9553 vector signed short vec_abs (vector signed short);
9554 vector signed int vec_abs (vector signed int);
9555 vector float vec_abs (vector float);
9557 vector signed char vec_abss (vector signed char);
9558 vector signed short vec_abss (vector signed short);
9559 vector signed int vec_abss (vector signed int);
9561 vector signed char vec_add (vector bool char, vector signed char);
9562 vector signed char vec_add (vector signed char, vector bool char);
9563 vector signed char vec_add (vector signed char, vector signed char);
9564 vector unsigned char vec_add (vector bool char, vector unsigned char);
9565 vector unsigned char vec_add (vector unsigned char, vector bool char);
9566 vector unsigned char vec_add (vector unsigned char,
9567 vector unsigned char);
9568 vector signed short vec_add (vector bool short, vector signed short);
9569 vector signed short vec_add (vector signed short, vector bool short);
9570 vector signed short vec_add (vector signed short, vector signed short);
9571 vector unsigned short vec_add (vector bool short,
9572 vector unsigned short);
9573 vector unsigned short vec_add (vector unsigned short,
9575 vector unsigned short vec_add (vector unsigned short,
9576 vector unsigned short);
9577 vector signed int vec_add (vector bool int, vector signed int);
9578 vector signed int vec_add (vector signed int, vector bool int);
9579 vector signed int vec_add (vector signed int, vector signed int);
9580 vector unsigned int vec_add (vector bool int, vector unsigned int);
9581 vector unsigned int vec_add (vector unsigned int, vector bool int);
9582 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
9583 vector float vec_add (vector float, vector float);
9585 vector float vec_vaddfp (vector float, vector float);
9587 vector signed int vec_vadduwm (vector bool int, vector signed int);
9588 vector signed int vec_vadduwm (vector signed int, vector bool int);
9589 vector signed int vec_vadduwm (vector signed int, vector signed int);
9590 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
9591 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
9592 vector unsigned int vec_vadduwm (vector unsigned int,
9593 vector unsigned int);
9595 vector signed short vec_vadduhm (vector bool short,
9596 vector signed short);
9597 vector signed short vec_vadduhm (vector signed short,
9599 vector signed short vec_vadduhm (vector signed short,
9600 vector signed short);
9601 vector unsigned short vec_vadduhm (vector bool short,
9602 vector unsigned short);
9603 vector unsigned short vec_vadduhm (vector unsigned short,
9605 vector unsigned short vec_vadduhm (vector unsigned short,
9606 vector unsigned short);
9608 vector signed char vec_vaddubm (vector bool char, vector signed char);
9609 vector signed char vec_vaddubm (vector signed char, vector bool char);
9610 vector signed char vec_vaddubm (vector signed char, vector signed char);
9611 vector unsigned char vec_vaddubm (vector bool char,
9612 vector unsigned char);
9613 vector unsigned char vec_vaddubm (vector unsigned char,
9615 vector unsigned char vec_vaddubm (vector unsigned char,
9616 vector unsigned char);
9618 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
9620 vector unsigned char vec_adds (vector bool char, vector unsigned char);
9621 vector unsigned char vec_adds (vector unsigned char, vector bool char);
9622 vector unsigned char vec_adds (vector unsigned char,
9623 vector unsigned char);
9624 vector signed char vec_adds (vector bool char, vector signed char);
9625 vector signed char vec_adds (vector signed char, vector bool char);
9626 vector signed char vec_adds (vector signed char, vector signed char);
9627 vector unsigned short vec_adds (vector bool short,
9628 vector unsigned short);
9629 vector unsigned short vec_adds (vector unsigned short,
9631 vector unsigned short vec_adds (vector unsigned short,
9632 vector unsigned short);
9633 vector signed short vec_adds (vector bool short, vector signed short);
9634 vector signed short vec_adds (vector signed short, vector bool short);
9635 vector signed short vec_adds (vector signed short, vector signed short);
9636 vector unsigned int vec_adds (vector bool int, vector unsigned int);
9637 vector unsigned int vec_adds (vector unsigned int, vector bool int);
9638 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
9639 vector signed int vec_adds (vector bool int, vector signed int);
9640 vector signed int vec_adds (vector signed int, vector bool int);
9641 vector signed int vec_adds (vector signed int, vector signed int);
9643 vector signed int vec_vaddsws (vector bool int, vector signed int);
9644 vector signed int vec_vaddsws (vector signed int, vector bool int);
9645 vector signed int vec_vaddsws (vector signed int, vector signed int);
9647 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
9648 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
9649 vector unsigned int vec_vadduws (vector unsigned int,
9650 vector unsigned int);
9652 vector signed short vec_vaddshs (vector bool short,
9653 vector signed short);
9654 vector signed short vec_vaddshs (vector signed short,
9656 vector signed short vec_vaddshs (vector signed short,
9657 vector signed short);
9659 vector unsigned short vec_vadduhs (vector bool short,
9660 vector unsigned short);
9661 vector unsigned short vec_vadduhs (vector unsigned short,
9663 vector unsigned short vec_vadduhs (vector unsigned short,
9664 vector unsigned short);
9666 vector signed char vec_vaddsbs (vector bool char, vector signed char);
9667 vector signed char vec_vaddsbs (vector signed char, vector bool char);
9668 vector signed char vec_vaddsbs (vector signed char, vector signed char);
9670 vector unsigned char vec_vaddubs (vector bool char,
9671 vector unsigned char);
9672 vector unsigned char vec_vaddubs (vector unsigned char,
9674 vector unsigned char vec_vaddubs (vector unsigned char,
9675 vector unsigned char);
9677 vector float vec_and (vector float, vector float);
9678 vector float vec_and (vector float, vector bool int);
9679 vector float vec_and (vector bool int, vector float);
9680 vector bool int vec_and (vector bool int, vector bool int);
9681 vector signed int vec_and (vector bool int, vector signed int);
9682 vector signed int vec_and (vector signed int, vector bool int);
9683 vector signed int vec_and (vector signed int, vector signed int);
9684 vector unsigned int vec_and (vector bool int, vector unsigned int);
9685 vector unsigned int vec_and (vector unsigned int, vector bool int);
9686 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
9687 vector bool short vec_and (vector bool short, vector bool short);
9688 vector signed short vec_and (vector bool short, vector signed short);
9689 vector signed short vec_and (vector signed short, vector bool short);
9690 vector signed short vec_and (vector signed short, vector signed short);
9691 vector unsigned short vec_and (vector bool short,
9692 vector unsigned short);
9693 vector unsigned short vec_and (vector unsigned short,
9695 vector unsigned short vec_and (vector unsigned short,
9696 vector unsigned short);
9697 vector signed char vec_and (vector bool char, vector signed char);
9698 vector bool char vec_and (vector bool char, vector bool char);
9699 vector signed char vec_and (vector signed char, vector bool char);
9700 vector signed char vec_and (vector signed char, vector signed char);
9701 vector unsigned char vec_and (vector bool char, vector unsigned char);
9702 vector unsigned char vec_and (vector unsigned char, vector bool char);
9703 vector unsigned char vec_and (vector unsigned char,
9704 vector unsigned char);
9706 vector float vec_andc (vector float, vector float);
9707 vector float vec_andc (vector float, vector bool int);
9708 vector float vec_andc (vector bool int, vector float);
9709 vector bool int vec_andc (vector bool int, vector bool int);
9710 vector signed int vec_andc (vector bool int, vector signed int);
9711 vector signed int vec_andc (vector signed int, vector bool int);
9712 vector signed int vec_andc (vector signed int, vector signed int);
9713 vector unsigned int vec_andc (vector bool int, vector unsigned int);
9714 vector unsigned int vec_andc (vector unsigned int, vector bool int);
9715 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
9716 vector bool short vec_andc (vector bool short, vector bool short);
9717 vector signed short vec_andc (vector bool short, vector signed short);
9718 vector signed short vec_andc (vector signed short, vector bool short);
9719 vector signed short vec_andc (vector signed short, vector signed short);
9720 vector unsigned short vec_andc (vector bool short,
9721 vector unsigned short);
9722 vector unsigned short vec_andc (vector unsigned short,
9724 vector unsigned short vec_andc (vector unsigned short,
9725 vector unsigned short);
9726 vector signed char vec_andc (vector bool char, vector signed char);
9727 vector bool char vec_andc (vector bool char, vector bool char);
9728 vector signed char vec_andc (vector signed char, vector bool char);
9729 vector signed char vec_andc (vector signed char, vector signed char);
9730 vector unsigned char vec_andc (vector bool char, vector unsigned char);
9731 vector unsigned char vec_andc (vector unsigned char, vector bool char);
9732 vector unsigned char vec_andc (vector unsigned char,
9733 vector unsigned char);
9735 vector unsigned char vec_avg (vector unsigned char,
9736 vector unsigned char);
9737 vector signed char vec_avg (vector signed char, vector signed char);
9738 vector unsigned short vec_avg (vector unsigned short,
9739 vector unsigned short);
9740 vector signed short vec_avg (vector signed short, vector signed short);
9741 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
9742 vector signed int vec_avg (vector signed int, vector signed int);
9744 vector signed int vec_vavgsw (vector signed int, vector signed int);
9746 vector unsigned int vec_vavguw (vector unsigned int,
9747 vector unsigned int);
9749 vector signed short vec_vavgsh (vector signed short,
9750 vector signed short);
9752 vector unsigned short vec_vavguh (vector unsigned short,
9753 vector unsigned short);
9755 vector signed char vec_vavgsb (vector signed char, vector signed char);
9757 vector unsigned char vec_vavgub (vector unsigned char,
9758 vector unsigned char);
9760 vector float vec_ceil (vector float);
9762 vector signed int vec_cmpb (vector float, vector float);
9764 vector bool char vec_cmpeq (vector signed char, vector signed char);
9765 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
9766 vector bool short vec_cmpeq (vector signed short, vector signed short);
9767 vector bool short vec_cmpeq (vector unsigned short,
9768 vector unsigned short);
9769 vector bool int vec_cmpeq (vector signed int, vector signed int);
9770 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
9771 vector bool int vec_cmpeq (vector float, vector float);
9773 vector bool int vec_vcmpeqfp (vector float, vector float);
9775 vector bool int vec_vcmpequw (vector signed int, vector signed int);
9776 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
9778 vector bool short vec_vcmpequh (vector signed short,
9779 vector signed short);
9780 vector bool short vec_vcmpequh (vector unsigned short,
9781 vector unsigned short);
9783 vector bool char vec_vcmpequb (vector signed char, vector signed char);
9784 vector bool char vec_vcmpequb (vector unsigned char,
9785 vector unsigned char);
9787 vector bool int vec_cmpge (vector float, vector float);
9789 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
9790 vector bool char vec_cmpgt (vector signed char, vector signed char);
9791 vector bool short vec_cmpgt (vector unsigned short,
9792 vector unsigned short);
9793 vector bool short vec_cmpgt (vector signed short, vector signed short);
9794 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
9795 vector bool int vec_cmpgt (vector signed int, vector signed int);
9796 vector bool int vec_cmpgt (vector float, vector float);
9798 vector bool int vec_vcmpgtfp (vector float, vector float);
9800 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
9802 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
9804 vector bool short vec_vcmpgtsh (vector signed short,
9805 vector signed short);
9807 vector bool short vec_vcmpgtuh (vector unsigned short,
9808 vector unsigned short);
9810 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
9812 vector bool char vec_vcmpgtub (vector unsigned char,
9813 vector unsigned char);
9815 vector bool int vec_cmple (vector float, vector float);
9817 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
9818 vector bool char vec_cmplt (vector signed char, vector signed char);
9819 vector bool short vec_cmplt (vector unsigned short,
9820 vector unsigned short);
9821 vector bool short vec_cmplt (vector signed short, vector signed short);
9822 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
9823 vector bool int vec_cmplt (vector signed int, vector signed int);
9824 vector bool int vec_cmplt (vector float, vector float);
9826 vector float vec_ctf (vector unsigned int, const int);
9827 vector float vec_ctf (vector signed int, const int);
9829 vector float vec_vcfsx (vector signed int, const int);
9831 vector float vec_vcfux (vector unsigned int, const int);
9833 vector signed int vec_cts (vector float, const int);
9835 vector unsigned int vec_ctu (vector float, const int);
9837 void vec_dss (const int);
9839 void vec_dssall (void);
9841 void vec_dst (const vector unsigned char *, int, const int);
9842 void vec_dst (const vector signed char *, int, const int);
9843 void vec_dst (const vector bool char *, int, const int);
9844 void vec_dst (const vector unsigned short *, int, const int);
9845 void vec_dst (const vector signed short *, int, const int);
9846 void vec_dst (const vector bool short *, int, const int);
9847 void vec_dst (const vector pixel *, int, const int);
9848 void vec_dst (const vector unsigned int *, int, const int);
9849 void vec_dst (const vector signed int *, int, const int);
9850 void vec_dst (const vector bool int *, int, const int);
9851 void vec_dst (const vector float *, int, const int);
9852 void vec_dst (const unsigned char *, int, const int);
9853 void vec_dst (const signed char *, int, const int);
9854 void vec_dst (const unsigned short *, int, const int);
9855 void vec_dst (const short *, int, const int);
9856 void vec_dst (const unsigned int *, int, const int);
9857 void vec_dst (const int *, int, const int);
9858 void vec_dst (const unsigned long *, int, const int);
9859 void vec_dst (const long *, int, const int);
9860 void vec_dst (const float *, int, const int);
9862 void vec_dstst (const vector unsigned char *, int, const int);
9863 void vec_dstst (const vector signed char *, int, const int);
9864 void vec_dstst (const vector bool char *, int, const int);
9865 void vec_dstst (const vector unsigned short *, int, const int);
9866 void vec_dstst (const vector signed short *, int, const int);
9867 void vec_dstst (const vector bool short *, int, const int);
9868 void vec_dstst (const vector pixel *, int, const int);
9869 void vec_dstst (const vector unsigned int *, int, const int);
9870 void vec_dstst (const vector signed int *, int, const int);
9871 void vec_dstst (const vector bool int *, int, const int);
9872 void vec_dstst (const vector float *, int, const int);
9873 void vec_dstst (const unsigned char *, int, const int);
9874 void vec_dstst (const signed char *, int, const int);
9875 void vec_dstst (const unsigned short *, int, const int);
9876 void vec_dstst (const short *, int, const int);
9877 void vec_dstst (const unsigned int *, int, const int);
9878 void vec_dstst (const int *, int, const int);
9879 void vec_dstst (const unsigned long *, int, const int);
9880 void vec_dstst (const long *, int, const int);
9881 void vec_dstst (const float *, int, const int);
9883 void vec_dststt (const vector unsigned char *, int, const int);
9884 void vec_dststt (const vector signed char *, int, const int);
9885 void vec_dststt (const vector bool char *, int, const int);
9886 void vec_dststt (const vector unsigned short *, int, const int);
9887 void vec_dststt (const vector signed short *, int, const int);
9888 void vec_dststt (const vector bool short *, int, const int);
9889 void vec_dststt (const vector pixel *, int, const int);
9890 void vec_dststt (const vector unsigned int *, int, const int);
9891 void vec_dststt (const vector signed int *, int, const int);
9892 void vec_dststt (const vector bool int *, int, const int);
9893 void vec_dststt (const vector float *, int, const int);
9894 void vec_dststt (const unsigned char *, int, const int);
9895 void vec_dststt (const signed char *, int, const int);
9896 void vec_dststt (const unsigned short *, int, const int);
9897 void vec_dststt (const short *, int, const int);
9898 void vec_dststt (const unsigned int *, int, const int);
9899 void vec_dststt (const int *, int, const int);
9900 void vec_dststt (const unsigned long *, int, const int);
9901 void vec_dststt (const long *, int, const int);
9902 void vec_dststt (const float *, int, const int);
9904 void vec_dstt (const vector unsigned char *, int, const int);
9905 void vec_dstt (const vector signed char *, int, const int);
9906 void vec_dstt (const vector bool char *, int, const int);
9907 void vec_dstt (const vector unsigned short *, int, const int);
9908 void vec_dstt (const vector signed short *, int, const int);
9909 void vec_dstt (const vector bool short *, int, const int);
9910 void vec_dstt (const vector pixel *, int, const int);
9911 void vec_dstt (const vector unsigned int *, int, const int);
9912 void vec_dstt (const vector signed int *, int, const int);
9913 void vec_dstt (const vector bool int *, int, const int);
9914 void vec_dstt (const vector float *, int, const int);
9915 void vec_dstt (const unsigned char *, int, const int);
9916 void vec_dstt (const signed char *, int, const int);
9917 void vec_dstt (const unsigned short *, int, const int);
9918 void vec_dstt (const short *, int, const int);
9919 void vec_dstt (const unsigned int *, int, const int);
9920 void vec_dstt (const int *, int, const int);
9921 void vec_dstt (const unsigned long *, int, const int);
9922 void vec_dstt (const long *, int, const int);
9923 void vec_dstt (const float *, int, const int);
9925 vector float vec_expte (vector float);
9927 vector float vec_floor (vector float);
9929 vector float vec_ld (int, const vector float *);
9930 vector float vec_ld (int, const float *);
9931 vector bool int vec_ld (int, const vector bool int *);
9932 vector signed int vec_ld (int, const vector signed int *);
9933 vector signed int vec_ld (int, const int *);
9934 vector signed int vec_ld (int, const long *);
9935 vector unsigned int vec_ld (int, const vector unsigned int *);
9936 vector unsigned int vec_ld (int, const unsigned int *);
9937 vector unsigned int vec_ld (int, const unsigned long *);
9938 vector bool short vec_ld (int, const vector bool short *);
9939 vector pixel vec_ld (int, const vector pixel *);
9940 vector signed short vec_ld (int, const vector signed short *);
9941 vector signed short vec_ld (int, const short *);
9942 vector unsigned short vec_ld (int, const vector unsigned short *);
9943 vector unsigned short vec_ld (int, const unsigned short *);
9944 vector bool char vec_ld (int, const vector bool char *);
9945 vector signed char vec_ld (int, const vector signed char *);
9946 vector signed char vec_ld (int, const signed char *);
9947 vector unsigned char vec_ld (int, const vector unsigned char *);
9948 vector unsigned char vec_ld (int, const unsigned char *);
9950 vector signed char vec_lde (int, const signed char *);
9951 vector unsigned char vec_lde (int, const unsigned char *);
9952 vector signed short vec_lde (int, const short *);
9953 vector unsigned short vec_lde (int, const unsigned short *);
9954 vector float vec_lde (int, const float *);
9955 vector signed int vec_lde (int, const int *);
9956 vector unsigned int vec_lde (int, const unsigned int *);
9957 vector signed int vec_lde (int, const long *);
9958 vector unsigned int vec_lde (int, const unsigned long *);
9960 vector float vec_lvewx (int, float *);
9961 vector signed int vec_lvewx (int, int *);
9962 vector unsigned int vec_lvewx (int, unsigned int *);
9963 vector signed int vec_lvewx (int, long *);
9964 vector unsigned int vec_lvewx (int, unsigned long *);
9966 vector signed short vec_lvehx (int, short *);
9967 vector unsigned short vec_lvehx (int, unsigned short *);
9969 vector signed char vec_lvebx (int, char *);
9970 vector unsigned char vec_lvebx (int, unsigned char *);
9972 vector float vec_ldl (int, const vector float *);
9973 vector float vec_ldl (int, const float *);
9974 vector bool int vec_ldl (int, const vector bool int *);
9975 vector signed int vec_ldl (int, const vector signed int *);
9976 vector signed int vec_ldl (int, const int *);
9977 vector signed int vec_ldl (int, const long *);
9978 vector unsigned int vec_ldl (int, const vector unsigned int *);
9979 vector unsigned int vec_ldl (int, const unsigned int *);
9980 vector unsigned int vec_ldl (int, const unsigned long *);
9981 vector bool short vec_ldl (int, const vector bool short *);
9982 vector pixel vec_ldl (int, const vector pixel *);
9983 vector signed short vec_ldl (int, const vector signed short *);
9984 vector signed short vec_ldl (int, const short *);
9985 vector unsigned short vec_ldl (int, const vector unsigned short *);
9986 vector unsigned short vec_ldl (int, const unsigned short *);
9987 vector bool char vec_ldl (int, const vector bool char *);
9988 vector signed char vec_ldl (int, const vector signed char *);
9989 vector signed char vec_ldl (int, const signed char *);
9990 vector unsigned char vec_ldl (int, const vector unsigned char *);
9991 vector unsigned char vec_ldl (int, const unsigned char *);
9993 vector float vec_loge (vector float);
9995 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
9996 vector unsigned char vec_lvsl (int, const volatile signed char *);
9997 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
9998 vector unsigned char vec_lvsl (int, const volatile short *);
9999 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
10000 vector unsigned char vec_lvsl (int, const volatile int *);
10001 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
10002 vector unsigned char vec_lvsl (int, const volatile long *);
10003 vector unsigned char vec_lvsl (int, const volatile float *);
10005 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
10006 vector unsigned char vec_lvsr (int, const volatile signed char *);
10007 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
10008 vector unsigned char vec_lvsr (int, const volatile short *);
10009 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
10010 vector unsigned char vec_lvsr (int, const volatile int *);
10011 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
10012 vector unsigned char vec_lvsr (int, const volatile long *);
10013 vector unsigned char vec_lvsr (int, const volatile float *);
10015 vector float vec_madd (vector float, vector float, vector float);
10017 vector signed short vec_madds (vector signed short,
10018 vector signed short,
10019 vector signed short);
10021 vector unsigned char vec_max (vector bool char, vector unsigned char);
10022 vector unsigned char vec_max (vector unsigned char, vector bool char);
10023 vector unsigned char vec_max (vector unsigned char,
10024 vector unsigned char);
10025 vector signed char vec_max (vector bool char, vector signed char);
10026 vector signed char vec_max (vector signed char, vector bool char);
10027 vector signed char vec_max (vector signed char, vector signed char);
10028 vector unsigned short vec_max (vector bool short,
10029 vector unsigned short);
10030 vector unsigned short vec_max (vector unsigned short,
10031 vector bool short);
10032 vector unsigned short vec_max (vector unsigned short,
10033 vector unsigned short);
10034 vector signed short vec_max (vector bool short, vector signed short);
10035 vector signed short vec_max (vector signed short, vector bool short);
10036 vector signed short vec_max (vector signed short, vector signed short);
10037 vector unsigned int vec_max (vector bool int, vector unsigned int);
10038 vector unsigned int vec_max (vector unsigned int, vector bool int);
10039 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
10040 vector signed int vec_max (vector bool int, vector signed int);
10041 vector signed int vec_max (vector signed int, vector bool int);
10042 vector signed int vec_max (vector signed int, vector signed int);
10043 vector float vec_max (vector float, vector float);
10045 vector float vec_vmaxfp (vector float, vector float);
10047 vector signed int vec_vmaxsw (vector bool int, vector signed int);
10048 vector signed int vec_vmaxsw (vector signed int, vector bool int);
10049 vector signed int vec_vmaxsw (vector signed int, vector signed int);
10051 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
10052 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
10053 vector unsigned int vec_vmaxuw (vector unsigned int,
10054 vector unsigned int);
10056 vector signed short vec_vmaxsh (vector bool short, vector signed short);
10057 vector signed short vec_vmaxsh (vector signed short, vector bool short);
10058 vector signed short vec_vmaxsh (vector signed short,
10059 vector signed short);
10061 vector unsigned short vec_vmaxuh (vector bool short,
10062 vector unsigned short);
10063 vector unsigned short vec_vmaxuh (vector unsigned short,
10064 vector bool short);
10065 vector unsigned short vec_vmaxuh (vector unsigned short,
10066 vector unsigned short);
10068 vector signed char vec_vmaxsb (vector bool char, vector signed char);
10069 vector signed char vec_vmaxsb (vector signed char, vector bool char);
10070 vector signed char vec_vmaxsb (vector signed char, vector signed char);
10072 vector unsigned char vec_vmaxub (vector bool char,
10073 vector unsigned char);
10074 vector unsigned char vec_vmaxub (vector unsigned char,
10076 vector unsigned char vec_vmaxub (vector unsigned char,
10077 vector unsigned char);
10079 vector bool char vec_mergeh (vector bool char, vector bool char);
10080 vector signed char vec_mergeh (vector signed char, vector signed char);
10081 vector unsigned char vec_mergeh (vector unsigned char,
10082 vector unsigned char);
10083 vector bool short vec_mergeh (vector bool short, vector bool short);
10084 vector pixel vec_mergeh (vector pixel, vector pixel);
10085 vector signed short vec_mergeh (vector signed short,
10086 vector signed short);
10087 vector unsigned short vec_mergeh (vector unsigned short,
10088 vector unsigned short);
10089 vector float vec_mergeh (vector float, vector float);
10090 vector bool int vec_mergeh (vector bool int, vector bool int);
10091 vector signed int vec_mergeh (vector signed int, vector signed int);
10092 vector unsigned int vec_mergeh (vector unsigned int,
10093 vector unsigned int);
10095 vector float vec_vmrghw (vector float, vector float);
10096 vector bool int vec_vmrghw (vector bool int, vector bool int);
10097 vector signed int vec_vmrghw (vector signed int, vector signed int);
10098 vector unsigned int vec_vmrghw (vector unsigned int,
10099 vector unsigned int);
10101 vector bool short vec_vmrghh (vector bool short, vector bool short);
10102 vector signed short vec_vmrghh (vector signed short,
10103 vector signed short);
10104 vector unsigned short vec_vmrghh (vector unsigned short,
10105 vector unsigned short);
10106 vector pixel vec_vmrghh (vector pixel, vector pixel);
10108 vector bool char vec_vmrghb (vector bool char, vector bool char);
10109 vector signed char vec_vmrghb (vector signed char, vector signed char);
10110 vector unsigned char vec_vmrghb (vector unsigned char,
10111 vector unsigned char);
10113 vector bool char vec_mergel (vector bool char, vector bool char);
10114 vector signed char vec_mergel (vector signed char, vector signed char);
10115 vector unsigned char vec_mergel (vector unsigned char,
10116 vector unsigned char);
10117 vector bool short vec_mergel (vector bool short, vector bool short);
10118 vector pixel vec_mergel (vector pixel, vector pixel);
10119 vector signed short vec_mergel (vector signed short,
10120 vector signed short);
10121 vector unsigned short vec_mergel (vector unsigned short,
10122 vector unsigned short);
10123 vector float vec_mergel (vector float, vector float);
10124 vector bool int vec_mergel (vector bool int, vector bool int);
10125 vector signed int vec_mergel (vector signed int, vector signed int);
10126 vector unsigned int vec_mergel (vector unsigned int,
10127 vector unsigned int);
10129 vector float vec_vmrglw (vector float, vector float);
10130 vector signed int vec_vmrglw (vector signed int, vector signed int);
10131 vector unsigned int vec_vmrglw (vector unsigned int,
10132 vector unsigned int);
10133 vector bool int vec_vmrglw (vector bool int, vector bool int);
10135 vector bool short vec_vmrglh (vector bool short, vector bool short);
10136 vector signed short vec_vmrglh (vector signed short,
10137 vector signed short);
10138 vector unsigned short vec_vmrglh (vector unsigned short,
10139 vector unsigned short);
10140 vector pixel vec_vmrglh (vector pixel, vector pixel);
10142 vector bool char vec_vmrglb (vector bool char, vector bool char);
10143 vector signed char vec_vmrglb (vector signed char, vector signed char);
10144 vector unsigned char vec_vmrglb (vector unsigned char,
10145 vector unsigned char);
10147 vector unsigned short vec_mfvscr (void);
10149 vector unsigned char vec_min (vector bool char, vector unsigned char);
10150 vector unsigned char vec_min (vector unsigned char, vector bool char);
10151 vector unsigned char vec_min (vector unsigned char,
10152 vector unsigned char);
10153 vector signed char vec_min (vector bool char, vector signed char);
10154 vector signed char vec_min (vector signed char, vector bool char);
10155 vector signed char vec_min (vector signed char, vector signed char);
10156 vector unsigned short vec_min (vector bool short,
10157 vector unsigned short);
10158 vector unsigned short vec_min (vector unsigned short,
10159 vector bool short);
10160 vector unsigned short vec_min (vector unsigned short,
10161 vector unsigned short);
10162 vector signed short vec_min (vector bool short, vector signed short);
10163 vector signed short vec_min (vector signed short, vector bool short);
10164 vector signed short vec_min (vector signed short, vector signed short);
10165 vector unsigned int vec_min (vector bool int, vector unsigned int);
10166 vector unsigned int vec_min (vector unsigned int, vector bool int);
10167 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
10168 vector signed int vec_min (vector bool int, vector signed int);
10169 vector signed int vec_min (vector signed int, vector bool int);
10170 vector signed int vec_min (vector signed int, vector signed int);
10171 vector float vec_min (vector float, vector float);
10173 vector float vec_vminfp (vector float, vector float);
10175 vector signed int vec_vminsw (vector bool int, vector signed int);
10176 vector signed int vec_vminsw (vector signed int, vector bool int);
10177 vector signed int vec_vminsw (vector signed int, vector signed int);
10179 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
10180 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
10181 vector unsigned int vec_vminuw (vector unsigned int,
10182 vector unsigned int);
10184 vector signed short vec_vminsh (vector bool short, vector signed short);
10185 vector signed short vec_vminsh (vector signed short, vector bool short);
10186 vector signed short vec_vminsh (vector signed short,
10187 vector signed short);
10189 vector unsigned short vec_vminuh (vector bool short,
10190 vector unsigned short);
10191 vector unsigned short vec_vminuh (vector unsigned short,
10192 vector bool short);
10193 vector unsigned short vec_vminuh (vector unsigned short,
10194 vector unsigned short);
10196 vector signed char vec_vminsb (vector bool char, vector signed char);
10197 vector signed char vec_vminsb (vector signed char, vector bool char);
10198 vector signed char vec_vminsb (vector signed char, vector signed char);
10200 vector unsigned char vec_vminub (vector bool char,
10201 vector unsigned char);
10202 vector unsigned char vec_vminub (vector unsigned char,
10204 vector unsigned char vec_vminub (vector unsigned char,
10205 vector unsigned char);
10207 vector signed short vec_mladd (vector signed short,
10208 vector signed short,
10209 vector signed short);
10210 vector signed short vec_mladd (vector signed short,
10211 vector unsigned short,
10212 vector unsigned short);
10213 vector signed short vec_mladd (vector unsigned short,
10214 vector signed short,
10215 vector signed short);
10216 vector unsigned short vec_mladd (vector unsigned short,
10217 vector unsigned short,
10218 vector unsigned short);
10220 vector signed short vec_mradds (vector signed short,
10221 vector signed short,
10222 vector signed short);
10224 vector unsigned int vec_msum (vector unsigned char,
10225 vector unsigned char,
10226 vector unsigned int);
10227 vector signed int vec_msum (vector signed char,
10228 vector unsigned char,
10229 vector signed int);
10230 vector unsigned int vec_msum (vector unsigned short,
10231 vector unsigned short,
10232 vector unsigned int);
10233 vector signed int vec_msum (vector signed short,
10234 vector signed short,
10235 vector signed int);
10237 vector signed int vec_vmsumshm (vector signed short,
10238 vector signed short,
10239 vector signed int);
10241 vector unsigned int vec_vmsumuhm (vector unsigned short,
10242 vector unsigned short,
10243 vector unsigned int);
10245 vector signed int vec_vmsummbm (vector signed char,
10246 vector unsigned char,
10247 vector signed int);
10249 vector unsigned int vec_vmsumubm (vector unsigned char,
10250 vector unsigned char,
10251 vector unsigned int);
10253 vector unsigned int vec_msums (vector unsigned short,
10254 vector unsigned short,
10255 vector unsigned int);
10256 vector signed int vec_msums (vector signed short,
10257 vector signed short,
10258 vector signed int);
10260 vector signed int vec_vmsumshs (vector signed short,
10261 vector signed short,
10262 vector signed int);
10264 vector unsigned int vec_vmsumuhs (vector unsigned short,
10265 vector unsigned short,
10266 vector unsigned int);
10268 void vec_mtvscr (vector signed int);
10269 void vec_mtvscr (vector unsigned int);
10270 void vec_mtvscr (vector bool int);
10271 void vec_mtvscr (vector signed short);
10272 void vec_mtvscr (vector unsigned short);
10273 void vec_mtvscr (vector bool short);
10274 void vec_mtvscr (vector pixel);
10275 void vec_mtvscr (vector signed char);
10276 void vec_mtvscr (vector unsigned char);
10277 void vec_mtvscr (vector bool char);
10279 vector unsigned short vec_mule (vector unsigned char,
10280 vector unsigned char);
10281 vector signed short vec_mule (vector signed char,
10282 vector signed char);
10283 vector unsigned int vec_mule (vector unsigned short,
10284 vector unsigned short);
10285 vector signed int vec_mule (vector signed short, vector signed short);
10287 vector signed int vec_vmulesh (vector signed short,
10288 vector signed short);
10290 vector unsigned int vec_vmuleuh (vector unsigned short,
10291 vector unsigned short);
10293 vector signed short vec_vmulesb (vector signed char,
10294 vector signed char);
10296 vector unsigned short vec_vmuleub (vector unsigned char,
10297 vector unsigned char);
10299 vector unsigned short vec_mulo (vector unsigned char,
10300 vector unsigned char);
10301 vector signed short vec_mulo (vector signed char, vector signed char);
10302 vector unsigned int vec_mulo (vector unsigned short,
10303 vector unsigned short);
10304 vector signed int vec_mulo (vector signed short, vector signed short);
10306 vector signed int vec_vmulosh (vector signed short,
10307 vector signed short);
10309 vector unsigned int vec_vmulouh (vector unsigned short,
10310 vector unsigned short);
10312 vector signed short vec_vmulosb (vector signed char,
10313 vector signed char);
10315 vector unsigned short vec_vmuloub (vector unsigned char,
10316 vector unsigned char);
10318 vector float vec_nmsub (vector float, vector float, vector float);
10320 vector float vec_nor (vector float, vector float);
10321 vector signed int vec_nor (vector signed int, vector signed int);
10322 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
10323 vector bool int vec_nor (vector bool int, vector bool int);
10324 vector signed short vec_nor (vector signed short, vector signed short);
10325 vector unsigned short vec_nor (vector unsigned short,
10326 vector unsigned short);
10327 vector bool short vec_nor (vector bool short, vector bool short);
10328 vector signed char vec_nor (vector signed char, vector signed char);
10329 vector unsigned char vec_nor (vector unsigned char,
10330 vector unsigned char);
10331 vector bool char vec_nor (vector bool char, vector bool char);
10333 vector float vec_or (vector float, vector float);
10334 vector float vec_or (vector float, vector bool int);
10335 vector float vec_or (vector bool int, vector float);
10336 vector bool int vec_or (vector bool int, vector bool int);
10337 vector signed int vec_or (vector bool int, vector signed int);
10338 vector signed int vec_or (vector signed int, vector bool int);
10339 vector signed int vec_or (vector signed int, vector signed int);
10340 vector unsigned int vec_or (vector bool int, vector unsigned int);
10341 vector unsigned int vec_or (vector unsigned int, vector bool int);
10342 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
10343 vector bool short vec_or (vector bool short, vector bool short);
10344 vector signed short vec_or (vector bool short, vector signed short);
10345 vector signed short vec_or (vector signed short, vector bool short);
10346 vector signed short vec_or (vector signed short, vector signed short);
10347 vector unsigned short vec_or (vector bool short, vector unsigned short);
10348 vector unsigned short vec_or (vector unsigned short, vector bool short);
10349 vector unsigned short vec_or (vector unsigned short,
10350 vector unsigned short);
10351 vector signed char vec_or (vector bool char, vector signed char);
10352 vector bool char vec_or (vector bool char, vector bool char);
10353 vector signed char vec_or (vector signed char, vector bool char);
10354 vector signed char vec_or (vector signed char, vector signed char);
10355 vector unsigned char vec_or (vector bool char, vector unsigned char);
10356 vector unsigned char vec_or (vector unsigned char, vector bool char);
10357 vector unsigned char vec_or (vector unsigned char,
10358 vector unsigned char);
10360 vector signed char vec_pack (vector signed short, vector signed short);
10361 vector unsigned char vec_pack (vector unsigned short,
10362 vector unsigned short);
10363 vector bool char vec_pack (vector bool short, vector bool short);
10364 vector signed short vec_pack (vector signed int, vector signed int);
10365 vector unsigned short vec_pack (vector unsigned int,
10366 vector unsigned int);
10367 vector bool short vec_pack (vector bool int, vector bool int);
10369 vector bool short vec_vpkuwum (vector bool int, vector bool int);
10370 vector signed short vec_vpkuwum (vector signed int, vector signed int);
10371 vector unsigned short vec_vpkuwum (vector unsigned int,
10372 vector unsigned int);
10374 vector bool char vec_vpkuhum (vector bool short, vector bool short);
10375 vector signed char vec_vpkuhum (vector signed short,
10376 vector signed short);
10377 vector unsigned char vec_vpkuhum (vector unsigned short,
10378 vector unsigned short);
10380 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
10382 vector unsigned char vec_packs (vector unsigned short,
10383 vector unsigned short);
10384 vector signed char vec_packs (vector signed short, vector signed short);
10385 vector unsigned short vec_packs (vector unsigned int,
10386 vector unsigned int);
10387 vector signed short vec_packs (vector signed int, vector signed int);
10389 vector signed short vec_vpkswss (vector signed int, vector signed int);
10391 vector unsigned short vec_vpkuwus (vector unsigned int,
10392 vector unsigned int);
10394 vector signed char vec_vpkshss (vector signed short,
10395 vector signed short);
10397 vector unsigned char vec_vpkuhus (vector unsigned short,
10398 vector unsigned short);
10400 vector unsigned char vec_packsu (vector unsigned short,
10401 vector unsigned short);
10402 vector unsigned char vec_packsu (vector signed short,
10403 vector signed short);
10404 vector unsigned short vec_packsu (vector unsigned int,
10405 vector unsigned int);
10406 vector unsigned short vec_packsu (vector signed int, vector signed int);
10408 vector unsigned short vec_vpkswus (vector signed int,
10409 vector signed int);
10411 vector unsigned char vec_vpkshus (vector signed short,
10412 vector signed short);
10414 vector float vec_perm (vector float,
10416 vector unsigned char);
10417 vector signed int vec_perm (vector signed int,
10419 vector unsigned char);
10420 vector unsigned int vec_perm (vector unsigned int,
10421 vector unsigned int,
10422 vector unsigned char);
10423 vector bool int vec_perm (vector bool int,
10425 vector unsigned char);
10426 vector signed short vec_perm (vector signed short,
10427 vector signed short,
10428 vector unsigned char);
10429 vector unsigned short vec_perm (vector unsigned short,
10430 vector unsigned short,
10431 vector unsigned char);
10432 vector bool short vec_perm (vector bool short,
10434 vector unsigned char);
10435 vector pixel vec_perm (vector pixel,
10437 vector unsigned char);
10438 vector signed char vec_perm (vector signed char,
10439 vector signed char,
10440 vector unsigned char);
10441 vector unsigned char vec_perm (vector unsigned char,
10442 vector unsigned char,
10443 vector unsigned char);
10444 vector bool char vec_perm (vector bool char,
10446 vector unsigned char);
10448 vector float vec_re (vector float);
10450 vector signed char vec_rl (vector signed char,
10451 vector unsigned char);
10452 vector unsigned char vec_rl (vector unsigned char,
10453 vector unsigned char);
10454 vector signed short vec_rl (vector signed short, vector unsigned short);
10455 vector unsigned short vec_rl (vector unsigned short,
10456 vector unsigned short);
10457 vector signed int vec_rl (vector signed int, vector unsigned int);
10458 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
10460 vector signed int vec_vrlw (vector signed int, vector unsigned int);
10461 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
10463 vector signed short vec_vrlh (vector signed short,
10464 vector unsigned short);
10465 vector unsigned short vec_vrlh (vector unsigned short,
10466 vector unsigned short);
10468 vector signed char vec_vrlb (vector signed char, vector unsigned char);
10469 vector unsigned char vec_vrlb (vector unsigned char,
10470 vector unsigned char);
10472 vector float vec_round (vector float);
10474 vector float vec_rsqrte (vector float);
10476 vector float vec_sel (vector float, vector float, vector bool int);
10477 vector float vec_sel (vector float, vector float, vector unsigned int);
10478 vector signed int vec_sel (vector signed int,
10481 vector signed int vec_sel (vector signed int,
10483 vector unsigned int);
10484 vector unsigned int vec_sel (vector unsigned int,
10485 vector unsigned int,
10487 vector unsigned int vec_sel (vector unsigned int,
10488 vector unsigned int,
10489 vector unsigned int);
10490 vector bool int vec_sel (vector bool int,
10493 vector bool int vec_sel (vector bool int,
10495 vector unsigned int);
10496 vector signed short vec_sel (vector signed short,
10497 vector signed short,
10498 vector bool short);
10499 vector signed short vec_sel (vector signed short,
10500 vector signed short,
10501 vector unsigned short);
10502 vector unsigned short vec_sel (vector unsigned short,
10503 vector unsigned short,
10504 vector bool short);
10505 vector unsigned short vec_sel (vector unsigned short,
10506 vector unsigned short,
10507 vector unsigned short);
10508 vector bool short vec_sel (vector bool short,
10510 vector bool short);
10511 vector bool short vec_sel (vector bool short,
10513 vector unsigned short);
10514 vector signed char vec_sel (vector signed char,
10515 vector signed char,
10517 vector signed char vec_sel (vector signed char,
10518 vector signed char,
10519 vector unsigned char);
10520 vector unsigned char vec_sel (vector unsigned char,
10521 vector unsigned char,
10523 vector unsigned char vec_sel (vector unsigned char,
10524 vector unsigned char,
10525 vector unsigned char);
10526 vector bool char vec_sel (vector bool char,
10529 vector bool char vec_sel (vector bool char,
10531 vector unsigned char);
10533 vector signed char vec_sl (vector signed char,
10534 vector unsigned char);
10535 vector unsigned char vec_sl (vector unsigned char,
10536 vector unsigned char);
10537 vector signed short vec_sl (vector signed short, vector unsigned short);
10538 vector unsigned short vec_sl (vector unsigned short,
10539 vector unsigned short);
10540 vector signed int vec_sl (vector signed int, vector unsigned int);
10541 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
10543 vector signed int vec_vslw (vector signed int, vector unsigned int);
10544 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
10546 vector signed short vec_vslh (vector signed short,
10547 vector unsigned short);
10548 vector unsigned short vec_vslh (vector unsigned short,
10549 vector unsigned short);
10551 vector signed char vec_vslb (vector signed char, vector unsigned char);
10552 vector unsigned char vec_vslb (vector unsigned char,
10553 vector unsigned char);
10555 vector float vec_sld (vector float, vector float, const int);
10556 vector signed int vec_sld (vector signed int,
10559 vector unsigned int vec_sld (vector unsigned int,
10560 vector unsigned int,
10562 vector bool int vec_sld (vector bool int,
10565 vector signed short vec_sld (vector signed short,
10566 vector signed short,
10568 vector unsigned short vec_sld (vector unsigned short,
10569 vector unsigned short,
10571 vector bool short vec_sld (vector bool short,
10574 vector pixel vec_sld (vector pixel,
10577 vector signed char vec_sld (vector signed char,
10578 vector signed char,
10580 vector unsigned char vec_sld (vector unsigned char,
10581 vector unsigned char,
10583 vector bool char vec_sld (vector bool char,
10587 vector signed int vec_sll (vector signed int,
10588 vector unsigned int);
10589 vector signed int vec_sll (vector signed int,
10590 vector unsigned short);
10591 vector signed int vec_sll (vector signed int,
10592 vector unsigned char);
10593 vector unsigned int vec_sll (vector unsigned int,
10594 vector unsigned int);
10595 vector unsigned int vec_sll (vector unsigned int,
10596 vector unsigned short);
10597 vector unsigned int vec_sll (vector unsigned int,
10598 vector unsigned char);
10599 vector bool int vec_sll (vector bool int,
10600 vector unsigned int);
10601 vector bool int vec_sll (vector bool int,
10602 vector unsigned short);
10603 vector bool int vec_sll (vector bool int,
10604 vector unsigned char);
10605 vector signed short vec_sll (vector signed short,
10606 vector unsigned int);
10607 vector signed short vec_sll (vector signed short,
10608 vector unsigned short);
10609 vector signed short vec_sll (vector signed short,
10610 vector unsigned char);
10611 vector unsigned short vec_sll (vector unsigned short,
10612 vector unsigned int);
10613 vector unsigned short vec_sll (vector unsigned short,
10614 vector unsigned short);
10615 vector unsigned short vec_sll (vector unsigned short,
10616 vector unsigned char);
10617 vector bool short vec_sll (vector bool short, vector unsigned int);
10618 vector bool short vec_sll (vector bool short, vector unsigned short);
10619 vector bool short vec_sll (vector bool short, vector unsigned char);
10620 vector pixel vec_sll (vector pixel, vector unsigned int);
10621 vector pixel vec_sll (vector pixel, vector unsigned short);
10622 vector pixel vec_sll (vector pixel, vector unsigned char);
10623 vector signed char vec_sll (vector signed char, vector unsigned int);
10624 vector signed char vec_sll (vector signed char, vector unsigned short);
10625 vector signed char vec_sll (vector signed char, vector unsigned char);
10626 vector unsigned char vec_sll (vector unsigned char,
10627 vector unsigned int);
10628 vector unsigned char vec_sll (vector unsigned char,
10629 vector unsigned short);
10630 vector unsigned char vec_sll (vector unsigned char,
10631 vector unsigned char);
10632 vector bool char vec_sll (vector bool char, vector unsigned int);
10633 vector bool char vec_sll (vector bool char, vector unsigned short);
10634 vector bool char vec_sll (vector bool char, vector unsigned char);
10636 vector float vec_slo (vector float, vector signed char);
10637 vector float vec_slo (vector float, vector unsigned char);
10638 vector signed int vec_slo (vector signed int, vector signed char);
10639 vector signed int vec_slo (vector signed int, vector unsigned char);
10640 vector unsigned int vec_slo (vector unsigned int, vector signed char);
10641 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
10642 vector signed short vec_slo (vector signed short, vector signed char);
10643 vector signed short vec_slo (vector signed short, vector unsigned char);
10644 vector unsigned short vec_slo (vector unsigned short,
10645 vector signed char);
10646 vector unsigned short vec_slo (vector unsigned short,
10647 vector unsigned char);
10648 vector pixel vec_slo (vector pixel, vector signed char);
10649 vector pixel vec_slo (vector pixel, vector unsigned char);
10650 vector signed char vec_slo (vector signed char, vector signed char);
10651 vector signed char vec_slo (vector signed char, vector unsigned char);
10652 vector unsigned char vec_slo (vector unsigned char, vector signed char);
10653 vector unsigned char vec_slo (vector unsigned char,
10654 vector unsigned char);
10656 vector signed char vec_splat (vector signed char, const int);
10657 vector unsigned char vec_splat (vector unsigned char, const int);
10658 vector bool char vec_splat (vector bool char, const int);
10659 vector signed short vec_splat (vector signed short, const int);
10660 vector unsigned short vec_splat (vector unsigned short, const int);
10661 vector bool short vec_splat (vector bool short, const int);
10662 vector pixel vec_splat (vector pixel, const int);
10663 vector float vec_splat (vector float, const int);
10664 vector signed int vec_splat (vector signed int, const int);
10665 vector unsigned int vec_splat (vector unsigned int, const int);
10666 vector bool int vec_splat (vector bool int, const int);
10668 vector float vec_vspltw (vector float, const int);
10669 vector signed int vec_vspltw (vector signed int, const int);
10670 vector unsigned int vec_vspltw (vector unsigned int, const int);
10671 vector bool int vec_vspltw (vector bool int, const int);
10673 vector bool short vec_vsplth (vector bool short, const int);
10674 vector signed short vec_vsplth (vector signed short, const int);
10675 vector unsigned short vec_vsplth (vector unsigned short, const int);
10676 vector pixel vec_vsplth (vector pixel, const int);
10678 vector signed char vec_vspltb (vector signed char, const int);
10679 vector unsigned char vec_vspltb (vector unsigned char, const int);
10680 vector bool char vec_vspltb (vector bool char, const int);
10682 vector signed char vec_splat_s8 (const int);
10684 vector signed short vec_splat_s16 (const int);
10686 vector signed int vec_splat_s32 (const int);
10688 vector unsigned char vec_splat_u8 (const int);
10690 vector unsigned short vec_splat_u16 (const int);
10692 vector unsigned int vec_splat_u32 (const int);
10694 vector signed char vec_sr (vector signed char, vector unsigned char);
10695 vector unsigned char vec_sr (vector unsigned char,
10696 vector unsigned char);
10697 vector signed short vec_sr (vector signed short,
10698 vector unsigned short);
10699 vector unsigned short vec_sr (vector unsigned short,
10700 vector unsigned short);
10701 vector signed int vec_sr (vector signed int, vector unsigned int);
10702 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
10704 vector signed int vec_vsrw (vector signed int, vector unsigned int);
10705 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
10707 vector signed short vec_vsrh (vector signed short,
10708 vector unsigned short);
10709 vector unsigned short vec_vsrh (vector unsigned short,
10710 vector unsigned short);
10712 vector signed char vec_vsrb (vector signed char, vector unsigned char);
10713 vector unsigned char vec_vsrb (vector unsigned char,
10714 vector unsigned char);
10716 vector signed char vec_sra (vector signed char, vector unsigned char);
10717 vector unsigned char vec_sra (vector unsigned char,
10718 vector unsigned char);
10719 vector signed short vec_sra (vector signed short,
10720 vector unsigned short);
10721 vector unsigned short vec_sra (vector unsigned short,
10722 vector unsigned short);
10723 vector signed int vec_sra (vector signed int, vector unsigned int);
10724 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
10726 vector signed int vec_vsraw (vector signed int, vector unsigned int);
10727 vector unsigned int vec_vsraw (vector unsigned int,
10728 vector unsigned int);
10730 vector signed short vec_vsrah (vector signed short,
10731 vector unsigned short);
10732 vector unsigned short vec_vsrah (vector unsigned short,
10733 vector unsigned short);
10735 vector signed char vec_vsrab (vector signed char, vector unsigned char);
10736 vector unsigned char vec_vsrab (vector unsigned char,
10737 vector unsigned char);
10739 vector signed int vec_srl (vector signed int, vector unsigned int);
10740 vector signed int vec_srl (vector signed int, vector unsigned short);
10741 vector signed int vec_srl (vector signed int, vector unsigned char);
10742 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
10743 vector unsigned int vec_srl (vector unsigned int,
10744 vector unsigned short);
10745 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
10746 vector bool int vec_srl (vector bool int, vector unsigned int);
10747 vector bool int vec_srl (vector bool int, vector unsigned short);
10748 vector bool int vec_srl (vector bool int, vector unsigned char);
10749 vector signed short vec_srl (vector signed short, vector unsigned int);
10750 vector signed short vec_srl (vector signed short,
10751 vector unsigned short);
10752 vector signed short vec_srl (vector signed short, vector unsigned char);
10753 vector unsigned short vec_srl (vector unsigned short,
10754 vector unsigned int);
10755 vector unsigned short vec_srl (vector unsigned short,
10756 vector unsigned short);
10757 vector unsigned short vec_srl (vector unsigned short,
10758 vector unsigned char);
10759 vector bool short vec_srl (vector bool short, vector unsigned int);
10760 vector bool short vec_srl (vector bool short, vector unsigned short);
10761 vector bool short vec_srl (vector bool short, vector unsigned char);
10762 vector pixel vec_srl (vector pixel, vector unsigned int);
10763 vector pixel vec_srl (vector pixel, vector unsigned short);
10764 vector pixel vec_srl (vector pixel, vector unsigned char);
10765 vector signed char vec_srl (vector signed char, vector unsigned int);
10766 vector signed char vec_srl (vector signed char, vector unsigned short);
10767 vector signed char vec_srl (vector signed char, vector unsigned char);
10768 vector unsigned char vec_srl (vector unsigned char,
10769 vector unsigned int);
10770 vector unsigned char vec_srl (vector unsigned char,
10771 vector unsigned short);
10772 vector unsigned char vec_srl (vector unsigned char,
10773 vector unsigned char);
10774 vector bool char vec_srl (vector bool char, vector unsigned int);
10775 vector bool char vec_srl (vector bool char, vector unsigned short);
10776 vector bool char vec_srl (vector bool char, vector unsigned char);
10778 vector float vec_sro (vector float, vector signed char);
10779 vector float vec_sro (vector float, vector unsigned char);
10780 vector signed int vec_sro (vector signed int, vector signed char);
10781 vector signed int vec_sro (vector signed int, vector unsigned char);
10782 vector unsigned int vec_sro (vector unsigned int, vector signed char);
10783 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
10784 vector signed short vec_sro (vector signed short, vector signed char);
10785 vector signed short vec_sro (vector signed short, vector unsigned char);
10786 vector unsigned short vec_sro (vector unsigned short,
10787 vector signed char);
10788 vector unsigned short vec_sro (vector unsigned short,
10789 vector unsigned char);
10790 vector pixel vec_sro (vector pixel, vector signed char);
10791 vector pixel vec_sro (vector pixel, vector unsigned char);
10792 vector signed char vec_sro (vector signed char, vector signed char);
10793 vector signed char vec_sro (vector signed char, vector unsigned char);
10794 vector unsigned char vec_sro (vector unsigned char, vector signed char);
10795 vector unsigned char vec_sro (vector unsigned char,
10796 vector unsigned char);
10798 void vec_st (vector float, int, vector float *);
10799 void vec_st (vector float, int, float *);
10800 void vec_st (vector signed int, int, vector signed int *);
10801 void vec_st (vector signed int, int, int *);
10802 void vec_st (vector unsigned int, int, vector unsigned int *);
10803 void vec_st (vector unsigned int, int, unsigned int *);
10804 void vec_st (vector bool int, int, vector bool int *);
10805 void vec_st (vector bool int, int, unsigned int *);
10806 void vec_st (vector bool int, int, int *);
10807 void vec_st (vector signed short, int, vector signed short *);
10808 void vec_st (vector signed short, int, short *);
10809 void vec_st (vector unsigned short, int, vector unsigned short *);
10810 void vec_st (vector unsigned short, int, unsigned short *);
10811 void vec_st (vector bool short, int, vector bool short *);
10812 void vec_st (vector bool short, int, unsigned short *);
10813 void vec_st (vector pixel, int, vector pixel *);
10814 void vec_st (vector pixel, int, unsigned short *);
10815 void vec_st (vector pixel, int, short *);
10816 void vec_st (vector bool short, int, short *);
10817 void vec_st (vector signed char, int, vector signed char *);
10818 void vec_st (vector signed char, int, signed char *);
10819 void vec_st (vector unsigned char, int, vector unsigned char *);
10820 void vec_st (vector unsigned char, int, unsigned char *);
10821 void vec_st (vector bool char, int, vector bool char *);
10822 void vec_st (vector bool char, int, unsigned char *);
10823 void vec_st (vector bool char, int, signed char *);
10825 void vec_ste (vector signed char, int, signed char *);
10826 void vec_ste (vector unsigned char, int, unsigned char *);
10827 void vec_ste (vector bool char, int, signed char *);
10828 void vec_ste (vector bool char, int, unsigned char *);
10829 void vec_ste (vector signed short, int, short *);
10830 void vec_ste (vector unsigned short, int, unsigned short *);
10831 void vec_ste (vector bool short, int, short *);
10832 void vec_ste (vector bool short, int, unsigned short *);
10833 void vec_ste (vector pixel, int, short *);
10834 void vec_ste (vector pixel, int, unsigned short *);
10835 void vec_ste (vector float, int, float *);
10836 void vec_ste (vector signed int, int, int *);
10837 void vec_ste (vector unsigned int, int, unsigned int *);
10838 void vec_ste (vector bool int, int, int *);
10839 void vec_ste (vector bool int, int, unsigned int *);
10841 void vec_stvewx (vector float, int, float *);
10842 void vec_stvewx (vector signed int, int, int *);
10843 void vec_stvewx (vector unsigned int, int, unsigned int *);
10844 void vec_stvewx (vector bool int, int, int *);
10845 void vec_stvewx (vector bool int, int, unsigned int *);
10847 void vec_stvehx (vector signed short, int, short *);
10848 void vec_stvehx (vector unsigned short, int, unsigned short *);
10849 void vec_stvehx (vector bool short, int, short *);
10850 void vec_stvehx (vector bool short, int, unsigned short *);
10851 void vec_stvehx (vector pixel, int, short *);
10852 void vec_stvehx (vector pixel, int, unsigned short *);
10854 void vec_stvebx (vector signed char, int, signed char *);
10855 void vec_stvebx (vector unsigned char, int, unsigned char *);
10856 void vec_stvebx (vector bool char, int, signed char *);
10857 void vec_stvebx (vector bool char, int, unsigned char *);
10859 void vec_stl (vector float, int, vector float *);
10860 void vec_stl (vector float, int, float *);
10861 void vec_stl (vector signed int, int, vector signed int *);
10862 void vec_stl (vector signed int, int, int *);
10863 void vec_stl (vector unsigned int, int, vector unsigned int *);
10864 void vec_stl (vector unsigned int, int, unsigned int *);
10865 void vec_stl (vector bool int, int, vector bool int *);
10866 void vec_stl (vector bool int, int, unsigned int *);
10867 void vec_stl (vector bool int, int, int *);
10868 void vec_stl (vector signed short, int, vector signed short *);
10869 void vec_stl (vector signed short, int, short *);
10870 void vec_stl (vector unsigned short, int, vector unsigned short *);
10871 void vec_stl (vector unsigned short, int, unsigned short *);
10872 void vec_stl (vector bool short, int, vector bool short *);
10873 void vec_stl (vector bool short, int, unsigned short *);
10874 void vec_stl (vector bool short, int, short *);
10875 void vec_stl (vector pixel, int, vector pixel *);
10876 void vec_stl (vector pixel, int, unsigned short *);
10877 void vec_stl (vector pixel, int, short *);
10878 void vec_stl (vector signed char, int, vector signed char *);
10879 void vec_stl (vector signed char, int, signed char *);
10880 void vec_stl (vector unsigned char, int, vector unsigned char *);
10881 void vec_stl (vector unsigned char, int, unsigned char *);
10882 void vec_stl (vector bool char, int, vector bool char *);
10883 void vec_stl (vector bool char, int, unsigned char *);
10884 void vec_stl (vector bool char, int, signed char *);
10886 vector signed char vec_sub (vector bool char, vector signed char);
10887 vector signed char vec_sub (vector signed char, vector bool char);
10888 vector signed char vec_sub (vector signed char, vector signed char);
10889 vector unsigned char vec_sub (vector bool char, vector unsigned char);
10890 vector unsigned char vec_sub (vector unsigned char, vector bool char);
10891 vector unsigned char vec_sub (vector unsigned char,
10892 vector unsigned char);
10893 vector signed short vec_sub (vector bool short, vector signed short);
10894 vector signed short vec_sub (vector signed short, vector bool short);
10895 vector signed short vec_sub (vector signed short, vector signed short);
10896 vector unsigned short vec_sub (vector bool short,
10897 vector unsigned short);
10898 vector unsigned short vec_sub (vector unsigned short,
10899 vector bool short);
10900 vector unsigned short vec_sub (vector unsigned short,
10901 vector unsigned short);
10902 vector signed int vec_sub (vector bool int, vector signed int);
10903 vector signed int vec_sub (vector signed int, vector bool int);
10904 vector signed int vec_sub (vector signed int, vector signed int);
10905 vector unsigned int vec_sub (vector bool int, vector unsigned int);
10906 vector unsigned int vec_sub (vector unsigned int, vector bool int);
10907 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
10908 vector float vec_sub (vector float, vector float);
10910 vector float vec_vsubfp (vector float, vector float);
10912 vector signed int vec_vsubuwm (vector bool int, vector signed int);
10913 vector signed int vec_vsubuwm (vector signed int, vector bool int);
10914 vector signed int vec_vsubuwm (vector signed int, vector signed int);
10915 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
10916 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
10917 vector unsigned int vec_vsubuwm (vector unsigned int,
10918 vector unsigned int);
10920 vector signed short vec_vsubuhm (vector bool short,
10921 vector signed short);
10922 vector signed short vec_vsubuhm (vector signed short,
10923 vector bool short);
10924 vector signed short vec_vsubuhm (vector signed short,
10925 vector signed short);
10926 vector unsigned short vec_vsubuhm (vector bool short,
10927 vector unsigned short);
10928 vector unsigned short vec_vsubuhm (vector unsigned short,
10929 vector bool short);
10930 vector unsigned short vec_vsubuhm (vector unsigned short,
10931 vector unsigned short);
10933 vector signed char vec_vsububm (vector bool char, vector signed char);
10934 vector signed char vec_vsububm (vector signed char, vector bool char);
10935 vector signed char vec_vsububm (vector signed char, vector signed char);
10936 vector unsigned char vec_vsububm (vector bool char,
10937 vector unsigned char);
10938 vector unsigned char vec_vsububm (vector unsigned char,
10940 vector unsigned char vec_vsububm (vector unsigned char,
10941 vector unsigned char);
10943 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
10945 vector unsigned char vec_subs (vector bool char, vector unsigned char);
10946 vector unsigned char vec_subs (vector unsigned char, vector bool char);
10947 vector unsigned char vec_subs (vector unsigned char,
10948 vector unsigned char);
10949 vector signed char vec_subs (vector bool char, vector signed char);
10950 vector signed char vec_subs (vector signed char, vector bool char);
10951 vector signed char vec_subs (vector signed char, vector signed char);
10952 vector unsigned short vec_subs (vector bool short,
10953 vector unsigned short);
10954 vector unsigned short vec_subs (vector unsigned short,
10955 vector bool short);
10956 vector unsigned short vec_subs (vector unsigned short,
10957 vector unsigned short);
10958 vector signed short vec_subs (vector bool short, vector signed short);
10959 vector signed short vec_subs (vector signed short, vector bool short);
10960 vector signed short vec_subs (vector signed short, vector signed short);
10961 vector unsigned int vec_subs (vector bool int, vector unsigned int);
10962 vector unsigned int vec_subs (vector unsigned int, vector bool int);
10963 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
10964 vector signed int vec_subs (vector bool int, vector signed int);
10965 vector signed int vec_subs (vector signed int, vector bool int);
10966 vector signed int vec_subs (vector signed int, vector signed int);
10968 vector signed int vec_vsubsws (vector bool int, vector signed int);
10969 vector signed int vec_vsubsws (vector signed int, vector bool int);
10970 vector signed int vec_vsubsws (vector signed int, vector signed int);
10972 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
10973 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
10974 vector unsigned int vec_vsubuws (vector unsigned int,
10975 vector unsigned int);
10977 vector signed short vec_vsubshs (vector bool short,
10978 vector signed short);
10979 vector signed short vec_vsubshs (vector signed short,
10980 vector bool short);
10981 vector signed short vec_vsubshs (vector signed short,
10982 vector signed short);
10984 vector unsigned short vec_vsubuhs (vector bool short,
10985 vector unsigned short);
10986 vector unsigned short vec_vsubuhs (vector unsigned short,
10987 vector bool short);
10988 vector unsigned short vec_vsubuhs (vector unsigned short,
10989 vector unsigned short);
10991 vector signed char vec_vsubsbs (vector bool char, vector signed char);
10992 vector signed char vec_vsubsbs (vector signed char, vector bool char);
10993 vector signed char vec_vsubsbs (vector signed char, vector signed char);
10995 vector unsigned char vec_vsububs (vector bool char,
10996 vector unsigned char);
10997 vector unsigned char vec_vsububs (vector unsigned char,
10999 vector unsigned char vec_vsububs (vector unsigned char,
11000 vector unsigned char);
11002 vector unsigned int vec_sum4s (vector unsigned char,
11003 vector unsigned int);
11004 vector signed int vec_sum4s (vector signed char, vector signed int);
11005 vector signed int vec_sum4s (vector signed short, vector signed int);
11007 vector signed int vec_vsum4shs (vector signed short, vector signed int);
11009 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
11011 vector unsigned int vec_vsum4ubs (vector unsigned char,
11012 vector unsigned int);
11014 vector signed int vec_sum2s (vector signed int, vector signed int);
11016 vector signed int vec_sums (vector signed int, vector signed int);
11018 vector float vec_trunc (vector float);
11020 vector signed short vec_unpackh (vector signed char);
11021 vector bool short vec_unpackh (vector bool char);
11022 vector signed int vec_unpackh (vector signed short);
11023 vector bool int vec_unpackh (vector bool short);
11024 vector unsigned int vec_unpackh (vector pixel);
11026 vector bool int vec_vupkhsh (vector bool short);
11027 vector signed int vec_vupkhsh (vector signed short);
11029 vector unsigned int vec_vupkhpx (vector pixel);
11031 vector bool short vec_vupkhsb (vector bool char);
11032 vector signed short vec_vupkhsb (vector signed char);
11034 vector signed short vec_unpackl (vector signed char);
11035 vector bool short vec_unpackl (vector bool char);
11036 vector unsigned int vec_unpackl (vector pixel);
11037 vector signed int vec_unpackl (vector signed short);
11038 vector bool int vec_unpackl (vector bool short);
11040 vector unsigned int vec_vupklpx (vector pixel);
11042 vector bool int vec_vupklsh (vector bool short);
11043 vector signed int vec_vupklsh (vector signed short);
11045 vector bool short vec_vupklsb (vector bool char);
11046 vector signed short vec_vupklsb (vector signed char);
11048 vector float vec_xor (vector float, vector float);
11049 vector float vec_xor (vector float, vector bool int);
11050 vector float vec_xor (vector bool int, vector float);
11051 vector bool int vec_xor (vector bool int, vector bool int);
11052 vector signed int vec_xor (vector bool int, vector signed int);
11053 vector signed int vec_xor (vector signed int, vector bool int);
11054 vector signed int vec_xor (vector signed int, vector signed int);
11055 vector unsigned int vec_xor (vector bool int, vector unsigned int);
11056 vector unsigned int vec_xor (vector unsigned int, vector bool int);
11057 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
11058 vector bool short vec_xor (vector bool short, vector bool short);
11059 vector signed short vec_xor (vector bool short, vector signed short);
11060 vector signed short vec_xor (vector signed short, vector bool short);
11061 vector signed short vec_xor (vector signed short, vector signed short);
11062 vector unsigned short vec_xor (vector bool short,
11063 vector unsigned short);
11064 vector unsigned short vec_xor (vector unsigned short,
11065 vector bool short);
11066 vector unsigned short vec_xor (vector unsigned short,
11067 vector unsigned short);
11068 vector signed char vec_xor (vector bool char, vector signed char);
11069 vector bool char vec_xor (vector bool char, vector bool char);
11070 vector signed char vec_xor (vector signed char, vector bool char);
11071 vector signed char vec_xor (vector signed char, vector signed char);
11072 vector unsigned char vec_xor (vector bool char, vector unsigned char);
11073 vector unsigned char vec_xor (vector unsigned char, vector bool char);
11074 vector unsigned char vec_xor (vector unsigned char,
11075 vector unsigned char);
11077 int vec_all_eq (vector signed char, vector bool char);
11078 int vec_all_eq (vector signed char, vector signed char);
11079 int vec_all_eq (vector unsigned char, vector bool char);
11080 int vec_all_eq (vector unsigned char, vector unsigned char);
11081 int vec_all_eq (vector bool char, vector bool char);
11082 int vec_all_eq (vector bool char, vector unsigned char);
11083 int vec_all_eq (vector bool char, vector signed char);
11084 int vec_all_eq (vector signed short, vector bool short);
11085 int vec_all_eq (vector signed short, vector signed short);
11086 int vec_all_eq (vector unsigned short, vector bool short);
11087 int vec_all_eq (vector unsigned short, vector unsigned short);
11088 int vec_all_eq (vector bool short, vector bool short);
11089 int vec_all_eq (vector bool short, vector unsigned short);
11090 int vec_all_eq (vector bool short, vector signed short);
11091 int vec_all_eq (vector pixel, vector pixel);
11092 int vec_all_eq (vector signed int, vector bool int);
11093 int vec_all_eq (vector signed int, vector signed int);
11094 int vec_all_eq (vector unsigned int, vector bool int);
11095 int vec_all_eq (vector unsigned int, vector unsigned int);
11096 int vec_all_eq (vector bool int, vector bool int);
11097 int vec_all_eq (vector bool int, vector unsigned int);
11098 int vec_all_eq (vector bool int, vector signed int);
11099 int vec_all_eq (vector float, vector float);
11101 int vec_all_ge (vector bool char, vector unsigned char);
11102 int vec_all_ge (vector unsigned char, vector bool char);
11103 int vec_all_ge (vector unsigned char, vector unsigned char);
11104 int vec_all_ge (vector bool char, vector signed char);
11105 int vec_all_ge (vector signed char, vector bool char);
11106 int vec_all_ge (vector signed char, vector signed char);
11107 int vec_all_ge (vector bool short, vector unsigned short);
11108 int vec_all_ge (vector unsigned short, vector bool short);
11109 int vec_all_ge (vector unsigned short, vector unsigned short);
11110 int vec_all_ge (vector signed short, vector signed short);
11111 int vec_all_ge (vector bool short, vector signed short);
11112 int vec_all_ge (vector signed short, vector bool short);
11113 int vec_all_ge (vector bool int, vector unsigned int);
11114 int vec_all_ge (vector unsigned int, vector bool int);
11115 int vec_all_ge (vector unsigned int, vector unsigned int);
11116 int vec_all_ge (vector bool int, vector signed int);
11117 int vec_all_ge (vector signed int, vector bool int);
11118 int vec_all_ge (vector signed int, vector signed int);
11119 int vec_all_ge (vector float, vector float);
11121 int vec_all_gt (vector bool char, vector unsigned char);
11122 int vec_all_gt (vector unsigned char, vector bool char);
11123 int vec_all_gt (vector unsigned char, vector unsigned char);
11124 int vec_all_gt (vector bool char, vector signed char);
11125 int vec_all_gt (vector signed char, vector bool char);
11126 int vec_all_gt (vector signed char, vector signed char);
11127 int vec_all_gt (vector bool short, vector unsigned short);
11128 int vec_all_gt (vector unsigned short, vector bool short);
11129 int vec_all_gt (vector unsigned short, vector unsigned short);
11130 int vec_all_gt (vector bool short, vector signed short);
11131 int vec_all_gt (vector signed short, vector bool short);
11132 int vec_all_gt (vector signed short, vector signed short);
11133 int vec_all_gt (vector bool int, vector unsigned int);
11134 int vec_all_gt (vector unsigned int, vector bool int);
11135 int vec_all_gt (vector unsigned int, vector unsigned int);
11136 int vec_all_gt (vector bool int, vector signed int);
11137 int vec_all_gt (vector signed int, vector bool int);
11138 int vec_all_gt (vector signed int, vector signed int);
11139 int vec_all_gt (vector float, vector float);
11141 int vec_all_in (vector float, vector float);
11143 int vec_all_le (vector bool char, vector unsigned char);
11144 int vec_all_le (vector unsigned char, vector bool char);
11145 int vec_all_le (vector unsigned char, vector unsigned char);
11146 int vec_all_le (vector bool char, vector signed char);
11147 int vec_all_le (vector signed char, vector bool char);
11148 int vec_all_le (vector signed char, vector signed char);
11149 int vec_all_le (vector bool short, vector unsigned short);
11150 int vec_all_le (vector unsigned short, vector bool short);
11151 int vec_all_le (vector unsigned short, vector unsigned short);
11152 int vec_all_le (vector bool short, vector signed short);
11153 int vec_all_le (vector signed short, vector bool short);
11154 int vec_all_le (vector signed short, vector signed short);
11155 int vec_all_le (vector bool int, vector unsigned int);
11156 int vec_all_le (vector unsigned int, vector bool int);
11157 int vec_all_le (vector unsigned int, vector unsigned int);
11158 int vec_all_le (vector bool int, vector signed int);
11159 int vec_all_le (vector signed int, vector bool int);
11160 int vec_all_le (vector signed int, vector signed int);
11161 int vec_all_le (vector float, vector float);
11163 int vec_all_lt (vector bool char, vector unsigned char);
11164 int vec_all_lt (vector unsigned char, vector bool char);
11165 int vec_all_lt (vector unsigned char, vector unsigned char);
11166 int vec_all_lt (vector bool char, vector signed char);
11167 int vec_all_lt (vector signed char, vector bool char);
11168 int vec_all_lt (vector signed char, vector signed char);
11169 int vec_all_lt (vector bool short, vector unsigned short);
11170 int vec_all_lt (vector unsigned short, vector bool short);
11171 int vec_all_lt (vector unsigned short, vector unsigned short);
11172 int vec_all_lt (vector bool short, vector signed short);
11173 int vec_all_lt (vector signed short, vector bool short);
11174 int vec_all_lt (vector signed short, vector signed short);
11175 int vec_all_lt (vector bool int, vector unsigned int);
11176 int vec_all_lt (vector unsigned int, vector bool int);
11177 int vec_all_lt (vector unsigned int, vector unsigned int);
11178 int vec_all_lt (vector bool int, vector signed int);
11179 int vec_all_lt (vector signed int, vector bool int);
11180 int vec_all_lt (vector signed int, vector signed int);
11181 int vec_all_lt (vector float, vector float);
11183 int vec_all_nan (vector float);
11185 int vec_all_ne (vector signed char, vector bool char);
11186 int vec_all_ne (vector signed char, vector signed char);
11187 int vec_all_ne (vector unsigned char, vector bool char);
11188 int vec_all_ne (vector unsigned char, vector unsigned char);
11189 int vec_all_ne (vector bool char, vector bool char);
11190 int vec_all_ne (vector bool char, vector unsigned char);
11191 int vec_all_ne (vector bool char, vector signed char);
11192 int vec_all_ne (vector signed short, vector bool short);
11193 int vec_all_ne (vector signed short, vector signed short);
11194 int vec_all_ne (vector unsigned short, vector bool short);
11195 int vec_all_ne (vector unsigned short, vector unsigned short);
11196 int vec_all_ne (vector bool short, vector bool short);
11197 int vec_all_ne (vector bool short, vector unsigned short);
11198 int vec_all_ne (vector bool short, vector signed short);
11199 int vec_all_ne (vector pixel, vector pixel);
11200 int vec_all_ne (vector signed int, vector bool int);
11201 int vec_all_ne (vector signed int, vector signed int);
11202 int vec_all_ne (vector unsigned int, vector bool int);
11203 int vec_all_ne (vector unsigned int, vector unsigned int);
11204 int vec_all_ne (vector bool int, vector bool int);
11205 int vec_all_ne (vector bool int, vector unsigned int);
11206 int vec_all_ne (vector bool int, vector signed int);
11207 int vec_all_ne (vector float, vector float);
11209 int vec_all_nge (vector float, vector float);
11211 int vec_all_ngt (vector float, vector float);
11213 int vec_all_nle (vector float, vector float);
11215 int vec_all_nlt (vector float, vector float);
11217 int vec_all_numeric (vector float);
11219 int vec_any_eq (vector signed char, vector bool char);
11220 int vec_any_eq (vector signed char, vector signed char);
11221 int vec_any_eq (vector unsigned char, vector bool char);
11222 int vec_any_eq (vector unsigned char, vector unsigned char);
11223 int vec_any_eq (vector bool char, vector bool char);
11224 int vec_any_eq (vector bool char, vector unsigned char);
11225 int vec_any_eq (vector bool char, vector signed char);
11226 int vec_any_eq (vector signed short, vector bool short);
11227 int vec_any_eq (vector signed short, vector signed short);
11228 int vec_any_eq (vector unsigned short, vector bool short);
11229 int vec_any_eq (vector unsigned short, vector unsigned short);
11230 int vec_any_eq (vector bool short, vector bool short);
11231 int vec_any_eq (vector bool short, vector unsigned short);
11232 int vec_any_eq (vector bool short, vector signed short);
11233 int vec_any_eq (vector pixel, vector pixel);
11234 int vec_any_eq (vector signed int, vector bool int);
11235 int vec_any_eq (vector signed int, vector signed int);
11236 int vec_any_eq (vector unsigned int, vector bool int);
11237 int vec_any_eq (vector unsigned int, vector unsigned int);
11238 int vec_any_eq (vector bool int, vector bool int);
11239 int vec_any_eq (vector bool int, vector unsigned int);
11240 int vec_any_eq (vector bool int, vector signed int);
11241 int vec_any_eq (vector float, vector float);
11243 int vec_any_ge (vector signed char, vector bool char);
11244 int vec_any_ge (vector unsigned char, vector bool char);
11245 int vec_any_ge (vector unsigned char, vector unsigned char);
11246 int vec_any_ge (vector signed char, vector signed char);
11247 int vec_any_ge (vector bool char, vector unsigned char);
11248 int vec_any_ge (vector bool char, vector signed char);
11249 int vec_any_ge (vector unsigned short, vector bool short);
11250 int vec_any_ge (vector unsigned short, vector unsigned short);
11251 int vec_any_ge (vector signed short, vector signed short);
11252 int vec_any_ge (vector signed short, vector bool short);
11253 int vec_any_ge (vector bool short, vector unsigned short);
11254 int vec_any_ge (vector bool short, vector signed short);
11255 int vec_any_ge (vector signed int, vector bool int);
11256 int vec_any_ge (vector unsigned int, vector bool int);
11257 int vec_any_ge (vector unsigned int, vector unsigned int);
11258 int vec_any_ge (vector signed int, vector signed int);
11259 int vec_any_ge (vector bool int, vector unsigned int);
11260 int vec_any_ge (vector bool int, vector signed int);
11261 int vec_any_ge (vector float, vector float);
11263 int vec_any_gt (vector bool char, vector unsigned char);
11264 int vec_any_gt (vector unsigned char, vector bool char);
11265 int vec_any_gt (vector unsigned char, vector unsigned char);
11266 int vec_any_gt (vector bool char, vector signed char);
11267 int vec_any_gt (vector signed char, vector bool char);
11268 int vec_any_gt (vector signed char, vector signed char);
11269 int vec_any_gt (vector bool short, vector unsigned short);
11270 int vec_any_gt (vector unsigned short, vector bool short);
11271 int vec_any_gt (vector unsigned short, vector unsigned short);
11272 int vec_any_gt (vector bool short, vector signed short);
11273 int vec_any_gt (vector signed short, vector bool short);
11274 int vec_any_gt (vector signed short, vector signed short);
11275 int vec_any_gt (vector bool int, vector unsigned int);
11276 int vec_any_gt (vector unsigned int, vector bool int);
11277 int vec_any_gt (vector unsigned int, vector unsigned int);
11278 int vec_any_gt (vector bool int, vector signed int);
11279 int vec_any_gt (vector signed int, vector bool int);
11280 int vec_any_gt (vector signed int, vector signed int);
11281 int vec_any_gt (vector float, vector float);
11283 int vec_any_le (vector bool char, vector unsigned char);
11284 int vec_any_le (vector unsigned char, vector bool char);
11285 int vec_any_le (vector unsigned char, vector unsigned char);
11286 int vec_any_le (vector bool char, vector signed char);
11287 int vec_any_le (vector signed char, vector bool char);
11288 int vec_any_le (vector signed char, vector signed char);
11289 int vec_any_le (vector bool short, vector unsigned short);
11290 int vec_any_le (vector unsigned short, vector bool short);
11291 int vec_any_le (vector unsigned short, vector unsigned short);
11292 int vec_any_le (vector bool short, vector signed short);
11293 int vec_any_le (vector signed short, vector bool short);
11294 int vec_any_le (vector signed short, vector signed short);
11295 int vec_any_le (vector bool int, vector unsigned int);
11296 int vec_any_le (vector unsigned int, vector bool int);
11297 int vec_any_le (vector unsigned int, vector unsigned int);
11298 int vec_any_le (vector bool int, vector signed int);
11299 int vec_any_le (vector signed int, vector bool int);
11300 int vec_any_le (vector signed int, vector signed int);
11301 int vec_any_le (vector float, vector float);
11303 int vec_any_lt (vector bool char, vector unsigned char);
11304 int vec_any_lt (vector unsigned char, vector bool char);
11305 int vec_any_lt (vector unsigned char, vector unsigned char);
11306 int vec_any_lt (vector bool char, vector signed char);
11307 int vec_any_lt (vector signed char, vector bool char);
11308 int vec_any_lt (vector signed char, vector signed char);
11309 int vec_any_lt (vector bool short, vector unsigned short);
11310 int vec_any_lt (vector unsigned short, vector bool short);
11311 int vec_any_lt (vector unsigned short, vector unsigned short);
11312 int vec_any_lt (vector bool short, vector signed short);
11313 int vec_any_lt (vector signed short, vector bool short);
11314 int vec_any_lt (vector signed short, vector signed short);
11315 int vec_any_lt (vector bool int, vector unsigned int);
11316 int vec_any_lt (vector unsigned int, vector bool int);
11317 int vec_any_lt (vector unsigned int, vector unsigned int);
11318 int vec_any_lt (vector bool int, vector signed int);
11319 int vec_any_lt (vector signed int, vector bool int);
11320 int vec_any_lt (vector signed int, vector signed int);
11321 int vec_any_lt (vector float, vector float);
11323 int vec_any_nan (vector float);
11325 int vec_any_ne (vector signed char, vector bool char);
11326 int vec_any_ne (vector signed char, vector signed char);
11327 int vec_any_ne (vector unsigned char, vector bool char);
11328 int vec_any_ne (vector unsigned char, vector unsigned char);
11329 int vec_any_ne (vector bool char, vector bool char);
11330 int vec_any_ne (vector bool char, vector unsigned char);
11331 int vec_any_ne (vector bool char, vector signed char);
11332 int vec_any_ne (vector signed short, vector bool short);
11333 int vec_any_ne (vector signed short, vector signed short);
11334 int vec_any_ne (vector unsigned short, vector bool short);
11335 int vec_any_ne (vector unsigned short, vector unsigned short);
11336 int vec_any_ne (vector bool short, vector bool short);
11337 int vec_any_ne (vector bool short, vector unsigned short);
11338 int vec_any_ne (vector bool short, vector signed short);
11339 int vec_any_ne (vector pixel, vector pixel);
11340 int vec_any_ne (vector signed int, vector bool int);
11341 int vec_any_ne (vector signed int, vector signed int);
11342 int vec_any_ne (vector unsigned int, vector bool int);
11343 int vec_any_ne (vector unsigned int, vector unsigned int);
11344 int vec_any_ne (vector bool int, vector bool int);
11345 int vec_any_ne (vector bool int, vector unsigned int);
11346 int vec_any_ne (vector bool int, vector signed int);
11347 int vec_any_ne (vector float, vector float);
11349 int vec_any_nge (vector float, vector float);
11351 int vec_any_ngt (vector float, vector float);
11353 int vec_any_nle (vector float, vector float);
11355 int vec_any_nlt (vector float, vector float);
11357 int vec_any_numeric (vector float);
11359 int vec_any_out (vector float, vector float);
11362 @node SPARC VIS Built-in Functions
11363 @subsection SPARC VIS Built-in Functions
11365 GCC supports SIMD operations on the SPARC using both the generic vector
11366 extensions (@pxref{Vector Extensions}) as well as built-in functions for
11367 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
11368 switch, the VIS extension is exposed as the following built-in functions:
11371 typedef int v2si __attribute__ ((vector_size (8)));
11372 typedef short v4hi __attribute__ ((vector_size (8)));
11373 typedef short v2hi __attribute__ ((vector_size (4)));
11374 typedef char v8qi __attribute__ ((vector_size (8)));
11375 typedef char v4qi __attribute__ ((vector_size (4)));
11377 void * __builtin_vis_alignaddr (void *, long);
11378 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
11379 v2si __builtin_vis_faligndatav2si (v2si, v2si);
11380 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
11381 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
11383 v4hi __builtin_vis_fexpand (v4qi);
11385 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
11386 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
11387 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
11388 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
11389 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
11390 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
11391 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
11393 v4qi __builtin_vis_fpack16 (v4hi);
11394 v8qi __builtin_vis_fpack32 (v2si, v2si);
11395 v2hi __builtin_vis_fpackfix (v2si);
11396 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
11398 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
11401 @node SPU Built-in Functions
11402 @subsection SPU Built-in Functions
11404 GCC provides extensions for the SPU processor as described in the
11405 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
11406 found at @uref{http://cell.scei.co.jp/} or
11407 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
11408 implementation differs in several ways.
11413 The optional extension of specifying vector constants in parentheses is
11417 A vector initializer requires no cast if the vector constant is of the
11418 same type as the variable it is initializing.
11421 If @code{signed} or @code{unsigned} is omitted, the signedness of the
11422 vector type is the default signedness of the base type. The default
11423 varies depending on the operating system, so a portable program should
11424 always specify the signedness.
11427 By default, the keyword @code{__vector} is added. The macro
11428 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
11432 GCC allows using a @code{typedef} name as the type specifier for a
11436 For C, overloaded functions are implemented with macros so the following
11440 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
11443 Since @code{spu_add} is a macro, the vector constant in the example
11444 is treated as four separate arguments. Wrap the entire argument in
11445 parentheses for this to work.
11448 The extended version of @code{__builtin_expect} is not supported.
11452 @emph{Note:} Only the interface described in the aforementioned
11453 specification is supported. Internally, GCC uses built-in functions to
11454 implement the required functionality, but these are not supported and
11455 are subject to change without notice.
11457 @node Target Format Checks
11458 @section Format Checks Specific to Particular Target Machines
11460 For some target machines, GCC supports additional options to the
11462 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
11465 * Solaris Format Checks::
11468 @node Solaris Format Checks
11469 @subsection Solaris Format Checks
11471 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
11472 check. @code{cmn_err} accepts a subset of the standard @code{printf}
11473 conversions, and the two-argument @code{%b} conversion for displaying
11474 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
11477 @section Pragmas Accepted by GCC
11481 GCC supports several types of pragmas, primarily in order to compile
11482 code originally written for other compilers. Note that in general
11483 we do not recommend the use of pragmas; @xref{Function Attributes},
11484 for further explanation.
11489 * RS/6000 and PowerPC Pragmas::
11491 * Solaris Pragmas::
11492 * Symbol-Renaming Pragmas::
11493 * Structure-Packing Pragmas::
11495 * Diagnostic Pragmas::
11496 * Visibility Pragmas::
11497 * Push/Pop Macro Pragmas::
11498 * Function Specific Option Pragmas::
11502 @subsection ARM Pragmas
11504 The ARM target defines pragmas for controlling the default addition of
11505 @code{long_call} and @code{short_call} attributes to functions.
11506 @xref{Function Attributes}, for information about the effects of these
11511 @cindex pragma, long_calls
11512 Set all subsequent functions to have the @code{long_call} attribute.
11514 @item no_long_calls
11515 @cindex pragma, no_long_calls
11516 Set all subsequent functions to have the @code{short_call} attribute.
11518 @item long_calls_off
11519 @cindex pragma, long_calls_off
11520 Do not affect the @code{long_call} or @code{short_call} attributes of
11521 subsequent functions.
11525 @subsection M32C Pragmas
11528 @item memregs @var{number}
11529 @cindex pragma, memregs
11530 Overrides the command line option @code{-memregs=} for the current
11531 file. Use with care! This pragma must be before any function in the
11532 file, and mixing different memregs values in different objects may
11533 make them incompatible. This pragma is useful when a
11534 performance-critical function uses a memreg for temporary values,
11535 as it may allow you to reduce the number of memregs used.
11539 @node RS/6000 and PowerPC Pragmas
11540 @subsection RS/6000 and PowerPC Pragmas
11542 The RS/6000 and PowerPC targets define one pragma for controlling
11543 whether or not the @code{longcall} attribute is added to function
11544 declarations by default. This pragma overrides the @option{-mlongcall}
11545 option, but not the @code{longcall} and @code{shortcall} attributes.
11546 @xref{RS/6000 and PowerPC Options}, for more information about when long
11547 calls are and are not necessary.
11551 @cindex pragma, longcall
11552 Apply the @code{longcall} attribute to all subsequent function
11556 Do not apply the @code{longcall} attribute to subsequent function
11560 @c Describe h8300 pragmas here.
11561 @c Describe sh pragmas here.
11562 @c Describe v850 pragmas here.
11564 @node Darwin Pragmas
11565 @subsection Darwin Pragmas
11567 The following pragmas are available for all architectures running the
11568 Darwin operating system. These are useful for compatibility with other
11572 @item mark @var{tokens}@dots{}
11573 @cindex pragma, mark
11574 This pragma is accepted, but has no effect.
11576 @item options align=@var{alignment}
11577 @cindex pragma, options align
11578 This pragma sets the alignment of fields in structures. The values of
11579 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
11580 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
11581 properly; to restore the previous setting, use @code{reset} for the
11584 @item segment @var{tokens}@dots{}
11585 @cindex pragma, segment
11586 This pragma is accepted, but has no effect.
11588 @item unused (@var{var} [, @var{var}]@dots{})
11589 @cindex pragma, unused
11590 This pragma declares variables to be possibly unused. GCC will not
11591 produce warnings for the listed variables. The effect is similar to
11592 that of the @code{unused} attribute, except that this pragma may appear
11593 anywhere within the variables' scopes.
11596 @node Solaris Pragmas
11597 @subsection Solaris Pragmas
11599 The Solaris target supports @code{#pragma redefine_extname}
11600 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
11601 @code{#pragma} directives for compatibility with the system compiler.
11604 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
11605 @cindex pragma, align
11607 Increase the minimum alignment of each @var{variable} to @var{alignment}.
11608 This is the same as GCC's @code{aligned} attribute @pxref{Variable
11609 Attributes}). Macro expansion occurs on the arguments to this pragma
11610 when compiling C and Objective-C@. It does not currently occur when
11611 compiling C++, but this is a bug which may be fixed in a future
11614 @item fini (@var{function} [, @var{function}]...)
11615 @cindex pragma, fini
11617 This pragma causes each listed @var{function} to be called after
11618 main, or during shared module unloading, by adding a call to the
11619 @code{.fini} section.
11621 @item init (@var{function} [, @var{function}]...)
11622 @cindex pragma, init
11624 This pragma causes each listed @var{function} to be called during
11625 initialization (before @code{main}) or during shared module loading, by
11626 adding a call to the @code{.init} section.
11630 @node Symbol-Renaming Pragmas
11631 @subsection Symbol-Renaming Pragmas
11633 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
11634 supports two @code{#pragma} directives which change the name used in
11635 assembly for a given declaration. These pragmas are only available on
11636 platforms whose system headers need them. To get this effect on all
11637 platforms supported by GCC, use the asm labels extension (@pxref{Asm
11641 @item redefine_extname @var{oldname} @var{newname}
11642 @cindex pragma, redefine_extname
11644 This pragma gives the C function @var{oldname} the assembly symbol
11645 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
11646 will be defined if this pragma is available (currently only on
11649 @item extern_prefix @var{string}
11650 @cindex pragma, extern_prefix
11652 This pragma causes all subsequent external function and variable
11653 declarations to have @var{string} prepended to their assembly symbols.
11654 This effect may be terminated with another @code{extern_prefix} pragma
11655 whose argument is an empty string. The preprocessor macro
11656 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
11657 available (currently only on Tru64 UNIX)@.
11660 These pragmas and the asm labels extension interact in a complicated
11661 manner. Here are some corner cases you may want to be aware of.
11664 @item Both pragmas silently apply only to declarations with external
11665 linkage. Asm labels do not have this restriction.
11667 @item In C++, both pragmas silently apply only to declarations with
11668 ``C'' linkage. Again, asm labels do not have this restriction.
11670 @item If any of the three ways of changing the assembly name of a
11671 declaration is applied to a declaration whose assembly name has
11672 already been determined (either by a previous use of one of these
11673 features, or because the compiler needed the assembly name in order to
11674 generate code), and the new name is different, a warning issues and
11675 the name does not change.
11677 @item The @var{oldname} used by @code{#pragma redefine_extname} is
11678 always the C-language name.
11680 @item If @code{#pragma extern_prefix} is in effect, and a declaration
11681 occurs with an asm label attached, the prefix is silently ignored for
11684 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
11685 apply to the same declaration, whichever triggered first wins, and a
11686 warning issues if they contradict each other. (We would like to have
11687 @code{#pragma redefine_extname} always win, for consistency with asm
11688 labels, but if @code{#pragma extern_prefix} triggers first we have no
11689 way of knowing that that happened.)
11692 @node Structure-Packing Pragmas
11693 @subsection Structure-Packing Pragmas
11695 For compatibility with Microsoft Windows compilers, GCC supports a
11696 set of @code{#pragma} directives which change the maximum alignment of
11697 members of structures (other than zero-width bitfields), unions, and
11698 classes subsequently defined. The @var{n} value below always is required
11699 to be a small power of two and specifies the new alignment in bytes.
11702 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
11703 @item @code{#pragma pack()} sets the alignment to the one that was in
11704 effect when compilation started (see also command line option
11705 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
11706 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
11707 setting on an internal stack and then optionally sets the new alignment.
11708 @item @code{#pragma pack(pop)} restores the alignment setting to the one
11709 saved at the top of the internal stack (and removes that stack entry).
11710 Note that @code{#pragma pack([@var{n}])} does not influence this internal
11711 stack; thus it is possible to have @code{#pragma pack(push)} followed by
11712 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
11713 @code{#pragma pack(pop)}.
11716 Some targets, e.g.@: i386 and powerpc, support the @code{ms_struct}
11717 @code{#pragma} which lays out a structure as the documented
11718 @code{__attribute__ ((ms_struct))}.
11720 @item @code{#pragma ms_struct on} turns on the layout for structures
11722 @item @code{#pragma ms_struct off} turns off the layout for structures
11724 @item @code{#pragma ms_struct reset} goes back to the default layout.
11728 @subsection Weak Pragmas
11730 For compatibility with SVR4, GCC supports a set of @code{#pragma}
11731 directives for declaring symbols to be weak, and defining weak
11735 @item #pragma weak @var{symbol}
11736 @cindex pragma, weak
11737 This pragma declares @var{symbol} to be weak, as if the declaration
11738 had the attribute of the same name. The pragma may appear before
11739 or after the declaration of @var{symbol}, but must appear before
11740 either its first use or its definition. It is not an error for
11741 @var{symbol} to never be defined at all.
11743 @item #pragma weak @var{symbol1} = @var{symbol2}
11744 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
11745 It is an error if @var{symbol2} is not defined in the current
11749 @node Diagnostic Pragmas
11750 @subsection Diagnostic Pragmas
11752 GCC allows the user to selectively enable or disable certain types of
11753 diagnostics, and change the kind of the diagnostic. For example, a
11754 project's policy might require that all sources compile with
11755 @option{-Werror} but certain files might have exceptions allowing
11756 specific types of warnings. Or, a project might selectively enable
11757 diagnostics and treat them as errors depending on which preprocessor
11758 macros are defined.
11761 @item #pragma GCC diagnostic @var{kind} @var{option}
11762 @cindex pragma, diagnostic
11764 Modifies the disposition of a diagnostic. Note that not all
11765 diagnostics are modifiable; at the moment only warnings (normally
11766 controlled by @samp{-W@dots{}}) can be controlled, and not all of them.
11767 Use @option{-fdiagnostics-show-option} to determine which diagnostics
11768 are controllable and which option controls them.
11770 @var{kind} is @samp{error} to treat this diagnostic as an error,
11771 @samp{warning} to treat it like a warning (even if @option{-Werror} is
11772 in effect), or @samp{ignored} if the diagnostic is to be ignored.
11773 @var{option} is a double quoted string which matches the command line
11777 #pragma GCC diagnostic warning "-Wformat"
11778 #pragma GCC diagnostic error "-Wformat"
11779 #pragma GCC diagnostic ignored "-Wformat"
11782 Note that these pragmas override any command line options. Also,
11783 while it is syntactically valid to put these pragmas anywhere in your
11784 sources, the only supported location for them is before any data or
11785 functions are defined. Doing otherwise may result in unpredictable
11786 results depending on how the optimizer manages your sources. If the
11787 same option is listed multiple times, the last one specified is the
11788 one that is in effect. This pragma is not intended to be a general
11789 purpose replacement for command line options, but for implementing
11790 strict control over project policies.
11794 GCC also offers a simple mechanism for printing messages during
11798 @item #pragma message @var{string}
11799 @cindex pragma, diagnostic
11801 Prints @var{string} as a compiler message on compilation. The message
11802 is informational only, and is neither a compilation warning nor an error.
11805 #pragma message "Compiling " __FILE__ "..."
11808 @var{string} may be parenthesized, and is printed with location
11809 information. For example,
11812 #define DO_PRAGMA(x) _Pragma (#x)
11813 #define TODO(x) DO_PRAGMA(message ("TODO - " #x))
11815 TODO(Remember to fix this)
11818 prints @samp{/tmp/file.c:4: note: #pragma message:
11819 TODO - Remember to fix this}.
11823 @node Visibility Pragmas
11824 @subsection Visibility Pragmas
11827 @item #pragma GCC visibility push(@var{visibility})
11828 @itemx #pragma GCC visibility pop
11829 @cindex pragma, visibility
11831 This pragma allows the user to set the visibility for multiple
11832 declarations without having to give each a visibility attribute
11833 @xref{Function Attributes}, for more information about visibility and
11834 the attribute syntax.
11836 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
11837 declarations. Class members and template specializations are not
11838 affected; if you want to override the visibility for a particular
11839 member or instantiation, you must use an attribute.
11844 @node Push/Pop Macro Pragmas
11845 @subsection Push/Pop Macro Pragmas
11847 For compatibility with Microsoft Windows compilers, GCC supports
11848 @samp{#pragma push_macro(@var{"macro_name"})}
11849 and @samp{#pragma pop_macro(@var{"macro_name"})}.
11852 @item #pragma push_macro(@var{"macro_name"})
11853 @cindex pragma, push_macro
11854 This pragma saves the value of the macro named as @var{macro_name} to
11855 the top of the stack for this macro.
11857 @item #pragma pop_macro(@var{"macro_name"})
11858 @cindex pragma, pop_macro
11859 This pragma sets the value of the macro named as @var{macro_name} to
11860 the value on top of the stack for this macro. If the stack for
11861 @var{macro_name} is empty, the value of the macro remains unchanged.
11868 #pragma push_macro("X")
11871 #pragma pop_macro("X")
11875 In this example, the definition of X as 1 is saved by @code{#pragma
11876 push_macro} and restored by @code{#pragma pop_macro}.
11878 @node Function Specific Option Pragmas
11879 @subsection Function Specific Option Pragmas
11882 @item #pragma GCC target (@var{"string"}...)
11883 @cindex pragma GCC target
11885 This pragma allows you to set target specific options for functions
11886 defined later in the source file. One or more strings can be
11887 specified. Each function that is defined after this point will be as
11888 if @code{attribute((target("STRING")))} was specified for that
11889 function. The parenthesis around the options is optional.
11890 @xref{Function Attributes}, for more information about the
11891 @code{target} attribute and the attribute syntax.
11893 The @samp{#pragma GCC target} pragma is not implemented in GCC
11894 versions earlier than 4.4, and is currently only implemented for the
11895 386 and x86_64 backends.
11899 @item #pragma GCC optimize (@var{"string"}...)
11900 @cindex pragma GCC optimize
11902 This pragma allows you to set global optimization options for functions
11903 defined later in the source file. One or more strings can be
11904 specified. Each function that is defined after this point will be as
11905 if @code{attribute((optimize("STRING")))} was specified for that
11906 function. The parenthesis around the options is optional.
11907 @xref{Function Attributes}, for more information about the
11908 @code{optimize} attribute and the attribute syntax.
11910 The @samp{#pragma GCC optimize} pragma is not implemented in GCC
11911 versions earlier than 4.4.
11915 @item #pragma GCC push_options
11916 @itemx #pragma GCC pop_options
11917 @cindex pragma GCC push_options
11918 @cindex pragma GCC pop_options
11920 These pragmas maintain a stack of the current target and optimization
11921 options. It is intended for include files where you temporarily want
11922 to switch to using a different @samp{#pragma GCC target} or
11923 @samp{#pragma GCC optimize} and then to pop back to the previous
11926 The @samp{#pragma GCC push_options} and @samp{#pragma GCC pop_options}
11927 pragmas are not implemented in GCC versions earlier than 4.4.
11931 @item #pragma GCC reset_options
11932 @cindex pragma GCC reset_options
11934 This pragma clears the current @code{#pragma GCC target} and
11935 @code{#pragma GCC optimize} to use the default switches as specified
11936 on the command line.
11938 The @samp{#pragma GCC reset_options} pragma is not implemented in GCC
11939 versions earlier than 4.4.
11942 @node Unnamed Fields
11943 @section Unnamed struct/union fields within structs/unions
11947 For compatibility with other compilers, GCC allows you to define
11948 a structure or union that contains, as fields, structures and unions
11949 without names. For example:
11962 In this example, the user would be able to access members of the unnamed
11963 union with code like @samp{foo.b}. Note that only unnamed structs and
11964 unions are allowed, you may not have, for example, an unnamed
11967 You must never create such structures that cause ambiguous field definitions.
11968 For example, this structure:
11979 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
11980 Such constructs are not supported and must be avoided. In the future,
11981 such constructs may be detected and treated as compilation errors.
11983 @opindex fms-extensions
11984 Unless @option{-fms-extensions} is used, the unnamed field must be a
11985 structure or union definition without a tag (for example, @samp{struct
11986 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
11987 also be a definition with a tag such as @samp{struct foo @{ int a;
11988 @};}, a reference to a previously defined structure or union such as
11989 @samp{struct foo;}, or a reference to a @code{typedef} name for a
11990 previously defined structure or union type.
11993 @section Thread-Local Storage
11994 @cindex Thread-Local Storage
11995 @cindex @acronym{TLS}
11998 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
11999 are allocated such that there is one instance of the variable per extant
12000 thread. The run-time model GCC uses to implement this originates
12001 in the IA-64 processor-specific ABI, but has since been migrated
12002 to other processors as well. It requires significant support from
12003 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
12004 system libraries (@file{libc.so} and @file{libpthread.so}), so it
12005 is not available everywhere.
12007 At the user level, the extension is visible with a new storage
12008 class keyword: @code{__thread}. For example:
12012 extern __thread struct state s;
12013 static __thread char *p;
12016 The @code{__thread} specifier may be used alone, with the @code{extern}
12017 or @code{static} specifiers, but with no other storage class specifier.
12018 When used with @code{extern} or @code{static}, @code{__thread} must appear
12019 immediately after the other storage class specifier.
12021 The @code{__thread} specifier may be applied to any global, file-scoped
12022 static, function-scoped static, or static data member of a class. It may
12023 not be applied to block-scoped automatic or non-static data member.
12025 When the address-of operator is applied to a thread-local variable, it is
12026 evaluated at run-time and returns the address of the current thread's
12027 instance of that variable. An address so obtained may be used by any
12028 thread. When a thread terminates, any pointers to thread-local variables
12029 in that thread become invalid.
12031 No static initialization may refer to the address of a thread-local variable.
12033 In C++, if an initializer is present for a thread-local variable, it must
12034 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
12037 See @uref{http://people.redhat.com/drepper/tls.pdf,
12038 ELF Handling For Thread-Local Storage} for a detailed explanation of
12039 the four thread-local storage addressing models, and how the run-time
12040 is expected to function.
12043 * C99 Thread-Local Edits::
12044 * C++98 Thread-Local Edits::
12047 @node C99 Thread-Local Edits
12048 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
12050 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
12051 that document the exact semantics of the language extension.
12055 @cite{5.1.2 Execution environments}
12057 Add new text after paragraph 1
12060 Within either execution environment, a @dfn{thread} is a flow of
12061 control within a program. It is implementation defined whether
12062 or not there may be more than one thread associated with a program.
12063 It is implementation defined how threads beyond the first are
12064 created, the name and type of the function called at thread
12065 startup, and how threads may be terminated. However, objects
12066 with thread storage duration shall be initialized before thread
12071 @cite{6.2.4 Storage durations of objects}
12073 Add new text before paragraph 3
12076 An object whose identifier is declared with the storage-class
12077 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
12078 Its lifetime is the entire execution of the thread, and its
12079 stored value is initialized only once, prior to thread startup.
12083 @cite{6.4.1 Keywords}
12085 Add @code{__thread}.
12088 @cite{6.7.1 Storage-class specifiers}
12090 Add @code{__thread} to the list of storage class specifiers in
12093 Change paragraph 2 to
12096 With the exception of @code{__thread}, at most one storage-class
12097 specifier may be given [@dots{}]. The @code{__thread} specifier may
12098 be used alone, or immediately following @code{extern} or
12102 Add new text after paragraph 6
12105 The declaration of an identifier for a variable that has
12106 block scope that specifies @code{__thread} shall also
12107 specify either @code{extern} or @code{static}.
12109 The @code{__thread} specifier shall be used only with
12114 @node C++98 Thread-Local Edits
12115 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
12117 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
12118 that document the exact semantics of the language extension.
12122 @b{[intro.execution]}
12124 New text after paragraph 4
12127 A @dfn{thread} is a flow of control within the abstract machine.
12128 It is implementation defined whether or not there may be more than
12132 New text after paragraph 7
12135 It is unspecified whether additional action must be taken to
12136 ensure when and whether side effects are visible to other threads.
12142 Add @code{__thread}.
12145 @b{[basic.start.main]}
12147 Add after paragraph 5
12150 The thread that begins execution at the @code{main} function is called
12151 the @dfn{main thread}. It is implementation defined how functions
12152 beginning threads other than the main thread are designated or typed.
12153 A function so designated, as well as the @code{main} function, is called
12154 a @dfn{thread startup function}. It is implementation defined what
12155 happens if a thread startup function returns. It is implementation
12156 defined what happens to other threads when any thread calls @code{exit}.
12160 @b{[basic.start.init]}
12162 Add after paragraph 4
12165 The storage for an object of thread storage duration shall be
12166 statically initialized before the first statement of the thread startup
12167 function. An object of thread storage duration shall not require
12168 dynamic initialization.
12172 @b{[basic.start.term]}
12174 Add after paragraph 3
12177 The type of an object with thread storage duration shall not have a
12178 non-trivial destructor, nor shall it be an array type whose elements
12179 (directly or indirectly) have non-trivial destructors.
12185 Add ``thread storage duration'' to the list in paragraph 1.
12190 Thread, static, and automatic storage durations are associated with
12191 objects introduced by declarations [@dots{}].
12194 Add @code{__thread} to the list of specifiers in paragraph 3.
12197 @b{[basic.stc.thread]}
12199 New section before @b{[basic.stc.static]}
12202 The keyword @code{__thread} applied to a non-local object gives the
12203 object thread storage duration.
12205 A local variable or class data member declared both @code{static}
12206 and @code{__thread} gives the variable or member thread storage
12211 @b{[basic.stc.static]}
12216 All objects which have neither thread storage duration, dynamic
12217 storage duration nor are local [@dots{}].
12223 Add @code{__thread} to the list in paragraph 1.
12228 With the exception of @code{__thread}, at most one
12229 @var{storage-class-specifier} shall appear in a given
12230 @var{decl-specifier-seq}. The @code{__thread} specifier may
12231 be used alone, or immediately following the @code{extern} or
12232 @code{static} specifiers. [@dots{}]
12235 Add after paragraph 5
12238 The @code{__thread} specifier can be applied only to the names of objects
12239 and to anonymous unions.
12245 Add after paragraph 6
12248 Non-@code{static} members shall not be @code{__thread}.
12252 @node Binary constants
12253 @section Binary constants using the @samp{0b} prefix
12254 @cindex Binary constants using the @samp{0b} prefix
12256 Integer constants can be written as binary constants, consisting of a
12257 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
12258 @samp{0B}. This is particularly useful in environments that operate a
12259 lot on the bit-level (like microcontrollers).
12261 The following statements are identical:
12270 The type of these constants follows the same rules as for octal or
12271 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
12274 @node C++ Extensions
12275 @chapter Extensions to the C++ Language
12276 @cindex extensions, C++ language
12277 @cindex C++ language extensions
12279 The GNU compiler provides these extensions to the C++ language (and you
12280 can also use most of the C language extensions in your C++ programs). If you
12281 want to write code that checks whether these features are available, you can
12282 test for the GNU compiler the same way as for C programs: check for a
12283 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
12284 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
12285 Predefined Macros,cpp,The GNU C Preprocessor}).
12288 * Volatiles:: What constitutes an access to a volatile object.
12289 * Restricted Pointers:: C99 restricted pointers and references.
12290 * Vague Linkage:: Where G++ puts inlines, vtables and such.
12291 * C++ Interface:: You can use a single C++ header file for both
12292 declarations and definitions.
12293 * Template Instantiation:: Methods for ensuring that exactly one copy of
12294 each needed template instantiation is emitted.
12295 * Bound member functions:: You can extract a function pointer to the
12296 method denoted by a @samp{->*} or @samp{.*} expression.
12297 * C++ Attributes:: Variable, function, and type attributes for C++ only.
12298 * Namespace Association:: Strong using-directives for namespace association.
12299 * Type Traits:: Compiler support for type traits
12300 * Java Exceptions:: Tweaking exception handling to work with Java.
12301 * Deprecated Features:: Things will disappear from g++.
12302 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
12306 @section When is a Volatile Object Accessed?
12307 @cindex accessing volatiles
12308 @cindex volatile read
12309 @cindex volatile write
12310 @cindex volatile access
12312 Both the C and C++ standard have the concept of volatile objects. These
12313 are normally accessed by pointers and used for accessing hardware. The
12314 standards encourage compilers to refrain from optimizations concerning
12315 accesses to volatile objects. The C standard leaves it implementation
12316 defined as to what constitutes a volatile access. The C++ standard omits
12317 to specify this, except to say that C++ should behave in a similar manner
12318 to C with respect to volatiles, where possible. The minimum either
12319 standard specifies is that at a sequence point all previous accesses to
12320 volatile objects have stabilized and no subsequent accesses have
12321 occurred. Thus an implementation is free to reorder and combine
12322 volatile accesses which occur between sequence points, but cannot do so
12323 for accesses across a sequence point. The use of volatiles does not
12324 allow you to violate the restriction on updating objects multiple times
12325 within a sequence point.
12327 @xref{Qualifiers implementation, , Volatile qualifier and the C compiler}.
12329 The behavior differs slightly between C and C++ in the non-obvious cases:
12332 volatile int *src = @var{somevalue};
12336 With C, such expressions are rvalues, and GCC interprets this either as a
12337 read of the volatile object being pointed to or only as request to evaluate
12338 the side-effects. The C++ standard specifies that such expressions do not
12339 undergo lvalue to rvalue conversion, and that the type of the dereferenced
12340 object may be incomplete. The C++ standard does not specify explicitly
12341 that it is this lvalue to rvalue conversion which may be responsible for
12342 causing an access. However, there is reason to believe that it is,
12343 because otherwise certain simple expressions become undefined. However,
12344 because it would surprise most programmers, G++ treats dereferencing a
12345 pointer to volatile object of complete type when the value is unused as
12346 GCC would do for an equivalent type in C@. When the object has incomplete
12347 type, G++ issues a warning; if you wish to force an error, you must
12348 force a conversion to rvalue with, for instance, a static cast.
12350 When using a reference to volatile, G++ does not treat equivalent
12351 expressions as accesses to volatiles, but instead issues a warning that
12352 no volatile is accessed. The rationale for this is that otherwise it
12353 becomes difficult to determine where volatile access occur, and not
12354 possible to ignore the return value from functions returning volatile
12355 references. Again, if you wish to force a read, cast the reference to
12358 @node Restricted Pointers
12359 @section Restricting Pointer Aliasing
12360 @cindex restricted pointers
12361 @cindex restricted references
12362 @cindex restricted this pointer
12364 As with the C front end, G++ understands the C99 feature of restricted pointers,
12365 specified with the @code{__restrict__}, or @code{__restrict} type
12366 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
12367 language flag, @code{restrict} is not a keyword in C++.
12369 In addition to allowing restricted pointers, you can specify restricted
12370 references, which indicate that the reference is not aliased in the local
12374 void fn (int *__restrict__ rptr, int &__restrict__ rref)
12381 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
12382 @var{rref} refers to a (different) unaliased integer.
12384 You may also specify whether a member function's @var{this} pointer is
12385 unaliased by using @code{__restrict__} as a member function qualifier.
12388 void T::fn () __restrict__
12395 Within the body of @code{T::fn}, @var{this} will have the effective
12396 definition @code{T *__restrict__ const this}. Notice that the
12397 interpretation of a @code{__restrict__} member function qualifier is
12398 different to that of @code{const} or @code{volatile} qualifier, in that it
12399 is applied to the pointer rather than the object. This is consistent with
12400 other compilers which implement restricted pointers.
12402 As with all outermost parameter qualifiers, @code{__restrict__} is
12403 ignored in function definition matching. This means you only need to
12404 specify @code{__restrict__} in a function definition, rather than
12405 in a function prototype as well.
12407 @node Vague Linkage
12408 @section Vague Linkage
12409 @cindex vague linkage
12411 There are several constructs in C++ which require space in the object
12412 file but are not clearly tied to a single translation unit. We say that
12413 these constructs have ``vague linkage''. Typically such constructs are
12414 emitted wherever they are needed, though sometimes we can be more
12418 @item Inline Functions
12419 Inline functions are typically defined in a header file which can be
12420 included in many different compilations. Hopefully they can usually be
12421 inlined, but sometimes an out-of-line copy is necessary, if the address
12422 of the function is taken or if inlining fails. In general, we emit an
12423 out-of-line copy in all translation units where one is needed. As an
12424 exception, we only emit inline virtual functions with the vtable, since
12425 it will always require a copy.
12427 Local static variables and string constants used in an inline function
12428 are also considered to have vague linkage, since they must be shared
12429 between all inlined and out-of-line instances of the function.
12433 C++ virtual functions are implemented in most compilers using a lookup
12434 table, known as a vtable. The vtable contains pointers to the virtual
12435 functions provided by a class, and each object of the class contains a
12436 pointer to its vtable (or vtables, in some multiple-inheritance
12437 situations). If the class declares any non-inline, non-pure virtual
12438 functions, the first one is chosen as the ``key method'' for the class,
12439 and the vtable is only emitted in the translation unit where the key
12442 @emph{Note:} If the chosen key method is later defined as inline, the
12443 vtable will still be emitted in every translation unit which defines it.
12444 Make sure that any inline virtuals are declared inline in the class
12445 body, even if they are not defined there.
12447 @item type_info objects
12450 C++ requires information about types to be written out in order to
12451 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
12452 For polymorphic classes (classes with virtual functions), the type_info
12453 object is written out along with the vtable so that @samp{dynamic_cast}
12454 can determine the dynamic type of a class object at runtime. For all
12455 other types, we write out the type_info object when it is used: when
12456 applying @samp{typeid} to an expression, throwing an object, or
12457 referring to a type in a catch clause or exception specification.
12459 @item Template Instantiations
12460 Most everything in this section also applies to template instantiations,
12461 but there are other options as well.
12462 @xref{Template Instantiation,,Where's the Template?}.
12466 When used with GNU ld version 2.8 or later on an ELF system such as
12467 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
12468 these constructs will be discarded at link time. This is known as
12471 On targets that don't support COMDAT, but do support weak symbols, GCC
12472 will use them. This way one copy will override all the others, but
12473 the unused copies will still take up space in the executable.
12475 For targets which do not support either COMDAT or weak symbols,
12476 most entities with vague linkage will be emitted as local symbols to
12477 avoid duplicate definition errors from the linker. This will not happen
12478 for local statics in inlines, however, as having multiple copies will
12479 almost certainly break things.
12481 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
12482 another way to control placement of these constructs.
12484 @node C++ Interface
12485 @section #pragma interface and implementation
12487 @cindex interface and implementation headers, C++
12488 @cindex C++ interface and implementation headers
12489 @cindex pragmas, interface and implementation
12491 @code{#pragma interface} and @code{#pragma implementation} provide the
12492 user with a way of explicitly directing the compiler to emit entities
12493 with vague linkage (and debugging information) in a particular
12496 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
12497 most cases, because of COMDAT support and the ``key method'' heuristic
12498 mentioned in @ref{Vague Linkage}. Using them can actually cause your
12499 program to grow due to unnecessary out-of-line copies of inline
12500 functions. Currently (3.4) the only benefit of these
12501 @code{#pragma}s is reduced duplication of debugging information, and
12502 that should be addressed soon on DWARF 2 targets with the use of
12506 @item #pragma interface
12507 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
12508 @kindex #pragma interface
12509 Use this directive in @emph{header files} that define object classes, to save
12510 space in most of the object files that use those classes. Normally,
12511 local copies of certain information (backup copies of inline member
12512 functions, debugging information, and the internal tables that implement
12513 virtual functions) must be kept in each object file that includes class
12514 definitions. You can use this pragma to avoid such duplication. When a
12515 header file containing @samp{#pragma interface} is included in a
12516 compilation, this auxiliary information will not be generated (unless
12517 the main input source file itself uses @samp{#pragma implementation}).
12518 Instead, the object files will contain references to be resolved at link
12521 The second form of this directive is useful for the case where you have
12522 multiple headers with the same name in different directories. If you
12523 use this form, you must specify the same string to @samp{#pragma
12526 @item #pragma implementation
12527 @itemx #pragma implementation "@var{objects}.h"
12528 @kindex #pragma implementation
12529 Use this pragma in a @emph{main input file}, when you want full output from
12530 included header files to be generated (and made globally visible). The
12531 included header file, in turn, should use @samp{#pragma interface}.
12532 Backup copies of inline member functions, debugging information, and the
12533 internal tables used to implement virtual functions are all generated in
12534 implementation files.
12536 @cindex implied @code{#pragma implementation}
12537 @cindex @code{#pragma implementation}, implied
12538 @cindex naming convention, implementation headers
12539 If you use @samp{#pragma implementation} with no argument, it applies to
12540 an include file with the same basename@footnote{A file's @dfn{basename}
12541 was the name stripped of all leading path information and of trailing
12542 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
12543 file. For example, in @file{allclass.cc}, giving just
12544 @samp{#pragma implementation}
12545 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
12547 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
12548 an implementation file whenever you would include it from
12549 @file{allclass.cc} even if you never specified @samp{#pragma
12550 implementation}. This was deemed to be more trouble than it was worth,
12551 however, and disabled.
12553 Use the string argument if you want a single implementation file to
12554 include code from multiple header files. (You must also use
12555 @samp{#include} to include the header file; @samp{#pragma
12556 implementation} only specifies how to use the file---it doesn't actually
12559 There is no way to split up the contents of a single header file into
12560 multiple implementation files.
12563 @cindex inlining and C++ pragmas
12564 @cindex C++ pragmas, effect on inlining
12565 @cindex pragmas in C++, effect on inlining
12566 @samp{#pragma implementation} and @samp{#pragma interface} also have an
12567 effect on function inlining.
12569 If you define a class in a header file marked with @samp{#pragma
12570 interface}, the effect on an inline function defined in that class is
12571 similar to an explicit @code{extern} declaration---the compiler emits
12572 no code at all to define an independent version of the function. Its
12573 definition is used only for inlining with its callers.
12575 @opindex fno-implement-inlines
12576 Conversely, when you include the same header file in a main source file
12577 that declares it as @samp{#pragma implementation}, the compiler emits
12578 code for the function itself; this defines a version of the function
12579 that can be found via pointers (or by callers compiled without
12580 inlining). If all calls to the function can be inlined, you can avoid
12581 emitting the function by compiling with @option{-fno-implement-inlines}.
12582 If any calls were not inlined, you will get linker errors.
12584 @node Template Instantiation
12585 @section Where's the Template?
12586 @cindex template instantiation
12588 C++ templates are the first language feature to require more
12589 intelligence from the environment than one usually finds on a UNIX
12590 system. Somehow the compiler and linker have to make sure that each
12591 template instance occurs exactly once in the executable if it is needed,
12592 and not at all otherwise. There are two basic approaches to this
12593 problem, which are referred to as the Borland model and the Cfront model.
12596 @item Borland model
12597 Borland C++ solved the template instantiation problem by adding the code
12598 equivalent of common blocks to their linker; the compiler emits template
12599 instances in each translation unit that uses them, and the linker
12600 collapses them together. The advantage of this model is that the linker
12601 only has to consider the object files themselves; there is no external
12602 complexity to worry about. This disadvantage is that compilation time
12603 is increased because the template code is being compiled repeatedly.
12604 Code written for this model tends to include definitions of all
12605 templates in the header file, since they must be seen to be
12609 The AT&T C++ translator, Cfront, solved the template instantiation
12610 problem by creating the notion of a template repository, an
12611 automatically maintained place where template instances are stored. A
12612 more modern version of the repository works as follows: As individual
12613 object files are built, the compiler places any template definitions and
12614 instantiations encountered in the repository. At link time, the link
12615 wrapper adds in the objects in the repository and compiles any needed
12616 instances that were not previously emitted. The advantages of this
12617 model are more optimal compilation speed and the ability to use the
12618 system linker; to implement the Borland model a compiler vendor also
12619 needs to replace the linker. The disadvantages are vastly increased
12620 complexity, and thus potential for error; for some code this can be
12621 just as transparent, but in practice it can been very difficult to build
12622 multiple programs in one directory and one program in multiple
12623 directories. Code written for this model tends to separate definitions
12624 of non-inline member templates into a separate file, which should be
12625 compiled separately.
12628 When used with GNU ld version 2.8 or later on an ELF system such as
12629 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
12630 Borland model. On other systems, G++ implements neither automatic
12633 A future version of G++ will support a hybrid model whereby the compiler
12634 will emit any instantiations for which the template definition is
12635 included in the compile, and store template definitions and
12636 instantiation context information into the object file for the rest.
12637 The link wrapper will extract that information as necessary and invoke
12638 the compiler to produce the remaining instantiations. The linker will
12639 then combine duplicate instantiations.
12641 In the mean time, you have the following options for dealing with
12642 template instantiations:
12647 Compile your template-using code with @option{-frepo}. The compiler will
12648 generate files with the extension @samp{.rpo} listing all of the
12649 template instantiations used in the corresponding object files which
12650 could be instantiated there; the link wrapper, @samp{collect2}, will
12651 then update the @samp{.rpo} files to tell the compiler where to place
12652 those instantiations and rebuild any affected object files. The
12653 link-time overhead is negligible after the first pass, as the compiler
12654 will continue to place the instantiations in the same files.
12656 This is your best option for application code written for the Borland
12657 model, as it will just work. Code written for the Cfront model will
12658 need to be modified so that the template definitions are available at
12659 one or more points of instantiation; usually this is as simple as adding
12660 @code{#include <tmethods.cc>} to the end of each template header.
12662 For library code, if you want the library to provide all of the template
12663 instantiations it needs, just try to link all of its object files
12664 together; the link will fail, but cause the instantiations to be
12665 generated as a side effect. Be warned, however, that this may cause
12666 conflicts if multiple libraries try to provide the same instantiations.
12667 For greater control, use explicit instantiation as described in the next
12671 @opindex fno-implicit-templates
12672 Compile your code with @option{-fno-implicit-templates} to disable the
12673 implicit generation of template instances, and explicitly instantiate
12674 all the ones you use. This approach requires more knowledge of exactly
12675 which instances you need than do the others, but it's less
12676 mysterious and allows greater control. You can scatter the explicit
12677 instantiations throughout your program, perhaps putting them in the
12678 translation units where the instances are used or the translation units
12679 that define the templates themselves; you can put all of the explicit
12680 instantiations you need into one big file; or you can create small files
12687 template class Foo<int>;
12688 template ostream& operator <<
12689 (ostream&, const Foo<int>&);
12692 for each of the instances you need, and create a template instantiation
12693 library from those.
12695 If you are using Cfront-model code, you can probably get away with not
12696 using @option{-fno-implicit-templates} when compiling files that don't
12697 @samp{#include} the member template definitions.
12699 If you use one big file to do the instantiations, you may want to
12700 compile it without @option{-fno-implicit-templates} so you get all of the
12701 instances required by your explicit instantiations (but not by any
12702 other files) without having to specify them as well.
12704 G++ has extended the template instantiation syntax given in the ISO
12705 standard to allow forward declaration of explicit instantiations
12706 (with @code{extern}), instantiation of the compiler support data for a
12707 template class (i.e.@: the vtable) without instantiating any of its
12708 members (with @code{inline}), and instantiation of only the static data
12709 members of a template class, without the support data or member
12710 functions (with (@code{static}):
12713 extern template int max (int, int);
12714 inline template class Foo<int>;
12715 static template class Foo<int>;
12719 Do nothing. Pretend G++ does implement automatic instantiation
12720 management. Code written for the Borland model will work fine, but
12721 each translation unit will contain instances of each of the templates it
12722 uses. In a large program, this can lead to an unacceptable amount of code
12726 @node Bound member functions
12727 @section Extracting the function pointer from a bound pointer to member function
12729 @cindex pointer to member function
12730 @cindex bound pointer to member function
12732 In C++, pointer to member functions (PMFs) are implemented using a wide
12733 pointer of sorts to handle all the possible call mechanisms; the PMF
12734 needs to store information about how to adjust the @samp{this} pointer,
12735 and if the function pointed to is virtual, where to find the vtable, and
12736 where in the vtable to look for the member function. If you are using
12737 PMFs in an inner loop, you should really reconsider that decision. If
12738 that is not an option, you can extract the pointer to the function that
12739 would be called for a given object/PMF pair and call it directly inside
12740 the inner loop, to save a bit of time.
12742 Note that you will still be paying the penalty for the call through a
12743 function pointer; on most modern architectures, such a call defeats the
12744 branch prediction features of the CPU@. This is also true of normal
12745 virtual function calls.
12747 The syntax for this extension is
12751 extern int (A::*fp)();
12752 typedef int (*fptr)(A *);
12754 fptr p = (fptr)(a.*fp);
12757 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
12758 no object is needed to obtain the address of the function. They can be
12759 converted to function pointers directly:
12762 fptr p1 = (fptr)(&A::foo);
12765 @opindex Wno-pmf-conversions
12766 You must specify @option{-Wno-pmf-conversions} to use this extension.
12768 @node C++ Attributes
12769 @section C++-Specific Variable, Function, and Type Attributes
12771 Some attributes only make sense for C++ programs.
12774 @item init_priority (@var{priority})
12775 @cindex init_priority attribute
12778 In Standard C++, objects defined at namespace scope are guaranteed to be
12779 initialized in an order in strict accordance with that of their definitions
12780 @emph{in a given translation unit}. No guarantee is made for initializations
12781 across translation units. However, GNU C++ allows users to control the
12782 order of initialization of objects defined at namespace scope with the
12783 @code{init_priority} attribute by specifying a relative @var{priority},
12784 a constant integral expression currently bounded between 101 and 65535
12785 inclusive. Lower numbers indicate a higher priority.
12787 In the following example, @code{A} would normally be created before
12788 @code{B}, but the @code{init_priority} attribute has reversed that order:
12791 Some_Class A __attribute__ ((init_priority (2000)));
12792 Some_Class B __attribute__ ((init_priority (543)));
12796 Note that the particular values of @var{priority} do not matter; only their
12799 @item java_interface
12800 @cindex java_interface attribute
12802 This type attribute informs C++ that the class is a Java interface. It may
12803 only be applied to classes declared within an @code{extern "Java"} block.
12804 Calls to methods declared in this interface will be dispatched using GCJ's
12805 interface table mechanism, instead of regular virtual table dispatch.
12809 See also @ref{Namespace Association}.
12811 @node Namespace Association
12812 @section Namespace Association
12814 @strong{Caution:} The semantics of this extension are not fully
12815 defined. Users should refrain from using this extension as its
12816 semantics may change subtly over time. It is possible that this
12817 extension will be removed in future versions of G++.
12819 A using-directive with @code{__attribute ((strong))} is stronger
12820 than a normal using-directive in two ways:
12824 Templates from the used namespace can be specialized and explicitly
12825 instantiated as though they were members of the using namespace.
12828 The using namespace is considered an associated namespace of all
12829 templates in the used namespace for purposes of argument-dependent
12833 The used namespace must be nested within the using namespace so that
12834 normal unqualified lookup works properly.
12836 This is useful for composing a namespace transparently from
12837 implementation namespaces. For example:
12842 template <class T> struct A @{ @};
12844 using namespace debug __attribute ((__strong__));
12845 template <> struct A<int> @{ @}; // @r{ok to specialize}
12847 template <class T> void f (A<T>);
12852 f (std::A<float>()); // @r{lookup finds} std::f
12858 @section Type Traits
12860 The C++ front-end implements syntactic extensions that allow to
12861 determine at compile time various characteristics of a type (or of a
12865 @item __has_nothrow_assign (type)
12866 If @code{type} is const qualified or is a reference type then the trait is
12867 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
12868 is true, else if @code{type} is a cv class or union type with copy assignment
12869 operators that are known not to throw an exception then the trait is true,
12870 else it is false. Requires: @code{type} shall be a complete type, an array
12871 type of unknown bound, or is a @code{void} type.
12873 @item __has_nothrow_copy (type)
12874 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
12875 @code{type} is a cv class or union type with copy constructors that
12876 are known not to throw an exception then the trait is true, else it is false.
12877 Requires: @code{type} shall be a complete type, an array type of
12878 unknown bound, or is a @code{void} type.
12880 @item __has_nothrow_constructor (type)
12881 If @code{__has_trivial_constructor (type)} is true then the trait is
12882 true, else if @code{type} is a cv class or union type (or array
12883 thereof) with a default constructor that is known not to throw an
12884 exception then the trait is true, else it is false. Requires:
12885 @code{type} shall be a complete type, an array type of unknown bound,
12886 or is a @code{void} type.
12888 @item __has_trivial_assign (type)
12889 If @code{type} is const qualified or is a reference type then the trait is
12890 false. Otherwise if @code{__is_pod (type)} is true then the trait is
12891 true, else if @code{type} is a cv class or union type with a trivial
12892 copy assignment ([class.copy]) then the trait is true, else it is
12893 false. Requires: @code{type} shall be a complete type, an array type
12894 of unknown bound, or is a @code{void} type.
12896 @item __has_trivial_copy (type)
12897 If @code{__is_pod (type)} is true or @code{type} is a reference type
12898 then the trait is true, else if @code{type} is a cv class or union type
12899 with a trivial copy constructor ([class.copy]) then the trait
12900 is true, else it is false. Requires: @code{type} shall be a complete
12901 type, an array type of unknown bound, or is a @code{void} type.
12903 @item __has_trivial_constructor (type)
12904 If @code{__is_pod (type)} is true then the trait is true, else if
12905 @code{type} is a cv class or union type (or array thereof) with a
12906 trivial default constructor ([class.ctor]) then the trait is true,
12907 else it is false. Requires: @code{type} shall be a complete type, an
12908 array type of unknown bound, or is a @code{void} type.
12910 @item __has_trivial_destructor (type)
12911 If @code{__is_pod (type)} is true or @code{type} is a reference type then
12912 the trait is true, else if @code{type} is a cv class or union type (or
12913 array thereof) with a trivial destructor ([class.dtor]) then the trait
12914 is true, else it is false. Requires: @code{type} shall be a complete
12915 type, an array type of unknown bound, or is a @code{void} type.
12917 @item __has_virtual_destructor (type)
12918 If @code{type} is a class type with a virtual destructor
12919 ([class.dtor]) then the trait is true, else it is false. Requires:
12920 @code{type} shall be a complete type, an array type of unknown bound,
12921 or is a @code{void} type.
12923 @item __is_abstract (type)
12924 If @code{type} is an abstract class ([class.abstract]) then the trait
12925 is true, else it is false. Requires: @code{type} shall be a complete
12926 type, an array type of unknown bound, or is a @code{void} type.
12928 @item __is_base_of (base_type, derived_type)
12929 If @code{base_type} is a base class of @code{derived_type}
12930 ([class.derived]) then the trait is true, otherwise it is false.
12931 Top-level cv qualifications of @code{base_type} and
12932 @code{derived_type} are ignored. For the purposes of this trait, a
12933 class type is considered is own base. Requires: if @code{__is_class
12934 (base_type)} and @code{__is_class (derived_type)} are true and
12935 @code{base_type} and @code{derived_type} are not the same type
12936 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
12937 type. Diagnostic is produced if this requirement is not met.
12939 @item __is_class (type)
12940 If @code{type} is a cv class type, and not a union type
12941 ([basic.compound]) the trait is true, else it is false.
12943 @item __is_empty (type)
12944 If @code{__is_class (type)} is false then the trait is false.
12945 Otherwise @code{type} is considered empty if and only if: @code{type}
12946 has no non-static data members, or all non-static data members, if
12947 any, are bit-fields of length 0, and @code{type} has no virtual
12948 members, and @code{type} has no virtual base classes, and @code{type}
12949 has no base classes @code{base_type} for which
12950 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
12951 be a complete type, an array type of unknown bound, or is a
12954 @item __is_enum (type)
12955 If @code{type} is a cv enumeration type ([basic.compound]) the trait is
12956 true, else it is false.
12958 @item __is_pod (type)
12959 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
12960 else it is false. Requires: @code{type} shall be a complete type,
12961 an array type of unknown bound, or is a @code{void} type.
12963 @item __is_polymorphic (type)
12964 If @code{type} is a polymorphic class ([class.virtual]) then the trait
12965 is true, else it is false. Requires: @code{type} shall be a complete
12966 type, an array type of unknown bound, or is a @code{void} type.
12968 @item __is_union (type)
12969 If @code{type} is a cv union type ([basic.compound]) the trait is
12970 true, else it is false.
12974 @node Java Exceptions
12975 @section Java Exceptions
12977 The Java language uses a slightly different exception handling model
12978 from C++. Normally, GNU C++ will automatically detect when you are
12979 writing C++ code that uses Java exceptions, and handle them
12980 appropriately. However, if C++ code only needs to execute destructors
12981 when Java exceptions are thrown through it, GCC will guess incorrectly.
12982 Sample problematic code is:
12985 struct S @{ ~S(); @};
12986 extern void bar(); // @r{is written in Java, and may throw exceptions}
12995 The usual effect of an incorrect guess is a link failure, complaining of
12996 a missing routine called @samp{__gxx_personality_v0}.
12998 You can inform the compiler that Java exceptions are to be used in a
12999 translation unit, irrespective of what it might think, by writing
13000 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
13001 @samp{#pragma} must appear before any functions that throw or catch
13002 exceptions, or run destructors when exceptions are thrown through them.
13004 You cannot mix Java and C++ exceptions in the same translation unit. It
13005 is believed to be safe to throw a C++ exception from one file through
13006 another file compiled for the Java exception model, or vice versa, but
13007 there may be bugs in this area.
13009 @node Deprecated Features
13010 @section Deprecated Features
13012 In the past, the GNU C++ compiler was extended to experiment with new
13013 features, at a time when the C++ language was still evolving. Now that
13014 the C++ standard is complete, some of those features are superseded by
13015 superior alternatives. Using the old features might cause a warning in
13016 some cases that the feature will be dropped in the future. In other
13017 cases, the feature might be gone already.
13019 While the list below is not exhaustive, it documents some of the options
13020 that are now deprecated:
13023 @item -fexternal-templates
13024 @itemx -falt-external-templates
13025 These are two of the many ways for G++ to implement template
13026 instantiation. @xref{Template Instantiation}. The C++ standard clearly
13027 defines how template definitions have to be organized across
13028 implementation units. G++ has an implicit instantiation mechanism that
13029 should work just fine for standard-conforming code.
13031 @item -fstrict-prototype
13032 @itemx -fno-strict-prototype
13033 Previously it was possible to use an empty prototype parameter list to
13034 indicate an unspecified number of parameters (like C), rather than no
13035 parameters, as C++ demands. This feature has been removed, except where
13036 it is required for backwards compatibility. @xref{Backwards Compatibility}.
13039 G++ allows a virtual function returning @samp{void *} to be overridden
13040 by one returning a different pointer type. This extension to the
13041 covariant return type rules is now deprecated and will be removed from a
13044 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
13045 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
13046 and are now removed from G++. Code using these operators should be
13047 modified to use @code{std::min} and @code{std::max} instead.
13049 The named return value extension has been deprecated, and is now
13052 The use of initializer lists with new expressions has been deprecated,
13053 and is now removed from G++.
13055 Floating and complex non-type template parameters have been deprecated,
13056 and are now removed from G++.
13058 The implicit typename extension has been deprecated and is now
13061 The use of default arguments in function pointers, function typedefs
13062 and other places where they are not permitted by the standard is
13063 deprecated and will be removed from a future version of G++.
13065 G++ allows floating-point literals to appear in integral constant expressions,
13066 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
13067 This extension is deprecated and will be removed from a future version.
13069 G++ allows static data members of const floating-point type to be declared
13070 with an initializer in a class definition. The standard only allows
13071 initializers for static members of const integral types and const
13072 enumeration types so this extension has been deprecated and will be removed
13073 from a future version.
13075 @node Backwards Compatibility
13076 @section Backwards Compatibility
13077 @cindex Backwards Compatibility
13078 @cindex ARM [Annotated C++ Reference Manual]
13080 Now that there is a definitive ISO standard C++, G++ has a specification
13081 to adhere to. The C++ language evolved over time, and features that
13082 used to be acceptable in previous drafts of the standard, such as the ARM
13083 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
13084 compilation of C++ written to such drafts, G++ contains some backwards
13085 compatibilities. @emph{All such backwards compatibility features are
13086 liable to disappear in future versions of G++.} They should be considered
13087 deprecated. @xref{Deprecated Features}.
13091 If a variable is declared at for scope, it used to remain in scope until
13092 the end of the scope which contained the for statement (rather than just
13093 within the for scope). G++ retains this, but issues a warning, if such a
13094 variable is accessed outside the for scope.
13096 @item Implicit C language
13097 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
13098 scope to set the language. On such systems, all header files are
13099 implicitly scoped inside a C language scope. Also, an empty prototype
13100 @code{()} will be treated as an unspecified number of arguments, rather
13101 than no arguments, as C++ demands.