1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1996, 1998, 1999, 2000, 2001,
2 @c 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
3 @c Free Software Foundation, Inc.
5 @c This is part of the GCC manual.
6 @c For copying conditions, see the file gcc.texi.
9 @chapter Extensions to the C Language Family
10 @cindex extensions, C language
11 @cindex C language extensions
14 GNU C provides several language features not found in ISO standard C@.
15 (The @option{-pedantic} option directs GCC to print a warning message if
16 any of these features is used.) To test for the availability of these
17 features in conditional compilation, check for a predefined macro
18 @code{__GNUC__}, which is always defined under GCC@.
20 These extensions are available in C and Objective-C@. Most of them are
21 also available in C++. @xref{C++ Extensions,,Extensions to the
22 C++ Language}, for extensions that apply @emph{only} to C++.
24 Some features that are in ISO C99 but not C89 or C++ are also, as
25 extensions, accepted by GCC in C89 mode and in C++.
28 * Statement Exprs:: Putting statements and declarations inside expressions.
29 * Local Labels:: Labels local to a block.
30 * Labels as Values:: Getting pointers to labels, and computed gotos.
31 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
32 * Constructing Calls:: Dispatching a call to another function.
33 * Typeof:: @code{typeof}: referring to the type of an expression.
34 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
35 * Long Long:: Double-word integers---@code{long long int}.
36 * Complex:: Data types for complex numbers.
37 * Floating Types:: Additional Floating Types.
38 * Half-Precision:: Half-Precision Floating Point.
39 * Decimal Float:: Decimal Floating Types.
40 * Hex Floats:: Hexadecimal floating-point constants.
41 * Fixed-Point:: Fixed-Point Types.
42 * Named Address Spaces::Named address spaces.
43 * Zero Length:: Zero-length arrays.
44 * Variable Length:: Arrays whose length is computed at run time.
45 * Empty Structures:: Structures with no members.
46 * Variadic Macros:: Macros with a variable number of arguments.
47 * Escaped Newlines:: Slightly looser rules for escaped newlines.
48 * Subscripting:: Any array can be subscripted, even if not an lvalue.
49 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
50 * Initializers:: Non-constant initializers.
51 * Compound Literals:: Compound literals give structures, unions
53 * Designated Inits:: Labeling elements of initializers.
54 * Cast to Union:: Casting to union type from any member of the union.
55 * Case Ranges:: `case 1 ... 9' and such.
56 * Mixed Declarations:: Mixing declarations and code.
57 * Function Attributes:: Declaring that functions have no side effects,
58 or that they can never return.
59 * Attribute Syntax:: Formal syntax for attributes.
60 * Function Prototypes:: Prototype declarations and old-style definitions.
61 * C++ Comments:: C++ comments are recognized.
62 * Dollar Signs:: Dollar sign is allowed in identifiers.
63 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
64 * Variable Attributes:: Specifying attributes of variables.
65 * Type Attributes:: Specifying attributes of types.
66 * Alignment:: Inquiring about the alignment of a type or variable.
67 * Inline:: Defining inline functions (as fast as macros).
68 * Extended Asm:: Assembler instructions with C expressions as operands.
69 (With them you can define ``built-in'' functions.)
70 * Constraints:: Constraints for asm operands
71 * Asm Labels:: Specifying the assembler name to use for a C symbol.
72 * Explicit Reg Vars:: Defining variables residing in specified registers.
73 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
74 * Incomplete Enums:: @code{enum foo;}, with details to follow.
75 * Function Names:: Printable strings which are the name of the current
77 * Return Address:: Getting the return or frame address of a function.
78 * Vector Extensions:: Using vector instructions through built-in functions.
79 * Offsetof:: Special syntax for implementing @code{offsetof}.
80 * Atomic Builtins:: Built-in functions for atomic memory access.
81 * Object Size Checking:: Built-in functions for limited buffer overflow
83 * Other Builtins:: Other built-in functions.
84 * Target Builtins:: Built-in functions specific to particular targets.
85 * Target Format Checks:: Format checks specific to particular targets.
86 * Pragmas:: Pragmas accepted by GCC.
87 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
88 * Thread-Local:: Per-thread variables.
89 * Binary constants:: Binary constants using the @samp{0b} prefix.
93 @section Statements and Declarations in Expressions
94 @cindex statements inside expressions
95 @cindex declarations inside expressions
96 @cindex expressions containing statements
97 @cindex macros, statements in expressions
99 @c the above section title wrapped and causes an underfull hbox.. i
100 @c changed it from "within" to "in". --mew 4feb93
101 A compound statement enclosed in parentheses may appear as an expression
102 in GNU C@. This allows you to use loops, switches, and local variables
103 within an expression.
105 Recall that a compound statement is a sequence of statements surrounded
106 by braces; in this construct, parentheses go around the braces. For
110 (@{ int y = foo (); int z;
117 is a valid (though slightly more complex than necessary) expression
118 for the absolute value of @code{foo ()}.
120 The last thing in the compound statement should be an expression
121 followed by a semicolon; the value of this subexpression serves as the
122 value of the entire construct. (If you use some other kind of statement
123 last within the braces, the construct has type @code{void}, and thus
124 effectively no value.)
126 This feature is especially useful in making macro definitions ``safe'' (so
127 that they evaluate each operand exactly once). For example, the
128 ``maximum'' function is commonly defined as a macro in standard C as
132 #define max(a,b) ((a) > (b) ? (a) : (b))
136 @cindex side effects, macro argument
137 But this definition computes either @var{a} or @var{b} twice, with bad
138 results if the operand has side effects. In GNU C, if you know the
139 type of the operands (here taken as @code{int}), you can define
140 the macro safely as follows:
143 #define maxint(a,b) \
144 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
147 Embedded statements are not allowed in constant expressions, such as
148 the value of an enumeration constant, the width of a bit-field, or
149 the initial value of a static variable.
151 If you don't know the type of the operand, you can still do this, but you
152 must use @code{typeof} (@pxref{Typeof}).
154 In G++, the result value of a statement expression undergoes array and
155 function pointer decay, and is returned by value to the enclosing
156 expression. For instance, if @code{A} is a class, then
165 will construct a temporary @code{A} object to hold the result of the
166 statement expression, and that will be used to invoke @code{Foo}.
167 Therefore the @code{this} pointer observed by @code{Foo} will not be the
170 Any temporaries created within a statement within a statement expression
171 will be destroyed at the statement's end. This makes statement
172 expressions inside macros slightly different from function calls. In
173 the latter case temporaries introduced during argument evaluation will
174 be destroyed at the end of the statement that includes the function
175 call. In the statement expression case they will be destroyed during
176 the statement expression. For instance,
179 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
180 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
190 will have different places where temporaries are destroyed. For the
191 @code{macro} case, the temporary @code{X} will be destroyed just after
192 the initialization of @code{b}. In the @code{function} case that
193 temporary will be destroyed when the function returns.
195 These considerations mean that it is probably a bad idea to use
196 statement-expressions of this form in header files that are designed to
197 work with C++. (Note that some versions of the GNU C Library contained
198 header files using statement-expression that lead to precisely this
201 Jumping into a statement expression with @code{goto} or using a
202 @code{switch} statement outside the statement expression with a
203 @code{case} or @code{default} label inside the statement expression is
204 not permitted. Jumping into a statement expression with a computed
205 @code{goto} (@pxref{Labels as Values}) yields undefined behavior.
206 Jumping out of a statement expression is permitted, but if the
207 statement expression is part of a larger expression then it is
208 unspecified which other subexpressions of that expression have been
209 evaluated except where the language definition requires certain
210 subexpressions to be evaluated before or after the statement
211 expression. In any case, as with a function call the evaluation of a
212 statement expression is not interleaved with the evaluation of other
213 parts of the containing expression. For example,
216 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
220 will call @code{foo} and @code{bar1} and will not call @code{baz} but
221 may or may not call @code{bar2}. If @code{bar2} is called, it will be
222 called after @code{foo} and before @code{bar1}
225 @section Locally Declared Labels
227 @cindex macros, local labels
229 GCC allows you to declare @dfn{local labels} in any nested block
230 scope. A local label is just like an ordinary label, but you can
231 only reference it (with a @code{goto} statement, or by taking its
232 address) within the block in which it was declared.
234 A local label declaration looks like this:
237 __label__ @var{label};
244 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
247 Local label declarations must come at the beginning of the block,
248 before any ordinary declarations or statements.
250 The label declaration defines the label @emph{name}, but does not define
251 the label itself. You must do this in the usual way, with
252 @code{@var{label}:}, within the statements of the statement expression.
254 The local label feature is useful for complex macros. If a macro
255 contains nested loops, a @code{goto} can be useful for breaking out of
256 them. However, an ordinary label whose scope is the whole function
257 cannot be used: if the macro can be expanded several times in one
258 function, the label will be multiply defined in that function. A
259 local label avoids this problem. For example:
262 #define SEARCH(value, array, target) \
265 typeof (target) _SEARCH_target = (target); \
266 typeof (*(array)) *_SEARCH_array = (array); \
269 for (i = 0; i < max; i++) \
270 for (j = 0; j < max; j++) \
271 if (_SEARCH_array[i][j] == _SEARCH_target) \
272 @{ (value) = i; goto found; @} \
278 This could also be written using a statement-expression:
281 #define SEARCH(array, target) \
284 typeof (target) _SEARCH_target = (target); \
285 typeof (*(array)) *_SEARCH_array = (array); \
288 for (i = 0; i < max; i++) \
289 for (j = 0; j < max; j++) \
290 if (_SEARCH_array[i][j] == _SEARCH_target) \
291 @{ value = i; goto found; @} \
298 Local label declarations also make the labels they declare visible to
299 nested functions, if there are any. @xref{Nested Functions}, for details.
301 @node Labels as Values
302 @section Labels as Values
303 @cindex labels as values
304 @cindex computed gotos
305 @cindex goto with computed label
306 @cindex address of a label
308 You can get the address of a label defined in the current function
309 (or a containing function) with the unary operator @samp{&&}. The
310 value has type @code{void *}. This value is a constant and can be used
311 wherever a constant of that type is valid. For example:
319 To use these values, you need to be able to jump to one. This is done
320 with the computed goto statement@footnote{The analogous feature in
321 Fortran is called an assigned goto, but that name seems inappropriate in
322 C, where one can do more than simply store label addresses in label
323 variables.}, @code{goto *@var{exp};}. For example,
330 Any expression of type @code{void *} is allowed.
332 One way of using these constants is in initializing a static array that
333 will serve as a jump table:
336 static void *array[] = @{ &&foo, &&bar, &&hack @};
339 Then you can select a label with indexing, like this:
346 Note that this does not check whether the subscript is in bounds---array
347 indexing in C never does that.
349 Such an array of label values serves a purpose much like that of the
350 @code{switch} statement. The @code{switch} statement is cleaner, so
351 use that rather than an array unless the problem does not fit a
352 @code{switch} statement very well.
354 Another use of label values is in an interpreter for threaded code.
355 The labels within the interpreter function can be stored in the
356 threaded code for super-fast dispatching.
358 You may not use this mechanism to jump to code in a different function.
359 If you do that, totally unpredictable things will happen. The best way to
360 avoid this is to store the label address only in automatic variables and
361 never pass it as an argument.
363 An alternate way to write the above example is
366 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
368 goto *(&&foo + array[i]);
372 This is more friendly to code living in shared libraries, as it reduces
373 the number of dynamic relocations that are needed, and by consequence,
374 allows the data to be read-only.
376 The @code{&&foo} expressions for the same label might have different
377 values if the containing function is inlined or cloned. If a program
378 relies on them being always the same,
379 @code{__attribute__((__noinline__,__noclone__))} should be used to
380 prevent inlining and cloning. If @code{&&foo} is used in a static
381 variable initializer, inlining and cloning is forbidden.
383 @node Nested Functions
384 @section Nested Functions
385 @cindex nested functions
386 @cindex downward funargs
389 A @dfn{nested function} is a function defined inside another function.
390 (Nested functions are not supported for GNU C++.) The nested function's
391 name is local to the block where it is defined. For example, here we
392 define a nested function named @code{square}, and call it twice:
396 foo (double a, double b)
398 double square (double z) @{ return z * z; @}
400 return square (a) + square (b);
405 The nested function can access all the variables of the containing
406 function that are visible at the point of its definition. This is
407 called @dfn{lexical scoping}. For example, here we show a nested
408 function which uses an inherited variable named @code{offset}:
412 bar (int *array, int offset, int size)
414 int access (int *array, int index)
415 @{ return array[index + offset]; @}
418 for (i = 0; i < size; i++)
419 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
424 Nested function definitions are permitted within functions in the places
425 where variable definitions are allowed; that is, in any block, mixed
426 with the other declarations and statements in the block.
428 It is possible to call the nested function from outside the scope of its
429 name by storing its address or passing the address to another function:
432 hack (int *array, int size)
434 void store (int index, int value)
435 @{ array[index] = value; @}
437 intermediate (store, size);
441 Here, the function @code{intermediate} receives the address of
442 @code{store} as an argument. If @code{intermediate} calls @code{store},
443 the arguments given to @code{store} are used to store into @code{array}.
444 But this technique works only so long as the containing function
445 (@code{hack}, in this example) does not exit.
447 If you try to call the nested function through its address after the
448 containing function has exited, all hell will break loose. If you try
449 to call it after a containing scope level has exited, and if it refers
450 to some of the variables that are no longer in scope, you may be lucky,
451 but it's not wise to take the risk. If, however, the nested function
452 does not refer to anything that has gone out of scope, you should be
455 GCC implements taking the address of a nested function using a technique
456 called @dfn{trampolines}. This technique was described in
457 @cite{Lexical Closures for C++} (Thomas M. Breuel, USENIX
458 C++ Conference Proceedings, October 17-21, 1988).
460 A nested function can jump to a label inherited from a containing
461 function, provided the label was explicitly declared in the containing
462 function (@pxref{Local Labels}). Such a jump returns instantly to the
463 containing function, exiting the nested function which did the
464 @code{goto} and any intermediate functions as well. Here is an example:
468 bar (int *array, int offset, int size)
471 int access (int *array, int index)
475 return array[index + offset];
479 for (i = 0; i < size; i++)
480 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
484 /* @r{Control comes here from @code{access}
485 if it detects an error.} */
492 A nested function always has no linkage. Declaring one with
493 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
494 before its definition, use @code{auto} (which is otherwise meaningless
495 for function declarations).
498 bar (int *array, int offset, int size)
501 auto int access (int *, int);
503 int access (int *array, int index)
507 return array[index + offset];
513 @node Constructing Calls
514 @section Constructing Function Calls
515 @cindex constructing calls
516 @cindex forwarding calls
518 Using the built-in functions described below, you can record
519 the arguments a function received, and call another function
520 with the same arguments, without knowing the number or types
523 You can also record the return value of that function call,
524 and later return that value, without knowing what data type
525 the function tried to return (as long as your caller expects
528 However, these built-in functions may interact badly with some
529 sophisticated features or other extensions of the language. It
530 is, therefore, not recommended to use them outside very simple
531 functions acting as mere forwarders for their arguments.
533 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
534 This built-in function returns a pointer to data
535 describing how to perform a call with the same arguments as were passed
536 to the current function.
538 The function saves the arg pointer register, structure value address,
539 and all registers that might be used to pass arguments to a function
540 into a block of memory allocated on the stack. Then it returns the
541 address of that block.
544 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
545 This built-in function invokes @var{function}
546 with a copy of the parameters described by @var{arguments}
549 The value of @var{arguments} should be the value returned by
550 @code{__builtin_apply_args}. The argument @var{size} specifies the size
551 of the stack argument data, in bytes.
553 This function returns a pointer to data describing
554 how to return whatever value was returned by @var{function}. The data
555 is saved in a block of memory allocated on the stack.
557 It is not always simple to compute the proper value for @var{size}. The
558 value is used by @code{__builtin_apply} to compute the amount of data
559 that should be pushed on the stack and copied from the incoming argument
563 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
564 This built-in function returns the value described by @var{result} from
565 the containing function. You should specify, for @var{result}, a value
566 returned by @code{__builtin_apply}.
569 @deftypefn {Built-in Function} __builtin_va_arg_pack ()
570 This built-in function represents all anonymous arguments of an inline
571 function. It can be used only in inline functions which will be always
572 inlined, never compiled as a separate function, such as those using
573 @code{__attribute__ ((__always_inline__))} or
574 @code{__attribute__ ((__gnu_inline__))} extern inline functions.
575 It must be only passed as last argument to some other function
576 with variable arguments. This is useful for writing small wrapper
577 inlines for variable argument functions, when using preprocessor
578 macros is undesirable. For example:
580 extern int myprintf (FILE *f, const char *format, ...);
581 extern inline __attribute__ ((__gnu_inline__)) int
582 myprintf (FILE *f, const char *format, ...)
584 int r = fprintf (f, "myprintf: ");
587 int s = fprintf (f, format, __builtin_va_arg_pack ());
595 @deftypefn {Built-in Function} __builtin_va_arg_pack_len ()
596 This built-in function returns the number of anonymous arguments of
597 an inline function. It can be used only in inline functions which
598 will be always inlined, never compiled as a separate function, such
599 as those using @code{__attribute__ ((__always_inline__))} or
600 @code{__attribute__ ((__gnu_inline__))} extern inline functions.
601 For example following will do link or runtime checking of open
602 arguments for optimized code:
605 extern inline __attribute__((__gnu_inline__)) int
606 myopen (const char *path, int oflag, ...)
608 if (__builtin_va_arg_pack_len () > 1)
609 warn_open_too_many_arguments ();
611 if (__builtin_constant_p (oflag))
613 if ((oflag & O_CREAT) != 0 && __builtin_va_arg_pack_len () < 1)
615 warn_open_missing_mode ();
616 return __open_2 (path, oflag);
618 return open (path, oflag, __builtin_va_arg_pack ());
621 if (__builtin_va_arg_pack_len () < 1)
622 return __open_2 (path, oflag);
624 return open (path, oflag, __builtin_va_arg_pack ());
631 @section Referring to a Type with @code{typeof}
634 @cindex macros, types of arguments
636 Another way to refer to the type of an expression is with @code{typeof}.
637 The syntax of using of this keyword looks like @code{sizeof}, but the
638 construct acts semantically like a type name defined with @code{typedef}.
640 There are two ways of writing the argument to @code{typeof}: with an
641 expression or with a type. Here is an example with an expression:
648 This assumes that @code{x} is an array of pointers to functions;
649 the type described is that of the values of the functions.
651 Here is an example with a typename as the argument:
658 Here the type described is that of pointers to @code{int}.
660 If you are writing a header file that must work when included in ISO C
661 programs, write @code{__typeof__} instead of @code{typeof}.
662 @xref{Alternate Keywords}.
664 A @code{typeof}-construct can be used anywhere a typedef name could be
665 used. For example, you can use it in a declaration, in a cast, or inside
666 of @code{sizeof} or @code{typeof}.
668 The operand of @code{typeof} is evaluated for its side effects if and
669 only if it is an expression of variably modified type or the name of
672 @code{typeof} is often useful in conjunction with the
673 statements-within-expressions feature. Here is how the two together can
674 be used to define a safe ``maximum'' macro that operates on any
675 arithmetic type and evaluates each of its arguments exactly once:
679 (@{ typeof (a) _a = (a); \
680 typeof (b) _b = (b); \
681 _a > _b ? _a : _b; @})
684 @cindex underscores in variables in macros
685 @cindex @samp{_} in variables in macros
686 @cindex local variables in macros
687 @cindex variables, local, in macros
688 @cindex macros, local variables in
690 The reason for using names that start with underscores for the local
691 variables is to avoid conflicts with variable names that occur within the
692 expressions that are substituted for @code{a} and @code{b}. Eventually we
693 hope to design a new form of declaration syntax that allows you to declare
694 variables whose scopes start only after their initializers; this will be a
695 more reliable way to prevent such conflicts.
698 Some more examples of the use of @code{typeof}:
702 This declares @code{y} with the type of what @code{x} points to.
709 This declares @code{y} as an array of such values.
716 This declares @code{y} as an array of pointers to characters:
719 typeof (typeof (char *)[4]) y;
723 It is equivalent to the following traditional C declaration:
729 To see the meaning of the declaration using @code{typeof}, and why it
730 might be a useful way to write, rewrite it with these macros:
733 #define pointer(T) typeof(T *)
734 #define array(T, N) typeof(T [N])
738 Now the declaration can be rewritten this way:
741 array (pointer (char), 4) y;
745 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
746 pointers to @code{char}.
749 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
750 a more limited extension which permitted one to write
753 typedef @var{T} = @var{expr};
757 with the effect of declaring @var{T} to have the type of the expression
758 @var{expr}. This extension does not work with GCC 3 (versions between
759 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
760 relies on it should be rewritten to use @code{typeof}:
763 typedef typeof(@var{expr}) @var{T};
767 This will work with all versions of GCC@.
770 @section Conditionals with Omitted Operands
771 @cindex conditional expressions, extensions
772 @cindex omitted middle-operands
773 @cindex middle-operands, omitted
774 @cindex extensions, @code{?:}
775 @cindex @code{?:} extensions
777 The middle operand in a conditional expression may be omitted. Then
778 if the first operand is nonzero, its value is the value of the conditional
781 Therefore, the expression
788 has the value of @code{x} if that is nonzero; otherwise, the value of
791 This example is perfectly equivalent to
797 @cindex side effect in ?:
798 @cindex ?: side effect
800 In this simple case, the ability to omit the middle operand is not
801 especially useful. When it becomes useful is when the first operand does,
802 or may (if it is a macro argument), contain a side effect. Then repeating
803 the operand in the middle would perform the side effect twice. Omitting
804 the middle operand uses the value already computed without the undesirable
805 effects of recomputing it.
808 @section Double-Word Integers
809 @cindex @code{long long} data types
810 @cindex double-word arithmetic
811 @cindex multiprecision arithmetic
812 @cindex @code{LL} integer suffix
813 @cindex @code{ULL} integer suffix
815 ISO C99 supports data types for integers that are at least 64 bits wide,
816 and as an extension GCC supports them in C89 mode and in C++.
817 Simply write @code{long long int} for a signed integer, or
818 @code{unsigned long long int} for an unsigned integer. To make an
819 integer constant of type @code{long long int}, add the suffix @samp{LL}
820 to the integer. To make an integer constant of type @code{unsigned long
821 long int}, add the suffix @samp{ULL} to the integer.
823 You can use these types in arithmetic like any other integer types.
824 Addition, subtraction, and bitwise boolean operations on these types
825 are open-coded on all types of machines. Multiplication is open-coded
826 if the machine supports fullword-to-doubleword a widening multiply
827 instruction. Division and shifts are open-coded only on machines that
828 provide special support. The operations that are not open-coded use
829 special library routines that come with GCC@.
831 There may be pitfalls when you use @code{long long} types for function
832 arguments, unless you declare function prototypes. If a function
833 expects type @code{int} for its argument, and you pass a value of type
834 @code{long long int}, confusion will result because the caller and the
835 subroutine will disagree about the number of bytes for the argument.
836 Likewise, if the function expects @code{long long int} and you pass
837 @code{int}. The best way to avoid such problems is to use prototypes.
840 @section Complex Numbers
841 @cindex complex numbers
842 @cindex @code{_Complex} keyword
843 @cindex @code{__complex__} keyword
845 ISO C99 supports complex floating data types, and as an extension GCC
846 supports them in C89 mode and in C++, and supports complex integer data
847 types which are not part of ISO C99. You can declare complex types
848 using the keyword @code{_Complex}. As an extension, the older GNU
849 keyword @code{__complex__} is also supported.
851 For example, @samp{_Complex double x;} declares @code{x} as a
852 variable whose real part and imaginary part are both of type
853 @code{double}. @samp{_Complex short int y;} declares @code{y} to
854 have real and imaginary parts of type @code{short int}; this is not
855 likely to be useful, but it shows that the set of complex types is
858 To write a constant with a complex data type, use the suffix @samp{i} or
859 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
860 has type @code{_Complex float} and @code{3i} has type
861 @code{_Complex int}. Such a constant always has a pure imaginary
862 value, but you can form any complex value you like by adding one to a
863 real constant. This is a GNU extension; if you have an ISO C99
864 conforming C library (such as GNU libc), and want to construct complex
865 constants of floating type, you should include @code{<complex.h>} and
866 use the macros @code{I} or @code{_Complex_I} instead.
868 @cindex @code{__real__} keyword
869 @cindex @code{__imag__} keyword
870 To extract the real part of a complex-valued expression @var{exp}, write
871 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
872 extract the imaginary part. This is a GNU extension; for values of
873 floating type, you should use the ISO C99 functions @code{crealf},
874 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
875 @code{cimagl}, declared in @code{<complex.h>} and also provided as
876 built-in functions by GCC@.
878 @cindex complex conjugation
879 The operator @samp{~} performs complex conjugation when used on a value
880 with a complex type. This is a GNU extension; for values of
881 floating type, you should use the ISO C99 functions @code{conjf},
882 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
883 provided as built-in functions by GCC@.
885 GCC can allocate complex automatic variables in a noncontiguous
886 fashion; it's even possible for the real part to be in a register while
887 the imaginary part is on the stack (or vice-versa). Only the DWARF2
888 debug info format can represent this, so use of DWARF2 is recommended.
889 If you are using the stabs debug info format, GCC describes a noncontiguous
890 complex variable as if it were two separate variables of noncomplex type.
891 If the variable's actual name is @code{foo}, the two fictitious
892 variables are named @code{foo$real} and @code{foo$imag}. You can
893 examine and set these two fictitious variables with your debugger.
896 @section Additional Floating Types
897 @cindex additional floating types
898 @cindex @code{__float80} data type
899 @cindex @code{__float128} data type
900 @cindex @code{w} floating point suffix
901 @cindex @code{q} floating point suffix
902 @cindex @code{W} floating point suffix
903 @cindex @code{Q} floating point suffix
905 As an extension, the GNU C compiler supports additional floating
906 types, @code{__float80} and @code{__float128} to support 80bit
907 (@code{XFmode}) and 128 bit (@code{TFmode}) floating types.
908 Support for additional types includes the arithmetic operators:
909 add, subtract, multiply, divide; unary arithmetic operators;
910 relational operators; equality operators; and conversions to and from
911 integer and other floating types. Use a suffix @samp{w} or @samp{W}
912 in a literal constant of type @code{__float80} and @samp{q} or @samp{Q}
913 for @code{_float128}. You can declare complex types using the
914 corresponding internal complex type, @code{XCmode} for @code{__float80}
915 type and @code{TCmode} for @code{__float128} type:
918 typedef _Complex float __attribute__((mode(TC))) _Complex128;
919 typedef _Complex float __attribute__((mode(XC))) _Complex80;
922 Not all targets support additional floating point types. @code{__float80}
923 and @code{__float128} types are supported on i386, x86_64 and ia64 targets.
926 @section Half-Precision Floating Point
927 @cindex half-precision floating point
928 @cindex @code{__fp16} data type
930 On ARM targets, GCC supports half-precision (16-bit) floating point via
931 the @code{__fp16} type. You must enable this type explicitly
932 with the @option{-mfp16-format} command-line option in order to use it.
934 ARM supports two incompatible representations for half-precision
935 floating-point values. You must choose one of the representations and
936 use it consistently in your program.
938 Specifying @option{-mfp16-format=ieee} selects the IEEE 754-2008 format.
939 This format can represent normalized values in the range of @math{2^{-14}} to 65504.
940 There are 11 bits of significand precision, approximately 3
943 Specifying @option{-mfp16-format=alternative} selects the ARM
944 alternative format. This representation is similar to the IEEE
945 format, but does not support infinities or NaNs. Instead, the range
946 of exponents is extended, so that this format can represent normalized
947 values in the range of @math{2^{-14}} to 131008.
949 The @code{__fp16} type is a storage format only. For purposes
950 of arithmetic and other operations, @code{__fp16} values in C or C++
951 expressions are automatically promoted to @code{float}. In addition,
952 you cannot declare a function with a return value or parameters
953 of type @code{__fp16}.
955 Note that conversions from @code{double} to @code{__fp16}
956 involve an intermediate conversion to @code{float}. Because
957 of rounding, this can sometimes produce a different result than a
960 ARM provides hardware support for conversions between
961 @code{__fp16} and @code{float} values
962 as an extension to VFP and NEON (Advanced SIMD). GCC generates
963 code using the instructions provided by this extension if you compile
964 with the options @option{-mfpu=neon-fp16 -mfloat-abi=softfp},
965 in addition to the @option{-mfp16-format} option to select
966 a half-precision format.
968 Language-level support for the @code{__fp16} data type is
969 independent of whether GCC generates code using hardware floating-point
970 instructions. In cases where hardware support is not specified, GCC
971 implements conversions between @code{__fp16} and @code{float} values
975 @section Decimal Floating Types
976 @cindex decimal floating types
977 @cindex @code{_Decimal32} data type
978 @cindex @code{_Decimal64} data type
979 @cindex @code{_Decimal128} data type
980 @cindex @code{df} integer suffix
981 @cindex @code{dd} integer suffix
982 @cindex @code{dl} integer suffix
983 @cindex @code{DF} integer suffix
984 @cindex @code{DD} integer suffix
985 @cindex @code{DL} integer suffix
987 As an extension, the GNU C compiler supports decimal floating types as
988 defined in the N1312 draft of ISO/IEC WDTR24732. Support for decimal
989 floating types in GCC will evolve as the draft technical report changes.
990 Calling conventions for any target might also change. Not all targets
991 support decimal floating types.
993 The decimal floating types are @code{_Decimal32}, @code{_Decimal64}, and
994 @code{_Decimal128}. They use a radix of ten, unlike the floating types
995 @code{float}, @code{double}, and @code{long double} whose radix is not
996 specified by the C standard but is usually two.
998 Support for decimal floating types includes the arithmetic operators
999 add, subtract, multiply, divide; unary arithmetic operators;
1000 relational operators; equality operators; and conversions to and from
1001 integer and other floating types. Use a suffix @samp{df} or
1002 @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd}
1003 or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for
1006 GCC support of decimal float as specified by the draft technical report
1011 When the value of a decimal floating type cannot be represented in the
1012 integer type to which it is being converted, the result is undefined
1013 rather than the result value specified by the draft technical report.
1016 GCC does not provide the C library functionality associated with
1017 @file{math.h}, @file{fenv.h}, @file{stdio.h}, @file{stdlib.h}, and
1018 @file{wchar.h}, which must come from a separate C library implementation.
1019 Because of this the GNU C compiler does not define macro
1020 @code{__STDC_DEC_FP__} to indicate that the implementation conforms to
1021 the technical report.
1024 Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128}
1025 are supported by the DWARF2 debug information format.
1031 ISO C99 supports floating-point numbers written not only in the usual
1032 decimal notation, such as @code{1.55e1}, but also numbers such as
1033 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
1034 supports this in C89 mode (except in some cases when strictly
1035 conforming) and in C++. In that format the
1036 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
1037 mandatory. The exponent is a decimal number that indicates the power of
1038 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
1045 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
1046 is the same as @code{1.55e1}.
1048 Unlike for floating-point numbers in the decimal notation the exponent
1049 is always required in the hexadecimal notation. Otherwise the compiler
1050 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
1051 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
1052 extension for floating-point constants of type @code{float}.
1055 @section Fixed-Point Types
1056 @cindex fixed-point types
1057 @cindex @code{_Fract} data type
1058 @cindex @code{_Accum} data type
1059 @cindex @code{_Sat} data type
1060 @cindex @code{hr} fixed-suffix
1061 @cindex @code{r} fixed-suffix
1062 @cindex @code{lr} fixed-suffix
1063 @cindex @code{llr} fixed-suffix
1064 @cindex @code{uhr} fixed-suffix
1065 @cindex @code{ur} fixed-suffix
1066 @cindex @code{ulr} fixed-suffix
1067 @cindex @code{ullr} fixed-suffix
1068 @cindex @code{hk} fixed-suffix
1069 @cindex @code{k} fixed-suffix
1070 @cindex @code{lk} fixed-suffix
1071 @cindex @code{llk} fixed-suffix
1072 @cindex @code{uhk} fixed-suffix
1073 @cindex @code{uk} fixed-suffix
1074 @cindex @code{ulk} fixed-suffix
1075 @cindex @code{ullk} fixed-suffix
1076 @cindex @code{HR} fixed-suffix
1077 @cindex @code{R} fixed-suffix
1078 @cindex @code{LR} fixed-suffix
1079 @cindex @code{LLR} fixed-suffix
1080 @cindex @code{UHR} fixed-suffix
1081 @cindex @code{UR} fixed-suffix
1082 @cindex @code{ULR} fixed-suffix
1083 @cindex @code{ULLR} fixed-suffix
1084 @cindex @code{HK} fixed-suffix
1085 @cindex @code{K} fixed-suffix
1086 @cindex @code{LK} fixed-suffix
1087 @cindex @code{LLK} fixed-suffix
1088 @cindex @code{UHK} fixed-suffix
1089 @cindex @code{UK} fixed-suffix
1090 @cindex @code{ULK} fixed-suffix
1091 @cindex @code{ULLK} fixed-suffix
1093 As an extension, the GNU C compiler supports fixed-point types as
1094 defined in the N1169 draft of ISO/IEC DTR 18037. Support for fixed-point
1095 types in GCC will evolve as the draft technical report changes.
1096 Calling conventions for any target might also change. Not all targets
1097 support fixed-point types.
1099 The fixed-point types are
1100 @code{short _Fract},
1103 @code{long long _Fract},
1104 @code{unsigned short _Fract},
1105 @code{unsigned _Fract},
1106 @code{unsigned long _Fract},
1107 @code{unsigned long long _Fract},
1108 @code{_Sat short _Fract},
1110 @code{_Sat long _Fract},
1111 @code{_Sat long long _Fract},
1112 @code{_Sat unsigned short _Fract},
1113 @code{_Sat unsigned _Fract},
1114 @code{_Sat unsigned long _Fract},
1115 @code{_Sat unsigned long long _Fract},
1116 @code{short _Accum},
1119 @code{long long _Accum},
1120 @code{unsigned short _Accum},
1121 @code{unsigned _Accum},
1122 @code{unsigned long _Accum},
1123 @code{unsigned long long _Accum},
1124 @code{_Sat short _Accum},
1126 @code{_Sat long _Accum},
1127 @code{_Sat long long _Accum},
1128 @code{_Sat unsigned short _Accum},
1129 @code{_Sat unsigned _Accum},
1130 @code{_Sat unsigned long _Accum},
1131 @code{_Sat unsigned long long _Accum}.
1133 Fixed-point data values contain fractional and optional integral parts.
1134 The format of fixed-point data varies and depends on the target machine.
1136 Support for fixed-point types includes:
1139 prefix and postfix increment and decrement operators (@code{++}, @code{--})
1141 unary arithmetic operators (@code{+}, @code{-}, @code{!})
1143 binary arithmetic operators (@code{+}, @code{-}, @code{*}, @code{/})
1145 binary shift operators (@code{<<}, @code{>>})
1147 relational operators (@code{<}, @code{<=}, @code{>=}, @code{>})
1149 equality operators (@code{==}, @code{!=})
1151 assignment operators (@code{+=}, @code{-=}, @code{*=}, @code{/=},
1152 @code{<<=}, @code{>>=})
1154 conversions to and from integer, floating-point, or fixed-point types
1157 Use a suffix in a fixed-point literal constant:
1159 @item @samp{hr} or @samp{HR} for @code{short _Fract} and
1160 @code{_Sat short _Fract}
1161 @item @samp{r} or @samp{R} for @code{_Fract} and @code{_Sat _Fract}
1162 @item @samp{lr} or @samp{LR} for @code{long _Fract} and
1163 @code{_Sat long _Fract}
1164 @item @samp{llr} or @samp{LLR} for @code{long long _Fract} and
1165 @code{_Sat long long _Fract}
1166 @item @samp{uhr} or @samp{UHR} for @code{unsigned short _Fract} and
1167 @code{_Sat unsigned short _Fract}
1168 @item @samp{ur} or @samp{UR} for @code{unsigned _Fract} and
1169 @code{_Sat unsigned _Fract}
1170 @item @samp{ulr} or @samp{ULR} for @code{unsigned long _Fract} and
1171 @code{_Sat unsigned long _Fract}
1172 @item @samp{ullr} or @samp{ULLR} for @code{unsigned long long _Fract}
1173 and @code{_Sat unsigned long long _Fract}
1174 @item @samp{hk} or @samp{HK} for @code{short _Accum} and
1175 @code{_Sat short _Accum}
1176 @item @samp{k} or @samp{K} for @code{_Accum} and @code{_Sat _Accum}
1177 @item @samp{lk} or @samp{LK} for @code{long _Accum} and
1178 @code{_Sat long _Accum}
1179 @item @samp{llk} or @samp{LLK} for @code{long long _Accum} and
1180 @code{_Sat long long _Accum}
1181 @item @samp{uhk} or @samp{UHK} for @code{unsigned short _Accum} and
1182 @code{_Sat unsigned short _Accum}
1183 @item @samp{uk} or @samp{UK} for @code{unsigned _Accum} and
1184 @code{_Sat unsigned _Accum}
1185 @item @samp{ulk} or @samp{ULK} for @code{unsigned long _Accum} and
1186 @code{_Sat unsigned long _Accum}
1187 @item @samp{ullk} or @samp{ULLK} for @code{unsigned long long _Accum}
1188 and @code{_Sat unsigned long long _Accum}
1191 GCC support of fixed-point types as specified by the draft technical report
1196 Pragmas to control overflow and rounding behaviors are not implemented.
1199 Fixed-point types are supported by the DWARF2 debug information format.
1201 @node Named Address Spaces
1202 @section Named address spaces
1203 @cindex named address spaces
1205 As an extension, the GNU C compiler supports named address spaces as
1206 defined in the N1275 draft of ISO/IEC DTR 18037. Support for named
1207 address spaces in GCC will evolve as the draft technical report changes.
1208 Calling conventions for any target might also change. At present, only
1209 the SPU target supports other address spaces. On the SPU target, for
1210 example, variables may be declared as belonging to another address space
1211 by qualifying the type with the @code{__ea} address space identifier:
1217 When the variable @code{i} is accessed, the compiler will generate
1218 special code to access this variable. It may use runtime library
1219 support, or generate special machine instructions to access that address
1222 The @code{__ea} identifier may be used exactly like any other C type
1223 qualifier (e.g., @code{const} or @code{volatile}). See the N1275
1224 document for more details.
1227 @section Arrays of Length Zero
1228 @cindex arrays of length zero
1229 @cindex zero-length arrays
1230 @cindex length-zero arrays
1231 @cindex flexible array members
1233 Zero-length arrays are allowed in GNU C@. They are very useful as the
1234 last element of a structure which is really a header for a variable-length
1243 struct line *thisline = (struct line *)
1244 malloc (sizeof (struct line) + this_length);
1245 thisline->length = this_length;
1248 In ISO C90, you would have to give @code{contents} a length of 1, which
1249 means either you waste space or complicate the argument to @code{malloc}.
1251 In ISO C99, you would use a @dfn{flexible array member}, which is
1252 slightly different in syntax and semantics:
1256 Flexible array members are written as @code{contents[]} without
1260 Flexible array members have incomplete type, and so the @code{sizeof}
1261 operator may not be applied. As a quirk of the original implementation
1262 of zero-length arrays, @code{sizeof} evaluates to zero.
1265 Flexible array members may only appear as the last member of a
1266 @code{struct} that is otherwise non-empty.
1269 A structure containing a flexible array member, or a union containing
1270 such a structure (possibly recursively), may not be a member of a
1271 structure or an element of an array. (However, these uses are
1272 permitted by GCC as extensions.)
1275 GCC versions before 3.0 allowed zero-length arrays to be statically
1276 initialized, as if they were flexible arrays. In addition to those
1277 cases that were useful, it also allowed initializations in situations
1278 that would corrupt later data. Non-empty initialization of zero-length
1279 arrays is now treated like any case where there are more initializer
1280 elements than the array holds, in that a suitable warning about "excess
1281 elements in array" is given, and the excess elements (all of them, in
1282 this case) are ignored.
1284 Instead GCC allows static initialization of flexible array members.
1285 This is equivalent to defining a new structure containing the original
1286 structure followed by an array of sufficient size to contain the data.
1287 I.e.@: in the following, @code{f1} is constructed as if it were declared
1293 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
1296 struct f1 f1; int data[3];
1297 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1301 The convenience of this extension is that @code{f1} has the desired
1302 type, eliminating the need to consistently refer to @code{f2.f1}.
1304 This has symmetry with normal static arrays, in that an array of
1305 unknown size is also written with @code{[]}.
1307 Of course, this extension only makes sense if the extra data comes at
1308 the end of a top-level object, as otherwise we would be overwriting
1309 data at subsequent offsets. To avoid undue complication and confusion
1310 with initialization of deeply nested arrays, we simply disallow any
1311 non-empty initialization except when the structure is the top-level
1312 object. For example:
1315 struct foo @{ int x; int y[]; @};
1316 struct bar @{ struct foo z; @};
1318 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1319 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1320 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1321 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1324 @node Empty Structures
1325 @section Structures With No Members
1326 @cindex empty structures
1327 @cindex zero-size structures
1329 GCC permits a C structure to have no members:
1336 The structure will have size zero. In C++, empty structures are part
1337 of the language. G++ treats empty structures as if they had a single
1338 member of type @code{char}.
1340 @node Variable Length
1341 @section Arrays of Variable Length
1342 @cindex variable-length arrays
1343 @cindex arrays of variable length
1346 Variable-length automatic arrays are allowed in ISO C99, and as an
1347 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1348 implementation of variable-length arrays does not yet conform in detail
1349 to the ISO C99 standard.) These arrays are
1350 declared like any other automatic arrays, but with a length that is not
1351 a constant expression. The storage is allocated at the point of
1352 declaration and deallocated when the brace-level is exited. For
1357 concat_fopen (char *s1, char *s2, char *mode)
1359 char str[strlen (s1) + strlen (s2) + 1];
1362 return fopen (str, mode);
1366 @cindex scope of a variable length array
1367 @cindex variable-length array scope
1368 @cindex deallocating variable length arrays
1369 Jumping or breaking out of the scope of the array name deallocates the
1370 storage. Jumping into the scope is not allowed; you get an error
1373 @cindex @code{alloca} vs variable-length arrays
1374 You can use the function @code{alloca} to get an effect much like
1375 variable-length arrays. The function @code{alloca} is available in
1376 many other C implementations (but not in all). On the other hand,
1377 variable-length arrays are more elegant.
1379 There are other differences between these two methods. Space allocated
1380 with @code{alloca} exists until the containing @emph{function} returns.
1381 The space for a variable-length array is deallocated as soon as the array
1382 name's scope ends. (If you use both variable-length arrays and
1383 @code{alloca} in the same function, deallocation of a variable-length array
1384 will also deallocate anything more recently allocated with @code{alloca}.)
1386 You can also use variable-length arrays as arguments to functions:
1390 tester (int len, char data[len][len])
1396 The length of an array is computed once when the storage is allocated
1397 and is remembered for the scope of the array in case you access it with
1400 If you want to pass the array first and the length afterward, you can
1401 use a forward declaration in the parameter list---another GNU extension.
1405 tester (int len; char data[len][len], int len)
1411 @cindex parameter forward declaration
1412 The @samp{int len} before the semicolon is a @dfn{parameter forward
1413 declaration}, and it serves the purpose of making the name @code{len}
1414 known when the declaration of @code{data} is parsed.
1416 You can write any number of such parameter forward declarations in the
1417 parameter list. They can be separated by commas or semicolons, but the
1418 last one must end with a semicolon, which is followed by the ``real''
1419 parameter declarations. Each forward declaration must match a ``real''
1420 declaration in parameter name and data type. ISO C99 does not support
1421 parameter forward declarations.
1423 @node Variadic Macros
1424 @section Macros with a Variable Number of Arguments.
1425 @cindex variable number of arguments
1426 @cindex macro with variable arguments
1427 @cindex rest argument (in macro)
1428 @cindex variadic macros
1430 In the ISO C standard of 1999, a macro can be declared to accept a
1431 variable number of arguments much as a function can. The syntax for
1432 defining the macro is similar to that of a function. Here is an
1436 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1439 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1440 such a macro, it represents the zero or more tokens until the closing
1441 parenthesis that ends the invocation, including any commas. This set of
1442 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1443 wherever it appears. See the CPP manual for more information.
1445 GCC has long supported variadic macros, and used a different syntax that
1446 allowed you to give a name to the variable arguments just like any other
1447 argument. Here is an example:
1450 #define debug(format, args...) fprintf (stderr, format, args)
1453 This is in all ways equivalent to the ISO C example above, but arguably
1454 more readable and descriptive.
1456 GNU CPP has two further variadic macro extensions, and permits them to
1457 be used with either of the above forms of macro definition.
1459 In standard C, you are not allowed to leave the variable argument out
1460 entirely; but you are allowed to pass an empty argument. For example,
1461 this invocation is invalid in ISO C, because there is no comma after
1468 GNU CPP permits you to completely omit the variable arguments in this
1469 way. In the above examples, the compiler would complain, though since
1470 the expansion of the macro still has the extra comma after the format
1473 To help solve this problem, CPP behaves specially for variable arguments
1474 used with the token paste operator, @samp{##}. If instead you write
1477 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1480 and if the variable arguments are omitted or empty, the @samp{##}
1481 operator causes the preprocessor to remove the comma before it. If you
1482 do provide some variable arguments in your macro invocation, GNU CPP
1483 does not complain about the paste operation and instead places the
1484 variable arguments after the comma. Just like any other pasted macro
1485 argument, these arguments are not macro expanded.
1487 @node Escaped Newlines
1488 @section Slightly Looser Rules for Escaped Newlines
1489 @cindex escaped newlines
1490 @cindex newlines (escaped)
1492 Recently, the preprocessor has relaxed its treatment of escaped
1493 newlines. Previously, the newline had to immediately follow a
1494 backslash. The current implementation allows whitespace in the form
1495 of spaces, horizontal and vertical tabs, and form feeds between the
1496 backslash and the subsequent newline. The preprocessor issues a
1497 warning, but treats it as a valid escaped newline and combines the two
1498 lines to form a single logical line. This works within comments and
1499 tokens, as well as between tokens. Comments are @emph{not} treated as
1500 whitespace for the purposes of this relaxation, since they have not
1501 yet been replaced with spaces.
1504 @section Non-Lvalue Arrays May Have Subscripts
1505 @cindex subscripting
1506 @cindex arrays, non-lvalue
1508 @cindex subscripting and function values
1509 In ISO C99, arrays that are not lvalues still decay to pointers, and
1510 may be subscripted, although they may not be modified or used after
1511 the next sequence point and the unary @samp{&} operator may not be
1512 applied to them. As an extension, GCC allows such arrays to be
1513 subscripted in C89 mode, though otherwise they do not decay to
1514 pointers outside C99 mode. For example,
1515 this is valid in GNU C though not valid in C89:
1519 struct foo @{int a[4];@};
1525 return f().a[index];
1531 @section Arithmetic on @code{void}- and Function-Pointers
1532 @cindex void pointers, arithmetic
1533 @cindex void, size of pointer to
1534 @cindex function pointers, arithmetic
1535 @cindex function, size of pointer to
1537 In GNU C, addition and subtraction operations are supported on pointers to
1538 @code{void} and on pointers to functions. This is done by treating the
1539 size of a @code{void} or of a function as 1.
1541 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1542 and on function types, and returns 1.
1544 @opindex Wpointer-arith
1545 The option @option{-Wpointer-arith} requests a warning if these extensions
1549 @section Non-Constant Initializers
1550 @cindex initializers, non-constant
1551 @cindex non-constant initializers
1553 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1554 automatic variable are not required to be constant expressions in GNU C@.
1555 Here is an example of an initializer with run-time varying elements:
1558 foo (float f, float g)
1560 float beat_freqs[2] = @{ f-g, f+g @};
1565 @node Compound Literals
1566 @section Compound Literals
1567 @cindex constructor expressions
1568 @cindex initializations in expressions
1569 @cindex structures, constructor expression
1570 @cindex expressions, constructor
1571 @cindex compound literals
1572 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1574 ISO C99 supports compound literals. A compound literal looks like
1575 a cast containing an initializer. Its value is an object of the
1576 type specified in the cast, containing the elements specified in
1577 the initializer; it is an lvalue. As an extension, GCC supports
1578 compound literals in C89 mode and in C++.
1580 Usually, the specified type is a structure. Assume that
1581 @code{struct foo} and @code{structure} are declared as shown:
1584 struct foo @{int a; char b[2];@} structure;
1588 Here is an example of constructing a @code{struct foo} with a compound literal:
1591 structure = ((struct foo) @{x + y, 'a', 0@});
1595 This is equivalent to writing the following:
1599 struct foo temp = @{x + y, 'a', 0@};
1604 You can also construct an array. If all the elements of the compound literal
1605 are (made up of) simple constant expressions, suitable for use in
1606 initializers of objects of static storage duration, then the compound
1607 literal can be coerced to a pointer to its first element and used in
1608 such an initializer, as shown here:
1611 char **foo = (char *[]) @{ "x", "y", "z" @};
1614 Compound literals for scalar types and union types are is
1615 also allowed, but then the compound literal is equivalent
1618 As a GNU extension, GCC allows initialization of objects with static storage
1619 duration by compound literals (which is not possible in ISO C99, because
1620 the initializer is not a constant).
1621 It is handled as if the object was initialized only with the bracket
1622 enclosed list if the types of the compound literal and the object match.
1623 The initializer list of the compound literal must be constant.
1624 If the object being initialized has array type of unknown size, the size is
1625 determined by compound literal size.
1628 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1629 static int y[] = (int []) @{1, 2, 3@};
1630 static int z[] = (int [3]) @{1@};
1634 The above lines are equivalent to the following:
1636 static struct foo x = @{1, 'a', 'b'@};
1637 static int y[] = @{1, 2, 3@};
1638 static int z[] = @{1, 0, 0@};
1641 @node Designated Inits
1642 @section Designated Initializers
1643 @cindex initializers with labeled elements
1644 @cindex labeled elements in initializers
1645 @cindex case labels in initializers
1646 @cindex designated initializers
1648 Standard C89 requires the elements of an initializer to appear in a fixed
1649 order, the same as the order of the elements in the array or structure
1652 In ISO C99 you can give the elements in any order, specifying the array
1653 indices or structure field names they apply to, and GNU C allows this as
1654 an extension in C89 mode as well. This extension is not
1655 implemented in GNU C++.
1657 To specify an array index, write
1658 @samp{[@var{index}] =} before the element value. For example,
1661 int a[6] = @{ [4] = 29, [2] = 15 @};
1668 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1672 The index values must be constant expressions, even if the array being
1673 initialized is automatic.
1675 An alternative syntax for this which has been obsolete since GCC 2.5 but
1676 GCC still accepts is to write @samp{[@var{index}]} before the element
1677 value, with no @samp{=}.
1679 To initialize a range of elements to the same value, write
1680 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1681 extension. For example,
1684 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1688 If the value in it has side-effects, the side-effects will happen only once,
1689 not for each initialized field by the range initializer.
1692 Note that the length of the array is the highest value specified
1695 In a structure initializer, specify the name of a field to initialize
1696 with @samp{.@var{fieldname} =} before the element value. For example,
1697 given the following structure,
1700 struct point @{ int x, y; @};
1704 the following initialization
1707 struct point p = @{ .y = yvalue, .x = xvalue @};
1714 struct point p = @{ xvalue, yvalue @};
1717 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1718 @samp{@var{fieldname}:}, as shown here:
1721 struct point p = @{ y: yvalue, x: xvalue @};
1725 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1726 @dfn{designator}. You can also use a designator (or the obsolete colon
1727 syntax) when initializing a union, to specify which element of the union
1728 should be used. For example,
1731 union foo @{ int i; double d; @};
1733 union foo f = @{ .d = 4 @};
1737 will convert 4 to a @code{double} to store it in the union using
1738 the second element. By contrast, casting 4 to type @code{union foo}
1739 would store it into the union as the integer @code{i}, since it is
1740 an integer. (@xref{Cast to Union}.)
1742 You can combine this technique of naming elements with ordinary C
1743 initialization of successive elements. Each initializer element that
1744 does not have a designator applies to the next consecutive element of the
1745 array or structure. For example,
1748 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1755 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1758 Labeling the elements of an array initializer is especially useful
1759 when the indices are characters or belong to an @code{enum} type.
1764 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1765 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1768 @cindex designator lists
1769 You can also write a series of @samp{.@var{fieldname}} and
1770 @samp{[@var{index}]} designators before an @samp{=} to specify a
1771 nested subobject to initialize; the list is taken relative to the
1772 subobject corresponding to the closest surrounding brace pair. For
1773 example, with the @samp{struct point} declaration above:
1776 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1780 If the same field is initialized multiple times, it will have value from
1781 the last initialization. If any such overridden initialization has
1782 side-effect, it is unspecified whether the side-effect happens or not.
1783 Currently, GCC will discard them and issue a warning.
1786 @section Case Ranges
1788 @cindex ranges in case statements
1790 You can specify a range of consecutive values in a single @code{case} label,
1794 case @var{low} ... @var{high}:
1798 This has the same effect as the proper number of individual @code{case}
1799 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1801 This feature is especially useful for ranges of ASCII character codes:
1807 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1808 it may be parsed wrong when you use it with integer values. For example,
1823 @section Cast to a Union Type
1824 @cindex cast to a union
1825 @cindex union, casting to a
1827 A cast to union type is similar to other casts, except that the type
1828 specified is a union type. You can specify the type either with
1829 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1830 a constructor though, not a cast, and hence does not yield an lvalue like
1831 normal casts. (@xref{Compound Literals}.)
1833 The types that may be cast to the union type are those of the members
1834 of the union. Thus, given the following union and variables:
1837 union foo @{ int i; double d; @};
1843 both @code{x} and @code{y} can be cast to type @code{union foo}.
1845 Using the cast as the right-hand side of an assignment to a variable of
1846 union type is equivalent to storing in a member of the union:
1851 u = (union foo) x @equiv{} u.i = x
1852 u = (union foo) y @equiv{} u.d = y
1855 You can also use the union cast as a function argument:
1858 void hack (union foo);
1860 hack ((union foo) x);
1863 @node Mixed Declarations
1864 @section Mixed Declarations and Code
1865 @cindex mixed declarations and code
1866 @cindex declarations, mixed with code
1867 @cindex code, mixed with declarations
1869 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1870 within compound statements. As an extension, GCC also allows this in
1871 C89 mode. For example, you could do:
1880 Each identifier is visible from where it is declared until the end of
1881 the enclosing block.
1883 @node Function Attributes
1884 @section Declaring Attributes of Functions
1885 @cindex function attributes
1886 @cindex declaring attributes of functions
1887 @cindex functions that never return
1888 @cindex functions that return more than once
1889 @cindex functions that have no side effects
1890 @cindex functions in arbitrary sections
1891 @cindex functions that behave like malloc
1892 @cindex @code{volatile} applied to function
1893 @cindex @code{const} applied to function
1894 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1895 @cindex functions with non-null pointer arguments
1896 @cindex functions that are passed arguments in registers on the 386
1897 @cindex functions that pop the argument stack on the 386
1898 @cindex functions that do not pop the argument stack on the 386
1899 @cindex functions that have different compilation options on the 386
1900 @cindex functions that have different optimization options
1902 In GNU C, you declare certain things about functions called in your program
1903 which help the compiler optimize function calls and check your code more
1906 The keyword @code{__attribute__} allows you to specify special
1907 attributes when making a declaration. This keyword is followed by an
1908 attribute specification inside double parentheses. The following
1909 attributes are currently defined for functions on all targets:
1910 @code{aligned}, @code{alloc_size}, @code{noreturn},
1911 @code{returns_twice}, @code{noinline}, @code{noclone},
1912 @code{always_inline}, @code{flatten}, @code{pure}, @code{const},
1913 @code{nothrow}, @code{sentinel}, @code{format}, @code{format_arg},
1914 @code{no_instrument_function}, @code{section}, @code{constructor},
1915 @code{destructor}, @code{used}, @code{unused}, @code{deprecated},
1916 @code{weak}, @code{malloc}, @code{alias}, @code{warn_unused_result},
1917 @code{nonnull}, @code{gnu_inline}, @code{externally_visible},
1918 @code{hot}, @code{cold}, @code{artificial}, @code{error} and
1919 @code{warning}. Several other attributes are defined for functions on
1920 particular target systems. Other attributes, including @code{section}
1921 are supported for variables declarations (@pxref{Variable Attributes})
1922 and for types (@pxref{Type Attributes}).
1924 You may also specify attributes with @samp{__} preceding and following
1925 each keyword. This allows you to use them in header files without
1926 being concerned about a possible macro of the same name. For example,
1927 you may use @code{__noreturn__} instead of @code{noreturn}.
1929 @xref{Attribute Syntax}, for details of the exact syntax for using
1933 @c Keep this table alphabetized by attribute name. Treat _ as space.
1935 @item alias ("@var{target}")
1936 @cindex @code{alias} attribute
1937 The @code{alias} attribute causes the declaration to be emitted as an
1938 alias for another symbol, which must be specified. For instance,
1941 void __f () @{ /* @r{Do something.} */; @}
1942 void f () __attribute__ ((weak, alias ("__f")));
1945 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
1946 mangled name for the target must be used. It is an error if @samp{__f}
1947 is not defined in the same translation unit.
1949 Not all target machines support this attribute.
1951 @item aligned (@var{alignment})
1952 @cindex @code{aligned} attribute
1953 This attribute specifies a minimum alignment for the function,
1956 You cannot use this attribute to decrease the alignment of a function,
1957 only to increase it. However, when you explicitly specify a function
1958 alignment this will override the effect of the
1959 @option{-falign-functions} (@pxref{Optimize Options}) option for this
1962 Note that the effectiveness of @code{aligned} attributes may be
1963 limited by inherent limitations in your linker. On many systems, the
1964 linker is only able to arrange for functions to be aligned up to a
1965 certain maximum alignment. (For some linkers, the maximum supported
1966 alignment may be very very small.) See your linker documentation for
1967 further information.
1969 The @code{aligned} attribute can also be used for variables and fields
1970 (@pxref{Variable Attributes}.)
1973 @cindex @code{alloc_size} attribute
1974 The @code{alloc_size} attribute is used to tell the compiler that the
1975 function return value points to memory, where the size is given by
1976 one or two of the functions parameters. GCC uses this
1977 information to improve the correctness of @code{__builtin_object_size}.
1979 The function parameter(s) denoting the allocated size are specified by
1980 one or two integer arguments supplied to the attribute. The allocated size
1981 is either the value of the single function argument specified or the product
1982 of the two function arguments specified. Argument numbering starts at
1988 void* my_calloc(size_t, size_t) __attribute__((alloc_size(1,2)))
1989 void my_realloc(void*, size_t) __attribute__((alloc_size(2)))
1992 declares that my_calloc will return memory of the size given by
1993 the product of parameter 1 and 2 and that my_realloc will return memory
1994 of the size given by parameter 2.
1997 @cindex @code{always_inline} function attribute
1998 Generally, functions are not inlined unless optimization is specified.
1999 For functions declared inline, this attribute inlines the function even
2000 if no optimization level was specified.
2003 @cindex @code{gnu_inline} function attribute
2004 This attribute should be used with a function which is also declared
2005 with the @code{inline} keyword. It directs GCC to treat the function
2006 as if it were defined in gnu89 mode even when compiling in C99 or
2009 If the function is declared @code{extern}, then this definition of the
2010 function is used only for inlining. In no case is the function
2011 compiled as a standalone function, not even if you take its address
2012 explicitly. Such an address becomes an external reference, as if you
2013 had only declared the function, and had not defined it. This has
2014 almost the effect of a macro. The way to use this is to put a
2015 function definition in a header file with this attribute, and put
2016 another copy of the function, without @code{extern}, in a library
2017 file. The definition in the header file will cause most calls to the
2018 function to be inlined. If any uses of the function remain, they will
2019 refer to the single copy in the library. Note that the two
2020 definitions of the functions need not be precisely the same, although
2021 if they do not have the same effect your program may behave oddly.
2023 In C, if the function is neither @code{extern} nor @code{static}, then
2024 the function is compiled as a standalone function, as well as being
2025 inlined where possible.
2027 This is how GCC traditionally handled functions declared
2028 @code{inline}. Since ISO C99 specifies a different semantics for
2029 @code{inline}, this function attribute is provided as a transition
2030 measure and as a useful feature in its own right. This attribute is
2031 available in GCC 4.1.3 and later. It is available if either of the
2032 preprocessor macros @code{__GNUC_GNU_INLINE__} or
2033 @code{__GNUC_STDC_INLINE__} are defined. @xref{Inline,,An Inline
2034 Function is As Fast As a Macro}.
2036 In C++, this attribute does not depend on @code{extern} in any way,
2037 but it still requires the @code{inline} keyword to enable its special
2041 @cindex @code{artificial} function attribute
2042 This attribute is useful for small inline wrappers which if possible
2043 should appear during debugging as a unit, depending on the debug
2044 info format it will either mean marking the function as artificial
2045 or using the caller location for all instructions within the inlined
2049 @cindex interrupt handler functions
2050 When added to an interrupt handler with the M32C port, causes the
2051 prologue and epilogue to use bank switching to preserve the registers
2052 rather than saving them on the stack.
2055 @cindex @code{flatten} function attribute
2056 Generally, inlining into a function is limited. For a function marked with
2057 this attribute, every call inside this function will be inlined, if possible.
2058 Whether the function itself is considered for inlining depends on its size and
2059 the current inlining parameters.
2061 @item error ("@var{message}")
2062 @cindex @code{error} function attribute
2063 If this attribute is used on a function declaration and a call to such a function
2064 is not eliminated through dead code elimination or other optimizations, an error
2065 which will include @var{message} will be diagnosed. This is useful
2066 for compile time checking, especially together with @code{__builtin_constant_p}
2067 and inline functions where checking the inline function arguments is not
2068 possible through @code{extern char [(condition) ? 1 : -1];} tricks.
2069 While it is possible to leave the function undefined and thus invoke
2070 a link failure, when using this attribute the problem will be diagnosed
2071 earlier and with exact location of the call even in presence of inline
2072 functions or when not emitting debugging information.
2074 @item warning ("@var{message}")
2075 @cindex @code{warning} function attribute
2076 If this attribute is used on a function declaration and a call to such a function
2077 is not eliminated through dead code elimination or other optimizations, a warning
2078 which will include @var{message} will be diagnosed. This is useful
2079 for compile time checking, especially together with @code{__builtin_constant_p}
2080 and inline functions. While it is possible to define the function with
2081 a message in @code{.gnu.warning*} section, when using this attribute the problem
2082 will be diagnosed earlier and with exact location of the call even in presence
2083 of inline functions or when not emitting debugging information.
2086 @cindex functions that do pop the argument stack on the 386
2088 On the Intel 386, the @code{cdecl} attribute causes the compiler to
2089 assume that the calling function will pop off the stack space used to
2090 pass arguments. This is
2091 useful to override the effects of the @option{-mrtd} switch.
2094 @cindex @code{const} function attribute
2095 Many functions do not examine any values except their arguments, and
2096 have no effects except the return value. Basically this is just slightly
2097 more strict class than the @code{pure} attribute below, since function is not
2098 allowed to read global memory.
2100 @cindex pointer arguments
2101 Note that a function that has pointer arguments and examines the data
2102 pointed to must @emph{not} be declared @code{const}. Likewise, a
2103 function that calls a non-@code{const} function usually must not be
2104 @code{const}. It does not make sense for a @code{const} function to
2107 The attribute @code{const} is not implemented in GCC versions earlier
2108 than 2.5. An alternative way to declare that a function has no side
2109 effects, which works in the current version and in some older versions,
2113 typedef int intfn ();
2115 extern const intfn square;
2118 This approach does not work in GNU C++ from 2.6.0 on, since the language
2119 specifies that the @samp{const} must be attached to the return value.
2123 @itemx constructor (@var{priority})
2124 @itemx destructor (@var{priority})
2125 @cindex @code{constructor} function attribute
2126 @cindex @code{destructor} function attribute
2127 The @code{constructor} attribute causes the function to be called
2128 automatically before execution enters @code{main ()}. Similarly, the
2129 @code{destructor} attribute causes the function to be called
2130 automatically after @code{main ()} has completed or @code{exit ()} has
2131 been called. Functions with these attributes are useful for
2132 initializing data that will be used implicitly during the execution of
2135 You may provide an optional integer priority to control the order in
2136 which constructor and destructor functions are run. A constructor
2137 with a smaller priority number runs before a constructor with a larger
2138 priority number; the opposite relationship holds for destructors. So,
2139 if you have a constructor that allocates a resource and a destructor
2140 that deallocates the same resource, both functions typically have the
2141 same priority. The priorities for constructor and destructor
2142 functions are the same as those specified for namespace-scope C++
2143 objects (@pxref{C++ Attributes}).
2145 These attributes are not currently implemented for Objective-C@.
2148 @itemx deprecated (@var{msg})
2149 @cindex @code{deprecated} attribute.
2150 The @code{deprecated} attribute results in a warning if the function
2151 is used anywhere in the source file. This is useful when identifying
2152 functions that are expected to be removed in a future version of a
2153 program. The warning also includes the location of the declaration
2154 of the deprecated function, to enable users to easily find further
2155 information about why the function is deprecated, or what they should
2156 do instead. Note that the warnings only occurs for uses:
2159 int old_fn () __attribute__ ((deprecated));
2161 int (*fn_ptr)() = old_fn;
2164 results in a warning on line 3 but not line 2. The optional msg
2165 argument, which must be a string, will be printed in the warning if
2168 The @code{deprecated} attribute can also be used for variables and
2169 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2172 @cindex @code{disinterrupt} attribute
2173 On MeP targets, this attribute causes the compiler to emit
2174 instructions to disable interrupts for the duration of the given
2178 @cindex @code{__declspec(dllexport)}
2179 On Microsoft Windows targets and Symbian OS targets the
2180 @code{dllexport} attribute causes the compiler to provide a global
2181 pointer to a pointer in a DLL, so that it can be referenced with the
2182 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
2183 name is formed by combining @code{_imp__} and the function or variable
2186 You can use @code{__declspec(dllexport)} as a synonym for
2187 @code{__attribute__ ((dllexport))} for compatibility with other
2190 On systems that support the @code{visibility} attribute, this
2191 attribute also implies ``default'' visibility. It is an error to
2192 explicitly specify any other visibility.
2194 Currently, the @code{dllexport} attribute is ignored for inlined
2195 functions, unless the @option{-fkeep-inline-functions} flag has been
2196 used. The attribute is also ignored for undefined symbols.
2198 When applied to C++ classes, the attribute marks defined non-inlined
2199 member functions and static data members as exports. Static consts
2200 initialized in-class are not marked unless they are also defined
2203 For Microsoft Windows targets there are alternative methods for
2204 including the symbol in the DLL's export table such as using a
2205 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
2206 the @option{--export-all} linker flag.
2209 @cindex @code{__declspec(dllimport)}
2210 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
2211 attribute causes the compiler to reference a function or variable via
2212 a global pointer to a pointer that is set up by the DLL exporting the
2213 symbol. The attribute implies @code{extern}. On Microsoft Windows
2214 targets, the pointer name is formed by combining @code{_imp__} and the
2215 function or variable name.
2217 You can use @code{__declspec(dllimport)} as a synonym for
2218 @code{__attribute__ ((dllimport))} for compatibility with other
2221 On systems that support the @code{visibility} attribute, this
2222 attribute also implies ``default'' visibility. It is an error to
2223 explicitly specify any other visibility.
2225 Currently, the attribute is ignored for inlined functions. If the
2226 attribute is applied to a symbol @emph{definition}, an error is reported.
2227 If a symbol previously declared @code{dllimport} is later defined, the
2228 attribute is ignored in subsequent references, and a warning is emitted.
2229 The attribute is also overridden by a subsequent declaration as
2232 When applied to C++ classes, the attribute marks non-inlined
2233 member functions and static data members as imports. However, the
2234 attribute is ignored for virtual methods to allow creation of vtables
2237 On the SH Symbian OS target the @code{dllimport} attribute also has
2238 another affect---it can cause the vtable and run-time type information
2239 for a class to be exported. This happens when the class has a
2240 dllimport'ed constructor or a non-inline, non-pure virtual function
2241 and, for either of those two conditions, the class also has an inline
2242 constructor or destructor and has a key function that is defined in
2243 the current translation unit.
2245 For Microsoft Windows based targets the use of the @code{dllimport}
2246 attribute on functions is not necessary, but provides a small
2247 performance benefit by eliminating a thunk in the DLL@. The use of the
2248 @code{dllimport} attribute on imported variables was required on older
2249 versions of the GNU linker, but can now be avoided by passing the
2250 @option{--enable-auto-import} switch to the GNU linker. As with
2251 functions, using the attribute for a variable eliminates a thunk in
2254 One drawback to using this attribute is that a pointer to a
2255 @emph{variable} marked as @code{dllimport} cannot be used as a constant
2256 address. However, a pointer to a @emph{function} with the
2257 @code{dllimport} attribute can be used as a constant initializer; in
2258 this case, the address of a stub function in the import lib is
2259 referenced. On Microsoft Windows targets, the attribute can be disabled
2260 for functions by setting the @option{-mnop-fun-dllimport} flag.
2263 @cindex eight bit data on the H8/300, H8/300H, and H8S
2264 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2265 variable should be placed into the eight bit data section.
2266 The compiler will generate more efficient code for certain operations
2267 on data in the eight bit data area. Note the eight bit data area is limited to
2270 You must use GAS and GLD from GNU binutils version 2.7 or later for
2271 this attribute to work correctly.
2273 @item exception_handler
2274 @cindex exception handler functions on the Blackfin processor
2275 Use this attribute on the Blackfin to indicate that the specified function
2276 is an exception handler. The compiler will generate function entry and
2277 exit sequences suitable for use in an exception handler when this
2278 attribute is present.
2280 @item externally_visible
2281 @cindex @code{externally_visible} attribute.
2282 This attribute, attached to a global variable or function, nullifies
2283 the effect of the @option{-fwhole-program} command-line option, so the
2284 object remains visible outside the current compilation unit.
2287 @cindex functions which handle memory bank switching
2288 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
2289 use a calling convention that takes care of switching memory banks when
2290 entering and leaving a function. This calling convention is also the
2291 default when using the @option{-mlong-calls} option.
2293 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
2294 to call and return from a function.
2296 On 68HC11 the compiler will generate a sequence of instructions
2297 to invoke a board-specific routine to switch the memory bank and call the
2298 real function. The board-specific routine simulates a @code{call}.
2299 At the end of a function, it will jump to a board-specific routine
2300 instead of using @code{rts}. The board-specific return routine simulates
2303 On MeP targets this causes the compiler to use a calling convention
2304 which assumes the called function is too far away for the built-in
2307 @item fast_interrupt
2308 @cindex interrupt handler functions
2309 Use this attribute on the M32C and RX ports to indicate that the specified
2310 function is a fast interrupt handler. This is just like the
2311 @code{interrupt} attribute, except that @code{freit} is used to return
2312 instead of @code{reit}.
2315 @cindex functions that pop the argument stack on the 386
2316 On the Intel 386, the @code{fastcall} attribute causes the compiler to
2317 pass the first argument (if of integral type) in the register ECX and
2318 the second argument (if of integral type) in the register EDX@. Subsequent
2319 and other typed arguments are passed on the stack. The called function will
2320 pop the arguments off the stack. If the number of arguments is variable all
2321 arguments are pushed on the stack.
2323 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2324 @cindex @code{format} function attribute
2326 The @code{format} attribute specifies that a function takes @code{printf},
2327 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
2328 should be type-checked against a format string. For example, the
2333 my_printf (void *my_object, const char *my_format, ...)
2334 __attribute__ ((format (printf, 2, 3)));
2338 causes the compiler to check the arguments in calls to @code{my_printf}
2339 for consistency with the @code{printf} style format string argument
2342 The parameter @var{archetype} determines how the format string is
2343 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime},
2344 @code{gnu_printf}, @code{gnu_scanf}, @code{gnu_strftime} or
2345 @code{strfmon}. (You can also use @code{__printf__},
2346 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) On
2347 MinGW targets, @code{ms_printf}, @code{ms_scanf}, and
2348 @code{ms_strftime} are also present.
2349 @var{archtype} values such as @code{printf} refer to the formats accepted
2350 by the system's C run-time library, while @code{gnu_} values always refer
2351 to the formats accepted by the GNU C Library. On Microsoft Windows
2352 targets, @code{ms_} values refer to the formats accepted by the
2353 @file{msvcrt.dll} library.
2354 The parameter @var{string-index}
2355 specifies which argument is the format string argument (starting
2356 from 1), while @var{first-to-check} is the number of the first
2357 argument to check against the format string. For functions
2358 where the arguments are not available to be checked (such as
2359 @code{vprintf}), specify the third parameter as zero. In this case the
2360 compiler only checks the format string for consistency. For
2361 @code{strftime} formats, the third parameter is required to be zero.
2362 Since non-static C++ methods have an implicit @code{this} argument, the
2363 arguments of such methods should be counted from two, not one, when
2364 giving values for @var{string-index} and @var{first-to-check}.
2366 In the example above, the format string (@code{my_format}) is the second
2367 argument of the function @code{my_print}, and the arguments to check
2368 start with the third argument, so the correct parameters for the format
2369 attribute are 2 and 3.
2371 @opindex ffreestanding
2372 @opindex fno-builtin
2373 The @code{format} attribute allows you to identify your own functions
2374 which take format strings as arguments, so that GCC can check the
2375 calls to these functions for errors. The compiler always (unless
2376 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
2377 for the standard library functions @code{printf}, @code{fprintf},
2378 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2379 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2380 warnings are requested (using @option{-Wformat}), so there is no need to
2381 modify the header file @file{stdio.h}. In C99 mode, the functions
2382 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2383 @code{vsscanf} are also checked. Except in strictly conforming C
2384 standard modes, the X/Open function @code{strfmon} is also checked as
2385 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2386 @xref{C Dialect Options,,Options Controlling C Dialect}.
2388 The target may provide additional types of format checks.
2389 @xref{Target Format Checks,,Format Checks Specific to Particular
2392 @item format_arg (@var{string-index})
2393 @cindex @code{format_arg} function attribute
2394 @opindex Wformat-nonliteral
2395 The @code{format_arg} attribute specifies that a function takes a format
2396 string for a @code{printf}, @code{scanf}, @code{strftime} or
2397 @code{strfmon} style function and modifies it (for example, to translate
2398 it into another language), so the result can be passed to a
2399 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2400 function (with the remaining arguments to the format function the same
2401 as they would have been for the unmodified string). For example, the
2406 my_dgettext (char *my_domain, const char *my_format)
2407 __attribute__ ((format_arg (2)));
2411 causes the compiler to check the arguments in calls to a @code{printf},
2412 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2413 format string argument is a call to the @code{my_dgettext} function, for
2414 consistency with the format string argument @code{my_format}. If the
2415 @code{format_arg} attribute had not been specified, all the compiler
2416 could tell in such calls to format functions would be that the format
2417 string argument is not constant; this would generate a warning when
2418 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2419 without the attribute.
2421 The parameter @var{string-index} specifies which argument is the format
2422 string argument (starting from one). Since non-static C++ methods have
2423 an implicit @code{this} argument, the arguments of such methods should
2424 be counted from two.
2426 The @code{format-arg} attribute allows you to identify your own
2427 functions which modify format strings, so that GCC can check the
2428 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2429 type function whose operands are a call to one of your own function.
2430 The compiler always treats @code{gettext}, @code{dgettext}, and
2431 @code{dcgettext} in this manner except when strict ISO C support is
2432 requested by @option{-ansi} or an appropriate @option{-std} option, or
2433 @option{-ffreestanding} or @option{-fno-builtin}
2434 is used. @xref{C Dialect Options,,Options
2435 Controlling C Dialect}.
2437 @item function_vector
2438 @cindex calling functions through the function vector on H8/300, M16C, M32C and SH2A processors
2439 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2440 function should be called through the function vector. Calling a
2441 function through the function vector will reduce code size, however;
2442 the function vector has a limited size (maximum 128 entries on the H8/300
2443 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2445 In SH2A target, this attribute declares a function to be called using the
2446 TBR relative addressing mode. The argument to this attribute is the entry
2447 number of the same function in a vector table containing all the TBR
2448 relative addressable functions. For the successful jump, register TBR
2449 should contain the start address of this TBR relative vector table.
2450 In the startup routine of the user application, user needs to care of this
2451 TBR register initialization. The TBR relative vector table can have at
2452 max 256 function entries. The jumps to these functions will be generated
2453 using a SH2A specific, non delayed branch instruction JSR/N @@(disp8,TBR).
2454 You must use GAS and GLD from GNU binutils version 2.7 or later for
2455 this attribute to work correctly.
2457 Please refer the example of M16C target, to see the use of this
2458 attribute while declaring a function,
2460 In an application, for a function being called once, this attribute will
2461 save at least 8 bytes of code; and if other successive calls are being
2462 made to the same function, it will save 2 bytes of code per each of these
2465 On M16C/M32C targets, the @code{function_vector} attribute declares a
2466 special page subroutine call function. Use of this attribute reduces
2467 the code size by 2 bytes for each call generated to the
2468 subroutine. The argument to the attribute is the vector number entry
2469 from the special page vector table which contains the 16 low-order
2470 bits of the subroutine's entry address. Each vector table has special
2471 page number (18 to 255) which are used in @code{jsrs} instruction.
2472 Jump addresses of the routines are generated by adding 0x0F0000 (in
2473 case of M16C targets) or 0xFF0000 (in case of M32C targets), to the 2
2474 byte addresses set in the vector table. Therefore you need to ensure
2475 that all the special page vector routines should get mapped within the
2476 address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF
2479 In the following example 2 bytes will be saved for each call to
2480 function @code{foo}.
2483 void foo (void) __attribute__((function_vector(0x18)));
2494 If functions are defined in one file and are called in another file,
2495 then be sure to write this declaration in both files.
2497 This attribute is ignored for R8C target.
2500 @cindex interrupt handler functions
2501 Use this attribute on the ARM, AVR, CRX, M32C, M32R/D, m68k, MeP, MIPS,
2502 RX and Xstormy16 ports to indicate that the specified function is an
2503 interrupt handler. The compiler will generate function entry and exit
2504 sequences suitable for use in an interrupt handler when this attribute
2507 Note, interrupt handlers for the Blackfin, H8/300, H8/300H, H8S, and
2508 SH processors can be specified via the @code{interrupt_handler} attribute.
2510 Note, on the AVR, interrupts will be enabled inside the function.
2512 Note, for the ARM, you can specify the kind of interrupt to be handled by
2513 adding an optional parameter to the interrupt attribute like this:
2516 void f () __attribute__ ((interrupt ("IRQ")));
2519 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2521 On ARMv7-M the interrupt type is ignored, and the attribute means the function
2522 may be called with a word aligned stack pointer.
2524 On MIPS targets, you can use the following attributes to modify the behavior
2525 of an interrupt handler:
2527 @item use_shadow_register_set
2528 @cindex @code{use_shadow_register_set} attribute
2529 Assume that the handler uses a shadow register set, instead of
2530 the main general-purpose registers.
2532 @item keep_interrupts_masked
2533 @cindex @code{keep_interrupts_masked} attribute
2534 Keep interrupts masked for the whole function. Without this attribute,
2535 GCC tries to reenable interrupts for as much of the function as it can.
2537 @item use_debug_exception_return
2538 @cindex @code{use_debug_exception_return} attribute
2539 Return using the @code{deret} instruction. Interrupt handlers that don't
2540 have this attribute return using @code{eret} instead.
2543 You can use any combination of these attributes, as shown below:
2545 void __attribute__ ((interrupt)) v0 ();
2546 void __attribute__ ((interrupt, use_shadow_register_set)) v1 ();
2547 void __attribute__ ((interrupt, keep_interrupts_masked)) v2 ();
2548 void __attribute__ ((interrupt, use_debug_exception_return)) v3 ();
2549 void __attribute__ ((interrupt, use_shadow_register_set,
2550 keep_interrupts_masked)) v4 ();
2551 void __attribute__ ((interrupt, use_shadow_register_set,
2552 use_debug_exception_return)) v5 ();
2553 void __attribute__ ((interrupt, keep_interrupts_masked,
2554 use_debug_exception_return)) v6 ();
2555 void __attribute__ ((interrupt, use_shadow_register_set,
2556 keep_interrupts_masked,
2557 use_debug_exception_return)) v7 ();
2560 @item interrupt_handler
2561 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
2562 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
2563 indicate that the specified function is an interrupt handler. The compiler
2564 will generate function entry and exit sequences suitable for use in an
2565 interrupt handler when this attribute is present.
2567 @item interrupt_thread
2568 @cindex interrupt thread functions on fido
2569 Use this attribute on fido, a subarchitecture of the m68k, to indicate
2570 that the specified function is an interrupt handler that is designed
2571 to run as a thread. The compiler omits generate prologue/epilogue
2572 sequences and replaces the return instruction with a @code{sleep}
2573 instruction. This attribute is available only on fido.
2576 @cindex interrupt service routines on ARM
2577 Use this attribute on ARM to write Interrupt Service Routines. This is an
2578 alias to the @code{interrupt} attribute above.
2581 @cindex User stack pointer in interrupts on the Blackfin
2582 When used together with @code{interrupt_handler}, @code{exception_handler}
2583 or @code{nmi_handler}, code will be generated to load the stack pointer
2584 from the USP register in the function prologue.
2587 @cindex @code{l1_text} function attribute
2588 This attribute specifies a function to be placed into L1 Instruction
2589 SRAM@. The function will be put into a specific section named @code{.l1.text}.
2590 With @option{-mfdpic}, function calls with a such function as the callee
2591 or caller will use inlined PLT.
2594 @cindex @code{l2} function attribute
2595 On the Blackfin, this attribute specifies a function to be placed into L2
2596 SRAM. The function will be put into a specific section named
2597 @code{.l1.text}. With @option{-mfdpic}, callers of such functions will use
2600 @item long_call/short_call
2601 @cindex indirect calls on ARM
2602 This attribute specifies how a particular function is called on
2603 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2604 command line switch and @code{#pragma long_calls} settings. The
2605 @code{long_call} attribute indicates that the function might be far
2606 away from the call site and require a different (more expensive)
2607 calling sequence. The @code{short_call} attribute always places
2608 the offset to the function from the call site into the @samp{BL}
2609 instruction directly.
2611 @item longcall/shortcall
2612 @cindex functions called via pointer on the RS/6000 and PowerPC
2613 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
2614 indicates that the function might be far away from the call site and
2615 require a different (more expensive) calling sequence. The
2616 @code{shortcall} attribute indicates that the function is always close
2617 enough for the shorter calling sequence to be used. These attributes
2618 override both the @option{-mlongcall} switch and, on the RS/6000 and
2619 PowerPC, the @code{#pragma longcall} setting.
2621 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2622 calls are necessary.
2624 @item long_call/near/far
2625 @cindex indirect calls on MIPS
2626 These attributes specify how a particular function is called on MIPS@.
2627 The attributes override the @option{-mlong-calls} (@pxref{MIPS Options})
2628 command-line switch. The @code{long_call} and @code{far} attributes are
2629 synonyms, and cause the compiler to always call
2630 the function by first loading its address into a register, and then using
2631 the contents of that register. The @code{near} attribute has the opposite
2632 effect; it specifies that non-PIC calls should be made using the more
2633 efficient @code{jal} instruction.
2636 @cindex @code{malloc} attribute
2637 The @code{malloc} attribute is used to tell the compiler that a function
2638 may be treated as if any non-@code{NULL} pointer it returns cannot
2639 alias any other pointer valid when the function returns.
2640 This will often improve optimization.
2641 Standard functions with this property include @code{malloc} and
2642 @code{calloc}. @code{realloc}-like functions have this property as
2643 long as the old pointer is never referred to (including comparing it
2644 to the new pointer) after the function returns a non-@code{NULL}
2647 @item mips16/nomips16
2648 @cindex @code{mips16} attribute
2649 @cindex @code{nomips16} attribute
2651 On MIPS targets, you can use the @code{mips16} and @code{nomips16}
2652 function attributes to locally select or turn off MIPS16 code generation.
2653 A function with the @code{mips16} attribute is emitted as MIPS16 code,
2654 while MIPS16 code generation is disabled for functions with the
2655 @code{nomips16} attribute. These attributes override the
2656 @option{-mips16} and @option{-mno-mips16} options on the command line
2657 (@pxref{MIPS Options}).
2659 When compiling files containing mixed MIPS16 and non-MIPS16 code, the
2660 preprocessor symbol @code{__mips16} reflects the setting on the command line,
2661 not that within individual functions. Mixed MIPS16 and non-MIPS16 code
2662 may interact badly with some GCC extensions such as @code{__builtin_apply}
2663 (@pxref{Constructing Calls}).
2665 @item model (@var{model-name})
2666 @cindex function addressability on the M32R/D
2667 @cindex variable addressability on the IA-64
2669 On the M32R/D, use this attribute to set the addressability of an
2670 object, and of the code generated for a function. The identifier
2671 @var{model-name} is one of @code{small}, @code{medium}, or
2672 @code{large}, representing each of the code models.
2674 Small model objects live in the lower 16MB of memory (so that their
2675 addresses can be loaded with the @code{ld24} instruction), and are
2676 callable with the @code{bl} instruction.
2678 Medium model objects may live anywhere in the 32-bit address space (the
2679 compiler will generate @code{seth/add3} instructions to load their addresses),
2680 and are callable with the @code{bl} instruction.
2682 Large model objects may live anywhere in the 32-bit address space (the
2683 compiler will generate @code{seth/add3} instructions to load their addresses),
2684 and may not be reachable with the @code{bl} instruction (the compiler will
2685 generate the much slower @code{seth/add3/jl} instruction sequence).
2687 On IA-64, use this attribute to set the addressability of an object.
2688 At present, the only supported identifier for @var{model-name} is
2689 @code{small}, indicating addressability via ``small'' (22-bit)
2690 addresses (so that their addresses can be loaded with the @code{addl}
2691 instruction). Caveat: such addressing is by definition not position
2692 independent and hence this attribute must not be used for objects
2693 defined by shared libraries.
2695 @item ms_abi/sysv_abi
2696 @cindex @code{ms_abi} attribute
2697 @cindex @code{sysv_abi} attribute
2699 On 64-bit x86_64-*-* targets, you can use an ABI attribute to indicate
2700 which calling convention should be used for a function. The @code{ms_abi}
2701 attribute tells the compiler to use the Microsoft ABI, while the
2702 @code{sysv_abi} attribute tells the compiler to use the ABI used on
2703 GNU/Linux and other systems. The default is to use the Microsoft ABI
2704 when targeting Windows. On all other systems, the default is the AMD ABI.
2706 Note, This feature is currently sorried out for Windows targets trying to
2708 @item ms_hook_prologue
2709 @cindex @code{ms_hook_prologue} attribute
2711 On 32 bit i[34567]86-*-* targets, you can use this function attribute to make
2712 gcc generate the "hot-patching" function prologue used in Win32 API
2713 functions in Microsoft Windows XP Service Pack 2 and newer. This requires
2714 support for the swap suffix in the assembler. (GNU Binutils 2.19.51 or later)
2717 @cindex function without a prologue/epilogue code
2718 Use this attribute on the ARM, AVR, IP2K, RX and SPU ports to indicate that
2719 the specified function does not need prologue/epilogue sequences generated by
2720 the compiler. It is up to the programmer to provide these sequences. The
2721 only statements that can be safely included in naked functions are
2722 @code{asm} statements that do not have operands. All other statements,
2723 including declarations of local variables, @code{if} statements, and so
2724 forth, should be avoided. Naked functions should be used to implement the
2725 body of an assembly function, while allowing the compiler to construct
2726 the requisite function declaration for the assembler.
2729 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2730 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2731 use the normal calling convention based on @code{jsr} and @code{rts}.
2732 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2735 On MeP targets this attribute causes the compiler to assume the called
2736 function is close enough to use the normal calling convention,
2737 overriding the @code{-mtf} command line option.
2740 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2741 Use this attribute together with @code{interrupt_handler},
2742 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2743 entry code should enable nested interrupts or exceptions.
2746 @cindex NMI handler functions on the Blackfin processor
2747 Use this attribute on the Blackfin to indicate that the specified function
2748 is an NMI handler. The compiler will generate function entry and
2749 exit sequences suitable for use in an NMI handler when this
2750 attribute is present.
2752 @item no_instrument_function
2753 @cindex @code{no_instrument_function} function attribute
2754 @opindex finstrument-functions
2755 If @option{-finstrument-functions} is given, profiling function calls will
2756 be generated at entry and exit of most user-compiled functions.
2757 Functions with this attribute will not be so instrumented.
2760 @cindex @code{noinline} function attribute
2761 This function attribute prevents a function from being considered for
2763 @c Don't enumerate the optimizations by name here; we try to be
2764 @c future-compatible with this mechanism.
2765 If the function does not have side-effects, there are optimizations
2766 other than inlining that causes function calls to be optimized away,
2767 although the function call is live. To keep such calls from being
2772 (@pxref{Extended Asm}) in the called function, to serve as a special
2776 @cindex @code{noclone} function attribute
2777 This function attribute prevents a function from being considered for
2778 cloning - a mechanism which produces specialized copies of functions
2779 and which is (currently) performed by interprocedural constant
2782 @item nonnull (@var{arg-index}, @dots{})
2783 @cindex @code{nonnull} function attribute
2784 The @code{nonnull} attribute specifies that some function parameters should
2785 be non-null pointers. For instance, the declaration:
2789 my_memcpy (void *dest, const void *src, size_t len)
2790 __attribute__((nonnull (1, 2)));
2794 causes the compiler to check that, in calls to @code{my_memcpy},
2795 arguments @var{dest} and @var{src} are non-null. If the compiler
2796 determines that a null pointer is passed in an argument slot marked
2797 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2798 is issued. The compiler may also choose to make optimizations based
2799 on the knowledge that certain function arguments will not be null.
2801 If no argument index list is given to the @code{nonnull} attribute,
2802 all pointer arguments are marked as non-null. To illustrate, the
2803 following declaration is equivalent to the previous example:
2807 my_memcpy (void *dest, const void *src, size_t len)
2808 __attribute__((nonnull));
2812 @cindex @code{noreturn} function attribute
2813 A few standard library functions, such as @code{abort} and @code{exit},
2814 cannot return. GCC knows this automatically. Some programs define
2815 their own functions that never return. You can declare them
2816 @code{noreturn} to tell the compiler this fact. For example,
2820 void fatal () __attribute__ ((noreturn));
2823 fatal (/* @r{@dots{}} */)
2825 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2831 The @code{noreturn} keyword tells the compiler to assume that
2832 @code{fatal} cannot return. It can then optimize without regard to what
2833 would happen if @code{fatal} ever did return. This makes slightly
2834 better code. More importantly, it helps avoid spurious warnings of
2835 uninitialized variables.
2837 The @code{noreturn} keyword does not affect the exceptional path when that
2838 applies: a @code{noreturn}-marked function may still return to the caller
2839 by throwing an exception or calling @code{longjmp}.
2841 Do not assume that registers saved by the calling function are
2842 restored before calling the @code{noreturn} function.
2844 It does not make sense for a @code{noreturn} function to have a return
2845 type other than @code{void}.
2847 The attribute @code{noreturn} is not implemented in GCC versions
2848 earlier than 2.5. An alternative way to declare that a function does
2849 not return, which works in the current version and in some older
2850 versions, is as follows:
2853 typedef void voidfn ();
2855 volatile voidfn fatal;
2858 This approach does not work in GNU C++.
2861 @cindex @code{nothrow} function attribute
2862 The @code{nothrow} attribute is used to inform the compiler that a
2863 function cannot throw an exception. For example, most functions in
2864 the standard C library can be guaranteed not to throw an exception
2865 with the notable exceptions of @code{qsort} and @code{bsearch} that
2866 take function pointer arguments. The @code{nothrow} attribute is not
2867 implemented in GCC versions earlier than 3.3.
2870 @cindex @code{optimize} function attribute
2871 The @code{optimize} attribute is used to specify that a function is to
2872 be compiled with different optimization options than specified on the
2873 command line. Arguments can either be numbers or strings. Numbers
2874 are assumed to be an optimization level. Strings that begin with
2875 @code{O} are assumed to be an optimization option, while other options
2876 are assumed to be used with a @code{-f} prefix. You can also use the
2877 @samp{#pragma GCC optimize} pragma to set the optimization options
2878 that affect more than one function.
2879 @xref{Function Specific Option Pragmas}, for details about the
2880 @samp{#pragma GCC optimize} pragma.
2882 This can be used for instance to have frequently executed functions
2883 compiled with more aggressive optimization options that produce faster
2884 and larger code, while other functions can be called with less
2888 @cindex @code{pcs} function attribute
2890 The @code{pcs} attribute can be used to control the calling convention
2891 used for a function on ARM. The attribute takes an argument that specifies
2892 the calling convention to use.
2894 When compiling using the AAPCS ABI (or a variant of that) then valid
2895 values for the argument are @code{"aapcs"} and @code{"aapcs-vfp"}. In
2896 order to use a variant other than @code{"aapcs"} then the compiler must
2897 be permitted to use the appropriate co-processor registers (i.e., the
2898 VFP registers must be available in order to use @code{"aapcs-vfp"}).
2902 /* Argument passed in r0, and result returned in r0+r1. */
2903 double f2d (float) __attribute__((pcs("aapcs")));
2906 Variadic functions always use the @code{"aapcs"} calling convention and
2907 the compiler will reject attempts to specify an alternative.
2910 @cindex @code{pure} function attribute
2911 Many functions have no effects except the return value and their
2912 return value depends only on the parameters and/or global variables.
2913 Such a function can be subject
2914 to common subexpression elimination and loop optimization just as an
2915 arithmetic operator would be. These functions should be declared
2916 with the attribute @code{pure}. For example,
2919 int square (int) __attribute__ ((pure));
2923 says that the hypothetical function @code{square} is safe to call
2924 fewer times than the program says.
2926 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2927 Interesting non-pure functions are functions with infinite loops or those
2928 depending on volatile memory or other system resource, that may change between
2929 two consecutive calls (such as @code{feof} in a multithreading environment).
2931 The attribute @code{pure} is not implemented in GCC versions earlier
2935 @cindex @code{hot} function attribute
2936 The @code{hot} attribute is used to inform the compiler that a function is a
2937 hot spot of the compiled program. The function is optimized more aggressively
2938 and on many target it is placed into special subsection of the text section so
2939 all hot functions appears close together improving locality.
2941 When profile feedback is available, via @option{-fprofile-use}, hot functions
2942 are automatically detected and this attribute is ignored.
2944 The @code{hot} attribute is not implemented in GCC versions earlier
2948 @cindex @code{cold} function attribute
2949 The @code{cold} attribute is used to inform the compiler that a function is
2950 unlikely executed. The function is optimized for size rather than speed and on
2951 many targets it is placed into special subsection of the text section so all
2952 cold functions appears close together improving code locality of non-cold parts
2953 of program. The paths leading to call of cold functions within code are marked
2954 as unlikely by the branch prediction mechanism. It is thus useful to mark
2955 functions used to handle unlikely conditions, such as @code{perror}, as cold to
2956 improve optimization of hot functions that do call marked functions in rare
2959 When profile feedback is available, via @option{-fprofile-use}, hot functions
2960 are automatically detected and this attribute is ignored.
2962 The @code{cold} attribute is not implemented in GCC versions earlier than 4.3.
2964 @item regparm (@var{number})
2965 @cindex @code{regparm} attribute
2966 @cindex functions that are passed arguments in registers on the 386
2967 On the Intel 386, the @code{regparm} attribute causes the compiler to
2968 pass arguments number one to @var{number} if they are of integral type
2969 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2970 take a variable number of arguments will continue to be passed all of their
2971 arguments on the stack.
2973 Beware that on some ELF systems this attribute is unsuitable for
2974 global functions in shared libraries with lazy binding (which is the
2975 default). Lazy binding will send the first call via resolving code in
2976 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2977 per the standard calling conventions. Solaris 8 is affected by this.
2978 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2979 safe since the loaders there save EAX, EDX and ECX. (Lazy binding can be
2980 disabled with the linker or the loader if desired, to avoid the
2984 @cindex @code{sseregparm} attribute
2985 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2986 causes the compiler to pass up to 3 floating point arguments in
2987 SSE registers instead of on the stack. Functions that take a
2988 variable number of arguments will continue to pass all of their
2989 floating point arguments on the stack.
2991 @item force_align_arg_pointer
2992 @cindex @code{force_align_arg_pointer} attribute
2993 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
2994 applied to individual function definitions, generating an alternate
2995 prologue and epilogue that realigns the runtime stack if necessary.
2996 This supports mixing legacy codes that run with a 4-byte aligned stack
2997 with modern codes that keep a 16-byte stack for SSE compatibility.
3000 @cindex @code{resbank} attribute
3001 On the SH2A target, this attribute enables the high-speed register
3002 saving and restoration using a register bank for @code{interrupt_handler}
3003 routines. Saving to the bank is performed automatically after the CPU
3004 accepts an interrupt that uses a register bank.
3006 The nineteen 32-bit registers comprising general register R0 to R14,
3007 control register GBR, and system registers MACH, MACL, and PR and the
3008 vector table address offset are saved into a register bank. Register
3009 banks are stacked in first-in last-out (FILO) sequence. Restoration
3010 from the bank is executed by issuing a RESBANK instruction.
3013 @cindex @code{returns_twice} attribute
3014 The @code{returns_twice} attribute tells the compiler that a function may
3015 return more than one time. The compiler will ensure that all registers
3016 are dead before calling such a function and will emit a warning about
3017 the variables that may be clobbered after the second return from the
3018 function. Examples of such functions are @code{setjmp} and @code{vfork}.
3019 The @code{longjmp}-like counterpart of such function, if any, might need
3020 to be marked with the @code{noreturn} attribute.
3023 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
3024 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
3025 all registers except the stack pointer should be saved in the prologue
3026 regardless of whether they are used or not.
3028 @item section ("@var{section-name}")
3029 @cindex @code{section} function attribute
3030 Normally, the compiler places the code it generates in the @code{text} section.
3031 Sometimes, however, you need additional sections, or you need certain
3032 particular functions to appear in special sections. The @code{section}
3033 attribute specifies that a function lives in a particular section.
3034 For example, the declaration:
3037 extern void foobar (void) __attribute__ ((section ("bar")));
3041 puts the function @code{foobar} in the @code{bar} section.
3043 Some file formats do not support arbitrary sections so the @code{section}
3044 attribute is not available on all platforms.
3045 If you need to map the entire contents of a module to a particular
3046 section, consider using the facilities of the linker instead.
3049 @cindex @code{sentinel} function attribute
3050 This function attribute ensures that a parameter in a function call is
3051 an explicit @code{NULL}. The attribute is only valid on variadic
3052 functions. By default, the sentinel is located at position zero, the
3053 last parameter of the function call. If an optional integer position
3054 argument P is supplied to the attribute, the sentinel must be located at
3055 position P counting backwards from the end of the argument list.
3058 __attribute__ ((sentinel))
3060 __attribute__ ((sentinel(0)))
3063 The attribute is automatically set with a position of 0 for the built-in
3064 functions @code{execl} and @code{execlp}. The built-in function
3065 @code{execle} has the attribute set with a position of 1.
3067 A valid @code{NULL} in this context is defined as zero with any pointer
3068 type. If your system defines the @code{NULL} macro with an integer type
3069 then you need to add an explicit cast. GCC replaces @code{stddef.h}
3070 with a copy that redefines NULL appropriately.
3072 The warnings for missing or incorrect sentinels are enabled with
3076 See long_call/short_call.
3079 See longcall/shortcall.
3082 @cindex signal handler functions on the AVR processors
3083 Use this attribute on the AVR to indicate that the specified
3084 function is a signal handler. The compiler will generate function
3085 entry and exit sequences suitable for use in a signal handler when this
3086 attribute is present. Interrupts will be disabled inside the function.
3089 Use this attribute on the SH to indicate an @code{interrupt_handler}
3090 function should switch to an alternate stack. It expects a string
3091 argument that names a global variable holding the address of the
3096 void f () __attribute__ ((interrupt_handler,
3097 sp_switch ("alt_stack")));
3101 @cindex functions that pop the argument stack on the 386
3102 On the Intel 386, the @code{stdcall} attribute causes the compiler to
3103 assume that the called function will pop off the stack space used to
3104 pass arguments, unless it takes a variable number of arguments.
3106 @item syscall_linkage
3107 @cindex @code{syscall_linkage} attribute
3108 This attribute is used to modify the IA64 calling convention by marking
3109 all input registers as live at all function exits. This makes it possible
3110 to restart a system call after an interrupt without having to save/restore
3111 the input registers. This also prevents kernel data from leaking into
3115 @cindex @code{target} function attribute
3116 The @code{target} attribute is used to specify that a function is to
3117 be compiled with different target options than specified on the
3118 command line. This can be used for instance to have functions
3119 compiled with a different ISA (instruction set architecture) than the
3120 default. You can also use the @samp{#pragma GCC target} pragma to set
3121 more than one function to be compiled with specific target options.
3122 @xref{Function Specific Option Pragmas}, for details about the
3123 @samp{#pragma GCC target} pragma.
3125 For instance on a 386, you could compile one function with
3126 @code{target("sse4.1,arch=core2")} and another with
3127 @code{target("sse4a,arch=amdfam10")} that would be equivalent to
3128 compiling the first function with @option{-msse4.1} and
3129 @option{-march=core2} options, and the second function with
3130 @option{-msse4a} and @option{-march=amdfam10} options. It is up to the
3131 user to make sure that a function is only invoked on a machine that
3132 supports the particular ISA it was compiled for (for example by using
3133 @code{cpuid} on 386 to determine what feature bits and architecture
3137 int core2_func (void) __attribute__ ((__target__ ("arch=core2")));
3138 int sse3_func (void) __attribute__ ((__target__ ("sse3")));
3141 On the 386, the following options are allowed:
3146 @cindex @code{target("abm")} attribute
3147 Enable/disable the generation of the advanced bit instructions.
3151 @cindex @code{target("aes")} attribute
3152 Enable/disable the generation of the AES instructions.
3156 @cindex @code{target("mmx")} attribute
3157 Enable/disable the generation of the MMX instructions.
3161 @cindex @code{target("pclmul")} attribute
3162 Enable/disable the generation of the PCLMUL instructions.
3166 @cindex @code{target("popcnt")} attribute
3167 Enable/disable the generation of the POPCNT instruction.
3171 @cindex @code{target("sse")} attribute
3172 Enable/disable the generation of the SSE instructions.
3176 @cindex @code{target("sse2")} attribute
3177 Enable/disable the generation of the SSE2 instructions.
3181 @cindex @code{target("sse3")} attribute
3182 Enable/disable the generation of the SSE3 instructions.
3186 @cindex @code{target("sse4")} attribute
3187 Enable/disable the generation of the SSE4 instructions (both SSE4.1
3192 @cindex @code{target("sse4.1")} attribute
3193 Enable/disable the generation of the sse4.1 instructions.
3197 @cindex @code{target("sse4.2")} attribute
3198 Enable/disable the generation of the sse4.2 instructions.
3202 @cindex @code{target("sse4a")} attribute
3203 Enable/disable the generation of the SSE4A instructions.
3207 @cindex @code{target("fma4")} attribute
3208 Enable/disable the generation of the FMA4 instructions.
3212 @cindex @code{target("xop")} attribute
3213 Enable/disable the generation of the XOP instructions.
3217 @cindex @code{target("lwp")} attribute
3218 Enable/disable the generation of the LWP instructions.
3222 @cindex @code{target("ssse3")} attribute
3223 Enable/disable the generation of the SSSE3 instructions.
3227 @cindex @code{target("cld")} attribute
3228 Enable/disable the generation of the CLD before string moves.
3230 @item fancy-math-387
3231 @itemx no-fancy-math-387
3232 @cindex @code{target("fancy-math-387")} attribute
3233 Enable/disable the generation of the @code{sin}, @code{cos}, and
3234 @code{sqrt} instructions on the 387 floating point unit.
3237 @itemx no-fused-madd
3238 @cindex @code{target("fused-madd")} attribute
3239 Enable/disable the generation of the fused multiply/add instructions.
3243 @cindex @code{target("ieee-fp")} attribute
3244 Enable/disable the generation of floating point that depends on IEEE arithmetic.
3246 @item inline-all-stringops
3247 @itemx no-inline-all-stringops
3248 @cindex @code{target("inline-all-stringops")} attribute
3249 Enable/disable inlining of string operations.
3251 @item inline-stringops-dynamically
3252 @itemx no-inline-stringops-dynamically
3253 @cindex @code{target("inline-stringops-dynamically")} attribute
3254 Enable/disable the generation of the inline code to do small string
3255 operations and calling the library routines for large operations.
3257 @item align-stringops
3258 @itemx no-align-stringops
3259 @cindex @code{target("align-stringops")} attribute
3260 Do/do not align destination of inlined string operations.
3264 @cindex @code{target("recip")} attribute
3265 Enable/disable the generation of RCPSS, RCPPS, RSQRTSS and RSQRTPS
3266 instructions followed an additional Newton-Raphson step instead of
3267 doing a floating point division.
3269 @item arch=@var{ARCH}
3270 @cindex @code{target("arch=@var{ARCH}")} attribute
3271 Specify the architecture to generate code for in compiling the function.
3273 @item tune=@var{TUNE}
3274 @cindex @code{target("tune=@var{TUNE}")} attribute
3275 Specify the architecture to tune for in compiling the function.
3277 @item fpmath=@var{FPMATH}
3278 @cindex @code{target("fpmath=@var{FPMATH}")} attribute
3279 Specify which floating point unit to use. The
3280 @code{target("fpmath=sse,387")} option must be specified as
3281 @code{target("fpmath=sse+387")} because the comma would separate
3285 On the 386, you can use either multiple strings to specify multiple
3286 options, or you can separate the option with a comma (@code{,}).
3288 On the 386, the inliner will not inline a function that has different
3289 target options than the caller, unless the callee has a subset of the
3290 target options of the caller. For example a function declared with
3291 @code{target("sse3")} can inline a function with
3292 @code{target("sse2")}, since @code{-msse3} implies @code{-msse2}.
3294 The @code{target} attribute is not implemented in GCC versions earlier
3295 than 4.4, and at present only the 386 uses it.
3298 @cindex tiny data section on the H8/300H and H8S
3299 Use this attribute on the H8/300H and H8S to indicate that the specified
3300 variable should be placed into the tiny data section.
3301 The compiler will generate more efficient code for loads and stores
3302 on data in the tiny data section. Note the tiny data area is limited to
3303 slightly under 32kbytes of data.
3306 Use this attribute on the SH for an @code{interrupt_handler} to return using
3307 @code{trapa} instead of @code{rte}. This attribute expects an integer
3308 argument specifying the trap number to be used.
3311 @cindex @code{unused} attribute.
3312 This attribute, attached to a function, means that the function is meant
3313 to be possibly unused. GCC will not produce a warning for this
3317 @cindex @code{used} attribute.
3318 This attribute, attached to a function, means that code must be emitted
3319 for the function even if it appears that the function is not referenced.
3320 This is useful, for example, when the function is referenced only in
3324 @cindex @code{version_id} attribute
3325 This IA64 HP-UX attribute, attached to a global variable or function, renames a
3326 symbol to contain a version string, thus allowing for function level
3327 versioning. HP-UX system header files may use version level functioning
3328 for some system calls.
3331 extern int foo () __attribute__((version_id ("20040821")));
3334 Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}.
3336 @item visibility ("@var{visibility_type}")
3337 @cindex @code{visibility} attribute
3338 This attribute affects the linkage of the declaration to which it is attached.
3339 There are four supported @var{visibility_type} values: default,
3340 hidden, protected or internal visibility.
3343 void __attribute__ ((visibility ("protected")))
3344 f () @{ /* @r{Do something.} */; @}
3345 int i __attribute__ ((visibility ("hidden")));
3348 The possible values of @var{visibility_type} correspond to the
3349 visibility settings in the ELF gABI.
3352 @c keep this list of visibilities in alphabetical order.
3355 Default visibility is the normal case for the object file format.
3356 This value is available for the visibility attribute to override other
3357 options that may change the assumed visibility of entities.
3359 On ELF, default visibility means that the declaration is visible to other
3360 modules and, in shared libraries, means that the declared entity may be
3363 On Darwin, default visibility means that the declaration is visible to
3366 Default visibility corresponds to ``external linkage'' in the language.
3369 Hidden visibility indicates that the entity declared will have a new
3370 form of linkage, which we'll call ``hidden linkage''. Two
3371 declarations of an object with hidden linkage refer to the same object
3372 if they are in the same shared object.
3375 Internal visibility is like hidden visibility, but with additional
3376 processor specific semantics. Unless otherwise specified by the
3377 psABI, GCC defines internal visibility to mean that a function is
3378 @emph{never} called from another module. Compare this with hidden
3379 functions which, while they cannot be referenced directly by other
3380 modules, can be referenced indirectly via function pointers. By
3381 indicating that a function cannot be called from outside the module,
3382 GCC may for instance omit the load of a PIC register since it is known
3383 that the calling function loaded the correct value.
3386 Protected visibility is like default visibility except that it
3387 indicates that references within the defining module will bind to the
3388 definition in that module. That is, the declared entity cannot be
3389 overridden by another module.
3393 All visibilities are supported on many, but not all, ELF targets
3394 (supported when the assembler supports the @samp{.visibility}
3395 pseudo-op). Default visibility is supported everywhere. Hidden
3396 visibility is supported on Darwin targets.
3398 The visibility attribute should be applied only to declarations which
3399 would otherwise have external linkage. The attribute should be applied
3400 consistently, so that the same entity should not be declared with
3401 different settings of the attribute.
3403 In C++, the visibility attribute applies to types as well as functions
3404 and objects, because in C++ types have linkage. A class must not have
3405 greater visibility than its non-static data member types and bases,
3406 and class members default to the visibility of their class. Also, a
3407 declaration without explicit visibility is limited to the visibility
3410 In C++, you can mark member functions and static member variables of a
3411 class with the visibility attribute. This is useful if you know a
3412 particular method or static member variable should only be used from
3413 one shared object; then you can mark it hidden while the rest of the
3414 class has default visibility. Care must be taken to avoid breaking
3415 the One Definition Rule; for example, it is usually not useful to mark
3416 an inline method as hidden without marking the whole class as hidden.
3418 A C++ namespace declaration can also have the visibility attribute.
3419 This attribute applies only to the particular namespace body, not to
3420 other definitions of the same namespace; it is equivalent to using
3421 @samp{#pragma GCC visibility} before and after the namespace
3422 definition (@pxref{Visibility Pragmas}).
3424 In C++, if a template argument has limited visibility, this
3425 restriction is implicitly propagated to the template instantiation.
3426 Otherwise, template instantiations and specializations default to the
3427 visibility of their template.
3429 If both the template and enclosing class have explicit visibility, the
3430 visibility from the template is used.
3433 @cindex @code{vliw} attribute
3434 On MeP, the @code{vliw} attribute tells the compiler to emit
3435 instructions in VLIW mode instead of core mode. Note that this
3436 attribute is not allowed unless a VLIW coprocessor has been configured
3437 and enabled through command line options.
3439 @item warn_unused_result
3440 @cindex @code{warn_unused_result} attribute
3441 The @code{warn_unused_result} attribute causes a warning to be emitted
3442 if a caller of the function with this attribute does not use its
3443 return value. This is useful for functions where not checking
3444 the result is either a security problem or always a bug, such as
3448 int fn () __attribute__ ((warn_unused_result));
3451 if (fn () < 0) return -1;
3457 results in warning on line 5.
3460 @cindex @code{weak} attribute
3461 The @code{weak} attribute causes the declaration to be emitted as a weak
3462 symbol rather than a global. This is primarily useful in defining
3463 library functions which can be overridden in user code, though it can
3464 also be used with non-function declarations. Weak symbols are supported
3465 for ELF targets, and also for a.out targets when using the GNU assembler
3469 @itemx weakref ("@var{target}")
3470 @cindex @code{weakref} attribute
3471 The @code{weakref} attribute marks a declaration as a weak reference.
3472 Without arguments, it should be accompanied by an @code{alias} attribute
3473 naming the target symbol. Optionally, the @var{target} may be given as
3474 an argument to @code{weakref} itself. In either case, @code{weakref}
3475 implicitly marks the declaration as @code{weak}. Without a
3476 @var{target}, given as an argument to @code{weakref} or to @code{alias},
3477 @code{weakref} is equivalent to @code{weak}.
3480 static int x() __attribute__ ((weakref ("y")));
3481 /* is equivalent to... */
3482 static int x() __attribute__ ((weak, weakref, alias ("y")));
3484 static int x() __attribute__ ((weakref));
3485 static int x() __attribute__ ((alias ("y")));
3488 A weak reference is an alias that does not by itself require a
3489 definition to be given for the target symbol. If the target symbol is
3490 only referenced through weak references, then the becomes a @code{weak}
3491 undefined symbol. If it is directly referenced, however, then such
3492 strong references prevail, and a definition will be required for the
3493 symbol, not necessarily in the same translation unit.
3495 The effect is equivalent to moving all references to the alias to a
3496 separate translation unit, renaming the alias to the aliased symbol,
3497 declaring it as weak, compiling the two separate translation units and
3498 performing a reloadable link on them.
3500 At present, a declaration to which @code{weakref} is attached can
3501 only be @code{static}.
3505 You can specify multiple attributes in a declaration by separating them
3506 by commas within the double parentheses or by immediately following an
3507 attribute declaration with another attribute declaration.
3509 @cindex @code{#pragma}, reason for not using
3510 @cindex pragma, reason for not using
3511 Some people object to the @code{__attribute__} feature, suggesting that
3512 ISO C's @code{#pragma} should be used instead. At the time
3513 @code{__attribute__} was designed, there were two reasons for not doing
3518 It is impossible to generate @code{#pragma} commands from a macro.
3521 There is no telling what the same @code{#pragma} might mean in another
3525 These two reasons applied to almost any application that might have been
3526 proposed for @code{#pragma}. It was basically a mistake to use
3527 @code{#pragma} for @emph{anything}.
3529 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
3530 to be generated from macros. In addition, a @code{#pragma GCC}
3531 namespace is now in use for GCC-specific pragmas. However, it has been
3532 found convenient to use @code{__attribute__} to achieve a natural
3533 attachment of attributes to their corresponding declarations, whereas
3534 @code{#pragma GCC} is of use for constructs that do not naturally form
3535 part of the grammar. @xref{Other Directives,,Miscellaneous
3536 Preprocessing Directives, cpp, The GNU C Preprocessor}.
3538 @node Attribute Syntax
3539 @section Attribute Syntax
3540 @cindex attribute syntax
3542 This section describes the syntax with which @code{__attribute__} may be
3543 used, and the constructs to which attribute specifiers bind, for the C
3544 language. Some details may vary for C++ and Objective-C@. Because of
3545 infelicities in the grammar for attributes, some forms described here
3546 may not be successfully parsed in all cases.
3548 There are some problems with the semantics of attributes in C++. For
3549 example, there are no manglings for attributes, although they may affect
3550 code generation, so problems may arise when attributed types are used in
3551 conjunction with templates or overloading. Similarly, @code{typeid}
3552 does not distinguish between types with different attributes. Support
3553 for attributes in C++ may be restricted in future to attributes on
3554 declarations only, but not on nested declarators.
3556 @xref{Function Attributes}, for details of the semantics of attributes
3557 applying to functions. @xref{Variable Attributes}, for details of the
3558 semantics of attributes applying to variables. @xref{Type Attributes},
3559 for details of the semantics of attributes applying to structure, union
3560 and enumerated types.
3562 An @dfn{attribute specifier} is of the form
3563 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
3564 is a possibly empty comma-separated sequence of @dfn{attributes}, where
3565 each attribute is one of the following:
3569 Empty. Empty attributes are ignored.
3572 A word (which may be an identifier such as @code{unused}, or a reserved
3573 word such as @code{const}).
3576 A word, followed by, in parentheses, parameters for the attribute.
3577 These parameters take one of the following forms:
3581 An identifier. For example, @code{mode} attributes use this form.
3584 An identifier followed by a comma and a non-empty comma-separated list
3585 of expressions. For example, @code{format} attributes use this form.
3588 A possibly empty comma-separated list of expressions. For example,
3589 @code{format_arg} attributes use this form with the list being a single
3590 integer constant expression, and @code{alias} attributes use this form
3591 with the list being a single string constant.
3595 An @dfn{attribute specifier list} is a sequence of one or more attribute
3596 specifiers, not separated by any other tokens.
3598 In GNU C, an attribute specifier list may appear after the colon following a
3599 label, other than a @code{case} or @code{default} label. The only
3600 attribute it makes sense to use after a label is @code{unused}. This
3601 feature is intended for code generated by programs which contains labels
3602 that may be unused but which is compiled with @option{-Wall}. It would
3603 not normally be appropriate to use in it human-written code, though it
3604 could be useful in cases where the code that jumps to the label is
3605 contained within an @code{#ifdef} conditional. GNU C++ only permits
3606 attributes on labels if the attribute specifier is immediately
3607 followed by a semicolon (i.e., the label applies to an empty
3608 statement). If the semicolon is missing, C++ label attributes are
3609 ambiguous, as it is permissible for a declaration, which could begin
3610 with an attribute list, to be labelled in C++. Declarations cannot be
3611 labelled in C90 or C99, so the ambiguity does not arise there.
3613 An attribute specifier list may appear as part of a @code{struct},
3614 @code{union} or @code{enum} specifier. It may go either immediately
3615 after the @code{struct}, @code{union} or @code{enum} keyword, or after
3616 the closing brace. The former syntax is preferred.
3617 Where attribute specifiers follow the closing brace, they are considered
3618 to relate to the structure, union or enumerated type defined, not to any
3619 enclosing declaration the type specifier appears in, and the type
3620 defined is not complete until after the attribute specifiers.
3621 @c Otherwise, there would be the following problems: a shift/reduce
3622 @c conflict between attributes binding the struct/union/enum and
3623 @c binding to the list of specifiers/qualifiers; and "aligned"
3624 @c attributes could use sizeof for the structure, but the size could be
3625 @c changed later by "packed" attributes.
3627 Otherwise, an attribute specifier appears as part of a declaration,
3628 counting declarations of unnamed parameters and type names, and relates
3629 to that declaration (which may be nested in another declaration, for
3630 example in the case of a parameter declaration), or to a particular declarator
3631 within a declaration. Where an
3632 attribute specifier is applied to a parameter declared as a function or
3633 an array, it should apply to the function or array rather than the
3634 pointer to which the parameter is implicitly converted, but this is not
3635 yet correctly implemented.
3637 Any list of specifiers and qualifiers at the start of a declaration may
3638 contain attribute specifiers, whether or not such a list may in that
3639 context contain storage class specifiers. (Some attributes, however,
3640 are essentially in the nature of storage class specifiers, and only make
3641 sense where storage class specifiers may be used; for example,
3642 @code{section}.) There is one necessary limitation to this syntax: the
3643 first old-style parameter declaration in a function definition cannot
3644 begin with an attribute specifier, because such an attribute applies to
3645 the function instead by syntax described below (which, however, is not
3646 yet implemented in this case). In some other cases, attribute
3647 specifiers are permitted by this grammar but not yet supported by the
3648 compiler. All attribute specifiers in this place relate to the
3649 declaration as a whole. In the obsolescent usage where a type of
3650 @code{int} is implied by the absence of type specifiers, such a list of
3651 specifiers and qualifiers may be an attribute specifier list with no
3652 other specifiers or qualifiers.
3654 At present, the first parameter in a function prototype must have some
3655 type specifier which is not an attribute specifier; this resolves an
3656 ambiguity in the interpretation of @code{void f(int
3657 (__attribute__((foo)) x))}, but is subject to change. At present, if
3658 the parentheses of a function declarator contain only attributes then
3659 those attributes are ignored, rather than yielding an error or warning
3660 or implying a single parameter of type int, but this is subject to
3663 An attribute specifier list may appear immediately before a declarator
3664 (other than the first) in a comma-separated list of declarators in a
3665 declaration of more than one identifier using a single list of
3666 specifiers and qualifiers. Such attribute specifiers apply
3667 only to the identifier before whose declarator they appear. For
3671 __attribute__((noreturn)) void d0 (void),
3672 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
3677 the @code{noreturn} attribute applies to all the functions
3678 declared; the @code{format} attribute only applies to @code{d1}.
3680 An attribute specifier list may appear immediately before the comma,
3681 @code{=} or semicolon terminating the declaration of an identifier other
3682 than a function definition. Such attribute specifiers apply
3683 to the declared object or function. Where an
3684 assembler name for an object or function is specified (@pxref{Asm
3685 Labels}), the attribute must follow the @code{asm}
3688 An attribute specifier list may, in future, be permitted to appear after
3689 the declarator in a function definition (before any old-style parameter
3690 declarations or the function body).
3692 Attribute specifiers may be mixed with type qualifiers appearing inside
3693 the @code{[]} of a parameter array declarator, in the C99 construct by
3694 which such qualifiers are applied to the pointer to which the array is
3695 implicitly converted. Such attribute specifiers apply to the pointer,
3696 not to the array, but at present this is not implemented and they are
3699 An attribute specifier list may appear at the start of a nested
3700 declarator. At present, there are some limitations in this usage: the
3701 attributes correctly apply to the declarator, but for most individual
3702 attributes the semantics this implies are not implemented.
3703 When attribute specifiers follow the @code{*} of a pointer
3704 declarator, they may be mixed with any type qualifiers present.
3705 The following describes the formal semantics of this syntax. It will make the
3706 most sense if you are familiar with the formal specification of
3707 declarators in the ISO C standard.
3709 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
3710 D1}, where @code{T} contains declaration specifiers that specify a type
3711 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
3712 contains an identifier @var{ident}. The type specified for @var{ident}
3713 for derived declarators whose type does not include an attribute
3714 specifier is as in the ISO C standard.
3716 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
3717 and the declaration @code{T D} specifies the type
3718 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3719 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3720 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
3722 If @code{D1} has the form @code{*
3723 @var{type-qualifier-and-attribute-specifier-list} D}, and the
3724 declaration @code{T D} specifies the type
3725 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
3726 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
3727 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
3733 void (__attribute__((noreturn)) ****f) (void);
3737 specifies the type ``pointer to pointer to pointer to pointer to
3738 non-returning function returning @code{void}''. As another example,
3741 char *__attribute__((aligned(8))) *f;
3745 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
3746 Note again that this does not work with most attributes; for example,
3747 the usage of @samp{aligned} and @samp{noreturn} attributes given above
3748 is not yet supported.
3750 For compatibility with existing code written for compiler versions that
3751 did not implement attributes on nested declarators, some laxity is
3752 allowed in the placing of attributes. If an attribute that only applies
3753 to types is applied to a declaration, it will be treated as applying to
3754 the type of that declaration. If an attribute that only applies to
3755 declarations is applied to the type of a declaration, it will be treated
3756 as applying to that declaration; and, for compatibility with code
3757 placing the attributes immediately before the identifier declared, such
3758 an attribute applied to a function return type will be treated as
3759 applying to the function type, and such an attribute applied to an array
3760 element type will be treated as applying to the array type. If an
3761 attribute that only applies to function types is applied to a
3762 pointer-to-function type, it will be treated as applying to the pointer
3763 target type; if such an attribute is applied to a function return type
3764 that is not a pointer-to-function type, it will be treated as applying
3765 to the function type.
3767 @node Function Prototypes
3768 @section Prototypes and Old-Style Function Definitions
3769 @cindex function prototype declarations
3770 @cindex old-style function definitions
3771 @cindex promotion of formal parameters
3773 GNU C extends ISO C to allow a function prototype to override a later
3774 old-style non-prototype definition. Consider the following example:
3777 /* @r{Use prototypes unless the compiler is old-fashioned.} */
3784 /* @r{Prototype function declaration.} */
3785 int isroot P((uid_t));
3787 /* @r{Old-style function definition.} */
3789 isroot (x) /* @r{??? lossage here ???} */
3796 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
3797 not allow this example, because subword arguments in old-style
3798 non-prototype definitions are promoted. Therefore in this example the
3799 function definition's argument is really an @code{int}, which does not
3800 match the prototype argument type of @code{short}.
3802 This restriction of ISO C makes it hard to write code that is portable
3803 to traditional C compilers, because the programmer does not know
3804 whether the @code{uid_t} type is @code{short}, @code{int}, or
3805 @code{long}. Therefore, in cases like these GNU C allows a prototype
3806 to override a later old-style definition. More precisely, in GNU C, a
3807 function prototype argument type overrides the argument type specified
3808 by a later old-style definition if the former type is the same as the
3809 latter type before promotion. Thus in GNU C the above example is
3810 equivalent to the following:
3823 GNU C++ does not support old-style function definitions, so this
3824 extension is irrelevant.
3827 @section C++ Style Comments
3829 @cindex C++ comments
3830 @cindex comments, C++ style
3832 In GNU C, you may use C++ style comments, which start with @samp{//} and
3833 continue until the end of the line. Many other C implementations allow
3834 such comments, and they are included in the 1999 C standard. However,
3835 C++ style comments are not recognized if you specify an @option{-std}
3836 option specifying a version of ISO C before C99, or @option{-ansi}
3837 (equivalent to @option{-std=c89}).
3840 @section Dollar Signs in Identifier Names
3842 @cindex dollar signs in identifier names
3843 @cindex identifier names, dollar signs in
3845 In GNU C, you may normally use dollar signs in identifier names.
3846 This is because many traditional C implementations allow such identifiers.
3847 However, dollar signs in identifiers are not supported on a few target
3848 machines, typically because the target assembler does not allow them.
3850 @node Character Escapes
3851 @section The Character @key{ESC} in Constants
3853 You can use the sequence @samp{\e} in a string or character constant to
3854 stand for the ASCII character @key{ESC}.
3857 @section Inquiring on Alignment of Types or Variables
3859 @cindex type alignment
3860 @cindex variable alignment
3862 The keyword @code{__alignof__} allows you to inquire about how an object
3863 is aligned, or the minimum alignment usually required by a type. Its
3864 syntax is just like @code{sizeof}.
3866 For example, if the target machine requires a @code{double} value to be
3867 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
3868 This is true on many RISC machines. On more traditional machine
3869 designs, @code{__alignof__ (double)} is 4 or even 2.
3871 Some machines never actually require alignment; they allow reference to any
3872 data type even at an odd address. For these machines, @code{__alignof__}
3873 reports the smallest alignment that GCC will give the data type, usually as
3874 mandated by the target ABI.
3876 If the operand of @code{__alignof__} is an lvalue rather than a type,
3877 its value is the required alignment for its type, taking into account
3878 any minimum alignment specified with GCC's @code{__attribute__}
3879 extension (@pxref{Variable Attributes}). For example, after this
3883 struct foo @{ int x; char y; @} foo1;
3887 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
3888 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
3890 It is an error to ask for the alignment of an incomplete type.
3892 @node Variable Attributes
3893 @section Specifying Attributes of Variables
3894 @cindex attribute of variables
3895 @cindex variable attributes
3897 The keyword @code{__attribute__} allows you to specify special
3898 attributes of variables or structure fields. This keyword is followed
3899 by an attribute specification inside double parentheses. Some
3900 attributes are currently defined generically for variables.
3901 Other attributes are defined for variables on particular target
3902 systems. Other attributes are available for functions
3903 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
3904 Other front ends might define more attributes
3905 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
3907 You may also specify attributes with @samp{__} preceding and following
3908 each keyword. This allows you to use them in header files without
3909 being concerned about a possible macro of the same name. For example,
3910 you may use @code{__aligned__} instead of @code{aligned}.
3912 @xref{Attribute Syntax}, for details of the exact syntax for using
3916 @cindex @code{aligned} attribute
3917 @item aligned (@var{alignment})
3918 This attribute specifies a minimum alignment for the variable or
3919 structure field, measured in bytes. For example, the declaration:
3922 int x __attribute__ ((aligned (16))) = 0;
3926 causes the compiler to allocate the global variable @code{x} on a
3927 16-byte boundary. On a 68040, this could be used in conjunction with
3928 an @code{asm} expression to access the @code{move16} instruction which
3929 requires 16-byte aligned operands.
3931 You can also specify the alignment of structure fields. For example, to
3932 create a double-word aligned @code{int} pair, you could write:
3935 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
3939 This is an alternative to creating a union with a @code{double} member
3940 that forces the union to be double-word aligned.
3942 As in the preceding examples, you can explicitly specify the alignment
3943 (in bytes) that you wish the compiler to use for a given variable or
3944 structure field. Alternatively, you can leave out the alignment factor
3945 and just ask the compiler to align a variable or field to the
3946 default alignment for the target architecture you are compiling for.
3947 The default alignment is sufficient for all scalar types, but may not be
3948 enough for all vector types on a target which supports vector operations.
3949 The default alignment is fixed for a particular target ABI.
3951 Gcc also provides a target specific macro @code{__BIGGEST_ALIGNMENT__},
3952 which is the largest alignment ever used for any data type on the
3953 target machine you are compiling for. For example, you could write:
3956 short array[3] __attribute__ ((aligned (__BIGGEST_ALIGNMENT__)));
3959 The compiler automatically sets the alignment for the declared
3960 variable or field to @code{__BIGGEST_ALIGNMENT__}. Doing this can
3961 often make copy operations more efficient, because the compiler can
3962 use whatever instructions copy the biggest chunks of memory when
3963 performing copies to or from the variables or fields that you have
3964 aligned this way. Note that the value of @code{__BIGGEST_ALIGNMENT__}
3965 may change depending on command line options.
3967 When used on a struct, or struct member, the @code{aligned} attribute can
3968 only increase the alignment; in order to decrease it, the @code{packed}
3969 attribute must be specified as well. When used as part of a typedef, the
3970 @code{aligned} attribute can both increase and decrease alignment, and
3971 specifying the @code{packed} attribute will generate a warning.
3973 Note that the effectiveness of @code{aligned} attributes may be limited
3974 by inherent limitations in your linker. On many systems, the linker is
3975 only able to arrange for variables to be aligned up to a certain maximum
3976 alignment. (For some linkers, the maximum supported alignment may
3977 be very very small.) If your linker is only able to align variables
3978 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3979 in an @code{__attribute__} will still only provide you with 8 byte
3980 alignment. See your linker documentation for further information.
3982 The @code{aligned} attribute can also be used for functions
3983 (@pxref{Function Attributes}.)
3985 @item cleanup (@var{cleanup_function})
3986 @cindex @code{cleanup} attribute
3987 The @code{cleanup} attribute runs a function when the variable goes
3988 out of scope. This attribute can only be applied to auto function
3989 scope variables; it may not be applied to parameters or variables
3990 with static storage duration. The function must take one parameter,
3991 a pointer to a type compatible with the variable. The return value
3992 of the function (if any) is ignored.
3994 If @option{-fexceptions} is enabled, then @var{cleanup_function}
3995 will be run during the stack unwinding that happens during the
3996 processing of the exception. Note that the @code{cleanup} attribute
3997 does not allow the exception to be caught, only to perform an action.
3998 It is undefined what happens if @var{cleanup_function} does not
4003 @cindex @code{common} attribute
4004 @cindex @code{nocommon} attribute
4007 The @code{common} attribute requests GCC to place a variable in
4008 ``common'' storage. The @code{nocommon} attribute requests the
4009 opposite---to allocate space for it directly.
4011 These attributes override the default chosen by the
4012 @option{-fno-common} and @option{-fcommon} flags respectively.
4015 @itemx deprecated (@var{msg})
4016 @cindex @code{deprecated} attribute
4017 The @code{deprecated} attribute results in a warning if the variable
4018 is used anywhere in the source file. This is useful when identifying
4019 variables that are expected to be removed in a future version of a
4020 program. The warning also includes the location of the declaration
4021 of the deprecated variable, to enable users to easily find further
4022 information about why the variable is deprecated, or what they should
4023 do instead. Note that the warning only occurs for uses:
4026 extern int old_var __attribute__ ((deprecated));
4028 int new_fn () @{ return old_var; @}
4031 results in a warning on line 3 but not line 2. The optional msg
4032 argument, which must be a string, will be printed in the warning if
4035 The @code{deprecated} attribute can also be used for functions and
4036 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
4038 @item mode (@var{mode})
4039 @cindex @code{mode} attribute
4040 This attribute specifies the data type for the declaration---whichever
4041 type corresponds to the mode @var{mode}. This in effect lets you
4042 request an integer or floating point type according to its width.
4044 You may also specify a mode of @samp{byte} or @samp{__byte__} to
4045 indicate the mode corresponding to a one-byte integer, @samp{word} or
4046 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
4047 or @samp{__pointer__} for the mode used to represent pointers.
4050 @cindex @code{packed} attribute
4051 The @code{packed} attribute specifies that a variable or structure field
4052 should have the smallest possible alignment---one byte for a variable,
4053 and one bit for a field, unless you specify a larger value with the
4054 @code{aligned} attribute.
4056 Here is a structure in which the field @code{x} is packed, so that it
4057 immediately follows @code{a}:
4063 int x[2] __attribute__ ((packed));
4067 @emph{Note:} The 4.1, 4.2 and 4.3 series of GCC ignore the
4068 @code{packed} attribute on bit-fields of type @code{char}. This has
4069 been fixed in GCC 4.4 but the change can lead to differences in the
4070 structure layout. See the documentation of
4071 @option{-Wpacked-bitfield-compat} for more information.
4073 @item section ("@var{section-name}")
4074 @cindex @code{section} variable attribute
4075 Normally, the compiler places the objects it generates in sections like
4076 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
4077 or you need certain particular variables to appear in special sections,
4078 for example to map to special hardware. The @code{section}
4079 attribute specifies that a variable (or function) lives in a particular
4080 section. For example, this small program uses several specific section names:
4083 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
4084 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
4085 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
4086 int init_data __attribute__ ((section ("INITDATA")));
4090 /* @r{Initialize stack pointer} */
4091 init_sp (stack + sizeof (stack));
4093 /* @r{Initialize initialized data} */
4094 memcpy (&init_data, &data, &edata - &data);
4096 /* @r{Turn on the serial ports} */
4103 Use the @code{section} attribute with
4104 @emph{global} variables and not @emph{local} variables,
4105 as shown in the example.
4107 You may use the @code{section} attribute with initialized or
4108 uninitialized global variables but the linker requires
4109 each object be defined once, with the exception that uninitialized
4110 variables tentatively go in the @code{common} (or @code{bss}) section
4111 and can be multiply ``defined''. Using the @code{section} attribute
4112 will change what section the variable goes into and may cause the
4113 linker to issue an error if an uninitialized variable has multiple
4114 definitions. You can force a variable to be initialized with the
4115 @option{-fno-common} flag or the @code{nocommon} attribute.
4117 Some file formats do not support arbitrary sections so the @code{section}
4118 attribute is not available on all platforms.
4119 If you need to map the entire contents of a module to a particular
4120 section, consider using the facilities of the linker instead.
4123 @cindex @code{shared} variable attribute
4124 On Microsoft Windows, in addition to putting variable definitions in a named
4125 section, the section can also be shared among all running copies of an
4126 executable or DLL@. For example, this small program defines shared data
4127 by putting it in a named section @code{shared} and marking the section
4131 int foo __attribute__((section ("shared"), shared)) = 0;
4136 /* @r{Read and write foo. All running
4137 copies see the same value.} */
4143 You may only use the @code{shared} attribute along with @code{section}
4144 attribute with a fully initialized global definition because of the way
4145 linkers work. See @code{section} attribute for more information.
4147 The @code{shared} attribute is only available on Microsoft Windows@.
4149 @item tls_model ("@var{tls_model}")
4150 @cindex @code{tls_model} attribute
4151 The @code{tls_model} attribute sets thread-local storage model
4152 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
4153 overriding @option{-ftls-model=} command line switch on a per-variable
4155 The @var{tls_model} argument should be one of @code{global-dynamic},
4156 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
4158 Not all targets support this attribute.
4161 This attribute, attached to a variable, means that the variable is meant
4162 to be possibly unused. GCC will not produce a warning for this
4166 This attribute, attached to a variable, means that the variable must be
4167 emitted even if it appears that the variable is not referenced.
4169 @item vector_size (@var{bytes})
4170 This attribute specifies the vector size for the variable, measured in
4171 bytes. For example, the declaration:
4174 int foo __attribute__ ((vector_size (16)));
4178 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
4179 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
4180 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
4182 This attribute is only applicable to integral and float scalars,
4183 although arrays, pointers, and function return values are allowed in
4184 conjunction with this construct.
4186 Aggregates with this attribute are invalid, even if they are of the same
4187 size as a corresponding scalar. For example, the declaration:
4190 struct S @{ int a; @};
4191 struct S __attribute__ ((vector_size (16))) foo;
4195 is invalid even if the size of the structure is the same as the size of
4199 The @code{selectany} attribute causes an initialized global variable to
4200 have link-once semantics. When multiple definitions of the variable are
4201 encountered by the linker, the first is selected and the remainder are
4202 discarded. Following usage by the Microsoft compiler, the linker is told
4203 @emph{not} to warn about size or content differences of the multiple
4206 Although the primary usage of this attribute is for POD types, the
4207 attribute can also be applied to global C++ objects that are initialized
4208 by a constructor. In this case, the static initialization and destruction
4209 code for the object is emitted in each translation defining the object,
4210 but the calls to the constructor and destructor are protected by a
4211 link-once guard variable.
4213 The @code{selectany} attribute is only available on Microsoft Windows
4214 targets. You can use @code{__declspec (selectany)} as a synonym for
4215 @code{__attribute__ ((selectany))} for compatibility with other
4219 The @code{weak} attribute is described in @ref{Function Attributes}.
4222 The @code{dllimport} attribute is described in @ref{Function Attributes}.
4225 The @code{dllexport} attribute is described in @ref{Function Attributes}.
4229 @subsection Blackfin Variable Attributes
4231 Three attributes are currently defined for the Blackfin.
4237 @cindex @code{l1_data} variable attribute
4238 @cindex @code{l1_data_A} variable attribute
4239 @cindex @code{l1_data_B} variable attribute
4240 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
4241 Variables with @code{l1_data} attribute will be put into the specific section
4242 named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into
4243 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
4244 attribute will be put into the specific section named @code{.l1.data.B}.
4247 @cindex @code{l2} variable attribute
4248 Use this attribute on the Blackfin to place the variable into L2 SRAM.
4249 Variables with @code{l2} attribute will be put into the specific section
4250 named @code{.l2.data}.
4253 @subsection M32R/D Variable Attributes
4255 One attribute is currently defined for the M32R/D@.
4258 @item model (@var{model-name})
4259 @cindex variable addressability on the M32R/D
4260 Use this attribute on the M32R/D to set the addressability of an object.
4261 The identifier @var{model-name} is one of @code{small}, @code{medium},
4262 or @code{large}, representing each of the code models.
4264 Small model objects live in the lower 16MB of memory (so that their
4265 addresses can be loaded with the @code{ld24} instruction).
4267 Medium and large model objects may live anywhere in the 32-bit address space
4268 (the compiler will generate @code{seth/add3} instructions to load their
4272 @anchor{MeP Variable Attributes}
4273 @subsection MeP Variable Attributes
4275 The MeP target has a number of addressing modes and busses. The
4276 @code{near} space spans the standard memory space's first 16 megabytes
4277 (24 bits). The @code{far} space spans the entire 32-bit memory space.
4278 The @code{based} space is a 128 byte region in the memory space which
4279 is addressed relative to the @code{$tp} register. The @code{tiny}
4280 space is a 65536 byte region relative to the @code{$gp} register. In
4281 addition to these memory regions, the MeP target has a separate 16-bit
4282 control bus which is specified with @code{cb} attributes.
4287 Any variable with the @code{based} attribute will be assigned to the
4288 @code{.based} section, and will be accessed with relative to the
4289 @code{$tp} register.
4292 Likewise, the @code{tiny} attribute assigned variables to the
4293 @code{.tiny} section, relative to the @code{$gp} register.
4296 Variables with the @code{near} attribute are assumed to have addresses
4297 that fit in a 24-bit addressing mode. This is the default for large
4298 variables (@code{-mtiny=4} is the default) but this attribute can
4299 override @code{-mtiny=} for small variables, or override @code{-ml}.
4302 Variables with the @code{far} attribute are addressed using a full
4303 32-bit address. Since this covers the entire memory space, this
4304 allows modules to make no assumptions about where variables might be
4308 @item io (@var{addr})
4309 Variables with the @code{io} attribute are used to address
4310 memory-mapped peripherals. If an address is specified, the variable
4311 is assigned that address, else it is not assigned an address (it is
4312 assumed some other module will assign an address). Example:
4315 int timer_count __attribute__((io(0x123)));
4319 @item cb (@var{addr})
4320 Variables with the @code{cb} attribute are used to access the control
4321 bus, using special instructions. @code{addr} indicates the control bus
4325 int cpu_clock __attribute__((cb(0x123)));
4330 @anchor{i386 Variable Attributes}
4331 @subsection i386 Variable Attributes
4333 Two attributes are currently defined for i386 configurations:
4334 @code{ms_struct} and @code{gcc_struct}
4339 @cindex @code{ms_struct} attribute
4340 @cindex @code{gcc_struct} attribute
4342 If @code{packed} is used on a structure, or if bit-fields are used
4343 it may be that the Microsoft ABI packs them differently
4344 than GCC would normally pack them. Particularly when moving packed
4345 data between functions compiled with GCC and the native Microsoft compiler
4346 (either via function call or as data in a file), it may be necessary to access
4349 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4350 compilers to match the native Microsoft compiler.
4352 The Microsoft structure layout algorithm is fairly simple with the exception
4353 of the bitfield packing:
4355 The padding and alignment of members of structures and whether a bit field
4356 can straddle a storage-unit boundary
4359 @item Structure members are stored sequentially in the order in which they are
4360 declared: the first member has the lowest memory address and the last member
4363 @item Every data object has an alignment-requirement. The alignment-requirement
4364 for all data except structures, unions, and arrays is either the size of the
4365 object or the current packing size (specified with either the aligned attribute
4366 or the pack pragma), whichever is less. For structures, unions, and arrays,
4367 the alignment-requirement is the largest alignment-requirement of its members.
4368 Every object is allocated an offset so that:
4370 offset % alignment-requirement == 0
4372 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
4373 unit if the integral types are the same size and if the next bit field fits
4374 into the current allocation unit without crossing the boundary imposed by the
4375 common alignment requirements of the bit fields.
4378 Handling of zero-length bitfields:
4380 MSVC interprets zero-length bitfields in the following ways:
4383 @item If a zero-length bitfield is inserted between two bitfields that would
4384 normally be coalesced, the bitfields will not be coalesced.
4391 unsigned long bf_1 : 12;
4393 unsigned long bf_2 : 12;
4397 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
4398 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
4400 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
4401 alignment of the zero-length bitfield is greater than the member that follows it,
4402 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
4422 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
4423 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
4424 bitfield will not affect the alignment of @code{bar} or, as a result, the size
4427 Taking this into account, it is important to note the following:
4430 @item If a zero-length bitfield follows a normal bitfield, the type of the
4431 zero-length bitfield may affect the alignment of the structure as whole. For
4432 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
4433 normal bitfield, and is of type short.
4435 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
4436 still affect the alignment of the structure:
4446 Here, @code{t4} will take up 4 bytes.
4449 @item Zero-length bitfields following non-bitfield members are ignored:
4460 Here, @code{t5} will take up 2 bytes.
4464 @subsection PowerPC Variable Attributes
4466 Three attributes currently are defined for PowerPC configurations:
4467 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4469 For full documentation of the struct attributes please see the
4470 documentation in @ref{i386 Variable Attributes}.
4472 For documentation of @code{altivec} attribute please see the
4473 documentation in @ref{PowerPC Type Attributes}.
4475 @subsection SPU Variable Attributes
4477 The SPU supports the @code{spu_vector} attribute for variables. For
4478 documentation of this attribute please see the documentation in
4479 @ref{SPU Type Attributes}.
4481 @subsection Xstormy16 Variable Attributes
4483 One attribute is currently defined for xstormy16 configurations:
4488 @cindex @code{below100} attribute
4490 If a variable has the @code{below100} attribute (@code{BELOW100} is
4491 allowed also), GCC will place the variable in the first 0x100 bytes of
4492 memory and use special opcodes to access it. Such variables will be
4493 placed in either the @code{.bss_below100} section or the
4494 @code{.data_below100} section.
4498 @subsection AVR Variable Attributes
4502 @cindex @code{progmem} variable attribute
4503 The @code{progmem} attribute is used on the AVR to place data in the Program
4504 Memory address space. The AVR is a Harvard Architecture processor and data
4505 normally resides in the Data Memory address space.
4508 @node Type Attributes
4509 @section Specifying Attributes of Types
4510 @cindex attribute of types
4511 @cindex type attributes
4513 The keyword @code{__attribute__} allows you to specify special
4514 attributes of @code{struct} and @code{union} types when you define
4515 such types. This keyword is followed by an attribute specification
4516 inside double parentheses. Seven attributes are currently defined for
4517 types: @code{aligned}, @code{packed}, @code{transparent_union},
4518 @code{unused}, @code{deprecated}, @code{visibility}, and
4519 @code{may_alias}. Other attributes are defined for functions
4520 (@pxref{Function Attributes}) and for variables (@pxref{Variable
4523 You may also specify any one of these attributes with @samp{__}
4524 preceding and following its keyword. This allows you to use these
4525 attributes in header files without being concerned about a possible
4526 macro of the same name. For example, you may use @code{__aligned__}
4527 instead of @code{aligned}.
4529 You may specify type attributes in an enum, struct or union type
4530 declaration or definition, or for other types in a @code{typedef}
4533 For an enum, struct or union type, you may specify attributes either
4534 between the enum, struct or union tag and the name of the type, or
4535 just past the closing curly brace of the @emph{definition}. The
4536 former syntax is preferred.
4538 @xref{Attribute Syntax}, for details of the exact syntax for using
4542 @cindex @code{aligned} attribute
4543 @item aligned (@var{alignment})
4544 This attribute specifies a minimum alignment (in bytes) for variables
4545 of the specified type. For example, the declarations:
4548 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
4549 typedef int more_aligned_int __attribute__ ((aligned (8)));
4553 force the compiler to insure (as far as it can) that each variable whose
4554 type is @code{struct S} or @code{more_aligned_int} will be allocated and
4555 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
4556 variables of type @code{struct S} aligned to 8-byte boundaries allows
4557 the compiler to use the @code{ldd} and @code{std} (doubleword load and
4558 store) instructions when copying one variable of type @code{struct S} to
4559 another, thus improving run-time efficiency.
4561 Note that the alignment of any given @code{struct} or @code{union} type
4562 is required by the ISO C standard to be at least a perfect multiple of
4563 the lowest common multiple of the alignments of all of the members of
4564 the @code{struct} or @code{union} in question. This means that you @emph{can}
4565 effectively adjust the alignment of a @code{struct} or @code{union}
4566 type by attaching an @code{aligned} attribute to any one of the members
4567 of such a type, but the notation illustrated in the example above is a
4568 more obvious, intuitive, and readable way to request the compiler to
4569 adjust the alignment of an entire @code{struct} or @code{union} type.
4571 As in the preceding example, you can explicitly specify the alignment
4572 (in bytes) that you wish the compiler to use for a given @code{struct}
4573 or @code{union} type. Alternatively, you can leave out the alignment factor
4574 and just ask the compiler to align a type to the maximum
4575 useful alignment for the target machine you are compiling for. For
4576 example, you could write:
4579 struct S @{ short f[3]; @} __attribute__ ((aligned));
4582 Whenever you leave out the alignment factor in an @code{aligned}
4583 attribute specification, the compiler automatically sets the alignment
4584 for the type to the largest alignment which is ever used for any data
4585 type on the target machine you are compiling for. Doing this can often
4586 make copy operations more efficient, because the compiler can use
4587 whatever instructions copy the biggest chunks of memory when performing
4588 copies to or from the variables which have types that you have aligned
4591 In the example above, if the size of each @code{short} is 2 bytes, then
4592 the size of the entire @code{struct S} type is 6 bytes. The smallest
4593 power of two which is greater than or equal to that is 8, so the
4594 compiler sets the alignment for the entire @code{struct S} type to 8
4597 Note that although you can ask the compiler to select a time-efficient
4598 alignment for a given type and then declare only individual stand-alone
4599 objects of that type, the compiler's ability to select a time-efficient
4600 alignment is primarily useful only when you plan to create arrays of
4601 variables having the relevant (efficiently aligned) type. If you
4602 declare or use arrays of variables of an efficiently-aligned type, then
4603 it is likely that your program will also be doing pointer arithmetic (or
4604 subscripting, which amounts to the same thing) on pointers to the
4605 relevant type, and the code that the compiler generates for these
4606 pointer arithmetic operations will often be more efficient for
4607 efficiently-aligned types than for other types.
4609 The @code{aligned} attribute can only increase the alignment; but you
4610 can decrease it by specifying @code{packed} as well. See below.
4612 Note that the effectiveness of @code{aligned} attributes may be limited
4613 by inherent limitations in your linker. On many systems, the linker is
4614 only able to arrange for variables to be aligned up to a certain maximum
4615 alignment. (For some linkers, the maximum supported alignment may
4616 be very very small.) If your linker is only able to align variables
4617 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
4618 in an @code{__attribute__} will still only provide you with 8 byte
4619 alignment. See your linker documentation for further information.
4622 This attribute, attached to @code{struct} or @code{union} type
4623 definition, specifies that each member (other than zero-width bitfields)
4624 of the structure or union is placed to minimize the memory required. When
4625 attached to an @code{enum} definition, it indicates that the smallest
4626 integral type should be used.
4628 @opindex fshort-enums
4629 Specifying this attribute for @code{struct} and @code{union} types is
4630 equivalent to specifying the @code{packed} attribute on each of the
4631 structure or union members. Specifying the @option{-fshort-enums}
4632 flag on the line is equivalent to specifying the @code{packed}
4633 attribute on all @code{enum} definitions.
4635 In the following example @code{struct my_packed_struct}'s members are
4636 packed closely together, but the internal layout of its @code{s} member
4637 is not packed---to do that, @code{struct my_unpacked_struct} would need to
4641 struct my_unpacked_struct
4647 struct __attribute__ ((__packed__)) my_packed_struct
4651 struct my_unpacked_struct s;
4655 You may only specify this attribute on the definition of an @code{enum},
4656 @code{struct} or @code{union}, not on a @code{typedef} which does not
4657 also define the enumerated type, structure or union.
4659 @item transparent_union
4660 This attribute, attached to a @code{union} type definition, indicates
4661 that any function parameter having that union type causes calls to that
4662 function to be treated in a special way.
4664 First, the argument corresponding to a transparent union type can be of
4665 any type in the union; no cast is required. Also, if the union contains
4666 a pointer type, the corresponding argument can be a null pointer
4667 constant or a void pointer expression; and if the union contains a void
4668 pointer type, the corresponding argument can be any pointer expression.
4669 If the union member type is a pointer, qualifiers like @code{const} on
4670 the referenced type must be respected, just as with normal pointer
4673 Second, the argument is passed to the function using the calling
4674 conventions of the first member of the transparent union, not the calling
4675 conventions of the union itself. All members of the union must have the
4676 same machine representation; this is necessary for this argument passing
4679 Transparent unions are designed for library functions that have multiple
4680 interfaces for compatibility reasons. For example, suppose the
4681 @code{wait} function must accept either a value of type @code{int *} to
4682 comply with Posix, or a value of type @code{union wait *} to comply with
4683 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
4684 @code{wait} would accept both kinds of arguments, but it would also
4685 accept any other pointer type and this would make argument type checking
4686 less useful. Instead, @code{<sys/wait.h>} might define the interface
4690 typedef union __attribute__ ((__transparent_union__))
4694 @} wait_status_ptr_t;
4696 pid_t wait (wait_status_ptr_t);
4699 This interface allows either @code{int *} or @code{union wait *}
4700 arguments to be passed, using the @code{int *} calling convention.
4701 The program can call @code{wait} with arguments of either type:
4704 int w1 () @{ int w; return wait (&w); @}
4705 int w2 () @{ union wait w; return wait (&w); @}
4708 With this interface, @code{wait}'s implementation might look like this:
4711 pid_t wait (wait_status_ptr_t p)
4713 return waitpid (-1, p.__ip, 0);
4718 When attached to a type (including a @code{union} or a @code{struct}),
4719 this attribute means that variables of that type are meant to appear
4720 possibly unused. GCC will not produce a warning for any variables of
4721 that type, even if the variable appears to do nothing. This is often
4722 the case with lock or thread classes, which are usually defined and then
4723 not referenced, but contain constructors and destructors that have
4724 nontrivial bookkeeping functions.
4727 @itemx deprecated (@var{msg})
4728 The @code{deprecated} attribute results in a warning if the type
4729 is used anywhere in the source file. This is useful when identifying
4730 types that are expected to be removed in a future version of a program.
4731 If possible, the warning also includes the location of the declaration
4732 of the deprecated type, to enable users to easily find further
4733 information about why the type is deprecated, or what they should do
4734 instead. Note that the warnings only occur for uses and then only
4735 if the type is being applied to an identifier that itself is not being
4736 declared as deprecated.
4739 typedef int T1 __attribute__ ((deprecated));
4743 typedef T1 T3 __attribute__ ((deprecated));
4744 T3 z __attribute__ ((deprecated));
4747 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
4748 warning is issued for line 4 because T2 is not explicitly
4749 deprecated. Line 5 has no warning because T3 is explicitly
4750 deprecated. Similarly for line 6. The optional msg
4751 argument, which must be a string, will be printed in the warning if
4754 The @code{deprecated} attribute can also be used for functions and
4755 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
4758 Accesses through pointers to types with this attribute are not subject
4759 to type-based alias analysis, but are instead assumed to be able to alias
4760 any other type of objects. In the context of 6.5/7 an lvalue expression
4761 dereferencing such a pointer is treated like having a character type.
4762 See @option{-fstrict-aliasing} for more information on aliasing issues.
4763 This extension exists to support some vector APIs, in which pointers to
4764 one vector type are permitted to alias pointers to a different vector type.
4766 Note that an object of a type with this attribute does not have any
4772 typedef short __attribute__((__may_alias__)) short_a;
4778 short_a *b = (short_a *) &a;
4782 if (a == 0x12345678)
4789 If you replaced @code{short_a} with @code{short} in the variable
4790 declaration, the above program would abort when compiled with
4791 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
4792 above in recent GCC versions.
4795 In C++, attribute visibility (@pxref{Function Attributes}) can also be
4796 applied to class, struct, union and enum types. Unlike other type
4797 attributes, the attribute must appear between the initial keyword and
4798 the name of the type; it cannot appear after the body of the type.
4800 Note that the type visibility is applied to vague linkage entities
4801 associated with the class (vtable, typeinfo node, etc.). In
4802 particular, if a class is thrown as an exception in one shared object
4803 and caught in another, the class must have default visibility.
4804 Otherwise the two shared objects will be unable to use the same
4805 typeinfo node and exception handling will break.
4809 @subsection ARM Type Attributes
4811 On those ARM targets that support @code{dllimport} (such as Symbian
4812 OS), you can use the @code{notshared} attribute to indicate that the
4813 virtual table and other similar data for a class should not be
4814 exported from a DLL@. For example:
4817 class __declspec(notshared) C @{
4819 __declspec(dllimport) C();
4823 __declspec(dllexport)
4827 In this code, @code{C::C} is exported from the current DLL, but the
4828 virtual table for @code{C} is not exported. (You can use
4829 @code{__attribute__} instead of @code{__declspec} if you prefer, but
4830 most Symbian OS code uses @code{__declspec}.)
4832 @anchor{MeP Type Attributes}
4833 @subsection MeP Type Attributes
4835 Many of the MeP variable attributes may be applied to types as well.
4836 Specifically, the @code{based}, @code{tiny}, @code{near}, and
4837 @code{far} attributes may be applied to either. The @code{io} and
4838 @code{cb} attributes may not be applied to types.
4840 @anchor{i386 Type Attributes}
4841 @subsection i386 Type Attributes
4843 Two attributes are currently defined for i386 configurations:
4844 @code{ms_struct} and @code{gcc_struct}.
4850 @cindex @code{ms_struct}
4851 @cindex @code{gcc_struct}
4853 If @code{packed} is used on a structure, or if bit-fields are used
4854 it may be that the Microsoft ABI packs them differently
4855 than GCC would normally pack them. Particularly when moving packed
4856 data between functions compiled with GCC and the native Microsoft compiler
4857 (either via function call or as data in a file), it may be necessary to access
4860 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4861 compilers to match the native Microsoft compiler.
4864 To specify multiple attributes, separate them by commas within the
4865 double parentheses: for example, @samp{__attribute__ ((aligned (16),
4868 @anchor{PowerPC Type Attributes}
4869 @subsection PowerPC Type Attributes
4871 Three attributes currently are defined for PowerPC configurations:
4872 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4874 For full documentation of the @code{ms_struct} and @code{gcc_struct}
4875 attributes please see the documentation in @ref{i386 Type Attributes}.
4877 The @code{altivec} attribute allows one to declare AltiVec vector data
4878 types supported by the AltiVec Programming Interface Manual. The
4879 attribute requires an argument to specify one of three vector types:
4880 @code{vector__}, @code{pixel__} (always followed by unsigned short),
4881 and @code{bool__} (always followed by unsigned).
4884 __attribute__((altivec(vector__)))
4885 __attribute__((altivec(pixel__))) unsigned short
4886 __attribute__((altivec(bool__))) unsigned
4889 These attributes mainly are intended to support the @code{__vector},
4890 @code{__pixel}, and @code{__bool} AltiVec keywords.
4892 @anchor{SPU Type Attributes}
4893 @subsection SPU Type Attributes
4895 The SPU supports the @code{spu_vector} attribute for types. This attribute
4896 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
4897 Language Extensions Specification. It is intended to support the
4898 @code{__vector} keyword.
4902 @section An Inline Function is As Fast As a Macro
4903 @cindex inline functions
4904 @cindex integrating function code
4906 @cindex macros, inline alternative
4908 By declaring a function inline, you can direct GCC to make
4909 calls to that function faster. One way GCC can achieve this is to
4910 integrate that function's code into the code for its callers. This
4911 makes execution faster by eliminating the function-call overhead; in
4912 addition, if any of the actual argument values are constant, their
4913 known values may permit simplifications at compile time so that not
4914 all of the inline function's code needs to be included. The effect on
4915 code size is less predictable; object code may be larger or smaller
4916 with function inlining, depending on the particular case. You can
4917 also direct GCC to try to integrate all ``simple enough'' functions
4918 into their callers with the option @option{-finline-functions}.
4920 GCC implements three different semantics of declaring a function
4921 inline. One is available with @option{-std=gnu89} or
4922 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
4923 on all inline declarations, another when @option{-std=c99} or
4924 @option{-std=gnu99} (without @option{-fgnu89-inline}), and the third
4925 is used when compiling C++.
4927 To declare a function inline, use the @code{inline} keyword in its
4928 declaration, like this:
4938 If you are writing a header file to be included in ISO C89 programs, write
4939 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
4941 The three types of inlining behave similarly in two important cases:
4942 when the @code{inline} keyword is used on a @code{static} function,
4943 like the example above, and when a function is first declared without
4944 using the @code{inline} keyword and then is defined with
4945 @code{inline}, like this:
4948 extern int inc (int *a);
4956 In both of these common cases, the program behaves the same as if you
4957 had not used the @code{inline} keyword, except for its speed.
4959 @cindex inline functions, omission of
4960 @opindex fkeep-inline-functions
4961 When a function is both inline and @code{static}, if all calls to the
4962 function are integrated into the caller, and the function's address is
4963 never used, then the function's own assembler code is never referenced.
4964 In this case, GCC does not actually output assembler code for the
4965 function, unless you specify the option @option{-fkeep-inline-functions}.
4966 Some calls cannot be integrated for various reasons (in particular,
4967 calls that precede the function's definition cannot be integrated, and
4968 neither can recursive calls within the definition). If there is a
4969 nonintegrated call, then the function is compiled to assembler code as
4970 usual. The function must also be compiled as usual if the program
4971 refers to its address, because that can't be inlined.
4974 Note that certain usages in a function definition can make it unsuitable
4975 for inline substitution. Among these usages are: use of varargs, use of
4976 alloca, use of variable sized data types (@pxref{Variable Length}),
4977 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
4978 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
4979 will warn when a function marked @code{inline} could not be substituted,
4980 and will give the reason for the failure.
4982 @cindex automatic @code{inline} for C++ member fns
4983 @cindex @code{inline} automatic for C++ member fns
4984 @cindex member fns, automatically @code{inline}
4985 @cindex C++ member fns, automatically @code{inline}
4986 @opindex fno-default-inline
4987 As required by ISO C++, GCC considers member functions defined within
4988 the body of a class to be marked inline even if they are
4989 not explicitly declared with the @code{inline} keyword. You can
4990 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
4991 Options,,Options Controlling C++ Dialect}.
4993 GCC does not inline any functions when not optimizing unless you specify
4994 the @samp{always_inline} attribute for the function, like this:
4997 /* @r{Prototype.} */
4998 inline void foo (const char) __attribute__((always_inline));
5001 The remainder of this section is specific to GNU C89 inlining.
5003 @cindex non-static inline function
5004 When an inline function is not @code{static}, then the compiler must assume
5005 that there may be calls from other source files; since a global symbol can
5006 be defined only once in any program, the function must not be defined in
5007 the other source files, so the calls therein cannot be integrated.
5008 Therefore, a non-@code{static} inline function is always compiled on its
5009 own in the usual fashion.
5011 If you specify both @code{inline} and @code{extern} in the function
5012 definition, then the definition is used only for inlining. In no case
5013 is the function compiled on its own, not even if you refer to its
5014 address explicitly. Such an address becomes an external reference, as
5015 if you had only declared the function, and had not defined it.
5017 This combination of @code{inline} and @code{extern} has almost the
5018 effect of a macro. The way to use it is to put a function definition in
5019 a header file with these keywords, and put another copy of the
5020 definition (lacking @code{inline} and @code{extern}) in a library file.
5021 The definition in the header file will cause most calls to the function
5022 to be inlined. If any uses of the function remain, they will refer to
5023 the single copy in the library.
5026 @section Assembler Instructions with C Expression Operands
5027 @cindex extended @code{asm}
5028 @cindex @code{asm} expressions
5029 @cindex assembler instructions
5032 In an assembler instruction using @code{asm}, you can specify the
5033 operands of the instruction using C expressions. This means you need not
5034 guess which registers or memory locations will contain the data you want
5037 You must specify an assembler instruction template much like what
5038 appears in a machine description, plus an operand constraint string for
5041 For example, here is how to use the 68881's @code{fsinx} instruction:
5044 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
5048 Here @code{angle} is the C expression for the input operand while
5049 @code{result} is that of the output operand. Each has @samp{"f"} as its
5050 operand constraint, saying that a floating point register is required.
5051 The @samp{=} in @samp{=f} indicates that the operand is an output; all
5052 output operands' constraints must use @samp{=}. The constraints use the
5053 same language used in the machine description (@pxref{Constraints}).
5055 Each operand is described by an operand-constraint string followed by
5056 the C expression in parentheses. A colon separates the assembler
5057 template from the first output operand and another separates the last
5058 output operand from the first input, if any. Commas separate the
5059 operands within each group. The total number of operands is currently
5060 limited to 30; this limitation may be lifted in some future version of
5063 If there are no output operands but there are input operands, you must
5064 place two consecutive colons surrounding the place where the output
5067 As of GCC version 3.1, it is also possible to specify input and output
5068 operands using symbolic names which can be referenced within the
5069 assembler code. These names are specified inside square brackets
5070 preceding the constraint string, and can be referenced inside the
5071 assembler code using @code{%[@var{name}]} instead of a percentage sign
5072 followed by the operand number. Using named operands the above example
5076 asm ("fsinx %[angle],%[output]"
5077 : [output] "=f" (result)
5078 : [angle] "f" (angle));
5082 Note that the symbolic operand names have no relation whatsoever to
5083 other C identifiers. You may use any name you like, even those of
5084 existing C symbols, but you must ensure that no two operands within the same
5085 assembler construct use the same symbolic name.
5087 Output operand expressions must be lvalues; the compiler can check this.
5088 The input operands need not be lvalues. The compiler cannot check
5089 whether the operands have data types that are reasonable for the
5090 instruction being executed. It does not parse the assembler instruction
5091 template and does not know what it means or even whether it is valid
5092 assembler input. The extended @code{asm} feature is most often used for
5093 machine instructions the compiler itself does not know exist. If
5094 the output expression cannot be directly addressed (for example, it is a
5095 bit-field), your constraint must allow a register. In that case, GCC
5096 will use the register as the output of the @code{asm}, and then store
5097 that register into the output.
5099 The ordinary output operands must be write-only; GCC will assume that
5100 the values in these operands before the instruction are dead and need
5101 not be generated. Extended asm supports input-output or read-write
5102 operands. Use the constraint character @samp{+} to indicate such an
5103 operand and list it with the output operands. You should only use
5104 read-write operands when the constraints for the operand (or the
5105 operand in which only some of the bits are to be changed) allow a
5108 You may, as an alternative, logically split its function into two
5109 separate operands, one input operand and one write-only output
5110 operand. The connection between them is expressed by constraints
5111 which say they need to be in the same location when the instruction
5112 executes. You can use the same C expression for both operands, or
5113 different expressions. For example, here we write the (fictitious)
5114 @samp{combine} instruction with @code{bar} as its read-only source
5115 operand and @code{foo} as its read-write destination:
5118 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
5122 The constraint @samp{"0"} for operand 1 says that it must occupy the
5123 same location as operand 0. A number in constraint is allowed only in
5124 an input operand and it must refer to an output operand.
5126 Only a number in the constraint can guarantee that one operand will be in
5127 the same place as another. The mere fact that @code{foo} is the value
5128 of both operands is not enough to guarantee that they will be in the
5129 same place in the generated assembler code. The following would not
5133 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
5136 Various optimizations or reloading could cause operands 0 and 1 to be in
5137 different registers; GCC knows no reason not to do so. For example, the
5138 compiler might find a copy of the value of @code{foo} in one register and
5139 use it for operand 1, but generate the output operand 0 in a different
5140 register (copying it afterward to @code{foo}'s own address). Of course,
5141 since the register for operand 1 is not even mentioned in the assembler
5142 code, the result will not work, but GCC can't tell that.
5144 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
5145 the operand number for a matching constraint. For example:
5148 asm ("cmoveq %1,%2,%[result]"
5149 : [result] "=r"(result)
5150 : "r" (test), "r"(new), "[result]"(old));
5153 Sometimes you need to make an @code{asm} operand be a specific register,
5154 but there's no matching constraint letter for that register @emph{by
5155 itself}. To force the operand into that register, use a local variable
5156 for the operand and specify the register in the variable declaration.
5157 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
5158 register constraint letter that matches the register:
5161 register int *p1 asm ("r0") = @dots{};
5162 register int *p2 asm ("r1") = @dots{};
5163 register int *result asm ("r0");
5164 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5167 @anchor{Example of asm with clobbered asm reg}
5168 In the above example, beware that a register that is call-clobbered by
5169 the target ABI will be overwritten by any function call in the
5170 assignment, including library calls for arithmetic operators.
5171 Also a register may be clobbered when generating some operations,
5172 like variable shift, memory copy or memory move on x86.
5173 Assuming it is a call-clobbered register, this may happen to @code{r0}
5174 above by the assignment to @code{p2}. If you have to use such a
5175 register, use temporary variables for expressions between the register
5180 register int *p1 asm ("r0") = @dots{};
5181 register int *p2 asm ("r1") = t1;
5182 register int *result asm ("r0");
5183 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
5186 Some instructions clobber specific hard registers. To describe this,
5187 write a third colon after the input operands, followed by the names of
5188 the clobbered hard registers (given as strings). Here is a realistic
5189 example for the VAX:
5192 asm volatile ("movc3 %0,%1,%2"
5193 : /* @r{no outputs} */
5194 : "g" (from), "g" (to), "g" (count)
5195 : "r0", "r1", "r2", "r3", "r4", "r5");
5198 You may not write a clobber description in a way that overlaps with an
5199 input or output operand. For example, you may not have an operand
5200 describing a register class with one member if you mention that register
5201 in the clobber list. Variables declared to live in specific registers
5202 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
5203 have no part mentioned in the clobber description.
5204 There is no way for you to specify that an input
5205 operand is modified without also specifying it as an output
5206 operand. Note that if all the output operands you specify are for this
5207 purpose (and hence unused), you will then also need to specify
5208 @code{volatile} for the @code{asm} construct, as described below, to
5209 prevent GCC from deleting the @code{asm} statement as unused.
5211 If you refer to a particular hardware register from the assembler code,
5212 you will probably have to list the register after the third colon to
5213 tell the compiler the register's value is modified. In some assemblers,
5214 the register names begin with @samp{%}; to produce one @samp{%} in the
5215 assembler code, you must write @samp{%%} in the input.
5217 If your assembler instruction can alter the condition code register, add
5218 @samp{cc} to the list of clobbered registers. GCC on some machines
5219 represents the condition codes as a specific hardware register;
5220 @samp{cc} serves to name this register. On other machines, the
5221 condition code is handled differently, and specifying @samp{cc} has no
5222 effect. But it is valid no matter what the machine.
5224 If your assembler instructions access memory in an unpredictable
5225 fashion, add @samp{memory} to the list of clobbered registers. This
5226 will cause GCC to not keep memory values cached in registers across the
5227 assembler instruction and not optimize stores or loads to that memory.
5228 You will also want to add the @code{volatile} keyword if the memory
5229 affected is not listed in the inputs or outputs of the @code{asm}, as
5230 the @samp{memory} clobber does not count as a side-effect of the
5231 @code{asm}. If you know how large the accessed memory is, you can add
5232 it as input or output but if this is not known, you should add
5233 @samp{memory}. As an example, if you access ten bytes of a string, you
5234 can use a memory input like:
5237 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
5240 Note that in the following example the memory input is necessary,
5241 otherwise GCC might optimize the store to @code{x} away:
5248 asm ("magic stuff accessing an 'int' pointed to by '%1'"
5249 "=&d" (r) : "a" (y), "m" (*y));
5254 You can put multiple assembler instructions together in a single
5255 @code{asm} template, separated by the characters normally used in assembly
5256 code for the system. A combination that works in most places is a newline
5257 to break the line, plus a tab character to move to the instruction field
5258 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
5259 assembler allows semicolons as a line-breaking character. Note that some
5260 assembler dialects use semicolons to start a comment.
5261 The input operands are guaranteed not to use any of the clobbered
5262 registers, and neither will the output operands' addresses, so you can
5263 read and write the clobbered registers as many times as you like. Here
5264 is an example of multiple instructions in a template; it assumes the
5265 subroutine @code{_foo} accepts arguments in registers 9 and 10:
5268 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
5270 : "g" (from), "g" (to)
5274 Unless an output operand has the @samp{&} constraint modifier, GCC
5275 may allocate it in the same register as an unrelated input operand, on
5276 the assumption the inputs are consumed before the outputs are produced.
5277 This assumption may be false if the assembler code actually consists of
5278 more than one instruction. In such a case, use @samp{&} for each output
5279 operand that may not overlap an input. @xref{Modifiers}.
5281 If you want to test the condition code produced by an assembler
5282 instruction, you must include a branch and a label in the @code{asm}
5283 construct, as follows:
5286 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
5292 This assumes your assembler supports local labels, as the GNU assembler
5293 and most Unix assemblers do.
5295 Speaking of labels, jumps from one @code{asm} to another are not
5296 supported. The compiler's optimizers do not know about these jumps, and
5297 therefore they cannot take account of them when deciding how to
5298 optimize. @xref{Extended asm with goto}.
5300 @cindex macros containing @code{asm}
5301 Usually the most convenient way to use these @code{asm} instructions is to
5302 encapsulate them in macros that look like functions. For example,
5306 (@{ double __value, __arg = (x); \
5307 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
5312 Here the variable @code{__arg} is used to make sure that the instruction
5313 operates on a proper @code{double} value, and to accept only those
5314 arguments @code{x} which can convert automatically to a @code{double}.
5316 Another way to make sure the instruction operates on the correct data
5317 type is to use a cast in the @code{asm}. This is different from using a
5318 variable @code{__arg} in that it converts more different types. For
5319 example, if the desired type were @code{int}, casting the argument to
5320 @code{int} would accept a pointer with no complaint, while assigning the
5321 argument to an @code{int} variable named @code{__arg} would warn about
5322 using a pointer unless the caller explicitly casts it.
5324 If an @code{asm} has output operands, GCC assumes for optimization
5325 purposes the instruction has no side effects except to change the output
5326 operands. This does not mean instructions with a side effect cannot be
5327 used, but you must be careful, because the compiler may eliminate them
5328 if the output operands aren't used, or move them out of loops, or
5329 replace two with one if they constitute a common subexpression. Also,
5330 if your instruction does have a side effect on a variable that otherwise
5331 appears not to change, the old value of the variable may be reused later
5332 if it happens to be found in a register.
5334 You can prevent an @code{asm} instruction from being deleted
5335 by writing the keyword @code{volatile} after
5336 the @code{asm}. For example:
5339 #define get_and_set_priority(new) \
5341 asm volatile ("get_and_set_priority %0, %1" \
5342 : "=g" (__old) : "g" (new)); \
5347 The @code{volatile} keyword indicates that the instruction has
5348 important side-effects. GCC will not delete a volatile @code{asm} if
5349 it is reachable. (The instruction can still be deleted if GCC can
5350 prove that control-flow will never reach the location of the
5351 instruction.) Note that even a volatile @code{asm} instruction
5352 can be moved relative to other code, including across jump
5353 instructions. For example, on many targets there is a system
5354 register which can be set to control the rounding mode of
5355 floating point operations. You might try
5356 setting it with a volatile @code{asm}, like this PowerPC example:
5359 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
5364 This will not work reliably, as the compiler may move the addition back
5365 before the volatile @code{asm}. To make it work you need to add an
5366 artificial dependency to the @code{asm} referencing a variable in the code
5367 you don't want moved, for example:
5370 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
5374 Similarly, you can't expect a
5375 sequence of volatile @code{asm} instructions to remain perfectly
5376 consecutive. If you want consecutive output, use a single @code{asm}.
5377 Also, GCC will perform some optimizations across a volatile @code{asm}
5378 instruction; GCC does not ``forget everything'' when it encounters
5379 a volatile @code{asm} instruction the way some other compilers do.
5381 An @code{asm} instruction without any output operands will be treated
5382 identically to a volatile @code{asm} instruction.
5384 It is a natural idea to look for a way to give access to the condition
5385 code left by the assembler instruction. However, when we attempted to
5386 implement this, we found no way to make it work reliably. The problem
5387 is that output operands might need reloading, which would result in
5388 additional following ``store'' instructions. On most machines, these
5389 instructions would alter the condition code before there was time to
5390 test it. This problem doesn't arise for ordinary ``test'' and
5391 ``compare'' instructions because they don't have any output operands.
5393 For reasons similar to those described above, it is not possible to give
5394 an assembler instruction access to the condition code left by previous
5397 @anchor{Extended asm with goto}
5398 As of GCC version 4.5, @code{asm goto} may be used to have the assembly
5399 jump to one or more C labels. In this form, a fifth section after the
5400 clobber list contains a list of all C labels to which the assembly may jump.
5401 Each label operand is implicitly self-named. The @code{asm} is also assumed
5402 to fall through to the next statement.
5404 This form of @code{asm} is restricted to not have outputs. This is due
5405 to a internal restriction in the compiler that control transfer instructions
5406 cannot have outputs. This restriction on @code{asm goto} may be lifted
5407 in some future version of the compiler. In the mean time, @code{asm goto}
5408 may include a memory clobber, and so leave outputs in memory.
5414 asm goto ("frob %%r5, %1; jc %l[error]; mov (%2), %%r5"
5415 : : "r"(x), "r"(&y) : "r5", "memory" : error);
5422 In this (inefficient) example, the @code{frob} instruction sets the
5423 carry bit to indicate an error. The @code{jc} instruction detects
5424 this and branches to the @code{error} label. Finally, the output
5425 of the @code{frob} instruction (@code{%r5}) is stored into the memory
5426 for variable @code{y}, which is later read by the @code{return} statement.
5432 asm goto ("mfsr %%r1, 123; jmp %%r1;"
5433 ".pushsection doit_table;"
5434 ".long %l0, %l1, %l2, %l3;"
5436 : : : "r1" : label1, label2, label3, label4);
5437 __builtin_unreachable ();
5452 In this (also inefficient) example, the @code{mfsr} instruction reads
5453 an address from some out-of-band machine register, and the following
5454 @code{jmp} instruction branches to that address. The address read by
5455 the @code{mfsr} instruction is assumed to have been previously set via
5456 some application-specific mechanism to be one of the four values stored
5457 in the @code{doit_table} section. Finally, the @code{asm} is followed
5458 by a call to @code{__builtin_unreachable} to indicate that the @code{asm}
5459 does not in fact fall through.
5462 #define TRACE1(NUM) \
5464 asm goto ("0: nop;" \
5465 ".pushsection trace_table;" \
5468 : : : : trace#NUM); \
5469 if (0) @{ trace#NUM: trace(); @} \
5471 #define TRACE TRACE1(__COUNTER__)
5474 In this example (which in fact inspired the @code{asm goto} feature)
5475 we want on rare occasions to call the @code{trace} function; on other
5476 occasions we'd like to keep the overhead to the absolute minimum.
5477 The normal code path consists of a single @code{nop} instruction.
5478 However, we record the address of this @code{nop} together with the
5479 address of a label that calls the @code{trace} function. This allows
5480 the @code{nop} instruction to be patched at runtime to be an
5481 unconditional branch to the stored label. It is assumed that an
5482 optimizing compiler will move the labeled block out of line, to
5483 optimize the fall through path from the @code{asm}.
5485 If you are writing a header file that should be includable in ISO C
5486 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
5489 @subsection Size of an @code{asm}
5491 Some targets require that GCC track the size of each instruction used in
5492 order to generate correct code. Because the final length of an
5493 @code{asm} is only known by the assembler, GCC must make an estimate as
5494 to how big it will be. The estimate is formed by counting the number of
5495 statements in the pattern of the @code{asm} and multiplying that by the
5496 length of the longest instruction on that processor. Statements in the
5497 @code{asm} are identified by newline characters and whatever statement
5498 separator characters are supported by the assembler; on most processors
5499 this is the `@code{;}' character.
5501 Normally, GCC's estimate is perfectly adequate to ensure that correct
5502 code is generated, but it is possible to confuse the compiler if you use
5503 pseudo instructions or assembler macros that expand into multiple real
5504 instructions or if you use assembler directives that expand to more
5505 space in the object file than would be needed for a single instruction.
5506 If this happens then the assembler will produce a diagnostic saying that
5507 a label is unreachable.
5509 @subsection i386 floating point asm operands
5511 There are several rules on the usage of stack-like regs in
5512 asm_operands insns. These rules apply only to the operands that are
5517 Given a set of input regs that die in an asm_operands, it is
5518 necessary to know which are implicitly popped by the asm, and
5519 which must be explicitly popped by gcc.
5521 An input reg that is implicitly popped by the asm must be
5522 explicitly clobbered, unless it is constrained to match an
5526 For any input reg that is implicitly popped by an asm, it is
5527 necessary to know how to adjust the stack to compensate for the pop.
5528 If any non-popped input is closer to the top of the reg-stack than
5529 the implicitly popped reg, it would not be possible to know what the
5530 stack looked like---it's not clear how the rest of the stack ``slides
5533 All implicitly popped input regs must be closer to the top of
5534 the reg-stack than any input that is not implicitly popped.
5536 It is possible that if an input dies in an insn, reload might
5537 use the input reg for an output reload. Consider this example:
5540 asm ("foo" : "=t" (a) : "f" (b));
5543 This asm says that input B is not popped by the asm, and that
5544 the asm pushes a result onto the reg-stack, i.e., the stack is one
5545 deeper after the asm than it was before. But, it is possible that
5546 reload will think that it can use the same reg for both the input and
5547 the output, if input B dies in this insn.
5549 If any input operand uses the @code{f} constraint, all output reg
5550 constraints must use the @code{&} earlyclobber.
5552 The asm above would be written as
5555 asm ("foo" : "=&t" (a) : "f" (b));
5559 Some operands need to be in particular places on the stack. All
5560 output operands fall in this category---there is no other way to
5561 know which regs the outputs appear in unless the user indicates
5562 this in the constraints.
5564 Output operands must specifically indicate which reg an output
5565 appears in after an asm. @code{=f} is not allowed: the operand
5566 constraints must select a class with a single reg.
5569 Output operands may not be ``inserted'' between existing stack regs.
5570 Since no 387 opcode uses a read/write operand, all output operands
5571 are dead before the asm_operands, and are pushed by the asm_operands.
5572 It makes no sense to push anywhere but the top of the reg-stack.
5574 Output operands must start at the top of the reg-stack: output
5575 operands may not ``skip'' a reg.
5578 Some asm statements may need extra stack space for internal
5579 calculations. This can be guaranteed by clobbering stack registers
5580 unrelated to the inputs and outputs.
5584 Here are a couple of reasonable asms to want to write. This asm
5585 takes one input, which is internally popped, and produces two outputs.
5588 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
5591 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
5592 and replaces them with one output. The user must code the @code{st(1)}
5593 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
5596 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
5602 @section Controlling Names Used in Assembler Code
5603 @cindex assembler names for identifiers
5604 @cindex names used in assembler code
5605 @cindex identifiers, names in assembler code
5607 You can specify the name to be used in the assembler code for a C
5608 function or variable by writing the @code{asm} (or @code{__asm__})
5609 keyword after the declarator as follows:
5612 int foo asm ("myfoo") = 2;
5616 This specifies that the name to be used for the variable @code{foo} in
5617 the assembler code should be @samp{myfoo} rather than the usual
5620 On systems where an underscore is normally prepended to the name of a C
5621 function or variable, this feature allows you to define names for the
5622 linker that do not start with an underscore.
5624 It does not make sense to use this feature with a non-static local
5625 variable since such variables do not have assembler names. If you are
5626 trying to put the variable in a particular register, see @ref{Explicit
5627 Reg Vars}. GCC presently accepts such code with a warning, but will
5628 probably be changed to issue an error, rather than a warning, in the
5631 You cannot use @code{asm} in this way in a function @emph{definition}; but
5632 you can get the same effect by writing a declaration for the function
5633 before its definition and putting @code{asm} there, like this:
5636 extern func () asm ("FUNC");
5643 It is up to you to make sure that the assembler names you choose do not
5644 conflict with any other assembler symbols. Also, you must not use a
5645 register name; that would produce completely invalid assembler code. GCC
5646 does not as yet have the ability to store static variables in registers.
5647 Perhaps that will be added.
5649 @node Explicit Reg Vars
5650 @section Variables in Specified Registers
5651 @cindex explicit register variables
5652 @cindex variables in specified registers
5653 @cindex specified registers
5654 @cindex registers, global allocation
5656 GNU C allows you to put a few global variables into specified hardware
5657 registers. You can also specify the register in which an ordinary
5658 register variable should be allocated.
5662 Global register variables reserve registers throughout the program.
5663 This may be useful in programs such as programming language
5664 interpreters which have a couple of global variables that are accessed
5668 Local register variables in specific registers do not reserve the
5669 registers, except at the point where they are used as input or output
5670 operands in an @code{asm} statement and the @code{asm} statement itself is
5671 not deleted. The compiler's data flow analysis is capable of determining
5672 where the specified registers contain live values, and where they are
5673 available for other uses. Stores into local register variables may be deleted
5674 when they appear to be dead according to dataflow analysis. References
5675 to local register variables may be deleted or moved or simplified.
5677 These local variables are sometimes convenient for use with the extended
5678 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
5679 output of the assembler instruction directly into a particular register.
5680 (This will work provided the register you specify fits the constraints
5681 specified for that operand in the @code{asm}.)
5689 @node Global Reg Vars
5690 @subsection Defining Global Register Variables
5691 @cindex global register variables
5692 @cindex registers, global variables in
5694 You can define a global register variable in GNU C like this:
5697 register int *foo asm ("a5");
5701 Here @code{a5} is the name of the register which should be used. Choose a
5702 register which is normally saved and restored by function calls on your
5703 machine, so that library routines will not clobber it.
5705 Naturally the register name is cpu-dependent, so you would need to
5706 conditionalize your program according to cpu type. The register
5707 @code{a5} would be a good choice on a 68000 for a variable of pointer
5708 type. On machines with register windows, be sure to choose a ``global''
5709 register that is not affected magically by the function call mechanism.
5711 In addition, operating systems on one type of cpu may differ in how they
5712 name the registers; then you would need additional conditionals. For
5713 example, some 68000 operating systems call this register @code{%a5}.
5715 Eventually there may be a way of asking the compiler to choose a register
5716 automatically, but first we need to figure out how it should choose and
5717 how to enable you to guide the choice. No solution is evident.
5719 Defining a global register variable in a certain register reserves that
5720 register entirely for this use, at least within the current compilation.
5721 The register will not be allocated for any other purpose in the functions
5722 in the current compilation. The register will not be saved and restored by
5723 these functions. Stores into this register are never deleted even if they
5724 would appear to be dead, but references may be deleted or moved or
5727 It is not safe to access the global register variables from signal
5728 handlers, or from more than one thread of control, because the system
5729 library routines may temporarily use the register for other things (unless
5730 you recompile them specially for the task at hand).
5732 @cindex @code{qsort}, and global register variables
5733 It is not safe for one function that uses a global register variable to
5734 call another such function @code{foo} by way of a third function
5735 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
5736 different source file in which the variable wasn't declared). This is
5737 because @code{lose} might save the register and put some other value there.
5738 For example, you can't expect a global register variable to be available in
5739 the comparison-function that you pass to @code{qsort}, since @code{qsort}
5740 might have put something else in that register. (If you are prepared to
5741 recompile @code{qsort} with the same global register variable, you can
5742 solve this problem.)
5744 If you want to recompile @code{qsort} or other source files which do not
5745 actually use your global register variable, so that they will not use that
5746 register for any other purpose, then it suffices to specify the compiler
5747 option @option{-ffixed-@var{reg}}. You need not actually add a global
5748 register declaration to their source code.
5750 A function which can alter the value of a global register variable cannot
5751 safely be called from a function compiled without this variable, because it
5752 could clobber the value the caller expects to find there on return.
5753 Therefore, the function which is the entry point into the part of the
5754 program that uses the global register variable must explicitly save and
5755 restore the value which belongs to its caller.
5757 @cindex register variable after @code{longjmp}
5758 @cindex global register after @code{longjmp}
5759 @cindex value after @code{longjmp}
5762 On most machines, @code{longjmp} will restore to each global register
5763 variable the value it had at the time of the @code{setjmp}. On some
5764 machines, however, @code{longjmp} will not change the value of global
5765 register variables. To be portable, the function that called @code{setjmp}
5766 should make other arrangements to save the values of the global register
5767 variables, and to restore them in a @code{longjmp}. This way, the same
5768 thing will happen regardless of what @code{longjmp} does.
5770 All global register variable declarations must precede all function
5771 definitions. If such a declaration could appear after function
5772 definitions, the declaration would be too late to prevent the register from
5773 being used for other purposes in the preceding functions.
5775 Global register variables may not have initial values, because an
5776 executable file has no means to supply initial contents for a register.
5778 On the SPARC, there are reports that g3 @dots{} g7 are suitable
5779 registers, but certain library functions, such as @code{getwd}, as well
5780 as the subroutines for division and remainder, modify g3 and g4. g1 and
5781 g2 are local temporaries.
5783 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
5784 Of course, it will not do to use more than a few of those.
5786 @node Local Reg Vars
5787 @subsection Specifying Registers for Local Variables
5788 @cindex local variables, specifying registers
5789 @cindex specifying registers for local variables
5790 @cindex registers for local variables
5792 You can define a local register variable with a specified register
5796 register int *foo asm ("a5");
5800 Here @code{a5} is the name of the register which should be used. Note
5801 that this is the same syntax used for defining global register
5802 variables, but for a local variable it would appear within a function.
5804 Naturally the register name is cpu-dependent, but this is not a
5805 problem, since specific registers are most often useful with explicit
5806 assembler instructions (@pxref{Extended Asm}). Both of these things
5807 generally require that you conditionalize your program according to
5810 In addition, operating systems on one type of cpu may differ in how they
5811 name the registers; then you would need additional conditionals. For
5812 example, some 68000 operating systems call this register @code{%a5}.
5814 Defining such a register variable does not reserve the register; it
5815 remains available for other uses in places where flow control determines
5816 the variable's value is not live.
5818 This option does not guarantee that GCC will generate code that has
5819 this variable in the register you specify at all times. You may not
5820 code an explicit reference to this register in the @emph{assembler
5821 instruction template} part of an @code{asm} statement and assume it will
5822 always refer to this variable. However, using the variable as an
5823 @code{asm} @emph{operand} guarantees that the specified register is used
5826 Stores into local register variables may be deleted when they appear to be dead
5827 according to dataflow analysis. References to local register variables may
5828 be deleted or moved or simplified.
5830 As for global register variables, it's recommended that you choose a
5831 register which is normally saved and restored by function calls on
5832 your machine, so that library routines will not clobber it. A common
5833 pitfall is to initialize multiple call-clobbered registers with
5834 arbitrary expressions, where a function call or library call for an
5835 arithmetic operator will overwrite a register value from a previous
5836 assignment, for example @code{r0} below:
5838 register int *p1 asm ("r0") = @dots{};
5839 register int *p2 asm ("r1") = @dots{};
5841 In those cases, a solution is to use a temporary variable for
5842 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
5844 @node Alternate Keywords
5845 @section Alternate Keywords
5846 @cindex alternate keywords
5847 @cindex keywords, alternate
5849 @option{-ansi} and the various @option{-std} options disable certain
5850 keywords. This causes trouble when you want to use GNU C extensions, or
5851 a general-purpose header file that should be usable by all programs,
5852 including ISO C programs. The keywords @code{asm}, @code{typeof} and
5853 @code{inline} are not available in programs compiled with
5854 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
5855 program compiled with @option{-std=c99}). The ISO C99 keyword
5856 @code{restrict} is only available when @option{-std=gnu99} (which will
5857 eventually be the default) or @option{-std=c99} (or the equivalent
5858 @option{-std=iso9899:1999}) is used.
5860 The way to solve these problems is to put @samp{__} at the beginning and
5861 end of each problematical keyword. For example, use @code{__asm__}
5862 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
5864 Other C compilers won't accept these alternative keywords; if you want to
5865 compile with another compiler, you can define the alternate keywords as
5866 macros to replace them with the customary keywords. It looks like this:
5874 @findex __extension__
5876 @option{-pedantic} and other options cause warnings for many GNU C extensions.
5878 prevent such warnings within one expression by writing
5879 @code{__extension__} before the expression. @code{__extension__} has no
5880 effect aside from this.
5882 @node Incomplete Enums
5883 @section Incomplete @code{enum} Types
5885 You can define an @code{enum} tag without specifying its possible values.
5886 This results in an incomplete type, much like what you get if you write
5887 @code{struct foo} without describing the elements. A later declaration
5888 which does specify the possible values completes the type.
5890 You can't allocate variables or storage using the type while it is
5891 incomplete. However, you can work with pointers to that type.
5893 This extension may not be very useful, but it makes the handling of
5894 @code{enum} more consistent with the way @code{struct} and @code{union}
5897 This extension is not supported by GNU C++.
5899 @node Function Names
5900 @section Function Names as Strings
5901 @cindex @code{__func__} identifier
5902 @cindex @code{__FUNCTION__} identifier
5903 @cindex @code{__PRETTY_FUNCTION__} identifier
5905 GCC provides three magic variables which hold the name of the current
5906 function, as a string. The first of these is @code{__func__}, which
5907 is part of the C99 standard:
5909 The identifier @code{__func__} is implicitly declared by the translator
5910 as if, immediately following the opening brace of each function
5911 definition, the declaration
5914 static const char __func__[] = "function-name";
5918 appeared, where function-name is the name of the lexically-enclosing
5919 function. This name is the unadorned name of the function.
5921 @code{__FUNCTION__} is another name for @code{__func__}. Older
5922 versions of GCC recognize only this name. However, it is not
5923 standardized. For maximum portability, we recommend you use
5924 @code{__func__}, but provide a fallback definition with the
5928 #if __STDC_VERSION__ < 199901L
5930 # define __func__ __FUNCTION__
5932 # define __func__ "<unknown>"
5937 In C, @code{__PRETTY_FUNCTION__} is yet another name for
5938 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
5939 the type signature of the function as well as its bare name. For
5940 example, this program:
5944 extern int printf (char *, ...);
5951 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
5952 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
5970 __PRETTY_FUNCTION__ = void a::sub(int)
5973 These identifiers are not preprocessor macros. In GCC 3.3 and
5974 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
5975 were treated as string literals; they could be used to initialize
5976 @code{char} arrays, and they could be concatenated with other string
5977 literals. GCC 3.4 and later treat them as variables, like
5978 @code{__func__}. In C++, @code{__FUNCTION__} and
5979 @code{__PRETTY_FUNCTION__} have always been variables.
5981 @node Return Address
5982 @section Getting the Return or Frame Address of a Function
5984 These functions may be used to get information about the callers of a
5987 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
5988 This function returns the return address of the current function, or of
5989 one of its callers. The @var{level} argument is number of frames to
5990 scan up the call stack. A value of @code{0} yields the return address
5991 of the current function, a value of @code{1} yields the return address
5992 of the caller of the current function, and so forth. When inlining
5993 the expected behavior is that the function will return the address of
5994 the function that will be returned to. To work around this behavior use
5995 the @code{noinline} function attribute.
5997 The @var{level} argument must be a constant integer.
5999 On some machines it may be impossible to determine the return address of
6000 any function other than the current one; in such cases, or when the top
6001 of the stack has been reached, this function will return @code{0} or a
6002 random value. In addition, @code{__builtin_frame_address} may be used
6003 to determine if the top of the stack has been reached.
6005 Additional post-processing of the returned value may be needed, see
6006 @code{__builtin_extract_return_address}.
6008 This function should only be used with a nonzero argument for debugging
6012 @deftypefn {Built-in Function} {void *} __builtin_extract_return_address (void *@var{addr})
6013 The address as returned by @code{__builtin_return_address} may have to be fed
6014 through this function to get the actual encoded address. For example, on the
6015 31-bit S/390 platform the highest bit has to be masked out, or on SPARC
6016 platforms an offset has to be added for the true next instruction to be
6019 If no fixup is needed, this function simply passes through @var{addr}.
6022 @deftypefn {Built-in Function} {void *} __builtin_frob_return_address (void *@var{addr})
6023 This function does the reverse of @code{__builtin_extract_return_address}.
6026 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
6027 This function is similar to @code{__builtin_return_address}, but it
6028 returns the address of the function frame rather than the return address
6029 of the function. Calling @code{__builtin_frame_address} with a value of
6030 @code{0} yields the frame address of the current function, a value of
6031 @code{1} yields the frame address of the caller of the current function,
6034 The frame is the area on the stack which holds local variables and saved
6035 registers. The frame address is normally the address of the first word
6036 pushed on to the stack by the function. However, the exact definition
6037 depends upon the processor and the calling convention. If the processor
6038 has a dedicated frame pointer register, and the function has a frame,
6039 then @code{__builtin_frame_address} will return the value of the frame
6042 On some machines it may be impossible to determine the frame address of
6043 any function other than the current one; in such cases, or when the top
6044 of the stack has been reached, this function will return @code{0} if
6045 the first frame pointer is properly initialized by the startup code.
6047 This function should only be used with a nonzero argument for debugging
6051 @node Vector Extensions
6052 @section Using vector instructions through built-in functions
6054 On some targets, the instruction set contains SIMD vector instructions that
6055 operate on multiple values contained in one large register at the same time.
6056 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
6059 The first step in using these extensions is to provide the necessary data
6060 types. This should be done using an appropriate @code{typedef}:
6063 typedef int v4si __attribute__ ((vector_size (16)));
6066 The @code{int} type specifies the base type, while the attribute specifies
6067 the vector size for the variable, measured in bytes. For example, the
6068 declaration above causes the compiler to set the mode for the @code{v4si}
6069 type to be 16 bytes wide and divided into @code{int} sized units. For
6070 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
6071 corresponding mode of @code{foo} will be @acronym{V4SI}.
6073 The @code{vector_size} attribute is only applicable to integral and
6074 float scalars, although arrays, pointers, and function return values
6075 are allowed in conjunction with this construct.
6077 All the basic integer types can be used as base types, both as signed
6078 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
6079 @code{long long}. In addition, @code{float} and @code{double} can be
6080 used to build floating-point vector types.
6082 Specifying a combination that is not valid for the current architecture
6083 will cause GCC to synthesize the instructions using a narrower mode.
6084 For example, if you specify a variable of type @code{V4SI} and your
6085 architecture does not allow for this specific SIMD type, GCC will
6086 produce code that uses 4 @code{SIs}.
6088 The types defined in this manner can be used with a subset of normal C
6089 operations. Currently, GCC will allow using the following operators
6090 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~, %}@.
6092 The operations behave like C++ @code{valarrays}. Addition is defined as
6093 the addition of the corresponding elements of the operands. For
6094 example, in the code below, each of the 4 elements in @var{a} will be
6095 added to the corresponding 4 elements in @var{b} and the resulting
6096 vector will be stored in @var{c}.
6099 typedef int v4si __attribute__ ((vector_size (16)));
6106 Subtraction, multiplication, division, and the logical operations
6107 operate in a similar manner. Likewise, the result of using the unary
6108 minus or complement operators on a vector type is a vector whose
6109 elements are the negative or complemented values of the corresponding
6110 elements in the operand.
6112 You can declare variables and use them in function calls and returns, as
6113 well as in assignments and some casts. You can specify a vector type as
6114 a return type for a function. Vector types can also be used as function
6115 arguments. It is possible to cast from one vector type to another,
6116 provided they are of the same size (in fact, you can also cast vectors
6117 to and from other datatypes of the same size).
6119 You cannot operate between vectors of different lengths or different
6120 signedness without a cast.
6122 A port that supports hardware vector operations, usually provides a set
6123 of built-in functions that can be used to operate on vectors. For
6124 example, a function to add two vectors and multiply the result by a
6125 third could look like this:
6128 v4si f (v4si a, v4si b, v4si c)
6130 v4si tmp = __builtin_addv4si (a, b);
6131 return __builtin_mulv4si (tmp, c);
6138 @findex __builtin_offsetof
6140 GCC implements for both C and C++ a syntactic extension to implement
6141 the @code{offsetof} macro.
6145 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
6147 offsetof_member_designator:
6149 | offsetof_member_designator "." @code{identifier}
6150 | offsetof_member_designator "[" @code{expr} "]"
6153 This extension is sufficient such that
6156 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
6159 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
6160 may be dependent. In either case, @var{member} may consist of a single
6161 identifier, or a sequence of member accesses and array references.
6163 @node Atomic Builtins
6164 @section Built-in functions for atomic memory access
6166 The following builtins are intended to be compatible with those described
6167 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
6168 section 7.4. As such, they depart from the normal GCC practice of using
6169 the ``__builtin_'' prefix, and further that they are overloaded such that
6170 they work on multiple types.
6172 The definition given in the Intel documentation allows only for the use of
6173 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
6174 counterparts. GCC will allow any integral scalar or pointer type that is
6175 1, 2, 4 or 8 bytes in length.
6177 Not all operations are supported by all target processors. If a particular
6178 operation cannot be implemented on the target processor, a warning will be
6179 generated and a call an external function will be generated. The external
6180 function will carry the same name as the builtin, with an additional suffix
6181 @samp{_@var{n}} where @var{n} is the size of the data type.
6183 @c ??? Should we have a mechanism to suppress this warning? This is almost
6184 @c useful for implementing the operation under the control of an external
6187 In most cases, these builtins are considered a @dfn{full barrier}. That is,
6188 no memory operand will be moved across the operation, either forward or
6189 backward. Further, instructions will be issued as necessary to prevent the
6190 processor from speculating loads across the operation and from queuing stores
6191 after the operation.
6193 All of the routines are described in the Intel documentation to take
6194 ``an optional list of variables protected by the memory barrier''. It's
6195 not clear what is meant by that; it could mean that @emph{only} the
6196 following variables are protected, or it could mean that these variables
6197 should in addition be protected. At present GCC ignores this list and
6198 protects all variables which are globally accessible. If in the future
6199 we make some use of this list, an empty list will continue to mean all
6200 globally accessible variables.
6203 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
6204 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
6205 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
6206 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
6207 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
6208 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
6209 @findex __sync_fetch_and_add
6210 @findex __sync_fetch_and_sub
6211 @findex __sync_fetch_and_or
6212 @findex __sync_fetch_and_and
6213 @findex __sync_fetch_and_xor
6214 @findex __sync_fetch_and_nand
6215 These builtins perform the operation suggested by the name, and
6216 returns the value that had previously been in memory. That is,
6219 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
6220 @{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; @} // nand
6223 @emph{Note:} GCC 4.4 and later implement @code{__sync_fetch_and_nand}
6224 builtin as @code{*ptr = ~(tmp & value)} instead of @code{*ptr = ~tmp & value}.
6226 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
6227 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
6228 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
6229 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
6230 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
6231 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
6232 @findex __sync_add_and_fetch
6233 @findex __sync_sub_and_fetch
6234 @findex __sync_or_and_fetch
6235 @findex __sync_and_and_fetch
6236 @findex __sync_xor_and_fetch
6237 @findex __sync_nand_and_fetch
6238 These builtins perform the operation suggested by the name, and
6239 return the new value. That is,
6242 @{ *ptr @var{op}= value; return *ptr; @}
6243 @{ *ptr = ~(*ptr & value); return *ptr; @} // nand
6246 @emph{Note:} GCC 4.4 and later implement @code{__sync_nand_and_fetch}
6247 builtin as @code{*ptr = ~(*ptr & value)} instead of
6248 @code{*ptr = ~*ptr & value}.
6250 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6251 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
6252 @findex __sync_bool_compare_and_swap
6253 @findex __sync_val_compare_and_swap
6254 These builtins perform an atomic compare and swap. That is, if the current
6255 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
6258 The ``bool'' version returns true if the comparison is successful and
6259 @var{newval} was written. The ``val'' version returns the contents
6260 of @code{*@var{ptr}} before the operation.
6262 @item __sync_synchronize (...)
6263 @findex __sync_synchronize
6264 This builtin issues a full memory barrier.
6266 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
6267 @findex __sync_lock_test_and_set
6268 This builtin, as described by Intel, is not a traditional test-and-set
6269 operation, but rather an atomic exchange operation. It writes @var{value}
6270 into @code{*@var{ptr}}, and returns the previous contents of
6273 Many targets have only minimal support for such locks, and do not support
6274 a full exchange operation. In this case, a target may support reduced
6275 functionality here by which the @emph{only} valid value to store is the
6276 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
6277 is implementation defined.
6279 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
6280 This means that references after the builtin cannot move to (or be
6281 speculated to) before the builtin, but previous memory stores may not
6282 be globally visible yet, and previous memory loads may not yet be
6285 @item void __sync_lock_release (@var{type} *ptr, ...)
6286 @findex __sync_lock_release
6287 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
6288 Normally this means writing the constant 0 to @code{*@var{ptr}}.
6290 This builtin is not a full barrier, but rather a @dfn{release barrier}.
6291 This means that all previous memory stores are globally visible, and all
6292 previous memory loads have been satisfied, but following memory reads
6293 are not prevented from being speculated to before the barrier.
6296 @node Object Size Checking
6297 @section Object Size Checking Builtins
6298 @findex __builtin_object_size
6299 @findex __builtin___memcpy_chk
6300 @findex __builtin___mempcpy_chk
6301 @findex __builtin___memmove_chk
6302 @findex __builtin___memset_chk
6303 @findex __builtin___strcpy_chk
6304 @findex __builtin___stpcpy_chk
6305 @findex __builtin___strncpy_chk
6306 @findex __builtin___strcat_chk
6307 @findex __builtin___strncat_chk
6308 @findex __builtin___sprintf_chk
6309 @findex __builtin___snprintf_chk
6310 @findex __builtin___vsprintf_chk
6311 @findex __builtin___vsnprintf_chk
6312 @findex __builtin___printf_chk
6313 @findex __builtin___vprintf_chk
6314 @findex __builtin___fprintf_chk
6315 @findex __builtin___vfprintf_chk
6317 GCC implements a limited buffer overflow protection mechanism
6318 that can prevent some buffer overflow attacks.
6320 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
6321 is a built-in construct that returns a constant number of bytes from
6322 @var{ptr} to the end of the object @var{ptr} pointer points to
6323 (if known at compile time). @code{__builtin_object_size} never evaluates
6324 its arguments for side-effects. If there are any side-effects in them, it
6325 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6326 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
6327 point to and all of them are known at compile time, the returned number
6328 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
6329 0 and minimum if nonzero. If it is not possible to determine which objects
6330 @var{ptr} points to at compile time, @code{__builtin_object_size} should
6331 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
6332 for @var{type} 2 or 3.
6334 @var{type} is an integer constant from 0 to 3. If the least significant
6335 bit is clear, objects are whole variables, if it is set, a closest
6336 surrounding subobject is considered the object a pointer points to.
6337 The second bit determines if maximum or minimum of remaining bytes
6341 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
6342 char *p = &var.buf1[1], *q = &var.b;
6344 /* Here the object p points to is var. */
6345 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
6346 /* The subobject p points to is var.buf1. */
6347 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
6348 /* The object q points to is var. */
6349 assert (__builtin_object_size (q, 0)
6350 == (char *) (&var + 1) - (char *) &var.b);
6351 /* The subobject q points to is var.b. */
6352 assert (__builtin_object_size (q, 1) == sizeof (var.b));
6356 There are built-in functions added for many common string operation
6357 functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk}
6358 built-in is provided. This built-in has an additional last argument,
6359 which is the number of bytes remaining in object the @var{dest}
6360 argument points to or @code{(size_t) -1} if the size is not known.
6362 The built-in functions are optimized into the normal string functions
6363 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
6364 it is known at compile time that the destination object will not
6365 be overflown. If the compiler can determine at compile time the
6366 object will be always overflown, it issues a warning.
6368 The intended use can be e.g.
6372 #define bos0(dest) __builtin_object_size (dest, 0)
6373 #define memcpy(dest, src, n) \
6374 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
6378 /* It is unknown what object p points to, so this is optimized
6379 into plain memcpy - no checking is possible. */
6380 memcpy (p, "abcde", n);
6381 /* Destination is known and length too. It is known at compile
6382 time there will be no overflow. */
6383 memcpy (&buf[5], "abcde", 5);
6384 /* Destination is known, but the length is not known at compile time.
6385 This will result in __memcpy_chk call that can check for overflow
6387 memcpy (&buf[5], "abcde", n);
6388 /* Destination is known and it is known at compile time there will
6389 be overflow. There will be a warning and __memcpy_chk call that
6390 will abort the program at runtime. */
6391 memcpy (&buf[6], "abcde", 5);
6394 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
6395 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
6396 @code{strcat} and @code{strncat}.
6398 There are also checking built-in functions for formatted output functions.
6400 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
6401 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6402 const char *fmt, ...);
6403 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
6405 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
6406 const char *fmt, va_list ap);
6409 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
6410 etc.@: functions and can contain implementation specific flags on what
6411 additional security measures the checking function might take, such as
6412 handling @code{%n} differently.
6414 The @var{os} argument is the object size @var{s} points to, like in the
6415 other built-in functions. There is a small difference in the behavior
6416 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
6417 optimized into the non-checking functions only if @var{flag} is 0, otherwise
6418 the checking function is called with @var{os} argument set to
6421 In addition to this, there are checking built-in functions
6422 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
6423 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
6424 These have just one additional argument, @var{flag}, right before
6425 format string @var{fmt}. If the compiler is able to optimize them to
6426 @code{fputc} etc.@: functions, it will, otherwise the checking function
6427 should be called and the @var{flag} argument passed to it.
6429 @node Other Builtins
6430 @section Other built-in functions provided by GCC
6431 @cindex built-in functions
6432 @findex __builtin_fpclassify
6433 @findex __builtin_isfinite
6434 @findex __builtin_isnormal
6435 @findex __builtin_isgreater
6436 @findex __builtin_isgreaterequal
6437 @findex __builtin_isinf_sign
6438 @findex __builtin_isless
6439 @findex __builtin_islessequal
6440 @findex __builtin_islessgreater
6441 @findex __builtin_isunordered
6442 @findex __builtin_powi
6443 @findex __builtin_powif
6444 @findex __builtin_powil
6602 @findex fprintf_unlocked
6604 @findex fputs_unlocked
6721 @findex printf_unlocked
6753 @findex significandf
6754 @findex significandl
6825 GCC provides a large number of built-in functions other than the ones
6826 mentioned above. Some of these are for internal use in the processing
6827 of exceptions or variable-length argument lists and will not be
6828 documented here because they may change from time to time; we do not
6829 recommend general use of these functions.
6831 The remaining functions are provided for optimization purposes.
6833 @opindex fno-builtin
6834 GCC includes built-in versions of many of the functions in the standard
6835 C library. The versions prefixed with @code{__builtin_} will always be
6836 treated as having the same meaning as the C library function even if you
6837 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
6838 Many of these functions are only optimized in certain cases; if they are
6839 not optimized in a particular case, a call to the library function will
6844 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
6845 @option{-std=c99}), the functions
6846 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
6847 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
6848 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
6849 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
6850 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
6851 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
6852 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
6853 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
6854 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
6855 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
6856 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
6857 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
6858 @code{signbitd64}, @code{signbitd128}, @code{significandf},
6859 @code{significandl}, @code{significand}, @code{sincosf},
6860 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
6861 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
6862 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
6863 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
6865 may be handled as built-in functions.
6866 All these functions have corresponding versions
6867 prefixed with @code{__builtin_}, which may be used even in strict C89
6870 The ISO C99 functions
6871 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
6872 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
6873 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
6874 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
6875 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
6876 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
6877 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
6878 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
6879 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
6880 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
6881 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
6882 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
6883 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
6884 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
6885 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
6886 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
6887 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
6888 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
6889 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
6890 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
6891 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
6892 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
6893 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
6894 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
6895 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
6896 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
6897 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
6898 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
6899 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
6900 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
6901 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
6902 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
6903 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
6904 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
6905 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
6906 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
6907 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
6908 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
6909 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
6910 are handled as built-in functions
6911 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
6913 There are also built-in versions of the ISO C99 functions
6914 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
6915 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
6916 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
6917 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
6918 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
6919 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
6920 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
6921 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
6922 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
6923 that are recognized in any mode since ISO C90 reserves these names for
6924 the purpose to which ISO C99 puts them. All these functions have
6925 corresponding versions prefixed with @code{__builtin_}.
6927 The ISO C94 functions
6928 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
6929 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
6930 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
6932 are handled as built-in functions
6933 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
6935 The ISO C90 functions
6936 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
6937 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
6938 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
6939 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
6940 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
6941 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
6942 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
6943 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
6944 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
6945 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
6946 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
6947 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
6948 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
6949 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
6950 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
6951 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
6952 are all recognized as built-in functions unless
6953 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
6954 is specified for an individual function). All of these functions have
6955 corresponding versions prefixed with @code{__builtin_}.
6957 GCC provides built-in versions of the ISO C99 floating point comparison
6958 macros that avoid raising exceptions for unordered operands. They have
6959 the same names as the standard macros ( @code{isgreater},
6960 @code{isgreaterequal}, @code{isless}, @code{islessequal},
6961 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
6962 prefixed. We intend for a library implementor to be able to simply
6963 @code{#define} each standard macro to its built-in equivalent.
6964 In the same fashion, GCC provides @code{fpclassify}, @code{isfinite},
6965 @code{isinf_sign} and @code{isnormal} built-ins used with
6966 @code{__builtin_} prefixed. The @code{isinf} and @code{isnan}
6967 builtins appear both with and without the @code{__builtin_} prefix.
6969 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
6971 You can use the built-in function @code{__builtin_types_compatible_p} to
6972 determine whether two types are the same.
6974 This built-in function returns 1 if the unqualified versions of the
6975 types @var{type1} and @var{type2} (which are types, not expressions) are
6976 compatible, 0 otherwise. The result of this built-in function can be
6977 used in integer constant expressions.
6979 This built-in function ignores top level qualifiers (e.g., @code{const},
6980 @code{volatile}). For example, @code{int} is equivalent to @code{const
6983 The type @code{int[]} and @code{int[5]} are compatible. On the other
6984 hand, @code{int} and @code{char *} are not compatible, even if the size
6985 of their types, on the particular architecture are the same. Also, the
6986 amount of pointer indirection is taken into account when determining
6987 similarity. Consequently, @code{short *} is not similar to
6988 @code{short **}. Furthermore, two types that are typedefed are
6989 considered compatible if their underlying types are compatible.
6991 An @code{enum} type is not considered to be compatible with another
6992 @code{enum} type even if both are compatible with the same integer
6993 type; this is what the C standard specifies.
6994 For example, @code{enum @{foo, bar@}} is not similar to
6995 @code{enum @{hot, dog@}}.
6997 You would typically use this function in code whose execution varies
6998 depending on the arguments' types. For example:
7003 typeof (x) tmp = (x); \
7004 if (__builtin_types_compatible_p (typeof (x), long double)) \
7005 tmp = foo_long_double (tmp); \
7006 else if (__builtin_types_compatible_p (typeof (x), double)) \
7007 tmp = foo_double (tmp); \
7008 else if (__builtin_types_compatible_p (typeof (x), float)) \
7009 tmp = foo_float (tmp); \
7016 @emph{Note:} This construct is only available for C@.
7020 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
7022 You can use the built-in function @code{__builtin_choose_expr} to
7023 evaluate code depending on the value of a constant expression. This
7024 built-in function returns @var{exp1} if @var{const_exp}, which is an
7025 integer constant expression, is nonzero. Otherwise it returns 0.
7027 This built-in function is analogous to the @samp{? :} operator in C,
7028 except that the expression returned has its type unaltered by promotion
7029 rules. Also, the built-in function does not evaluate the expression
7030 that was not chosen. For example, if @var{const_exp} evaluates to true,
7031 @var{exp2} is not evaluated even if it has side-effects.
7033 This built-in function can return an lvalue if the chosen argument is an
7036 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
7037 type. Similarly, if @var{exp2} is returned, its return type is the same
7044 __builtin_choose_expr ( \
7045 __builtin_types_compatible_p (typeof (x), double), \
7047 __builtin_choose_expr ( \
7048 __builtin_types_compatible_p (typeof (x), float), \
7050 /* @r{The void expression results in a compile-time error} \
7051 @r{when assigning the result to something.} */ \
7055 @emph{Note:} This construct is only available for C@. Furthermore, the
7056 unused expression (@var{exp1} or @var{exp2} depending on the value of
7057 @var{const_exp}) may still generate syntax errors. This may change in
7062 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
7063 You can use the built-in function @code{__builtin_constant_p} to
7064 determine if a value is known to be constant at compile-time and hence
7065 that GCC can perform constant-folding on expressions involving that
7066 value. The argument of the function is the value to test. The function
7067 returns the integer 1 if the argument is known to be a compile-time
7068 constant and 0 if it is not known to be a compile-time constant. A
7069 return of 0 does not indicate that the value is @emph{not} a constant,
7070 but merely that GCC cannot prove it is a constant with the specified
7071 value of the @option{-O} option.
7073 You would typically use this function in an embedded application where
7074 memory was a critical resource. If you have some complex calculation,
7075 you may want it to be folded if it involves constants, but need to call
7076 a function if it does not. For example:
7079 #define Scale_Value(X) \
7080 (__builtin_constant_p (X) \
7081 ? ((X) * SCALE + OFFSET) : Scale (X))
7084 You may use this built-in function in either a macro or an inline
7085 function. However, if you use it in an inlined function and pass an
7086 argument of the function as the argument to the built-in, GCC will
7087 never return 1 when you call the inline function with a string constant
7088 or compound literal (@pxref{Compound Literals}) and will not return 1
7089 when you pass a constant numeric value to the inline function unless you
7090 specify the @option{-O} option.
7092 You may also use @code{__builtin_constant_p} in initializers for static
7093 data. For instance, you can write
7096 static const int table[] = @{
7097 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
7103 This is an acceptable initializer even if @var{EXPRESSION} is not a
7104 constant expression, including the case where
7105 @code{__builtin_constant_p} returns 1 because @var{EXPRESSION} can be
7106 folded to a constant but @var{EXPRESSION} contains operands that would
7107 not otherwise be permitted in a static initializer (for example,
7108 @code{0 && foo ()}). GCC must be more conservative about evaluating the
7109 built-in in this case, because it has no opportunity to perform
7112 Previous versions of GCC did not accept this built-in in data
7113 initializers. The earliest version where it is completely safe is
7117 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
7118 @opindex fprofile-arcs
7119 You may use @code{__builtin_expect} to provide the compiler with
7120 branch prediction information. In general, you should prefer to
7121 use actual profile feedback for this (@option{-fprofile-arcs}), as
7122 programmers are notoriously bad at predicting how their programs
7123 actually perform. However, there are applications in which this
7124 data is hard to collect.
7126 The return value is the value of @var{exp}, which should be an integral
7127 expression. The semantics of the built-in are that it is expected that
7128 @var{exp} == @var{c}. For example:
7131 if (__builtin_expect (x, 0))
7136 would indicate that we do not expect to call @code{foo}, since
7137 we expect @code{x} to be zero. Since you are limited to integral
7138 expressions for @var{exp}, you should use constructions such as
7141 if (__builtin_expect (ptr != NULL, 1))
7146 when testing pointer or floating-point values.
7149 @deftypefn {Built-in Function} void __builtin_trap (void)
7150 This function causes the program to exit abnormally. GCC implements
7151 this function by using a target-dependent mechanism (such as
7152 intentionally executing an illegal instruction) or by calling
7153 @code{abort}. The mechanism used may vary from release to release so
7154 you should not rely on any particular implementation.
7157 @deftypefn {Built-in Function} void __builtin_unreachable (void)
7158 If control flow reaches the point of the @code{__builtin_unreachable},
7159 the program is undefined. It is useful in situations where the
7160 compiler cannot deduce the unreachability of the code.
7162 One such case is immediately following an @code{asm} statement that
7163 will either never terminate, or one that transfers control elsewhere
7164 and never returns. In this example, without the
7165 @code{__builtin_unreachable}, GCC would issue a warning that control
7166 reaches the end of a non-void function. It would also generate code
7167 to return after the @code{asm}.
7170 int f (int c, int v)
7178 asm("jmp error_handler");
7179 __builtin_unreachable ();
7184 Because the @code{asm} statement unconditionally transfers control out
7185 of the function, control will never reach the end of the function
7186 body. The @code{__builtin_unreachable} is in fact unreachable and
7187 communicates this fact to the compiler.
7189 Another use for @code{__builtin_unreachable} is following a call a
7190 function that never returns but that is not declared
7191 @code{__attribute__((noreturn))}, as in this example:
7194 void function_that_never_returns (void);
7204 function_that_never_returns ();
7205 __builtin_unreachable ();
7212 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
7213 This function is used to flush the processor's instruction cache for
7214 the region of memory between @var{begin} inclusive and @var{end}
7215 exclusive. Some targets require that the instruction cache be
7216 flushed, after modifying memory containing code, in order to obtain
7217 deterministic behavior.
7219 If the target does not require instruction cache flushes,
7220 @code{__builtin___clear_cache} has no effect. Otherwise either
7221 instructions are emitted in-line to clear the instruction cache or a
7222 call to the @code{__clear_cache} function in libgcc is made.
7225 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
7226 This function is used to minimize cache-miss latency by moving data into
7227 a cache before it is accessed.
7228 You can insert calls to @code{__builtin_prefetch} into code for which
7229 you know addresses of data in memory that is likely to be accessed soon.
7230 If the target supports them, data prefetch instructions will be generated.
7231 If the prefetch is done early enough before the access then the data will
7232 be in the cache by the time it is accessed.
7234 The value of @var{addr} is the address of the memory to prefetch.
7235 There are two optional arguments, @var{rw} and @var{locality}.
7236 The value of @var{rw} is a compile-time constant one or zero; one
7237 means that the prefetch is preparing for a write to the memory address
7238 and zero, the default, means that the prefetch is preparing for a read.
7239 The value @var{locality} must be a compile-time constant integer between
7240 zero and three. A value of zero means that the data has no temporal
7241 locality, so it need not be left in the cache after the access. A value
7242 of three means that the data has a high degree of temporal locality and
7243 should be left in all levels of cache possible. Values of one and two
7244 mean, respectively, a low or moderate degree of temporal locality. The
7248 for (i = 0; i < n; i++)
7251 __builtin_prefetch (&a[i+j], 1, 1);
7252 __builtin_prefetch (&b[i+j], 0, 1);
7257 Data prefetch does not generate faults if @var{addr} is invalid, but
7258 the address expression itself must be valid. For example, a prefetch
7259 of @code{p->next} will not fault if @code{p->next} is not a valid
7260 address, but evaluation will fault if @code{p} is not a valid address.
7262 If the target does not support data prefetch, the address expression
7263 is evaluated if it includes side effects but no other code is generated
7264 and GCC does not issue a warning.
7267 @deftypefn {Built-in Function} double __builtin_huge_val (void)
7268 Returns a positive infinity, if supported by the floating-point format,
7269 else @code{DBL_MAX}. This function is suitable for implementing the
7270 ISO C macro @code{HUGE_VAL}.
7273 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
7274 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
7277 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
7278 Similar to @code{__builtin_huge_val}, except the return
7279 type is @code{long double}.
7282 @deftypefn {Built-in Function} int __builtin_fpclassify (int, int, int, int, int, ...)
7283 This built-in implements the C99 fpclassify functionality. The first
7284 five int arguments should be the target library's notion of the
7285 possible FP classes and are used for return values. They must be
7286 constant values and they must appear in this order: @code{FP_NAN},
7287 @code{FP_INFINITE}, @code{FP_NORMAL}, @code{FP_SUBNORMAL} and
7288 @code{FP_ZERO}. The ellipsis is for exactly one floating point value
7289 to classify. GCC treats the last argument as type-generic, which
7290 means it does not do default promotion from float to double.
7293 @deftypefn {Built-in Function} double __builtin_inf (void)
7294 Similar to @code{__builtin_huge_val}, except a warning is generated
7295 if the target floating-point format does not support infinities.
7298 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
7299 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
7302 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
7303 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
7306 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
7307 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
7310 @deftypefn {Built-in Function} float __builtin_inff (void)
7311 Similar to @code{__builtin_inf}, except the return type is @code{float}.
7312 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
7315 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
7316 Similar to @code{__builtin_inf}, except the return
7317 type is @code{long double}.
7320 @deftypefn {Built-in Function} int __builtin_isinf_sign (...)
7321 Similar to @code{isinf}, except the return value will be negative for
7322 an argument of @code{-Inf}. Note while the parameter list is an
7323 ellipsis, this function only accepts exactly one floating point
7324 argument. GCC treats this parameter as type-generic, which means it
7325 does not do default promotion from float to double.
7328 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
7329 This is an implementation of the ISO C99 function @code{nan}.
7331 Since ISO C99 defines this function in terms of @code{strtod}, which we
7332 do not implement, a description of the parsing is in order. The string
7333 is parsed as by @code{strtol}; that is, the base is recognized by
7334 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
7335 in the significand such that the least significant bit of the number
7336 is at the least significant bit of the significand. The number is
7337 truncated to fit the significand field provided. The significand is
7338 forced to be a quiet NaN@.
7340 This function, if given a string literal all of which would have been
7341 consumed by strtol, is evaluated early enough that it is considered a
7342 compile-time constant.
7345 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
7346 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
7349 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
7350 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
7353 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
7354 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
7357 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
7358 Similar to @code{__builtin_nan}, except the return type is @code{float}.
7361 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
7362 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
7365 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
7366 Similar to @code{__builtin_nan}, except the significand is forced
7367 to be a signaling NaN@. The @code{nans} function is proposed by
7368 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
7371 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
7372 Similar to @code{__builtin_nans}, except the return type is @code{float}.
7375 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
7376 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
7379 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
7380 Returns one plus the index of the least significant 1-bit of @var{x}, or
7381 if @var{x} is zero, returns zero.
7384 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
7385 Returns the number of leading 0-bits in @var{x}, starting at the most
7386 significant bit position. If @var{x} is 0, the result is undefined.
7389 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
7390 Returns the number of trailing 0-bits in @var{x}, starting at the least
7391 significant bit position. If @var{x} is 0, the result is undefined.
7394 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
7395 Returns the number of 1-bits in @var{x}.
7398 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
7399 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
7403 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
7404 Similar to @code{__builtin_ffs}, except the argument type is
7405 @code{unsigned long}.
7408 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
7409 Similar to @code{__builtin_clz}, except the argument type is
7410 @code{unsigned long}.
7413 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
7414 Similar to @code{__builtin_ctz}, except the argument type is
7415 @code{unsigned long}.
7418 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
7419 Similar to @code{__builtin_popcount}, except the argument type is
7420 @code{unsigned long}.
7423 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
7424 Similar to @code{__builtin_parity}, except the argument type is
7425 @code{unsigned long}.
7428 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
7429 Similar to @code{__builtin_ffs}, except the argument type is
7430 @code{unsigned long long}.
7433 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
7434 Similar to @code{__builtin_clz}, except the argument type is
7435 @code{unsigned long long}.
7438 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
7439 Similar to @code{__builtin_ctz}, except the argument type is
7440 @code{unsigned long long}.
7443 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
7444 Similar to @code{__builtin_popcount}, except the argument type is
7445 @code{unsigned long long}.
7448 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
7449 Similar to @code{__builtin_parity}, except the argument type is
7450 @code{unsigned long long}.
7453 @deftypefn {Built-in Function} double __builtin_powi (double, int)
7454 Returns the first argument raised to the power of the second. Unlike the
7455 @code{pow} function no guarantees about precision and rounding are made.
7458 @deftypefn {Built-in Function} float __builtin_powif (float, int)
7459 Similar to @code{__builtin_powi}, except the argument and return types
7463 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
7464 Similar to @code{__builtin_powi}, except the argument and return types
7465 are @code{long double}.
7468 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
7469 Returns @var{x} with the order of the bytes reversed; for example,
7470 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
7474 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
7475 Similar to @code{__builtin_bswap32}, except the argument and return types
7479 @node Target Builtins
7480 @section Built-in Functions Specific to Particular Target Machines
7482 On some target machines, GCC supports many built-in functions specific
7483 to those machines. Generally these generate calls to specific machine
7484 instructions, but allow the compiler to schedule those calls.
7487 * Alpha Built-in Functions::
7488 * ARM iWMMXt Built-in Functions::
7489 * ARM NEON Intrinsics::
7490 * Blackfin Built-in Functions::
7491 * FR-V Built-in Functions::
7492 * X86 Built-in Functions::
7493 * MIPS DSP Built-in Functions::
7494 * MIPS Paired-Single Support::
7495 * MIPS Loongson Built-in Functions::
7496 * Other MIPS Built-in Functions::
7497 * picoChip Built-in Functions::
7498 * PowerPC AltiVec/VSX Built-in Functions::
7499 * RX Built-in Functions::
7500 * SPARC VIS Built-in Functions::
7501 * SPU Built-in Functions::
7504 @node Alpha Built-in Functions
7505 @subsection Alpha Built-in Functions
7507 These built-in functions are available for the Alpha family of
7508 processors, depending on the command-line switches used.
7510 The following built-in functions are always available. They
7511 all generate the machine instruction that is part of the name.
7514 long __builtin_alpha_implver (void)
7515 long __builtin_alpha_rpcc (void)
7516 long __builtin_alpha_amask (long)
7517 long __builtin_alpha_cmpbge (long, long)
7518 long __builtin_alpha_extbl (long, long)
7519 long __builtin_alpha_extwl (long, long)
7520 long __builtin_alpha_extll (long, long)
7521 long __builtin_alpha_extql (long, long)
7522 long __builtin_alpha_extwh (long, long)
7523 long __builtin_alpha_extlh (long, long)
7524 long __builtin_alpha_extqh (long, long)
7525 long __builtin_alpha_insbl (long, long)
7526 long __builtin_alpha_inswl (long, long)
7527 long __builtin_alpha_insll (long, long)
7528 long __builtin_alpha_insql (long, long)
7529 long __builtin_alpha_inswh (long, long)
7530 long __builtin_alpha_inslh (long, long)
7531 long __builtin_alpha_insqh (long, long)
7532 long __builtin_alpha_mskbl (long, long)
7533 long __builtin_alpha_mskwl (long, long)
7534 long __builtin_alpha_mskll (long, long)
7535 long __builtin_alpha_mskql (long, long)
7536 long __builtin_alpha_mskwh (long, long)
7537 long __builtin_alpha_msklh (long, long)
7538 long __builtin_alpha_mskqh (long, long)
7539 long __builtin_alpha_umulh (long, long)
7540 long __builtin_alpha_zap (long, long)
7541 long __builtin_alpha_zapnot (long, long)
7544 The following built-in functions are always with @option{-mmax}
7545 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
7546 later. They all generate the machine instruction that is part
7550 long __builtin_alpha_pklb (long)
7551 long __builtin_alpha_pkwb (long)
7552 long __builtin_alpha_unpkbl (long)
7553 long __builtin_alpha_unpkbw (long)
7554 long __builtin_alpha_minub8 (long, long)
7555 long __builtin_alpha_minsb8 (long, long)
7556 long __builtin_alpha_minuw4 (long, long)
7557 long __builtin_alpha_minsw4 (long, long)
7558 long __builtin_alpha_maxub8 (long, long)
7559 long __builtin_alpha_maxsb8 (long, long)
7560 long __builtin_alpha_maxuw4 (long, long)
7561 long __builtin_alpha_maxsw4 (long, long)
7562 long __builtin_alpha_perr (long, long)
7565 The following built-in functions are always with @option{-mcix}
7566 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
7567 later. They all generate the machine instruction that is part
7571 long __builtin_alpha_cttz (long)
7572 long __builtin_alpha_ctlz (long)
7573 long __builtin_alpha_ctpop (long)
7576 The following builtins are available on systems that use the OSF/1
7577 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
7578 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
7579 @code{rdval} and @code{wrval}.
7582 void *__builtin_thread_pointer (void)
7583 void __builtin_set_thread_pointer (void *)
7586 @node ARM iWMMXt Built-in Functions
7587 @subsection ARM iWMMXt Built-in Functions
7589 These built-in functions are available for the ARM family of
7590 processors when the @option{-mcpu=iwmmxt} switch is used:
7593 typedef int v2si __attribute__ ((vector_size (8)));
7594 typedef short v4hi __attribute__ ((vector_size (8)));
7595 typedef char v8qi __attribute__ ((vector_size (8)));
7597 int __builtin_arm_getwcx (int)
7598 void __builtin_arm_setwcx (int, int)
7599 int __builtin_arm_textrmsb (v8qi, int)
7600 int __builtin_arm_textrmsh (v4hi, int)
7601 int __builtin_arm_textrmsw (v2si, int)
7602 int __builtin_arm_textrmub (v8qi, int)
7603 int __builtin_arm_textrmuh (v4hi, int)
7604 int __builtin_arm_textrmuw (v2si, int)
7605 v8qi __builtin_arm_tinsrb (v8qi, int)
7606 v4hi __builtin_arm_tinsrh (v4hi, int)
7607 v2si __builtin_arm_tinsrw (v2si, int)
7608 long long __builtin_arm_tmia (long long, int, int)
7609 long long __builtin_arm_tmiabb (long long, int, int)
7610 long long __builtin_arm_tmiabt (long long, int, int)
7611 long long __builtin_arm_tmiaph (long long, int, int)
7612 long long __builtin_arm_tmiatb (long long, int, int)
7613 long long __builtin_arm_tmiatt (long long, int, int)
7614 int __builtin_arm_tmovmskb (v8qi)
7615 int __builtin_arm_tmovmskh (v4hi)
7616 int __builtin_arm_tmovmskw (v2si)
7617 long long __builtin_arm_waccb (v8qi)
7618 long long __builtin_arm_wacch (v4hi)
7619 long long __builtin_arm_waccw (v2si)
7620 v8qi __builtin_arm_waddb (v8qi, v8qi)
7621 v8qi __builtin_arm_waddbss (v8qi, v8qi)
7622 v8qi __builtin_arm_waddbus (v8qi, v8qi)
7623 v4hi __builtin_arm_waddh (v4hi, v4hi)
7624 v4hi __builtin_arm_waddhss (v4hi, v4hi)
7625 v4hi __builtin_arm_waddhus (v4hi, v4hi)
7626 v2si __builtin_arm_waddw (v2si, v2si)
7627 v2si __builtin_arm_waddwss (v2si, v2si)
7628 v2si __builtin_arm_waddwus (v2si, v2si)
7629 v8qi __builtin_arm_walign (v8qi, v8qi, int)
7630 long long __builtin_arm_wand(long long, long long)
7631 long long __builtin_arm_wandn (long long, long long)
7632 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
7633 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
7634 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
7635 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
7636 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
7637 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
7638 v2si __builtin_arm_wcmpeqw (v2si, v2si)
7639 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
7640 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
7641 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
7642 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
7643 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
7644 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
7645 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
7646 long long __builtin_arm_wmacsz (v4hi, v4hi)
7647 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
7648 long long __builtin_arm_wmacuz (v4hi, v4hi)
7649 v4hi __builtin_arm_wmadds (v4hi, v4hi)
7650 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
7651 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
7652 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
7653 v2si __builtin_arm_wmaxsw (v2si, v2si)
7654 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
7655 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
7656 v2si __builtin_arm_wmaxuw (v2si, v2si)
7657 v8qi __builtin_arm_wminsb (v8qi, v8qi)
7658 v4hi __builtin_arm_wminsh (v4hi, v4hi)
7659 v2si __builtin_arm_wminsw (v2si, v2si)
7660 v8qi __builtin_arm_wminub (v8qi, v8qi)
7661 v4hi __builtin_arm_wminuh (v4hi, v4hi)
7662 v2si __builtin_arm_wminuw (v2si, v2si)
7663 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
7664 v4hi __builtin_arm_wmulul (v4hi, v4hi)
7665 v4hi __builtin_arm_wmulum (v4hi, v4hi)
7666 long long __builtin_arm_wor (long long, long long)
7667 v2si __builtin_arm_wpackdss (long long, long long)
7668 v2si __builtin_arm_wpackdus (long long, long long)
7669 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
7670 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
7671 v4hi __builtin_arm_wpackwss (v2si, v2si)
7672 v4hi __builtin_arm_wpackwus (v2si, v2si)
7673 long long __builtin_arm_wrord (long long, long long)
7674 long long __builtin_arm_wrordi (long long, int)
7675 v4hi __builtin_arm_wrorh (v4hi, long long)
7676 v4hi __builtin_arm_wrorhi (v4hi, int)
7677 v2si __builtin_arm_wrorw (v2si, long long)
7678 v2si __builtin_arm_wrorwi (v2si, int)
7679 v2si __builtin_arm_wsadb (v8qi, v8qi)
7680 v2si __builtin_arm_wsadbz (v8qi, v8qi)
7681 v2si __builtin_arm_wsadh (v4hi, v4hi)
7682 v2si __builtin_arm_wsadhz (v4hi, v4hi)
7683 v4hi __builtin_arm_wshufh (v4hi, int)
7684 long long __builtin_arm_wslld (long long, long long)
7685 long long __builtin_arm_wslldi (long long, int)
7686 v4hi __builtin_arm_wsllh (v4hi, long long)
7687 v4hi __builtin_arm_wsllhi (v4hi, int)
7688 v2si __builtin_arm_wsllw (v2si, long long)
7689 v2si __builtin_arm_wsllwi (v2si, int)
7690 long long __builtin_arm_wsrad (long long, long long)
7691 long long __builtin_arm_wsradi (long long, int)
7692 v4hi __builtin_arm_wsrah (v4hi, long long)
7693 v4hi __builtin_arm_wsrahi (v4hi, int)
7694 v2si __builtin_arm_wsraw (v2si, long long)
7695 v2si __builtin_arm_wsrawi (v2si, int)
7696 long long __builtin_arm_wsrld (long long, long long)
7697 long long __builtin_arm_wsrldi (long long, int)
7698 v4hi __builtin_arm_wsrlh (v4hi, long long)
7699 v4hi __builtin_arm_wsrlhi (v4hi, int)
7700 v2si __builtin_arm_wsrlw (v2si, long long)
7701 v2si __builtin_arm_wsrlwi (v2si, int)
7702 v8qi __builtin_arm_wsubb (v8qi, v8qi)
7703 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
7704 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
7705 v4hi __builtin_arm_wsubh (v4hi, v4hi)
7706 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
7707 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
7708 v2si __builtin_arm_wsubw (v2si, v2si)
7709 v2si __builtin_arm_wsubwss (v2si, v2si)
7710 v2si __builtin_arm_wsubwus (v2si, v2si)
7711 v4hi __builtin_arm_wunpckehsb (v8qi)
7712 v2si __builtin_arm_wunpckehsh (v4hi)
7713 long long __builtin_arm_wunpckehsw (v2si)
7714 v4hi __builtin_arm_wunpckehub (v8qi)
7715 v2si __builtin_arm_wunpckehuh (v4hi)
7716 long long __builtin_arm_wunpckehuw (v2si)
7717 v4hi __builtin_arm_wunpckelsb (v8qi)
7718 v2si __builtin_arm_wunpckelsh (v4hi)
7719 long long __builtin_arm_wunpckelsw (v2si)
7720 v4hi __builtin_arm_wunpckelub (v8qi)
7721 v2si __builtin_arm_wunpckeluh (v4hi)
7722 long long __builtin_arm_wunpckeluw (v2si)
7723 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
7724 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
7725 v2si __builtin_arm_wunpckihw (v2si, v2si)
7726 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
7727 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
7728 v2si __builtin_arm_wunpckilw (v2si, v2si)
7729 long long __builtin_arm_wxor (long long, long long)
7730 long long __builtin_arm_wzero ()
7733 @node ARM NEON Intrinsics
7734 @subsection ARM NEON Intrinsics
7736 These built-in intrinsics for the ARM Advanced SIMD extension are available
7737 when the @option{-mfpu=neon} switch is used:
7739 @include arm-neon-intrinsics.texi
7741 @node Blackfin Built-in Functions
7742 @subsection Blackfin Built-in Functions
7744 Currently, there are two Blackfin-specific built-in functions. These are
7745 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
7746 using inline assembly; by using these built-in functions the compiler can
7747 automatically add workarounds for hardware errata involving these
7748 instructions. These functions are named as follows:
7751 void __builtin_bfin_csync (void)
7752 void __builtin_bfin_ssync (void)
7755 @node FR-V Built-in Functions
7756 @subsection FR-V Built-in Functions
7758 GCC provides many FR-V-specific built-in functions. In general,
7759 these functions are intended to be compatible with those described
7760 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
7761 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
7762 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
7763 pointer rather than by value.
7765 Most of the functions are named after specific FR-V instructions.
7766 Such functions are said to be ``directly mapped'' and are summarized
7767 here in tabular form.
7771 * Directly-mapped Integer Functions::
7772 * Directly-mapped Media Functions::
7773 * Raw read/write Functions::
7774 * Other Built-in Functions::
7777 @node Argument Types
7778 @subsubsection Argument Types
7780 The arguments to the built-in functions can be divided into three groups:
7781 register numbers, compile-time constants and run-time values. In order
7782 to make this classification clear at a glance, the arguments and return
7783 values are given the following pseudo types:
7785 @multitable @columnfractions .20 .30 .15 .35
7786 @item Pseudo type @tab Real C type @tab Constant? @tab Description
7787 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
7788 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
7789 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
7790 @item @code{uw2} @tab @code{unsigned long long} @tab No
7791 @tab an unsigned doubleword
7792 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
7793 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
7794 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
7795 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
7798 These pseudo types are not defined by GCC, they are simply a notational
7799 convenience used in this manual.
7801 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
7802 and @code{sw2} are evaluated at run time. They correspond to
7803 register operands in the underlying FR-V instructions.
7805 @code{const} arguments represent immediate operands in the underlying
7806 FR-V instructions. They must be compile-time constants.
7808 @code{acc} arguments are evaluated at compile time and specify the number
7809 of an accumulator register. For example, an @code{acc} argument of 2
7810 will select the ACC2 register.
7812 @code{iacc} arguments are similar to @code{acc} arguments but specify the
7813 number of an IACC register. See @pxref{Other Built-in Functions}
7816 @node Directly-mapped Integer Functions
7817 @subsubsection Directly-mapped Integer Functions
7819 The functions listed below map directly to FR-V I-type instructions.
7821 @multitable @columnfractions .45 .32 .23
7822 @item Function prototype @tab Example usage @tab Assembly output
7823 @item @code{sw1 __ADDSS (sw1, sw1)}
7824 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
7825 @tab @code{ADDSS @var{a},@var{b},@var{c}}
7826 @item @code{sw1 __SCAN (sw1, sw1)}
7827 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
7828 @tab @code{SCAN @var{a},@var{b},@var{c}}
7829 @item @code{sw1 __SCUTSS (sw1)}
7830 @tab @code{@var{b} = __SCUTSS (@var{a})}
7831 @tab @code{SCUTSS @var{a},@var{b}}
7832 @item @code{sw1 __SLASS (sw1, sw1)}
7833 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
7834 @tab @code{SLASS @var{a},@var{b},@var{c}}
7835 @item @code{void __SMASS (sw1, sw1)}
7836 @tab @code{__SMASS (@var{a}, @var{b})}
7837 @tab @code{SMASS @var{a},@var{b}}
7838 @item @code{void __SMSSS (sw1, sw1)}
7839 @tab @code{__SMSSS (@var{a}, @var{b})}
7840 @tab @code{SMSSS @var{a},@var{b}}
7841 @item @code{void __SMU (sw1, sw1)}
7842 @tab @code{__SMU (@var{a}, @var{b})}
7843 @tab @code{SMU @var{a},@var{b}}
7844 @item @code{sw2 __SMUL (sw1, sw1)}
7845 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
7846 @tab @code{SMUL @var{a},@var{b},@var{c}}
7847 @item @code{sw1 __SUBSS (sw1, sw1)}
7848 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
7849 @tab @code{SUBSS @var{a},@var{b},@var{c}}
7850 @item @code{uw2 __UMUL (uw1, uw1)}
7851 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
7852 @tab @code{UMUL @var{a},@var{b},@var{c}}
7855 @node Directly-mapped Media Functions
7856 @subsubsection Directly-mapped Media Functions
7858 The functions listed below map directly to FR-V M-type instructions.
7860 @multitable @columnfractions .45 .32 .23
7861 @item Function prototype @tab Example usage @tab Assembly output
7862 @item @code{uw1 __MABSHS (sw1)}
7863 @tab @code{@var{b} = __MABSHS (@var{a})}
7864 @tab @code{MABSHS @var{a},@var{b}}
7865 @item @code{void __MADDACCS (acc, acc)}
7866 @tab @code{__MADDACCS (@var{b}, @var{a})}
7867 @tab @code{MADDACCS @var{a},@var{b}}
7868 @item @code{sw1 __MADDHSS (sw1, sw1)}
7869 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
7870 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
7871 @item @code{uw1 __MADDHUS (uw1, uw1)}
7872 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
7873 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
7874 @item @code{uw1 __MAND (uw1, uw1)}
7875 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
7876 @tab @code{MAND @var{a},@var{b},@var{c}}
7877 @item @code{void __MASACCS (acc, acc)}
7878 @tab @code{__MASACCS (@var{b}, @var{a})}
7879 @tab @code{MASACCS @var{a},@var{b}}
7880 @item @code{uw1 __MAVEH (uw1, uw1)}
7881 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
7882 @tab @code{MAVEH @var{a},@var{b},@var{c}}
7883 @item @code{uw2 __MBTOH (uw1)}
7884 @tab @code{@var{b} = __MBTOH (@var{a})}
7885 @tab @code{MBTOH @var{a},@var{b}}
7886 @item @code{void __MBTOHE (uw1 *, uw1)}
7887 @tab @code{__MBTOHE (&@var{b}, @var{a})}
7888 @tab @code{MBTOHE @var{a},@var{b}}
7889 @item @code{void __MCLRACC (acc)}
7890 @tab @code{__MCLRACC (@var{a})}
7891 @tab @code{MCLRACC @var{a}}
7892 @item @code{void __MCLRACCA (void)}
7893 @tab @code{__MCLRACCA ()}
7894 @tab @code{MCLRACCA}
7895 @item @code{uw1 __Mcop1 (uw1, uw1)}
7896 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
7897 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
7898 @item @code{uw1 __Mcop2 (uw1, uw1)}
7899 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
7900 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
7901 @item @code{uw1 __MCPLHI (uw2, const)}
7902 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
7903 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
7904 @item @code{uw1 __MCPLI (uw2, const)}
7905 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
7906 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
7907 @item @code{void __MCPXIS (acc, sw1, sw1)}
7908 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
7909 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
7910 @item @code{void __MCPXIU (acc, uw1, uw1)}
7911 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
7912 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
7913 @item @code{void __MCPXRS (acc, sw1, sw1)}
7914 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
7915 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
7916 @item @code{void __MCPXRU (acc, uw1, uw1)}
7917 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
7918 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
7919 @item @code{uw1 __MCUT (acc, uw1)}
7920 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
7921 @tab @code{MCUT @var{a},@var{b},@var{c}}
7922 @item @code{uw1 __MCUTSS (acc, sw1)}
7923 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
7924 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
7925 @item @code{void __MDADDACCS (acc, acc)}
7926 @tab @code{__MDADDACCS (@var{b}, @var{a})}
7927 @tab @code{MDADDACCS @var{a},@var{b}}
7928 @item @code{void __MDASACCS (acc, acc)}
7929 @tab @code{__MDASACCS (@var{b}, @var{a})}
7930 @tab @code{MDASACCS @var{a},@var{b}}
7931 @item @code{uw2 __MDCUTSSI (acc, const)}
7932 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
7933 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
7934 @item @code{uw2 __MDPACKH (uw2, uw2)}
7935 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
7936 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
7937 @item @code{uw2 __MDROTLI (uw2, const)}
7938 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
7939 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
7940 @item @code{void __MDSUBACCS (acc, acc)}
7941 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
7942 @tab @code{MDSUBACCS @var{a},@var{b}}
7943 @item @code{void __MDUNPACKH (uw1 *, uw2)}
7944 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
7945 @tab @code{MDUNPACKH @var{a},@var{b}}
7946 @item @code{uw2 __MEXPDHD (uw1, const)}
7947 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
7948 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
7949 @item @code{uw1 __MEXPDHW (uw1, const)}
7950 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
7951 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
7952 @item @code{uw1 __MHDSETH (uw1, const)}
7953 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
7954 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
7955 @item @code{sw1 __MHDSETS (const)}
7956 @tab @code{@var{b} = __MHDSETS (@var{a})}
7957 @tab @code{MHDSETS #@var{a},@var{b}}
7958 @item @code{uw1 __MHSETHIH (uw1, const)}
7959 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
7960 @tab @code{MHSETHIH #@var{a},@var{b}}
7961 @item @code{sw1 __MHSETHIS (sw1, const)}
7962 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
7963 @tab @code{MHSETHIS #@var{a},@var{b}}
7964 @item @code{uw1 __MHSETLOH (uw1, const)}
7965 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
7966 @tab @code{MHSETLOH #@var{a},@var{b}}
7967 @item @code{sw1 __MHSETLOS (sw1, const)}
7968 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
7969 @tab @code{MHSETLOS #@var{a},@var{b}}
7970 @item @code{uw1 __MHTOB (uw2)}
7971 @tab @code{@var{b} = __MHTOB (@var{a})}
7972 @tab @code{MHTOB @var{a},@var{b}}
7973 @item @code{void __MMACHS (acc, sw1, sw1)}
7974 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
7975 @tab @code{MMACHS @var{a},@var{b},@var{c}}
7976 @item @code{void __MMACHU (acc, uw1, uw1)}
7977 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
7978 @tab @code{MMACHU @var{a},@var{b},@var{c}}
7979 @item @code{void __MMRDHS (acc, sw1, sw1)}
7980 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
7981 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
7982 @item @code{void __MMRDHU (acc, uw1, uw1)}
7983 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
7984 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
7985 @item @code{void __MMULHS (acc, sw1, sw1)}
7986 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
7987 @tab @code{MMULHS @var{a},@var{b},@var{c}}
7988 @item @code{void __MMULHU (acc, uw1, uw1)}
7989 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
7990 @tab @code{MMULHU @var{a},@var{b},@var{c}}
7991 @item @code{void __MMULXHS (acc, sw1, sw1)}
7992 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
7993 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
7994 @item @code{void __MMULXHU (acc, uw1, uw1)}
7995 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
7996 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
7997 @item @code{uw1 __MNOT (uw1)}
7998 @tab @code{@var{b} = __MNOT (@var{a})}
7999 @tab @code{MNOT @var{a},@var{b}}
8000 @item @code{uw1 __MOR (uw1, uw1)}
8001 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
8002 @tab @code{MOR @var{a},@var{b},@var{c}}
8003 @item @code{uw1 __MPACKH (uh, uh)}
8004 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
8005 @tab @code{MPACKH @var{a},@var{b},@var{c}}
8006 @item @code{sw2 __MQADDHSS (sw2, sw2)}
8007 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
8008 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
8009 @item @code{uw2 __MQADDHUS (uw2, uw2)}
8010 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
8011 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
8012 @item @code{void __MQCPXIS (acc, sw2, sw2)}
8013 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
8014 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
8015 @item @code{void __MQCPXIU (acc, uw2, uw2)}
8016 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
8017 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
8018 @item @code{void __MQCPXRS (acc, sw2, sw2)}
8019 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
8020 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
8021 @item @code{void __MQCPXRU (acc, uw2, uw2)}
8022 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
8023 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
8024 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
8025 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
8026 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
8027 @item @code{sw2 __MQLMTHS (sw2, sw2)}
8028 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
8029 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
8030 @item @code{void __MQMACHS (acc, sw2, sw2)}
8031 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
8032 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
8033 @item @code{void __MQMACHU (acc, uw2, uw2)}
8034 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
8035 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
8036 @item @code{void __MQMACXHS (acc, sw2, sw2)}
8037 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
8038 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
8039 @item @code{void __MQMULHS (acc, sw2, sw2)}
8040 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
8041 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
8042 @item @code{void __MQMULHU (acc, uw2, uw2)}
8043 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
8044 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
8045 @item @code{void __MQMULXHS (acc, sw2, sw2)}
8046 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
8047 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
8048 @item @code{void __MQMULXHU (acc, uw2, uw2)}
8049 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
8050 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
8051 @item @code{sw2 __MQSATHS (sw2, sw2)}
8052 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
8053 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
8054 @item @code{uw2 __MQSLLHI (uw2, int)}
8055 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
8056 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
8057 @item @code{sw2 __MQSRAHI (sw2, int)}
8058 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
8059 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
8060 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
8061 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
8062 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
8063 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
8064 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
8065 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
8066 @item @code{void __MQXMACHS (acc, sw2, sw2)}
8067 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
8068 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
8069 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
8070 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
8071 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
8072 @item @code{uw1 __MRDACC (acc)}
8073 @tab @code{@var{b} = __MRDACC (@var{a})}
8074 @tab @code{MRDACC @var{a},@var{b}}
8075 @item @code{uw1 __MRDACCG (acc)}
8076 @tab @code{@var{b} = __MRDACCG (@var{a})}
8077 @tab @code{MRDACCG @var{a},@var{b}}
8078 @item @code{uw1 __MROTLI (uw1, const)}
8079 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
8080 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
8081 @item @code{uw1 __MROTRI (uw1, const)}
8082 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
8083 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
8084 @item @code{sw1 __MSATHS (sw1, sw1)}
8085 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
8086 @tab @code{MSATHS @var{a},@var{b},@var{c}}
8087 @item @code{uw1 __MSATHU (uw1, uw1)}
8088 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
8089 @tab @code{MSATHU @var{a},@var{b},@var{c}}
8090 @item @code{uw1 __MSLLHI (uw1, const)}
8091 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
8092 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
8093 @item @code{sw1 __MSRAHI (sw1, const)}
8094 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
8095 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
8096 @item @code{uw1 __MSRLHI (uw1, const)}
8097 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
8098 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
8099 @item @code{void __MSUBACCS (acc, acc)}
8100 @tab @code{__MSUBACCS (@var{b}, @var{a})}
8101 @tab @code{MSUBACCS @var{a},@var{b}}
8102 @item @code{sw1 __MSUBHSS (sw1, sw1)}
8103 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
8104 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
8105 @item @code{uw1 __MSUBHUS (uw1, uw1)}
8106 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
8107 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
8108 @item @code{void __MTRAP (void)}
8109 @tab @code{__MTRAP ()}
8111 @item @code{uw2 __MUNPACKH (uw1)}
8112 @tab @code{@var{b} = __MUNPACKH (@var{a})}
8113 @tab @code{MUNPACKH @var{a},@var{b}}
8114 @item @code{uw1 __MWCUT (uw2, uw1)}
8115 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
8116 @tab @code{MWCUT @var{a},@var{b},@var{c}}
8117 @item @code{void __MWTACC (acc, uw1)}
8118 @tab @code{__MWTACC (@var{b}, @var{a})}
8119 @tab @code{MWTACC @var{a},@var{b}}
8120 @item @code{void __MWTACCG (acc, uw1)}
8121 @tab @code{__MWTACCG (@var{b}, @var{a})}
8122 @tab @code{MWTACCG @var{a},@var{b}}
8123 @item @code{uw1 __MXOR (uw1, uw1)}
8124 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
8125 @tab @code{MXOR @var{a},@var{b},@var{c}}
8128 @node Raw read/write Functions
8129 @subsubsection Raw read/write Functions
8131 This sections describes built-in functions related to read and write
8132 instructions to access memory. These functions generate
8133 @code{membar} instructions to flush the I/O load and stores where
8134 appropriate, as described in Fujitsu's manual described above.
8138 @item unsigned char __builtin_read8 (void *@var{data})
8139 @item unsigned short __builtin_read16 (void *@var{data})
8140 @item unsigned long __builtin_read32 (void *@var{data})
8141 @item unsigned long long __builtin_read64 (void *@var{data})
8143 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
8144 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
8145 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
8146 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
8149 @node Other Built-in Functions
8150 @subsubsection Other Built-in Functions
8152 This section describes built-in functions that are not named after
8153 a specific FR-V instruction.
8156 @item sw2 __IACCreadll (iacc @var{reg})
8157 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
8158 for future expansion and must be 0.
8160 @item sw1 __IACCreadl (iacc @var{reg})
8161 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
8162 Other values of @var{reg} are rejected as invalid.
8164 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
8165 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
8166 is reserved for future expansion and must be 0.
8168 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
8169 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
8170 is 1. Other values of @var{reg} are rejected as invalid.
8172 @item void __data_prefetch0 (const void *@var{x})
8173 Use the @code{dcpl} instruction to load the contents of address @var{x}
8174 into the data cache.
8176 @item void __data_prefetch (const void *@var{x})
8177 Use the @code{nldub} instruction to load the contents of address @var{x}
8178 into the data cache. The instruction will be issued in slot I1@.
8181 @node X86 Built-in Functions
8182 @subsection X86 Built-in Functions
8184 These built-in functions are available for the i386 and x86-64 family
8185 of computers, depending on the command-line switches used.
8187 Note that, if you specify command-line switches such as @option{-msse},
8188 the compiler could use the extended instruction sets even if the built-ins
8189 are not used explicitly in the program. For this reason, applications
8190 which perform runtime CPU detection must compile separate files for each
8191 supported architecture, using the appropriate flags. In particular,
8192 the file containing the CPU detection code should be compiled without
8195 The following machine modes are available for use with MMX built-in functions
8196 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
8197 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
8198 vector of eight 8-bit integers. Some of the built-in functions operate on
8199 MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode.
8201 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
8202 of two 32-bit floating point values.
8204 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
8205 floating point values. Some instructions use a vector of four 32-bit
8206 integers, these use @code{V4SI}. Finally, some instructions operate on an
8207 entire vector register, interpreting it as a 128-bit integer, these use mode
8210 In 64-bit mode, the x86-64 family of processors uses additional built-in
8211 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
8212 floating point and @code{TC} 128-bit complex floating point values.
8214 The following floating point built-in functions are available in 64-bit
8215 mode. All of them implement the function that is part of the name.
8218 __float128 __builtin_fabsq (__float128)
8219 __float128 __builtin_copysignq (__float128, __float128)
8222 The following floating point built-in functions are made available in the
8226 @item __float128 __builtin_infq (void)
8227 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
8228 @findex __builtin_infq
8230 @item __float128 __builtin_huge_valq (void)
8231 Similar to @code{__builtin_huge_val}, except the return type is @code{__float128}.
8232 @findex __builtin_huge_valq
8235 The following built-in functions are made available by @option{-mmmx}.
8236 All of them generate the machine instruction that is part of the name.
8239 v8qi __builtin_ia32_paddb (v8qi, v8qi)
8240 v4hi __builtin_ia32_paddw (v4hi, v4hi)
8241 v2si __builtin_ia32_paddd (v2si, v2si)
8242 v8qi __builtin_ia32_psubb (v8qi, v8qi)
8243 v4hi __builtin_ia32_psubw (v4hi, v4hi)
8244 v2si __builtin_ia32_psubd (v2si, v2si)
8245 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
8246 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
8247 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
8248 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
8249 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
8250 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
8251 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
8252 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
8253 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
8254 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
8255 di __builtin_ia32_pand (di, di)
8256 di __builtin_ia32_pandn (di,di)
8257 di __builtin_ia32_por (di, di)
8258 di __builtin_ia32_pxor (di, di)
8259 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
8260 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
8261 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
8262 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
8263 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
8264 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
8265 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
8266 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
8267 v2si __builtin_ia32_punpckhdq (v2si, v2si)
8268 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
8269 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
8270 v2si __builtin_ia32_punpckldq (v2si, v2si)
8271 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
8272 v4hi __builtin_ia32_packssdw (v2si, v2si)
8273 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
8275 v4hi __builtin_ia32_psllw (v4hi, v4hi)
8276 v2si __builtin_ia32_pslld (v2si, v2si)
8277 v1di __builtin_ia32_psllq (v1di, v1di)
8278 v4hi __builtin_ia32_psrlw (v4hi, v4hi)
8279 v2si __builtin_ia32_psrld (v2si, v2si)
8280 v1di __builtin_ia32_psrlq (v1di, v1di)
8281 v4hi __builtin_ia32_psraw (v4hi, v4hi)
8282 v2si __builtin_ia32_psrad (v2si, v2si)
8283 v4hi __builtin_ia32_psllwi (v4hi, int)
8284 v2si __builtin_ia32_pslldi (v2si, int)
8285 v1di __builtin_ia32_psllqi (v1di, int)
8286 v4hi __builtin_ia32_psrlwi (v4hi, int)
8287 v2si __builtin_ia32_psrldi (v2si, int)
8288 v1di __builtin_ia32_psrlqi (v1di, int)
8289 v4hi __builtin_ia32_psrawi (v4hi, int)
8290 v2si __builtin_ia32_psradi (v2si, int)
8294 The following built-in functions are made available either with
8295 @option{-msse}, or with a combination of @option{-m3dnow} and
8296 @option{-march=athlon}. All of them generate the machine
8297 instruction that is part of the name.
8300 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
8301 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
8302 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
8303 v1di __builtin_ia32_psadbw (v8qi, v8qi)
8304 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
8305 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
8306 v8qi __builtin_ia32_pminub (v8qi, v8qi)
8307 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
8308 int __builtin_ia32_pextrw (v4hi, int)
8309 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
8310 int __builtin_ia32_pmovmskb (v8qi)
8311 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
8312 void __builtin_ia32_movntq (di *, di)
8313 void __builtin_ia32_sfence (void)
8316 The following built-in functions are available when @option{-msse} is used.
8317 All of them generate the machine instruction that is part of the name.
8320 int __builtin_ia32_comieq (v4sf, v4sf)
8321 int __builtin_ia32_comineq (v4sf, v4sf)
8322 int __builtin_ia32_comilt (v4sf, v4sf)
8323 int __builtin_ia32_comile (v4sf, v4sf)
8324 int __builtin_ia32_comigt (v4sf, v4sf)
8325 int __builtin_ia32_comige (v4sf, v4sf)
8326 int __builtin_ia32_ucomieq (v4sf, v4sf)
8327 int __builtin_ia32_ucomineq (v4sf, v4sf)
8328 int __builtin_ia32_ucomilt (v4sf, v4sf)
8329 int __builtin_ia32_ucomile (v4sf, v4sf)
8330 int __builtin_ia32_ucomigt (v4sf, v4sf)
8331 int __builtin_ia32_ucomige (v4sf, v4sf)
8332 v4sf __builtin_ia32_addps (v4sf, v4sf)
8333 v4sf __builtin_ia32_subps (v4sf, v4sf)
8334 v4sf __builtin_ia32_mulps (v4sf, v4sf)
8335 v4sf __builtin_ia32_divps (v4sf, v4sf)
8336 v4sf __builtin_ia32_addss (v4sf, v4sf)
8337 v4sf __builtin_ia32_subss (v4sf, v4sf)
8338 v4sf __builtin_ia32_mulss (v4sf, v4sf)
8339 v4sf __builtin_ia32_divss (v4sf, v4sf)
8340 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
8341 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
8342 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
8343 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
8344 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
8345 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
8346 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
8347 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
8348 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
8349 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
8350 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
8351 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
8352 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
8353 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
8354 v4si __builtin_ia32_cmpless (v4sf, v4sf)
8355 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
8356 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
8357 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
8358 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
8359 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
8360 v4sf __builtin_ia32_maxps (v4sf, v4sf)
8361 v4sf __builtin_ia32_maxss (v4sf, v4sf)
8362 v4sf __builtin_ia32_minps (v4sf, v4sf)
8363 v4sf __builtin_ia32_minss (v4sf, v4sf)
8364 v4sf __builtin_ia32_andps (v4sf, v4sf)
8365 v4sf __builtin_ia32_andnps (v4sf, v4sf)
8366 v4sf __builtin_ia32_orps (v4sf, v4sf)
8367 v4sf __builtin_ia32_xorps (v4sf, v4sf)
8368 v4sf __builtin_ia32_movss (v4sf, v4sf)
8369 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
8370 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
8371 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
8372 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
8373 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
8374 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
8375 v2si __builtin_ia32_cvtps2pi (v4sf)
8376 int __builtin_ia32_cvtss2si (v4sf)
8377 v2si __builtin_ia32_cvttps2pi (v4sf)
8378 int __builtin_ia32_cvttss2si (v4sf)
8379 v4sf __builtin_ia32_rcpps (v4sf)
8380 v4sf __builtin_ia32_rsqrtps (v4sf)
8381 v4sf __builtin_ia32_sqrtps (v4sf)
8382 v4sf __builtin_ia32_rcpss (v4sf)
8383 v4sf __builtin_ia32_rsqrtss (v4sf)
8384 v4sf __builtin_ia32_sqrtss (v4sf)
8385 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
8386 void __builtin_ia32_movntps (float *, v4sf)
8387 int __builtin_ia32_movmskps (v4sf)
8390 The following built-in functions are available when @option{-msse} is used.
8393 @item v4sf __builtin_ia32_loadaps (float *)
8394 Generates the @code{movaps} machine instruction as a load from memory.
8395 @item void __builtin_ia32_storeaps (float *, v4sf)
8396 Generates the @code{movaps} machine instruction as a store to memory.
8397 @item v4sf __builtin_ia32_loadups (float *)
8398 Generates the @code{movups} machine instruction as a load from memory.
8399 @item void __builtin_ia32_storeups (float *, v4sf)
8400 Generates the @code{movups} machine instruction as a store to memory.
8401 @item v4sf __builtin_ia32_loadsss (float *)
8402 Generates the @code{movss} machine instruction as a load from memory.
8403 @item void __builtin_ia32_storess (float *, v4sf)
8404 Generates the @code{movss} machine instruction as a store to memory.
8405 @item v4sf __builtin_ia32_loadhps (v4sf, const v2sf *)
8406 Generates the @code{movhps} machine instruction as a load from memory.
8407 @item v4sf __builtin_ia32_loadlps (v4sf, const v2sf *)
8408 Generates the @code{movlps} machine instruction as a load from memory
8409 @item void __builtin_ia32_storehps (v2sf *, v4sf)
8410 Generates the @code{movhps} machine instruction as a store to memory.
8411 @item void __builtin_ia32_storelps (v2sf *, v4sf)
8412 Generates the @code{movlps} machine instruction as a store to memory.
8415 The following built-in functions are available when @option{-msse2} is used.
8416 All of them generate the machine instruction that is part of the name.
8419 int __builtin_ia32_comisdeq (v2df, v2df)
8420 int __builtin_ia32_comisdlt (v2df, v2df)
8421 int __builtin_ia32_comisdle (v2df, v2df)
8422 int __builtin_ia32_comisdgt (v2df, v2df)
8423 int __builtin_ia32_comisdge (v2df, v2df)
8424 int __builtin_ia32_comisdneq (v2df, v2df)
8425 int __builtin_ia32_ucomisdeq (v2df, v2df)
8426 int __builtin_ia32_ucomisdlt (v2df, v2df)
8427 int __builtin_ia32_ucomisdle (v2df, v2df)
8428 int __builtin_ia32_ucomisdgt (v2df, v2df)
8429 int __builtin_ia32_ucomisdge (v2df, v2df)
8430 int __builtin_ia32_ucomisdneq (v2df, v2df)
8431 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
8432 v2df __builtin_ia32_cmpltpd (v2df, v2df)
8433 v2df __builtin_ia32_cmplepd (v2df, v2df)
8434 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
8435 v2df __builtin_ia32_cmpgepd (v2df, v2df)
8436 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
8437 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
8438 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
8439 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
8440 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
8441 v2df __builtin_ia32_cmpngepd (v2df, v2df)
8442 v2df __builtin_ia32_cmpordpd (v2df, v2df)
8443 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
8444 v2df __builtin_ia32_cmpltsd (v2df, v2df)
8445 v2df __builtin_ia32_cmplesd (v2df, v2df)
8446 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
8447 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
8448 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
8449 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
8450 v2df __builtin_ia32_cmpordsd (v2df, v2df)
8451 v2di __builtin_ia32_paddq (v2di, v2di)
8452 v2di __builtin_ia32_psubq (v2di, v2di)
8453 v2df __builtin_ia32_addpd (v2df, v2df)
8454 v2df __builtin_ia32_subpd (v2df, v2df)
8455 v2df __builtin_ia32_mulpd (v2df, v2df)
8456 v2df __builtin_ia32_divpd (v2df, v2df)
8457 v2df __builtin_ia32_addsd (v2df, v2df)
8458 v2df __builtin_ia32_subsd (v2df, v2df)
8459 v2df __builtin_ia32_mulsd (v2df, v2df)
8460 v2df __builtin_ia32_divsd (v2df, v2df)
8461 v2df __builtin_ia32_minpd (v2df, v2df)
8462 v2df __builtin_ia32_maxpd (v2df, v2df)
8463 v2df __builtin_ia32_minsd (v2df, v2df)
8464 v2df __builtin_ia32_maxsd (v2df, v2df)
8465 v2df __builtin_ia32_andpd (v2df, v2df)
8466 v2df __builtin_ia32_andnpd (v2df, v2df)
8467 v2df __builtin_ia32_orpd (v2df, v2df)
8468 v2df __builtin_ia32_xorpd (v2df, v2df)
8469 v2df __builtin_ia32_movsd (v2df, v2df)
8470 v2df __builtin_ia32_unpckhpd (v2df, v2df)
8471 v2df __builtin_ia32_unpcklpd (v2df, v2df)
8472 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
8473 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
8474 v4si __builtin_ia32_paddd128 (v4si, v4si)
8475 v2di __builtin_ia32_paddq128 (v2di, v2di)
8476 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
8477 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
8478 v4si __builtin_ia32_psubd128 (v4si, v4si)
8479 v2di __builtin_ia32_psubq128 (v2di, v2di)
8480 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
8481 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
8482 v2di __builtin_ia32_pand128 (v2di, v2di)
8483 v2di __builtin_ia32_pandn128 (v2di, v2di)
8484 v2di __builtin_ia32_por128 (v2di, v2di)
8485 v2di __builtin_ia32_pxor128 (v2di, v2di)
8486 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
8487 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
8488 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
8489 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
8490 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
8491 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
8492 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
8493 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
8494 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
8495 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
8496 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
8497 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
8498 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
8499 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
8500 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
8501 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
8502 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
8503 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
8504 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
8505 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
8506 v16qi __builtin_ia32_packsswb128 (v8hi, v8hi)
8507 v8hi __builtin_ia32_packssdw128 (v4si, v4si)
8508 v16qi __builtin_ia32_packuswb128 (v8hi, v8hi)
8509 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
8510 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
8511 v2df __builtin_ia32_loadupd (double *)
8512 void __builtin_ia32_storeupd (double *, v2df)
8513 v2df __builtin_ia32_loadhpd (v2df, double const *)
8514 v2df __builtin_ia32_loadlpd (v2df, double const *)
8515 int __builtin_ia32_movmskpd (v2df)
8516 int __builtin_ia32_pmovmskb128 (v16qi)
8517 void __builtin_ia32_movnti (int *, int)
8518 void __builtin_ia32_movntpd (double *, v2df)
8519 void __builtin_ia32_movntdq (v2df *, v2df)
8520 v4si __builtin_ia32_pshufd (v4si, int)
8521 v8hi __builtin_ia32_pshuflw (v8hi, int)
8522 v8hi __builtin_ia32_pshufhw (v8hi, int)
8523 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
8524 v2df __builtin_ia32_sqrtpd (v2df)
8525 v2df __builtin_ia32_sqrtsd (v2df)
8526 v2df __builtin_ia32_shufpd (v2df, v2df, int)
8527 v2df __builtin_ia32_cvtdq2pd (v4si)
8528 v4sf __builtin_ia32_cvtdq2ps (v4si)
8529 v4si __builtin_ia32_cvtpd2dq (v2df)
8530 v2si __builtin_ia32_cvtpd2pi (v2df)
8531 v4sf __builtin_ia32_cvtpd2ps (v2df)
8532 v4si __builtin_ia32_cvttpd2dq (v2df)
8533 v2si __builtin_ia32_cvttpd2pi (v2df)
8534 v2df __builtin_ia32_cvtpi2pd (v2si)
8535 int __builtin_ia32_cvtsd2si (v2df)
8536 int __builtin_ia32_cvttsd2si (v2df)
8537 long long __builtin_ia32_cvtsd2si64 (v2df)
8538 long long __builtin_ia32_cvttsd2si64 (v2df)
8539 v4si __builtin_ia32_cvtps2dq (v4sf)
8540 v2df __builtin_ia32_cvtps2pd (v4sf)
8541 v4si __builtin_ia32_cvttps2dq (v4sf)
8542 v2df __builtin_ia32_cvtsi2sd (v2df, int)
8543 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
8544 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
8545 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
8546 void __builtin_ia32_clflush (const void *)
8547 void __builtin_ia32_lfence (void)
8548 void __builtin_ia32_mfence (void)
8549 v16qi __builtin_ia32_loaddqu (const char *)
8550 void __builtin_ia32_storedqu (char *, v16qi)
8551 v1di __builtin_ia32_pmuludq (v2si, v2si)
8552 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
8553 v8hi __builtin_ia32_psllw128 (v8hi, v8hi)
8554 v4si __builtin_ia32_pslld128 (v4si, v4si)
8555 v2di __builtin_ia32_psllq128 (v2di, v2di)
8556 v8hi __builtin_ia32_psrlw128 (v8hi, v8hi)
8557 v4si __builtin_ia32_psrld128 (v4si, v4si)
8558 v2di __builtin_ia32_psrlq128 (v2di, v2di)
8559 v8hi __builtin_ia32_psraw128 (v8hi, v8hi)
8560 v4si __builtin_ia32_psrad128 (v4si, v4si)
8561 v2di __builtin_ia32_pslldqi128 (v2di, int)
8562 v8hi __builtin_ia32_psllwi128 (v8hi, int)
8563 v4si __builtin_ia32_pslldi128 (v4si, int)
8564 v2di __builtin_ia32_psllqi128 (v2di, int)
8565 v2di __builtin_ia32_psrldqi128 (v2di, int)
8566 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
8567 v4si __builtin_ia32_psrldi128 (v4si, int)
8568 v2di __builtin_ia32_psrlqi128 (v2di, int)
8569 v8hi __builtin_ia32_psrawi128 (v8hi, int)
8570 v4si __builtin_ia32_psradi128 (v4si, int)
8571 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
8572 v2di __builtin_ia32_movq128 (v2di)
8575 The following built-in functions are available when @option{-msse3} is used.
8576 All of them generate the machine instruction that is part of the name.
8579 v2df __builtin_ia32_addsubpd (v2df, v2df)
8580 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
8581 v2df __builtin_ia32_haddpd (v2df, v2df)
8582 v4sf __builtin_ia32_haddps (v4sf, v4sf)
8583 v2df __builtin_ia32_hsubpd (v2df, v2df)
8584 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
8585 v16qi __builtin_ia32_lddqu (char const *)
8586 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
8587 v2df __builtin_ia32_movddup (v2df)
8588 v4sf __builtin_ia32_movshdup (v4sf)
8589 v4sf __builtin_ia32_movsldup (v4sf)
8590 void __builtin_ia32_mwait (unsigned int, unsigned int)
8593 The following built-in functions are available when @option{-msse3} is used.
8596 @item v2df __builtin_ia32_loadddup (double const *)
8597 Generates the @code{movddup} machine instruction as a load from memory.
8600 The following built-in functions are available when @option{-mssse3} is used.
8601 All of them generate the machine instruction that is part of the name
8605 v2si __builtin_ia32_phaddd (v2si, v2si)
8606 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
8607 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
8608 v2si __builtin_ia32_phsubd (v2si, v2si)
8609 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
8610 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
8611 v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi)
8612 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
8613 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
8614 v8qi __builtin_ia32_psignb (v8qi, v8qi)
8615 v2si __builtin_ia32_psignd (v2si, v2si)
8616 v4hi __builtin_ia32_psignw (v4hi, v4hi)
8617 v1di __builtin_ia32_palignr (v1di, v1di, int)
8618 v8qi __builtin_ia32_pabsb (v8qi)
8619 v2si __builtin_ia32_pabsd (v2si)
8620 v4hi __builtin_ia32_pabsw (v4hi)
8623 The following built-in functions are available when @option{-mssse3} is used.
8624 All of them generate the machine instruction that is part of the name
8628 v4si __builtin_ia32_phaddd128 (v4si, v4si)
8629 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
8630 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
8631 v4si __builtin_ia32_phsubd128 (v4si, v4si)
8632 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
8633 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
8634 v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
8635 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
8636 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
8637 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
8638 v4si __builtin_ia32_psignd128 (v4si, v4si)
8639 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
8640 v2di __builtin_ia32_palignr128 (v2di, v2di, int)
8641 v16qi __builtin_ia32_pabsb128 (v16qi)
8642 v4si __builtin_ia32_pabsd128 (v4si)
8643 v8hi __builtin_ia32_pabsw128 (v8hi)
8646 The following built-in functions are available when @option{-msse4.1} is
8647 used. All of them generate the machine instruction that is part of the
8651 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
8652 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
8653 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
8654 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
8655 v2df __builtin_ia32_dppd (v2df, v2df, const int)
8656 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
8657 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
8658 v2di __builtin_ia32_movntdqa (v2di *);
8659 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
8660 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
8661 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
8662 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
8663 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
8664 v8hi __builtin_ia32_phminposuw128 (v8hi)
8665 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
8666 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
8667 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
8668 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
8669 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
8670 v4si __builtin_ia32_pminsd128 (v4si, v4si)
8671 v4si __builtin_ia32_pminud128 (v4si, v4si)
8672 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
8673 v4si __builtin_ia32_pmovsxbd128 (v16qi)
8674 v2di __builtin_ia32_pmovsxbq128 (v16qi)
8675 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
8676 v2di __builtin_ia32_pmovsxdq128 (v4si)
8677 v4si __builtin_ia32_pmovsxwd128 (v8hi)
8678 v2di __builtin_ia32_pmovsxwq128 (v8hi)
8679 v4si __builtin_ia32_pmovzxbd128 (v16qi)
8680 v2di __builtin_ia32_pmovzxbq128 (v16qi)
8681 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
8682 v2di __builtin_ia32_pmovzxdq128 (v4si)
8683 v4si __builtin_ia32_pmovzxwd128 (v8hi)
8684 v2di __builtin_ia32_pmovzxwq128 (v8hi)
8685 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
8686 v4si __builtin_ia32_pmulld128 (v4si, v4si)
8687 int __builtin_ia32_ptestc128 (v2di, v2di)
8688 int __builtin_ia32_ptestnzc128 (v2di, v2di)
8689 int __builtin_ia32_ptestz128 (v2di, v2di)
8690 v2df __builtin_ia32_roundpd (v2df, const int)
8691 v4sf __builtin_ia32_roundps (v4sf, const int)
8692 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
8693 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
8696 The following built-in functions are available when @option{-msse4.1} is
8700 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
8701 Generates the @code{insertps} machine instruction.
8702 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
8703 Generates the @code{pextrb} machine instruction.
8704 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
8705 Generates the @code{pinsrb} machine instruction.
8706 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
8707 Generates the @code{pinsrd} machine instruction.
8708 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
8709 Generates the @code{pinsrq} machine instruction in 64bit mode.
8712 The following built-in functions are changed to generate new SSE4.1
8713 instructions when @option{-msse4.1} is used.
8716 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
8717 Generates the @code{extractps} machine instruction.
8718 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
8719 Generates the @code{pextrd} machine instruction.
8720 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
8721 Generates the @code{pextrq} machine instruction in 64bit mode.
8724 The following built-in functions are available when @option{-msse4.2} is
8725 used. All of them generate the machine instruction that is part of the
8729 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
8730 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
8731 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
8732 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
8733 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
8734 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
8735 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
8736 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
8737 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
8738 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
8739 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
8740 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
8741 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
8742 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
8743 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
8746 The following built-in functions are available when @option{-msse4.2} is
8750 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
8751 Generates the @code{crc32b} machine instruction.
8752 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
8753 Generates the @code{crc32w} machine instruction.
8754 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
8755 Generates the @code{crc32l} machine instruction.
8756 @item unsigned long long __builtin_ia32_crc32di (unsigned long long, unsigned long long)
8757 Generates the @code{crc32q} machine instruction.
8760 The following built-in functions are changed to generate new SSE4.2
8761 instructions when @option{-msse4.2} is used.
8764 @item int __builtin_popcount (unsigned int)
8765 Generates the @code{popcntl} machine instruction.
8766 @item int __builtin_popcountl (unsigned long)
8767 Generates the @code{popcntl} or @code{popcntq} machine instruction,
8768 depending on the size of @code{unsigned long}.
8769 @item int __builtin_popcountll (unsigned long long)
8770 Generates the @code{popcntq} machine instruction.
8773 The following built-in functions are available when @option{-mavx} is
8774 used. All of them generate the machine instruction that is part of the
8778 v4df __builtin_ia32_addpd256 (v4df,v4df)
8779 v8sf __builtin_ia32_addps256 (v8sf,v8sf)
8780 v4df __builtin_ia32_addsubpd256 (v4df,v4df)
8781 v8sf __builtin_ia32_addsubps256 (v8sf,v8sf)
8782 v4df __builtin_ia32_andnpd256 (v4df,v4df)
8783 v8sf __builtin_ia32_andnps256 (v8sf,v8sf)
8784 v4df __builtin_ia32_andpd256 (v4df,v4df)
8785 v8sf __builtin_ia32_andps256 (v8sf,v8sf)
8786 v4df __builtin_ia32_blendpd256 (v4df,v4df,int)
8787 v8sf __builtin_ia32_blendps256 (v8sf,v8sf,int)
8788 v4df __builtin_ia32_blendvpd256 (v4df,v4df,v4df)
8789 v8sf __builtin_ia32_blendvps256 (v8sf,v8sf,v8sf)
8790 v2df __builtin_ia32_cmppd (v2df,v2df,int)
8791 v4df __builtin_ia32_cmppd256 (v4df,v4df,int)
8792 v4sf __builtin_ia32_cmpps (v4sf,v4sf,int)
8793 v8sf __builtin_ia32_cmpps256 (v8sf,v8sf,int)
8794 v2df __builtin_ia32_cmpsd (v2df,v2df,int)
8795 v4sf __builtin_ia32_cmpss (v4sf,v4sf,int)
8796 v4df __builtin_ia32_cvtdq2pd256 (v4si)
8797 v8sf __builtin_ia32_cvtdq2ps256 (v8si)
8798 v4si __builtin_ia32_cvtpd2dq256 (v4df)
8799 v4sf __builtin_ia32_cvtpd2ps256 (v4df)
8800 v8si __builtin_ia32_cvtps2dq256 (v8sf)
8801 v4df __builtin_ia32_cvtps2pd256 (v4sf)
8802 v4si __builtin_ia32_cvttpd2dq256 (v4df)
8803 v8si __builtin_ia32_cvttps2dq256 (v8sf)
8804 v4df __builtin_ia32_divpd256 (v4df,v4df)
8805 v8sf __builtin_ia32_divps256 (v8sf,v8sf)
8806 v8sf __builtin_ia32_dpps256 (v8sf,v8sf,int)
8807 v4df __builtin_ia32_haddpd256 (v4df,v4df)
8808 v8sf __builtin_ia32_haddps256 (v8sf,v8sf)
8809 v4df __builtin_ia32_hsubpd256 (v4df,v4df)
8810 v8sf __builtin_ia32_hsubps256 (v8sf,v8sf)
8811 v32qi __builtin_ia32_lddqu256 (pcchar)
8812 v32qi __builtin_ia32_loaddqu256 (pcchar)
8813 v4df __builtin_ia32_loadupd256 (pcdouble)
8814 v8sf __builtin_ia32_loadups256 (pcfloat)
8815 v2df __builtin_ia32_maskloadpd (pcv2df,v2df)
8816 v4df __builtin_ia32_maskloadpd256 (pcv4df,v4df)
8817 v4sf __builtin_ia32_maskloadps (pcv4sf,v4sf)
8818 v8sf __builtin_ia32_maskloadps256 (pcv8sf,v8sf)
8819 void __builtin_ia32_maskstorepd (pv2df,v2df,v2df)
8820 void __builtin_ia32_maskstorepd256 (pv4df,v4df,v4df)
8821 void __builtin_ia32_maskstoreps (pv4sf,v4sf,v4sf)
8822 void __builtin_ia32_maskstoreps256 (pv8sf,v8sf,v8sf)
8823 v4df __builtin_ia32_maxpd256 (v4df,v4df)
8824 v8sf __builtin_ia32_maxps256 (v8sf,v8sf)
8825 v4df __builtin_ia32_minpd256 (v4df,v4df)
8826 v8sf __builtin_ia32_minps256 (v8sf,v8sf)
8827 v4df __builtin_ia32_movddup256 (v4df)
8828 int __builtin_ia32_movmskpd256 (v4df)
8829 int __builtin_ia32_movmskps256 (v8sf)
8830 v8sf __builtin_ia32_movshdup256 (v8sf)
8831 v8sf __builtin_ia32_movsldup256 (v8sf)
8832 v4df __builtin_ia32_mulpd256 (v4df,v4df)
8833 v8sf __builtin_ia32_mulps256 (v8sf,v8sf)
8834 v4df __builtin_ia32_orpd256 (v4df,v4df)
8835 v8sf __builtin_ia32_orps256 (v8sf,v8sf)
8836 v2df __builtin_ia32_pd_pd256 (v4df)
8837 v4df __builtin_ia32_pd256_pd (v2df)
8838 v4sf __builtin_ia32_ps_ps256 (v8sf)
8839 v8sf __builtin_ia32_ps256_ps (v4sf)
8840 int __builtin_ia32_ptestc256 (v4di,v4di,ptest)
8841 int __builtin_ia32_ptestnzc256 (v4di,v4di,ptest)
8842 int __builtin_ia32_ptestz256 (v4di,v4di,ptest)
8843 v8sf __builtin_ia32_rcpps256 (v8sf)
8844 v4df __builtin_ia32_roundpd256 (v4df,int)
8845 v8sf __builtin_ia32_roundps256 (v8sf,int)
8846 v8sf __builtin_ia32_rsqrtps_nr256 (v8sf)
8847 v8sf __builtin_ia32_rsqrtps256 (v8sf)
8848 v4df __builtin_ia32_shufpd256 (v4df,v4df,int)
8849 v8sf __builtin_ia32_shufps256 (v8sf,v8sf,int)
8850 v4si __builtin_ia32_si_si256 (v8si)
8851 v8si __builtin_ia32_si256_si (v4si)
8852 v4df __builtin_ia32_sqrtpd256 (v4df)
8853 v8sf __builtin_ia32_sqrtps_nr256 (v8sf)
8854 v8sf __builtin_ia32_sqrtps256 (v8sf)
8855 void __builtin_ia32_storedqu256 (pchar,v32qi)
8856 void __builtin_ia32_storeupd256 (pdouble,v4df)
8857 void __builtin_ia32_storeups256 (pfloat,v8sf)
8858 v4df __builtin_ia32_subpd256 (v4df,v4df)
8859 v8sf __builtin_ia32_subps256 (v8sf,v8sf)
8860 v4df __builtin_ia32_unpckhpd256 (v4df,v4df)
8861 v8sf __builtin_ia32_unpckhps256 (v8sf,v8sf)
8862 v4df __builtin_ia32_unpcklpd256 (v4df,v4df)
8863 v8sf __builtin_ia32_unpcklps256 (v8sf,v8sf)
8864 v4df __builtin_ia32_vbroadcastf128_pd256 (pcv2df)
8865 v8sf __builtin_ia32_vbroadcastf128_ps256 (pcv4sf)
8866 v4df __builtin_ia32_vbroadcastsd256 (pcdouble)
8867 v4sf __builtin_ia32_vbroadcastss (pcfloat)
8868 v8sf __builtin_ia32_vbroadcastss256 (pcfloat)
8869 v2df __builtin_ia32_vextractf128_pd256 (v4df,int)
8870 v4sf __builtin_ia32_vextractf128_ps256 (v8sf,int)
8871 v4si __builtin_ia32_vextractf128_si256 (v8si,int)
8872 v4df __builtin_ia32_vinsertf128_pd256 (v4df,v2df,int)
8873 v8sf __builtin_ia32_vinsertf128_ps256 (v8sf,v4sf,int)
8874 v8si __builtin_ia32_vinsertf128_si256 (v8si,v4si,int)
8875 v4df __builtin_ia32_vperm2f128_pd256 (v4df,v4df,int)
8876 v8sf __builtin_ia32_vperm2f128_ps256 (v8sf,v8sf,int)
8877 v8si __builtin_ia32_vperm2f128_si256 (v8si,v8si,int)
8878 v2df __builtin_ia32_vpermil2pd (v2df,v2df,v2di,int)
8879 v4df __builtin_ia32_vpermil2pd256 (v4df,v4df,v4di,int)
8880 v4sf __builtin_ia32_vpermil2ps (v4sf,v4sf,v4si,int)
8881 v8sf __builtin_ia32_vpermil2ps256 (v8sf,v8sf,v8si,int)
8882 v2df __builtin_ia32_vpermilpd (v2df,int)
8883 v4df __builtin_ia32_vpermilpd256 (v4df,int)
8884 v4sf __builtin_ia32_vpermilps (v4sf,int)
8885 v8sf __builtin_ia32_vpermilps256 (v8sf,int)
8886 v2df __builtin_ia32_vpermilvarpd (v2df,v2di)
8887 v4df __builtin_ia32_vpermilvarpd256 (v4df,v4di)
8888 v4sf __builtin_ia32_vpermilvarps (v4sf,v4si)
8889 v8sf __builtin_ia32_vpermilvarps256 (v8sf,v8si)
8890 int __builtin_ia32_vtestcpd (v2df,v2df,ptest)
8891 int __builtin_ia32_vtestcpd256 (v4df,v4df,ptest)
8892 int __builtin_ia32_vtestcps (v4sf,v4sf,ptest)
8893 int __builtin_ia32_vtestcps256 (v8sf,v8sf,ptest)
8894 int __builtin_ia32_vtestnzcpd (v2df,v2df,ptest)
8895 int __builtin_ia32_vtestnzcpd256 (v4df,v4df,ptest)
8896 int __builtin_ia32_vtestnzcps (v4sf,v4sf,ptest)
8897 int __builtin_ia32_vtestnzcps256 (v8sf,v8sf,ptest)
8898 int __builtin_ia32_vtestzpd (v2df,v2df,ptest)
8899 int __builtin_ia32_vtestzpd256 (v4df,v4df,ptest)
8900 int __builtin_ia32_vtestzps (v4sf,v4sf,ptest)
8901 int __builtin_ia32_vtestzps256 (v8sf,v8sf,ptest)
8902 void __builtin_ia32_vzeroall (void)
8903 void __builtin_ia32_vzeroupper (void)
8904 v4df __builtin_ia32_xorpd256 (v4df,v4df)
8905 v8sf __builtin_ia32_xorps256 (v8sf,v8sf)
8908 The following built-in functions are available when @option{-maes} is
8909 used. All of them generate the machine instruction that is part of the
8913 v2di __builtin_ia32_aesenc128 (v2di, v2di)
8914 v2di __builtin_ia32_aesenclast128 (v2di, v2di)
8915 v2di __builtin_ia32_aesdec128 (v2di, v2di)
8916 v2di __builtin_ia32_aesdeclast128 (v2di, v2di)
8917 v2di __builtin_ia32_aeskeygenassist128 (v2di, const int)
8918 v2di __builtin_ia32_aesimc128 (v2di)
8921 The following built-in function is available when @option{-mpclmul} is
8925 @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int)
8926 Generates the @code{pclmulqdq} machine instruction.
8929 The following built-in functions are available when @option{-msse4a} is used.
8930 All of them generate the machine instruction that is part of the name.
8933 void __builtin_ia32_movntsd (double *, v2df)
8934 void __builtin_ia32_movntss (float *, v4sf)
8935 v2di __builtin_ia32_extrq (v2di, v16qi)
8936 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
8937 v2di __builtin_ia32_insertq (v2di, v2di)
8938 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
8941 The following built-in functions are available when @option{-mxop} is used.
8943 v2df __builtin_ia32_vfrczpd (v2df)
8944 v4sf __builtin_ia32_vfrczps (v4sf)
8945 v2df __builtin_ia32_vfrczsd (v2df, v2df)
8946 v4sf __builtin_ia32_vfrczss (v4sf, v4sf)
8947 v4df __builtin_ia32_vfrczpd256 (v4df)
8948 v8sf __builtin_ia32_vfrczps256 (v8sf)
8949 v2di __builtin_ia32_vpcmov (v2di, v2di, v2di)
8950 v2di __builtin_ia32_vpcmov_v2di (v2di, v2di, v2di)
8951 v4si __builtin_ia32_vpcmov_v4si (v4si, v4si, v4si)
8952 v8hi __builtin_ia32_vpcmov_v8hi (v8hi, v8hi, v8hi)
8953 v16qi __builtin_ia32_vpcmov_v16qi (v16qi, v16qi, v16qi)
8954 v2df __builtin_ia32_vpcmov_v2df (v2df, v2df, v2df)
8955 v4sf __builtin_ia32_vpcmov_v4sf (v4sf, v4sf, v4sf)
8956 v4di __builtin_ia32_vpcmov_v4di256 (v4di, v4di, v4di)
8957 v8si __builtin_ia32_vpcmov_v8si256 (v8si, v8si, v8si)
8958 v16hi __builtin_ia32_vpcmov_v16hi256 (v16hi, v16hi, v16hi)
8959 v32qi __builtin_ia32_vpcmov_v32qi256 (v32qi, v32qi, v32qi)
8960 v4df __builtin_ia32_vpcmov_v4df256 (v4df, v4df, v4df)
8961 v8sf __builtin_ia32_vpcmov_v8sf256 (v8sf, v8sf, v8sf)
8962 v16qi __builtin_ia32_vpcomeqb (v16qi, v16qi)
8963 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
8964 v4si __builtin_ia32_vpcomeqd (v4si, v4si)
8965 v2di __builtin_ia32_vpcomeqq (v2di, v2di)
8966 v16qi __builtin_ia32_vpcomequb (v16qi, v16qi)
8967 v4si __builtin_ia32_vpcomequd (v4si, v4si)
8968 v2di __builtin_ia32_vpcomequq (v2di, v2di)
8969 v8hi __builtin_ia32_vpcomequw (v8hi, v8hi)
8970 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
8971 v16qi __builtin_ia32_vpcomfalseb (v16qi, v16qi)
8972 v4si __builtin_ia32_vpcomfalsed (v4si, v4si)
8973 v2di __builtin_ia32_vpcomfalseq (v2di, v2di)
8974 v16qi __builtin_ia32_vpcomfalseub (v16qi, v16qi)
8975 v4si __builtin_ia32_vpcomfalseud (v4si, v4si)
8976 v2di __builtin_ia32_vpcomfalseuq (v2di, v2di)
8977 v8hi __builtin_ia32_vpcomfalseuw (v8hi, v8hi)
8978 v8hi __builtin_ia32_vpcomfalsew (v8hi, v8hi)
8979 v16qi __builtin_ia32_vpcomgeb (v16qi, v16qi)
8980 v4si __builtin_ia32_vpcomged (v4si, v4si)
8981 v2di __builtin_ia32_vpcomgeq (v2di, v2di)
8982 v16qi __builtin_ia32_vpcomgeub (v16qi, v16qi)
8983 v4si __builtin_ia32_vpcomgeud (v4si, v4si)
8984 v2di __builtin_ia32_vpcomgeuq (v2di, v2di)
8985 v8hi __builtin_ia32_vpcomgeuw (v8hi, v8hi)
8986 v8hi __builtin_ia32_vpcomgew (v8hi, v8hi)
8987 v16qi __builtin_ia32_vpcomgtb (v16qi, v16qi)
8988 v4si __builtin_ia32_vpcomgtd (v4si, v4si)
8989 v2di __builtin_ia32_vpcomgtq (v2di, v2di)
8990 v16qi __builtin_ia32_vpcomgtub (v16qi, v16qi)
8991 v4si __builtin_ia32_vpcomgtud (v4si, v4si)
8992 v2di __builtin_ia32_vpcomgtuq (v2di, v2di)
8993 v8hi __builtin_ia32_vpcomgtuw (v8hi, v8hi)
8994 v8hi __builtin_ia32_vpcomgtw (v8hi, v8hi)
8995 v16qi __builtin_ia32_vpcomleb (v16qi, v16qi)
8996 v4si __builtin_ia32_vpcomled (v4si, v4si)
8997 v2di __builtin_ia32_vpcomleq (v2di, v2di)
8998 v16qi __builtin_ia32_vpcomleub (v16qi, v16qi)
8999 v4si __builtin_ia32_vpcomleud (v4si, v4si)
9000 v2di __builtin_ia32_vpcomleuq (v2di, v2di)
9001 v8hi __builtin_ia32_vpcomleuw (v8hi, v8hi)
9002 v8hi __builtin_ia32_vpcomlew (v8hi, v8hi)
9003 v16qi __builtin_ia32_vpcomltb (v16qi, v16qi)
9004 v4si __builtin_ia32_vpcomltd (v4si, v4si)
9005 v2di __builtin_ia32_vpcomltq (v2di, v2di)
9006 v16qi __builtin_ia32_vpcomltub (v16qi, v16qi)
9007 v4si __builtin_ia32_vpcomltud (v4si, v4si)
9008 v2di __builtin_ia32_vpcomltuq (v2di, v2di)
9009 v8hi __builtin_ia32_vpcomltuw (v8hi, v8hi)
9010 v8hi __builtin_ia32_vpcomltw (v8hi, v8hi)
9011 v16qi __builtin_ia32_vpcomneb (v16qi, v16qi)
9012 v4si __builtin_ia32_vpcomned (v4si, v4si)
9013 v2di __builtin_ia32_vpcomneq (v2di, v2di)
9014 v16qi __builtin_ia32_vpcomneub (v16qi, v16qi)
9015 v4si __builtin_ia32_vpcomneud (v4si, v4si)
9016 v2di __builtin_ia32_vpcomneuq (v2di, v2di)
9017 v8hi __builtin_ia32_vpcomneuw (v8hi, v8hi)
9018 v8hi __builtin_ia32_vpcomnew (v8hi, v8hi)
9019 v16qi __builtin_ia32_vpcomtrueb (v16qi, v16qi)
9020 v4si __builtin_ia32_vpcomtrued (v4si, v4si)
9021 v2di __builtin_ia32_vpcomtrueq (v2di, v2di)
9022 v16qi __builtin_ia32_vpcomtrueub (v16qi, v16qi)
9023 v4si __builtin_ia32_vpcomtrueud (v4si, v4si)
9024 v2di __builtin_ia32_vpcomtrueuq (v2di, v2di)
9025 v8hi __builtin_ia32_vpcomtrueuw (v8hi, v8hi)
9026 v8hi __builtin_ia32_vpcomtruew (v8hi, v8hi)
9027 v4si __builtin_ia32_vphaddbd (v16qi)
9028 v2di __builtin_ia32_vphaddbq (v16qi)
9029 v8hi __builtin_ia32_vphaddbw (v16qi)
9030 v2di __builtin_ia32_vphadddq (v4si)
9031 v4si __builtin_ia32_vphaddubd (v16qi)
9032 v2di __builtin_ia32_vphaddubq (v16qi)
9033 v8hi __builtin_ia32_vphaddubw (v16qi)
9034 v2di __builtin_ia32_vphaddudq (v4si)
9035 v4si __builtin_ia32_vphadduwd (v8hi)
9036 v2di __builtin_ia32_vphadduwq (v8hi)
9037 v4si __builtin_ia32_vphaddwd (v8hi)
9038 v2di __builtin_ia32_vphaddwq (v8hi)
9039 v8hi __builtin_ia32_vphsubbw (v16qi)
9040 v2di __builtin_ia32_vphsubdq (v4si)
9041 v4si __builtin_ia32_vphsubwd (v8hi)
9042 v4si __builtin_ia32_vpmacsdd (v4si, v4si, v4si)
9043 v2di __builtin_ia32_vpmacsdqh (v4si, v4si, v2di)
9044 v2di __builtin_ia32_vpmacsdql (v4si, v4si, v2di)
9045 v4si __builtin_ia32_vpmacssdd (v4si, v4si, v4si)
9046 v2di __builtin_ia32_vpmacssdqh (v4si, v4si, v2di)
9047 v2di __builtin_ia32_vpmacssdql (v4si, v4si, v2di)
9048 v4si __builtin_ia32_vpmacsswd (v8hi, v8hi, v4si)
9049 v8hi __builtin_ia32_vpmacssww (v8hi, v8hi, v8hi)
9050 v4si __builtin_ia32_vpmacswd (v8hi, v8hi, v4si)
9051 v8hi __builtin_ia32_vpmacsww (v8hi, v8hi, v8hi)
9052 v4si __builtin_ia32_vpmadcsswd (v8hi, v8hi, v4si)
9053 v4si __builtin_ia32_vpmadcswd (v8hi, v8hi, v4si)
9054 v16qi __builtin_ia32_vpperm (v16qi, v16qi, v16qi)
9055 v16qi __builtin_ia32_vprotb (v16qi, v16qi)
9056 v4si __builtin_ia32_vprotd (v4si, v4si)
9057 v2di __builtin_ia32_vprotq (v2di, v2di)
9058 v8hi __builtin_ia32_vprotw (v8hi, v8hi)
9059 v16qi __builtin_ia32_vpshab (v16qi, v16qi)
9060 v4si __builtin_ia32_vpshad (v4si, v4si)
9061 v2di __builtin_ia32_vpshaq (v2di, v2di)
9062 v8hi __builtin_ia32_vpshaw (v8hi, v8hi)
9063 v16qi __builtin_ia32_vpshlb (v16qi, v16qi)
9064 v4si __builtin_ia32_vpshld (v4si, v4si)
9065 v2di __builtin_ia32_vpshlq (v2di, v2di)
9066 v8hi __builtin_ia32_vpshlw (v8hi, v8hi)
9069 The following built-in functions are available when @option{-mfma4} is used.
9070 All of them generate the machine instruction that is part of the name
9074 v2df __builtin_ia32_fmaddpd (v2df, v2df, v2df)
9075 v4sf __builtin_ia32_fmaddps (v4sf, v4sf, v4sf)
9076 v2df __builtin_ia32_fmaddsd (v2df, v2df, v2df)
9077 v4sf __builtin_ia32_fmaddss (v4sf, v4sf, v4sf)
9078 v2df __builtin_ia32_fmsubpd (v2df, v2df, v2df)
9079 v4sf __builtin_ia32_fmsubps (v4sf, v4sf, v4sf)
9080 v2df __builtin_ia32_fmsubsd (v2df, v2df, v2df)
9081 v4sf __builtin_ia32_fmsubss (v4sf, v4sf, v4sf)
9082 v2df __builtin_ia32_fnmaddpd (v2df, v2df, v2df)
9083 v4sf __builtin_ia32_fnmaddps (v4sf, v4sf, v4sf)
9084 v2df __builtin_ia32_fnmaddsd (v2df, v2df, v2df)
9085 v4sf __builtin_ia32_fnmaddss (v4sf, v4sf, v4sf)
9086 v2df __builtin_ia32_fnmsubpd (v2df, v2df, v2df)
9087 v4sf __builtin_ia32_fnmsubps (v4sf, v4sf, v4sf)
9088 v2df __builtin_ia32_fnmsubsd (v2df, v2df, v2df)
9089 v4sf __builtin_ia32_fnmsubss (v4sf, v4sf, v4sf)
9090 v2df __builtin_ia32_fmaddsubpd (v2df, v2df, v2df)
9091 v4sf __builtin_ia32_fmaddsubps (v4sf, v4sf, v4sf)
9092 v2df __builtin_ia32_fmsubaddpd (v2df, v2df, v2df)
9093 v4sf __builtin_ia32_fmsubaddps (v4sf, v4sf, v4sf)
9094 v4df __builtin_ia32_fmaddpd256 (v4df, v4df, v4df)
9095 v8sf __builtin_ia32_fmaddps256 (v8sf, v8sf, v8sf)
9096 v4df __builtin_ia32_fmsubpd256 (v4df, v4df, v4df)
9097 v8sf __builtin_ia32_fmsubps256 (v8sf, v8sf, v8sf)
9098 v4df __builtin_ia32_fnmaddpd256 (v4df, v4df, v4df)
9099 v8sf __builtin_ia32_fnmaddps256 (v8sf, v8sf, v8sf)
9100 v4df __builtin_ia32_fnmsubpd256 (v4df, v4df, v4df)
9101 v8sf __builtin_ia32_fnmsubps256 (v8sf, v8sf, v8sf)
9102 v4df __builtin_ia32_fmaddsubpd256 (v4df, v4df, v4df)
9103 v8sf __builtin_ia32_fmaddsubps256 (v8sf, v8sf, v8sf)
9104 v4df __builtin_ia32_fmsubaddpd256 (v4df, v4df, v4df)
9105 v8sf __builtin_ia32_fmsubaddps256 (v8sf, v8sf, v8sf)
9109 The following built-in functions are available when @option{-mlwp} is used.
9112 void __builtin_ia32_llwpcb16 (void *);
9113 void __builtin_ia32_llwpcb32 (void *);
9114 void __builtin_ia32_llwpcb64 (void *);
9115 void * __builtin_ia32_llwpcb16 (void);
9116 void * __builtin_ia32_llwpcb32 (void);
9117 void * __builtin_ia32_llwpcb64 (void);
9118 void __builtin_ia32_lwpval16 (unsigned short, unsigned int, unsigned short)
9119 void __builtin_ia32_lwpval32 (unsigned int, unsigned int, unsigned int)
9120 void __builtin_ia32_lwpval64 (unsigned __int64, unsigned int, unsigned int)
9121 unsigned char __builtin_ia32_lwpins16 (unsigned short, unsigned int, unsigned short)
9122 unsigned char __builtin_ia32_lwpins32 (unsigned int, unsigned int, unsigned int)
9123 unsigned char __builtin_ia32_lwpins64 (unsigned __int64, unsigned int, unsigned int)
9126 The following built-in functions are available when @option{-m3dnow} is used.
9127 All of them generate the machine instruction that is part of the name.
9130 void __builtin_ia32_femms (void)
9131 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
9132 v2si __builtin_ia32_pf2id (v2sf)
9133 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
9134 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
9135 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
9136 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
9137 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
9138 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
9139 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
9140 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
9141 v2sf __builtin_ia32_pfrcp (v2sf)
9142 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
9143 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
9144 v2sf __builtin_ia32_pfrsqrt (v2sf)
9145 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
9146 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
9147 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
9148 v2sf __builtin_ia32_pi2fd (v2si)
9149 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
9152 The following built-in functions are available when both @option{-m3dnow}
9153 and @option{-march=athlon} are used. All of them generate the machine
9154 instruction that is part of the name.
9157 v2si __builtin_ia32_pf2iw (v2sf)
9158 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
9159 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
9160 v2sf __builtin_ia32_pi2fw (v2si)
9161 v2sf __builtin_ia32_pswapdsf (v2sf)
9162 v2si __builtin_ia32_pswapdsi (v2si)
9165 @node MIPS DSP Built-in Functions
9166 @subsection MIPS DSP Built-in Functions
9168 The MIPS DSP Application-Specific Extension (ASE) includes new
9169 instructions that are designed to improve the performance of DSP and
9170 media applications. It provides instructions that operate on packed
9171 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
9173 GCC supports MIPS DSP operations using both the generic
9174 vector extensions (@pxref{Vector Extensions}) and a collection of
9175 MIPS-specific built-in functions. Both kinds of support are
9176 enabled by the @option{-mdsp} command-line option.
9178 Revision 2 of the ASE was introduced in the second half of 2006.
9179 This revision adds extra instructions to the original ASE, but is
9180 otherwise backwards-compatible with it. You can select revision 2
9181 using the command-line option @option{-mdspr2}; this option implies
9184 The SCOUNT and POS bits of the DSP control register are global. The
9185 WRDSP, EXTPDP, EXTPDPV and MTHLIP instructions modify the SCOUNT and
9186 POS bits. During optimization, the compiler will not delete these
9187 instructions and it will not delete calls to functions containing
9190 At present, GCC only provides support for operations on 32-bit
9191 vectors. The vector type associated with 8-bit integer data is
9192 usually called @code{v4i8}, the vector type associated with Q7
9193 is usually called @code{v4q7}, the vector type associated with 16-bit
9194 integer data is usually called @code{v2i16}, and the vector type
9195 associated with Q15 is usually called @code{v2q15}. They can be
9196 defined in C as follows:
9199 typedef signed char v4i8 __attribute__ ((vector_size(4)));
9200 typedef signed char v4q7 __attribute__ ((vector_size(4)));
9201 typedef short v2i16 __attribute__ ((vector_size(4)));
9202 typedef short v2q15 __attribute__ ((vector_size(4)));
9205 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
9206 initialized in the same way as aggregates. For example:
9209 v4i8 a = @{1, 2, 3, 4@};
9211 b = (v4i8) @{5, 6, 7, 8@};
9213 v2q15 c = @{0x0fcb, 0x3a75@};
9215 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
9218 @emph{Note:} The CPU's endianness determines the order in which values
9219 are packed. On little-endian targets, the first value is the least
9220 significant and the last value is the most significant. The opposite
9221 order applies to big-endian targets. For example, the code above will
9222 set the lowest byte of @code{a} to @code{1} on little-endian targets
9223 and @code{4} on big-endian targets.
9225 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
9226 representation. As shown in this example, the integer representation
9227 of a Q7 value can be obtained by multiplying the fractional value by
9228 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
9229 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
9232 The table below lists the @code{v4i8} and @code{v2q15} operations for which
9233 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
9234 and @code{c} and @code{d} are @code{v2q15} values.
9236 @multitable @columnfractions .50 .50
9237 @item C code @tab MIPS instruction
9238 @item @code{a + b} @tab @code{addu.qb}
9239 @item @code{c + d} @tab @code{addq.ph}
9240 @item @code{a - b} @tab @code{subu.qb}
9241 @item @code{c - d} @tab @code{subq.ph}
9244 The table below lists the @code{v2i16} operation for which
9245 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
9246 @code{v2i16} values.
9248 @multitable @columnfractions .50 .50
9249 @item C code @tab MIPS instruction
9250 @item @code{e * f} @tab @code{mul.ph}
9253 It is easier to describe the DSP built-in functions if we first define
9254 the following types:
9259 typedef unsigned int ui32;
9260 typedef long long a64;
9263 @code{q31} and @code{i32} are actually the same as @code{int}, but we
9264 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
9265 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
9266 @code{long long}, but we use @code{a64} to indicate values that will
9267 be placed in one of the four DSP accumulators (@code{$ac0},
9268 @code{$ac1}, @code{$ac2} or @code{$ac3}).
9270 Also, some built-in functions prefer or require immediate numbers as
9271 parameters, because the corresponding DSP instructions accept both immediate
9272 numbers and register operands, or accept immediate numbers only. The
9273 immediate parameters are listed as follows.
9282 imm_n32_31: -32 to 31.
9283 imm_n512_511: -512 to 511.
9286 The following built-in functions map directly to a particular MIPS DSP
9287 instruction. Please refer to the architecture specification
9288 for details on what each instruction does.
9291 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
9292 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
9293 q31 __builtin_mips_addq_s_w (q31, q31)
9294 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
9295 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
9296 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
9297 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
9298 q31 __builtin_mips_subq_s_w (q31, q31)
9299 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
9300 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
9301 i32 __builtin_mips_addsc (i32, i32)
9302 i32 __builtin_mips_addwc (i32, i32)
9303 i32 __builtin_mips_modsub (i32, i32)
9304 i32 __builtin_mips_raddu_w_qb (v4i8)
9305 v2q15 __builtin_mips_absq_s_ph (v2q15)
9306 q31 __builtin_mips_absq_s_w (q31)
9307 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
9308 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
9309 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
9310 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
9311 q31 __builtin_mips_preceq_w_phl (v2q15)
9312 q31 __builtin_mips_preceq_w_phr (v2q15)
9313 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
9314 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
9315 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
9316 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
9317 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
9318 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
9319 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
9320 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
9321 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
9322 v4i8 __builtin_mips_shll_qb (v4i8, i32)
9323 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
9324 v2q15 __builtin_mips_shll_ph (v2q15, i32)
9325 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
9326 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
9327 q31 __builtin_mips_shll_s_w (q31, imm0_31)
9328 q31 __builtin_mips_shll_s_w (q31, i32)
9329 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
9330 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
9331 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
9332 v2q15 __builtin_mips_shra_ph (v2q15, i32)
9333 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
9334 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
9335 q31 __builtin_mips_shra_r_w (q31, imm0_31)
9336 q31 __builtin_mips_shra_r_w (q31, i32)
9337 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
9338 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
9339 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
9340 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
9341 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
9342 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
9343 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
9344 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
9345 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
9346 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
9347 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
9348 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
9349 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
9350 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
9351 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
9352 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
9353 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
9354 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
9355 i32 __builtin_mips_bitrev (i32)
9356 i32 __builtin_mips_insv (i32, i32)
9357 v4i8 __builtin_mips_repl_qb (imm0_255)
9358 v4i8 __builtin_mips_repl_qb (i32)
9359 v2q15 __builtin_mips_repl_ph (imm_n512_511)
9360 v2q15 __builtin_mips_repl_ph (i32)
9361 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
9362 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
9363 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
9364 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
9365 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
9366 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
9367 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
9368 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
9369 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
9370 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
9371 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
9372 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
9373 i32 __builtin_mips_extr_w (a64, imm0_31)
9374 i32 __builtin_mips_extr_w (a64, i32)
9375 i32 __builtin_mips_extr_r_w (a64, imm0_31)
9376 i32 __builtin_mips_extr_s_h (a64, i32)
9377 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
9378 i32 __builtin_mips_extr_rs_w (a64, i32)
9379 i32 __builtin_mips_extr_s_h (a64, imm0_31)
9380 i32 __builtin_mips_extr_r_w (a64, i32)
9381 i32 __builtin_mips_extp (a64, imm0_31)
9382 i32 __builtin_mips_extp (a64, i32)
9383 i32 __builtin_mips_extpdp (a64, imm0_31)
9384 i32 __builtin_mips_extpdp (a64, i32)
9385 a64 __builtin_mips_shilo (a64, imm_n32_31)
9386 a64 __builtin_mips_shilo (a64, i32)
9387 a64 __builtin_mips_mthlip (a64, i32)
9388 void __builtin_mips_wrdsp (i32, imm0_63)
9389 i32 __builtin_mips_rddsp (imm0_63)
9390 i32 __builtin_mips_lbux (void *, i32)
9391 i32 __builtin_mips_lhx (void *, i32)
9392 i32 __builtin_mips_lwx (void *, i32)
9393 i32 __builtin_mips_bposge32 (void)
9396 The following built-in functions map directly to a particular MIPS DSP REV 2
9397 instruction. Please refer to the architecture specification
9398 for details on what each instruction does.
9401 v4q7 __builtin_mips_absq_s_qb (v4q7);
9402 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
9403 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
9404 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
9405 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
9406 i32 __builtin_mips_append (i32, i32, imm0_31);
9407 i32 __builtin_mips_balign (i32, i32, imm0_3);
9408 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
9409 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
9410 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
9411 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
9412 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
9413 a64 __builtin_mips_madd (a64, i32, i32);
9414 a64 __builtin_mips_maddu (a64, ui32, ui32);
9415 a64 __builtin_mips_msub (a64, i32, i32);
9416 a64 __builtin_mips_msubu (a64, ui32, ui32);
9417 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
9418 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
9419 q31 __builtin_mips_mulq_rs_w (q31, q31);
9420 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
9421 q31 __builtin_mips_mulq_s_w (q31, q31);
9422 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
9423 a64 __builtin_mips_mult (i32, i32);
9424 a64 __builtin_mips_multu (ui32, ui32);
9425 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
9426 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
9427 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
9428 i32 __builtin_mips_prepend (i32, i32, imm0_31);
9429 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
9430 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
9431 v4i8 __builtin_mips_shra_qb (v4i8, i32);
9432 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
9433 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
9434 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
9435 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
9436 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
9437 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
9438 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
9439 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
9440 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
9441 q31 __builtin_mips_addqh_w (q31, q31);
9442 q31 __builtin_mips_addqh_r_w (q31, q31);
9443 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
9444 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
9445 q31 __builtin_mips_subqh_w (q31, q31);
9446 q31 __builtin_mips_subqh_r_w (q31, q31);
9447 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
9448 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
9449 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
9450 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
9451 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
9452 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
9456 @node MIPS Paired-Single Support
9457 @subsection MIPS Paired-Single Support
9459 The MIPS64 architecture includes a number of instructions that
9460 operate on pairs of single-precision floating-point values.
9461 Each pair is packed into a 64-bit floating-point register,
9462 with one element being designated the ``upper half'' and
9463 the other being designated the ``lower half''.
9465 GCC supports paired-single operations using both the generic
9466 vector extensions (@pxref{Vector Extensions}) and a collection of
9467 MIPS-specific built-in functions. Both kinds of support are
9468 enabled by the @option{-mpaired-single} command-line option.
9470 The vector type associated with paired-single values is usually
9471 called @code{v2sf}. It can be defined in C as follows:
9474 typedef float v2sf __attribute__ ((vector_size (8)));
9477 @code{v2sf} values are initialized in the same way as aggregates.
9481 v2sf a = @{1.5, 9.1@};
9484 b = (v2sf) @{e, f@};
9487 @emph{Note:} The CPU's endianness determines which value is stored in
9488 the upper half of a register and which value is stored in the lower half.
9489 On little-endian targets, the first value is the lower one and the second
9490 value is the upper one. The opposite order applies to big-endian targets.
9491 For example, the code above will set the lower half of @code{a} to
9492 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
9494 @node MIPS Loongson Built-in Functions
9495 @subsection MIPS Loongson Built-in Functions
9497 GCC provides intrinsics to access the SIMD instructions provided by the
9498 ST Microelectronics Loongson-2E and -2F processors. These intrinsics,
9499 available after inclusion of the @code{loongson.h} header file,
9500 operate on the following 64-bit vector types:
9503 @item @code{uint8x8_t}, a vector of eight unsigned 8-bit integers;
9504 @item @code{uint16x4_t}, a vector of four unsigned 16-bit integers;
9505 @item @code{uint32x2_t}, a vector of two unsigned 32-bit integers;
9506 @item @code{int8x8_t}, a vector of eight signed 8-bit integers;
9507 @item @code{int16x4_t}, a vector of four signed 16-bit integers;
9508 @item @code{int32x2_t}, a vector of two signed 32-bit integers.
9511 The intrinsics provided are listed below; each is named after the
9512 machine instruction to which it corresponds, with suffixes added as
9513 appropriate to distinguish intrinsics that expand to the same machine
9514 instruction yet have different argument types. Refer to the architecture
9515 documentation for a description of the functionality of each
9519 int16x4_t packsswh (int32x2_t s, int32x2_t t);
9520 int8x8_t packsshb (int16x4_t s, int16x4_t t);
9521 uint8x8_t packushb (uint16x4_t s, uint16x4_t t);
9522 uint32x2_t paddw_u (uint32x2_t s, uint32x2_t t);
9523 uint16x4_t paddh_u (uint16x4_t s, uint16x4_t t);
9524 uint8x8_t paddb_u (uint8x8_t s, uint8x8_t t);
9525 int32x2_t paddw_s (int32x2_t s, int32x2_t t);
9526 int16x4_t paddh_s (int16x4_t s, int16x4_t t);
9527 int8x8_t paddb_s (int8x8_t s, int8x8_t t);
9528 uint64_t paddd_u (uint64_t s, uint64_t t);
9529 int64_t paddd_s (int64_t s, int64_t t);
9530 int16x4_t paddsh (int16x4_t s, int16x4_t t);
9531 int8x8_t paddsb (int8x8_t s, int8x8_t t);
9532 uint16x4_t paddush (uint16x4_t s, uint16x4_t t);
9533 uint8x8_t paddusb (uint8x8_t s, uint8x8_t t);
9534 uint64_t pandn_ud (uint64_t s, uint64_t t);
9535 uint32x2_t pandn_uw (uint32x2_t s, uint32x2_t t);
9536 uint16x4_t pandn_uh (uint16x4_t s, uint16x4_t t);
9537 uint8x8_t pandn_ub (uint8x8_t s, uint8x8_t t);
9538 int64_t pandn_sd (int64_t s, int64_t t);
9539 int32x2_t pandn_sw (int32x2_t s, int32x2_t t);
9540 int16x4_t pandn_sh (int16x4_t s, int16x4_t t);
9541 int8x8_t pandn_sb (int8x8_t s, int8x8_t t);
9542 uint16x4_t pavgh (uint16x4_t s, uint16x4_t t);
9543 uint8x8_t pavgb (uint8x8_t s, uint8x8_t t);
9544 uint32x2_t pcmpeqw_u (uint32x2_t s, uint32x2_t t);
9545 uint16x4_t pcmpeqh_u (uint16x4_t s, uint16x4_t t);
9546 uint8x8_t pcmpeqb_u (uint8x8_t s, uint8x8_t t);
9547 int32x2_t pcmpeqw_s (int32x2_t s, int32x2_t t);
9548 int16x4_t pcmpeqh_s (int16x4_t s, int16x4_t t);
9549 int8x8_t pcmpeqb_s (int8x8_t s, int8x8_t t);
9550 uint32x2_t pcmpgtw_u (uint32x2_t s, uint32x2_t t);
9551 uint16x4_t pcmpgth_u (uint16x4_t s, uint16x4_t t);
9552 uint8x8_t pcmpgtb_u (uint8x8_t s, uint8x8_t t);
9553 int32x2_t pcmpgtw_s (int32x2_t s, int32x2_t t);
9554 int16x4_t pcmpgth_s (int16x4_t s, int16x4_t t);
9555 int8x8_t pcmpgtb_s (int8x8_t s, int8x8_t t);
9556 uint16x4_t pextrh_u (uint16x4_t s, int field);
9557 int16x4_t pextrh_s (int16x4_t s, int field);
9558 uint16x4_t pinsrh_0_u (uint16x4_t s, uint16x4_t t);
9559 uint16x4_t pinsrh_1_u (uint16x4_t s, uint16x4_t t);
9560 uint16x4_t pinsrh_2_u (uint16x4_t s, uint16x4_t t);
9561 uint16x4_t pinsrh_3_u (uint16x4_t s, uint16x4_t t);
9562 int16x4_t pinsrh_0_s (int16x4_t s, int16x4_t t);
9563 int16x4_t pinsrh_1_s (int16x4_t s, int16x4_t t);
9564 int16x4_t pinsrh_2_s (int16x4_t s, int16x4_t t);
9565 int16x4_t pinsrh_3_s (int16x4_t s, int16x4_t t);
9566 int32x2_t pmaddhw (int16x4_t s, int16x4_t t);
9567 int16x4_t pmaxsh (int16x4_t s, int16x4_t t);
9568 uint8x8_t pmaxub (uint8x8_t s, uint8x8_t t);
9569 int16x4_t pminsh (int16x4_t s, int16x4_t t);
9570 uint8x8_t pminub (uint8x8_t s, uint8x8_t t);
9571 uint8x8_t pmovmskb_u (uint8x8_t s);
9572 int8x8_t pmovmskb_s (int8x8_t s);
9573 uint16x4_t pmulhuh (uint16x4_t s, uint16x4_t t);
9574 int16x4_t pmulhh (int16x4_t s, int16x4_t t);
9575 int16x4_t pmullh (int16x4_t s, int16x4_t t);
9576 int64_t pmuluw (uint32x2_t s, uint32x2_t t);
9577 uint8x8_t pasubub (uint8x8_t s, uint8x8_t t);
9578 uint16x4_t biadd (uint8x8_t s);
9579 uint16x4_t psadbh (uint8x8_t s, uint8x8_t t);
9580 uint16x4_t pshufh_u (uint16x4_t dest, uint16x4_t s, uint8_t order);
9581 int16x4_t pshufh_s (int16x4_t dest, int16x4_t s, uint8_t order);
9582 uint16x4_t psllh_u (uint16x4_t s, uint8_t amount);
9583 int16x4_t psllh_s (int16x4_t s, uint8_t amount);
9584 uint32x2_t psllw_u (uint32x2_t s, uint8_t amount);
9585 int32x2_t psllw_s (int32x2_t s, uint8_t amount);
9586 uint16x4_t psrlh_u (uint16x4_t s, uint8_t amount);
9587 int16x4_t psrlh_s (int16x4_t s, uint8_t amount);
9588 uint32x2_t psrlw_u (uint32x2_t s, uint8_t amount);
9589 int32x2_t psrlw_s (int32x2_t s, uint8_t amount);
9590 uint16x4_t psrah_u (uint16x4_t s, uint8_t amount);
9591 int16x4_t psrah_s (int16x4_t s, uint8_t amount);
9592 uint32x2_t psraw_u (uint32x2_t s, uint8_t amount);
9593 int32x2_t psraw_s (int32x2_t s, uint8_t amount);
9594 uint32x2_t psubw_u (uint32x2_t s, uint32x2_t t);
9595 uint16x4_t psubh_u (uint16x4_t s, uint16x4_t t);
9596 uint8x8_t psubb_u (uint8x8_t s, uint8x8_t t);
9597 int32x2_t psubw_s (int32x2_t s, int32x2_t t);
9598 int16x4_t psubh_s (int16x4_t s, int16x4_t t);
9599 int8x8_t psubb_s (int8x8_t s, int8x8_t t);
9600 uint64_t psubd_u (uint64_t s, uint64_t t);
9601 int64_t psubd_s (int64_t s, int64_t t);
9602 int16x4_t psubsh (int16x4_t s, int16x4_t t);
9603 int8x8_t psubsb (int8x8_t s, int8x8_t t);
9604 uint16x4_t psubush (uint16x4_t s, uint16x4_t t);
9605 uint8x8_t psubusb (uint8x8_t s, uint8x8_t t);
9606 uint32x2_t punpckhwd_u (uint32x2_t s, uint32x2_t t);
9607 uint16x4_t punpckhhw_u (uint16x4_t s, uint16x4_t t);
9608 uint8x8_t punpckhbh_u (uint8x8_t s, uint8x8_t t);
9609 int32x2_t punpckhwd_s (int32x2_t s, int32x2_t t);
9610 int16x4_t punpckhhw_s (int16x4_t s, int16x4_t t);
9611 int8x8_t punpckhbh_s (int8x8_t s, int8x8_t t);
9612 uint32x2_t punpcklwd_u (uint32x2_t s, uint32x2_t t);
9613 uint16x4_t punpcklhw_u (uint16x4_t s, uint16x4_t t);
9614 uint8x8_t punpcklbh_u (uint8x8_t s, uint8x8_t t);
9615 int32x2_t punpcklwd_s (int32x2_t s, int32x2_t t);
9616 int16x4_t punpcklhw_s (int16x4_t s, int16x4_t t);
9617 int8x8_t punpcklbh_s (int8x8_t s, int8x8_t t);
9621 * Paired-Single Arithmetic::
9622 * Paired-Single Built-in Functions::
9623 * MIPS-3D Built-in Functions::
9626 @node Paired-Single Arithmetic
9627 @subsubsection Paired-Single Arithmetic
9629 The table below lists the @code{v2sf} operations for which hardware
9630 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
9631 values and @code{x} is an integral value.
9633 @multitable @columnfractions .50 .50
9634 @item C code @tab MIPS instruction
9635 @item @code{a + b} @tab @code{add.ps}
9636 @item @code{a - b} @tab @code{sub.ps}
9637 @item @code{-a} @tab @code{neg.ps}
9638 @item @code{a * b} @tab @code{mul.ps}
9639 @item @code{a * b + c} @tab @code{madd.ps}
9640 @item @code{a * b - c} @tab @code{msub.ps}
9641 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
9642 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
9643 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
9646 Note that the multiply-accumulate instructions can be disabled
9647 using the command-line option @code{-mno-fused-madd}.
9649 @node Paired-Single Built-in Functions
9650 @subsubsection Paired-Single Built-in Functions
9652 The following paired-single functions map directly to a particular
9653 MIPS instruction. Please refer to the architecture specification
9654 for details on what each instruction does.
9657 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
9658 Pair lower lower (@code{pll.ps}).
9660 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
9661 Pair upper lower (@code{pul.ps}).
9663 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
9664 Pair lower upper (@code{plu.ps}).
9666 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
9667 Pair upper upper (@code{puu.ps}).
9669 @item v2sf __builtin_mips_cvt_ps_s (float, float)
9670 Convert pair to paired single (@code{cvt.ps.s}).
9672 @item float __builtin_mips_cvt_s_pl (v2sf)
9673 Convert pair lower to single (@code{cvt.s.pl}).
9675 @item float __builtin_mips_cvt_s_pu (v2sf)
9676 Convert pair upper to single (@code{cvt.s.pu}).
9678 @item v2sf __builtin_mips_abs_ps (v2sf)
9679 Absolute value (@code{abs.ps}).
9681 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
9682 Align variable (@code{alnv.ps}).
9684 @emph{Note:} The value of the third parameter must be 0 or 4
9685 modulo 8, otherwise the result will be unpredictable. Please read the
9686 instruction description for details.
9689 The following multi-instruction functions are also available.
9690 In each case, @var{cond} can be any of the 16 floating-point conditions:
9691 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
9692 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
9693 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
9696 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9697 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9698 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
9699 @code{movt.ps}/@code{movf.ps}).
9701 The @code{movt} functions return the value @var{x} computed by:
9704 c.@var{cond}.ps @var{cc},@var{a},@var{b}
9705 mov.ps @var{x},@var{c}
9706 movt.ps @var{x},@var{d},@var{cc}
9709 The @code{movf} functions are similar but use @code{movf.ps} instead
9712 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9713 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9714 Comparison of two paired-single values (@code{c.@var{cond}.ps},
9715 @code{bc1t}/@code{bc1f}).
9717 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
9718 and return either the upper or lower half of the result. For example:
9722 if (__builtin_mips_upper_c_eq_ps (a, b))
9723 upper_halves_are_equal ();
9725 upper_halves_are_unequal ();
9727 if (__builtin_mips_lower_c_eq_ps (a, b))
9728 lower_halves_are_equal ();
9730 lower_halves_are_unequal ();
9734 @node MIPS-3D Built-in Functions
9735 @subsubsection MIPS-3D Built-in Functions
9737 The MIPS-3D Application-Specific Extension (ASE) includes additional
9738 paired-single instructions that are designed to improve the performance
9739 of 3D graphics operations. Support for these instructions is controlled
9740 by the @option{-mips3d} command-line option.
9742 The functions listed below map directly to a particular MIPS-3D
9743 instruction. Please refer to the architecture specification for
9744 more details on what each instruction does.
9747 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
9748 Reduction add (@code{addr.ps}).
9750 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
9751 Reduction multiply (@code{mulr.ps}).
9753 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
9754 Convert paired single to paired word (@code{cvt.pw.ps}).
9756 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
9757 Convert paired word to paired single (@code{cvt.ps.pw}).
9759 @item float __builtin_mips_recip1_s (float)
9760 @itemx double __builtin_mips_recip1_d (double)
9761 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
9762 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
9764 @item float __builtin_mips_recip2_s (float, float)
9765 @itemx double __builtin_mips_recip2_d (double, double)
9766 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
9767 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
9769 @item float __builtin_mips_rsqrt1_s (float)
9770 @itemx double __builtin_mips_rsqrt1_d (double)
9771 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
9772 Reduced precision reciprocal square root (sequence step 1)
9773 (@code{rsqrt1.@var{fmt}}).
9775 @item float __builtin_mips_rsqrt2_s (float, float)
9776 @itemx double __builtin_mips_rsqrt2_d (double, double)
9777 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
9778 Reduced precision reciprocal square root (sequence step 2)
9779 (@code{rsqrt2.@var{fmt}}).
9782 The following multi-instruction functions are also available.
9783 In each case, @var{cond} can be any of the 16 floating-point conditions:
9784 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
9785 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
9786 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
9789 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
9790 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
9791 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
9792 @code{bc1t}/@code{bc1f}).
9794 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
9795 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
9800 if (__builtin_mips_cabs_eq_s (a, b))
9806 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9807 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9808 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
9809 @code{bc1t}/@code{bc1f}).
9811 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
9812 and return either the upper or lower half of the result. For example:
9816 if (__builtin_mips_upper_cabs_eq_ps (a, b))
9817 upper_halves_are_equal ();
9819 upper_halves_are_unequal ();
9821 if (__builtin_mips_lower_cabs_eq_ps (a, b))
9822 lower_halves_are_equal ();
9824 lower_halves_are_unequal ();
9827 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9828 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9829 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
9830 @code{movt.ps}/@code{movf.ps}).
9832 The @code{movt} functions return the value @var{x} computed by:
9835 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
9836 mov.ps @var{x},@var{c}
9837 movt.ps @var{x},@var{d},@var{cc}
9840 The @code{movf} functions are similar but use @code{movf.ps} instead
9843 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9844 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9845 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9846 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
9847 Comparison of two paired-single values
9848 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
9849 @code{bc1any2t}/@code{bc1any2f}).
9851 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
9852 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
9853 result is true and the @code{all} forms return true if both results are true.
9858 if (__builtin_mips_any_c_eq_ps (a, b))
9863 if (__builtin_mips_all_c_eq_ps (a, b))
9869 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9870 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9871 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9872 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
9873 Comparison of four paired-single values
9874 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
9875 @code{bc1any4t}/@code{bc1any4f}).
9877 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
9878 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
9879 The @code{any} forms return true if any of the four results are true
9880 and the @code{all} forms return true if all four results are true.
9885 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
9890 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
9897 @node picoChip Built-in Functions
9898 @subsection picoChip Built-in Functions
9900 GCC provides an interface to selected machine instructions from the
9901 picoChip instruction set.
9904 @item int __builtin_sbc (int @var{value})
9905 Sign bit count. Return the number of consecutive bits in @var{value}
9906 which have the same value as the sign-bit. The result is the number of
9907 leading sign bits minus one, giving the number of redundant sign bits in
9910 @item int __builtin_byteswap (int @var{value})
9911 Byte swap. Return the result of swapping the upper and lower bytes of
9914 @item int __builtin_brev (int @var{value})
9915 Bit reversal. Return the result of reversing the bits in
9916 @var{value}. Bit 15 is swapped with bit 0, bit 14 is swapped with bit 1,
9919 @item int __builtin_adds (int @var{x}, int @var{y})
9920 Saturating addition. Return the result of adding @var{x} and @var{y},
9921 storing the value 32767 if the result overflows.
9923 @item int __builtin_subs (int @var{x}, int @var{y})
9924 Saturating subtraction. Return the result of subtracting @var{y} from
9925 @var{x}, storing the value @minus{}32768 if the result overflows.
9927 @item void __builtin_halt (void)
9928 Halt. The processor will stop execution. This built-in is useful for
9929 implementing assertions.
9933 @node Other MIPS Built-in Functions
9934 @subsection Other MIPS Built-in Functions
9936 GCC provides other MIPS-specific built-in functions:
9939 @item void __builtin_mips_cache (int @var{op}, const volatile void *@var{addr})
9940 Insert a @samp{cache} instruction with operands @var{op} and @var{addr}.
9941 GCC defines the preprocessor macro @code{___GCC_HAVE_BUILTIN_MIPS_CACHE}
9942 when this function is available.
9945 @node PowerPC AltiVec/VSX Built-in Functions
9946 @subsection PowerPC AltiVec Built-in Functions
9948 GCC provides an interface for the PowerPC family of processors to access
9949 the AltiVec operations described in Motorola's AltiVec Programming
9950 Interface Manual. The interface is made available by including
9951 @code{<altivec.h>} and using @option{-maltivec} and
9952 @option{-mabi=altivec}. The interface supports the following vector
9956 vector unsigned char
9960 vector unsigned short
9971 If @option{-mvsx} is used the following additional vector types are
9975 vector unsigned long
9980 The long types are only implemented for 64-bit code generation, and
9981 the long type is only used in the floating point/integer conversion
9984 GCC's implementation of the high-level language interface available from
9985 C and C++ code differs from Motorola's documentation in several ways.
9990 A vector constant is a list of constant expressions within curly braces.
9993 A vector initializer requires no cast if the vector constant is of the
9994 same type as the variable it is initializing.
9997 If @code{signed} or @code{unsigned} is omitted, the signedness of the
9998 vector type is the default signedness of the base type. The default
9999 varies depending on the operating system, so a portable program should
10000 always specify the signedness.
10003 Compiling with @option{-maltivec} adds keywords @code{__vector},
10004 @code{vector}, @code{__pixel}, @code{pixel}, @code{__bool} and
10005 @code{bool}. When compiling ISO C, the context-sensitive substitution
10006 of the keywords @code{vector}, @code{pixel} and @code{bool} is
10007 disabled. To use them, you must include @code{<altivec.h>} instead.
10010 GCC allows using a @code{typedef} name as the type specifier for a
10014 For C, overloaded functions are implemented with macros so the following
10018 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
10021 Since @code{vec_add} is a macro, the vector constant in the example
10022 is treated as four separate arguments. Wrap the entire argument in
10023 parentheses for this to work.
10026 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
10027 Internally, GCC uses built-in functions to achieve the functionality in
10028 the aforementioned header file, but they are not supported and are
10029 subject to change without notice.
10031 The following interfaces are supported for the generic and specific
10032 AltiVec operations and the AltiVec predicates. In cases where there
10033 is a direct mapping between generic and specific operations, only the
10034 generic names are shown here, although the specific operations can also
10037 Arguments that are documented as @code{const int} require literal
10038 integral values within the range required for that operation.
10041 vector signed char vec_abs (vector signed char);
10042 vector signed short vec_abs (vector signed short);
10043 vector signed int vec_abs (vector signed int);
10044 vector float vec_abs (vector float);
10046 vector signed char vec_abss (vector signed char);
10047 vector signed short vec_abss (vector signed short);
10048 vector signed int vec_abss (vector signed int);
10050 vector signed char vec_add (vector bool char, vector signed char);
10051 vector signed char vec_add (vector signed char, vector bool char);
10052 vector signed char vec_add (vector signed char, vector signed char);
10053 vector unsigned char vec_add (vector bool char, vector unsigned char);
10054 vector unsigned char vec_add (vector unsigned char, vector bool char);
10055 vector unsigned char vec_add (vector unsigned char,
10056 vector unsigned char);
10057 vector signed short vec_add (vector bool short, vector signed short);
10058 vector signed short vec_add (vector signed short, vector bool short);
10059 vector signed short vec_add (vector signed short, vector signed short);
10060 vector unsigned short vec_add (vector bool short,
10061 vector unsigned short);
10062 vector unsigned short vec_add (vector unsigned short,
10063 vector bool short);
10064 vector unsigned short vec_add (vector unsigned short,
10065 vector unsigned short);
10066 vector signed int vec_add (vector bool int, vector signed int);
10067 vector signed int vec_add (vector signed int, vector bool int);
10068 vector signed int vec_add (vector signed int, vector signed int);
10069 vector unsigned int vec_add (vector bool int, vector unsigned int);
10070 vector unsigned int vec_add (vector unsigned int, vector bool int);
10071 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
10072 vector float vec_add (vector float, vector float);
10074 vector float vec_vaddfp (vector float, vector float);
10076 vector signed int vec_vadduwm (vector bool int, vector signed int);
10077 vector signed int vec_vadduwm (vector signed int, vector bool int);
10078 vector signed int vec_vadduwm (vector signed int, vector signed int);
10079 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
10080 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
10081 vector unsigned int vec_vadduwm (vector unsigned int,
10082 vector unsigned int);
10084 vector signed short vec_vadduhm (vector bool short,
10085 vector signed short);
10086 vector signed short vec_vadduhm (vector signed short,
10087 vector bool short);
10088 vector signed short vec_vadduhm (vector signed short,
10089 vector signed short);
10090 vector unsigned short vec_vadduhm (vector bool short,
10091 vector unsigned short);
10092 vector unsigned short vec_vadduhm (vector unsigned short,
10093 vector bool short);
10094 vector unsigned short vec_vadduhm (vector unsigned short,
10095 vector unsigned short);
10097 vector signed char vec_vaddubm (vector bool char, vector signed char);
10098 vector signed char vec_vaddubm (vector signed char, vector bool char);
10099 vector signed char vec_vaddubm (vector signed char, vector signed char);
10100 vector unsigned char vec_vaddubm (vector bool char,
10101 vector unsigned char);
10102 vector unsigned char vec_vaddubm (vector unsigned char,
10104 vector unsigned char vec_vaddubm (vector unsigned char,
10105 vector unsigned char);
10107 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
10109 vector unsigned char vec_adds (vector bool char, vector unsigned char);
10110 vector unsigned char vec_adds (vector unsigned char, vector bool char);
10111 vector unsigned char vec_adds (vector unsigned char,
10112 vector unsigned char);
10113 vector signed char vec_adds (vector bool char, vector signed char);
10114 vector signed char vec_adds (vector signed char, vector bool char);
10115 vector signed char vec_adds (vector signed char, vector signed char);
10116 vector unsigned short vec_adds (vector bool short,
10117 vector unsigned short);
10118 vector unsigned short vec_adds (vector unsigned short,
10119 vector bool short);
10120 vector unsigned short vec_adds (vector unsigned short,
10121 vector unsigned short);
10122 vector signed short vec_adds (vector bool short, vector signed short);
10123 vector signed short vec_adds (vector signed short, vector bool short);
10124 vector signed short vec_adds (vector signed short, vector signed short);
10125 vector unsigned int vec_adds (vector bool int, vector unsigned int);
10126 vector unsigned int vec_adds (vector unsigned int, vector bool int);
10127 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
10128 vector signed int vec_adds (vector bool int, vector signed int);
10129 vector signed int vec_adds (vector signed int, vector bool int);
10130 vector signed int vec_adds (vector signed int, vector signed int);
10132 vector signed int vec_vaddsws (vector bool int, vector signed int);
10133 vector signed int vec_vaddsws (vector signed int, vector bool int);
10134 vector signed int vec_vaddsws (vector signed int, vector signed int);
10136 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
10137 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
10138 vector unsigned int vec_vadduws (vector unsigned int,
10139 vector unsigned int);
10141 vector signed short vec_vaddshs (vector bool short,
10142 vector signed short);
10143 vector signed short vec_vaddshs (vector signed short,
10144 vector bool short);
10145 vector signed short vec_vaddshs (vector signed short,
10146 vector signed short);
10148 vector unsigned short vec_vadduhs (vector bool short,
10149 vector unsigned short);
10150 vector unsigned short vec_vadduhs (vector unsigned short,
10151 vector bool short);
10152 vector unsigned short vec_vadduhs (vector unsigned short,
10153 vector unsigned short);
10155 vector signed char vec_vaddsbs (vector bool char, vector signed char);
10156 vector signed char vec_vaddsbs (vector signed char, vector bool char);
10157 vector signed char vec_vaddsbs (vector signed char, vector signed char);
10159 vector unsigned char vec_vaddubs (vector bool char,
10160 vector unsigned char);
10161 vector unsigned char vec_vaddubs (vector unsigned char,
10163 vector unsigned char vec_vaddubs (vector unsigned char,
10164 vector unsigned char);
10166 vector float vec_and (vector float, vector float);
10167 vector float vec_and (vector float, vector bool int);
10168 vector float vec_and (vector bool int, vector float);
10169 vector bool int vec_and (vector bool int, vector bool int);
10170 vector signed int vec_and (vector bool int, vector signed int);
10171 vector signed int vec_and (vector signed int, vector bool int);
10172 vector signed int vec_and (vector signed int, vector signed int);
10173 vector unsigned int vec_and (vector bool int, vector unsigned int);
10174 vector unsigned int vec_and (vector unsigned int, vector bool int);
10175 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
10176 vector bool short vec_and (vector bool short, vector bool short);
10177 vector signed short vec_and (vector bool short, vector signed short);
10178 vector signed short vec_and (vector signed short, vector bool short);
10179 vector signed short vec_and (vector signed short, vector signed short);
10180 vector unsigned short vec_and (vector bool short,
10181 vector unsigned short);
10182 vector unsigned short vec_and (vector unsigned short,
10183 vector bool short);
10184 vector unsigned short vec_and (vector unsigned short,
10185 vector unsigned short);
10186 vector signed char vec_and (vector bool char, vector signed char);
10187 vector bool char vec_and (vector bool char, vector bool char);
10188 vector signed char vec_and (vector signed char, vector bool char);
10189 vector signed char vec_and (vector signed char, vector signed char);
10190 vector unsigned char vec_and (vector bool char, vector unsigned char);
10191 vector unsigned char vec_and (vector unsigned char, vector bool char);
10192 vector unsigned char vec_and (vector unsigned char,
10193 vector unsigned char);
10195 vector float vec_andc (vector float, vector float);
10196 vector float vec_andc (vector float, vector bool int);
10197 vector float vec_andc (vector bool int, vector float);
10198 vector bool int vec_andc (vector bool int, vector bool int);
10199 vector signed int vec_andc (vector bool int, vector signed int);
10200 vector signed int vec_andc (vector signed int, vector bool int);
10201 vector signed int vec_andc (vector signed int, vector signed int);
10202 vector unsigned int vec_andc (vector bool int, vector unsigned int);
10203 vector unsigned int vec_andc (vector unsigned int, vector bool int);
10204 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
10205 vector bool short vec_andc (vector bool short, vector bool short);
10206 vector signed short vec_andc (vector bool short, vector signed short);
10207 vector signed short vec_andc (vector signed short, vector bool short);
10208 vector signed short vec_andc (vector signed short, vector signed short);
10209 vector unsigned short vec_andc (vector bool short,
10210 vector unsigned short);
10211 vector unsigned short vec_andc (vector unsigned short,
10212 vector bool short);
10213 vector unsigned short vec_andc (vector unsigned short,
10214 vector unsigned short);
10215 vector signed char vec_andc (vector bool char, vector signed char);
10216 vector bool char vec_andc (vector bool char, vector bool char);
10217 vector signed char vec_andc (vector signed char, vector bool char);
10218 vector signed char vec_andc (vector signed char, vector signed char);
10219 vector unsigned char vec_andc (vector bool char, vector unsigned char);
10220 vector unsigned char vec_andc (vector unsigned char, vector bool char);
10221 vector unsigned char vec_andc (vector unsigned char,
10222 vector unsigned char);
10224 vector unsigned char vec_avg (vector unsigned char,
10225 vector unsigned char);
10226 vector signed char vec_avg (vector signed char, vector signed char);
10227 vector unsigned short vec_avg (vector unsigned short,
10228 vector unsigned short);
10229 vector signed short vec_avg (vector signed short, vector signed short);
10230 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
10231 vector signed int vec_avg (vector signed int, vector signed int);
10233 vector signed int vec_vavgsw (vector signed int, vector signed int);
10235 vector unsigned int vec_vavguw (vector unsigned int,
10236 vector unsigned int);
10238 vector signed short vec_vavgsh (vector signed short,
10239 vector signed short);
10241 vector unsigned short vec_vavguh (vector unsigned short,
10242 vector unsigned short);
10244 vector signed char vec_vavgsb (vector signed char, vector signed char);
10246 vector unsigned char vec_vavgub (vector unsigned char,
10247 vector unsigned char);
10249 vector float vec_copysign (vector float);
10251 vector float vec_ceil (vector float);
10253 vector signed int vec_cmpb (vector float, vector float);
10255 vector bool char vec_cmpeq (vector signed char, vector signed char);
10256 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
10257 vector bool short vec_cmpeq (vector signed short, vector signed short);
10258 vector bool short vec_cmpeq (vector unsigned short,
10259 vector unsigned short);
10260 vector bool int vec_cmpeq (vector signed int, vector signed int);
10261 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
10262 vector bool int vec_cmpeq (vector float, vector float);
10264 vector bool int vec_vcmpeqfp (vector float, vector float);
10266 vector bool int vec_vcmpequw (vector signed int, vector signed int);
10267 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
10269 vector bool short vec_vcmpequh (vector signed short,
10270 vector signed short);
10271 vector bool short vec_vcmpequh (vector unsigned short,
10272 vector unsigned short);
10274 vector bool char vec_vcmpequb (vector signed char, vector signed char);
10275 vector bool char vec_vcmpequb (vector unsigned char,
10276 vector unsigned char);
10278 vector bool int vec_cmpge (vector float, vector float);
10280 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
10281 vector bool char vec_cmpgt (vector signed char, vector signed char);
10282 vector bool short vec_cmpgt (vector unsigned short,
10283 vector unsigned short);
10284 vector bool short vec_cmpgt (vector signed short, vector signed short);
10285 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
10286 vector bool int vec_cmpgt (vector signed int, vector signed int);
10287 vector bool int vec_cmpgt (vector float, vector float);
10289 vector bool int vec_vcmpgtfp (vector float, vector float);
10291 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
10293 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
10295 vector bool short vec_vcmpgtsh (vector signed short,
10296 vector signed short);
10298 vector bool short vec_vcmpgtuh (vector unsigned short,
10299 vector unsigned short);
10301 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
10303 vector bool char vec_vcmpgtub (vector unsigned char,
10304 vector unsigned char);
10306 vector bool int vec_cmple (vector float, vector float);
10308 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
10309 vector bool char vec_cmplt (vector signed char, vector signed char);
10310 vector bool short vec_cmplt (vector unsigned short,
10311 vector unsigned short);
10312 vector bool short vec_cmplt (vector signed short, vector signed short);
10313 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
10314 vector bool int vec_cmplt (vector signed int, vector signed int);
10315 vector bool int vec_cmplt (vector float, vector float);
10317 vector float vec_ctf (vector unsigned int, const int);
10318 vector float vec_ctf (vector signed int, const int);
10320 vector float vec_vcfsx (vector signed int, const int);
10322 vector float vec_vcfux (vector unsigned int, const int);
10324 vector signed int vec_cts (vector float, const int);
10326 vector unsigned int vec_ctu (vector float, const int);
10328 void vec_dss (const int);
10330 void vec_dssall (void);
10332 void vec_dst (const vector unsigned char *, int, const int);
10333 void vec_dst (const vector signed char *, int, const int);
10334 void vec_dst (const vector bool char *, int, const int);
10335 void vec_dst (const vector unsigned short *, int, const int);
10336 void vec_dst (const vector signed short *, int, const int);
10337 void vec_dst (const vector bool short *, int, const int);
10338 void vec_dst (const vector pixel *, int, const int);
10339 void vec_dst (const vector unsigned int *, int, const int);
10340 void vec_dst (const vector signed int *, int, const int);
10341 void vec_dst (const vector bool int *, int, const int);
10342 void vec_dst (const vector float *, int, const int);
10343 void vec_dst (const unsigned char *, int, const int);
10344 void vec_dst (const signed char *, int, const int);
10345 void vec_dst (const unsigned short *, int, const int);
10346 void vec_dst (const short *, int, const int);
10347 void vec_dst (const unsigned int *, int, const int);
10348 void vec_dst (const int *, int, const int);
10349 void vec_dst (const unsigned long *, int, const int);
10350 void vec_dst (const long *, int, const int);
10351 void vec_dst (const float *, int, const int);
10353 void vec_dstst (const vector unsigned char *, int, const int);
10354 void vec_dstst (const vector signed char *, int, const int);
10355 void vec_dstst (const vector bool char *, int, const int);
10356 void vec_dstst (const vector unsigned short *, int, const int);
10357 void vec_dstst (const vector signed short *, int, const int);
10358 void vec_dstst (const vector bool short *, int, const int);
10359 void vec_dstst (const vector pixel *, int, const int);
10360 void vec_dstst (const vector unsigned int *, int, const int);
10361 void vec_dstst (const vector signed int *, int, const int);
10362 void vec_dstst (const vector bool int *, int, const int);
10363 void vec_dstst (const vector float *, int, const int);
10364 void vec_dstst (const unsigned char *, int, const int);
10365 void vec_dstst (const signed char *, int, const int);
10366 void vec_dstst (const unsigned short *, int, const int);
10367 void vec_dstst (const short *, int, const int);
10368 void vec_dstst (const unsigned int *, int, const int);
10369 void vec_dstst (const int *, int, const int);
10370 void vec_dstst (const unsigned long *, int, const int);
10371 void vec_dstst (const long *, int, const int);
10372 void vec_dstst (const float *, int, const int);
10374 void vec_dststt (const vector unsigned char *, int, const int);
10375 void vec_dststt (const vector signed char *, int, const int);
10376 void vec_dststt (const vector bool char *, int, const int);
10377 void vec_dststt (const vector unsigned short *, int, const int);
10378 void vec_dststt (const vector signed short *, int, const int);
10379 void vec_dststt (const vector bool short *, int, const int);
10380 void vec_dststt (const vector pixel *, int, const int);
10381 void vec_dststt (const vector unsigned int *, int, const int);
10382 void vec_dststt (const vector signed int *, int, const int);
10383 void vec_dststt (const vector bool int *, int, const int);
10384 void vec_dststt (const vector float *, int, const int);
10385 void vec_dststt (const unsigned char *, int, const int);
10386 void vec_dststt (const signed char *, int, const int);
10387 void vec_dststt (const unsigned short *, int, const int);
10388 void vec_dststt (const short *, int, const int);
10389 void vec_dststt (const unsigned int *, int, const int);
10390 void vec_dststt (const int *, int, const int);
10391 void vec_dststt (const unsigned long *, int, const int);
10392 void vec_dststt (const long *, int, const int);
10393 void vec_dststt (const float *, int, const int);
10395 void vec_dstt (const vector unsigned char *, int, const int);
10396 void vec_dstt (const vector signed char *, int, const int);
10397 void vec_dstt (const vector bool char *, int, const int);
10398 void vec_dstt (const vector unsigned short *, int, const int);
10399 void vec_dstt (const vector signed short *, int, const int);
10400 void vec_dstt (const vector bool short *, int, const int);
10401 void vec_dstt (const vector pixel *, int, const int);
10402 void vec_dstt (const vector unsigned int *, int, const int);
10403 void vec_dstt (const vector signed int *, int, const int);
10404 void vec_dstt (const vector bool int *, int, const int);
10405 void vec_dstt (const vector float *, int, const int);
10406 void vec_dstt (const unsigned char *, int, const int);
10407 void vec_dstt (const signed char *, int, const int);
10408 void vec_dstt (const unsigned short *, int, const int);
10409 void vec_dstt (const short *, int, const int);
10410 void vec_dstt (const unsigned int *, int, const int);
10411 void vec_dstt (const int *, int, const int);
10412 void vec_dstt (const unsigned long *, int, const int);
10413 void vec_dstt (const long *, int, const int);
10414 void vec_dstt (const float *, int, const int);
10416 vector float vec_expte (vector float);
10418 vector float vec_floor (vector float);
10420 vector float vec_ld (int, const vector float *);
10421 vector float vec_ld (int, const float *);
10422 vector bool int vec_ld (int, const vector bool int *);
10423 vector signed int vec_ld (int, const vector signed int *);
10424 vector signed int vec_ld (int, const int *);
10425 vector signed int vec_ld (int, const long *);
10426 vector unsigned int vec_ld (int, const vector unsigned int *);
10427 vector unsigned int vec_ld (int, const unsigned int *);
10428 vector unsigned int vec_ld (int, const unsigned long *);
10429 vector bool short vec_ld (int, const vector bool short *);
10430 vector pixel vec_ld (int, const vector pixel *);
10431 vector signed short vec_ld (int, const vector signed short *);
10432 vector signed short vec_ld (int, const short *);
10433 vector unsigned short vec_ld (int, const vector unsigned short *);
10434 vector unsigned short vec_ld (int, const unsigned short *);
10435 vector bool char vec_ld (int, const vector bool char *);
10436 vector signed char vec_ld (int, const vector signed char *);
10437 vector signed char vec_ld (int, const signed char *);
10438 vector unsigned char vec_ld (int, const vector unsigned char *);
10439 vector unsigned char vec_ld (int, const unsigned char *);
10441 vector signed char vec_lde (int, const signed char *);
10442 vector unsigned char vec_lde (int, const unsigned char *);
10443 vector signed short vec_lde (int, const short *);
10444 vector unsigned short vec_lde (int, const unsigned short *);
10445 vector float vec_lde (int, const float *);
10446 vector signed int vec_lde (int, const int *);
10447 vector unsigned int vec_lde (int, const unsigned int *);
10448 vector signed int vec_lde (int, const long *);
10449 vector unsigned int vec_lde (int, const unsigned long *);
10451 vector float vec_lvewx (int, float *);
10452 vector signed int vec_lvewx (int, int *);
10453 vector unsigned int vec_lvewx (int, unsigned int *);
10454 vector signed int vec_lvewx (int, long *);
10455 vector unsigned int vec_lvewx (int, unsigned long *);
10457 vector signed short vec_lvehx (int, short *);
10458 vector unsigned short vec_lvehx (int, unsigned short *);
10460 vector signed char vec_lvebx (int, char *);
10461 vector unsigned char vec_lvebx (int, unsigned char *);
10463 vector float vec_ldl (int, const vector float *);
10464 vector float vec_ldl (int, const float *);
10465 vector bool int vec_ldl (int, const vector bool int *);
10466 vector signed int vec_ldl (int, const vector signed int *);
10467 vector signed int vec_ldl (int, const int *);
10468 vector signed int vec_ldl (int, const long *);
10469 vector unsigned int vec_ldl (int, const vector unsigned int *);
10470 vector unsigned int vec_ldl (int, const unsigned int *);
10471 vector unsigned int vec_ldl (int, const unsigned long *);
10472 vector bool short vec_ldl (int, const vector bool short *);
10473 vector pixel vec_ldl (int, const vector pixel *);
10474 vector signed short vec_ldl (int, const vector signed short *);
10475 vector signed short vec_ldl (int, const short *);
10476 vector unsigned short vec_ldl (int, const vector unsigned short *);
10477 vector unsigned short vec_ldl (int, const unsigned short *);
10478 vector bool char vec_ldl (int, const vector bool char *);
10479 vector signed char vec_ldl (int, const vector signed char *);
10480 vector signed char vec_ldl (int, const signed char *);
10481 vector unsigned char vec_ldl (int, const vector unsigned char *);
10482 vector unsigned char vec_ldl (int, const unsigned char *);
10484 vector float vec_loge (vector float);
10486 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
10487 vector unsigned char vec_lvsl (int, const volatile signed char *);
10488 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
10489 vector unsigned char vec_lvsl (int, const volatile short *);
10490 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
10491 vector unsigned char vec_lvsl (int, const volatile int *);
10492 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
10493 vector unsigned char vec_lvsl (int, const volatile long *);
10494 vector unsigned char vec_lvsl (int, const volatile float *);
10496 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
10497 vector unsigned char vec_lvsr (int, const volatile signed char *);
10498 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
10499 vector unsigned char vec_lvsr (int, const volatile short *);
10500 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
10501 vector unsigned char vec_lvsr (int, const volatile int *);
10502 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
10503 vector unsigned char vec_lvsr (int, const volatile long *);
10504 vector unsigned char vec_lvsr (int, const volatile float *);
10506 vector float vec_madd (vector float, vector float, vector float);
10508 vector signed short vec_madds (vector signed short,
10509 vector signed short,
10510 vector signed short);
10512 vector unsigned char vec_max (vector bool char, vector unsigned char);
10513 vector unsigned char vec_max (vector unsigned char, vector bool char);
10514 vector unsigned char vec_max (vector unsigned char,
10515 vector unsigned char);
10516 vector signed char vec_max (vector bool char, vector signed char);
10517 vector signed char vec_max (vector signed char, vector bool char);
10518 vector signed char vec_max (vector signed char, vector signed char);
10519 vector unsigned short vec_max (vector bool short,
10520 vector unsigned short);
10521 vector unsigned short vec_max (vector unsigned short,
10522 vector bool short);
10523 vector unsigned short vec_max (vector unsigned short,
10524 vector unsigned short);
10525 vector signed short vec_max (vector bool short, vector signed short);
10526 vector signed short vec_max (vector signed short, vector bool short);
10527 vector signed short vec_max (vector signed short, vector signed short);
10528 vector unsigned int vec_max (vector bool int, vector unsigned int);
10529 vector unsigned int vec_max (vector unsigned int, vector bool int);
10530 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
10531 vector signed int vec_max (vector bool int, vector signed int);
10532 vector signed int vec_max (vector signed int, vector bool int);
10533 vector signed int vec_max (vector signed int, vector signed int);
10534 vector float vec_max (vector float, vector float);
10536 vector float vec_vmaxfp (vector float, vector float);
10538 vector signed int vec_vmaxsw (vector bool int, vector signed int);
10539 vector signed int vec_vmaxsw (vector signed int, vector bool int);
10540 vector signed int vec_vmaxsw (vector signed int, vector signed int);
10542 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
10543 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
10544 vector unsigned int vec_vmaxuw (vector unsigned int,
10545 vector unsigned int);
10547 vector signed short vec_vmaxsh (vector bool short, vector signed short);
10548 vector signed short vec_vmaxsh (vector signed short, vector bool short);
10549 vector signed short vec_vmaxsh (vector signed short,
10550 vector signed short);
10552 vector unsigned short vec_vmaxuh (vector bool short,
10553 vector unsigned short);
10554 vector unsigned short vec_vmaxuh (vector unsigned short,
10555 vector bool short);
10556 vector unsigned short vec_vmaxuh (vector unsigned short,
10557 vector unsigned short);
10559 vector signed char vec_vmaxsb (vector bool char, vector signed char);
10560 vector signed char vec_vmaxsb (vector signed char, vector bool char);
10561 vector signed char vec_vmaxsb (vector signed char, vector signed char);
10563 vector unsigned char vec_vmaxub (vector bool char,
10564 vector unsigned char);
10565 vector unsigned char vec_vmaxub (vector unsigned char,
10567 vector unsigned char vec_vmaxub (vector unsigned char,
10568 vector unsigned char);
10570 vector bool char vec_mergeh (vector bool char, vector bool char);
10571 vector signed char vec_mergeh (vector signed char, vector signed char);
10572 vector unsigned char vec_mergeh (vector unsigned char,
10573 vector unsigned char);
10574 vector bool short vec_mergeh (vector bool short, vector bool short);
10575 vector pixel vec_mergeh (vector pixel, vector pixel);
10576 vector signed short vec_mergeh (vector signed short,
10577 vector signed short);
10578 vector unsigned short vec_mergeh (vector unsigned short,
10579 vector unsigned short);
10580 vector float vec_mergeh (vector float, vector float);
10581 vector bool int vec_mergeh (vector bool int, vector bool int);
10582 vector signed int vec_mergeh (vector signed int, vector signed int);
10583 vector unsigned int vec_mergeh (vector unsigned int,
10584 vector unsigned int);
10586 vector float vec_vmrghw (vector float, vector float);
10587 vector bool int vec_vmrghw (vector bool int, vector bool int);
10588 vector signed int vec_vmrghw (vector signed int, vector signed int);
10589 vector unsigned int vec_vmrghw (vector unsigned int,
10590 vector unsigned int);
10592 vector bool short vec_vmrghh (vector bool short, vector bool short);
10593 vector signed short vec_vmrghh (vector signed short,
10594 vector signed short);
10595 vector unsigned short vec_vmrghh (vector unsigned short,
10596 vector unsigned short);
10597 vector pixel vec_vmrghh (vector pixel, vector pixel);
10599 vector bool char vec_vmrghb (vector bool char, vector bool char);
10600 vector signed char vec_vmrghb (vector signed char, vector signed char);
10601 vector unsigned char vec_vmrghb (vector unsigned char,
10602 vector unsigned char);
10604 vector bool char vec_mergel (vector bool char, vector bool char);
10605 vector signed char vec_mergel (vector signed char, vector signed char);
10606 vector unsigned char vec_mergel (vector unsigned char,
10607 vector unsigned char);
10608 vector bool short vec_mergel (vector bool short, vector bool short);
10609 vector pixel vec_mergel (vector pixel, vector pixel);
10610 vector signed short vec_mergel (vector signed short,
10611 vector signed short);
10612 vector unsigned short vec_mergel (vector unsigned short,
10613 vector unsigned short);
10614 vector float vec_mergel (vector float, vector float);
10615 vector bool int vec_mergel (vector bool int, vector bool int);
10616 vector signed int vec_mergel (vector signed int, vector signed int);
10617 vector unsigned int vec_mergel (vector unsigned int,
10618 vector unsigned int);
10620 vector float vec_vmrglw (vector float, vector float);
10621 vector signed int vec_vmrglw (vector signed int, vector signed int);
10622 vector unsigned int vec_vmrglw (vector unsigned int,
10623 vector unsigned int);
10624 vector bool int vec_vmrglw (vector bool int, vector bool int);
10626 vector bool short vec_vmrglh (vector bool short, vector bool short);
10627 vector signed short vec_vmrglh (vector signed short,
10628 vector signed short);
10629 vector unsigned short vec_vmrglh (vector unsigned short,
10630 vector unsigned short);
10631 vector pixel vec_vmrglh (vector pixel, vector pixel);
10633 vector bool char vec_vmrglb (vector bool char, vector bool char);
10634 vector signed char vec_vmrglb (vector signed char, vector signed char);
10635 vector unsigned char vec_vmrglb (vector unsigned char,
10636 vector unsigned char);
10638 vector unsigned short vec_mfvscr (void);
10640 vector unsigned char vec_min (vector bool char, vector unsigned char);
10641 vector unsigned char vec_min (vector unsigned char, vector bool char);
10642 vector unsigned char vec_min (vector unsigned char,
10643 vector unsigned char);
10644 vector signed char vec_min (vector bool char, vector signed char);
10645 vector signed char vec_min (vector signed char, vector bool char);
10646 vector signed char vec_min (vector signed char, vector signed char);
10647 vector unsigned short vec_min (vector bool short,
10648 vector unsigned short);
10649 vector unsigned short vec_min (vector unsigned short,
10650 vector bool short);
10651 vector unsigned short vec_min (vector unsigned short,
10652 vector unsigned short);
10653 vector signed short vec_min (vector bool short, vector signed short);
10654 vector signed short vec_min (vector signed short, vector bool short);
10655 vector signed short vec_min (vector signed short, vector signed short);
10656 vector unsigned int vec_min (vector bool int, vector unsigned int);
10657 vector unsigned int vec_min (vector unsigned int, vector bool int);
10658 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
10659 vector signed int vec_min (vector bool int, vector signed int);
10660 vector signed int vec_min (vector signed int, vector bool int);
10661 vector signed int vec_min (vector signed int, vector signed int);
10662 vector float vec_min (vector float, vector float);
10664 vector float vec_vminfp (vector float, vector float);
10666 vector signed int vec_vminsw (vector bool int, vector signed int);
10667 vector signed int vec_vminsw (vector signed int, vector bool int);
10668 vector signed int vec_vminsw (vector signed int, vector signed int);
10670 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
10671 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
10672 vector unsigned int vec_vminuw (vector unsigned int,
10673 vector unsigned int);
10675 vector signed short vec_vminsh (vector bool short, vector signed short);
10676 vector signed short vec_vminsh (vector signed short, vector bool short);
10677 vector signed short vec_vminsh (vector signed short,
10678 vector signed short);
10680 vector unsigned short vec_vminuh (vector bool short,
10681 vector unsigned short);
10682 vector unsigned short vec_vminuh (vector unsigned short,
10683 vector bool short);
10684 vector unsigned short vec_vminuh (vector unsigned short,
10685 vector unsigned short);
10687 vector signed char vec_vminsb (vector bool char, vector signed char);
10688 vector signed char vec_vminsb (vector signed char, vector bool char);
10689 vector signed char vec_vminsb (vector signed char, vector signed char);
10691 vector unsigned char vec_vminub (vector bool char,
10692 vector unsigned char);
10693 vector unsigned char vec_vminub (vector unsigned char,
10695 vector unsigned char vec_vminub (vector unsigned char,
10696 vector unsigned char);
10698 vector signed short vec_mladd (vector signed short,
10699 vector signed short,
10700 vector signed short);
10701 vector signed short vec_mladd (vector signed short,
10702 vector unsigned short,
10703 vector unsigned short);
10704 vector signed short vec_mladd (vector unsigned short,
10705 vector signed short,
10706 vector signed short);
10707 vector unsigned short vec_mladd (vector unsigned short,
10708 vector unsigned short,
10709 vector unsigned short);
10711 vector signed short vec_mradds (vector signed short,
10712 vector signed short,
10713 vector signed short);
10715 vector unsigned int vec_msum (vector unsigned char,
10716 vector unsigned char,
10717 vector unsigned int);
10718 vector signed int vec_msum (vector signed char,
10719 vector unsigned char,
10720 vector signed int);
10721 vector unsigned int vec_msum (vector unsigned short,
10722 vector unsigned short,
10723 vector unsigned int);
10724 vector signed int vec_msum (vector signed short,
10725 vector signed short,
10726 vector signed int);
10728 vector signed int vec_vmsumshm (vector signed short,
10729 vector signed short,
10730 vector signed int);
10732 vector unsigned int vec_vmsumuhm (vector unsigned short,
10733 vector unsigned short,
10734 vector unsigned int);
10736 vector signed int vec_vmsummbm (vector signed char,
10737 vector unsigned char,
10738 vector signed int);
10740 vector unsigned int vec_vmsumubm (vector unsigned char,
10741 vector unsigned char,
10742 vector unsigned int);
10744 vector unsigned int vec_msums (vector unsigned short,
10745 vector unsigned short,
10746 vector unsigned int);
10747 vector signed int vec_msums (vector signed short,
10748 vector signed short,
10749 vector signed int);
10751 vector signed int vec_vmsumshs (vector signed short,
10752 vector signed short,
10753 vector signed int);
10755 vector unsigned int vec_vmsumuhs (vector unsigned short,
10756 vector unsigned short,
10757 vector unsigned int);
10759 void vec_mtvscr (vector signed int);
10760 void vec_mtvscr (vector unsigned int);
10761 void vec_mtvscr (vector bool int);
10762 void vec_mtvscr (vector signed short);
10763 void vec_mtvscr (vector unsigned short);
10764 void vec_mtvscr (vector bool short);
10765 void vec_mtvscr (vector pixel);
10766 void vec_mtvscr (vector signed char);
10767 void vec_mtvscr (vector unsigned char);
10768 void vec_mtvscr (vector bool char);
10770 vector unsigned short vec_mule (vector unsigned char,
10771 vector unsigned char);
10772 vector signed short vec_mule (vector signed char,
10773 vector signed char);
10774 vector unsigned int vec_mule (vector unsigned short,
10775 vector unsigned short);
10776 vector signed int vec_mule (vector signed short, vector signed short);
10778 vector signed int vec_vmulesh (vector signed short,
10779 vector signed short);
10781 vector unsigned int vec_vmuleuh (vector unsigned short,
10782 vector unsigned short);
10784 vector signed short vec_vmulesb (vector signed char,
10785 vector signed char);
10787 vector unsigned short vec_vmuleub (vector unsigned char,
10788 vector unsigned char);
10790 vector unsigned short vec_mulo (vector unsigned char,
10791 vector unsigned char);
10792 vector signed short vec_mulo (vector signed char, vector signed char);
10793 vector unsigned int vec_mulo (vector unsigned short,
10794 vector unsigned short);
10795 vector signed int vec_mulo (vector signed short, vector signed short);
10797 vector signed int vec_vmulosh (vector signed short,
10798 vector signed short);
10800 vector unsigned int vec_vmulouh (vector unsigned short,
10801 vector unsigned short);
10803 vector signed short vec_vmulosb (vector signed char,
10804 vector signed char);
10806 vector unsigned short vec_vmuloub (vector unsigned char,
10807 vector unsigned char);
10809 vector float vec_nmsub (vector float, vector float, vector float);
10811 vector float vec_nor (vector float, vector float);
10812 vector signed int vec_nor (vector signed int, vector signed int);
10813 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
10814 vector bool int vec_nor (vector bool int, vector bool int);
10815 vector signed short vec_nor (vector signed short, vector signed short);
10816 vector unsigned short vec_nor (vector unsigned short,
10817 vector unsigned short);
10818 vector bool short vec_nor (vector bool short, vector bool short);
10819 vector signed char vec_nor (vector signed char, vector signed char);
10820 vector unsigned char vec_nor (vector unsigned char,
10821 vector unsigned char);
10822 vector bool char vec_nor (vector bool char, vector bool char);
10824 vector float vec_or (vector float, vector float);
10825 vector float vec_or (vector float, vector bool int);
10826 vector float vec_or (vector bool int, vector float);
10827 vector bool int vec_or (vector bool int, vector bool int);
10828 vector signed int vec_or (vector bool int, vector signed int);
10829 vector signed int vec_or (vector signed int, vector bool int);
10830 vector signed int vec_or (vector signed int, vector signed int);
10831 vector unsigned int vec_or (vector bool int, vector unsigned int);
10832 vector unsigned int vec_or (vector unsigned int, vector bool int);
10833 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
10834 vector bool short vec_or (vector bool short, vector bool short);
10835 vector signed short vec_or (vector bool short, vector signed short);
10836 vector signed short vec_or (vector signed short, vector bool short);
10837 vector signed short vec_or (vector signed short, vector signed short);
10838 vector unsigned short vec_or (vector bool short, vector unsigned short);
10839 vector unsigned short vec_or (vector unsigned short, vector bool short);
10840 vector unsigned short vec_or (vector unsigned short,
10841 vector unsigned short);
10842 vector signed char vec_or (vector bool char, vector signed char);
10843 vector bool char vec_or (vector bool char, vector bool char);
10844 vector signed char vec_or (vector signed char, vector bool char);
10845 vector signed char vec_or (vector signed char, vector signed char);
10846 vector unsigned char vec_or (vector bool char, vector unsigned char);
10847 vector unsigned char vec_or (vector unsigned char, vector bool char);
10848 vector unsigned char vec_or (vector unsigned char,
10849 vector unsigned char);
10851 vector signed char vec_pack (vector signed short, vector signed short);
10852 vector unsigned char vec_pack (vector unsigned short,
10853 vector unsigned short);
10854 vector bool char vec_pack (vector bool short, vector bool short);
10855 vector signed short vec_pack (vector signed int, vector signed int);
10856 vector unsigned short vec_pack (vector unsigned int,
10857 vector unsigned int);
10858 vector bool short vec_pack (vector bool int, vector bool int);
10860 vector bool short vec_vpkuwum (vector bool int, vector bool int);
10861 vector signed short vec_vpkuwum (vector signed int, vector signed int);
10862 vector unsigned short vec_vpkuwum (vector unsigned int,
10863 vector unsigned int);
10865 vector bool char vec_vpkuhum (vector bool short, vector bool short);
10866 vector signed char vec_vpkuhum (vector signed short,
10867 vector signed short);
10868 vector unsigned char vec_vpkuhum (vector unsigned short,
10869 vector unsigned short);
10871 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
10873 vector unsigned char vec_packs (vector unsigned short,
10874 vector unsigned short);
10875 vector signed char vec_packs (vector signed short, vector signed short);
10876 vector unsigned short vec_packs (vector unsigned int,
10877 vector unsigned int);
10878 vector signed short vec_packs (vector signed int, vector signed int);
10880 vector signed short vec_vpkswss (vector signed int, vector signed int);
10882 vector unsigned short vec_vpkuwus (vector unsigned int,
10883 vector unsigned int);
10885 vector signed char vec_vpkshss (vector signed short,
10886 vector signed short);
10888 vector unsigned char vec_vpkuhus (vector unsigned short,
10889 vector unsigned short);
10891 vector unsigned char vec_packsu (vector unsigned short,
10892 vector unsigned short);
10893 vector unsigned char vec_packsu (vector signed short,
10894 vector signed short);
10895 vector unsigned short vec_packsu (vector unsigned int,
10896 vector unsigned int);
10897 vector unsigned short vec_packsu (vector signed int, vector signed int);
10899 vector unsigned short vec_vpkswus (vector signed int,
10900 vector signed int);
10902 vector unsigned char vec_vpkshus (vector signed short,
10903 vector signed short);
10905 vector float vec_perm (vector float,
10907 vector unsigned char);
10908 vector signed int vec_perm (vector signed int,
10910 vector unsigned char);
10911 vector unsigned int vec_perm (vector unsigned int,
10912 vector unsigned int,
10913 vector unsigned char);
10914 vector bool int vec_perm (vector bool int,
10916 vector unsigned char);
10917 vector signed short vec_perm (vector signed short,
10918 vector signed short,
10919 vector unsigned char);
10920 vector unsigned short vec_perm (vector unsigned short,
10921 vector unsigned short,
10922 vector unsigned char);
10923 vector bool short vec_perm (vector bool short,
10925 vector unsigned char);
10926 vector pixel vec_perm (vector pixel,
10928 vector unsigned char);
10929 vector signed char vec_perm (vector signed char,
10930 vector signed char,
10931 vector unsigned char);
10932 vector unsigned char vec_perm (vector unsigned char,
10933 vector unsigned char,
10934 vector unsigned char);
10935 vector bool char vec_perm (vector bool char,
10937 vector unsigned char);
10939 vector float vec_re (vector float);
10941 vector signed char vec_rl (vector signed char,
10942 vector unsigned char);
10943 vector unsigned char vec_rl (vector unsigned char,
10944 vector unsigned char);
10945 vector signed short vec_rl (vector signed short, vector unsigned short);
10946 vector unsigned short vec_rl (vector unsigned short,
10947 vector unsigned short);
10948 vector signed int vec_rl (vector signed int, vector unsigned int);
10949 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
10951 vector signed int vec_vrlw (vector signed int, vector unsigned int);
10952 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
10954 vector signed short vec_vrlh (vector signed short,
10955 vector unsigned short);
10956 vector unsigned short vec_vrlh (vector unsigned short,
10957 vector unsigned short);
10959 vector signed char vec_vrlb (vector signed char, vector unsigned char);
10960 vector unsigned char vec_vrlb (vector unsigned char,
10961 vector unsigned char);
10963 vector float vec_round (vector float);
10965 vector float vec_rsqrte (vector float);
10967 vector float vec_sel (vector float, vector float, vector bool int);
10968 vector float vec_sel (vector float, vector float, vector unsigned int);
10969 vector signed int vec_sel (vector signed int,
10972 vector signed int vec_sel (vector signed int,
10974 vector unsigned int);
10975 vector unsigned int vec_sel (vector unsigned int,
10976 vector unsigned int,
10978 vector unsigned int vec_sel (vector unsigned int,
10979 vector unsigned int,
10980 vector unsigned int);
10981 vector bool int vec_sel (vector bool int,
10984 vector bool int vec_sel (vector bool int,
10986 vector unsigned int);
10987 vector signed short vec_sel (vector signed short,
10988 vector signed short,
10989 vector bool short);
10990 vector signed short vec_sel (vector signed short,
10991 vector signed short,
10992 vector unsigned short);
10993 vector unsigned short vec_sel (vector unsigned short,
10994 vector unsigned short,
10995 vector bool short);
10996 vector unsigned short vec_sel (vector unsigned short,
10997 vector unsigned short,
10998 vector unsigned short);
10999 vector bool short vec_sel (vector bool short,
11001 vector bool short);
11002 vector bool short vec_sel (vector bool short,
11004 vector unsigned short);
11005 vector signed char vec_sel (vector signed char,
11006 vector signed char,
11008 vector signed char vec_sel (vector signed char,
11009 vector signed char,
11010 vector unsigned char);
11011 vector unsigned char vec_sel (vector unsigned char,
11012 vector unsigned char,
11014 vector unsigned char vec_sel (vector unsigned char,
11015 vector unsigned char,
11016 vector unsigned char);
11017 vector bool char vec_sel (vector bool char,
11020 vector bool char vec_sel (vector bool char,
11022 vector unsigned char);
11024 vector signed char vec_sl (vector signed char,
11025 vector unsigned char);
11026 vector unsigned char vec_sl (vector unsigned char,
11027 vector unsigned char);
11028 vector signed short vec_sl (vector signed short, vector unsigned short);
11029 vector unsigned short vec_sl (vector unsigned short,
11030 vector unsigned short);
11031 vector signed int vec_sl (vector signed int, vector unsigned int);
11032 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
11034 vector signed int vec_vslw (vector signed int, vector unsigned int);
11035 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
11037 vector signed short vec_vslh (vector signed short,
11038 vector unsigned short);
11039 vector unsigned short vec_vslh (vector unsigned short,
11040 vector unsigned short);
11042 vector signed char vec_vslb (vector signed char, vector unsigned char);
11043 vector unsigned char vec_vslb (vector unsigned char,
11044 vector unsigned char);
11046 vector float vec_sld (vector float, vector float, const int);
11047 vector signed int vec_sld (vector signed int,
11050 vector unsigned int vec_sld (vector unsigned int,
11051 vector unsigned int,
11053 vector bool int vec_sld (vector bool int,
11056 vector signed short vec_sld (vector signed short,
11057 vector signed short,
11059 vector unsigned short vec_sld (vector unsigned short,
11060 vector unsigned short,
11062 vector bool short vec_sld (vector bool short,
11065 vector pixel vec_sld (vector pixel,
11068 vector signed char vec_sld (vector signed char,
11069 vector signed char,
11071 vector unsigned char vec_sld (vector unsigned char,
11072 vector unsigned char,
11074 vector bool char vec_sld (vector bool char,
11078 vector signed int vec_sll (vector signed int,
11079 vector unsigned int);
11080 vector signed int vec_sll (vector signed int,
11081 vector unsigned short);
11082 vector signed int vec_sll (vector signed int,
11083 vector unsigned char);
11084 vector unsigned int vec_sll (vector unsigned int,
11085 vector unsigned int);
11086 vector unsigned int vec_sll (vector unsigned int,
11087 vector unsigned short);
11088 vector unsigned int vec_sll (vector unsigned int,
11089 vector unsigned char);
11090 vector bool int vec_sll (vector bool int,
11091 vector unsigned int);
11092 vector bool int vec_sll (vector bool int,
11093 vector unsigned short);
11094 vector bool int vec_sll (vector bool int,
11095 vector unsigned char);
11096 vector signed short vec_sll (vector signed short,
11097 vector unsigned int);
11098 vector signed short vec_sll (vector signed short,
11099 vector unsigned short);
11100 vector signed short vec_sll (vector signed short,
11101 vector unsigned char);
11102 vector unsigned short vec_sll (vector unsigned short,
11103 vector unsigned int);
11104 vector unsigned short vec_sll (vector unsigned short,
11105 vector unsigned short);
11106 vector unsigned short vec_sll (vector unsigned short,
11107 vector unsigned char);
11108 vector bool short vec_sll (vector bool short, vector unsigned int);
11109 vector bool short vec_sll (vector bool short, vector unsigned short);
11110 vector bool short vec_sll (vector bool short, vector unsigned char);
11111 vector pixel vec_sll (vector pixel, vector unsigned int);
11112 vector pixel vec_sll (vector pixel, vector unsigned short);
11113 vector pixel vec_sll (vector pixel, vector unsigned char);
11114 vector signed char vec_sll (vector signed char, vector unsigned int);
11115 vector signed char vec_sll (vector signed char, vector unsigned short);
11116 vector signed char vec_sll (vector signed char, vector unsigned char);
11117 vector unsigned char vec_sll (vector unsigned char,
11118 vector unsigned int);
11119 vector unsigned char vec_sll (vector unsigned char,
11120 vector unsigned short);
11121 vector unsigned char vec_sll (vector unsigned char,
11122 vector unsigned char);
11123 vector bool char vec_sll (vector bool char, vector unsigned int);
11124 vector bool char vec_sll (vector bool char, vector unsigned short);
11125 vector bool char vec_sll (vector bool char, vector unsigned char);
11127 vector float vec_slo (vector float, vector signed char);
11128 vector float vec_slo (vector float, vector unsigned char);
11129 vector signed int vec_slo (vector signed int, vector signed char);
11130 vector signed int vec_slo (vector signed int, vector unsigned char);
11131 vector unsigned int vec_slo (vector unsigned int, vector signed char);
11132 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
11133 vector signed short vec_slo (vector signed short, vector signed char);
11134 vector signed short vec_slo (vector signed short, vector unsigned char);
11135 vector unsigned short vec_slo (vector unsigned short,
11136 vector signed char);
11137 vector unsigned short vec_slo (vector unsigned short,
11138 vector unsigned char);
11139 vector pixel vec_slo (vector pixel, vector signed char);
11140 vector pixel vec_slo (vector pixel, vector unsigned char);
11141 vector signed char vec_slo (vector signed char, vector signed char);
11142 vector signed char vec_slo (vector signed char, vector unsigned char);
11143 vector unsigned char vec_slo (vector unsigned char, vector signed char);
11144 vector unsigned char vec_slo (vector unsigned char,
11145 vector unsigned char);
11147 vector signed char vec_splat (vector signed char, const int);
11148 vector unsigned char vec_splat (vector unsigned char, const int);
11149 vector bool char vec_splat (vector bool char, const int);
11150 vector signed short vec_splat (vector signed short, const int);
11151 vector unsigned short vec_splat (vector unsigned short, const int);
11152 vector bool short vec_splat (vector bool short, const int);
11153 vector pixel vec_splat (vector pixel, const int);
11154 vector float vec_splat (vector float, const int);
11155 vector signed int vec_splat (vector signed int, const int);
11156 vector unsigned int vec_splat (vector unsigned int, const int);
11157 vector bool int vec_splat (vector bool int, const int);
11159 vector float vec_vspltw (vector float, const int);
11160 vector signed int vec_vspltw (vector signed int, const int);
11161 vector unsigned int vec_vspltw (vector unsigned int, const int);
11162 vector bool int vec_vspltw (vector bool int, const int);
11164 vector bool short vec_vsplth (vector bool short, const int);
11165 vector signed short vec_vsplth (vector signed short, const int);
11166 vector unsigned short vec_vsplth (vector unsigned short, const int);
11167 vector pixel vec_vsplth (vector pixel, const int);
11169 vector signed char vec_vspltb (vector signed char, const int);
11170 vector unsigned char vec_vspltb (vector unsigned char, const int);
11171 vector bool char vec_vspltb (vector bool char, const int);
11173 vector signed char vec_splat_s8 (const int);
11175 vector signed short vec_splat_s16 (const int);
11177 vector signed int vec_splat_s32 (const int);
11179 vector unsigned char vec_splat_u8 (const int);
11181 vector unsigned short vec_splat_u16 (const int);
11183 vector unsigned int vec_splat_u32 (const int);
11185 vector signed char vec_sr (vector signed char, vector unsigned char);
11186 vector unsigned char vec_sr (vector unsigned char,
11187 vector unsigned char);
11188 vector signed short vec_sr (vector signed short,
11189 vector unsigned short);
11190 vector unsigned short vec_sr (vector unsigned short,
11191 vector unsigned short);
11192 vector signed int vec_sr (vector signed int, vector unsigned int);
11193 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
11195 vector signed int vec_vsrw (vector signed int, vector unsigned int);
11196 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
11198 vector signed short vec_vsrh (vector signed short,
11199 vector unsigned short);
11200 vector unsigned short vec_vsrh (vector unsigned short,
11201 vector unsigned short);
11203 vector signed char vec_vsrb (vector signed char, vector unsigned char);
11204 vector unsigned char vec_vsrb (vector unsigned char,
11205 vector unsigned char);
11207 vector signed char vec_sra (vector signed char, vector unsigned char);
11208 vector unsigned char vec_sra (vector unsigned char,
11209 vector unsigned char);
11210 vector signed short vec_sra (vector signed short,
11211 vector unsigned short);
11212 vector unsigned short vec_sra (vector unsigned short,
11213 vector unsigned short);
11214 vector signed int vec_sra (vector signed int, vector unsigned int);
11215 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
11217 vector signed int vec_vsraw (vector signed int, vector unsigned int);
11218 vector unsigned int vec_vsraw (vector unsigned int,
11219 vector unsigned int);
11221 vector signed short vec_vsrah (vector signed short,
11222 vector unsigned short);
11223 vector unsigned short vec_vsrah (vector unsigned short,
11224 vector unsigned short);
11226 vector signed char vec_vsrab (vector signed char, vector unsigned char);
11227 vector unsigned char vec_vsrab (vector unsigned char,
11228 vector unsigned char);
11230 vector signed int vec_srl (vector signed int, vector unsigned int);
11231 vector signed int vec_srl (vector signed int, vector unsigned short);
11232 vector signed int vec_srl (vector signed int, vector unsigned char);
11233 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
11234 vector unsigned int vec_srl (vector unsigned int,
11235 vector unsigned short);
11236 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
11237 vector bool int vec_srl (vector bool int, vector unsigned int);
11238 vector bool int vec_srl (vector bool int, vector unsigned short);
11239 vector bool int vec_srl (vector bool int, vector unsigned char);
11240 vector signed short vec_srl (vector signed short, vector unsigned int);
11241 vector signed short vec_srl (vector signed short,
11242 vector unsigned short);
11243 vector signed short vec_srl (vector signed short, vector unsigned char);
11244 vector unsigned short vec_srl (vector unsigned short,
11245 vector unsigned int);
11246 vector unsigned short vec_srl (vector unsigned short,
11247 vector unsigned short);
11248 vector unsigned short vec_srl (vector unsigned short,
11249 vector unsigned char);
11250 vector bool short vec_srl (vector bool short, vector unsigned int);
11251 vector bool short vec_srl (vector bool short, vector unsigned short);
11252 vector bool short vec_srl (vector bool short, vector unsigned char);
11253 vector pixel vec_srl (vector pixel, vector unsigned int);
11254 vector pixel vec_srl (vector pixel, vector unsigned short);
11255 vector pixel vec_srl (vector pixel, vector unsigned char);
11256 vector signed char vec_srl (vector signed char, vector unsigned int);
11257 vector signed char vec_srl (vector signed char, vector unsigned short);
11258 vector signed char vec_srl (vector signed char, vector unsigned char);
11259 vector unsigned char vec_srl (vector unsigned char,
11260 vector unsigned int);
11261 vector unsigned char vec_srl (vector unsigned char,
11262 vector unsigned short);
11263 vector unsigned char vec_srl (vector unsigned char,
11264 vector unsigned char);
11265 vector bool char vec_srl (vector bool char, vector unsigned int);
11266 vector bool char vec_srl (vector bool char, vector unsigned short);
11267 vector bool char vec_srl (vector bool char, vector unsigned char);
11269 vector float vec_sro (vector float, vector signed char);
11270 vector float vec_sro (vector float, vector unsigned char);
11271 vector signed int vec_sro (vector signed int, vector signed char);
11272 vector signed int vec_sro (vector signed int, vector unsigned char);
11273 vector unsigned int vec_sro (vector unsigned int, vector signed char);
11274 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
11275 vector signed short vec_sro (vector signed short, vector signed char);
11276 vector signed short vec_sro (vector signed short, vector unsigned char);
11277 vector unsigned short vec_sro (vector unsigned short,
11278 vector signed char);
11279 vector unsigned short vec_sro (vector unsigned short,
11280 vector unsigned char);
11281 vector pixel vec_sro (vector pixel, vector signed char);
11282 vector pixel vec_sro (vector pixel, vector unsigned char);
11283 vector signed char vec_sro (vector signed char, vector signed char);
11284 vector signed char vec_sro (vector signed char, vector unsigned char);
11285 vector unsigned char vec_sro (vector unsigned char, vector signed char);
11286 vector unsigned char vec_sro (vector unsigned char,
11287 vector unsigned char);
11289 void vec_st (vector float, int, vector float *);
11290 void vec_st (vector float, int, float *);
11291 void vec_st (vector signed int, int, vector signed int *);
11292 void vec_st (vector signed int, int, int *);
11293 void vec_st (vector unsigned int, int, vector unsigned int *);
11294 void vec_st (vector unsigned int, int, unsigned int *);
11295 void vec_st (vector bool int, int, vector bool int *);
11296 void vec_st (vector bool int, int, unsigned int *);
11297 void vec_st (vector bool int, int, int *);
11298 void vec_st (vector signed short, int, vector signed short *);
11299 void vec_st (vector signed short, int, short *);
11300 void vec_st (vector unsigned short, int, vector unsigned short *);
11301 void vec_st (vector unsigned short, int, unsigned short *);
11302 void vec_st (vector bool short, int, vector bool short *);
11303 void vec_st (vector bool short, int, unsigned short *);
11304 void vec_st (vector pixel, int, vector pixel *);
11305 void vec_st (vector pixel, int, unsigned short *);
11306 void vec_st (vector pixel, int, short *);
11307 void vec_st (vector bool short, int, short *);
11308 void vec_st (vector signed char, int, vector signed char *);
11309 void vec_st (vector signed char, int, signed char *);
11310 void vec_st (vector unsigned char, int, vector unsigned char *);
11311 void vec_st (vector unsigned char, int, unsigned char *);
11312 void vec_st (vector bool char, int, vector bool char *);
11313 void vec_st (vector bool char, int, unsigned char *);
11314 void vec_st (vector bool char, int, signed char *);
11316 void vec_ste (vector signed char, int, signed char *);
11317 void vec_ste (vector unsigned char, int, unsigned char *);
11318 void vec_ste (vector bool char, int, signed char *);
11319 void vec_ste (vector bool char, int, unsigned char *);
11320 void vec_ste (vector signed short, int, short *);
11321 void vec_ste (vector unsigned short, int, unsigned short *);
11322 void vec_ste (vector bool short, int, short *);
11323 void vec_ste (vector bool short, int, unsigned short *);
11324 void vec_ste (vector pixel, int, short *);
11325 void vec_ste (vector pixel, int, unsigned short *);
11326 void vec_ste (vector float, int, float *);
11327 void vec_ste (vector signed int, int, int *);
11328 void vec_ste (vector unsigned int, int, unsigned int *);
11329 void vec_ste (vector bool int, int, int *);
11330 void vec_ste (vector bool int, int, unsigned int *);
11332 void vec_stvewx (vector float, int, float *);
11333 void vec_stvewx (vector signed int, int, int *);
11334 void vec_stvewx (vector unsigned int, int, unsigned int *);
11335 void vec_stvewx (vector bool int, int, int *);
11336 void vec_stvewx (vector bool int, int, unsigned int *);
11338 void vec_stvehx (vector signed short, int, short *);
11339 void vec_stvehx (vector unsigned short, int, unsigned short *);
11340 void vec_stvehx (vector bool short, int, short *);
11341 void vec_stvehx (vector bool short, int, unsigned short *);
11342 void vec_stvehx (vector pixel, int, short *);
11343 void vec_stvehx (vector pixel, int, unsigned short *);
11345 void vec_stvebx (vector signed char, int, signed char *);
11346 void vec_stvebx (vector unsigned char, int, unsigned char *);
11347 void vec_stvebx (vector bool char, int, signed char *);
11348 void vec_stvebx (vector bool char, int, unsigned char *);
11350 void vec_stl (vector float, int, vector float *);
11351 void vec_stl (vector float, int, float *);
11352 void vec_stl (vector signed int, int, vector signed int *);
11353 void vec_stl (vector signed int, int, int *);
11354 void vec_stl (vector unsigned int, int, vector unsigned int *);
11355 void vec_stl (vector unsigned int, int, unsigned int *);
11356 void vec_stl (vector bool int, int, vector bool int *);
11357 void vec_stl (vector bool int, int, unsigned int *);
11358 void vec_stl (vector bool int, int, int *);
11359 void vec_stl (vector signed short, int, vector signed short *);
11360 void vec_stl (vector signed short, int, short *);
11361 void vec_stl (vector unsigned short, int, vector unsigned short *);
11362 void vec_stl (vector unsigned short, int, unsigned short *);
11363 void vec_stl (vector bool short, int, vector bool short *);
11364 void vec_stl (vector bool short, int, unsigned short *);
11365 void vec_stl (vector bool short, int, short *);
11366 void vec_stl (vector pixel, int, vector pixel *);
11367 void vec_stl (vector pixel, int, unsigned short *);
11368 void vec_stl (vector pixel, int, short *);
11369 void vec_stl (vector signed char, int, vector signed char *);
11370 void vec_stl (vector signed char, int, signed char *);
11371 void vec_stl (vector unsigned char, int, vector unsigned char *);
11372 void vec_stl (vector unsigned char, int, unsigned char *);
11373 void vec_stl (vector bool char, int, vector bool char *);
11374 void vec_stl (vector bool char, int, unsigned char *);
11375 void vec_stl (vector bool char, int, signed char *);
11377 vector signed char vec_sub (vector bool char, vector signed char);
11378 vector signed char vec_sub (vector signed char, vector bool char);
11379 vector signed char vec_sub (vector signed char, vector signed char);
11380 vector unsigned char vec_sub (vector bool char, vector unsigned char);
11381 vector unsigned char vec_sub (vector unsigned char, vector bool char);
11382 vector unsigned char vec_sub (vector unsigned char,
11383 vector unsigned char);
11384 vector signed short vec_sub (vector bool short, vector signed short);
11385 vector signed short vec_sub (vector signed short, vector bool short);
11386 vector signed short vec_sub (vector signed short, vector signed short);
11387 vector unsigned short vec_sub (vector bool short,
11388 vector unsigned short);
11389 vector unsigned short vec_sub (vector unsigned short,
11390 vector bool short);
11391 vector unsigned short vec_sub (vector unsigned short,
11392 vector unsigned short);
11393 vector signed int vec_sub (vector bool int, vector signed int);
11394 vector signed int vec_sub (vector signed int, vector bool int);
11395 vector signed int vec_sub (vector signed int, vector signed int);
11396 vector unsigned int vec_sub (vector bool int, vector unsigned int);
11397 vector unsigned int vec_sub (vector unsigned int, vector bool int);
11398 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
11399 vector float vec_sub (vector float, vector float);
11401 vector float vec_vsubfp (vector float, vector float);
11403 vector signed int vec_vsubuwm (vector bool int, vector signed int);
11404 vector signed int vec_vsubuwm (vector signed int, vector bool int);
11405 vector signed int vec_vsubuwm (vector signed int, vector signed int);
11406 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
11407 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
11408 vector unsigned int vec_vsubuwm (vector unsigned int,
11409 vector unsigned int);
11411 vector signed short vec_vsubuhm (vector bool short,
11412 vector signed short);
11413 vector signed short vec_vsubuhm (vector signed short,
11414 vector bool short);
11415 vector signed short vec_vsubuhm (vector signed short,
11416 vector signed short);
11417 vector unsigned short vec_vsubuhm (vector bool short,
11418 vector unsigned short);
11419 vector unsigned short vec_vsubuhm (vector unsigned short,
11420 vector bool short);
11421 vector unsigned short vec_vsubuhm (vector unsigned short,
11422 vector unsigned short);
11424 vector signed char vec_vsububm (vector bool char, vector signed char);
11425 vector signed char vec_vsububm (vector signed char, vector bool char);
11426 vector signed char vec_vsububm (vector signed char, vector signed char);
11427 vector unsigned char vec_vsububm (vector bool char,
11428 vector unsigned char);
11429 vector unsigned char vec_vsububm (vector unsigned char,
11431 vector unsigned char vec_vsububm (vector unsigned char,
11432 vector unsigned char);
11434 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
11436 vector unsigned char vec_subs (vector bool char, vector unsigned char);
11437 vector unsigned char vec_subs (vector unsigned char, vector bool char);
11438 vector unsigned char vec_subs (vector unsigned char,
11439 vector unsigned char);
11440 vector signed char vec_subs (vector bool char, vector signed char);
11441 vector signed char vec_subs (vector signed char, vector bool char);
11442 vector signed char vec_subs (vector signed char, vector signed char);
11443 vector unsigned short vec_subs (vector bool short,
11444 vector unsigned short);
11445 vector unsigned short vec_subs (vector unsigned short,
11446 vector bool short);
11447 vector unsigned short vec_subs (vector unsigned short,
11448 vector unsigned short);
11449 vector signed short vec_subs (vector bool short, vector signed short);
11450 vector signed short vec_subs (vector signed short, vector bool short);
11451 vector signed short vec_subs (vector signed short, vector signed short);
11452 vector unsigned int vec_subs (vector bool int, vector unsigned int);
11453 vector unsigned int vec_subs (vector unsigned int, vector bool int);
11454 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
11455 vector signed int vec_subs (vector bool int, vector signed int);
11456 vector signed int vec_subs (vector signed int, vector bool int);
11457 vector signed int vec_subs (vector signed int, vector signed int);
11459 vector signed int vec_vsubsws (vector bool int, vector signed int);
11460 vector signed int vec_vsubsws (vector signed int, vector bool int);
11461 vector signed int vec_vsubsws (vector signed int, vector signed int);
11463 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
11464 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
11465 vector unsigned int vec_vsubuws (vector unsigned int,
11466 vector unsigned int);
11468 vector signed short vec_vsubshs (vector bool short,
11469 vector signed short);
11470 vector signed short vec_vsubshs (vector signed short,
11471 vector bool short);
11472 vector signed short vec_vsubshs (vector signed short,
11473 vector signed short);
11475 vector unsigned short vec_vsubuhs (vector bool short,
11476 vector unsigned short);
11477 vector unsigned short vec_vsubuhs (vector unsigned short,
11478 vector bool short);
11479 vector unsigned short vec_vsubuhs (vector unsigned short,
11480 vector unsigned short);
11482 vector signed char vec_vsubsbs (vector bool char, vector signed char);
11483 vector signed char vec_vsubsbs (vector signed char, vector bool char);
11484 vector signed char vec_vsubsbs (vector signed char, vector signed char);
11486 vector unsigned char vec_vsububs (vector bool char,
11487 vector unsigned char);
11488 vector unsigned char vec_vsububs (vector unsigned char,
11490 vector unsigned char vec_vsububs (vector unsigned char,
11491 vector unsigned char);
11493 vector unsigned int vec_sum4s (vector unsigned char,
11494 vector unsigned int);
11495 vector signed int vec_sum4s (vector signed char, vector signed int);
11496 vector signed int vec_sum4s (vector signed short, vector signed int);
11498 vector signed int vec_vsum4shs (vector signed short, vector signed int);
11500 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
11502 vector unsigned int vec_vsum4ubs (vector unsigned char,
11503 vector unsigned int);
11505 vector signed int vec_sum2s (vector signed int, vector signed int);
11507 vector signed int vec_sums (vector signed int, vector signed int);
11509 vector float vec_trunc (vector float);
11511 vector signed short vec_unpackh (vector signed char);
11512 vector bool short vec_unpackh (vector bool char);
11513 vector signed int vec_unpackh (vector signed short);
11514 vector bool int vec_unpackh (vector bool short);
11515 vector unsigned int vec_unpackh (vector pixel);
11517 vector bool int vec_vupkhsh (vector bool short);
11518 vector signed int vec_vupkhsh (vector signed short);
11520 vector unsigned int vec_vupkhpx (vector pixel);
11522 vector bool short vec_vupkhsb (vector bool char);
11523 vector signed short vec_vupkhsb (vector signed char);
11525 vector signed short vec_unpackl (vector signed char);
11526 vector bool short vec_unpackl (vector bool char);
11527 vector unsigned int vec_unpackl (vector pixel);
11528 vector signed int vec_unpackl (vector signed short);
11529 vector bool int vec_unpackl (vector bool short);
11531 vector unsigned int vec_vupklpx (vector pixel);
11533 vector bool int vec_vupklsh (vector bool short);
11534 vector signed int vec_vupklsh (vector signed short);
11536 vector bool short vec_vupklsb (vector bool char);
11537 vector signed short vec_vupklsb (vector signed char);
11539 vector float vec_xor (vector float, vector float);
11540 vector float vec_xor (vector float, vector bool int);
11541 vector float vec_xor (vector bool int, vector float);
11542 vector bool int vec_xor (vector bool int, vector bool int);
11543 vector signed int vec_xor (vector bool int, vector signed int);
11544 vector signed int vec_xor (vector signed int, vector bool int);
11545 vector signed int vec_xor (vector signed int, vector signed int);
11546 vector unsigned int vec_xor (vector bool int, vector unsigned int);
11547 vector unsigned int vec_xor (vector unsigned int, vector bool int);
11548 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
11549 vector bool short vec_xor (vector bool short, vector bool short);
11550 vector signed short vec_xor (vector bool short, vector signed short);
11551 vector signed short vec_xor (vector signed short, vector bool short);
11552 vector signed short vec_xor (vector signed short, vector signed short);
11553 vector unsigned short vec_xor (vector bool short,
11554 vector unsigned short);
11555 vector unsigned short vec_xor (vector unsigned short,
11556 vector bool short);
11557 vector unsigned short vec_xor (vector unsigned short,
11558 vector unsigned short);
11559 vector signed char vec_xor (vector bool char, vector signed char);
11560 vector bool char vec_xor (vector bool char, vector bool char);
11561 vector signed char vec_xor (vector signed char, vector bool char);
11562 vector signed char vec_xor (vector signed char, vector signed char);
11563 vector unsigned char vec_xor (vector bool char, vector unsigned char);
11564 vector unsigned char vec_xor (vector unsigned char, vector bool char);
11565 vector unsigned char vec_xor (vector unsigned char,
11566 vector unsigned char);
11568 int vec_all_eq (vector signed char, vector bool char);
11569 int vec_all_eq (vector signed char, vector signed char);
11570 int vec_all_eq (vector unsigned char, vector bool char);
11571 int vec_all_eq (vector unsigned char, vector unsigned char);
11572 int vec_all_eq (vector bool char, vector bool char);
11573 int vec_all_eq (vector bool char, vector unsigned char);
11574 int vec_all_eq (vector bool char, vector signed char);
11575 int vec_all_eq (vector signed short, vector bool short);
11576 int vec_all_eq (vector signed short, vector signed short);
11577 int vec_all_eq (vector unsigned short, vector bool short);
11578 int vec_all_eq (vector unsigned short, vector unsigned short);
11579 int vec_all_eq (vector bool short, vector bool short);
11580 int vec_all_eq (vector bool short, vector unsigned short);
11581 int vec_all_eq (vector bool short, vector signed short);
11582 int vec_all_eq (vector pixel, vector pixel);
11583 int vec_all_eq (vector signed int, vector bool int);
11584 int vec_all_eq (vector signed int, vector signed int);
11585 int vec_all_eq (vector unsigned int, vector bool int);
11586 int vec_all_eq (vector unsigned int, vector unsigned int);
11587 int vec_all_eq (vector bool int, vector bool int);
11588 int vec_all_eq (vector bool int, vector unsigned int);
11589 int vec_all_eq (vector bool int, vector signed int);
11590 int vec_all_eq (vector float, vector float);
11592 int vec_all_ge (vector bool char, vector unsigned char);
11593 int vec_all_ge (vector unsigned char, vector bool char);
11594 int vec_all_ge (vector unsigned char, vector unsigned char);
11595 int vec_all_ge (vector bool char, vector signed char);
11596 int vec_all_ge (vector signed char, vector bool char);
11597 int vec_all_ge (vector signed char, vector signed char);
11598 int vec_all_ge (vector bool short, vector unsigned short);
11599 int vec_all_ge (vector unsigned short, vector bool short);
11600 int vec_all_ge (vector unsigned short, vector unsigned short);
11601 int vec_all_ge (vector signed short, vector signed short);
11602 int vec_all_ge (vector bool short, vector signed short);
11603 int vec_all_ge (vector signed short, vector bool short);
11604 int vec_all_ge (vector bool int, vector unsigned int);
11605 int vec_all_ge (vector unsigned int, vector bool int);
11606 int vec_all_ge (vector unsigned int, vector unsigned int);
11607 int vec_all_ge (vector bool int, vector signed int);
11608 int vec_all_ge (vector signed int, vector bool int);
11609 int vec_all_ge (vector signed int, vector signed int);
11610 int vec_all_ge (vector float, vector float);
11612 int vec_all_gt (vector bool char, vector unsigned char);
11613 int vec_all_gt (vector unsigned char, vector bool char);
11614 int vec_all_gt (vector unsigned char, vector unsigned char);
11615 int vec_all_gt (vector bool char, vector signed char);
11616 int vec_all_gt (vector signed char, vector bool char);
11617 int vec_all_gt (vector signed char, vector signed char);
11618 int vec_all_gt (vector bool short, vector unsigned short);
11619 int vec_all_gt (vector unsigned short, vector bool short);
11620 int vec_all_gt (vector unsigned short, vector unsigned short);
11621 int vec_all_gt (vector bool short, vector signed short);
11622 int vec_all_gt (vector signed short, vector bool short);
11623 int vec_all_gt (vector signed short, vector signed short);
11624 int vec_all_gt (vector bool int, vector unsigned int);
11625 int vec_all_gt (vector unsigned int, vector bool int);
11626 int vec_all_gt (vector unsigned int, vector unsigned int);
11627 int vec_all_gt (vector bool int, vector signed int);
11628 int vec_all_gt (vector signed int, vector bool int);
11629 int vec_all_gt (vector signed int, vector signed int);
11630 int vec_all_gt (vector float, vector float);
11632 int vec_all_in (vector float, vector float);
11634 int vec_all_le (vector bool char, vector unsigned char);
11635 int vec_all_le (vector unsigned char, vector bool char);
11636 int vec_all_le (vector unsigned char, vector unsigned char);
11637 int vec_all_le (vector bool char, vector signed char);
11638 int vec_all_le (vector signed char, vector bool char);
11639 int vec_all_le (vector signed char, vector signed char);
11640 int vec_all_le (vector bool short, vector unsigned short);
11641 int vec_all_le (vector unsigned short, vector bool short);
11642 int vec_all_le (vector unsigned short, vector unsigned short);
11643 int vec_all_le (vector bool short, vector signed short);
11644 int vec_all_le (vector signed short, vector bool short);
11645 int vec_all_le (vector signed short, vector signed short);
11646 int vec_all_le (vector bool int, vector unsigned int);
11647 int vec_all_le (vector unsigned int, vector bool int);
11648 int vec_all_le (vector unsigned int, vector unsigned int);
11649 int vec_all_le (vector bool int, vector signed int);
11650 int vec_all_le (vector signed int, vector bool int);
11651 int vec_all_le (vector signed int, vector signed int);
11652 int vec_all_le (vector float, vector float);
11654 int vec_all_lt (vector bool char, vector unsigned char);
11655 int vec_all_lt (vector unsigned char, vector bool char);
11656 int vec_all_lt (vector unsigned char, vector unsigned char);
11657 int vec_all_lt (vector bool char, vector signed char);
11658 int vec_all_lt (vector signed char, vector bool char);
11659 int vec_all_lt (vector signed char, vector signed char);
11660 int vec_all_lt (vector bool short, vector unsigned short);
11661 int vec_all_lt (vector unsigned short, vector bool short);
11662 int vec_all_lt (vector unsigned short, vector unsigned short);
11663 int vec_all_lt (vector bool short, vector signed short);
11664 int vec_all_lt (vector signed short, vector bool short);
11665 int vec_all_lt (vector signed short, vector signed short);
11666 int vec_all_lt (vector bool int, vector unsigned int);
11667 int vec_all_lt (vector unsigned int, vector bool int);
11668 int vec_all_lt (vector unsigned int, vector unsigned int);
11669 int vec_all_lt (vector bool int, vector signed int);
11670 int vec_all_lt (vector signed int, vector bool int);
11671 int vec_all_lt (vector signed int, vector signed int);
11672 int vec_all_lt (vector float, vector float);
11674 int vec_all_nan (vector float);
11676 int vec_all_ne (vector signed char, vector bool char);
11677 int vec_all_ne (vector signed char, vector signed char);
11678 int vec_all_ne (vector unsigned char, vector bool char);
11679 int vec_all_ne (vector unsigned char, vector unsigned char);
11680 int vec_all_ne (vector bool char, vector bool char);
11681 int vec_all_ne (vector bool char, vector unsigned char);
11682 int vec_all_ne (vector bool char, vector signed char);
11683 int vec_all_ne (vector signed short, vector bool short);
11684 int vec_all_ne (vector signed short, vector signed short);
11685 int vec_all_ne (vector unsigned short, vector bool short);
11686 int vec_all_ne (vector unsigned short, vector unsigned short);
11687 int vec_all_ne (vector bool short, vector bool short);
11688 int vec_all_ne (vector bool short, vector unsigned short);
11689 int vec_all_ne (vector bool short, vector signed short);
11690 int vec_all_ne (vector pixel, vector pixel);
11691 int vec_all_ne (vector signed int, vector bool int);
11692 int vec_all_ne (vector signed int, vector signed int);
11693 int vec_all_ne (vector unsigned int, vector bool int);
11694 int vec_all_ne (vector unsigned int, vector unsigned int);
11695 int vec_all_ne (vector bool int, vector bool int);
11696 int vec_all_ne (vector bool int, vector unsigned int);
11697 int vec_all_ne (vector bool int, vector signed int);
11698 int vec_all_ne (vector float, vector float);
11700 int vec_all_nge (vector float, vector float);
11702 int vec_all_ngt (vector float, vector float);
11704 int vec_all_nle (vector float, vector float);
11706 int vec_all_nlt (vector float, vector float);
11708 int vec_all_numeric (vector float);
11710 int vec_any_eq (vector signed char, vector bool char);
11711 int vec_any_eq (vector signed char, vector signed char);
11712 int vec_any_eq (vector unsigned char, vector bool char);
11713 int vec_any_eq (vector unsigned char, vector unsigned char);
11714 int vec_any_eq (vector bool char, vector bool char);
11715 int vec_any_eq (vector bool char, vector unsigned char);
11716 int vec_any_eq (vector bool char, vector signed char);
11717 int vec_any_eq (vector signed short, vector bool short);
11718 int vec_any_eq (vector signed short, vector signed short);
11719 int vec_any_eq (vector unsigned short, vector bool short);
11720 int vec_any_eq (vector unsigned short, vector unsigned short);
11721 int vec_any_eq (vector bool short, vector bool short);
11722 int vec_any_eq (vector bool short, vector unsigned short);
11723 int vec_any_eq (vector bool short, vector signed short);
11724 int vec_any_eq (vector pixel, vector pixel);
11725 int vec_any_eq (vector signed int, vector bool int);
11726 int vec_any_eq (vector signed int, vector signed int);
11727 int vec_any_eq (vector unsigned int, vector bool int);
11728 int vec_any_eq (vector unsigned int, vector unsigned int);
11729 int vec_any_eq (vector bool int, vector bool int);
11730 int vec_any_eq (vector bool int, vector unsigned int);
11731 int vec_any_eq (vector bool int, vector signed int);
11732 int vec_any_eq (vector float, vector float);
11734 int vec_any_ge (vector signed char, vector bool char);
11735 int vec_any_ge (vector unsigned char, vector bool char);
11736 int vec_any_ge (vector unsigned char, vector unsigned char);
11737 int vec_any_ge (vector signed char, vector signed char);
11738 int vec_any_ge (vector bool char, vector unsigned char);
11739 int vec_any_ge (vector bool char, vector signed char);
11740 int vec_any_ge (vector unsigned short, vector bool short);
11741 int vec_any_ge (vector unsigned short, vector unsigned short);
11742 int vec_any_ge (vector signed short, vector signed short);
11743 int vec_any_ge (vector signed short, vector bool short);
11744 int vec_any_ge (vector bool short, vector unsigned short);
11745 int vec_any_ge (vector bool short, vector signed short);
11746 int vec_any_ge (vector signed int, vector bool int);
11747 int vec_any_ge (vector unsigned int, vector bool int);
11748 int vec_any_ge (vector unsigned int, vector unsigned int);
11749 int vec_any_ge (vector signed int, vector signed int);
11750 int vec_any_ge (vector bool int, vector unsigned int);
11751 int vec_any_ge (vector bool int, vector signed int);
11752 int vec_any_ge (vector float, vector float);
11754 int vec_any_gt (vector bool char, vector unsigned char);
11755 int vec_any_gt (vector unsigned char, vector bool char);
11756 int vec_any_gt (vector unsigned char, vector unsigned char);
11757 int vec_any_gt (vector bool char, vector signed char);
11758 int vec_any_gt (vector signed char, vector bool char);
11759 int vec_any_gt (vector signed char, vector signed char);
11760 int vec_any_gt (vector bool short, vector unsigned short);
11761 int vec_any_gt (vector unsigned short, vector bool short);
11762 int vec_any_gt (vector unsigned short, vector unsigned short);
11763 int vec_any_gt (vector bool short, vector signed short);
11764 int vec_any_gt (vector signed short, vector bool short);
11765 int vec_any_gt (vector signed short, vector signed short);
11766 int vec_any_gt (vector bool int, vector unsigned int);
11767 int vec_any_gt (vector unsigned int, vector bool int);
11768 int vec_any_gt (vector unsigned int, vector unsigned int);
11769 int vec_any_gt (vector bool int, vector signed int);
11770 int vec_any_gt (vector signed int, vector bool int);
11771 int vec_any_gt (vector signed int, vector signed int);
11772 int vec_any_gt (vector float, vector float);
11774 int vec_any_le (vector bool char, vector unsigned char);
11775 int vec_any_le (vector unsigned char, vector bool char);
11776 int vec_any_le (vector unsigned char, vector unsigned char);
11777 int vec_any_le (vector bool char, vector signed char);
11778 int vec_any_le (vector signed char, vector bool char);
11779 int vec_any_le (vector signed char, vector signed char);
11780 int vec_any_le (vector bool short, vector unsigned short);
11781 int vec_any_le (vector unsigned short, vector bool short);
11782 int vec_any_le (vector unsigned short, vector unsigned short);
11783 int vec_any_le (vector bool short, vector signed short);
11784 int vec_any_le (vector signed short, vector bool short);
11785 int vec_any_le (vector signed short, vector signed short);
11786 int vec_any_le (vector bool int, vector unsigned int);
11787 int vec_any_le (vector unsigned int, vector bool int);
11788 int vec_any_le (vector unsigned int, vector unsigned int);
11789 int vec_any_le (vector bool int, vector signed int);
11790 int vec_any_le (vector signed int, vector bool int);
11791 int vec_any_le (vector signed int, vector signed int);
11792 int vec_any_le (vector float, vector float);
11794 int vec_any_lt (vector bool char, vector unsigned char);
11795 int vec_any_lt (vector unsigned char, vector bool char);
11796 int vec_any_lt (vector unsigned char, vector unsigned char);
11797 int vec_any_lt (vector bool char, vector signed char);
11798 int vec_any_lt (vector signed char, vector bool char);
11799 int vec_any_lt (vector signed char, vector signed char);
11800 int vec_any_lt (vector bool short, vector unsigned short);
11801 int vec_any_lt (vector unsigned short, vector bool short);
11802 int vec_any_lt (vector unsigned short, vector unsigned short);
11803 int vec_any_lt (vector bool short, vector signed short);
11804 int vec_any_lt (vector signed short, vector bool short);
11805 int vec_any_lt (vector signed short, vector signed short);
11806 int vec_any_lt (vector bool int, vector unsigned int);
11807 int vec_any_lt (vector unsigned int, vector bool int);
11808 int vec_any_lt (vector unsigned int, vector unsigned int);
11809 int vec_any_lt (vector bool int, vector signed int);
11810 int vec_any_lt (vector signed int, vector bool int);
11811 int vec_any_lt (vector signed int, vector signed int);
11812 int vec_any_lt (vector float, vector float);
11814 int vec_any_nan (vector float);
11816 int vec_any_ne (vector signed char, vector bool char);
11817 int vec_any_ne (vector signed char, vector signed char);
11818 int vec_any_ne (vector unsigned char, vector bool char);
11819 int vec_any_ne (vector unsigned char, vector unsigned char);
11820 int vec_any_ne (vector bool char, vector bool char);
11821 int vec_any_ne (vector bool char, vector unsigned char);
11822 int vec_any_ne (vector bool char, vector signed char);
11823 int vec_any_ne (vector signed short, vector bool short);
11824 int vec_any_ne (vector signed short, vector signed short);
11825 int vec_any_ne (vector unsigned short, vector bool short);
11826 int vec_any_ne (vector unsigned short, vector unsigned short);
11827 int vec_any_ne (vector bool short, vector bool short);
11828 int vec_any_ne (vector bool short, vector unsigned short);
11829 int vec_any_ne (vector bool short, vector signed short);
11830 int vec_any_ne (vector pixel, vector pixel);
11831 int vec_any_ne (vector signed int, vector bool int);
11832 int vec_any_ne (vector signed int, vector signed int);
11833 int vec_any_ne (vector unsigned int, vector bool int);
11834 int vec_any_ne (vector unsigned int, vector unsigned int);
11835 int vec_any_ne (vector bool int, vector bool int);
11836 int vec_any_ne (vector bool int, vector unsigned int);
11837 int vec_any_ne (vector bool int, vector signed int);
11838 int vec_any_ne (vector float, vector float);
11840 int vec_any_nge (vector float, vector float);
11842 int vec_any_ngt (vector float, vector float);
11844 int vec_any_nle (vector float, vector float);
11846 int vec_any_nlt (vector float, vector float);
11848 int vec_any_numeric (vector float);
11850 int vec_any_out (vector float, vector float);
11853 If the vector/scalar (VSX) instruction set is available, the following
11854 additional functions are available:
11857 vector double vec_abs (vector double);
11858 vector double vec_add (vector double, vector double);
11859 vector double vec_and (vector double, vector double);
11860 vector double vec_and (vector double, vector bool long);
11861 vector double vec_and (vector bool long, vector double);
11862 vector double vec_andc (vector double, vector double);
11863 vector double vec_andc (vector double, vector bool long);
11864 vector double vec_andc (vector bool long, vector double);
11865 vector double vec_ceil (vector double);
11866 vector bool long vec_cmpeq (vector double, vector double);
11867 vector bool long vec_cmpge (vector double, vector double);
11868 vector bool long vec_cmpgt (vector double, vector double);
11869 vector bool long vec_cmple (vector double, vector double);
11870 vector bool long vec_cmplt (vector double, vector double);
11871 vector float vec_div (vector float, vector float);
11872 vector double vec_div (vector double, vector double);
11873 vector double vec_floor (vector double);
11874 vector double vec_madd (vector double, vector double, vector double);
11875 vector double vec_max (vector double, vector double);
11876 vector double vec_min (vector double, vector double);
11877 vector float vec_msub (vector float, vector float, vector float);
11878 vector double vec_msub (vector double, vector double, vector double);
11879 vector float vec_mul (vector float, vector float);
11880 vector double vec_mul (vector double, vector double);
11881 vector float vec_nearbyint (vector float);
11882 vector double vec_nearbyint (vector double);
11883 vector float vec_nmadd (vector float, vector float, vector float);
11884 vector double vec_nmadd (vector double, vector double, vector double);
11885 vector double vec_nmsub (vector double, vector double, vector double);
11886 vector double vec_nor (vector double, vector double);
11887 vector double vec_or (vector double, vector double);
11888 vector double vec_or (vector double, vector bool long);
11889 vector double vec_or (vector bool long, vector double);
11890 vector double vec_perm (vector double,
11892 vector unsigned char);
11893 vector float vec_rint (vector float);
11894 vector double vec_rint (vector double);
11895 vector double vec_sel (vector double, vector double, vector bool long);
11896 vector double vec_sel (vector double, vector double, vector unsigned long);
11897 vector double vec_sub (vector double, vector double);
11898 vector float vec_sqrt (vector float);
11899 vector double vec_sqrt (vector double);
11900 vector double vec_trunc (vector double);
11901 vector double vec_xor (vector double, vector double);
11902 vector double vec_xor (vector double, vector bool long);
11903 vector double vec_xor (vector bool long, vector double);
11904 int vec_all_eq (vector double, vector double);
11905 int vec_all_ge (vector double, vector double);
11906 int vec_all_gt (vector double, vector double);
11907 int vec_all_le (vector double, vector double);
11908 int vec_all_lt (vector double, vector double);
11909 int vec_all_nan (vector double);
11910 int vec_all_ne (vector double, vector double);
11911 int vec_all_nge (vector double, vector double);
11912 int vec_all_ngt (vector double, vector double);
11913 int vec_all_nle (vector double, vector double);
11914 int vec_all_nlt (vector double, vector double);
11915 int vec_all_numeric (vector double);
11916 int vec_any_eq (vector double, vector double);
11917 int vec_any_ge (vector double, vector double);
11918 int vec_any_gt (vector double, vector double);
11919 int vec_any_le (vector double, vector double);
11920 int vec_any_lt (vector double, vector double);
11921 int vec_any_nan (vector double);
11922 int vec_any_ne (vector double, vector double);
11923 int vec_any_nge (vector double, vector double);
11924 int vec_any_ngt (vector double, vector double);
11925 int vec_any_nle (vector double, vector double);
11926 int vec_any_nlt (vector double, vector double);
11927 int vec_any_numeric (vector double);
11930 GCC provides a few other builtins on Powerpc to access certain instructions:
11932 float __builtin_recipdivf (float, float);
11933 float __builtin_rsqrtf (float);
11934 double __builtin_recipdiv (double, double);
11935 long __builtin_bpermd (long, long);
11936 int __builtin_bswap16 (int);
11939 @node RX Built-in Functions
11940 @subsection RX Built-in Functions
11941 GCC supports some of the RX instructions which cannot be expressed in
11942 the C programming language via the use of built-in functions. The
11943 following functions are supported:
11945 @deftypefn {Built-in Function} void __builtin_rx_brk (void)
11946 Generates the @code{brk} machine instruction.
11949 @deftypefn {Built-in Function} void __builtin_rx_clrpsw (int)
11950 Generates the @code{clrpsw} machine instruction to clear the specified
11951 bit in the processor status word.
11954 @deftypefn {Built-in Function} void __builtin_rx_int (int)
11955 Generates the @code{int} machine instruction to generate an interrupt
11956 with the specified value.
11959 @deftypefn {Built-in Function} void __builtin_rx_machi (int, int)
11960 Generates the @code{machi} machine instruction to add the result of
11961 multiplying the top 16-bits of the two arguments into the
11965 @deftypefn {Built-in Function} void __builtin_rx_maclo (int, int)
11966 Generates the @code{maclo} machine instruction to add the result of
11967 multiplying the bottom 16-bits of the two arguments into the
11971 @deftypefn {Built-in Function} void __builtin_rx_mulhi (int, int)
11972 Generates the @code{mulhi} machine instruction to place the result of
11973 multiplying the top 16-bits of the two arguments into the
11977 @deftypefn {Built-in Function} void __builtin_rx_mullo (int, int)
11978 Generates the @code{mullo} machine instruction to place the result of
11979 multiplying the bottom 16-bits of the two arguments into the
11983 @deftypefn {Built-in Function} int __builtin_rx_mvfachi (void)
11984 Generates the @code{mvfachi} machine instruction to read the top
11985 32-bits of the accumulator.
11988 @deftypefn {Built-in Function} int __builtin_rx_mvfacmi (void)
11989 Generates the @code{mvfacmi} machine instruction to read the middle
11990 32-bits of the accumulator.
11993 @deftypefn {Built-in Function} int __builtin_rx_mvfc (int)
11994 Generates the @code{mvfc} machine instruction which reads the control
11995 register specified in its argument and returns its value.
11998 @deftypefn {Built-in Function} void __builtin_rx_mvtachi (int)
11999 Generates the @code{mvtachi} machine instruction to set the top
12000 32-bits of the accumulator.
12003 @deftypefn {Built-in Function} void __builtin_rx_mvtaclo (int)
12004 Generates the @code{mvtaclo} machine instruction to set the bottom
12005 32-bits of the accumulator.
12008 @deftypefn {Built-in Function} void __builtin_rx_mvtc (int reg, int val)
12009 Generates the @code{mvtc} machine instruction which sets control
12010 register number @code{reg} to @code{val}.
12013 @deftypefn {Built-in Function} void __builtin_rx_mvtipl (int)
12014 Generates the @code{mvtipl} machine instruction set the interrupt
12018 @deftypefn {Built-in Function} void __builtin_rx_racw (int)
12019 Generates the @code{racw} machine instruction to round the accumulator
12020 according to the specified mode.
12023 @deftypefn {Built-in Function} int __builtin_rx_revw (int)
12024 Generates the @code{revw} machine instruction which swaps the bytes in
12025 the argument so that bits 0--7 now occupy bits 8--15 and vice versa,
12026 and also bits 16--23 occupy bits 24--31 and vice versa.
12029 @deftypefn {Built-in Function} void __builtin_rx_rmpa (void)
12030 Generates the @code{rmpa} machine instruction which initiates a
12031 repeated multiply and accumulate sequence.
12034 @deftypefn {Built-in Function} void __builtin_rx_round (float)
12035 Generates the @code{round} machine instruction which returns the
12036 floating point argument rounded according to the current rounding mode
12037 set in the floating point status word register.
12040 @deftypefn {Built-in Function} int __builtin_rx_sat (int)
12041 Generates the @code{sat} machine instruction which returns the
12042 saturated value of the argument.
12045 @deftypefn {Built-in Function} void __builtin_rx_setpsw (int)
12046 Generates the @code{setpsw} machine instruction to set the specified
12047 bit in the processor status word.
12050 @deftypefn {Built-in Function} void __builtin_rx_wait (void)
12051 Generates the @code{wait} machine instruction.
12054 @node SPARC VIS Built-in Functions
12055 @subsection SPARC VIS Built-in Functions
12057 GCC supports SIMD operations on the SPARC using both the generic vector
12058 extensions (@pxref{Vector Extensions}) as well as built-in functions for
12059 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
12060 switch, the VIS extension is exposed as the following built-in functions:
12063 typedef int v2si __attribute__ ((vector_size (8)));
12064 typedef short v4hi __attribute__ ((vector_size (8)));
12065 typedef short v2hi __attribute__ ((vector_size (4)));
12066 typedef char v8qi __attribute__ ((vector_size (8)));
12067 typedef char v4qi __attribute__ ((vector_size (4)));
12069 void * __builtin_vis_alignaddr (void *, long);
12070 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
12071 v2si __builtin_vis_faligndatav2si (v2si, v2si);
12072 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
12073 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
12075 v4hi __builtin_vis_fexpand (v4qi);
12077 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
12078 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
12079 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
12080 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
12081 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
12082 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
12083 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
12085 v4qi __builtin_vis_fpack16 (v4hi);
12086 v8qi __builtin_vis_fpack32 (v2si, v2si);
12087 v2hi __builtin_vis_fpackfix (v2si);
12088 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
12090 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
12093 @node SPU Built-in Functions
12094 @subsection SPU Built-in Functions
12096 GCC provides extensions for the SPU processor as described in the
12097 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
12098 found at @uref{http://cell.scei.co.jp/} or
12099 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
12100 implementation differs in several ways.
12105 The optional extension of specifying vector constants in parentheses is
12109 A vector initializer requires no cast if the vector constant is of the
12110 same type as the variable it is initializing.
12113 If @code{signed} or @code{unsigned} is omitted, the signedness of the
12114 vector type is the default signedness of the base type. The default
12115 varies depending on the operating system, so a portable program should
12116 always specify the signedness.
12119 By default, the keyword @code{__vector} is added. The macro
12120 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
12124 GCC allows using a @code{typedef} name as the type specifier for a
12128 For C, overloaded functions are implemented with macros so the following
12132 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
12135 Since @code{spu_add} is a macro, the vector constant in the example
12136 is treated as four separate arguments. Wrap the entire argument in
12137 parentheses for this to work.
12140 The extended version of @code{__builtin_expect} is not supported.
12144 @emph{Note:} Only the interface described in the aforementioned
12145 specification is supported. Internally, GCC uses built-in functions to
12146 implement the required functionality, but these are not supported and
12147 are subject to change without notice.
12149 @node Target Format Checks
12150 @section Format Checks Specific to Particular Target Machines
12152 For some target machines, GCC supports additional options to the
12154 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
12157 * Solaris Format Checks::
12160 @node Solaris Format Checks
12161 @subsection Solaris Format Checks
12163 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
12164 check. @code{cmn_err} accepts a subset of the standard @code{printf}
12165 conversions, and the two-argument @code{%b} conversion for displaying
12166 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
12169 @section Pragmas Accepted by GCC
12173 GCC supports several types of pragmas, primarily in order to compile
12174 code originally written for other compilers. Note that in general
12175 we do not recommend the use of pragmas; @xref{Function Attributes},
12176 for further explanation.
12182 * RS/6000 and PowerPC Pragmas::
12184 * Solaris Pragmas::
12185 * Symbol-Renaming Pragmas::
12186 * Structure-Packing Pragmas::
12188 * Diagnostic Pragmas::
12189 * Visibility Pragmas::
12190 * Push/Pop Macro Pragmas::
12191 * Function Specific Option Pragmas::
12195 @subsection ARM Pragmas
12197 The ARM target defines pragmas for controlling the default addition of
12198 @code{long_call} and @code{short_call} attributes to functions.
12199 @xref{Function Attributes}, for information about the effects of these
12204 @cindex pragma, long_calls
12205 Set all subsequent functions to have the @code{long_call} attribute.
12207 @item no_long_calls
12208 @cindex pragma, no_long_calls
12209 Set all subsequent functions to have the @code{short_call} attribute.
12211 @item long_calls_off
12212 @cindex pragma, long_calls_off
12213 Do not affect the @code{long_call} or @code{short_call} attributes of
12214 subsequent functions.
12218 @subsection M32C Pragmas
12221 @item memregs @var{number}
12222 @cindex pragma, memregs
12223 Overrides the command line option @code{-memregs=} for the current
12224 file. Use with care! This pragma must be before any function in the
12225 file, and mixing different memregs values in different objects may
12226 make them incompatible. This pragma is useful when a
12227 performance-critical function uses a memreg for temporary values,
12228 as it may allow you to reduce the number of memregs used.
12233 @subsection MeP Pragmas
12237 @item custom io_volatile (on|off)
12238 @cindex pragma, custom io_volatile
12239 Overrides the command line option @code{-mio-volatile} for the current
12240 file. Note that for compatibility with future GCC releases, this
12241 option should only be used once before any @code{io} variables in each
12244 @item GCC coprocessor available @var{registers}
12245 @cindex pragma, coprocessor available
12246 Specifies which coprocessor registers are available to the register
12247 allocator. @var{registers} may be a single register, register range
12248 separated by ellipses, or comma-separated list of those. Example:
12251 #pragma GCC coprocessor available $c0...$c10, $c28
12254 @item GCC coprocessor call_saved @var{registers}
12255 @cindex pragma, coprocessor call_saved
12256 Specifies which coprocessor registers are to be saved and restored by
12257 any function using them. @var{registers} may be a single register,
12258 register range separated by ellipses, or comma-separated list of
12262 #pragma GCC coprocessor call_saved $c4...$c6, $c31
12265 @item GCC coprocessor subclass '(A|B|C|D)' = @var{registers}
12266 @cindex pragma, coprocessor subclass
12267 Creates and defines a register class. These register classes can be
12268 used by inline @code{asm} constructs. @var{registers} may be a single
12269 register, register range separated by ellipses, or comma-separated
12270 list of those. Example:
12273 #pragma GCC coprocessor subclass 'B' = $c2, $c4, $c6
12275 asm ("cpfoo %0" : "=B" (x));
12278 @item GCC disinterrupt @var{name} , @var{name} @dots{}
12279 @cindex pragma, disinterrupt
12280 For the named functions, the compiler adds code to disable interrupts
12281 for the duration of those functions. Any functions so named, which
12282 are not encountered in the source, cause a warning that the pragma was
12283 not used. Examples:
12286 #pragma disinterrupt foo
12287 #pragma disinterrupt bar, grill
12288 int foo () @{ @dots{} @}
12291 @item GCC call @var{name} , @var{name} @dots{}
12292 @cindex pragma, call
12293 For the named functions, the compiler always uses a register-indirect
12294 call model when calling the named functions. Examples:
12303 @node RS/6000 and PowerPC Pragmas
12304 @subsection RS/6000 and PowerPC Pragmas
12306 The RS/6000 and PowerPC targets define one pragma for controlling
12307 whether or not the @code{longcall} attribute is added to function
12308 declarations by default. This pragma overrides the @option{-mlongcall}
12309 option, but not the @code{longcall} and @code{shortcall} attributes.
12310 @xref{RS/6000 and PowerPC Options}, for more information about when long
12311 calls are and are not necessary.
12315 @cindex pragma, longcall
12316 Apply the @code{longcall} attribute to all subsequent function
12320 Do not apply the @code{longcall} attribute to subsequent function
12324 @c Describe h8300 pragmas here.
12325 @c Describe sh pragmas here.
12326 @c Describe v850 pragmas here.
12328 @node Darwin Pragmas
12329 @subsection Darwin Pragmas
12331 The following pragmas are available for all architectures running the
12332 Darwin operating system. These are useful for compatibility with other
12336 @item mark @var{tokens}@dots{}
12337 @cindex pragma, mark
12338 This pragma is accepted, but has no effect.
12340 @item options align=@var{alignment}
12341 @cindex pragma, options align
12342 This pragma sets the alignment of fields in structures. The values of
12343 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
12344 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
12345 properly; to restore the previous setting, use @code{reset} for the
12348 @item segment @var{tokens}@dots{}
12349 @cindex pragma, segment
12350 This pragma is accepted, but has no effect.
12352 @item unused (@var{var} [, @var{var}]@dots{})
12353 @cindex pragma, unused
12354 This pragma declares variables to be possibly unused. GCC will not
12355 produce warnings for the listed variables. The effect is similar to
12356 that of the @code{unused} attribute, except that this pragma may appear
12357 anywhere within the variables' scopes.
12360 @node Solaris Pragmas
12361 @subsection Solaris Pragmas
12363 The Solaris target supports @code{#pragma redefine_extname}
12364 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
12365 @code{#pragma} directives for compatibility with the system compiler.
12368 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
12369 @cindex pragma, align
12371 Increase the minimum alignment of each @var{variable} to @var{alignment}.
12372 This is the same as GCC's @code{aligned} attribute @pxref{Variable
12373 Attributes}). Macro expansion occurs on the arguments to this pragma
12374 when compiling C and Objective-C@. It does not currently occur when
12375 compiling C++, but this is a bug which may be fixed in a future
12378 @item fini (@var{function} [, @var{function}]...)
12379 @cindex pragma, fini
12381 This pragma causes each listed @var{function} to be called after
12382 main, or during shared module unloading, by adding a call to the
12383 @code{.fini} section.
12385 @item init (@var{function} [, @var{function}]...)
12386 @cindex pragma, init
12388 This pragma causes each listed @var{function} to be called during
12389 initialization (before @code{main}) or during shared module loading, by
12390 adding a call to the @code{.init} section.
12394 @node Symbol-Renaming Pragmas
12395 @subsection Symbol-Renaming Pragmas
12397 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
12398 supports two @code{#pragma} directives which change the name used in
12399 assembly for a given declaration. @code{#pragma_extern_prefix} is only
12400 available on platforms whose system headers need it. To get this effect
12401 on all platforms supported by GCC, use the asm labels extension (@pxref{Asm
12405 @item redefine_extname @var{oldname} @var{newname}
12406 @cindex pragma, redefine_extname
12408 This pragma gives the C function @var{oldname} the assembly symbol
12409 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
12410 will be defined if this pragma is available (currently on all platforms).
12412 @item extern_prefix @var{string}
12413 @cindex pragma, extern_prefix
12415 This pragma causes all subsequent external function and variable
12416 declarations to have @var{string} prepended to their assembly symbols.
12417 This effect may be terminated with another @code{extern_prefix} pragma
12418 whose argument is an empty string. The preprocessor macro
12419 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
12420 available (currently only on Tru64 UNIX)@.
12423 These pragmas and the asm labels extension interact in a complicated
12424 manner. Here are some corner cases you may want to be aware of.
12427 @item Both pragmas silently apply only to declarations with external
12428 linkage. Asm labels do not have this restriction.
12430 @item In C++, both pragmas silently apply only to declarations with
12431 ``C'' linkage. Again, asm labels do not have this restriction.
12433 @item If any of the three ways of changing the assembly name of a
12434 declaration is applied to a declaration whose assembly name has
12435 already been determined (either by a previous use of one of these
12436 features, or because the compiler needed the assembly name in order to
12437 generate code), and the new name is different, a warning issues and
12438 the name does not change.
12440 @item The @var{oldname} used by @code{#pragma redefine_extname} is
12441 always the C-language name.
12443 @item If @code{#pragma extern_prefix} is in effect, and a declaration
12444 occurs with an asm label attached, the prefix is silently ignored for
12447 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
12448 apply to the same declaration, whichever triggered first wins, and a
12449 warning issues if they contradict each other. (We would like to have
12450 @code{#pragma redefine_extname} always win, for consistency with asm
12451 labels, but if @code{#pragma extern_prefix} triggers first we have no
12452 way of knowing that that happened.)
12455 @node Structure-Packing Pragmas
12456 @subsection Structure-Packing Pragmas
12458 For compatibility with Microsoft Windows compilers, GCC supports a
12459 set of @code{#pragma} directives which change the maximum alignment of
12460 members of structures (other than zero-width bitfields), unions, and
12461 classes subsequently defined. The @var{n} value below always is required
12462 to be a small power of two and specifies the new alignment in bytes.
12465 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
12466 @item @code{#pragma pack()} sets the alignment to the one that was in
12467 effect when compilation started (see also command line option
12468 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
12469 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
12470 setting on an internal stack and then optionally sets the new alignment.
12471 @item @code{#pragma pack(pop)} restores the alignment setting to the one
12472 saved at the top of the internal stack (and removes that stack entry).
12473 Note that @code{#pragma pack([@var{n}])} does not influence this internal
12474 stack; thus it is possible to have @code{#pragma pack(push)} followed by
12475 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
12476 @code{#pragma pack(pop)}.
12479 Some targets, e.g.@: i386 and powerpc, support the @code{ms_struct}
12480 @code{#pragma} which lays out a structure as the documented
12481 @code{__attribute__ ((ms_struct))}.
12483 @item @code{#pragma ms_struct on} turns on the layout for structures
12485 @item @code{#pragma ms_struct off} turns off the layout for structures
12487 @item @code{#pragma ms_struct reset} goes back to the default layout.
12491 @subsection Weak Pragmas
12493 For compatibility with SVR4, GCC supports a set of @code{#pragma}
12494 directives for declaring symbols to be weak, and defining weak
12498 @item #pragma weak @var{symbol}
12499 @cindex pragma, weak
12500 This pragma declares @var{symbol} to be weak, as if the declaration
12501 had the attribute of the same name. The pragma may appear before
12502 or after the declaration of @var{symbol}, but must appear before
12503 either its first use or its definition. It is not an error for
12504 @var{symbol} to never be defined at all.
12506 @item #pragma weak @var{symbol1} = @var{symbol2}
12507 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
12508 It is an error if @var{symbol2} is not defined in the current
12512 @node Diagnostic Pragmas
12513 @subsection Diagnostic Pragmas
12515 GCC allows the user to selectively enable or disable certain types of
12516 diagnostics, and change the kind of the diagnostic. For example, a
12517 project's policy might require that all sources compile with
12518 @option{-Werror} but certain files might have exceptions allowing
12519 specific types of warnings. Or, a project might selectively enable
12520 diagnostics and treat them as errors depending on which preprocessor
12521 macros are defined.
12524 @item #pragma GCC diagnostic @var{kind} @var{option}
12525 @cindex pragma, diagnostic
12527 Modifies the disposition of a diagnostic. Note that not all
12528 diagnostics are modifiable; at the moment only warnings (normally
12529 controlled by @samp{-W@dots{}}) can be controlled, and not all of them.
12530 Use @option{-fdiagnostics-show-option} to determine which diagnostics
12531 are controllable and which option controls them.
12533 @var{kind} is @samp{error} to treat this diagnostic as an error,
12534 @samp{warning} to treat it like a warning (even if @option{-Werror} is
12535 in effect), or @samp{ignored} if the diagnostic is to be ignored.
12536 @var{option} is a double quoted string which matches the command line
12540 #pragma GCC diagnostic warning "-Wformat"
12541 #pragma GCC diagnostic error "-Wformat"
12542 #pragma GCC diagnostic ignored "-Wformat"
12545 Note that these pragmas override any command line options. Also,
12546 while it is syntactically valid to put these pragmas anywhere in your
12547 sources, the only supported location for them is before any data or
12548 functions are defined. Doing otherwise may result in unpredictable
12549 results depending on how the optimizer manages your sources. If the
12550 same option is listed multiple times, the last one specified is the
12551 one that is in effect. This pragma is not intended to be a general
12552 purpose replacement for command line options, but for implementing
12553 strict control over project policies.
12557 GCC also offers a simple mechanism for printing messages during
12561 @item #pragma message @var{string}
12562 @cindex pragma, diagnostic
12564 Prints @var{string} as a compiler message on compilation. The message
12565 is informational only, and is neither a compilation warning nor an error.
12568 #pragma message "Compiling " __FILE__ "..."
12571 @var{string} may be parenthesized, and is printed with location
12572 information. For example,
12575 #define DO_PRAGMA(x) _Pragma (#x)
12576 #define TODO(x) DO_PRAGMA(message ("TODO - " #x))
12578 TODO(Remember to fix this)
12581 prints @samp{/tmp/file.c:4: note: #pragma message:
12582 TODO - Remember to fix this}.
12586 @node Visibility Pragmas
12587 @subsection Visibility Pragmas
12590 @item #pragma GCC visibility push(@var{visibility})
12591 @itemx #pragma GCC visibility pop
12592 @cindex pragma, visibility
12594 This pragma allows the user to set the visibility for multiple
12595 declarations without having to give each a visibility attribute
12596 @xref{Function Attributes}, for more information about visibility and
12597 the attribute syntax.
12599 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
12600 declarations. Class members and template specializations are not
12601 affected; if you want to override the visibility for a particular
12602 member or instantiation, you must use an attribute.
12607 @node Push/Pop Macro Pragmas
12608 @subsection Push/Pop Macro Pragmas
12610 For compatibility with Microsoft Windows compilers, GCC supports
12611 @samp{#pragma push_macro(@var{"macro_name"})}
12612 and @samp{#pragma pop_macro(@var{"macro_name"})}.
12615 @item #pragma push_macro(@var{"macro_name"})
12616 @cindex pragma, push_macro
12617 This pragma saves the value of the macro named as @var{macro_name} to
12618 the top of the stack for this macro.
12620 @item #pragma pop_macro(@var{"macro_name"})
12621 @cindex pragma, pop_macro
12622 This pragma sets the value of the macro named as @var{macro_name} to
12623 the value on top of the stack for this macro. If the stack for
12624 @var{macro_name} is empty, the value of the macro remains unchanged.
12631 #pragma push_macro("X")
12634 #pragma pop_macro("X")
12638 In this example, the definition of X as 1 is saved by @code{#pragma
12639 push_macro} and restored by @code{#pragma pop_macro}.
12641 @node Function Specific Option Pragmas
12642 @subsection Function Specific Option Pragmas
12645 @item #pragma GCC target (@var{"string"}...)
12646 @cindex pragma GCC target
12648 This pragma allows you to set target specific options for functions
12649 defined later in the source file. One or more strings can be
12650 specified. Each function that is defined after this point will be as
12651 if @code{attribute((target("STRING")))} was specified for that
12652 function. The parenthesis around the options is optional.
12653 @xref{Function Attributes}, for more information about the
12654 @code{target} attribute and the attribute syntax.
12656 The @samp{#pragma GCC target} pragma is not implemented in GCC
12657 versions earlier than 4.4, and is currently only implemented for the
12658 386 and x86_64 backends.
12662 @item #pragma GCC optimize (@var{"string"}...)
12663 @cindex pragma GCC optimize
12665 This pragma allows you to set global optimization options for functions
12666 defined later in the source file. One or more strings can be
12667 specified. Each function that is defined after this point will be as
12668 if @code{attribute((optimize("STRING")))} was specified for that
12669 function. The parenthesis around the options is optional.
12670 @xref{Function Attributes}, for more information about the
12671 @code{optimize} attribute and the attribute syntax.
12673 The @samp{#pragma GCC optimize} pragma is not implemented in GCC
12674 versions earlier than 4.4.
12678 @item #pragma GCC push_options
12679 @itemx #pragma GCC pop_options
12680 @cindex pragma GCC push_options
12681 @cindex pragma GCC pop_options
12683 These pragmas maintain a stack of the current target and optimization
12684 options. It is intended for include files where you temporarily want
12685 to switch to using a different @samp{#pragma GCC target} or
12686 @samp{#pragma GCC optimize} and then to pop back to the previous
12689 The @samp{#pragma GCC push_options} and @samp{#pragma GCC pop_options}
12690 pragmas are not implemented in GCC versions earlier than 4.4.
12694 @item #pragma GCC reset_options
12695 @cindex pragma GCC reset_options
12697 This pragma clears the current @code{#pragma GCC target} and
12698 @code{#pragma GCC optimize} to use the default switches as specified
12699 on the command line.
12701 The @samp{#pragma GCC reset_options} pragma is not implemented in GCC
12702 versions earlier than 4.4.
12705 @node Unnamed Fields
12706 @section Unnamed struct/union fields within structs/unions
12710 For compatibility with other compilers, GCC allows you to define
12711 a structure or union that contains, as fields, structures and unions
12712 without names. For example:
12725 In this example, the user would be able to access members of the unnamed
12726 union with code like @samp{foo.b}. Note that only unnamed structs and
12727 unions are allowed, you may not have, for example, an unnamed
12730 You must never create such structures that cause ambiguous field definitions.
12731 For example, this structure:
12742 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
12743 Such constructs are not supported and must be avoided. In the future,
12744 such constructs may be detected and treated as compilation errors.
12746 @opindex fms-extensions
12747 Unless @option{-fms-extensions} is used, the unnamed field must be a
12748 structure or union definition without a tag (for example, @samp{struct
12749 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
12750 also be a definition with a tag such as @samp{struct foo @{ int a;
12751 @};}, a reference to a previously defined structure or union such as
12752 @samp{struct foo;}, or a reference to a @code{typedef} name for a
12753 previously defined structure or union type.
12756 @section Thread-Local Storage
12757 @cindex Thread-Local Storage
12758 @cindex @acronym{TLS}
12761 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
12762 are allocated such that there is one instance of the variable per extant
12763 thread. The run-time model GCC uses to implement this originates
12764 in the IA-64 processor-specific ABI, but has since been migrated
12765 to other processors as well. It requires significant support from
12766 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
12767 system libraries (@file{libc.so} and @file{libpthread.so}), so it
12768 is not available everywhere.
12770 At the user level, the extension is visible with a new storage
12771 class keyword: @code{__thread}. For example:
12775 extern __thread struct state s;
12776 static __thread char *p;
12779 The @code{__thread} specifier may be used alone, with the @code{extern}
12780 or @code{static} specifiers, but with no other storage class specifier.
12781 When used with @code{extern} or @code{static}, @code{__thread} must appear
12782 immediately after the other storage class specifier.
12784 The @code{__thread} specifier may be applied to any global, file-scoped
12785 static, function-scoped static, or static data member of a class. It may
12786 not be applied to block-scoped automatic or non-static data member.
12788 When the address-of operator is applied to a thread-local variable, it is
12789 evaluated at run-time and returns the address of the current thread's
12790 instance of that variable. An address so obtained may be used by any
12791 thread. When a thread terminates, any pointers to thread-local variables
12792 in that thread become invalid.
12794 No static initialization may refer to the address of a thread-local variable.
12796 In C++, if an initializer is present for a thread-local variable, it must
12797 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
12800 See @uref{http://people.redhat.com/drepper/tls.pdf,
12801 ELF Handling For Thread-Local Storage} for a detailed explanation of
12802 the four thread-local storage addressing models, and how the run-time
12803 is expected to function.
12806 * C99 Thread-Local Edits::
12807 * C++98 Thread-Local Edits::
12810 @node C99 Thread-Local Edits
12811 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
12813 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
12814 that document the exact semantics of the language extension.
12818 @cite{5.1.2 Execution environments}
12820 Add new text after paragraph 1
12823 Within either execution environment, a @dfn{thread} is a flow of
12824 control within a program. It is implementation defined whether
12825 or not there may be more than one thread associated with a program.
12826 It is implementation defined how threads beyond the first are
12827 created, the name and type of the function called at thread
12828 startup, and how threads may be terminated. However, objects
12829 with thread storage duration shall be initialized before thread
12834 @cite{6.2.4 Storage durations of objects}
12836 Add new text before paragraph 3
12839 An object whose identifier is declared with the storage-class
12840 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
12841 Its lifetime is the entire execution of the thread, and its
12842 stored value is initialized only once, prior to thread startup.
12846 @cite{6.4.1 Keywords}
12848 Add @code{__thread}.
12851 @cite{6.7.1 Storage-class specifiers}
12853 Add @code{__thread} to the list of storage class specifiers in
12856 Change paragraph 2 to
12859 With the exception of @code{__thread}, at most one storage-class
12860 specifier may be given [@dots{}]. The @code{__thread} specifier may
12861 be used alone, or immediately following @code{extern} or
12865 Add new text after paragraph 6
12868 The declaration of an identifier for a variable that has
12869 block scope that specifies @code{__thread} shall also
12870 specify either @code{extern} or @code{static}.
12872 The @code{__thread} specifier shall be used only with
12877 @node C++98 Thread-Local Edits
12878 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
12880 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
12881 that document the exact semantics of the language extension.
12885 @b{[intro.execution]}
12887 New text after paragraph 4
12890 A @dfn{thread} is a flow of control within the abstract machine.
12891 It is implementation defined whether or not there may be more than
12895 New text after paragraph 7
12898 It is unspecified whether additional action must be taken to
12899 ensure when and whether side effects are visible to other threads.
12905 Add @code{__thread}.
12908 @b{[basic.start.main]}
12910 Add after paragraph 5
12913 The thread that begins execution at the @code{main} function is called
12914 the @dfn{main thread}. It is implementation defined how functions
12915 beginning threads other than the main thread are designated or typed.
12916 A function so designated, as well as the @code{main} function, is called
12917 a @dfn{thread startup function}. It is implementation defined what
12918 happens if a thread startup function returns. It is implementation
12919 defined what happens to other threads when any thread calls @code{exit}.
12923 @b{[basic.start.init]}
12925 Add after paragraph 4
12928 The storage for an object of thread storage duration shall be
12929 statically initialized before the first statement of the thread startup
12930 function. An object of thread storage duration shall not require
12931 dynamic initialization.
12935 @b{[basic.start.term]}
12937 Add after paragraph 3
12940 The type of an object with thread storage duration shall not have a
12941 non-trivial destructor, nor shall it be an array type whose elements
12942 (directly or indirectly) have non-trivial destructors.
12948 Add ``thread storage duration'' to the list in paragraph 1.
12953 Thread, static, and automatic storage durations are associated with
12954 objects introduced by declarations [@dots{}].
12957 Add @code{__thread} to the list of specifiers in paragraph 3.
12960 @b{[basic.stc.thread]}
12962 New section before @b{[basic.stc.static]}
12965 The keyword @code{__thread} applied to a non-local object gives the
12966 object thread storage duration.
12968 A local variable or class data member declared both @code{static}
12969 and @code{__thread} gives the variable or member thread storage
12974 @b{[basic.stc.static]}
12979 All objects which have neither thread storage duration, dynamic
12980 storage duration nor are local [@dots{}].
12986 Add @code{__thread} to the list in paragraph 1.
12991 With the exception of @code{__thread}, at most one
12992 @var{storage-class-specifier} shall appear in a given
12993 @var{decl-specifier-seq}. The @code{__thread} specifier may
12994 be used alone, or immediately following the @code{extern} or
12995 @code{static} specifiers. [@dots{}]
12998 Add after paragraph 5
13001 The @code{__thread} specifier can be applied only to the names of objects
13002 and to anonymous unions.
13008 Add after paragraph 6
13011 Non-@code{static} members shall not be @code{__thread}.
13015 @node Binary constants
13016 @section Binary constants using the @samp{0b} prefix
13017 @cindex Binary constants using the @samp{0b} prefix
13019 Integer constants can be written as binary constants, consisting of a
13020 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
13021 @samp{0B}. This is particularly useful in environments that operate a
13022 lot on the bit-level (like microcontrollers).
13024 The following statements are identical:
13033 The type of these constants follows the same rules as for octal or
13034 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
13037 @node C++ Extensions
13038 @chapter Extensions to the C++ Language
13039 @cindex extensions, C++ language
13040 @cindex C++ language extensions
13042 The GNU compiler provides these extensions to the C++ language (and you
13043 can also use most of the C language extensions in your C++ programs). If you
13044 want to write code that checks whether these features are available, you can
13045 test for the GNU compiler the same way as for C programs: check for a
13046 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
13047 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
13048 Predefined Macros,cpp,The GNU C Preprocessor}).
13051 * Volatiles:: What constitutes an access to a volatile object.
13052 * Restricted Pointers:: C99 restricted pointers and references.
13053 * Vague Linkage:: Where G++ puts inlines, vtables and such.
13054 * C++ Interface:: You can use a single C++ header file for both
13055 declarations and definitions.
13056 * Template Instantiation:: Methods for ensuring that exactly one copy of
13057 each needed template instantiation is emitted.
13058 * Bound member functions:: You can extract a function pointer to the
13059 method denoted by a @samp{->*} or @samp{.*} expression.
13060 * C++ Attributes:: Variable, function, and type attributes for C++ only.
13061 * Namespace Association:: Strong using-directives for namespace association.
13062 * Type Traits:: Compiler support for type traits
13063 * Java Exceptions:: Tweaking exception handling to work with Java.
13064 * Deprecated Features:: Things will disappear from g++.
13065 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
13069 @section When is a Volatile Object Accessed?
13070 @cindex accessing volatiles
13071 @cindex volatile read
13072 @cindex volatile write
13073 @cindex volatile access
13075 Both the C and C++ standard have the concept of volatile objects. These
13076 are normally accessed by pointers and used for accessing hardware. The
13077 standards encourage compilers to refrain from optimizations concerning
13078 accesses to volatile objects. The C standard leaves it implementation
13079 defined as to what constitutes a volatile access. The C++ standard omits
13080 to specify this, except to say that C++ should behave in a similar manner
13081 to C with respect to volatiles, where possible. The minimum either
13082 standard specifies is that at a sequence point all previous accesses to
13083 volatile objects have stabilized and no subsequent accesses have
13084 occurred. Thus an implementation is free to reorder and combine
13085 volatile accesses which occur between sequence points, but cannot do so
13086 for accesses across a sequence point. The use of volatiles does not
13087 allow you to violate the restriction on updating objects multiple times
13088 within a sequence point.
13090 @xref{Qualifiers implementation, , Volatile qualifier and the C compiler}.
13092 The behavior differs slightly between C and C++ in the non-obvious cases:
13095 volatile int *src = @var{somevalue};
13099 With C, such expressions are rvalues, and GCC interprets this either as a
13100 read of the volatile object being pointed to or only as request to evaluate
13101 the side-effects. The C++ standard specifies that such expressions do not
13102 undergo lvalue to rvalue conversion, and that the type of the dereferenced
13103 object may be incomplete. The C++ standard does not specify explicitly
13104 that it is this lvalue to rvalue conversion which may be responsible for
13105 causing an access. However, there is reason to believe that it is,
13106 because otherwise certain simple expressions become undefined. However,
13107 because it would surprise most programmers, G++ treats dereferencing a
13108 pointer to volatile object of complete type when the value is unused as
13109 GCC would do for an equivalent type in C@. When the object has incomplete
13110 type, G++ issues a warning; if you wish to force an error, you must
13111 force a conversion to rvalue with, for instance, a static cast.
13113 When using a reference to volatile, G++ does not treat equivalent
13114 expressions as accesses to volatiles, but instead issues a warning that
13115 no volatile is accessed. The rationale for this is that otherwise it
13116 becomes difficult to determine where volatile access occur, and not
13117 possible to ignore the return value from functions returning volatile
13118 references. Again, if you wish to force a read, cast the reference to
13121 @node Restricted Pointers
13122 @section Restricting Pointer Aliasing
13123 @cindex restricted pointers
13124 @cindex restricted references
13125 @cindex restricted this pointer
13127 As with the C front end, G++ understands the C99 feature of restricted pointers,
13128 specified with the @code{__restrict__}, or @code{__restrict} type
13129 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
13130 language flag, @code{restrict} is not a keyword in C++.
13132 In addition to allowing restricted pointers, you can specify restricted
13133 references, which indicate that the reference is not aliased in the local
13137 void fn (int *__restrict__ rptr, int &__restrict__ rref)
13144 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
13145 @var{rref} refers to a (different) unaliased integer.
13147 You may also specify whether a member function's @var{this} pointer is
13148 unaliased by using @code{__restrict__} as a member function qualifier.
13151 void T::fn () __restrict__
13158 Within the body of @code{T::fn}, @var{this} will have the effective
13159 definition @code{T *__restrict__ const this}. Notice that the
13160 interpretation of a @code{__restrict__} member function qualifier is
13161 different to that of @code{const} or @code{volatile} qualifier, in that it
13162 is applied to the pointer rather than the object. This is consistent with
13163 other compilers which implement restricted pointers.
13165 As with all outermost parameter qualifiers, @code{__restrict__} is
13166 ignored in function definition matching. This means you only need to
13167 specify @code{__restrict__} in a function definition, rather than
13168 in a function prototype as well.
13170 @node Vague Linkage
13171 @section Vague Linkage
13172 @cindex vague linkage
13174 There are several constructs in C++ which require space in the object
13175 file but are not clearly tied to a single translation unit. We say that
13176 these constructs have ``vague linkage''. Typically such constructs are
13177 emitted wherever they are needed, though sometimes we can be more
13181 @item Inline Functions
13182 Inline functions are typically defined in a header file which can be
13183 included in many different compilations. Hopefully they can usually be
13184 inlined, but sometimes an out-of-line copy is necessary, if the address
13185 of the function is taken or if inlining fails. In general, we emit an
13186 out-of-line copy in all translation units where one is needed. As an
13187 exception, we only emit inline virtual functions with the vtable, since
13188 it will always require a copy.
13190 Local static variables and string constants used in an inline function
13191 are also considered to have vague linkage, since they must be shared
13192 between all inlined and out-of-line instances of the function.
13196 C++ virtual functions are implemented in most compilers using a lookup
13197 table, known as a vtable. The vtable contains pointers to the virtual
13198 functions provided by a class, and each object of the class contains a
13199 pointer to its vtable (or vtables, in some multiple-inheritance
13200 situations). If the class declares any non-inline, non-pure virtual
13201 functions, the first one is chosen as the ``key method'' for the class,
13202 and the vtable is only emitted in the translation unit where the key
13205 @emph{Note:} If the chosen key method is later defined as inline, the
13206 vtable will still be emitted in every translation unit which defines it.
13207 Make sure that any inline virtuals are declared inline in the class
13208 body, even if they are not defined there.
13210 @item type_info objects
13213 C++ requires information about types to be written out in order to
13214 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
13215 For polymorphic classes (classes with virtual functions), the type_info
13216 object is written out along with the vtable so that @samp{dynamic_cast}
13217 can determine the dynamic type of a class object at runtime. For all
13218 other types, we write out the type_info object when it is used: when
13219 applying @samp{typeid} to an expression, throwing an object, or
13220 referring to a type in a catch clause or exception specification.
13222 @item Template Instantiations
13223 Most everything in this section also applies to template instantiations,
13224 but there are other options as well.
13225 @xref{Template Instantiation,,Where's the Template?}.
13229 When used with GNU ld version 2.8 or later on an ELF system such as
13230 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
13231 these constructs will be discarded at link time. This is known as
13234 On targets that don't support COMDAT, but do support weak symbols, GCC
13235 will use them. This way one copy will override all the others, but
13236 the unused copies will still take up space in the executable.
13238 For targets which do not support either COMDAT or weak symbols,
13239 most entities with vague linkage will be emitted as local symbols to
13240 avoid duplicate definition errors from the linker. This will not happen
13241 for local statics in inlines, however, as having multiple copies will
13242 almost certainly break things.
13244 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
13245 another way to control placement of these constructs.
13247 @node C++ Interface
13248 @section #pragma interface and implementation
13250 @cindex interface and implementation headers, C++
13251 @cindex C++ interface and implementation headers
13252 @cindex pragmas, interface and implementation
13254 @code{#pragma interface} and @code{#pragma implementation} provide the
13255 user with a way of explicitly directing the compiler to emit entities
13256 with vague linkage (and debugging information) in a particular
13259 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
13260 most cases, because of COMDAT support and the ``key method'' heuristic
13261 mentioned in @ref{Vague Linkage}. Using them can actually cause your
13262 program to grow due to unnecessary out-of-line copies of inline
13263 functions. Currently (3.4) the only benefit of these
13264 @code{#pragma}s is reduced duplication of debugging information, and
13265 that should be addressed soon on DWARF 2 targets with the use of
13269 @item #pragma interface
13270 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
13271 @kindex #pragma interface
13272 Use this directive in @emph{header files} that define object classes, to save
13273 space in most of the object files that use those classes. Normally,
13274 local copies of certain information (backup copies of inline member
13275 functions, debugging information, and the internal tables that implement
13276 virtual functions) must be kept in each object file that includes class
13277 definitions. You can use this pragma to avoid such duplication. When a
13278 header file containing @samp{#pragma interface} is included in a
13279 compilation, this auxiliary information will not be generated (unless
13280 the main input source file itself uses @samp{#pragma implementation}).
13281 Instead, the object files will contain references to be resolved at link
13284 The second form of this directive is useful for the case where you have
13285 multiple headers with the same name in different directories. If you
13286 use this form, you must specify the same string to @samp{#pragma
13289 @item #pragma implementation
13290 @itemx #pragma implementation "@var{objects}.h"
13291 @kindex #pragma implementation
13292 Use this pragma in a @emph{main input file}, when you want full output from
13293 included header files to be generated (and made globally visible). The
13294 included header file, in turn, should use @samp{#pragma interface}.
13295 Backup copies of inline member functions, debugging information, and the
13296 internal tables used to implement virtual functions are all generated in
13297 implementation files.
13299 @cindex implied @code{#pragma implementation}
13300 @cindex @code{#pragma implementation}, implied
13301 @cindex naming convention, implementation headers
13302 If you use @samp{#pragma implementation} with no argument, it applies to
13303 an include file with the same basename@footnote{A file's @dfn{basename}
13304 was the name stripped of all leading path information and of trailing
13305 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
13306 file. For example, in @file{allclass.cc}, giving just
13307 @samp{#pragma implementation}
13308 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
13310 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
13311 an implementation file whenever you would include it from
13312 @file{allclass.cc} even if you never specified @samp{#pragma
13313 implementation}. This was deemed to be more trouble than it was worth,
13314 however, and disabled.
13316 Use the string argument if you want a single implementation file to
13317 include code from multiple header files. (You must also use
13318 @samp{#include} to include the header file; @samp{#pragma
13319 implementation} only specifies how to use the file---it doesn't actually
13322 There is no way to split up the contents of a single header file into
13323 multiple implementation files.
13326 @cindex inlining and C++ pragmas
13327 @cindex C++ pragmas, effect on inlining
13328 @cindex pragmas in C++, effect on inlining
13329 @samp{#pragma implementation} and @samp{#pragma interface} also have an
13330 effect on function inlining.
13332 If you define a class in a header file marked with @samp{#pragma
13333 interface}, the effect on an inline function defined in that class is
13334 similar to an explicit @code{extern} declaration---the compiler emits
13335 no code at all to define an independent version of the function. Its
13336 definition is used only for inlining with its callers.
13338 @opindex fno-implement-inlines
13339 Conversely, when you include the same header file in a main source file
13340 that declares it as @samp{#pragma implementation}, the compiler emits
13341 code for the function itself; this defines a version of the function
13342 that can be found via pointers (or by callers compiled without
13343 inlining). If all calls to the function can be inlined, you can avoid
13344 emitting the function by compiling with @option{-fno-implement-inlines}.
13345 If any calls were not inlined, you will get linker errors.
13347 @node Template Instantiation
13348 @section Where's the Template?
13349 @cindex template instantiation
13351 C++ templates are the first language feature to require more
13352 intelligence from the environment than one usually finds on a UNIX
13353 system. Somehow the compiler and linker have to make sure that each
13354 template instance occurs exactly once in the executable if it is needed,
13355 and not at all otherwise. There are two basic approaches to this
13356 problem, which are referred to as the Borland model and the Cfront model.
13359 @item Borland model
13360 Borland C++ solved the template instantiation problem by adding the code
13361 equivalent of common blocks to their linker; the compiler emits template
13362 instances in each translation unit that uses them, and the linker
13363 collapses them together. The advantage of this model is that the linker
13364 only has to consider the object files themselves; there is no external
13365 complexity to worry about. This disadvantage is that compilation time
13366 is increased because the template code is being compiled repeatedly.
13367 Code written for this model tends to include definitions of all
13368 templates in the header file, since they must be seen to be
13372 The AT&T C++ translator, Cfront, solved the template instantiation
13373 problem by creating the notion of a template repository, an
13374 automatically maintained place where template instances are stored. A
13375 more modern version of the repository works as follows: As individual
13376 object files are built, the compiler places any template definitions and
13377 instantiations encountered in the repository. At link time, the link
13378 wrapper adds in the objects in the repository and compiles any needed
13379 instances that were not previously emitted. The advantages of this
13380 model are more optimal compilation speed and the ability to use the
13381 system linker; to implement the Borland model a compiler vendor also
13382 needs to replace the linker. The disadvantages are vastly increased
13383 complexity, and thus potential for error; for some code this can be
13384 just as transparent, but in practice it can been very difficult to build
13385 multiple programs in one directory and one program in multiple
13386 directories. Code written for this model tends to separate definitions
13387 of non-inline member templates into a separate file, which should be
13388 compiled separately.
13391 When used with GNU ld version 2.8 or later on an ELF system such as
13392 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
13393 Borland model. On other systems, G++ implements neither automatic
13396 A future version of G++ will support a hybrid model whereby the compiler
13397 will emit any instantiations for which the template definition is
13398 included in the compile, and store template definitions and
13399 instantiation context information into the object file for the rest.
13400 The link wrapper will extract that information as necessary and invoke
13401 the compiler to produce the remaining instantiations. The linker will
13402 then combine duplicate instantiations.
13404 In the mean time, you have the following options for dealing with
13405 template instantiations:
13410 Compile your template-using code with @option{-frepo}. The compiler will
13411 generate files with the extension @samp{.rpo} listing all of the
13412 template instantiations used in the corresponding object files which
13413 could be instantiated there; the link wrapper, @samp{collect2}, will
13414 then update the @samp{.rpo} files to tell the compiler where to place
13415 those instantiations and rebuild any affected object files. The
13416 link-time overhead is negligible after the first pass, as the compiler
13417 will continue to place the instantiations in the same files.
13419 This is your best option for application code written for the Borland
13420 model, as it will just work. Code written for the Cfront model will
13421 need to be modified so that the template definitions are available at
13422 one or more points of instantiation; usually this is as simple as adding
13423 @code{#include <tmethods.cc>} to the end of each template header.
13425 For library code, if you want the library to provide all of the template
13426 instantiations it needs, just try to link all of its object files
13427 together; the link will fail, but cause the instantiations to be
13428 generated as a side effect. Be warned, however, that this may cause
13429 conflicts if multiple libraries try to provide the same instantiations.
13430 For greater control, use explicit instantiation as described in the next
13434 @opindex fno-implicit-templates
13435 Compile your code with @option{-fno-implicit-templates} to disable the
13436 implicit generation of template instances, and explicitly instantiate
13437 all the ones you use. This approach requires more knowledge of exactly
13438 which instances you need than do the others, but it's less
13439 mysterious and allows greater control. You can scatter the explicit
13440 instantiations throughout your program, perhaps putting them in the
13441 translation units where the instances are used or the translation units
13442 that define the templates themselves; you can put all of the explicit
13443 instantiations you need into one big file; or you can create small files
13450 template class Foo<int>;
13451 template ostream& operator <<
13452 (ostream&, const Foo<int>&);
13455 for each of the instances you need, and create a template instantiation
13456 library from those.
13458 If you are using Cfront-model code, you can probably get away with not
13459 using @option{-fno-implicit-templates} when compiling files that don't
13460 @samp{#include} the member template definitions.
13462 If you use one big file to do the instantiations, you may want to
13463 compile it without @option{-fno-implicit-templates} so you get all of the
13464 instances required by your explicit instantiations (but not by any
13465 other files) without having to specify them as well.
13467 G++ has extended the template instantiation syntax given in the ISO
13468 standard to allow forward declaration of explicit instantiations
13469 (with @code{extern}), instantiation of the compiler support data for a
13470 template class (i.e.@: the vtable) without instantiating any of its
13471 members (with @code{inline}), and instantiation of only the static data
13472 members of a template class, without the support data or member
13473 functions (with (@code{static}):
13476 extern template int max (int, int);
13477 inline template class Foo<int>;
13478 static template class Foo<int>;
13482 Do nothing. Pretend G++ does implement automatic instantiation
13483 management. Code written for the Borland model will work fine, but
13484 each translation unit will contain instances of each of the templates it
13485 uses. In a large program, this can lead to an unacceptable amount of code
13489 @node Bound member functions
13490 @section Extracting the function pointer from a bound pointer to member function
13492 @cindex pointer to member function
13493 @cindex bound pointer to member function
13495 In C++, pointer to member functions (PMFs) are implemented using a wide
13496 pointer of sorts to handle all the possible call mechanisms; the PMF
13497 needs to store information about how to adjust the @samp{this} pointer,
13498 and if the function pointed to is virtual, where to find the vtable, and
13499 where in the vtable to look for the member function. If you are using
13500 PMFs in an inner loop, you should really reconsider that decision. If
13501 that is not an option, you can extract the pointer to the function that
13502 would be called for a given object/PMF pair and call it directly inside
13503 the inner loop, to save a bit of time.
13505 Note that you will still be paying the penalty for the call through a
13506 function pointer; on most modern architectures, such a call defeats the
13507 branch prediction features of the CPU@. This is also true of normal
13508 virtual function calls.
13510 The syntax for this extension is
13514 extern int (A::*fp)();
13515 typedef int (*fptr)(A *);
13517 fptr p = (fptr)(a.*fp);
13520 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
13521 no object is needed to obtain the address of the function. They can be
13522 converted to function pointers directly:
13525 fptr p1 = (fptr)(&A::foo);
13528 @opindex Wno-pmf-conversions
13529 You must specify @option{-Wno-pmf-conversions} to use this extension.
13531 @node C++ Attributes
13532 @section C++-Specific Variable, Function, and Type Attributes
13534 Some attributes only make sense for C++ programs.
13537 @item init_priority (@var{priority})
13538 @cindex init_priority attribute
13541 In Standard C++, objects defined at namespace scope are guaranteed to be
13542 initialized in an order in strict accordance with that of their definitions
13543 @emph{in a given translation unit}. No guarantee is made for initializations
13544 across translation units. However, GNU C++ allows users to control the
13545 order of initialization of objects defined at namespace scope with the
13546 @code{init_priority} attribute by specifying a relative @var{priority},
13547 a constant integral expression currently bounded between 101 and 65535
13548 inclusive. Lower numbers indicate a higher priority.
13550 In the following example, @code{A} would normally be created before
13551 @code{B}, but the @code{init_priority} attribute has reversed that order:
13554 Some_Class A __attribute__ ((init_priority (2000)));
13555 Some_Class B __attribute__ ((init_priority (543)));
13559 Note that the particular values of @var{priority} do not matter; only their
13562 @item java_interface
13563 @cindex java_interface attribute
13565 This type attribute informs C++ that the class is a Java interface. It may
13566 only be applied to classes declared within an @code{extern "Java"} block.
13567 Calls to methods declared in this interface will be dispatched using GCJ's
13568 interface table mechanism, instead of regular virtual table dispatch.
13572 See also @ref{Namespace Association}.
13574 @node Namespace Association
13575 @section Namespace Association
13577 @strong{Caution:} The semantics of this extension are not fully
13578 defined. Users should refrain from using this extension as its
13579 semantics may change subtly over time. It is possible that this
13580 extension will be removed in future versions of G++.
13582 A using-directive with @code{__attribute ((strong))} is stronger
13583 than a normal using-directive in two ways:
13587 Templates from the used namespace can be specialized and explicitly
13588 instantiated as though they were members of the using namespace.
13591 The using namespace is considered an associated namespace of all
13592 templates in the used namespace for purposes of argument-dependent
13596 The used namespace must be nested within the using namespace so that
13597 normal unqualified lookup works properly.
13599 This is useful for composing a namespace transparently from
13600 implementation namespaces. For example:
13605 template <class T> struct A @{ @};
13607 using namespace debug __attribute ((__strong__));
13608 template <> struct A<int> @{ @}; // @r{ok to specialize}
13610 template <class T> void f (A<T>);
13615 f (std::A<float>()); // @r{lookup finds} std::f
13621 @section Type Traits
13623 The C++ front-end implements syntactic extensions that allow to
13624 determine at compile time various characteristics of a type (or of a
13628 @item __has_nothrow_assign (type)
13629 If @code{type} is const qualified or is a reference type then the trait is
13630 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
13631 is true, else if @code{type} is a cv class or union type with copy assignment
13632 operators that are known not to throw an exception then the trait is true,
13633 else it is false. Requires: @code{type} shall be a complete type, an array
13634 type of unknown bound, or is a @code{void} type.
13636 @item __has_nothrow_copy (type)
13637 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
13638 @code{type} is a cv class or union type with copy constructors that
13639 are known not to throw an exception then the trait is true, else it is false.
13640 Requires: @code{type} shall be a complete type, an array type of
13641 unknown bound, or is a @code{void} type.
13643 @item __has_nothrow_constructor (type)
13644 If @code{__has_trivial_constructor (type)} is true then the trait is
13645 true, else if @code{type} is a cv class or union type (or array
13646 thereof) with a default constructor that is known not to throw an
13647 exception then the trait is true, else it is false. Requires:
13648 @code{type} shall be a complete type, an array type of unknown bound,
13649 or is a @code{void} type.
13651 @item __has_trivial_assign (type)
13652 If @code{type} is const qualified or is a reference type then the trait is
13653 false. Otherwise if @code{__is_pod (type)} is true then the trait is
13654 true, else if @code{type} is a cv class or union type with a trivial
13655 copy assignment ([class.copy]) then the trait is true, else it is
13656 false. Requires: @code{type} shall be a complete type, an array type
13657 of unknown bound, or is a @code{void} type.
13659 @item __has_trivial_copy (type)
13660 If @code{__is_pod (type)} is true or @code{type} is a reference type
13661 then the trait is true, else if @code{type} is a cv class or union type
13662 with a trivial copy constructor ([class.copy]) then the trait
13663 is true, else it is false. Requires: @code{type} shall be a complete
13664 type, an array type of unknown bound, or is a @code{void} type.
13666 @item __has_trivial_constructor (type)
13667 If @code{__is_pod (type)} is true then the trait is true, else if
13668 @code{type} is a cv class or union type (or array thereof) with a
13669 trivial default constructor ([class.ctor]) then the trait is true,
13670 else it is false. Requires: @code{type} shall be a complete type, an
13671 array type of unknown bound, or is a @code{void} type.
13673 @item __has_trivial_destructor (type)
13674 If @code{__is_pod (type)} is true or @code{type} is a reference type then
13675 the trait is true, else if @code{type} is a cv class or union type (or
13676 array thereof) with a trivial destructor ([class.dtor]) then the trait
13677 is true, else it is false. Requires: @code{type} shall be a complete
13678 type, an array type of unknown bound, or is a @code{void} type.
13680 @item __has_virtual_destructor (type)
13681 If @code{type} is a class type with a virtual destructor
13682 ([class.dtor]) then the trait is true, else it is false. Requires:
13683 @code{type} shall be a complete type, an array type of unknown bound,
13684 or is a @code{void} type.
13686 @item __is_abstract (type)
13687 If @code{type} is an abstract class ([class.abstract]) then the trait
13688 is true, else it is false. Requires: @code{type} shall be a complete
13689 type, an array type of unknown bound, or is a @code{void} type.
13691 @item __is_base_of (base_type, derived_type)
13692 If @code{base_type} is a base class of @code{derived_type}
13693 ([class.derived]) then the trait is true, otherwise it is false.
13694 Top-level cv qualifications of @code{base_type} and
13695 @code{derived_type} are ignored. For the purposes of this trait, a
13696 class type is considered is own base. Requires: if @code{__is_class
13697 (base_type)} and @code{__is_class (derived_type)} are true and
13698 @code{base_type} and @code{derived_type} are not the same type
13699 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
13700 type. Diagnostic is produced if this requirement is not met.
13702 @item __is_class (type)
13703 If @code{type} is a cv class type, and not a union type
13704 ([basic.compound]) the trait is true, else it is false.
13706 @item __is_empty (type)
13707 If @code{__is_class (type)} is false then the trait is false.
13708 Otherwise @code{type} is considered empty if and only if: @code{type}
13709 has no non-static data members, or all non-static data members, if
13710 any, are bit-fields of length 0, and @code{type} has no virtual
13711 members, and @code{type} has no virtual base classes, and @code{type}
13712 has no base classes @code{base_type} for which
13713 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
13714 be a complete type, an array type of unknown bound, or is a
13717 @item __is_enum (type)
13718 If @code{type} is a cv enumeration type ([basic.compound]) the trait is
13719 true, else it is false.
13721 @item __is_pod (type)
13722 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
13723 else it is false. Requires: @code{type} shall be a complete type,
13724 an array type of unknown bound, or is a @code{void} type.
13726 @item __is_polymorphic (type)
13727 If @code{type} is a polymorphic class ([class.virtual]) then the trait
13728 is true, else it is false. Requires: @code{type} shall be a complete
13729 type, an array type of unknown bound, or is a @code{void} type.
13731 @item __is_union (type)
13732 If @code{type} is a cv union type ([basic.compound]) the trait is
13733 true, else it is false.
13737 @node Java Exceptions
13738 @section Java Exceptions
13740 The Java language uses a slightly different exception handling model
13741 from C++. Normally, GNU C++ will automatically detect when you are
13742 writing C++ code that uses Java exceptions, and handle them
13743 appropriately. However, if C++ code only needs to execute destructors
13744 when Java exceptions are thrown through it, GCC will guess incorrectly.
13745 Sample problematic code is:
13748 struct S @{ ~S(); @};
13749 extern void bar(); // @r{is written in Java, and may throw exceptions}
13758 The usual effect of an incorrect guess is a link failure, complaining of
13759 a missing routine called @samp{__gxx_personality_v0}.
13761 You can inform the compiler that Java exceptions are to be used in a
13762 translation unit, irrespective of what it might think, by writing
13763 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
13764 @samp{#pragma} must appear before any functions that throw or catch
13765 exceptions, or run destructors when exceptions are thrown through them.
13767 You cannot mix Java and C++ exceptions in the same translation unit. It
13768 is believed to be safe to throw a C++ exception from one file through
13769 another file compiled for the Java exception model, or vice versa, but
13770 there may be bugs in this area.
13772 @node Deprecated Features
13773 @section Deprecated Features
13775 In the past, the GNU C++ compiler was extended to experiment with new
13776 features, at a time when the C++ language was still evolving. Now that
13777 the C++ standard is complete, some of those features are superseded by
13778 superior alternatives. Using the old features might cause a warning in
13779 some cases that the feature will be dropped in the future. In other
13780 cases, the feature might be gone already.
13782 While the list below is not exhaustive, it documents some of the options
13783 that are now deprecated:
13786 @item -fexternal-templates
13787 @itemx -falt-external-templates
13788 These are two of the many ways for G++ to implement template
13789 instantiation. @xref{Template Instantiation}. The C++ standard clearly
13790 defines how template definitions have to be organized across
13791 implementation units. G++ has an implicit instantiation mechanism that
13792 should work just fine for standard-conforming code.
13794 @item -fstrict-prototype
13795 @itemx -fno-strict-prototype
13796 Previously it was possible to use an empty prototype parameter list to
13797 indicate an unspecified number of parameters (like C), rather than no
13798 parameters, as C++ demands. This feature has been removed, except where
13799 it is required for backwards compatibility. @xref{Backwards Compatibility}.
13802 G++ allows a virtual function returning @samp{void *} to be overridden
13803 by one returning a different pointer type. This extension to the
13804 covariant return type rules is now deprecated and will be removed from a
13807 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
13808 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
13809 and are now removed from G++. Code using these operators should be
13810 modified to use @code{std::min} and @code{std::max} instead.
13812 The named return value extension has been deprecated, and is now
13815 The use of initializer lists with new expressions has been deprecated,
13816 and is now removed from G++.
13818 Floating and complex non-type template parameters have been deprecated,
13819 and are now removed from G++.
13821 The implicit typename extension has been deprecated and is now
13824 The use of default arguments in function pointers, function typedefs
13825 and other places where they are not permitted by the standard is
13826 deprecated and will be removed from a future version of G++.
13828 G++ allows floating-point literals to appear in integral constant expressions,
13829 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
13830 This extension is deprecated and will be removed from a future version.
13832 G++ allows static data members of const floating-point type to be declared
13833 with an initializer in a class definition. The standard only allows
13834 initializers for static members of const integral types and const
13835 enumeration types so this extension has been deprecated and will be removed
13836 from a future version.
13838 @node Backwards Compatibility
13839 @section Backwards Compatibility
13840 @cindex Backwards Compatibility
13841 @cindex ARM [Annotated C++ Reference Manual]
13843 Now that there is a definitive ISO standard C++, G++ has a specification
13844 to adhere to. The C++ language evolved over time, and features that
13845 used to be acceptable in previous drafts of the standard, such as the ARM
13846 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
13847 compilation of C++ written to such drafts, G++ contains some backwards
13848 compatibilities. @emph{All such backwards compatibility features are
13849 liable to disappear in future versions of G++.} They should be considered
13850 deprecated. @xref{Deprecated Features}.
13854 If a variable is declared at for scope, it used to remain in scope until
13855 the end of the scope which contained the for statement (rather than just
13856 within the for scope). G++ retains this, but issues a warning, if such a
13857 variable is accessed outside the for scope.
13859 @item Implicit C language
13860 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
13861 scope to set the language. On such systems, all header files are
13862 implicitly scoped inside a C language scope. Also, an empty prototype
13863 @code{()} will be treated as an unspecified number of arguments, rather
13864 than no arguments, as C++ demands.