1 @c Copyright (C) 1988-2015 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
7 @chapter Extensions to the C Language Family
8 @cindex extensions, C language
9 @cindex C language extensions
12 GNU C provides several language features not found in ISO standard C@.
13 (The @option{-pedantic} option directs GCC to print a warning message if
14 any of these features is used.) To test for the availability of these
15 features in conditional compilation, check for a predefined macro
16 @code{__GNUC__}, which is always defined under GCC@.
18 These extensions are available in C and Objective-C@. Most of them are
19 also available in C++. @xref{C++ Extensions,,Extensions to the
20 C++ Language}, for extensions that apply @emph{only} to C++.
22 Some features that are in ISO C99 but not C90 or C++ are also, as
23 extensions, accepted by GCC in C90 mode and in C++.
26 * Statement Exprs:: Putting statements and declarations inside expressions.
27 * Local Labels:: Labels local to a block.
28 * Labels as Values:: Getting pointers to labels, and computed gotos.
29 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
30 * Constructing Calls:: Dispatching a call to another function.
31 * Typeof:: @code{typeof}: referring to the type of an expression.
32 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
33 * __int128:: 128-bit integers---@code{__int128}.
34 * Long Long:: Double-word integers---@code{long long int}.
35 * Complex:: Data types for complex numbers.
36 * Floating Types:: Additional Floating Types.
37 * Half-Precision:: Half-Precision Floating Point.
38 * Decimal Float:: Decimal Floating Types.
39 * Hex Floats:: Hexadecimal floating-point constants.
40 * Fixed-Point:: Fixed-Point Types.
41 * Named Address Spaces::Named address spaces.
42 * Zero Length:: Zero-length arrays.
43 * Empty Structures:: Structures with no members.
44 * Variable Length:: Arrays whose length is computed at run time.
45 * Variadic Macros:: Macros with a variable number of arguments.
46 * Escaped Newlines:: Slightly looser rules for escaped newlines.
47 * Subscripting:: Any array can be subscripted, even if not an lvalue.
48 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
49 * Pointers to Arrays:: Pointers to arrays with qualifiers work as expected.
50 * Initializers:: Non-constant initializers.
51 * Compound Literals:: Compound literals give structures, unions
53 * Designated Inits:: Labeling elements of initializers.
54 * Case Ranges:: `case 1 ... 9' and such.
55 * Cast to Union:: Casting to union type from any member of the union.
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 * Variable Attributes:: Specifying attributes of variables.
60 * Type Attributes:: Specifying attributes of types.
61 * Label Attributes:: Specifying attributes on labels.
62 * Enumerator Attributes:: Specifying attributes on enumerators.
63 * Attribute Syntax:: Formal syntax for attributes.
64 * Function Prototypes:: Prototype declarations and old-style definitions.
65 * C++ Comments:: C++ comments are recognized.
66 * Dollar Signs:: Dollar sign is allowed in identifiers.
67 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
68 * Alignment:: Inquiring about the alignment of a type or variable.
69 * Inline:: Defining inline functions (as fast as macros).
70 * Volatiles:: What constitutes an access to a volatile object.
71 * Using Assembly Language with C:: Instructions and extensions for interfacing C with assembler.
72 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
73 * Incomplete Enums:: @code{enum foo;}, with details to follow.
74 * Function Names:: Printable strings which are the name of the current
76 * Return Address:: Getting the return or frame address of a function.
77 * Vector Extensions:: Using vector instructions through built-in functions.
78 * Offsetof:: Special syntax for implementing @code{offsetof}.
79 * __sync Builtins:: Legacy built-in functions for atomic memory access.
80 * __atomic Builtins:: Atomic built-in functions with memory model.
81 * Integer Overflow Builtins:: Built-in functions to perform arithmetics and
82 arithmetic overflow checking.
83 * x86 specific memory model extensions for transactional memory:: x86 memory models.
84 * Object Size Checking:: Built-in functions for limited buffer overflow
86 * Pointer Bounds Checker builtins:: Built-in functions for Pointer Bounds Checker.
87 * Cilk Plus Builtins:: Built-in functions for the Cilk Plus language extension.
88 * Other Builtins:: Other built-in functions.
89 * Target Builtins:: Built-in functions specific to particular targets.
90 * Target Format Checks:: Format checks specific to particular targets.
91 * Pragmas:: Pragmas accepted by GCC.
92 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
93 * Thread-Local:: Per-thread variables.
94 * Binary constants:: Binary constants using the @samp{0b} prefix.
98 @section Statements and Declarations in Expressions
99 @cindex statements inside expressions
100 @cindex declarations inside expressions
101 @cindex expressions containing statements
102 @cindex macros, statements in expressions
104 @c the above section title wrapped and causes an underfull hbox.. i
105 @c changed it from "within" to "in". --mew 4feb93
106 A compound statement enclosed in parentheses may appear as an expression
107 in GNU C@. This allows you to use loops, switches, and local variables
108 within an expression.
110 Recall that a compound statement is a sequence of statements surrounded
111 by braces; in this construct, parentheses go around the braces. For
115 (@{ int y = foo (); int z;
122 is a valid (though slightly more complex than necessary) expression
123 for the absolute value of @code{foo ()}.
125 The last thing in the compound statement should be an expression
126 followed by a semicolon; the value of this subexpression serves as the
127 value of the entire construct. (If you use some other kind of statement
128 last within the braces, the construct has type @code{void}, and thus
129 effectively no value.)
131 This feature is especially useful in making macro definitions ``safe'' (so
132 that they evaluate each operand exactly once). For example, the
133 ``maximum'' function is commonly defined as a macro in standard C as
137 #define max(a,b) ((a) > (b) ? (a) : (b))
141 @cindex side effects, macro argument
142 But this definition computes either @var{a} or @var{b} twice, with bad
143 results if the operand has side effects. In GNU C, if you know the
144 type of the operands (here taken as @code{int}), you can define
145 the macro safely as follows:
148 #define maxint(a,b) \
149 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
152 Embedded statements are not allowed in constant expressions, such as
153 the value of an enumeration constant, the width of a bit-field, or
154 the initial value of a static variable.
156 If you don't know the type of the operand, you can still do this, but you
157 must use @code{typeof} or @code{__auto_type} (@pxref{Typeof}).
159 In G++, the result value of a statement expression undergoes array and
160 function pointer decay, and is returned by value to the enclosing
161 expression. For instance, if @code{A} is a class, then
170 constructs a temporary @code{A} object to hold the result of the
171 statement expression, and that is used to invoke @code{Foo}.
172 Therefore the @code{this} pointer observed by @code{Foo} is not the
175 In a statement expression, any temporaries created within a statement
176 are destroyed at that statement's end. This makes statement
177 expressions inside macros slightly different from function calls. In
178 the latter case temporaries introduced during argument evaluation are
179 destroyed at the end of the statement that includes the function
180 call. In the statement expression case they are destroyed during
181 the statement expression. For instance,
184 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
185 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
195 has different places where temporaries are destroyed. For the
196 @code{macro} case, the temporary @code{X} is destroyed just after
197 the initialization of @code{b}. In the @code{function} case that
198 temporary is destroyed when the function returns.
200 These considerations mean that it is probably a bad idea to use
201 statement expressions of this form in header files that are designed to
202 work with C++. (Note that some versions of the GNU C Library contained
203 header files using statement expressions that lead to precisely this
206 Jumping into a statement expression with @code{goto} or using a
207 @code{switch} statement outside the statement expression with a
208 @code{case} or @code{default} label inside the statement expression is
209 not permitted. Jumping into a statement expression with a computed
210 @code{goto} (@pxref{Labels as Values}) has undefined behavior.
211 Jumping out of a statement expression is permitted, but if the
212 statement expression is part of a larger expression then it is
213 unspecified which other subexpressions of that expression have been
214 evaluated except where the language definition requires certain
215 subexpressions to be evaluated before or after the statement
216 expression. In any case, as with a function call, the evaluation of a
217 statement expression is not interleaved with the evaluation of other
218 parts of the containing expression. For example,
221 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
225 calls @code{foo} and @code{bar1} and does not call @code{baz} but
226 may or may not call @code{bar2}. If @code{bar2} is called, it is
227 called after @code{foo} and before @code{bar1}.
230 @section Locally Declared Labels
232 @cindex macros, local labels
234 GCC allows you to declare @dfn{local labels} in any nested block
235 scope. A local label is just like an ordinary label, but you can
236 only reference it (with a @code{goto} statement, or by taking its
237 address) within the block in which it is declared.
239 A local label declaration looks like this:
242 __label__ @var{label};
249 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
252 Local label declarations must come at the beginning of the block,
253 before any ordinary declarations or statements.
255 The label declaration defines the label @emph{name}, but does not define
256 the label itself. You must do this in the usual way, with
257 @code{@var{label}:}, within the statements of the statement expression.
259 The local label feature is useful for complex macros. If a macro
260 contains nested loops, a @code{goto} can be useful for breaking out of
261 them. However, an ordinary label whose scope is the whole function
262 cannot be used: if the macro can be expanded several times in one
263 function, the label is multiply defined in that function. A
264 local label avoids this problem. For example:
267 #define SEARCH(value, array, target) \
270 typeof (target) _SEARCH_target = (target); \
271 typeof (*(array)) *_SEARCH_array = (array); \
274 for (i = 0; i < max; i++) \
275 for (j = 0; j < max; j++) \
276 if (_SEARCH_array[i][j] == _SEARCH_target) \
277 @{ (value) = i; goto found; @} \
283 This could also be written using a statement expression:
286 #define SEARCH(array, target) \
289 typeof (target) _SEARCH_target = (target); \
290 typeof (*(array)) *_SEARCH_array = (array); \
293 for (i = 0; i < max; i++) \
294 for (j = 0; j < max; j++) \
295 if (_SEARCH_array[i][j] == _SEARCH_target) \
296 @{ value = i; goto found; @} \
303 Local label declarations also make the labels they declare visible to
304 nested functions, if there are any. @xref{Nested Functions}, for details.
306 @node Labels as Values
307 @section Labels as Values
308 @cindex labels as values
309 @cindex computed gotos
310 @cindex goto with computed label
311 @cindex address of a label
313 You can get the address of a label defined in the current function
314 (or a containing function) with the unary operator @samp{&&}. The
315 value has type @code{void *}. This value is a constant and can be used
316 wherever a constant of that type is valid. For example:
324 To use these values, you need to be able to jump to one. This is done
325 with the computed goto statement@footnote{The analogous feature in
326 Fortran is called an assigned goto, but that name seems inappropriate in
327 C, where one can do more than simply store label addresses in label
328 variables.}, @code{goto *@var{exp};}. For example,
335 Any expression of type @code{void *} is allowed.
337 One way of using these constants is in initializing a static array that
338 serves as a jump table:
341 static void *array[] = @{ &&foo, &&bar, &&hack @};
345 Then you can select a label with indexing, like this:
352 Note that this does not check whether the subscript is in bounds---array
353 indexing in C never does that.
355 Such an array of label values serves a purpose much like that of the
356 @code{switch} statement. The @code{switch} statement is cleaner, so
357 use that rather than an array unless the problem does not fit a
358 @code{switch} statement very well.
360 Another use of label values is in an interpreter for threaded code.
361 The labels within the interpreter function can be stored in the
362 threaded code for super-fast dispatching.
364 You may not use this mechanism to jump to code in a different function.
365 If you do that, totally unpredictable things happen. The best way to
366 avoid this is to store the label address only in automatic variables and
367 never pass it as an argument.
369 An alternate way to write the above example is
372 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
374 goto *(&&foo + array[i]);
378 This is more friendly to code living in shared libraries, as it reduces
379 the number of dynamic relocations that are needed, and by consequence,
380 allows the data to be read-only.
381 This alternative with label differences is not supported for the AVR target,
382 please use the first approach for AVR programs.
384 The @code{&&foo} expressions for the same label might have different
385 values if the containing function is inlined or cloned. If a program
386 relies on them being always the same,
387 @code{__attribute__((__noinline__,__noclone__))} should be used to
388 prevent inlining and cloning. If @code{&&foo} is used in a static
389 variable initializer, inlining and cloning is forbidden.
391 @node Nested Functions
392 @section Nested Functions
393 @cindex nested functions
394 @cindex downward funargs
397 A @dfn{nested function} is a function defined inside another function.
398 Nested functions are supported as an extension in GNU C, but are not
399 supported by GNU C++.
401 The nested function's name is local to the block where it is defined.
402 For example, here we define a nested function named @code{square}, and
407 foo (double a, double b)
409 double square (double z) @{ return z * z; @}
411 return square (a) + square (b);
416 The nested function can access all the variables of the containing
417 function that are visible at the point of its definition. This is
418 called @dfn{lexical scoping}. For example, here we show a nested
419 function which uses an inherited variable named @code{offset}:
423 bar (int *array, int offset, int size)
425 int access (int *array, int index)
426 @{ return array[index + offset]; @}
429 for (i = 0; i < size; i++)
430 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
435 Nested function definitions are permitted within functions in the places
436 where variable definitions are allowed; that is, in any block, mixed
437 with the other declarations and statements in the block.
439 It is possible to call the nested function from outside the scope of its
440 name by storing its address or passing the address to another function:
443 hack (int *array, int size)
445 void store (int index, int value)
446 @{ array[index] = value; @}
448 intermediate (store, size);
452 Here, the function @code{intermediate} receives the address of
453 @code{store} as an argument. If @code{intermediate} calls @code{store},
454 the arguments given to @code{store} are used to store into @code{array}.
455 But this technique works only so long as the containing function
456 (@code{hack}, in this example) does not exit.
458 If you try to call the nested function through its address after the
459 containing function exits, all hell breaks loose. If you try
460 to call it after a containing scope level exits, and if it refers
461 to some of the variables that are no longer in scope, you may be lucky,
462 but it's not wise to take the risk. If, however, the nested function
463 does not refer to anything that has gone out of scope, you should be
466 GCC implements taking the address of a nested function using a technique
467 called @dfn{trampolines}. This technique was described in
468 @cite{Lexical Closures for C++} (Thomas M. Breuel, USENIX
469 C++ Conference Proceedings, October 17-21, 1988).
471 A nested function can jump to a label inherited from a containing
472 function, provided the label is explicitly declared in the containing
473 function (@pxref{Local Labels}). Such a jump returns instantly to the
474 containing function, exiting the nested function that did the
475 @code{goto} and any intermediate functions as well. Here is an example:
479 bar (int *array, int offset, int size)
482 int access (int *array, int index)
486 return array[index + offset];
490 for (i = 0; i < size; i++)
491 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
495 /* @r{Control comes here from @code{access}
496 if it detects an error.} */
503 A nested function always has no linkage. Declaring one with
504 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
505 before its definition, use @code{auto} (which is otherwise meaningless
506 for function declarations).
509 bar (int *array, int offset, int size)
512 auto int access (int *, int);
514 int access (int *array, int index)
518 return array[index + offset];
524 @node Constructing Calls
525 @section Constructing Function Calls
526 @cindex constructing calls
527 @cindex forwarding calls
529 Using the built-in functions described below, you can record
530 the arguments a function received, and call another function
531 with the same arguments, without knowing the number or types
534 You can also record the return value of that function call,
535 and later return that value, without knowing what data type
536 the function tried to return (as long as your caller expects
539 However, these built-in functions may interact badly with some
540 sophisticated features or other extensions of the language. It
541 is, therefore, not recommended to use them outside very simple
542 functions acting as mere forwarders for their arguments.
544 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
545 This built-in function returns a pointer to data
546 describing how to perform a call with the same arguments as are passed
547 to the current function.
549 The function saves the arg pointer register, structure value address,
550 and all registers that might be used to pass arguments to a function
551 into a block of memory allocated on the stack. Then it returns the
552 address of that block.
555 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
556 This built-in function invokes @var{function}
557 with a copy of the parameters described by @var{arguments}
560 The value of @var{arguments} should be the value returned by
561 @code{__builtin_apply_args}. The argument @var{size} specifies the size
562 of the stack argument data, in bytes.
564 This function returns a pointer to data describing
565 how to return whatever value is returned by @var{function}. The data
566 is saved in a block of memory allocated on the stack.
568 It is not always simple to compute the proper value for @var{size}. The
569 value is used by @code{__builtin_apply} to compute the amount of data
570 that should be pushed on the stack and copied from the incoming argument
574 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
575 This built-in function returns the value described by @var{result} from
576 the containing function. You should specify, for @var{result}, a value
577 returned by @code{__builtin_apply}.
580 @deftypefn {Built-in Function} {} __builtin_va_arg_pack ()
581 This built-in function represents all anonymous arguments of an inline
582 function. It can be used only in inline functions that are always
583 inlined, never compiled as a separate function, such as those using
584 @code{__attribute__ ((__always_inline__))} or
585 @code{__attribute__ ((__gnu_inline__))} extern inline functions.
586 It must be only passed as last argument to some other function
587 with variable arguments. This is useful for writing small wrapper
588 inlines for variable argument functions, when using preprocessor
589 macros is undesirable. For example:
591 extern int myprintf (FILE *f, const char *format, ...);
592 extern inline __attribute__ ((__gnu_inline__)) int
593 myprintf (FILE *f, const char *format, ...)
595 int r = fprintf (f, "myprintf: ");
598 int s = fprintf (f, format, __builtin_va_arg_pack ());
606 @deftypefn {Built-in Function} {size_t} __builtin_va_arg_pack_len ()
607 This built-in function returns the number of anonymous arguments of
608 an inline function. It can be used only in inline functions that
609 are always inlined, never compiled as a separate function, such
610 as those using @code{__attribute__ ((__always_inline__))} or
611 @code{__attribute__ ((__gnu_inline__))} extern inline functions.
612 For example following does link- or run-time checking of open
613 arguments for optimized code:
616 extern inline __attribute__((__gnu_inline__)) int
617 myopen (const char *path, int oflag, ...)
619 if (__builtin_va_arg_pack_len () > 1)
620 warn_open_too_many_arguments ();
622 if (__builtin_constant_p (oflag))
624 if ((oflag & O_CREAT) != 0 && __builtin_va_arg_pack_len () < 1)
626 warn_open_missing_mode ();
627 return __open_2 (path, oflag);
629 return open (path, oflag, __builtin_va_arg_pack ());
632 if (__builtin_va_arg_pack_len () < 1)
633 return __open_2 (path, oflag);
635 return open (path, oflag, __builtin_va_arg_pack ());
642 @section Referring to a Type with @code{typeof}
645 @cindex macros, types of arguments
647 Another way to refer to the type of an expression is with @code{typeof}.
648 The syntax of using of this keyword looks like @code{sizeof}, but the
649 construct acts semantically like a type name defined with @code{typedef}.
651 There are two ways of writing the argument to @code{typeof}: with an
652 expression or with a type. Here is an example with an expression:
659 This assumes that @code{x} is an array of pointers to functions;
660 the type described is that of the values of the functions.
662 Here is an example with a typename as the argument:
669 Here the type described is that of pointers to @code{int}.
671 If you are writing a header file that must work when included in ISO C
672 programs, write @code{__typeof__} instead of @code{typeof}.
673 @xref{Alternate Keywords}.
675 A @code{typeof} construct can be used anywhere a typedef name can be
676 used. For example, you can use it in a declaration, in a cast, or inside
677 of @code{sizeof} or @code{typeof}.
679 The operand of @code{typeof} is evaluated for its side effects if and
680 only if it is an expression of variably modified type or the name of
683 @code{typeof} is often useful in conjunction with
684 statement expressions (@pxref{Statement Exprs}).
685 Here is how the two together can
686 be used to define a safe ``maximum'' macro which operates on any
687 arithmetic type and evaluates each of its arguments exactly once:
691 (@{ typeof (a) _a = (a); \
692 typeof (b) _b = (b); \
693 _a > _b ? _a : _b; @})
696 @cindex underscores in variables in macros
697 @cindex @samp{_} in variables in macros
698 @cindex local variables in macros
699 @cindex variables, local, in macros
700 @cindex macros, local variables in
702 The reason for using names that start with underscores for the local
703 variables is to avoid conflicts with variable names that occur within the
704 expressions that are substituted for @code{a} and @code{b}. Eventually we
705 hope to design a new form of declaration syntax that allows you to declare
706 variables whose scopes start only after their initializers; this will be a
707 more reliable way to prevent such conflicts.
710 Some more examples of the use of @code{typeof}:
714 This declares @code{y} with the type of what @code{x} points to.
721 This declares @code{y} as an array of such values.
728 This declares @code{y} as an array of pointers to characters:
731 typeof (typeof (char *)[4]) y;
735 It is equivalent to the following traditional C declaration:
741 To see the meaning of the declaration using @code{typeof}, and why it
742 might be a useful way to write, rewrite it with these macros:
745 #define pointer(T) typeof(T *)
746 #define array(T, N) typeof(T [N])
750 Now the declaration can be rewritten this way:
753 array (pointer (char), 4) y;
757 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
758 pointers to @code{char}.
761 In GNU C, but not GNU C++, you may also declare the type of a variable
762 as @code{__auto_type}. In that case, the declaration must declare
763 only one variable, whose declarator must just be an identifier, the
764 declaration must be initialized, and the type of the variable is
765 determined by the initializer; the name of the variable is not in
766 scope until after the initializer. (In C++, you should use C++11
767 @code{auto} for this purpose.) Using @code{__auto_type}, the
768 ``maximum'' macro above could be written as:
772 (@{ __auto_type _a = (a); \
773 __auto_type _b = (b); \
774 _a > _b ? _a : _b; @})
777 Using @code{__auto_type} instead of @code{typeof} has two advantages:
780 @item Each argument to the macro appears only once in the expansion of
781 the macro. This prevents the size of the macro expansion growing
782 exponentially when calls to such macros are nested inside arguments of
785 @item If the argument to the macro has variably modified type, it is
786 evaluated only once when using @code{__auto_type}, but twice if
787 @code{typeof} is used.
791 @section Conditionals with Omitted Operands
792 @cindex conditional expressions, extensions
793 @cindex omitted middle-operands
794 @cindex middle-operands, omitted
795 @cindex extensions, @code{?:}
796 @cindex @code{?:} extensions
798 The middle operand in a conditional expression may be omitted. Then
799 if the first operand is nonzero, its value is the value of the conditional
802 Therefore, the expression
809 has the value of @code{x} if that is nonzero; otherwise, the value of
812 This example is perfectly equivalent to
818 @cindex side effect in @code{?:}
819 @cindex @code{?:} side effect
821 In this simple case, the ability to omit the middle operand is not
822 especially useful. When it becomes useful is when the first operand does,
823 or may (if it is a macro argument), contain a side effect. Then repeating
824 the operand in the middle would perform the side effect twice. Omitting
825 the middle operand uses the value already computed without the undesirable
826 effects of recomputing it.
829 @section 128-bit Integers
830 @cindex @code{__int128} data types
832 As an extension the integer scalar type @code{__int128} is supported for
833 targets which have an integer mode wide enough to hold 128 bits.
834 Simply write @code{__int128} for a signed 128-bit integer, or
835 @code{unsigned __int128} for an unsigned 128-bit integer. There is no
836 support in GCC for expressing an integer constant of type @code{__int128}
837 for targets with @code{long long} integer less than 128 bits wide.
840 @section Double-Word Integers
841 @cindex @code{long long} data types
842 @cindex double-word arithmetic
843 @cindex multiprecision arithmetic
844 @cindex @code{LL} integer suffix
845 @cindex @code{ULL} integer suffix
847 ISO C99 supports data types for integers that are at least 64 bits wide,
848 and as an extension GCC supports them in C90 mode and in C++.
849 Simply write @code{long long int} for a signed integer, or
850 @code{unsigned long long int} for an unsigned integer. To make an
851 integer constant of type @code{long long int}, add the suffix @samp{LL}
852 to the integer. To make an integer constant of type @code{unsigned long
853 long int}, add the suffix @samp{ULL} to the integer.
855 You can use these types in arithmetic like any other integer types.
856 Addition, subtraction, and bitwise boolean operations on these types
857 are open-coded on all types of machines. Multiplication is open-coded
858 if the machine supports a fullword-to-doubleword widening multiply
859 instruction. Division and shifts are open-coded only on machines that
860 provide special support. The operations that are not open-coded use
861 special library routines that come with GCC@.
863 There may be pitfalls when you use @code{long long} types for function
864 arguments without function prototypes. If a function
865 expects type @code{int} for its argument, and you pass a value of type
866 @code{long long int}, confusion results because the caller and the
867 subroutine disagree about the number of bytes for the argument.
868 Likewise, if the function expects @code{long long int} and you pass
869 @code{int}. The best way to avoid such problems is to use prototypes.
872 @section Complex Numbers
873 @cindex complex numbers
874 @cindex @code{_Complex} keyword
875 @cindex @code{__complex__} keyword
877 ISO C99 supports complex floating data types, and as an extension GCC
878 supports them in C90 mode and in C++. GCC also supports complex integer data
879 types which are not part of ISO C99. You can declare complex types
880 using the keyword @code{_Complex}. As an extension, the older GNU
881 keyword @code{__complex__} is also supported.
883 For example, @samp{_Complex double x;} declares @code{x} as a
884 variable whose real part and imaginary part are both of type
885 @code{double}. @samp{_Complex short int y;} declares @code{y} to
886 have real and imaginary parts of type @code{short int}; this is not
887 likely to be useful, but it shows that the set of complex types is
890 To write a constant with a complex data type, use the suffix @samp{i} or
891 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
892 has type @code{_Complex float} and @code{3i} has type
893 @code{_Complex int}. Such a constant always has a pure imaginary
894 value, but you can form any complex value you like by adding one to a
895 real constant. This is a GNU extension; if you have an ISO C99
896 conforming C library (such as the GNU C Library), and want to construct complex
897 constants of floating type, you should include @code{<complex.h>} and
898 use the macros @code{I} or @code{_Complex_I} instead.
900 @cindex @code{__real__} keyword
901 @cindex @code{__imag__} keyword
902 To extract the real part of a complex-valued expression @var{exp}, write
903 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
904 extract the imaginary part. This is a GNU extension; for values of
905 floating type, you should use the ISO C99 functions @code{crealf},
906 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
907 @code{cimagl}, declared in @code{<complex.h>} and also provided as
908 built-in functions by GCC@.
910 @cindex complex conjugation
911 The operator @samp{~} performs complex conjugation when used on a value
912 with a complex type. This is a GNU extension; for values of
913 floating type, you should use the ISO C99 functions @code{conjf},
914 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
915 provided as built-in functions by GCC@.
917 GCC can allocate complex automatic variables in a noncontiguous
918 fashion; it's even possible for the real part to be in a register while
919 the imaginary part is on the stack (or vice versa). Only the DWARF 2
920 debug info format can represent this, so use of DWARF 2 is recommended.
921 If you are using the stabs debug info format, GCC describes a noncontiguous
922 complex variable as if it were two separate variables of noncomplex type.
923 If the variable's actual name is @code{foo}, the two fictitious
924 variables are named @code{foo$real} and @code{foo$imag}. You can
925 examine and set these two fictitious variables with your debugger.
928 @section Additional Floating Types
929 @cindex additional floating types
930 @cindex @code{__float80} data type
931 @cindex @code{__float128} data type
932 @cindex @code{__ibm128} data type
933 @cindex @code{w} floating point suffix
934 @cindex @code{q} floating point suffix
935 @cindex @code{W} floating point suffix
936 @cindex @code{Q} floating point suffix
938 As an extension, GNU C supports additional floating
939 types, @code{__float80} and @code{__float128} to support 80-bit
940 (@code{XFmode}) and 128-bit (@code{TFmode}) floating types.
941 Support for additional types includes the arithmetic operators:
942 add, subtract, multiply, divide; unary arithmetic operators;
943 relational operators; equality operators; and conversions to and from
944 integer and other floating types. Use a suffix @samp{w} or @samp{W}
945 in a literal constant of type @code{__float80} or type
946 @code{__ibm128}. Use a suffix @samp{q} or @samp{Q} for @code{_float128}.
948 On the i386, x86_64, IA-64, and HP-UX targets, you can declare complex
949 types using the corresponding internal complex type, @code{XCmode} for
950 @code{__float80} type and @code{TCmode} for @code{__float128} type:
953 typedef _Complex float __attribute__((mode(TC))) _Complex128;
954 typedef _Complex float __attribute__((mode(XC))) _Complex80;
957 On PowerPC Linux, Freebsd and Darwin systems, the default for
958 @code{long double} is to use the IBM extended floating point format
959 that uses a pair of @code{double} values to extend the precision.
960 This means that the mode @code{TCmode} was already used by the
961 traditional IBM long double format, and you would need to use the mode
965 typedef _Complex float __attribute__((mode(KC))) _Complex128;
968 Not all targets support additional floating-point types. @code{__float80}
969 and @code{__float128} types are supported on x86 and IA-64 targets.
970 The @code{__float128} type is supported on hppa HP-UX.
971 The @code{__float128} type is supported on PowerPC systems by default
972 if the vector scalar instruction set (VSX) is enabled.
974 On the PowerPC, @code{__ibm128} provides access to the IBM extended
975 double format, and it is intended to be used by the library functions
976 that handle conversions if/when long double is changed to be IEEE
977 128-bit floating point.
980 @section Half-Precision Floating Point
981 @cindex half-precision floating point
982 @cindex @code{__fp16} data type
984 On ARM targets, GCC supports half-precision (16-bit) floating point via
985 the @code{__fp16} type. You must enable this type explicitly
986 with the @option{-mfp16-format} command-line option in order to use it.
988 ARM supports two incompatible representations for half-precision
989 floating-point values. You must choose one of the representations and
990 use it consistently in your program.
992 Specifying @option{-mfp16-format=ieee} selects the IEEE 754-2008 format.
993 This format can represent normalized values in the range of @math{2^{-14}} to 65504.
994 There are 11 bits of significand precision, approximately 3
997 Specifying @option{-mfp16-format=alternative} selects the ARM
998 alternative format. This representation is similar to the IEEE
999 format, but does not support infinities or NaNs. Instead, the range
1000 of exponents is extended, so that this format can represent normalized
1001 values in the range of @math{2^{-14}} to 131008.
1003 The @code{__fp16} type is a storage format only. For purposes
1004 of arithmetic and other operations, @code{__fp16} values in C or C++
1005 expressions are automatically promoted to @code{float}. In addition,
1006 you cannot declare a function with a return value or parameters
1007 of type @code{__fp16}.
1009 Note that conversions from @code{double} to @code{__fp16}
1010 involve an intermediate conversion to @code{float}. Because
1011 of rounding, this can sometimes produce a different result than a
1014 ARM provides hardware support for conversions between
1015 @code{__fp16} and @code{float} values
1016 as an extension to VFP and NEON (Advanced SIMD). GCC generates
1017 code using these hardware instructions if you compile with
1018 options to select an FPU that provides them;
1019 for example, @option{-mfpu=neon-fp16 -mfloat-abi=softfp},
1020 in addition to the @option{-mfp16-format} option to select
1021 a half-precision format.
1023 Language-level support for the @code{__fp16} data type is
1024 independent of whether GCC generates code using hardware floating-point
1025 instructions. In cases where hardware support is not specified, GCC
1026 implements conversions between @code{__fp16} and @code{float} values
1030 @section Decimal Floating Types
1031 @cindex decimal floating types
1032 @cindex @code{_Decimal32} data type
1033 @cindex @code{_Decimal64} data type
1034 @cindex @code{_Decimal128} data type
1035 @cindex @code{df} integer suffix
1036 @cindex @code{dd} integer suffix
1037 @cindex @code{dl} integer suffix
1038 @cindex @code{DF} integer suffix
1039 @cindex @code{DD} integer suffix
1040 @cindex @code{DL} integer suffix
1042 As an extension, GNU C supports decimal floating types as
1043 defined in the N1312 draft of ISO/IEC WDTR24732. Support for decimal
1044 floating types in GCC will evolve as the draft technical report changes.
1045 Calling conventions for any target might also change. Not all targets
1046 support decimal floating types.
1048 The decimal floating types are @code{_Decimal32}, @code{_Decimal64}, and
1049 @code{_Decimal128}. They use a radix of ten, unlike the floating types
1050 @code{float}, @code{double}, and @code{long double} whose radix is not
1051 specified by the C standard but is usually two.
1053 Support for decimal floating types includes the arithmetic operators
1054 add, subtract, multiply, divide; unary arithmetic operators;
1055 relational operators; equality operators; and conversions to and from
1056 integer and other floating types. Use a suffix @samp{df} or
1057 @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd}
1058 or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for
1061 GCC support of decimal float as specified by the draft technical report
1066 When the value of a decimal floating type cannot be represented in the
1067 integer type to which it is being converted, the result is undefined
1068 rather than the result value specified by the draft technical report.
1071 GCC does not provide the C library functionality associated with
1072 @file{math.h}, @file{fenv.h}, @file{stdio.h}, @file{stdlib.h}, and
1073 @file{wchar.h}, which must come from a separate C library implementation.
1074 Because of this the GNU C compiler does not define macro
1075 @code{__STDC_DEC_FP__} to indicate that the implementation conforms to
1076 the technical report.
1079 Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128}
1080 are supported by the DWARF 2 debug information format.
1086 ISO C99 supports floating-point numbers written not only in the usual
1087 decimal notation, such as @code{1.55e1}, but also numbers such as
1088 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
1089 supports this in C90 mode (except in some cases when strictly
1090 conforming) and in C++. In that format the
1091 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
1092 mandatory. The exponent is a decimal number that indicates the power of
1093 2 by which the significant part is multiplied. Thus @samp{0x1.f} is
1100 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
1101 is the same as @code{1.55e1}.
1103 Unlike for floating-point numbers in the decimal notation the exponent
1104 is always required in the hexadecimal notation. Otherwise the compiler
1105 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
1106 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
1107 extension for floating-point constants of type @code{float}.
1110 @section Fixed-Point Types
1111 @cindex fixed-point types
1112 @cindex @code{_Fract} data type
1113 @cindex @code{_Accum} data type
1114 @cindex @code{_Sat} data type
1115 @cindex @code{hr} fixed-suffix
1116 @cindex @code{r} fixed-suffix
1117 @cindex @code{lr} fixed-suffix
1118 @cindex @code{llr} fixed-suffix
1119 @cindex @code{uhr} fixed-suffix
1120 @cindex @code{ur} fixed-suffix
1121 @cindex @code{ulr} fixed-suffix
1122 @cindex @code{ullr} fixed-suffix
1123 @cindex @code{hk} fixed-suffix
1124 @cindex @code{k} fixed-suffix
1125 @cindex @code{lk} fixed-suffix
1126 @cindex @code{llk} fixed-suffix
1127 @cindex @code{uhk} fixed-suffix
1128 @cindex @code{uk} fixed-suffix
1129 @cindex @code{ulk} fixed-suffix
1130 @cindex @code{ullk} fixed-suffix
1131 @cindex @code{HR} fixed-suffix
1132 @cindex @code{R} fixed-suffix
1133 @cindex @code{LR} fixed-suffix
1134 @cindex @code{LLR} fixed-suffix
1135 @cindex @code{UHR} fixed-suffix
1136 @cindex @code{UR} fixed-suffix
1137 @cindex @code{ULR} fixed-suffix
1138 @cindex @code{ULLR} fixed-suffix
1139 @cindex @code{HK} fixed-suffix
1140 @cindex @code{K} fixed-suffix
1141 @cindex @code{LK} fixed-suffix
1142 @cindex @code{LLK} fixed-suffix
1143 @cindex @code{UHK} fixed-suffix
1144 @cindex @code{UK} fixed-suffix
1145 @cindex @code{ULK} fixed-suffix
1146 @cindex @code{ULLK} fixed-suffix
1148 As an extension, GNU C supports fixed-point types as
1149 defined in the N1169 draft of ISO/IEC DTR 18037. Support for fixed-point
1150 types in GCC will evolve as the draft technical report changes.
1151 Calling conventions for any target might also change. Not all targets
1152 support fixed-point types.
1154 The fixed-point types are
1155 @code{short _Fract},
1158 @code{long long _Fract},
1159 @code{unsigned short _Fract},
1160 @code{unsigned _Fract},
1161 @code{unsigned long _Fract},
1162 @code{unsigned long long _Fract},
1163 @code{_Sat short _Fract},
1165 @code{_Sat long _Fract},
1166 @code{_Sat long long _Fract},
1167 @code{_Sat unsigned short _Fract},
1168 @code{_Sat unsigned _Fract},
1169 @code{_Sat unsigned long _Fract},
1170 @code{_Sat unsigned long long _Fract},
1171 @code{short _Accum},
1174 @code{long long _Accum},
1175 @code{unsigned short _Accum},
1176 @code{unsigned _Accum},
1177 @code{unsigned long _Accum},
1178 @code{unsigned long long _Accum},
1179 @code{_Sat short _Accum},
1181 @code{_Sat long _Accum},
1182 @code{_Sat long long _Accum},
1183 @code{_Sat unsigned short _Accum},
1184 @code{_Sat unsigned _Accum},
1185 @code{_Sat unsigned long _Accum},
1186 @code{_Sat unsigned long long _Accum}.
1188 Fixed-point data values contain fractional and optional integral parts.
1189 The format of fixed-point data varies and depends on the target machine.
1191 Support for fixed-point types includes:
1194 prefix and postfix increment and decrement operators (@code{++}, @code{--})
1196 unary arithmetic operators (@code{+}, @code{-}, @code{!})
1198 binary arithmetic operators (@code{+}, @code{-}, @code{*}, @code{/})
1200 binary shift operators (@code{<<}, @code{>>})
1202 relational operators (@code{<}, @code{<=}, @code{>=}, @code{>})
1204 equality operators (@code{==}, @code{!=})
1206 assignment operators (@code{+=}, @code{-=}, @code{*=}, @code{/=},
1207 @code{<<=}, @code{>>=})
1209 conversions to and from integer, floating-point, or fixed-point types
1212 Use a suffix in a fixed-point literal constant:
1214 @item @samp{hr} or @samp{HR} for @code{short _Fract} and
1215 @code{_Sat short _Fract}
1216 @item @samp{r} or @samp{R} for @code{_Fract} and @code{_Sat _Fract}
1217 @item @samp{lr} or @samp{LR} for @code{long _Fract} and
1218 @code{_Sat long _Fract}
1219 @item @samp{llr} or @samp{LLR} for @code{long long _Fract} and
1220 @code{_Sat long long _Fract}
1221 @item @samp{uhr} or @samp{UHR} for @code{unsigned short _Fract} and
1222 @code{_Sat unsigned short _Fract}
1223 @item @samp{ur} or @samp{UR} for @code{unsigned _Fract} and
1224 @code{_Sat unsigned _Fract}
1225 @item @samp{ulr} or @samp{ULR} for @code{unsigned long _Fract} and
1226 @code{_Sat unsigned long _Fract}
1227 @item @samp{ullr} or @samp{ULLR} for @code{unsigned long long _Fract}
1228 and @code{_Sat unsigned long long _Fract}
1229 @item @samp{hk} or @samp{HK} for @code{short _Accum} and
1230 @code{_Sat short _Accum}
1231 @item @samp{k} or @samp{K} for @code{_Accum} and @code{_Sat _Accum}
1232 @item @samp{lk} or @samp{LK} for @code{long _Accum} and
1233 @code{_Sat long _Accum}
1234 @item @samp{llk} or @samp{LLK} for @code{long long _Accum} and
1235 @code{_Sat long long _Accum}
1236 @item @samp{uhk} or @samp{UHK} for @code{unsigned short _Accum} and
1237 @code{_Sat unsigned short _Accum}
1238 @item @samp{uk} or @samp{UK} for @code{unsigned _Accum} and
1239 @code{_Sat unsigned _Accum}
1240 @item @samp{ulk} or @samp{ULK} for @code{unsigned long _Accum} and
1241 @code{_Sat unsigned long _Accum}
1242 @item @samp{ullk} or @samp{ULLK} for @code{unsigned long long _Accum}
1243 and @code{_Sat unsigned long long _Accum}
1246 GCC support of fixed-point types as specified by the draft technical report
1251 Pragmas to control overflow and rounding behaviors are not implemented.
1254 Fixed-point types are supported by the DWARF 2 debug information format.
1256 @node Named Address Spaces
1257 @section Named Address Spaces
1258 @cindex Named Address Spaces
1260 As an extension, GNU C supports named address spaces as
1261 defined in the N1275 draft of ISO/IEC DTR 18037. Support for named
1262 address spaces in GCC will evolve as the draft technical report
1263 changes. Calling conventions for any target might also change. At
1264 present, only the AVR, SPU, M32C, RL78, and x86 targets support
1265 address spaces other than the generic address space.
1267 Address space identifiers may be used exactly like any other C type
1268 qualifier (e.g., @code{const} or @code{volatile}). See the N1275
1269 document for more details.
1271 @anchor{AVR Named Address Spaces}
1272 @subsection AVR Named Address Spaces
1274 On the AVR target, there are several address spaces that can be used
1275 in order to put read-only data into the flash memory and access that
1276 data by means of the special instructions @code{LPM} or @code{ELPM}
1277 needed to read from flash.
1279 Per default, any data including read-only data is located in RAM
1280 (the generic address space) so that non-generic address spaces are
1281 needed to locate read-only data in flash memory
1282 @emph{and} to generate the right instructions to access this data
1283 without using (inline) assembler code.
1287 @cindex @code{__flash} AVR Named Address Spaces
1288 The @code{__flash} qualifier locates data in the
1289 @code{.progmem.data} section. Data is read using the @code{LPM}
1290 instruction. Pointers to this address space are 16 bits wide.
1297 @cindex @code{__flash1} AVR Named Address Spaces
1298 @cindex @code{__flash2} AVR Named Address Spaces
1299 @cindex @code{__flash3} AVR Named Address Spaces
1300 @cindex @code{__flash4} AVR Named Address Spaces
1301 @cindex @code{__flash5} AVR Named Address Spaces
1302 These are 16-bit address spaces locating data in section
1303 @code{.progmem@var{N}.data} where @var{N} refers to
1304 address space @code{__flash@var{N}}.
1305 The compiler sets the @code{RAMPZ} segment register appropriately
1306 before reading data by means of the @code{ELPM} instruction.
1309 @cindex @code{__memx} AVR Named Address Spaces
1310 This is a 24-bit address space that linearizes flash and RAM:
1311 If the high bit of the address is set, data is read from
1312 RAM using the lower two bytes as RAM address.
1313 If the high bit of the address is clear, data is read from flash
1314 with @code{RAMPZ} set according to the high byte of the address.
1315 @xref{AVR Built-in Functions,,@code{__builtin_avr_flash_segment}}.
1317 Objects in this address space are located in @code{.progmemx.data}.
1323 char my_read (const __flash char ** p)
1325 /* p is a pointer to RAM that points to a pointer to flash.
1326 The first indirection of p reads that flash pointer
1327 from RAM and the second indirection reads a char from this
1333 /* Locate array[] in flash memory */
1334 const __flash int array[] = @{ 3, 5, 7, 11, 13, 17, 19 @};
1340 /* Return 17 by reading from flash memory */
1341 return array[array[i]];
1346 For each named address space supported by avr-gcc there is an equally
1347 named but uppercase built-in macro defined.
1348 The purpose is to facilitate testing if respective address space
1349 support is available or not:
1353 const __flash int var = 1;
1360 #include <avr/pgmspace.h> /* From AVR-LibC */
1362 const int var PROGMEM = 1;
1366 return (int) pgm_read_word (&var);
1368 #endif /* __FLASH */
1372 Notice that attribute @ref{AVR Variable Attributes,,@code{progmem}}
1373 locates data in flash but
1374 accesses to these data read from generic address space, i.e.@:
1376 so that you need special accessors like @code{pgm_read_byte}
1377 from @w{@uref{http://nongnu.org/avr-libc/user-manual/,AVR-LibC}}
1378 together with attribute @code{progmem}.
1381 @b{Limitations and caveats}
1385 Reading across the 64@tie{}KiB section boundary of
1386 the @code{__flash} or @code{__flash@var{N}} address spaces
1387 shows undefined behavior. The only address space that
1388 supports reading across the 64@tie{}KiB flash segment boundaries is
1392 If you use one of the @code{__flash@var{N}} address spaces
1393 you must arrange your linker script to locate the
1394 @code{.progmem@var{N}.data} sections according to your needs.
1397 Any data or pointers to the non-generic address spaces must
1398 be qualified as @code{const}, i.e.@: as read-only data.
1399 This still applies if the data in one of these address
1400 spaces like software version number or calibration lookup table are intended to
1401 be changed after load time by, say, a boot loader. In this case
1402 the right qualification is @code{const} @code{volatile} so that the compiler
1403 must not optimize away known values or insert them
1404 as immediates into operands of instructions.
1407 The following code initializes a variable @code{pfoo}
1408 located in static storage with a 24-bit address:
1410 extern const __memx char foo;
1411 const __memx void *pfoo = &foo;
1415 Such code requires at least binutils 2.23, see
1416 @w{@uref{http://sourceware.org/PR13503,PR13503}}.
1420 @subsection M32C Named Address Spaces
1421 @cindex @code{__far} M32C Named Address Spaces
1423 On the M32C target, with the R8C and M16C CPU variants, variables
1424 qualified with @code{__far} are accessed using 32-bit addresses in
1425 order to access memory beyond the first 64@tie{}Ki bytes. If
1426 @code{__far} is used with the M32CM or M32C CPU variants, it has no
1429 @subsection RL78 Named Address Spaces
1430 @cindex @code{__far} RL78 Named Address Spaces
1432 On the RL78 target, variables qualified with @code{__far} are accessed
1433 with 32-bit pointers (20-bit addresses) rather than the default 16-bit
1434 addresses. Non-far variables are assumed to appear in the topmost
1435 64@tie{}KiB of the address space.
1437 @subsection SPU Named Address Spaces
1438 @cindex @code{__ea} SPU Named Address Spaces
1440 On the SPU target variables may be declared as
1441 belonging to another address space by qualifying the type with the
1442 @code{__ea} address space identifier:
1449 The compiler generates special code to access the variable @code{i}.
1450 It may use runtime library
1451 support, or generate special machine instructions to access that address
1454 @subsection x86 Named Address Spaces
1455 @cindex x86 named address spaces
1457 On the x86 target, variables may be declared as being relative
1458 to the @code{%fs} or @code{%gs} segments.
1463 @cindex @code{__seg_fs} x86 named address space
1464 @cindex @code{__seg_gs} x86 named address space
1465 The object is accessed with the respective segment override prefix.
1467 The respective segment base must be set via some method specific to
1468 the operating system. Rather than require an expensive system call
1469 to retrieve the segment base, these address spaces are not considered
1470 to be subspaces of the generic (flat) address space. This means that
1471 explicit casts are required to convert pointers between these address
1472 spaces and the generic address space. In practice the application
1473 should cast to @code{uintptr_t} and apply the segment base offset
1474 that it installed previously.
1476 The preprocessor symbols @code{__SEG_FS} and @code{__SEG_GS} are
1477 defined when these address spaces are supported.
1480 @cindex @code{__seg_tls} x86 named address space
1481 Some operating systems define either the @code{%fs} or @code{%gs}
1482 segment as the thread-local storage base for each thread. Objects
1483 within this address space are accessed with the appropriate
1484 segment override prefix.
1486 The pointer located at address 0 within the segment contains the
1487 offset of the segment within the generic address space. Thus this
1488 address space is considered a subspace of the generic address space,
1489 and the known segment offset is applied when converting addresses
1490 to and from the generic address space.
1492 The preprocessor symbol @code{__SEG_TLS} is defined when this
1493 address space is supported.
1498 @section Arrays of Length Zero
1499 @cindex arrays of length zero
1500 @cindex zero-length arrays
1501 @cindex length-zero arrays
1502 @cindex flexible array members
1504 Zero-length arrays are allowed in GNU C@. They are very useful as the
1505 last element of a structure that is really a header for a variable-length
1514 struct line *thisline = (struct line *)
1515 malloc (sizeof (struct line) + this_length);
1516 thisline->length = this_length;
1519 In ISO C90, you would have to give @code{contents} a length of 1, which
1520 means either you waste space or complicate the argument to @code{malloc}.
1522 In ISO C99, you would use a @dfn{flexible array member}, which is
1523 slightly different in syntax and semantics:
1527 Flexible array members are written as @code{contents[]} without
1531 Flexible array members have incomplete type, and so the @code{sizeof}
1532 operator may not be applied. As a quirk of the original implementation
1533 of zero-length arrays, @code{sizeof} evaluates to zero.
1536 Flexible array members may only appear as the last member of a
1537 @code{struct} that is otherwise non-empty.
1540 A structure containing a flexible array member, or a union containing
1541 such a structure (possibly recursively), may not be a member of a
1542 structure or an element of an array. (However, these uses are
1543 permitted by GCC as extensions.)
1546 Non-empty initialization of zero-length
1547 arrays is treated like any case where there are more initializer
1548 elements than the array holds, in that a suitable warning about ``excess
1549 elements in array'' is given, and the excess elements (all of them, in
1550 this case) are ignored.
1552 GCC allows static initialization of flexible array members.
1553 This is equivalent to defining a new structure containing the original
1554 structure followed by an array of sufficient size to contain the data.
1555 E.g.@: in the following, @code{f1} is constructed as if it were declared
1561 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
1564 struct f1 f1; int data[3];
1565 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1569 The convenience of this extension is that @code{f1} has the desired
1570 type, eliminating the need to consistently refer to @code{f2.f1}.
1572 This has symmetry with normal static arrays, in that an array of
1573 unknown size is also written with @code{[]}.
1575 Of course, this extension only makes sense if the extra data comes at
1576 the end of a top-level object, as otherwise we would be overwriting
1577 data at subsequent offsets. To avoid undue complication and confusion
1578 with initialization of deeply nested arrays, we simply disallow any
1579 non-empty initialization except when the structure is the top-level
1580 object. For example:
1583 struct foo @{ int x; int y[]; @};
1584 struct bar @{ struct foo z; @};
1586 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1587 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1588 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1589 struct foo d[1] = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1592 @node Empty Structures
1593 @section Structures with No Members
1594 @cindex empty structures
1595 @cindex zero-size structures
1597 GCC permits a C structure to have no members:
1604 The structure has size zero. In C++, empty structures are part
1605 of the language. G++ treats empty structures as if they had a single
1606 member of type @code{char}.
1608 @node Variable Length
1609 @section Arrays of Variable Length
1610 @cindex variable-length arrays
1611 @cindex arrays of variable length
1614 Variable-length automatic arrays are allowed in ISO C99, and as an
1615 extension GCC accepts them in C90 mode and in C++. These arrays are
1616 declared like any other automatic arrays, but with a length that is not
1617 a constant expression. The storage is allocated at the point of
1618 declaration and deallocated when the block scope containing the declaration
1624 concat_fopen (char *s1, char *s2, char *mode)
1626 char str[strlen (s1) + strlen (s2) + 1];
1629 return fopen (str, mode);
1633 @cindex scope of a variable length array
1634 @cindex variable-length array scope
1635 @cindex deallocating variable length arrays
1636 Jumping or breaking out of the scope of the array name deallocates the
1637 storage. Jumping into the scope is not allowed; you get an error
1640 @cindex variable-length array in a structure
1641 As an extension, GCC accepts variable-length arrays as a member of
1642 a structure or a union. For example:
1648 struct S @{ int x[n]; @};
1652 @cindex @code{alloca} vs variable-length arrays
1653 You can use the function @code{alloca} to get an effect much like
1654 variable-length arrays. The function @code{alloca} is available in
1655 many other C implementations (but not in all). On the other hand,
1656 variable-length arrays are more elegant.
1658 There are other differences between these two methods. Space allocated
1659 with @code{alloca} exists until the containing @emph{function} returns.
1660 The space for a variable-length array is deallocated as soon as the array
1661 name's scope ends. (If you use both variable-length arrays and
1662 @code{alloca} in the same function, deallocation of a variable-length array
1663 also deallocates anything more recently allocated with @code{alloca}.)
1665 You can also use variable-length arrays as arguments to functions:
1669 tester (int len, char data[len][len])
1675 The length of an array is computed once when the storage is allocated
1676 and is remembered for the scope of the array in case you access it with
1679 If you want to pass the array first and the length afterward, you can
1680 use a forward declaration in the parameter list---another GNU extension.
1684 tester (int len; char data[len][len], int len)
1690 @cindex parameter forward declaration
1691 The @samp{int len} before the semicolon is a @dfn{parameter forward
1692 declaration}, and it serves the purpose of making the name @code{len}
1693 known when the declaration of @code{data} is parsed.
1695 You can write any number of such parameter forward declarations in the
1696 parameter list. They can be separated by commas or semicolons, but the
1697 last one must end with a semicolon, which is followed by the ``real''
1698 parameter declarations. Each forward declaration must match a ``real''
1699 declaration in parameter name and data type. ISO C99 does not support
1700 parameter forward declarations.
1702 @node Variadic Macros
1703 @section Macros with a Variable Number of Arguments.
1704 @cindex variable number of arguments
1705 @cindex macro with variable arguments
1706 @cindex rest argument (in macro)
1707 @cindex variadic macros
1709 In the ISO C standard of 1999, a macro can be declared to accept a
1710 variable number of arguments much as a function can. The syntax for
1711 defining the macro is similar to that of a function. Here is an
1715 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1719 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1720 such a macro, it represents the zero or more tokens until the closing
1721 parenthesis that ends the invocation, including any commas. This set of
1722 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1723 wherever it appears. See the CPP manual for more information.
1725 GCC has long supported variadic macros, and used a different syntax that
1726 allowed you to give a name to the variable arguments just like any other
1727 argument. Here is an example:
1730 #define debug(format, args...) fprintf (stderr, format, args)
1734 This is in all ways equivalent to the ISO C example above, but arguably
1735 more readable and descriptive.
1737 GNU CPP has two further variadic macro extensions, and permits them to
1738 be used with either of the above forms of macro definition.
1740 In standard C, you are not allowed to leave the variable argument out
1741 entirely; but you are allowed to pass an empty argument. For example,
1742 this invocation is invalid in ISO C, because there is no comma after
1749 GNU CPP permits you to completely omit the variable arguments in this
1750 way. In the above examples, the compiler would complain, though since
1751 the expansion of the macro still has the extra comma after the format
1754 To help solve this problem, CPP behaves specially for variable arguments
1755 used with the token paste operator, @samp{##}. If instead you write
1758 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1762 and if the variable arguments are omitted or empty, the @samp{##}
1763 operator causes the preprocessor to remove the comma before it. If you
1764 do provide some variable arguments in your macro invocation, GNU CPP
1765 does not complain about the paste operation and instead places the
1766 variable arguments after the comma. Just like any other pasted macro
1767 argument, these arguments are not macro expanded.
1769 @node Escaped Newlines
1770 @section Slightly Looser Rules for Escaped Newlines
1771 @cindex escaped newlines
1772 @cindex newlines (escaped)
1774 The preprocessor treatment of escaped newlines is more relaxed
1775 than that specified by the C90 standard, which requires the newline
1776 to immediately follow a backslash.
1777 GCC's implementation allows whitespace in the form
1778 of spaces, horizontal and vertical tabs, and form feeds between the
1779 backslash and the subsequent newline. The preprocessor issues a
1780 warning, but treats it as a valid escaped newline and combines the two
1781 lines to form a single logical line. This works within comments and
1782 tokens, as well as between tokens. Comments are @emph{not} treated as
1783 whitespace for the purposes of this relaxation, since they have not
1784 yet been replaced with spaces.
1787 @section Non-Lvalue Arrays May Have Subscripts
1788 @cindex subscripting
1789 @cindex arrays, non-lvalue
1791 @cindex subscripting and function values
1792 In ISO C99, arrays that are not lvalues still decay to pointers, and
1793 may be subscripted, although they may not be modified or used after
1794 the next sequence point and the unary @samp{&} operator may not be
1795 applied to them. As an extension, GNU C allows such arrays to be
1796 subscripted in C90 mode, though otherwise they do not decay to
1797 pointers outside C99 mode. For example,
1798 this is valid in GNU C though not valid in C90:
1802 struct foo @{int a[4];@};
1808 return f().a[index];
1814 @section Arithmetic on @code{void}- and Function-Pointers
1815 @cindex void pointers, arithmetic
1816 @cindex void, size of pointer to
1817 @cindex function pointers, arithmetic
1818 @cindex function, size of pointer to
1820 In GNU C, addition and subtraction operations are supported on pointers to
1821 @code{void} and on pointers to functions. This is done by treating the
1822 size of a @code{void} or of a function as 1.
1824 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1825 and on function types, and returns 1.
1827 @opindex Wpointer-arith
1828 The option @option{-Wpointer-arith} requests a warning if these extensions
1831 @node Pointers to Arrays
1832 @section Pointers to Arrays with Qualifiers Work as Expected
1833 @cindex pointers to arrays
1834 @cindex const qualifier
1836 In GNU C, pointers to arrays with qualifiers work similar to pointers
1837 to other qualified types. For example, a value of type @code{int (*)[5]}
1838 can be used to initialize a variable of type @code{const int (*)[5]}.
1839 These types are incompatible in ISO C because the @code{const} qualifier
1840 is formally attached to the element type of the array and not the
1845 transpose (int N, int M, double out[M][N], const double in[N][M]);
1849 transpose(3, 2, y, x);
1853 @section Non-Constant Initializers
1854 @cindex initializers, non-constant
1855 @cindex non-constant initializers
1857 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1858 automatic variable are not required to be constant expressions in GNU C@.
1859 Here is an example of an initializer with run-time varying elements:
1862 foo (float f, float g)
1864 float beat_freqs[2] = @{ f-g, f+g @};
1869 @node Compound Literals
1870 @section Compound Literals
1871 @cindex constructor expressions
1872 @cindex initializations in expressions
1873 @cindex structures, constructor expression
1874 @cindex expressions, constructor
1875 @cindex compound literals
1876 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1878 ISO C99 supports compound literals. A compound literal looks like
1879 a cast containing an initializer. Its value is an object of the
1880 type specified in the cast, containing the elements specified in
1881 the initializer; it is an lvalue. As an extension, GCC supports
1882 compound literals in C90 mode and in C++, though the semantics are
1883 somewhat different in C++.
1885 Usually, the specified type is a structure. Assume that
1886 @code{struct foo} and @code{structure} are declared as shown:
1889 struct foo @{int a; char b[2];@} structure;
1893 Here is an example of constructing a @code{struct foo} with a compound literal:
1896 structure = ((struct foo) @{x + y, 'a', 0@});
1900 This is equivalent to writing the following:
1904 struct foo temp = @{x + y, 'a', 0@};
1909 You can also construct an array, though this is dangerous in C++, as
1910 explained below. If all the elements of the compound literal are
1911 (made up of) simple constant expressions, suitable for use in
1912 initializers of objects of static storage duration, then the compound
1913 literal can be coerced to a pointer to its first element and used in
1914 such an initializer, as shown here:
1917 char **foo = (char *[]) @{ "x", "y", "z" @};
1920 Compound literals for scalar types and union types are
1921 also allowed, but then the compound literal is equivalent
1924 As a GNU extension, GCC allows initialization of objects with static storage
1925 duration by compound literals (which is not possible in ISO C99, because
1926 the initializer is not a constant).
1927 It is handled as if the object is initialized only with the bracket
1928 enclosed list if the types of the compound literal and the object match.
1929 The initializer list of the compound literal must be constant.
1930 If the object being initialized has array type of unknown size, the size is
1931 determined by compound literal size.
1934 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1935 static int y[] = (int []) @{1, 2, 3@};
1936 static int z[] = (int [3]) @{1@};
1940 The above lines are equivalent to the following:
1942 static struct foo x = @{1, 'a', 'b'@};
1943 static int y[] = @{1, 2, 3@};
1944 static int z[] = @{1, 0, 0@};
1947 In C, a compound literal designates an unnamed object with static or
1948 automatic storage duration. In C++, a compound literal designates a
1949 temporary object, which only lives until the end of its
1950 full-expression. As a result, well-defined C code that takes the
1951 address of a subobject of a compound literal can be undefined in C++,
1952 so the C++ compiler rejects the conversion of a temporary array to a pointer.
1953 For instance, if the array compound literal example above appeared
1954 inside a function, any subsequent use of @samp{foo} in C++ has
1955 undefined behavior because the lifetime of the array ends after the
1956 declaration of @samp{foo}.
1958 As an optimization, the C++ compiler sometimes gives array compound
1959 literals longer lifetimes: when the array either appears outside a
1960 function or has const-qualified type. If @samp{foo} and its
1961 initializer had elements of @samp{char *const} type rather than
1962 @samp{char *}, or if @samp{foo} were a global variable, the array
1963 would have static storage duration. But it is probably safest just to
1964 avoid the use of array compound literals in code compiled as C++.
1966 @node Designated Inits
1967 @section Designated Initializers
1968 @cindex initializers with labeled elements
1969 @cindex labeled elements in initializers
1970 @cindex case labels in initializers
1971 @cindex designated initializers
1973 Standard C90 requires the elements of an initializer to appear in a fixed
1974 order, the same as the order of the elements in the array or structure
1977 In ISO C99 you can give the elements in any order, specifying the array
1978 indices or structure field names they apply to, and GNU C allows this as
1979 an extension in C90 mode as well. This extension is not
1980 implemented in GNU C++.
1982 To specify an array index, write
1983 @samp{[@var{index}] =} before the element value. For example,
1986 int a[6] = @{ [4] = 29, [2] = 15 @};
1993 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1997 The index values must be constant expressions, even if the array being
1998 initialized is automatic.
2000 An alternative syntax for this that has been obsolete since GCC 2.5 but
2001 GCC still accepts is to write @samp{[@var{index}]} before the element
2002 value, with no @samp{=}.
2004 To initialize a range of elements to the same value, write
2005 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
2006 extension. For example,
2009 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
2013 If the value in it has side-effects, the side-effects happen only once,
2014 not for each initialized field by the range initializer.
2017 Note that the length of the array is the highest value specified
2020 In a structure initializer, specify the name of a field to initialize
2021 with @samp{.@var{fieldname} =} before the element value. For example,
2022 given the following structure,
2025 struct point @{ int x, y; @};
2029 the following initialization
2032 struct point p = @{ .y = yvalue, .x = xvalue @};
2039 struct point p = @{ xvalue, yvalue @};
2042 Another syntax that has the same meaning, obsolete since GCC 2.5, is
2043 @samp{@var{fieldname}:}, as shown here:
2046 struct point p = @{ y: yvalue, x: xvalue @};
2049 Omitted field members are implicitly initialized the same as objects
2050 that have static storage duration.
2053 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
2054 @dfn{designator}. You can also use a designator (or the obsolete colon
2055 syntax) when initializing a union, to specify which element of the union
2056 should be used. For example,
2059 union foo @{ int i; double d; @};
2061 union foo f = @{ .d = 4 @};
2065 converts 4 to a @code{double} to store it in the union using
2066 the second element. By contrast, casting 4 to type @code{union foo}
2067 stores it into the union as the integer @code{i}, since it is
2068 an integer. (@xref{Cast to Union}.)
2070 You can combine this technique of naming elements with ordinary C
2071 initialization of successive elements. Each initializer element that
2072 does not have a designator applies to the next consecutive element of the
2073 array or structure. For example,
2076 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
2083 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
2086 Labeling the elements of an array initializer is especially useful
2087 when the indices are characters or belong to an @code{enum} type.
2092 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
2093 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
2096 @cindex designator lists
2097 You can also write a series of @samp{.@var{fieldname}} and
2098 @samp{[@var{index}]} designators before an @samp{=} to specify a
2099 nested subobject to initialize; the list is taken relative to the
2100 subobject corresponding to the closest surrounding brace pair. For
2101 example, with the @samp{struct point} declaration above:
2104 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
2108 If the same field is initialized multiple times, it has the value from
2109 the last initialization. If any such overridden initialization has
2110 side-effect, it is unspecified whether the side-effect happens or not.
2111 Currently, GCC discards them and issues a warning.
2114 @section Case Ranges
2116 @cindex ranges in case statements
2118 You can specify a range of consecutive values in a single @code{case} label,
2122 case @var{low} ... @var{high}:
2126 This has the same effect as the proper number of individual @code{case}
2127 labels, one for each integer value from @var{low} to @var{high}, inclusive.
2129 This feature is especially useful for ranges of ASCII character codes:
2135 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
2136 it may be parsed wrong when you use it with integer values. For example,
2151 @section Cast to a Union Type
2152 @cindex cast to a union
2153 @cindex union, casting to a
2155 A cast to union type is similar to other casts, except that the type
2156 specified is a union type. You can specify the type either with
2157 @code{union @var{tag}} or with a typedef name. A cast to union is actually
2158 a constructor, not a cast, and hence does not yield an lvalue like
2159 normal casts. (@xref{Compound Literals}.)
2161 The types that may be cast to the union type are those of the members
2162 of the union. Thus, given the following union and variables:
2165 union foo @{ int i; double d; @};
2171 both @code{x} and @code{y} can be cast to type @code{union foo}.
2173 Using the cast as the right-hand side of an assignment to a variable of
2174 union type is equivalent to storing in a member of the union:
2179 u = (union foo) x @equiv{} u.i = x
2180 u = (union foo) y @equiv{} u.d = y
2183 You can also use the union cast as a function argument:
2186 void hack (union foo);
2188 hack ((union foo) x);
2191 @node Mixed Declarations
2192 @section Mixed Declarations and Code
2193 @cindex mixed declarations and code
2194 @cindex declarations, mixed with code
2195 @cindex code, mixed with declarations
2197 ISO C99 and ISO C++ allow declarations and code to be freely mixed
2198 within compound statements. As an extension, GNU C also allows this in
2199 C90 mode. For example, you could do:
2208 Each identifier is visible from where it is declared until the end of
2209 the enclosing block.
2211 @node Function Attributes
2212 @section Declaring Attributes of Functions
2213 @cindex function attributes
2214 @cindex declaring attributes of functions
2215 @cindex @code{volatile} applied to function
2216 @cindex @code{const} applied to function
2218 In GNU C, you can use function attributes to declare certain things
2219 about functions called in your program which help the compiler
2220 optimize calls and check your code more carefully. For example, you
2221 can use attributes to declare that a function never returns
2222 (@code{noreturn}), returns a value depending only on its arguments
2223 (@code{pure}), or has @code{printf}-style arguments (@code{format}).
2225 You can also use attributes to control memory placement, code
2226 generation options or call/return conventions within the function
2227 being annotated. Many of these attributes are target-specific. For
2228 example, many targets support attributes for defining interrupt
2229 handler functions, which typically must follow special register usage
2230 and return conventions.
2232 Function attributes are introduced by the @code{__attribute__} keyword
2233 on a declaration, followed by an attribute specification inside double
2234 parentheses. You can specify multiple attributes in a declaration by
2235 separating them by commas within the double parentheses or by
2236 immediately following an attribute declaration with another attribute
2237 declaration. @xref{Attribute Syntax}, for the exact rules on
2238 attribute syntax and placement.
2240 GCC also supports attributes on
2241 variable declarations (@pxref{Variable Attributes}),
2242 labels (@pxref{Label Attributes}),
2243 enumerators (@pxref{Enumerator Attributes}),
2244 and types (@pxref{Type Attributes}).
2246 There is some overlap between the purposes of attributes and pragmas
2247 (@pxref{Pragmas,,Pragmas Accepted by GCC}). It has been
2248 found convenient to use @code{__attribute__} to achieve a natural
2249 attachment of attributes to their corresponding declarations, whereas
2250 @code{#pragma} is of use for compatibility with other compilers
2251 or constructs that do not naturally form part of the grammar.
2253 In addition to the attributes documented here,
2254 GCC plugins may provide their own attributes.
2257 * Common Function Attributes::
2258 * AArch64 Function Attributes::
2259 * ARC Function Attributes::
2260 * ARM Function Attributes::
2261 * AVR Function Attributes::
2262 * Blackfin Function Attributes::
2263 * CR16 Function Attributes::
2264 * Epiphany Function Attributes::
2265 * H8/300 Function Attributes::
2266 * IA-64 Function Attributes::
2267 * M32C Function Attributes::
2268 * M32R/D Function Attributes::
2269 * m68k Function Attributes::
2270 * MCORE Function Attributes::
2271 * MeP Function Attributes::
2272 * MicroBlaze Function Attributes::
2273 * Microsoft Windows Function Attributes::
2274 * MIPS Function Attributes::
2275 * MSP430 Function Attributes::
2276 * NDS32 Function Attributes::
2277 * Nios II Function Attributes::
2278 * PowerPC Function Attributes::
2279 * RL78 Function Attributes::
2280 * RX Function Attributes::
2281 * S/390 Function Attributes::
2282 * SH Function Attributes::
2283 * SPU Function Attributes::
2284 * Symbian OS Function Attributes::
2285 * Visium Function Attributes::
2286 * x86 Function Attributes::
2287 * Xstormy16 Function Attributes::
2290 @node Common Function Attributes
2291 @subsection Common Function Attributes
2293 The following attributes are supported on most targets.
2296 @c Keep this table alphabetized by attribute name. Treat _ as space.
2298 @item alias ("@var{target}")
2299 @cindex @code{alias} function attribute
2300 The @code{alias} attribute causes the declaration to be emitted as an
2301 alias for another symbol, which must be specified. For instance,
2304 void __f () @{ /* @r{Do something.} */; @}
2305 void f () __attribute__ ((weak, alias ("__f")));
2309 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
2310 mangled name for the target must be used. It is an error if @samp{__f}
2311 is not defined in the same translation unit.
2313 This attribute requires assembler and object file support,
2314 and may not be available on all targets.
2316 @item aligned (@var{alignment})
2317 @cindex @code{aligned} function attribute
2318 This attribute specifies a minimum alignment for the function,
2321 You cannot use this attribute to decrease the alignment of a function,
2322 only to increase it. However, when you explicitly specify a function
2323 alignment this overrides the effect of the
2324 @option{-falign-functions} (@pxref{Optimize Options}) option for this
2327 Note that the effectiveness of @code{aligned} attributes may be
2328 limited by inherent limitations in your linker. On many systems, the
2329 linker is only able to arrange for functions to be aligned up to a
2330 certain maximum alignment. (For some linkers, the maximum supported
2331 alignment may be very very small.) See your linker documentation for
2332 further information.
2334 The @code{aligned} attribute can also be used for variables and fields
2335 (@pxref{Variable Attributes}.)
2338 @cindex @code{alloc_align} function attribute
2339 The @code{alloc_align} attribute is used to tell the compiler that the
2340 function return value points to memory, where the returned pointer minimum
2341 alignment is given by one of the functions parameters. GCC uses this
2342 information to improve pointer alignment analysis.
2344 The function parameter denoting the allocated alignment is specified by
2345 one integer argument, whose number is the argument of the attribute.
2346 Argument numbering starts at one.
2351 void* my_memalign(size_t, size_t) __attribute__((alloc_align(1)))
2355 declares that @code{my_memalign} returns memory with minimum alignment
2356 given by parameter 1.
2359 @cindex @code{alloc_size} function attribute
2360 The @code{alloc_size} attribute is used to tell the compiler that the
2361 function return value points to memory, where the size is given by
2362 one or two of the functions parameters. GCC uses this
2363 information to improve the correctness of @code{__builtin_object_size}.
2365 The function parameter(s) denoting the allocated size are specified by
2366 one or two integer arguments supplied to the attribute. The allocated size
2367 is either the value of the single function argument specified or the product
2368 of the two function arguments specified. Argument numbering starts at
2374 void* my_calloc(size_t, size_t) __attribute__((alloc_size(1,2)))
2375 void* my_realloc(void*, size_t) __attribute__((alloc_size(2)))
2379 declares that @code{my_calloc} returns memory of the size given by
2380 the product of parameter 1 and 2 and that @code{my_realloc} returns memory
2381 of the size given by parameter 2.
2384 @cindex @code{always_inline} function attribute
2385 Generally, functions are not inlined unless optimization is specified.
2386 For functions declared inline, this attribute inlines the function
2387 independent of any restrictions that otherwise apply to inlining.
2388 Failure to inline such a function is diagnosed as an error.
2389 Note that if such a function is called indirectly the compiler may
2390 or may not inline it depending on optimization level and a failure
2391 to inline an indirect call may or may not be diagnosed.
2394 @cindex @code{artificial} function attribute
2395 This attribute is useful for small inline wrappers that if possible
2396 should appear during debugging as a unit. Depending on the debug
2397 info format it either means marking the function as artificial
2398 or using the caller location for all instructions within the inlined
2401 @item assume_aligned
2402 @cindex @code{assume_aligned} function attribute
2403 The @code{assume_aligned} attribute is used to tell the compiler that the
2404 function return value points to memory, where the returned pointer minimum
2405 alignment is given by the first argument.
2406 If the attribute has two arguments, the second argument is misalignment offset.
2411 void* my_alloc1(size_t) __attribute__((assume_aligned(16)))
2412 void* my_alloc2(size_t) __attribute__((assume_aligned(32, 8)))
2416 declares that @code{my_alloc1} returns 16-byte aligned pointer and
2417 that @code{my_alloc2} returns a pointer whose value modulo 32 is equal
2420 @item bnd_instrument
2421 @cindex @code{bnd_instrument} function attribute
2422 The @code{bnd_instrument} attribute on functions is used to inform the
2423 compiler that the function should be instrumented when compiled
2424 with the @option{-fchkp-instrument-marked-only} option.
2427 @cindex @code{bnd_legacy} function attribute
2428 @cindex Pointer Bounds Checker attributes
2429 The @code{bnd_legacy} attribute on functions is used to inform the
2430 compiler that the function should not be instrumented when compiled
2431 with the @option{-fcheck-pointer-bounds} option.
2434 @cindex @code{cold} function attribute
2435 The @code{cold} attribute on functions is used to inform the compiler that
2436 the function is unlikely to be executed. The function is optimized for
2437 size rather than speed and on many targets it is placed into a special
2438 subsection of the text section so all cold functions appear close together,
2439 improving code locality of non-cold parts of program. The paths leading
2440 to calls of cold functions within code are marked as unlikely by the branch
2441 prediction mechanism. It is thus useful to mark functions used to handle
2442 unlikely conditions, such as @code{perror}, as cold to improve optimization
2443 of hot functions that do call marked functions in rare occasions.
2445 When profile feedback is available, via @option{-fprofile-use}, cold functions
2446 are automatically detected and this attribute is ignored.
2449 @cindex @code{const} function attribute
2450 @cindex functions that have no side effects
2451 Many functions do not examine any values except their arguments, and
2452 have no effects except the return value. Basically this is just slightly
2453 more strict class than the @code{pure} attribute below, since function is not
2454 allowed to read global memory.
2456 @cindex pointer arguments
2457 Note that a function that has pointer arguments and examines the data
2458 pointed to must @emph{not} be declared @code{const}. Likewise, a
2459 function that calls a non-@code{const} function usually must not be
2460 @code{const}. It does not make sense for a @code{const} function to
2465 @itemx constructor (@var{priority})
2466 @itemx destructor (@var{priority})
2467 @cindex @code{constructor} function attribute
2468 @cindex @code{destructor} function attribute
2469 The @code{constructor} attribute causes the function to be called
2470 automatically before execution enters @code{main ()}. Similarly, the
2471 @code{destructor} attribute causes the function to be called
2472 automatically after @code{main ()} completes or @code{exit ()} is
2473 called. Functions with these attributes are useful for
2474 initializing data that is used implicitly during the execution of
2477 You may provide an optional integer priority to control the order in
2478 which constructor and destructor functions are run. A constructor
2479 with a smaller priority number runs before a constructor with a larger
2480 priority number; the opposite relationship holds for destructors. So,
2481 if you have a constructor that allocates a resource and a destructor
2482 that deallocates the same resource, both functions typically have the
2483 same priority. The priorities for constructor and destructor
2484 functions are the same as those specified for namespace-scope C++
2485 objects (@pxref{C++ Attributes}).
2487 These attributes are not currently implemented for Objective-C@.
2490 @itemx deprecated (@var{msg})
2491 @cindex @code{deprecated} function attribute
2492 The @code{deprecated} attribute results in a warning if the function
2493 is used anywhere in the source file. This is useful when identifying
2494 functions that are expected to be removed in a future version of a
2495 program. The warning also includes the location of the declaration
2496 of the deprecated function, to enable users to easily find further
2497 information about why the function is deprecated, or what they should
2498 do instead. Note that the warnings only occurs for uses:
2501 int old_fn () __attribute__ ((deprecated));
2503 int (*fn_ptr)() = old_fn;
2507 results in a warning on line 3 but not line 2. The optional @var{msg}
2508 argument, which must be a string, is printed in the warning if
2511 The @code{deprecated} attribute can also be used for variables and
2512 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
2514 @item error ("@var{message}")
2515 @itemx warning ("@var{message}")
2516 @cindex @code{error} function attribute
2517 @cindex @code{warning} function attribute
2518 If the @code{error} or @code{warning} attribute
2519 is used on a function declaration and a call to such a function
2520 is not eliminated through dead code elimination or other optimizations,
2521 an error or warning (respectively) that includes @var{message} is diagnosed.
2523 for compile-time checking, especially together with @code{__builtin_constant_p}
2524 and inline functions where checking the inline function arguments is not
2525 possible through @code{extern char [(condition) ? 1 : -1];} tricks.
2527 While it is possible to leave the function undefined and thus invoke
2528 a link failure (to define the function with
2529 a message in @code{.gnu.warning*} section),
2530 when using these attributes the problem is diagnosed
2531 earlier and with exact location of the call even in presence of inline
2532 functions or when not emitting debugging information.
2534 @item externally_visible
2535 @cindex @code{externally_visible} function attribute
2536 This attribute, attached to a global variable or function, nullifies
2537 the effect of the @option{-fwhole-program} command-line option, so the
2538 object remains visible outside the current compilation unit.
2540 If @option{-fwhole-program} is used together with @option{-flto} and
2541 @command{gold} is used as the linker plugin,
2542 @code{externally_visible} attributes are automatically added to functions
2543 (not variable yet due to a current @command{gold} issue)
2544 that are accessed outside of LTO objects according to resolution file
2545 produced by @command{gold}.
2546 For other linkers that cannot generate resolution file,
2547 explicit @code{externally_visible} attributes are still necessary.
2550 @cindex @code{flatten} function attribute
2551 Generally, inlining into a function is limited. For a function marked with
2552 this attribute, every call inside this function is inlined, if possible.
2553 Whether the function itself is considered for inlining depends on its size and
2554 the current inlining parameters.
2556 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
2557 @cindex @code{format} function attribute
2558 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
2560 The @code{format} attribute specifies that a function takes @code{printf},
2561 @code{scanf}, @code{strftime} or @code{strfmon} style arguments that
2562 should be type-checked against a format string. For example, the
2567 my_printf (void *my_object, const char *my_format, ...)
2568 __attribute__ ((format (printf, 2, 3)));
2572 causes the compiler to check the arguments in calls to @code{my_printf}
2573 for consistency with the @code{printf} style format string argument
2576 The parameter @var{archetype} determines how the format string is
2577 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime},
2578 @code{gnu_printf}, @code{gnu_scanf}, @code{gnu_strftime} or
2579 @code{strfmon}. (You can also use @code{__printf__},
2580 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) On
2581 MinGW targets, @code{ms_printf}, @code{ms_scanf}, and
2582 @code{ms_strftime} are also present.
2583 @var{archetype} values such as @code{printf} refer to the formats accepted
2584 by the system's C runtime library,
2585 while values prefixed with @samp{gnu_} always refer
2586 to the formats accepted by the GNU C Library. On Microsoft Windows
2587 targets, values prefixed with @samp{ms_} refer to the formats accepted by the
2588 @file{msvcrt.dll} library.
2589 The parameter @var{string-index}
2590 specifies which argument is the format string argument (starting
2591 from 1), while @var{first-to-check} is the number of the first
2592 argument to check against the format string. For functions
2593 where the arguments are not available to be checked (such as
2594 @code{vprintf}), specify the third parameter as zero. In this case the
2595 compiler only checks the format string for consistency. For
2596 @code{strftime} formats, the third parameter is required to be zero.
2597 Since non-static C++ methods have an implicit @code{this} argument, the
2598 arguments of such methods should be counted from two, not one, when
2599 giving values for @var{string-index} and @var{first-to-check}.
2601 In the example above, the format string (@code{my_format}) is the second
2602 argument of the function @code{my_print}, and the arguments to check
2603 start with the third argument, so the correct parameters for the format
2604 attribute are 2 and 3.
2606 @opindex ffreestanding
2607 @opindex fno-builtin
2608 The @code{format} attribute allows you to identify your own functions
2609 that take format strings as arguments, so that GCC can check the
2610 calls to these functions for errors. The compiler always (unless
2611 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
2612 for the standard library functions @code{printf}, @code{fprintf},
2613 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2614 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2615 warnings are requested (using @option{-Wformat}), so there is no need to
2616 modify the header file @file{stdio.h}. In C99 mode, the functions
2617 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2618 @code{vsscanf} are also checked. Except in strictly conforming C
2619 standard modes, the X/Open function @code{strfmon} is also checked as
2620 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2621 @xref{C Dialect Options,,Options Controlling C Dialect}.
2623 For Objective-C dialects, @code{NSString} (or @code{__NSString__}) is
2624 recognized in the same context. Declarations including these format attributes
2625 are parsed for correct syntax, however the result of checking of such format
2626 strings is not yet defined, and is not carried out by this version of the
2629 The target may also provide additional types of format checks.
2630 @xref{Target Format Checks,,Format Checks Specific to Particular
2633 @item format_arg (@var{string-index})
2634 @cindex @code{format_arg} function attribute
2635 @opindex Wformat-nonliteral
2636 The @code{format_arg} attribute specifies that a function takes a format
2637 string for a @code{printf}, @code{scanf}, @code{strftime} or
2638 @code{strfmon} style function and modifies it (for example, to translate
2639 it into another language), so the result can be passed to a
2640 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2641 function (with the remaining arguments to the format function the same
2642 as they would have been for the unmodified string). For example, the
2647 my_dgettext (char *my_domain, const char *my_format)
2648 __attribute__ ((format_arg (2)));
2652 causes the compiler to check the arguments in calls to a @code{printf},
2653 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2654 format string argument is a call to the @code{my_dgettext} function, for
2655 consistency with the format string argument @code{my_format}. If the
2656 @code{format_arg} attribute had not been specified, all the compiler
2657 could tell in such calls to format functions would be that the format
2658 string argument is not constant; this would generate a warning when
2659 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2660 without the attribute.
2662 The parameter @var{string-index} specifies which argument is the format
2663 string argument (starting from one). Since non-static C++ methods have
2664 an implicit @code{this} argument, the arguments of such methods should
2665 be counted from two.
2667 The @code{format_arg} attribute allows you to identify your own
2668 functions that modify format strings, so that GCC can check the
2669 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2670 type function whose operands are a call to one of your own function.
2671 The compiler always treats @code{gettext}, @code{dgettext}, and
2672 @code{dcgettext} in this manner except when strict ISO C support is
2673 requested by @option{-ansi} or an appropriate @option{-std} option, or
2674 @option{-ffreestanding} or @option{-fno-builtin}
2675 is used. @xref{C Dialect Options,,Options
2676 Controlling C Dialect}.
2678 For Objective-C dialects, the @code{format-arg} attribute may refer to an
2679 @code{NSString} reference for compatibility with the @code{format} attribute
2682 The target may also allow additional types in @code{format-arg} attributes.
2683 @xref{Target Format Checks,,Format Checks Specific to Particular
2687 @cindex @code{gnu_inline} function attribute
2688 This attribute should be used with a function that is also declared
2689 with the @code{inline} keyword. It directs GCC to treat the function
2690 as if it were defined in gnu90 mode even when compiling in C99 or
2693 If the function is declared @code{extern}, then this definition of the
2694 function is used only for inlining. In no case is the function
2695 compiled as a standalone function, not even if you take its address
2696 explicitly. Such an address becomes an external reference, as if you
2697 had only declared the function, and had not defined it. This has
2698 almost the effect of a macro. The way to use this is to put a
2699 function definition in a header file with this attribute, and put
2700 another copy of the function, without @code{extern}, in a library
2701 file. The definition in the header file causes most calls to the
2702 function to be inlined. If any uses of the function remain, they
2703 refer to the single copy in the library. Note that the two
2704 definitions of the functions need not be precisely the same, although
2705 if they do not have the same effect your program may behave oddly.
2707 In C, if the function is neither @code{extern} nor @code{static}, then
2708 the function is compiled as a standalone function, as well as being
2709 inlined where possible.
2711 This is how GCC traditionally handled functions declared
2712 @code{inline}. Since ISO C99 specifies a different semantics for
2713 @code{inline}, this function attribute is provided as a transition
2714 measure and as a useful feature in its own right. This attribute is
2715 available in GCC 4.1.3 and later. It is available if either of the
2716 preprocessor macros @code{__GNUC_GNU_INLINE__} or
2717 @code{__GNUC_STDC_INLINE__} are defined. @xref{Inline,,An Inline
2718 Function is As Fast As a Macro}.
2720 In C++, this attribute does not depend on @code{extern} in any way,
2721 but it still requires the @code{inline} keyword to enable its special
2725 @cindex @code{hot} function attribute
2726 The @code{hot} attribute on a function is used to inform the compiler that
2727 the function is a hot spot of the compiled program. The function is
2728 optimized more aggressively and on many targets it is placed into a special
2729 subsection of the text section so all hot functions appear close together,
2732 When profile feedback is available, via @option{-fprofile-use}, hot functions
2733 are automatically detected and this attribute is ignored.
2735 @item ifunc ("@var{resolver}")
2736 @cindex @code{ifunc} function attribute
2737 @cindex indirect functions
2738 @cindex functions that are dynamically resolved
2739 The @code{ifunc} attribute is used to mark a function as an indirect
2740 function using the STT_GNU_IFUNC symbol type extension to the ELF
2741 standard. This allows the resolution of the symbol value to be
2742 determined dynamically at load time, and an optimized version of the
2743 routine can be selected for the particular processor or other system
2744 characteristics determined then. To use this attribute, first define
2745 the implementation functions available, and a resolver function that
2746 returns a pointer to the selected implementation function. The
2747 implementation functions' declarations must match the API of the
2748 function being implemented, the resolver's declaration is be a
2749 function returning pointer to void function returning void:
2752 void *my_memcpy (void *dst, const void *src, size_t len)
2757 static void (*resolve_memcpy (void)) (void)
2759 return my_memcpy; // we'll just always select this routine
2764 The exported header file declaring the function the user calls would
2768 extern void *memcpy (void *, const void *, size_t);
2772 allowing the user to call this as a regular function, unaware of the
2773 implementation. Finally, the indirect function needs to be defined in
2774 the same translation unit as the resolver function:
2777 void *memcpy (void *, const void *, size_t)
2778 __attribute__ ((ifunc ("resolve_memcpy")));
2781 Indirect functions cannot be weak. Binutils version 2.20.1 or higher
2782 and GNU C Library version 2.11.1 are required to use this feature.
2785 @itemx interrupt_handler
2786 Many GCC back ends support attributes to indicate that a function is
2787 an interrupt handler, which tells the compiler to generate function
2788 entry and exit sequences that differ from those from regular
2789 functions. The exact syntax and behavior are target-specific;
2790 refer to the following subsections for details.
2793 @cindex @code{leaf} function attribute
2794 Calls to external functions with this attribute must return to the current
2795 compilation unit only by return or by exception handling. In particular, leaf
2796 functions are not allowed to call callback function passed to it from the current
2797 compilation unit or directly call functions exported by the unit or longjmp
2798 into the unit. Leaf function might still call functions from other compilation
2799 units and thus they are not necessarily leaf in the sense that they contain no
2800 function calls at all.
2802 The attribute is intended for library functions to improve dataflow analysis.
2803 The compiler takes the hint that any data not escaping the current compilation unit can
2804 not be used or modified by the leaf function. For example, the @code{sin} function
2805 is a leaf function, but @code{qsort} is not.
2807 Note that leaf functions might invoke signals and signal handlers might be
2808 defined in the current compilation unit and use static variables. The only
2809 compliant way to write such a signal handler is to declare such variables
2812 The attribute has no effect on functions defined within the current compilation
2813 unit. This is to allow easy merging of multiple compilation units into one,
2814 for example, by using the link-time optimization. For this reason the
2815 attribute is not allowed on types to annotate indirect calls.
2819 @cindex @code{malloc} function attribute
2820 @cindex functions that behave like malloc
2821 This tells the compiler that a function is @code{malloc}-like, i.e.,
2822 that the pointer @var{P} returned by the function cannot alias any
2823 other pointer valid when the function returns, and moreover no
2824 pointers to valid objects occur in any storage addressed by @var{P}.
2826 Using this attribute can improve optimization. Functions like
2827 @code{malloc} and @code{calloc} have this property because they return
2828 a pointer to uninitialized or zeroed-out storage. However, functions
2829 like @code{realloc} do not have this property, as they can return a
2830 pointer to storage containing pointers.
2833 @cindex @code{no_icf} function attribute
2834 This function attribute prevents a functions from being merged with another
2835 semantically equivalent function.
2837 @item no_instrument_function
2838 @cindex @code{no_instrument_function} function attribute
2839 @opindex finstrument-functions
2840 If @option{-finstrument-functions} is given, profiling function calls are
2841 generated at entry and exit of most user-compiled functions.
2842 Functions with this attribute are not so instrumented.
2845 @cindex @code{no_reorder} function attribute
2846 Do not reorder functions or variables marked @code{no_reorder}
2847 against each other or top level assembler statements the executable.
2848 The actual order in the program will depend on the linker command
2849 line. Static variables marked like this are also not removed.
2850 This has a similar effect
2851 as the @option{-fno-toplevel-reorder} option, but only applies to the
2854 @item no_sanitize_address
2855 @itemx no_address_safety_analysis
2856 @cindex @code{no_sanitize_address} function attribute
2857 The @code{no_sanitize_address} attribute on functions is used
2858 to inform the compiler that it should not instrument memory accesses
2859 in the function when compiling with the @option{-fsanitize=address} option.
2860 The @code{no_address_safety_analysis} is a deprecated alias of the
2861 @code{no_sanitize_address} attribute, new code should use
2862 @code{no_sanitize_address}.
2864 @item no_sanitize_thread
2865 @cindex @code{no_sanitize_thread} function attribute
2866 The @code{no_sanitize_thread} attribute on functions is used
2867 to inform the compiler that it should not instrument memory accesses
2868 in the function when compiling with the @option{-fsanitize=thread} option.
2870 @item no_sanitize_undefined
2871 @cindex @code{no_sanitize_undefined} function attribute
2872 The @code{no_sanitize_undefined} attribute on functions is used
2873 to inform the compiler that it should not check for undefined behavior
2874 in the function when compiling with the @option{-fsanitize=undefined} option.
2876 @item no_split_stack
2877 @cindex @code{no_split_stack} function attribute
2878 @opindex fsplit-stack
2879 If @option{-fsplit-stack} is given, functions have a small
2880 prologue which decides whether to split the stack. Functions with the
2881 @code{no_split_stack} attribute do not have that prologue, and thus
2882 may run with only a small amount of stack space available.
2885 @cindex @code{noclone} function attribute
2886 This function attribute prevents a function from being considered for
2887 cloning---a mechanism that produces specialized copies of functions
2888 and which is (currently) performed by interprocedural constant
2892 @cindex @code{noinline} function attribute
2893 This function attribute prevents a function from being considered for
2895 @c Don't enumerate the optimizations by name here; we try to be
2896 @c future-compatible with this mechanism.
2897 If the function does not have side-effects, there are optimizations
2898 other than inlining that cause function calls to be optimized away,
2899 although the function call is live. To keep such calls from being
2906 (@pxref{Extended Asm}) in the called function, to serve as a special
2909 @item nonnull (@var{arg-index}, @dots{})
2910 @cindex @code{nonnull} function attribute
2911 @cindex functions with non-null pointer arguments
2912 The @code{nonnull} attribute specifies that some function parameters should
2913 be non-null pointers. For instance, the declaration:
2917 my_memcpy (void *dest, const void *src, size_t len)
2918 __attribute__((nonnull (1, 2)));
2922 causes the compiler to check that, in calls to @code{my_memcpy},
2923 arguments @var{dest} and @var{src} are non-null. If the compiler
2924 determines that a null pointer is passed in an argument slot marked
2925 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2926 is issued. The compiler may also choose to make optimizations based
2927 on the knowledge that certain function arguments will never be null.
2929 If no argument index list is given to the @code{nonnull} attribute,
2930 all pointer arguments are marked as non-null. To illustrate, the
2931 following declaration is equivalent to the previous example:
2935 my_memcpy (void *dest, const void *src, size_t len)
2936 __attribute__((nonnull));
2940 @cindex @code{noreturn} function attribute
2941 @cindex functions that never return
2942 A few standard library functions, such as @code{abort} and @code{exit},
2943 cannot return. GCC knows this automatically. Some programs define
2944 their own functions that never return. You can declare them
2945 @code{noreturn} to tell the compiler this fact. For example,
2949 void fatal () __attribute__ ((noreturn));
2952 fatal (/* @r{@dots{}} */)
2954 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2960 The @code{noreturn} keyword tells the compiler to assume that
2961 @code{fatal} cannot return. It can then optimize without regard to what
2962 would happen if @code{fatal} ever did return. This makes slightly
2963 better code. More importantly, it helps avoid spurious warnings of
2964 uninitialized variables.
2966 The @code{noreturn} keyword does not affect the exceptional path when that
2967 applies: a @code{noreturn}-marked function may still return to the caller
2968 by throwing an exception or calling @code{longjmp}.
2970 Do not assume that registers saved by the calling function are
2971 restored before calling the @code{noreturn} function.
2973 It does not make sense for a @code{noreturn} function to have a return
2974 type other than @code{void}.
2977 @cindex @code{nothrow} function attribute
2978 The @code{nothrow} attribute is used to inform the compiler that a
2979 function cannot throw an exception. For example, most functions in
2980 the standard C library can be guaranteed not to throw an exception
2981 with the notable exceptions of @code{qsort} and @code{bsearch} that
2982 take function pointer arguments.
2985 @cindex @code{noplt} function attribute
2986 The @code{noplt} attribute is the counterpart to option @option{-fno-plt} and
2987 does not use PLT for calls to functions marked with this attribute in position
2992 /* Externally defined function foo. */
2993 int foo () __attribute__ ((noplt));
2996 main (/* @r{@dots{}} */)
3005 The @code{noplt} attribute on function foo tells the compiler to assume that
3006 the function foo is externally defined and the call to foo must avoid the PLT
3007 in position independent code.
3009 Additionally, a few targets also convert calls to those functions that are
3010 marked to not use the PLT to use the GOT instead for non-position independent
3014 @cindex @code{optimize} function attribute
3015 The @code{optimize} attribute is used to specify that a function is to
3016 be compiled with different optimization options than specified on the
3017 command line. Arguments can either be numbers or strings. Numbers
3018 are assumed to be an optimization level. Strings that begin with
3019 @code{O} are assumed to be an optimization option, while other options
3020 are assumed to be used with a @code{-f} prefix. You can also use the
3021 @samp{#pragma GCC optimize} pragma to set the optimization options
3022 that affect more than one function.
3023 @xref{Function Specific Option Pragmas}, for details about the
3024 @samp{#pragma GCC optimize} pragma.
3026 This can be used for instance to have frequently-executed functions
3027 compiled with more aggressive optimization options that produce faster
3028 and larger code, while other functions can be compiled with less
3032 @cindex @code{pure} function attribute
3033 @cindex functions that have no side effects
3034 Many functions have no effects except the return value and their
3035 return value depends only on the parameters and/or global variables.
3036 Such a function can be subject
3037 to common subexpression elimination and loop optimization just as an
3038 arithmetic operator would be. These functions should be declared
3039 with the attribute @code{pure}. For example,
3042 int square (int) __attribute__ ((pure));
3046 says that the hypothetical function @code{square} is safe to call
3047 fewer times than the program says.
3049 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
3050 Interesting non-pure functions are functions with infinite loops or those
3051 depending on volatile memory or other system resource, that may change between
3052 two consecutive calls (such as @code{feof} in a multithreading environment).
3054 @item returns_nonnull
3055 @cindex @code{returns_nonnull} function attribute
3056 The @code{returns_nonnull} attribute specifies that the function
3057 return value should be a non-null pointer. For instance, the declaration:
3061 mymalloc (size_t len) __attribute__((returns_nonnull));
3065 lets the compiler optimize callers based on the knowledge
3066 that the return value will never be null.
3069 @cindex @code{returns_twice} function attribute
3070 @cindex functions that return more than once
3071 The @code{returns_twice} attribute tells the compiler that a function may
3072 return more than one time. The compiler ensures that all registers
3073 are dead before calling such a function and emits a warning about
3074 the variables that may be clobbered after the second return from the
3075 function. Examples of such functions are @code{setjmp} and @code{vfork}.
3076 The @code{longjmp}-like counterpart of such function, if any, might need
3077 to be marked with the @code{noreturn} attribute.
3079 @item section ("@var{section-name}")
3080 @cindex @code{section} function attribute
3081 @cindex functions in arbitrary sections
3082 Normally, the compiler places the code it generates in the @code{text} section.
3083 Sometimes, however, you need additional sections, or you need certain
3084 particular functions to appear in special sections. The @code{section}
3085 attribute specifies that a function lives in a particular section.
3086 For example, the declaration:
3089 extern void foobar (void) __attribute__ ((section ("bar")));
3093 puts the function @code{foobar} in the @code{bar} section.
3095 Some file formats do not support arbitrary sections so the @code{section}
3096 attribute is not available on all platforms.
3097 If you need to map the entire contents of a module to a particular
3098 section, consider using the facilities of the linker instead.
3101 @cindex @code{sentinel} function attribute
3102 This function attribute ensures that a parameter in a function call is
3103 an explicit @code{NULL}. The attribute is only valid on variadic
3104 functions. By default, the sentinel is located at position zero, the
3105 last parameter of the function call. If an optional integer position
3106 argument P is supplied to the attribute, the sentinel must be located at
3107 position P counting backwards from the end of the argument list.
3110 __attribute__ ((sentinel))
3112 __attribute__ ((sentinel(0)))
3115 The attribute is automatically set with a position of 0 for the built-in
3116 functions @code{execl} and @code{execlp}. The built-in function
3117 @code{execle} has the attribute set with a position of 1.
3119 A valid @code{NULL} in this context is defined as zero with any pointer
3120 type. If your system defines the @code{NULL} macro with an integer type
3121 then you need to add an explicit cast. GCC replaces @code{stddef.h}
3122 with a copy that redefines NULL appropriately.
3124 The warnings for missing or incorrect sentinels are enabled with
3128 @cindex @code{stack_protect} function attribute
3129 This function attribute make a stack protection of the function if
3130 flags @option{fstack-protector} or @option{fstack-protector-strong}
3131 or @option{fstack-protector-explicit} are set.
3133 @item target_clones (@var{options})
3134 @cindex @code{target_clones} function attribute
3135 The @code{target_clones} attribute is used to specify that a function is to
3136 be cloned into multiple versions compiled with different target options
3137 than specified on the command line. The supported options and restrictions
3138 are the same as for @code{target} attribute.
3140 For instance on an x86, you could compile a function with
3141 @code{target_clones("sse4.1,avx")}. It will create 2 function clones,
3142 one compiled with @option{-msse4.1} and another with @option{-mavx}.
3143 At the function call it will create resolver @code{ifunc}, that will
3144 dynamically call a clone suitable for current architecture.
3146 @item target (@var{options})
3147 @cindex @code{target} function attribute
3148 Multiple target back ends implement the @code{target} attribute
3149 to specify that a function is to
3150 be compiled with different target options than specified on the
3151 command line. This can be used for instance to have functions
3152 compiled with a different ISA (instruction set architecture) than the
3153 default. You can also use the @samp{#pragma GCC target} pragma to set
3154 more than one function to be compiled with specific target options.
3155 @xref{Function Specific Option Pragmas}, for details about the
3156 @samp{#pragma GCC target} pragma.
3158 For instance, on an x86, you could declare one function with the
3159 @code{target("sse4.1,arch=core2")} attribute and another with
3160 @code{target("sse4a,arch=amdfam10")}. This is equivalent to
3161 compiling the first function with @option{-msse4.1} and
3162 @option{-march=core2} options, and the second function with
3163 @option{-msse4a} and @option{-march=amdfam10} options. It is up to you
3164 to make sure that a function is only invoked on a machine that
3165 supports the particular ISA it is compiled for (for example by using
3166 @code{cpuid} on x86 to determine what feature bits and architecture
3170 int core2_func (void) __attribute__ ((__target__ ("arch=core2")));
3171 int sse3_func (void) __attribute__ ((__target__ ("sse3")));
3174 You can either use multiple
3175 strings separated by commas to specify multiple options,
3176 or separate the options with a comma (@samp{,}) within a single string.
3178 The options supported are specific to each target; refer to @ref{x86
3179 Function Attributes}, @ref{PowerPC Function Attributes},
3180 @ref{ARM Function Attributes},and @ref{Nios II Function Attributes},
3184 @cindex @code{unused} function attribute
3185 This attribute, attached to a function, means that the function is meant
3186 to be possibly unused. GCC does not produce a warning for this
3190 @cindex @code{used} function attribute
3191 This attribute, attached to a function, means that code must be emitted
3192 for the function even if it appears that the function is not referenced.
3193 This is useful, for example, when the function is referenced only in
3196 When applied to a member function of a C++ class template, the
3197 attribute also means that the function is instantiated if the
3198 class itself is instantiated.
3200 @item visibility ("@var{visibility_type}")
3201 @cindex @code{visibility} function attribute
3202 This attribute affects the linkage of the declaration to which it is attached.
3203 There are four supported @var{visibility_type} values: default,
3204 hidden, protected or internal visibility.
3207 void __attribute__ ((visibility ("protected")))
3208 f () @{ /* @r{Do something.} */; @}
3209 int i __attribute__ ((visibility ("hidden")));
3212 The possible values of @var{visibility_type} correspond to the
3213 visibility settings in the ELF gABI.
3216 @c keep this list of visibilities in alphabetical order.
3219 Default visibility is the normal case for the object file format.
3220 This value is available for the visibility attribute to override other
3221 options that may change the assumed visibility of entities.
3223 On ELF, default visibility means that the declaration is visible to other
3224 modules and, in shared libraries, means that the declared entity may be
3227 On Darwin, default visibility means that the declaration is visible to
3230 Default visibility corresponds to ``external linkage'' in the language.
3233 Hidden visibility indicates that the entity declared has a new
3234 form of linkage, which we call ``hidden linkage''. Two
3235 declarations of an object with hidden linkage refer to the same object
3236 if they are in the same shared object.
3239 Internal visibility is like hidden visibility, but with additional
3240 processor specific semantics. Unless otherwise specified by the
3241 psABI, GCC defines internal visibility to mean that a function is
3242 @emph{never} called from another module. Compare this with hidden
3243 functions which, while they cannot be referenced directly by other
3244 modules, can be referenced indirectly via function pointers. By
3245 indicating that a function cannot be called from outside the module,
3246 GCC may for instance omit the load of a PIC register since it is known
3247 that the calling function loaded the correct value.
3250 Protected visibility is like default visibility except that it
3251 indicates that references within the defining module bind to the
3252 definition in that module. That is, the declared entity cannot be
3253 overridden by another module.
3257 All visibilities are supported on many, but not all, ELF targets
3258 (supported when the assembler supports the @samp{.visibility}
3259 pseudo-op). Default visibility is supported everywhere. Hidden
3260 visibility is supported on Darwin targets.
3262 The visibility attribute should be applied only to declarations that
3263 would otherwise have external linkage. The attribute should be applied
3264 consistently, so that the same entity should not be declared with
3265 different settings of the attribute.
3267 In C++, the visibility attribute applies to types as well as functions
3268 and objects, because in C++ types have linkage. A class must not have
3269 greater visibility than its non-static data member types and bases,
3270 and class members default to the visibility of their class. Also, a
3271 declaration without explicit visibility is limited to the visibility
3274 In C++, you can mark member functions and static member variables of a
3275 class with the visibility attribute. This is useful if you know a
3276 particular method or static member variable should only be used from
3277 one shared object; then you can mark it hidden while the rest of the
3278 class has default visibility. Care must be taken to avoid breaking
3279 the One Definition Rule; for example, it is usually not useful to mark
3280 an inline method as hidden without marking the whole class as hidden.
3282 A C++ namespace declaration can also have the visibility attribute.
3285 namespace nspace1 __attribute__ ((visibility ("protected")))
3286 @{ /* @r{Do something.} */; @}
3289 This attribute applies only to the particular namespace body, not to
3290 other definitions of the same namespace; it is equivalent to using
3291 @samp{#pragma GCC visibility} before and after the namespace
3292 definition (@pxref{Visibility Pragmas}).
3294 In C++, if a template argument has limited visibility, this
3295 restriction is implicitly propagated to the template instantiation.
3296 Otherwise, template instantiations and specializations default to the
3297 visibility of their template.
3299 If both the template and enclosing class have explicit visibility, the
3300 visibility from the template is used.
3302 @item warn_unused_result
3303 @cindex @code{warn_unused_result} function attribute
3304 The @code{warn_unused_result} attribute causes a warning to be emitted
3305 if a caller of the function with this attribute does not use its
3306 return value. This is useful for functions where not checking
3307 the result is either a security problem or always a bug, such as
3311 int fn () __attribute__ ((warn_unused_result));
3314 if (fn () < 0) return -1;
3321 results in warning on line 5.
3324 @cindex @code{weak} function attribute
3325 The @code{weak} attribute causes the declaration to be emitted as a weak
3326 symbol rather than a global. This is primarily useful in defining
3327 library functions that can be overridden in user code, though it can
3328 also be used with non-function declarations. Weak symbols are supported
3329 for ELF targets, and also for a.out targets when using the GNU assembler
3333 @itemx weakref ("@var{target}")
3334 @cindex @code{weakref} function attribute
3335 The @code{weakref} attribute marks a declaration as a weak reference.
3336 Without arguments, it should be accompanied by an @code{alias} attribute
3337 naming the target symbol. Optionally, the @var{target} may be given as
3338 an argument to @code{weakref} itself. In either case, @code{weakref}
3339 implicitly marks the declaration as @code{weak}. Without a
3340 @var{target}, given as an argument to @code{weakref} or to @code{alias},
3341 @code{weakref} is equivalent to @code{weak}.
3344 static int x() __attribute__ ((weakref ("y")));
3345 /* is equivalent to... */
3346 static int x() __attribute__ ((weak, weakref, alias ("y")));
3348 static int x() __attribute__ ((weakref));
3349 static int x() __attribute__ ((alias ("y")));
3352 A weak reference is an alias that does not by itself require a
3353 definition to be given for the target symbol. If the target symbol is
3354 only referenced through weak references, then it becomes a @code{weak}
3355 undefined symbol. If it is directly referenced, however, then such
3356 strong references prevail, and a definition is required for the
3357 symbol, not necessarily in the same translation unit.
3359 The effect is equivalent to moving all references to the alias to a
3360 separate translation unit, renaming the alias to the aliased symbol,
3361 declaring it as weak, compiling the two separate translation units and
3362 performing a reloadable link on them.
3364 At present, a declaration to which @code{weakref} is attached can
3365 only be @code{static}.
3370 @cindex lower memory region on the MSP430
3371 @cindex upper memory region on the MSP430
3372 @cindex either memory region on the MSP430
3373 On the MSP430 target these attributes can be used to specify whether
3374 the function or variable should be placed into low memory, high
3375 memory, or the placement should be left to the linker to decide. The
3376 attributes are only significant if compiling for the MSP430X
3379 The attributes work in conjunction with a linker script that has been
3380 augmented to specify where to place sections with a @code{.lower} and
3381 a @code{.upper} prefix. So for example as well as placing the
3382 @code{.data} section the script would also specify the placement of a
3383 @code{.lower.data} and a @code{.upper.data} section. The intention
3384 being that @code{lower} sections are placed into a small but easier to
3385 access memory region and the upper sections are placed into a larger, but
3386 slower to access region.
3388 The @code{either} attribute is special. It tells the linker to place
3389 the object into the corresponding @code{lower} section if there is
3390 room for it. If there is insufficient room then the object is placed
3391 into the corresponding @code{upper} section instead. Note - the
3392 placement algorithm is not very sophisticated. It will not attempt to
3393 find an optimal packing of the @code{lower} sections. It just makes
3394 one pass over the objects and does the best that it can. Using the
3395 @option{-ffunction-sections} and @option{-fdata-sections} command line
3396 options can help the packing however, since they produce smaller,
3397 easier to pack regions.
3400 On the MSP430 a function can be given the @code{reentant} attribute.
3401 This makes the function disable interrupts upon entry and enable
3402 interrupts upon exit. Reentrant functions cannot be @code{naked}.
3405 On the MSP430 a function can be given the @code{critical} attribute.
3406 This makes the function disable interrupts upon entry and restore the
3407 previous interrupt enabled/disabled state upon exit. A function
3408 cannot have both the @code{reentrant} and @code{critical} attributes.
3409 Critical functions cannot be @code{naked}.
3412 On the MSP430 a function can be given the @code{wakeup} attribute.
3413 Such a function must also have the @code{interrupt} attribute. When a
3414 function with the @code{wakeup} attribute exists the processor will be
3415 woken up from any low-power state in which it may be residing.
3419 @c This is the end of the target-independent attribute table
3421 @node AArch64 Function Attributes
3422 @subsection AArch64 Function Attributes
3424 The following target-specific function attributes are available for the
3425 AArch64 target. For the most part, these options mirror the behavior of
3426 similar command-line options (@pxref{AArch64 Options}), but on a
3430 @item general-regs-only
3431 @cindex @code{general-regs-only} function attribute, AArch64
3432 Indicates that no floating-point or Advanced SIMD registers should be
3433 used when generating code for this function. If the function explicitly
3434 uses floating-point code, then the compiler gives an error. This is
3435 the same behavior as that of the command-line option
3436 @option{-mgeneral-regs-only}.
3438 @item fix-cortex-a53-835769
3439 @cindex @code{fix-cortex-a53-835769} function attribute, AArch64
3440 Indicates that the workaround for the Cortex-A53 erratum 835769 should be
3441 applied to this function. To explicitly disable the workaround for this
3442 function specify the negated form: @code{no-fix-cortex-a53-835769}.
3443 This corresponds to the behavior of the command line options
3444 @option{-mfix-cortex-a53-835769} and @option{-mno-fix-cortex-a53-835769}.
3447 @cindex @code{cmodel=} function attribute, AArch64
3448 Indicates that code should be generated for a particular code model for
3449 this function. The behavior and permissible arguments are the same as
3450 for the command line option @option{-mcmodel=}.
3453 @cindex @code{strict-align} function attribute, AArch64
3454 Indicates that the compiler should not assume that unaligned memory references
3455 are handled by the system. The behavior is the same as for the command-line
3456 option @option{-mstrict-align}.
3458 @item omit-leaf-frame-pointer
3459 @cindex @code{omit-leaf-frame-pointer} function attribute, AArch64
3460 Indicates that the frame pointer should be omitted for a leaf function call.
3461 To keep the frame pointer, the inverse attribute
3462 @code{no-omit-leaf-frame-pointer} can be specified. These attributes have
3463 the same behavior as the command-line options @option{-momit-leaf-frame-pointer}
3464 and @option{-mno-omit-leaf-frame-pointer}.
3467 @cindex @code{tls-dialect=} function attribute, AArch64
3468 Specifies the TLS dialect to use for this function. The behavior and
3469 permissible arguments are the same as for the command-line option
3470 @option{-mtls-dialect=}.
3473 @cindex @code{arch=} function attribute, AArch64
3474 Specifies the architecture version and architectural extensions to use
3475 for this function. The behavior and permissible arguments are the same as
3476 for the @option{-march=} command-line option.
3479 @cindex @code{tune=} function attribute, AArch64
3480 Specifies the core for which to tune the performance of this function.
3481 The behavior and permissible arguments are the same as for the @option{-mtune=}
3482 command-line option.
3485 @cindex @code{cpu=} function attribute, AArch64
3486 Specifies the core for which to tune the performance of this function and also
3487 whose architectural features to use. The behavior and valid arguments are the
3488 same as for the @option{-mcpu=} command-line option.
3492 The above target attributes can be specified as follows:
3495 __attribute__((target("@var{attr-string}")))
3503 where @code{@var{attr-string}} is one of the attribute strings specified above.
3505 Additionally, the architectural extension string may be specified on its
3506 own. This can be used to turn on and off particular architectural extensions
3507 without having to specify a particular architecture version or core. Example:
3510 __attribute__((target("+crc+nocrypto")))
3518 In this example @code{target("+crc+nocrypto")} enables the @code{crc}
3519 extension and disables the @code{crypto} extension for the function @code{foo}
3520 without modifying an existing @option{-march=} or @option{-mcpu} option.
3522 Multiple target function attributes can be specified by separating them with
3523 a comma. For example:
3525 __attribute__((target("arch=armv8-a+crc+crypto,tune=cortex-a53")))
3533 is valid and compiles function @code{foo} for ARMv8-A with @code{crc}
3534 and @code{crypto} extensions and tunes it for @code{cortex-a53}.
3536 @subsubsection Inlining rules
3537 Specifying target attributes on individual functions or performing link-time
3538 optimization across translation units compiled with different target options
3539 can affect function inlining rules:
3541 In particular, a caller function can inline a callee function only if the
3542 architectural features available to the callee are a subset of the features
3543 available to the caller.
3544 For example: A function @code{foo} compiled with @option{-march=armv8-a+crc},
3545 or tagged with the equivalent @code{arch=armv8-a+crc} attribute,
3546 can inline a function @code{bar} compiled with @option{-march=armv8-a+nocrc}
3547 because the all the architectural features that function @code{bar} requires
3548 are available to function @code{foo}. Conversely, function @code{bar} cannot
3549 inline function @code{foo}.
3551 Additionally inlining a function compiled with @option{-mstrict-align} into a
3552 function compiled without @code{-mstrict-align} is not allowed.
3553 However, inlining a function compiled without @option{-mstrict-align} into a
3554 function compiled with @option{-mstrict-align} is allowed.
3556 Note that CPU tuning options and attributes such as the @option{-mcpu=},
3557 @option{-mtune=} do not inhibit inlining unless the CPU specified by the
3558 @option{-mcpu=} option or the @code{cpu=} attribute conflicts with the
3559 architectural feature rules specified above.
3561 @node ARC Function Attributes
3562 @subsection ARC Function Attributes
3564 These function attributes are supported by the ARC back end:
3568 @cindex @code{interrupt} function attribute, ARC
3569 Use this attribute to indicate
3570 that the specified function is an interrupt handler. The compiler generates
3571 function entry and exit sequences suitable for use in an interrupt handler
3572 when this attribute is present.
3574 On the ARC, you must specify the kind of interrupt to be handled
3575 in a parameter to the interrupt attribute like this:
3578 void f () __attribute__ ((interrupt ("ilink1")));
3581 Permissible values for this parameter are: @w{@code{ilink1}} and
3587 @cindex @code{long_call} function attribute, ARC
3588 @cindex @code{medium_call} function attribute, ARC
3589 @cindex @code{short_call} function attribute, ARC
3590 @cindex indirect calls, ARC
3591 These attributes specify how a particular function is called.
3592 These attributes override the
3593 @option{-mlong-calls} and @option{-mmedium-calls} (@pxref{ARC Options})
3594 command-line switches and @code{#pragma long_calls} settings.
3596 For ARC, a function marked with the @code{long_call} attribute is
3597 always called using register-indirect jump-and-link instructions,
3598 thereby enabling the called function to be placed anywhere within the
3599 32-bit address space. A function marked with the @code{medium_call}
3600 attribute will always be close enough to be called with an unconditional
3601 branch-and-link instruction, which has a 25-bit offset from
3602 the call site. A function marked with the @code{short_call}
3603 attribute will always be close enough to be called with a conditional
3604 branch-and-link instruction, which has a 21-bit offset from
3608 @node ARM Function Attributes
3609 @subsection ARM Function Attributes
3611 These function attributes are supported for ARM targets:
3615 @cindex @code{interrupt} function attribute, ARM
3616 Use this attribute to indicate
3617 that the specified function is an interrupt handler. The compiler generates
3618 function entry and exit sequences suitable for use in an interrupt handler
3619 when this attribute is present.
3621 You can specify the kind of interrupt to be handled by
3622 adding an optional parameter to the interrupt attribute like this:
3625 void f () __attribute__ ((interrupt ("IRQ")));
3629 Permissible values for this parameter are: @code{IRQ}, @code{FIQ},
3630 @code{SWI}, @code{ABORT} and @code{UNDEF}.
3632 On ARMv7-M the interrupt type is ignored, and the attribute means the function
3633 may be called with a word-aligned stack pointer.
3636 @cindex @code{isr} function attribute, ARM
3637 Use this attribute on ARM to write Interrupt Service Routines. This is an
3638 alias to the @code{interrupt} attribute above.
3642 @cindex @code{long_call} function attribute, ARM
3643 @cindex @code{short_call} function attribute, ARM
3644 @cindex indirect calls, ARM
3645 These attributes specify how a particular function is called.
3646 These attributes override the
3647 @option{-mlong-calls} (@pxref{ARM Options})
3648 command-line switch and @code{#pragma long_calls} settings. For ARM, the
3649 @code{long_call} attribute indicates that the function might be far
3650 away from the call site and require a different (more expensive)
3651 calling sequence. The @code{short_call} attribute always places
3652 the offset to the function from the call site into the @samp{BL}
3653 instruction directly.
3656 @cindex @code{naked} function attribute, ARM
3657 This attribute allows the compiler to construct the
3658 requisite function declaration, while allowing the body of the
3659 function to be assembly code. The specified function will not have
3660 prologue/epilogue sequences generated by the compiler. Only basic
3661 @code{asm} statements can safely be included in naked functions
3662 (@pxref{Basic Asm}). While using extended @code{asm} or a mixture of
3663 basic @code{asm} and C code may appear to work, they cannot be
3664 depended upon to work reliably and are not supported.
3667 @cindex @code{pcs} function attribute, ARM
3669 The @code{pcs} attribute can be used to control the calling convention
3670 used for a function on ARM. The attribute takes an argument that specifies
3671 the calling convention to use.
3673 When compiling using the AAPCS ABI (or a variant of it) then valid
3674 values for the argument are @code{"aapcs"} and @code{"aapcs-vfp"}. In
3675 order to use a variant other than @code{"aapcs"} then the compiler must
3676 be permitted to use the appropriate co-processor registers (i.e., the
3677 VFP registers must be available in order to use @code{"aapcs-vfp"}).
3681 /* Argument passed in r0, and result returned in r0+r1. */
3682 double f2d (float) __attribute__((pcs("aapcs")));
3685 Variadic functions always use the @code{"aapcs"} calling convention and
3686 the compiler rejects attempts to specify an alternative.
3688 @item target (@var{options})
3689 @cindex @code{target} function attribute
3690 As discussed in @ref{Common Function Attributes}, this attribute
3691 allows specification of target-specific compilation options.
3693 On ARM, the following options are allowed:
3697 @cindex @code{target("thumb")} function attribute, ARM
3698 Force code generation in the Thumb (T16/T32) ISA, depending on the
3702 @cindex @code{target("arm")} function attribute, ARM
3703 Force code generation in the ARM (A32) ISA.
3706 Functions from different modes can be inlined in the caller's mode.
3710 @node AVR Function Attributes
3711 @subsection AVR Function Attributes
3713 These function attributes are supported by the AVR back end:
3717 @cindex @code{interrupt} function attribute, AVR
3718 Use this attribute to indicate
3719 that the specified function is an interrupt handler. The compiler generates
3720 function entry and exit sequences suitable for use in an interrupt handler
3721 when this attribute is present.
3723 On the AVR, the hardware globally disables interrupts when an
3724 interrupt is executed. The first instruction of an interrupt handler
3725 declared with this attribute is a @code{SEI} instruction to
3726 re-enable interrupts. See also the @code{signal} function attribute
3727 that does not insert a @code{SEI} instruction. If both @code{signal} and
3728 @code{interrupt} are specified for the same function, @code{signal}
3729 is silently ignored.
3732 @cindex @code{naked} function attribute, AVR
3733 This attribute allows the compiler to construct the
3734 requisite function declaration, while allowing the body of the
3735 function to be assembly code. The specified function will not have
3736 prologue/epilogue sequences generated by the compiler. Only basic
3737 @code{asm} statements can safely be included in naked functions
3738 (@pxref{Basic Asm}). While using extended @code{asm} or a mixture of
3739 basic @code{asm} and C code may appear to work, they cannot be
3740 depended upon to work reliably and are not supported.
3744 @cindex @code{OS_main} function attribute, AVR
3745 @cindex @code{OS_task} function attribute, AVR
3746 On AVR, functions with the @code{OS_main} or @code{OS_task} attribute
3747 do not save/restore any call-saved register in their prologue/epilogue.
3749 The @code{OS_main} attribute can be used when there @emph{is
3750 guarantee} that interrupts are disabled at the time when the function
3751 is entered. This saves resources when the stack pointer has to be
3752 changed to set up a frame for local variables.
3754 The @code{OS_task} attribute can be used when there is @emph{no
3755 guarantee} that interrupts are disabled at that time when the function
3756 is entered like for, e@.g@. task functions in a multi-threading operating
3757 system. In that case, changing the stack pointer register is
3758 guarded by save/clear/restore of the global interrupt enable flag.
3760 The differences to the @code{naked} function attribute are:
3762 @item @code{naked} functions do not have a return instruction whereas
3763 @code{OS_main} and @code{OS_task} functions have a @code{RET} or
3764 @code{RETI} return instruction.
3765 @item @code{naked} functions do not set up a frame for local variables
3766 or a frame pointer whereas @code{OS_main} and @code{OS_task} do this
3771 @cindex @code{signal} function attribute, AVR
3772 Use this attribute on the AVR to indicate that the specified
3773 function is an interrupt handler. The compiler generates function
3774 entry and exit sequences suitable for use in an interrupt handler when this
3775 attribute is present.
3777 See also the @code{interrupt} function attribute.
3779 The AVR hardware globally disables interrupts when an interrupt is executed.
3780 Interrupt handler functions defined with the @code{signal} attribute
3781 do not re-enable interrupts. It is save to enable interrupts in a
3782 @code{signal} handler. This ``save'' only applies to the code
3783 generated by the compiler and not to the IRQ layout of the
3784 application which is responsibility of the application.
3786 If both @code{signal} and @code{interrupt} are specified for the same
3787 function, @code{signal} is silently ignored.
3790 @node Blackfin Function Attributes
3791 @subsection Blackfin Function Attributes
3793 These function attributes are supported by the Blackfin back end:
3797 @item exception_handler
3798 @cindex @code{exception_handler} function attribute
3799 @cindex exception handler functions, Blackfin
3800 Use this attribute on the Blackfin to indicate that the specified function
3801 is an exception handler. The compiler generates function entry and
3802 exit sequences suitable for use in an exception handler when this
3803 attribute is present.
3805 @item interrupt_handler
3806 @cindex @code{interrupt_handler} function attribute, Blackfin
3807 Use this attribute to
3808 indicate that the specified function is an interrupt handler. The compiler
3809 generates function entry and exit sequences suitable for use in an
3810 interrupt handler when this attribute is present.
3813 @cindex @code{kspisusp} function attribute, Blackfin
3814 @cindex User stack pointer in interrupts on the Blackfin
3815 When used together with @code{interrupt_handler}, @code{exception_handler}
3816 or @code{nmi_handler}, code is generated to load the stack pointer
3817 from the USP register in the function prologue.
3820 @cindex @code{l1_text} function attribute, Blackfin
3821 This attribute specifies a function to be placed into L1 Instruction
3822 SRAM@. The function is put into a specific section named @code{.l1.text}.
3823 With @option{-mfdpic}, function calls with a such function as the callee
3824 or caller uses inlined PLT.
3827 @cindex @code{l2} function attribute, Blackfin
3828 This attribute specifies a function to be placed into L2
3829 SRAM. The function is put into a specific section named
3830 @code{.l2.text}. With @option{-mfdpic}, callers of such functions use
3835 @cindex indirect calls, Blackfin
3836 @cindex @code{longcall} function attribute, Blackfin
3837 @cindex @code{shortcall} function attribute, Blackfin
3838 The @code{longcall} attribute
3839 indicates that the function might be far away from the call site and
3840 require a different (more expensive) calling sequence. The
3841 @code{shortcall} attribute indicates that the function is always close
3842 enough for the shorter calling sequence to be used. These attributes
3843 override the @option{-mlongcall} switch.
3846 @cindex @code{nesting} function attribute, Blackfin
3847 @cindex Allow nesting in an interrupt handler on the Blackfin processor
3848 Use this attribute together with @code{interrupt_handler},
3849 @code{exception_handler} or @code{nmi_handler} to indicate that the function
3850 entry code should enable nested interrupts or exceptions.
3853 @cindex @code{nmi_handler} function attribute, Blackfin
3854 @cindex NMI handler functions on the Blackfin processor
3855 Use this attribute on the Blackfin to indicate that the specified function
3856 is an NMI handler. The compiler generates function entry and
3857 exit sequences suitable for use in an NMI handler when this
3858 attribute is present.
3861 @cindex @code{saveall} function attribute, Blackfin
3862 @cindex save all registers on the Blackfin
3863 Use this attribute to indicate that
3864 all registers except the stack pointer should be saved in the prologue
3865 regardless of whether they are used or not.
3868 @node CR16 Function Attributes
3869 @subsection CR16 Function Attributes
3871 These function attributes are supported by the CR16 back end:
3875 @cindex @code{interrupt} function attribute, CR16
3876 Use this attribute to indicate
3877 that the specified function is an interrupt handler. The compiler generates
3878 function entry and exit sequences suitable for use in an interrupt handler
3879 when this attribute is present.
3882 @node Epiphany Function Attributes
3883 @subsection Epiphany Function Attributes
3885 These function attributes are supported by the Epiphany back end:
3889 @cindex @code{disinterrupt} function attribute, Epiphany
3890 This attribute causes the compiler to emit
3891 instructions to disable interrupts for the duration of the given
3894 @item forwarder_section
3895 @cindex @code{forwarder_section} function attribute, Epiphany
3896 This attribute modifies the behavior of an interrupt handler.
3897 The interrupt handler may be in external memory which cannot be
3898 reached by a branch instruction, so generate a local memory trampoline
3899 to transfer control. The single parameter identifies the section where
3900 the trampoline is placed.
3903 @cindex @code{interrupt} function attribute, Epiphany
3904 Use this attribute to indicate
3905 that the specified function is an interrupt handler. The compiler generates
3906 function entry and exit sequences suitable for use in an interrupt handler
3907 when this attribute is present. It may also generate
3908 a special section with code to initialize the interrupt vector table.
3910 On Epiphany targets one or more optional parameters can be added like this:
3913 void __attribute__ ((interrupt ("dma0, dma1"))) universal_dma_handler ();
3916 Permissible values for these parameters are: @w{@code{reset}},
3917 @w{@code{software_exception}}, @w{@code{page_miss}},
3918 @w{@code{timer0}}, @w{@code{timer1}}, @w{@code{message}},
3919 @w{@code{dma0}}, @w{@code{dma1}}, @w{@code{wand}} and @w{@code{swi}}.
3920 Multiple parameters indicate that multiple entries in the interrupt
3921 vector table should be initialized for this function, i.e.@: for each
3922 parameter @w{@var{name}}, a jump to the function is emitted in
3923 the section @w{ivt_entry_@var{name}}. The parameter(s) may be omitted
3924 entirely, in which case no interrupt vector table entry is provided.
3926 Note that interrupts are enabled inside the function
3927 unless the @code{disinterrupt} attribute is also specified.
3929 The following examples are all valid uses of these attributes on
3932 void __attribute__ ((interrupt)) universal_handler ();
3933 void __attribute__ ((interrupt ("dma1"))) dma1_handler ();
3934 void __attribute__ ((interrupt ("dma0, dma1")))
3935 universal_dma_handler ();
3936 void __attribute__ ((interrupt ("timer0"), disinterrupt))
3937 fast_timer_handler ();
3938 void __attribute__ ((interrupt ("dma0, dma1"),
3939 forwarder_section ("tramp")))
3940 external_dma_handler ();
3945 @cindex @code{long_call} function attribute, Epiphany
3946 @cindex @code{short_call} function attribute, Epiphany
3947 @cindex indirect calls, Epiphany
3948 These attributes specify how a particular function is called.
3949 These attributes override the
3950 @option{-mlong-calls} (@pxref{Adapteva Epiphany Options})
3951 command-line switch and @code{#pragma long_calls} settings.
3955 @node H8/300 Function Attributes
3956 @subsection H8/300 Function Attributes
3958 These function attributes are available for H8/300 targets:
3961 @item function_vector
3962 @cindex @code{function_vector} function attribute, H8/300
3963 Use this attribute on the H8/300, H8/300H, and H8S to indicate
3964 that the specified function should be called through the function vector.
3965 Calling a function through the function vector reduces code size; however,
3966 the function vector has a limited size (maximum 128 entries on the H8/300
3967 and 64 entries on the H8/300H and H8S)
3968 and shares space with the interrupt vector.
3970 @item interrupt_handler
3971 @cindex @code{interrupt_handler} function attribute, H8/300
3972 Use this attribute on the H8/300, H8/300H, and H8S to
3973 indicate that the specified function is an interrupt handler. The compiler
3974 generates function entry and exit sequences suitable for use in an
3975 interrupt handler when this attribute is present.
3978 @cindex @code{saveall} function attribute, H8/300
3979 @cindex save all registers on the H8/300, H8/300H, and H8S
3980 Use this attribute on the H8/300, H8/300H, and H8S to indicate that
3981 all registers except the stack pointer should be saved in the prologue
3982 regardless of whether they are used or not.
3985 @node IA-64 Function Attributes
3986 @subsection IA-64 Function Attributes
3988 These function attributes are supported on IA-64 targets:
3991 @item syscall_linkage
3992 @cindex @code{syscall_linkage} function attribute, IA-64
3993 This attribute is used to modify the IA-64 calling convention by marking
3994 all input registers as live at all function exits. This makes it possible
3995 to restart a system call after an interrupt without having to save/restore
3996 the input registers. This also prevents kernel data from leaking into
4000 @cindex @code{version_id} function attribute, IA-64
4001 This IA-64 HP-UX attribute, attached to a global variable or function, renames a
4002 symbol to contain a version string, thus allowing for function level
4003 versioning. HP-UX system header files may use function level versioning
4004 for some system calls.
4007 extern int foo () __attribute__((version_id ("20040821")));
4011 Calls to @code{foo} are mapped to calls to @code{foo@{20040821@}}.
4014 @node M32C Function Attributes
4015 @subsection M32C Function Attributes
4017 These function attributes are supported by the M32C back end:
4021 @cindex @code{bank_switch} function attribute, M32C
4022 When added to an interrupt handler with the M32C port, causes the
4023 prologue and epilogue to use bank switching to preserve the registers
4024 rather than saving them on the stack.
4026 @item fast_interrupt
4027 @cindex @code{fast_interrupt} function attribute, M32C
4028 Use this attribute on the M32C port to indicate that the specified
4029 function is a fast interrupt handler. This is just like the
4030 @code{interrupt} attribute, except that @code{freit} is used to return
4031 instead of @code{reit}.
4033 @item function_vector
4034 @cindex @code{function_vector} function attribute, M16C/M32C
4035 On M16C/M32C targets, the @code{function_vector} attribute declares a
4036 special page subroutine call function. Use of this attribute reduces
4037 the code size by 2 bytes for each call generated to the
4038 subroutine. The argument to the attribute is the vector number entry
4039 from the special page vector table which contains the 16 low-order
4040 bits of the subroutine's entry address. Each vector table has special
4041 page number (18 to 255) that is used in @code{jsrs} instructions.
4042 Jump addresses of the routines are generated by adding 0x0F0000 (in
4043 case of M16C targets) or 0xFF0000 (in case of M32C targets), to the
4044 2-byte addresses set in the vector table. Therefore you need to ensure
4045 that all the special page vector routines should get mapped within the
4046 address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF
4049 In the following example 2 bytes are saved for each call to
4050 function @code{foo}.
4053 void foo (void) __attribute__((function_vector(0x18)));
4064 If functions are defined in one file and are called in another file,
4065 then be sure to write this declaration in both files.
4067 This attribute is ignored for R8C target.
4070 @cindex @code{interrupt} function attribute, M32C
4071 Use this attribute to indicate
4072 that the specified function is an interrupt handler. The compiler generates
4073 function entry and exit sequences suitable for use in an interrupt handler
4074 when this attribute is present.
4077 @node M32R/D Function Attributes
4078 @subsection M32R/D Function Attributes
4080 These function attributes are supported by the M32R/D back end:
4084 @cindex @code{interrupt} function attribute, M32R/D
4085 Use this attribute to indicate
4086 that the specified function is an interrupt handler. The compiler generates
4087 function entry and exit sequences suitable for use in an interrupt handler
4088 when this attribute is present.
4090 @item model (@var{model-name})
4091 @cindex @code{model} function attribute, M32R/D
4092 @cindex function addressability on the M32R/D
4094 On the M32R/D, use this attribute to set the addressability of an
4095 object, and of the code generated for a function. The identifier
4096 @var{model-name} is one of @code{small}, @code{medium}, or
4097 @code{large}, representing each of the code models.
4099 Small model objects live in the lower 16MB of memory (so that their
4100 addresses can be loaded with the @code{ld24} instruction), and are
4101 callable with the @code{bl} instruction.
4103 Medium model objects may live anywhere in the 32-bit address space (the
4104 compiler generates @code{seth/add3} instructions to load their addresses),
4105 and are callable with the @code{bl} instruction.
4107 Large model objects may live anywhere in the 32-bit address space (the
4108 compiler generates @code{seth/add3} instructions to load their addresses),
4109 and may not be reachable with the @code{bl} instruction (the compiler
4110 generates the much slower @code{seth/add3/jl} instruction sequence).
4113 @node m68k Function Attributes
4114 @subsection m68k Function Attributes
4116 These function attributes are supported by the m68k back end:
4120 @itemx interrupt_handler
4121 @cindex @code{interrupt} function attribute, m68k
4122 @cindex @code{interrupt_handler} function attribute, m68k
4123 Use this attribute to
4124 indicate that the specified function is an interrupt handler. The compiler
4125 generates function entry and exit sequences suitable for use in an
4126 interrupt handler when this attribute is present. Either name may be used.
4128 @item interrupt_thread
4129 @cindex @code{interrupt_thread} function attribute, fido
4130 Use this attribute on fido, a subarchitecture of the m68k, to indicate
4131 that the specified function is an interrupt handler that is designed
4132 to run as a thread. The compiler omits generate prologue/epilogue
4133 sequences and replaces the return instruction with a @code{sleep}
4134 instruction. This attribute is available only on fido.
4137 @node MCORE Function Attributes
4138 @subsection MCORE Function Attributes
4140 These function attributes are supported by the MCORE back end:
4144 @cindex @code{naked} function attribute, MCORE
4145 This attribute allows the compiler to construct the
4146 requisite function declaration, while allowing the body of the
4147 function to be assembly code. The specified function will not have
4148 prologue/epilogue sequences generated by the compiler. Only basic
4149 @code{asm} statements can safely be included in naked functions
4150 (@pxref{Basic Asm}). While using extended @code{asm} or a mixture of
4151 basic @code{asm} and C code may appear to work, they cannot be
4152 depended upon to work reliably and are not supported.
4155 @node MeP Function Attributes
4156 @subsection MeP Function Attributes
4158 These function attributes are supported by the MeP back end:
4162 @cindex @code{disinterrupt} function attribute, MeP
4163 On MeP targets, this attribute causes the compiler to emit
4164 instructions to disable interrupts for the duration of the given
4168 @cindex @code{interrupt} function attribute, MeP
4169 Use this attribute to indicate
4170 that the specified function is an interrupt handler. The compiler generates
4171 function entry and exit sequences suitable for use in an interrupt handler
4172 when this attribute is present.
4175 @cindex @code{near} function attribute, MeP
4176 This attribute causes the compiler to assume the called
4177 function is close enough to use the normal calling convention,
4178 overriding the @option{-mtf} command-line option.
4181 @cindex @code{far} function attribute, MeP
4182 On MeP targets this causes the compiler to use a calling convention
4183 that assumes the called function is too far away for the built-in
4187 @cindex @code{vliw} function attribute, MeP
4188 The @code{vliw} attribute tells the compiler to emit
4189 instructions in VLIW mode instead of core mode. Note that this
4190 attribute is not allowed unless a VLIW coprocessor has been configured
4191 and enabled through command-line options.
4194 @node MicroBlaze Function Attributes
4195 @subsection MicroBlaze Function Attributes
4197 These function attributes are supported on MicroBlaze targets:
4200 @item save_volatiles
4201 @cindex @code{save_volatiles} function attribute, MicroBlaze
4202 Use this attribute to indicate that the function is
4203 an interrupt handler. All volatile registers (in addition to non-volatile
4204 registers) are saved in the function prologue. If the function is a leaf
4205 function, only volatiles used by the function are saved. A normal function
4206 return is generated instead of a return from interrupt.
4209 @cindex @code{break_handler} function attribute, MicroBlaze
4210 @cindex break handler functions
4211 Use this attribute to indicate that
4212 the specified function is a break handler. The compiler generates function
4213 entry and exit sequences suitable for use in an break handler when this
4214 attribute is present. The return from @code{break_handler} is done through
4215 the @code{rtbd} instead of @code{rtsd}.
4218 void f () __attribute__ ((break_handler));
4222 @node Microsoft Windows Function Attributes
4223 @subsection Microsoft Windows Function Attributes
4225 The following attributes are available on Microsoft Windows and Symbian OS
4230 @cindex @code{dllexport} function attribute
4231 @cindex @code{__declspec(dllexport)}
4232 On Microsoft Windows targets and Symbian OS targets the
4233 @code{dllexport} attribute causes the compiler to provide a global
4234 pointer to a pointer in a DLL, so that it can be referenced with the
4235 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
4236 name is formed by combining @code{_imp__} and the function or variable
4239 You can use @code{__declspec(dllexport)} as a synonym for
4240 @code{__attribute__ ((dllexport))} for compatibility with other
4243 On systems that support the @code{visibility} attribute, this
4244 attribute also implies ``default'' visibility. It is an error to
4245 explicitly specify any other visibility.
4247 GCC's default behavior is to emit all inline functions with the
4248 @code{dllexport} attribute. Since this can cause object file-size bloat,
4249 you can use @option{-fno-keep-inline-dllexport}, which tells GCC to
4250 ignore the attribute for inlined functions unless the
4251 @option{-fkeep-inline-functions} flag is used instead.
4253 The attribute is ignored for undefined symbols.
4255 When applied to C++ classes, the attribute marks defined non-inlined
4256 member functions and static data members as exports. Static consts
4257 initialized in-class are not marked unless they are also defined
4260 For Microsoft Windows targets there are alternative methods for
4261 including the symbol in the DLL's export table such as using a
4262 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
4263 the @option{--export-all} linker flag.
4266 @cindex @code{dllimport} function attribute
4267 @cindex @code{__declspec(dllimport)}
4268 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
4269 attribute causes the compiler to reference a function or variable via
4270 a global pointer to a pointer that is set up by the DLL exporting the
4271 symbol. The attribute implies @code{extern}. On Microsoft Windows
4272 targets, the pointer name is formed by combining @code{_imp__} and the
4273 function or variable name.
4275 You can use @code{__declspec(dllimport)} as a synonym for
4276 @code{__attribute__ ((dllimport))} for compatibility with other
4279 On systems that support the @code{visibility} attribute, this
4280 attribute also implies ``default'' visibility. It is an error to
4281 explicitly specify any other visibility.
4283 Currently, the attribute is ignored for inlined functions. If the
4284 attribute is applied to a symbol @emph{definition}, an error is reported.
4285 If a symbol previously declared @code{dllimport} is later defined, the
4286 attribute is ignored in subsequent references, and a warning is emitted.
4287 The attribute is also overridden by a subsequent declaration as
4290 When applied to C++ classes, the attribute marks non-inlined
4291 member functions and static data members as imports. However, the
4292 attribute is ignored for virtual methods to allow creation of vtables
4295 On the SH Symbian OS target the @code{dllimport} attribute also has
4296 another affect---it can cause the vtable and run-time type information
4297 for a class to be exported. This happens when the class has a
4298 dllimported constructor or a non-inline, non-pure virtual function
4299 and, for either of those two conditions, the class also has an inline
4300 constructor or destructor and has a key function that is defined in
4301 the current translation unit.
4303 For Microsoft Windows targets the use of the @code{dllimport}
4304 attribute on functions is not necessary, but provides a small
4305 performance benefit by eliminating a thunk in the DLL@. The use of the
4306 @code{dllimport} attribute on imported variables can be avoided by passing the
4307 @option{--enable-auto-import} switch to the GNU linker. As with
4308 functions, using the attribute for a variable eliminates a thunk in
4311 One drawback to using this attribute is that a pointer to a
4312 @emph{variable} marked as @code{dllimport} cannot be used as a constant
4313 address. However, a pointer to a @emph{function} with the
4314 @code{dllimport} attribute can be used as a constant initializer; in
4315 this case, the address of a stub function in the import lib is
4316 referenced. On Microsoft Windows targets, the attribute can be disabled
4317 for functions by setting the @option{-mnop-fun-dllimport} flag.
4320 @node MIPS Function Attributes
4321 @subsection MIPS Function Attributes
4323 These function attributes are supported by the MIPS back end:
4327 @cindex @code{interrupt} function attribute, MIPS
4328 Use this attribute to indicate that the specified function is an interrupt
4329 handler. The compiler generates function entry and exit sequences suitable
4330 for use in an interrupt handler when this attribute is present.
4331 An optional argument is supported for the interrupt attribute which allows
4332 the interrupt mode to be described. By default GCC assumes the external
4333 interrupt controller (EIC) mode is in use, this can be explicitly set using
4334 @code{eic}. When interrupts are non-masked then the requested Interrupt
4335 Priority Level (IPL) is copied to the current IPL which has the effect of only
4336 enabling higher priority interrupts. To use vectored interrupt mode use
4337 the argument @code{vector=[sw0|sw1|hw0|hw1|hw2|hw3|hw4|hw5]}, this will change
4338 the behaviour of the non-masked interrupt support and GCC will arrange to mask
4339 all interrupts from sw0 up to and including the specified interrupt vector.
4341 You can use the following attributes to modify the behavior
4342 of an interrupt handler:
4344 @item use_shadow_register_set
4345 @cindex @code{use_shadow_register_set} function attribute, MIPS
4346 Assume that the handler uses a shadow register set, instead of
4347 the main general-purpose registers. An optional argument @code{intstack} is
4348 supported to indicate that the shadow register set contains a valid stack
4351 @item keep_interrupts_masked
4352 @cindex @code{keep_interrupts_masked} function attribute, MIPS
4353 Keep interrupts masked for the whole function. Without this attribute,
4354 GCC tries to reenable interrupts for as much of the function as it can.
4356 @item use_debug_exception_return
4357 @cindex @code{use_debug_exception_return} function attribute, MIPS
4358 Return using the @code{deret} instruction. Interrupt handlers that don't
4359 have this attribute return using @code{eret} instead.
4362 You can use any combination of these attributes, as shown below:
4364 void __attribute__ ((interrupt)) v0 ();
4365 void __attribute__ ((interrupt, use_shadow_register_set)) v1 ();
4366 void __attribute__ ((interrupt, keep_interrupts_masked)) v2 ();
4367 void __attribute__ ((interrupt, use_debug_exception_return)) v3 ();
4368 void __attribute__ ((interrupt, use_shadow_register_set,
4369 keep_interrupts_masked)) v4 ();
4370 void __attribute__ ((interrupt, use_shadow_register_set,
4371 use_debug_exception_return)) v5 ();
4372 void __attribute__ ((interrupt, keep_interrupts_masked,
4373 use_debug_exception_return)) v6 ();
4374 void __attribute__ ((interrupt, use_shadow_register_set,
4375 keep_interrupts_masked,
4376 use_debug_exception_return)) v7 ();
4377 void __attribute__ ((interrupt("eic"))) v8 ();
4378 void __attribute__ ((interrupt("vector=hw3"))) v9 ();
4384 @cindex indirect calls, MIPS
4385 @cindex @code{long_call} function attribute, MIPS
4386 @cindex @code{near} function attribute, MIPS
4387 @cindex @code{far} function attribute, MIPS
4388 These attributes specify how a particular function is called on MIPS@.
4389 The attributes override the @option{-mlong-calls} (@pxref{MIPS Options})
4390 command-line switch. The @code{long_call} and @code{far} attributes are
4391 synonyms, and cause the compiler to always call
4392 the function by first loading its address into a register, and then using
4393 the contents of that register. The @code{near} attribute has the opposite
4394 effect; it specifies that non-PIC calls should be made using the more
4395 efficient @code{jal} instruction.
4399 @cindex @code{mips16} function attribute, MIPS
4400 @cindex @code{nomips16} function attribute, MIPS
4402 On MIPS targets, you can use the @code{mips16} and @code{nomips16}
4403 function attributes to locally select or turn off MIPS16 code generation.
4404 A function with the @code{mips16} attribute is emitted as MIPS16 code,
4405 while MIPS16 code generation is disabled for functions with the
4406 @code{nomips16} attribute. These attributes override the
4407 @option{-mips16} and @option{-mno-mips16} options on the command line
4408 (@pxref{MIPS Options}).
4410 When compiling files containing mixed MIPS16 and non-MIPS16 code, the
4411 preprocessor symbol @code{__mips16} reflects the setting on the command line,
4412 not that within individual functions. Mixed MIPS16 and non-MIPS16 code
4413 may interact badly with some GCC extensions such as @code{__builtin_apply}
4414 (@pxref{Constructing Calls}).
4416 @item micromips, MIPS
4417 @itemx nomicromips, MIPS
4418 @cindex @code{micromips} function attribute
4419 @cindex @code{nomicromips} function attribute
4421 On MIPS targets, you can use the @code{micromips} and @code{nomicromips}
4422 function attributes to locally select or turn off microMIPS code generation.
4423 A function with the @code{micromips} attribute is emitted as microMIPS code,
4424 while microMIPS code generation is disabled for functions with the
4425 @code{nomicromips} attribute. These attributes override the
4426 @option{-mmicromips} and @option{-mno-micromips} options on the command line
4427 (@pxref{MIPS Options}).
4429 When compiling files containing mixed microMIPS and non-microMIPS code, the
4430 preprocessor symbol @code{__mips_micromips} reflects the setting on the
4432 not that within individual functions. Mixed microMIPS and non-microMIPS code
4433 may interact badly with some GCC extensions such as @code{__builtin_apply}
4434 (@pxref{Constructing Calls}).
4437 @cindex @code{nocompression} function attribute, MIPS
4438 On MIPS targets, you can use the @code{nocompression} function attribute
4439 to locally turn off MIPS16 and microMIPS code generation. This attribute
4440 overrides the @option{-mips16} and @option{-mmicromips} options on the
4441 command line (@pxref{MIPS Options}).
4444 @node MSP430 Function Attributes
4445 @subsection MSP430 Function Attributes
4447 These function attributes are supported by the MSP430 back end:
4451 @cindex @code{critical} function attribute, MSP430
4452 Critical functions disable interrupts upon entry and restore the
4453 previous interrupt state upon exit. Critical functions cannot also
4454 have the @code{naked} or @code{reentrant} attributes. They can have
4455 the @code{interrupt} attribute.
4458 @cindex @code{interrupt} function attribute, MSP430
4459 Use this attribute to indicate
4460 that the specified function is an interrupt handler. The compiler generates
4461 function entry and exit sequences suitable for use in an interrupt handler
4462 when this attribute is present.
4464 You can provide an argument to the interrupt
4465 attribute which specifies a name or number. If the argument is a
4466 number it indicates the slot in the interrupt vector table (0 - 31) to
4467 which this handler should be assigned. If the argument is a name it
4468 is treated as a symbolic name for the vector slot. These names should
4469 match up with appropriate entries in the linker script. By default
4470 the names @code{watchdog} for vector 26, @code{nmi} for vector 30 and
4471 @code{reset} for vector 31 are recognized.
4474 @cindex @code{naked} function attribute, MSP430
4475 This attribute allows the compiler to construct the
4476 requisite function declaration, while allowing the body of the
4477 function to be assembly code. The specified function will not have
4478 prologue/epilogue sequences generated by the compiler. Only basic
4479 @code{asm} statements can safely be included in naked functions
4480 (@pxref{Basic Asm}). While using extended @code{asm} or a mixture of
4481 basic @code{asm} and C code may appear to work, they cannot be
4482 depended upon to work reliably and are not supported.
4485 @cindex @code{reentrant} function attribute, MSP430
4486 Reentrant functions disable interrupts upon entry and enable them
4487 upon exit. Reentrant functions cannot also have the @code{naked}
4488 or @code{critical} attributes. They can have the @code{interrupt}
4492 @cindex @code{wakeup} function attribute, MSP430
4493 This attribute only applies to interrupt functions. It is silently
4494 ignored if applied to a non-interrupt function. A wakeup interrupt
4495 function will rouse the processor from any low-power state that it
4496 might be in when the function exits.
4499 @node NDS32 Function Attributes
4500 @subsection NDS32 Function Attributes
4502 These function attributes are supported by the NDS32 back end:
4506 @cindex @code{exception} function attribute
4507 @cindex exception handler functions, NDS32
4508 Use this attribute on the NDS32 target to indicate that the specified function
4509 is an exception handler. The compiler will generate corresponding sections
4510 for use in an exception handler.
4513 @cindex @code{interrupt} function attribute, NDS32
4514 On NDS32 target, this attribute indicates that the specified function
4515 is an interrupt handler. The compiler generates corresponding sections
4516 for use in an interrupt handler. You can use the following attributes
4517 to modify the behavior:
4520 @cindex @code{nested} function attribute, NDS32
4521 This interrupt service routine is interruptible.
4523 @cindex @code{not_nested} function attribute, NDS32
4524 This interrupt service routine is not interruptible.
4526 @cindex @code{nested_ready} function attribute, NDS32
4527 This interrupt service routine is interruptible after @code{PSW.GIE}
4528 (global interrupt enable) is set. This allows interrupt service routine to
4529 finish some short critical code before enabling interrupts.
4531 @cindex @code{save_all} function attribute, NDS32
4532 The system will help save all registers into stack before entering
4535 @cindex @code{partial_save} function attribute, NDS32
4536 The system will help save caller registers into stack before entering
4541 @cindex @code{naked} function attribute, NDS32
4542 This attribute allows the compiler to construct the
4543 requisite function declaration, while allowing the body of the
4544 function to be assembly code. The specified function will not have
4545 prologue/epilogue sequences generated by the compiler. Only basic
4546 @code{asm} statements can safely be included in naked functions
4547 (@pxref{Basic Asm}). While using extended @code{asm} or a mixture of
4548 basic @code{asm} and C code may appear to work, they cannot be
4549 depended upon to work reliably and are not supported.
4552 @cindex @code{reset} function attribute, NDS32
4553 @cindex reset handler functions
4554 Use this attribute on the NDS32 target to indicate that the specified function
4555 is a reset handler. The compiler will generate corresponding sections
4556 for use in a reset handler. You can use the following attributes
4557 to provide extra exception handling:
4560 @cindex @code{nmi} function attribute, NDS32
4561 Provide a user-defined function to handle NMI exception.
4563 @cindex @code{warm} function attribute, NDS32
4564 Provide a user-defined function to handle warm reset exception.
4568 @node Nios II Function Attributes
4569 @subsection Nios II Function Attributes
4571 These function attributes are supported by the Nios II back end:
4574 @item target (@var{options})
4575 @cindex @code{target} function attribute
4576 As discussed in @ref{Common Function Attributes}, this attribute
4577 allows specification of target-specific compilation options.
4579 When compiling for Nios II, the following options are allowed:
4582 @item custom-@var{insn}=@var{N}
4583 @itemx no-custom-@var{insn}
4584 @cindex @code{target("custom-@var{insn}=@var{N}")} function attribute, Nios II
4585 @cindex @code{target("no-custom-@var{insn}")} function attribute, Nios II
4586 Each @samp{custom-@var{insn}=@var{N}} attribute locally enables use of a
4587 custom instruction with encoding @var{N} when generating code that uses
4588 @var{insn}. Similarly, @samp{no-custom-@var{insn}} locally inhibits use of
4589 the custom instruction @var{insn}.
4590 These target attributes correspond to the
4591 @option{-mcustom-@var{insn}=@var{N}} and @option{-mno-custom-@var{insn}}
4592 command-line options, and support the same set of @var{insn} keywords.
4593 @xref{Nios II Options}, for more information.
4595 @item custom-fpu-cfg=@var{name}
4596 @cindex @code{target("custom-fpu-cfg=@var{name}")} function attribute, Nios II
4597 This attribute corresponds to the @option{-mcustom-fpu-cfg=@var{name}}
4598 command-line option, to select a predefined set of custom instructions
4600 @xref{Nios II Options}, for more information.
4604 @node PowerPC Function Attributes
4605 @subsection PowerPC Function Attributes
4607 These function attributes are supported by the PowerPC back end:
4612 @cindex indirect calls, PowerPC
4613 @cindex @code{longcall} function attribute, PowerPC
4614 @cindex @code{shortcall} function attribute, PowerPC
4615 The @code{longcall} attribute
4616 indicates that the function might be far away from the call site and
4617 require a different (more expensive) calling sequence. The
4618 @code{shortcall} attribute indicates that the function is always close
4619 enough for the shorter calling sequence to be used. These attributes
4620 override both the @option{-mlongcall} switch and
4621 the @code{#pragma longcall} setting.
4623 @xref{RS/6000 and PowerPC Options}, for more information on whether long
4624 calls are necessary.
4626 @item target (@var{options})
4627 @cindex @code{target} function attribute
4628 As discussed in @ref{Common Function Attributes}, this attribute
4629 allows specification of target-specific compilation options.
4631 On the PowerPC, the following options are allowed:
4636 @cindex @code{target("altivec")} function attribute, PowerPC
4637 Generate code that uses (does not use) AltiVec instructions. In
4638 32-bit code, you cannot enable AltiVec instructions unless
4639 @option{-mabi=altivec} is used on the command line.
4643 @cindex @code{target("cmpb")} function attribute, PowerPC
4644 Generate code that uses (does not use) the compare bytes instruction
4645 implemented on the POWER6 processor and other processors that support
4646 the PowerPC V2.05 architecture.
4650 @cindex @code{target("dlmzb")} function attribute, PowerPC
4651 Generate code that uses (does not use) the string-search @samp{dlmzb}
4652 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
4653 generated by default when targeting those processors.
4657 @cindex @code{target("fprnd")} function attribute, PowerPC
4658 Generate code that uses (does not use) the FP round to integer
4659 instructions implemented on the POWER5+ processor and other processors
4660 that support the PowerPC V2.03 architecture.
4664 @cindex @code{target("hard-dfp")} function attribute, PowerPC
4665 Generate code that uses (does not use) the decimal floating-point
4666 instructions implemented on some POWER processors.
4670 @cindex @code{target("isel")} function attribute, PowerPC
4671 Generate code that uses (does not use) ISEL instruction.
4675 @cindex @code{target("mfcrf")} function attribute, PowerPC
4676 Generate code that uses (does not use) the move from condition
4677 register field instruction implemented on the POWER4 processor and
4678 other processors that support the PowerPC V2.01 architecture.
4682 @cindex @code{target("mfpgpr")} function attribute, PowerPC
4683 Generate code that uses (does not use) the FP move to/from general
4684 purpose register instructions implemented on the POWER6X processor and
4685 other processors that support the extended PowerPC V2.05 architecture.
4689 @cindex @code{target("mulhw")} function attribute, PowerPC
4690 Generate code that uses (does not use) the half-word multiply and
4691 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
4692 These instructions are generated by default when targeting those
4697 @cindex @code{target("multiple")} function attribute, PowerPC
4698 Generate code that uses (does not use) the load multiple word
4699 instructions and the store multiple word instructions.
4703 @cindex @code{target("update")} function attribute, PowerPC
4704 Generate code that uses (does not use) the load or store instructions
4705 that update the base register to the address of the calculated memory
4710 @cindex @code{target("popcntb")} function attribute, PowerPC
4711 Generate code that uses (does not use) the popcount and double-precision
4712 FP reciprocal estimate instruction implemented on the POWER5
4713 processor and other processors that support the PowerPC V2.02
4718 @cindex @code{target("popcntd")} function attribute, PowerPC
4719 Generate code that uses (does not use) the popcount instruction
4720 implemented on the POWER7 processor and other processors that support
4721 the PowerPC V2.06 architecture.
4723 @item powerpc-gfxopt
4724 @itemx no-powerpc-gfxopt
4725 @cindex @code{target("powerpc-gfxopt")} function attribute, PowerPC
4726 Generate code that uses (does not use) the optional PowerPC
4727 architecture instructions in the Graphics group, including
4728 floating-point select.
4731 @itemx no-powerpc-gpopt
4732 @cindex @code{target("powerpc-gpopt")} function attribute, PowerPC
4733 Generate code that uses (does not use) the optional PowerPC
4734 architecture instructions in the General Purpose group, including
4735 floating-point square root.
4737 @item recip-precision
4738 @itemx no-recip-precision
4739 @cindex @code{target("recip-precision")} function attribute, PowerPC
4740 Assume (do not assume) that the reciprocal estimate instructions
4741 provide higher-precision estimates than is mandated by the PowerPC
4746 @cindex @code{target("string")} function attribute, PowerPC
4747 Generate code that uses (does not use) the load string instructions
4748 and the store string word instructions to save multiple registers and
4749 do small block moves.
4753 @cindex @code{target("vsx")} function attribute, PowerPC
4754 Generate code that uses (does not use) vector/scalar (VSX)
4755 instructions, and also enable the use of built-in functions that allow
4756 more direct access to the VSX instruction set. In 32-bit code, you
4757 cannot enable VSX or AltiVec instructions unless
4758 @option{-mabi=altivec} is used on the command line.
4762 @cindex @code{target("friz")} function attribute, PowerPC
4763 Generate (do not generate) the @code{friz} instruction when the
4764 @option{-funsafe-math-optimizations} option is used to optimize
4765 rounding a floating-point value to 64-bit integer and back to floating
4766 point. The @code{friz} instruction does not return the same value if
4767 the floating-point number is too large to fit in an integer.
4769 @item avoid-indexed-addresses
4770 @itemx no-avoid-indexed-addresses
4771 @cindex @code{target("avoid-indexed-addresses")} function attribute, PowerPC
4772 Generate code that tries to avoid (not avoid) the use of indexed load
4773 or store instructions.
4777 @cindex @code{target("paired")} function attribute, PowerPC
4778 Generate code that uses (does not use) the generation of PAIRED simd
4783 @cindex @code{target("longcall")} function attribute, PowerPC
4784 Generate code that assumes (does not assume) that all calls are far
4785 away so that a longer more expensive calling sequence is required.
4788 @cindex @code{target("cpu=@var{CPU}")} function attribute, PowerPC
4789 Specify the architecture to generate code for when compiling the
4790 function. If you select the @code{target("cpu=power7")} attribute when
4791 generating 32-bit code, VSX and AltiVec instructions are not generated
4792 unless you use the @option{-mabi=altivec} option on the command line.
4794 @item tune=@var{TUNE}
4795 @cindex @code{target("tune=@var{TUNE}")} function attribute, PowerPC
4796 Specify the architecture to tune for when compiling the function. If
4797 you do not specify the @code{target("tune=@var{TUNE}")} attribute and
4798 you do specify the @code{target("cpu=@var{CPU}")} attribute,
4799 compilation tunes for the @var{CPU} architecture, and not the
4800 default tuning specified on the command line.
4803 On the PowerPC, the inliner does not inline a
4804 function that has different target options than the caller, unless the
4805 callee has a subset of the target options of the caller.
4808 @node RL78 Function Attributes
4809 @subsection RL78 Function Attributes
4811 These function attributes are supported by the RL78 back end:
4815 @itemx brk_interrupt
4816 @cindex @code{interrupt} function attribute, RL78
4817 @cindex @code{brk_interrupt} function attribute, RL78
4818 These attributes indicate
4819 that the specified function is an interrupt handler. The compiler generates
4820 function entry and exit sequences suitable for use in an interrupt handler
4821 when this attribute is present.
4823 Use @code{brk_interrupt} instead of @code{interrupt} for
4824 handlers intended to be used with the @code{BRK} opcode (i.e.@: those
4825 that must end with @code{RETB} instead of @code{RETI}).
4828 @cindex @code{naked} function attribute, RL78
4829 This attribute allows the compiler to construct the
4830 requisite function declaration, while allowing the body of the
4831 function to be assembly code. The specified function will not have
4832 prologue/epilogue sequences generated by the compiler. Only basic
4833 @code{asm} statements can safely be included in naked functions
4834 (@pxref{Basic Asm}). While using extended @code{asm} or a mixture of
4835 basic @code{asm} and C code may appear to work, they cannot be
4836 depended upon to work reliably and are not supported.
4839 @node RX Function Attributes
4840 @subsection RX Function Attributes
4842 These function attributes are supported by the RX back end:
4845 @item fast_interrupt
4846 @cindex @code{fast_interrupt} function attribute, RX
4847 Use this attribute on the RX port to indicate that the specified
4848 function is a fast interrupt handler. This is just like the
4849 @code{interrupt} attribute, except that @code{freit} is used to return
4850 instead of @code{reit}.
4853 @cindex @code{interrupt} function attribute, RX
4854 Use this attribute to indicate
4855 that the specified function is an interrupt handler. The compiler generates
4856 function entry and exit sequences suitable for use in an interrupt handler
4857 when this attribute is present.
4859 On RX targets, you may specify one or more vector numbers as arguments
4860 to the attribute, as well as naming an alternate table name.
4861 Parameters are handled sequentially, so one handler can be assigned to
4862 multiple entries in multiple tables. One may also pass the magic
4863 string @code{"$default"} which causes the function to be used for any
4864 unfilled slots in the current table.
4866 This example shows a simple assignment of a function to one vector in
4867 the default table (note that preprocessor macros may be used for
4868 chip-specific symbolic vector names):
4870 void __attribute__ ((interrupt (5))) txd1_handler ();
4873 This example assigns a function to two slots in the default table
4874 (using preprocessor macros defined elsewhere) and makes it the default
4875 for the @code{dct} table:
4877 void __attribute__ ((interrupt (RXD1_VECT,RXD2_VECT,"dct","$default")))
4882 @cindex @code{naked} function attribute, RX
4883 This attribute allows the compiler to construct the
4884 requisite function declaration, while allowing the body of the
4885 function to be assembly code. The specified function will not have
4886 prologue/epilogue sequences generated by the compiler. Only basic
4887 @code{asm} statements can safely be included in naked functions
4888 (@pxref{Basic Asm}). While using extended @code{asm} or a mixture of
4889 basic @code{asm} and C code may appear to work, they cannot be
4890 depended upon to work reliably and are not supported.
4893 @cindex @code{vector} function attribute, RX
4894 This RX attribute is similar to the @code{interrupt} attribute, including its
4895 parameters, but does not make the function an interrupt-handler type
4896 function (i.e. it retains the normal C function calling ABI). See the
4897 @code{interrupt} attribute for a description of its arguments.
4900 @node S/390 Function Attributes
4901 @subsection S/390 Function Attributes
4903 These function attributes are supported on the S/390:
4906 @item hotpatch (@var{halfwords-before-function-label},@var{halfwords-after-function-label})
4907 @cindex @code{hotpatch} function attribute, S/390
4909 On S/390 System z targets, you can use this function attribute to
4910 make GCC generate a ``hot-patching'' function prologue. If the
4911 @option{-mhotpatch=} command-line option is used at the same time,
4912 the @code{hotpatch} attribute takes precedence. The first of the
4913 two arguments specifies the number of halfwords to be added before
4914 the function label. A second argument can be used to specify the
4915 number of halfwords to be added after the function label. For
4916 both arguments the maximum allowed value is 1000000.
4918 If both arguments are zero, hotpatching is disabled.
4921 @node SH Function Attributes
4922 @subsection SH Function Attributes
4924 These function attributes are supported on the SH family of processors:
4927 @item function_vector
4928 @cindex @code{function_vector} function attribute, SH
4929 @cindex calling functions through the function vector on SH2A
4930 On SH2A targets, this attribute declares a function to be called using the
4931 TBR relative addressing mode. The argument to this attribute is the entry
4932 number of the same function in a vector table containing all the TBR
4933 relative addressable functions. For correct operation the TBR must be setup
4934 accordingly to point to the start of the vector table before any functions with
4935 this attribute are invoked. Usually a good place to do the initialization is
4936 the startup routine. The TBR relative vector table can have at max 256 function
4937 entries. The jumps to these functions are generated using a SH2A specific,
4938 non delayed branch instruction JSR/N @@(disp8,TBR). You must use GAS and GLD
4939 from GNU binutils version 2.7 or later for this attribute to work correctly.
4941 In an application, for a function being called once, this attribute
4942 saves at least 8 bytes of code; and if other successive calls are being
4943 made to the same function, it saves 2 bytes of code per each of these
4946 @item interrupt_handler
4947 @cindex @code{interrupt_handler} function attribute, SH
4948 Use this attribute to
4949 indicate that the specified function is an interrupt handler. The compiler
4950 generates function entry and exit sequences suitable for use in an
4951 interrupt handler when this attribute is present.
4953 @item nosave_low_regs
4954 @cindex @code{nosave_low_regs} function attribute, SH
4955 Use this attribute on SH targets to indicate that an @code{interrupt_handler}
4956 function should not save and restore registers R0..R7. This can be used on SH3*
4957 and SH4* targets that have a second R0..R7 register bank for non-reentrant
4961 @cindex @code{renesas} function attribute, SH
4962 On SH targets this attribute specifies that the function or struct follows the
4966 @cindex @code{resbank} function attribute, SH
4967 On the SH2A target, this attribute enables the high-speed register
4968 saving and restoration using a register bank for @code{interrupt_handler}
4969 routines. Saving to the bank is performed automatically after the CPU
4970 accepts an interrupt that uses a register bank.
4972 The nineteen 32-bit registers comprising general register R0 to R14,
4973 control register GBR, and system registers MACH, MACL, and PR and the
4974 vector table address offset are saved into a register bank. Register
4975 banks are stacked in first-in last-out (FILO) sequence. Restoration
4976 from the bank is executed by issuing a RESBANK instruction.
4979 @cindex @code{sp_switch} function attribute, SH
4980 Use this attribute on the SH to indicate an @code{interrupt_handler}
4981 function should switch to an alternate stack. It expects a string
4982 argument that names a global variable holding the address of the
4987 void f () __attribute__ ((interrupt_handler,
4988 sp_switch ("alt_stack")));
4992 @cindex @code{trap_exit} function attribute, SH
4993 Use this attribute on the SH for an @code{interrupt_handler} to return using
4994 @code{trapa} instead of @code{rte}. This attribute expects an integer
4995 argument specifying the trap number to be used.
4998 @cindex @code{trapa_handler} function attribute, SH
4999 On SH targets this function attribute is similar to @code{interrupt_handler}
5000 but it does not save and restore all registers.
5003 @node SPU Function Attributes
5004 @subsection SPU Function Attributes
5006 These function attributes are supported by the SPU back end:
5010 @cindex @code{naked} function attribute, SPU
5011 This attribute allows the compiler to construct the
5012 requisite function declaration, while allowing the body of the
5013 function to be assembly code. The specified function will not have
5014 prologue/epilogue sequences generated by the compiler. Only basic
5015 @code{asm} statements can safely be included in naked functions
5016 (@pxref{Basic Asm}). While using extended @code{asm} or a mixture of
5017 basic @code{asm} and C code may appear to work, they cannot be
5018 depended upon to work reliably and are not supported.
5021 @node Symbian OS Function Attributes
5022 @subsection Symbian OS Function Attributes
5024 @xref{Microsoft Windows Function Attributes}, for discussion of the
5025 @code{dllexport} and @code{dllimport} attributes.
5027 @node Visium Function Attributes
5028 @subsection Visium Function Attributes
5030 These function attributes are supported by the Visium back end:
5034 @cindex @code{interrupt} function attribute, Visium
5035 Use this attribute to indicate
5036 that the specified function is an interrupt handler. The compiler generates
5037 function entry and exit sequences suitable for use in an interrupt handler
5038 when this attribute is present.
5041 @node x86 Function Attributes
5042 @subsection x86 Function Attributes
5044 These function attributes are supported by the x86 back end:
5048 @cindex @code{cdecl} function attribute, x86-32
5049 @cindex functions that pop the argument stack on x86-32
5051 On the x86-32 targets, the @code{cdecl} attribute causes the compiler to
5052 assume that the calling function pops off the stack space used to
5053 pass arguments. This is
5054 useful to override the effects of the @option{-mrtd} switch.
5057 @cindex @code{fastcall} function attribute, x86-32
5058 @cindex functions that pop the argument stack on x86-32
5059 On x86-32 targets, the @code{fastcall} attribute causes the compiler to
5060 pass the first argument (if of integral type) in the register ECX and
5061 the second argument (if of integral type) in the register EDX@. Subsequent
5062 and other typed arguments are passed on the stack. The called function
5063 pops the arguments off the stack. If the number of arguments is variable all
5064 arguments are pushed on the stack.
5067 @cindex @code{thiscall} function attribute, x86-32
5068 @cindex functions that pop the argument stack on x86-32
5069 On x86-32 targets, the @code{thiscall} attribute causes the compiler to
5070 pass the first argument (if of integral type) in the register ECX.
5071 Subsequent and other typed arguments are passed on the stack. The called
5072 function pops the arguments off the stack.
5073 If the number of arguments is variable all arguments are pushed on the
5075 The @code{thiscall} attribute is intended for C++ non-static member functions.
5076 As a GCC extension, this calling convention can be used for C functions
5077 and for static member methods.
5081 @cindex @code{ms_abi} function attribute, x86
5082 @cindex @code{sysv_abi} function attribute, x86
5084 On 32-bit and 64-bit x86 targets, you can use an ABI attribute
5085 to indicate which calling convention should be used for a function. The
5086 @code{ms_abi} attribute tells the compiler to use the Microsoft ABI,
5087 while the @code{sysv_abi} attribute tells the compiler to use the ABI
5088 used on GNU/Linux and other systems. The default is to use the Microsoft ABI
5089 when targeting Windows. On all other systems, the default is the x86/AMD ABI.
5091 Note, the @code{ms_abi} attribute for Microsoft Windows 64-bit targets currently
5092 requires the @option{-maccumulate-outgoing-args} option.
5094 @item callee_pop_aggregate_return (@var{number})
5095 @cindex @code{callee_pop_aggregate_return} function attribute, x86
5097 On x86-32 targets, you can use this attribute to control how
5098 aggregates are returned in memory. If the caller is responsible for
5099 popping the hidden pointer together with the rest of the arguments, specify
5100 @var{number} equal to zero. If callee is responsible for popping the
5101 hidden pointer, specify @var{number} equal to one.
5103 The default x86-32 ABI assumes that the callee pops the
5104 stack for hidden pointer. However, on x86-32 Microsoft Windows targets,
5105 the compiler assumes that the
5106 caller pops the stack for hidden pointer.
5108 @item ms_hook_prologue
5109 @cindex @code{ms_hook_prologue} function attribute, x86
5111 On 32-bit and 64-bit x86 targets, you can use
5112 this function attribute to make GCC generate the ``hot-patching'' function
5113 prologue used in Win32 API functions in Microsoft Windows XP Service Pack 2
5116 @item regparm (@var{number})
5117 @cindex @code{regparm} function attribute, x86
5118 @cindex functions that are passed arguments in registers on x86-32
5119 On x86-32 targets, the @code{regparm} attribute causes the compiler to
5120 pass arguments number one to @var{number} if they are of integral type
5121 in registers EAX, EDX, and ECX instead of on the stack. Functions that
5122 take a variable number of arguments continue to be passed all of their
5123 arguments on the stack.
5125 Beware that on some ELF systems this attribute is unsuitable for
5126 global functions in shared libraries with lazy binding (which is the
5127 default). Lazy binding sends the first call via resolving code in
5128 the loader, which might assume EAX, EDX and ECX can be clobbered, as
5129 per the standard calling conventions. Solaris 8 is affected by this.
5130 Systems with the GNU C Library version 2.1 or higher
5131 and FreeBSD are believed to be
5132 safe since the loaders there save EAX, EDX and ECX. (Lazy binding can be
5133 disabled with the linker or the loader if desired, to avoid the
5137 @cindex @code{sseregparm} function attribute, x86
5138 On x86-32 targets with SSE support, the @code{sseregparm} attribute
5139 causes the compiler to pass up to 3 floating-point arguments in
5140 SSE registers instead of on the stack. Functions that take a
5141 variable number of arguments continue to pass all of their
5142 floating-point arguments on the stack.
5144 @item force_align_arg_pointer
5145 @cindex @code{force_align_arg_pointer} function attribute, x86
5146 On x86 targets, the @code{force_align_arg_pointer} attribute may be
5147 applied to individual function definitions, generating an alternate
5148 prologue and epilogue that realigns the run-time stack if necessary.
5149 This supports mixing legacy codes that run with a 4-byte aligned stack
5150 with modern codes that keep a 16-byte stack for SSE compatibility.
5153 @cindex @code{stdcall} function attribute, x86-32
5154 @cindex functions that pop the argument stack on x86-32
5155 On x86-32 targets, the @code{stdcall} attribute causes the compiler to
5156 assume that the called function pops off the stack space used to
5157 pass arguments, unless it takes a variable number of arguments.
5159 @item target (@var{options})
5160 @cindex @code{target} function attribute
5161 As discussed in @ref{Common Function Attributes}, this attribute
5162 allows specification of target-specific compilation options.
5164 On the x86, the following options are allowed:
5168 @cindex @code{target("abm")} function attribute, x86
5169 Enable/disable the generation of the advanced bit instructions.
5173 @cindex @code{target("aes")} function attribute, x86
5174 Enable/disable the generation of the AES instructions.
5177 @cindex @code{target("default")} function attribute, x86
5178 @xref{Function Multiversioning}, where it is used to specify the
5179 default function version.
5183 @cindex @code{target("mmx")} function attribute, x86
5184 Enable/disable the generation of the MMX instructions.
5188 @cindex @code{target("pclmul")} function attribute, x86
5189 Enable/disable the generation of the PCLMUL instructions.
5193 @cindex @code{target("popcnt")} function attribute, x86
5194 Enable/disable the generation of the POPCNT instruction.
5198 @cindex @code{target("sse")} function attribute, x86
5199 Enable/disable the generation of the SSE instructions.
5203 @cindex @code{target("sse2")} function attribute, x86
5204 Enable/disable the generation of the SSE2 instructions.
5208 @cindex @code{target("sse3")} function attribute, x86
5209 Enable/disable the generation of the SSE3 instructions.
5213 @cindex @code{target("sse4")} function attribute, x86
5214 Enable/disable the generation of the SSE4 instructions (both SSE4.1
5219 @cindex @code{target("sse4.1")} function attribute, x86
5220 Enable/disable the generation of the sse4.1 instructions.
5224 @cindex @code{target("sse4.2")} function attribute, x86
5225 Enable/disable the generation of the sse4.2 instructions.
5229 @cindex @code{target("sse4a")} function attribute, x86
5230 Enable/disable the generation of the SSE4A instructions.
5234 @cindex @code{target("fma4")} function attribute, x86
5235 Enable/disable the generation of the FMA4 instructions.
5239 @cindex @code{target("xop")} function attribute, x86
5240 Enable/disable the generation of the XOP instructions.
5244 @cindex @code{target("lwp")} function attribute, x86
5245 Enable/disable the generation of the LWP instructions.
5249 @cindex @code{target("ssse3")} function attribute, x86
5250 Enable/disable the generation of the SSSE3 instructions.
5254 @cindex @code{target("cld")} function attribute, x86
5255 Enable/disable the generation of the CLD before string moves.
5257 @item fancy-math-387
5258 @itemx no-fancy-math-387
5259 @cindex @code{target("fancy-math-387")} function attribute, x86
5260 Enable/disable the generation of the @code{sin}, @code{cos}, and
5261 @code{sqrt} instructions on the 387 floating-point unit.
5264 @itemx no-fused-madd
5265 @cindex @code{target("fused-madd")} function attribute, x86
5266 Enable/disable the generation of the fused multiply/add instructions.
5270 @cindex @code{target("ieee-fp")} function attribute, x86
5271 Enable/disable the generation of floating point that depends on IEEE arithmetic.
5273 @item inline-all-stringops
5274 @itemx no-inline-all-stringops
5275 @cindex @code{target("inline-all-stringops")} function attribute, x86
5276 Enable/disable inlining of string operations.
5278 @item inline-stringops-dynamically
5279 @itemx no-inline-stringops-dynamically
5280 @cindex @code{target("inline-stringops-dynamically")} function attribute, x86
5281 Enable/disable the generation of the inline code to do small string
5282 operations and calling the library routines for large operations.
5284 @item align-stringops
5285 @itemx no-align-stringops
5286 @cindex @code{target("align-stringops")} function attribute, x86
5287 Do/do not align destination of inlined string operations.
5291 @cindex @code{target("recip")} function attribute, x86
5292 Enable/disable the generation of RCPSS, RCPPS, RSQRTSS and RSQRTPS
5293 instructions followed an additional Newton-Raphson step instead of
5294 doing a floating-point division.
5296 @item arch=@var{ARCH}
5297 @cindex @code{target("arch=@var{ARCH}")} function attribute, x86
5298 Specify the architecture to generate code for in compiling the function.
5300 @item tune=@var{TUNE}
5301 @cindex @code{target("tune=@var{TUNE}")} function attribute, x86
5302 Specify the architecture to tune for in compiling the function.
5304 @item fpmath=@var{FPMATH}
5305 @cindex @code{target("fpmath=@var{FPMATH}")} function attribute, x86
5306 Specify which floating-point unit to use. You must specify the
5307 @code{target("fpmath=sse,387")} option as
5308 @code{target("fpmath=sse+387")} because the comma would separate
5312 On the x86, the inliner does not inline a
5313 function that has different target options than the caller, unless the
5314 callee has a subset of the target options of the caller. For example
5315 a function declared with @code{target("sse3")} can inline a function
5316 with @code{target("sse2")}, since @code{-msse3} implies @code{-msse2}.
5319 @node Xstormy16 Function Attributes
5320 @subsection Xstormy16 Function Attributes
5322 These function attributes are supported by the Xstormy16 back end:
5326 @cindex @code{interrupt} function attribute, Xstormy16
5327 Use this attribute to indicate
5328 that the specified function is an interrupt handler. The compiler generates
5329 function entry and exit sequences suitable for use in an interrupt handler
5330 when this attribute is present.
5333 @node Variable Attributes
5334 @section Specifying Attributes of Variables
5335 @cindex attribute of variables
5336 @cindex variable attributes
5338 The keyword @code{__attribute__} allows you to specify special
5339 attributes of variables or structure fields. This keyword is followed
5340 by an attribute specification inside double parentheses. Some
5341 attributes are currently defined generically for variables.
5342 Other attributes are defined for variables on particular target
5343 systems. Other attributes are available for functions
5344 (@pxref{Function Attributes}), labels (@pxref{Label Attributes}),
5345 enumerators (@pxref{Enumerator Attributes}), and for types
5346 (@pxref{Type Attributes}).
5347 Other front ends might define more attributes
5348 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
5350 @xref{Attribute Syntax}, for details of the exact syntax for using
5354 * Common Variable Attributes::
5355 * AVR Variable Attributes::
5356 * Blackfin Variable Attributes::
5357 * H8/300 Variable Attributes::
5358 * IA-64 Variable Attributes::
5359 * M32R/D Variable Attributes::
5360 * MeP Variable Attributes::
5361 * Microsoft Windows Variable Attributes::
5362 * MSP430 Variable Attributes::
5363 * PowerPC Variable Attributes::
5364 * SPU Variable Attributes::
5365 * x86 Variable Attributes::
5366 * Xstormy16 Variable Attributes::
5369 @node Common Variable Attributes
5370 @subsection Common Variable Attributes
5372 The following attributes are supported on most targets.
5375 @cindex @code{aligned} variable attribute
5376 @item aligned (@var{alignment})
5377 This attribute specifies a minimum alignment for the variable or
5378 structure field, measured in bytes. For example, the declaration:
5381 int x __attribute__ ((aligned (16))) = 0;
5385 causes the compiler to allocate the global variable @code{x} on a
5386 16-byte boundary. On a 68040, this could be used in conjunction with
5387 an @code{asm} expression to access the @code{move16} instruction which
5388 requires 16-byte aligned operands.
5390 You can also specify the alignment of structure fields. For example, to
5391 create a double-word aligned @code{int} pair, you could write:
5394 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
5398 This is an alternative to creating a union with a @code{double} member,
5399 which forces the union to be double-word aligned.
5401 As in the preceding examples, you can explicitly specify the alignment
5402 (in bytes) that you wish the compiler to use for a given variable or
5403 structure field. Alternatively, you can leave out the alignment factor
5404 and just ask the compiler to align a variable or field to the
5405 default alignment for the target architecture you are compiling for.
5406 The default alignment is sufficient for all scalar types, but may not be
5407 enough for all vector types on a target that supports vector operations.
5408 The default alignment is fixed for a particular target ABI.
5410 GCC also provides a target specific macro @code{__BIGGEST_ALIGNMENT__},
5411 which is the largest alignment ever used for any data type on the
5412 target machine you are compiling for. For example, you could write:
5415 short array[3] __attribute__ ((aligned (__BIGGEST_ALIGNMENT__)));
5418 The compiler automatically sets the alignment for the declared
5419 variable or field to @code{__BIGGEST_ALIGNMENT__}. Doing this can
5420 often make copy operations more efficient, because the compiler can
5421 use whatever instructions copy the biggest chunks of memory when
5422 performing copies to or from the variables or fields that you have
5423 aligned this way. Note that the value of @code{__BIGGEST_ALIGNMENT__}
5424 may change depending on command-line options.
5426 When used on a struct, or struct member, the @code{aligned} attribute can
5427 only increase the alignment; in order to decrease it, the @code{packed}
5428 attribute must be specified as well. When used as part of a typedef, the
5429 @code{aligned} attribute can both increase and decrease alignment, and
5430 specifying the @code{packed} attribute generates a warning.
5432 Note that the effectiveness of @code{aligned} attributes may be limited
5433 by inherent limitations in your linker. On many systems, the linker is
5434 only able to arrange for variables to be aligned up to a certain maximum
5435 alignment. (For some linkers, the maximum supported alignment may
5436 be very very small.) If your linker is only able to align variables
5437 up to a maximum of 8-byte alignment, then specifying @code{aligned(16)}
5438 in an @code{__attribute__} still only provides you with 8-byte
5439 alignment. See your linker documentation for further information.
5441 The @code{aligned} attribute can also be used for functions
5442 (@pxref{Common Function Attributes}.)
5444 @item cleanup (@var{cleanup_function})
5445 @cindex @code{cleanup} variable attribute
5446 The @code{cleanup} attribute runs a function when the variable goes
5447 out of scope. This attribute can only be applied to auto function
5448 scope variables; it may not be applied to parameters or variables
5449 with static storage duration. The function must take one parameter,
5450 a pointer to a type compatible with the variable. The return value
5451 of the function (if any) is ignored.
5453 If @option{-fexceptions} is enabled, then @var{cleanup_function}
5454 is run during the stack unwinding that happens during the
5455 processing of the exception. Note that the @code{cleanup} attribute
5456 does not allow the exception to be caught, only to perform an action.
5457 It is undefined what happens if @var{cleanup_function} does not
5462 @cindex @code{common} variable attribute
5463 @cindex @code{nocommon} variable attribute
5466 The @code{common} attribute requests GCC to place a variable in
5467 ``common'' storage. The @code{nocommon} attribute requests the
5468 opposite---to allocate space for it directly.
5470 These attributes override the default chosen by the
5471 @option{-fno-common} and @option{-fcommon} flags respectively.
5474 @itemx deprecated (@var{msg})
5475 @cindex @code{deprecated} variable attribute
5476 The @code{deprecated} attribute results in a warning if the variable
5477 is used anywhere in the source file. This is useful when identifying
5478 variables that are expected to be removed in a future version of a
5479 program. The warning also includes the location of the declaration
5480 of the deprecated variable, to enable users to easily find further
5481 information about why the variable is deprecated, or what they should
5482 do instead. Note that the warning only occurs for uses:
5485 extern int old_var __attribute__ ((deprecated));
5487 int new_fn () @{ return old_var; @}
5491 results in a warning on line 3 but not line 2. The optional @var{msg}
5492 argument, which must be a string, is printed in the warning if
5495 The @code{deprecated} attribute can also be used for functions and
5496 types (@pxref{Common Function Attributes},
5497 @pxref{Common Type Attributes}).
5499 @item mode (@var{mode})
5500 @cindex @code{mode} variable attribute
5501 This attribute specifies the data type for the declaration---whichever
5502 type corresponds to the mode @var{mode}. This in effect lets you
5503 request an integer or floating-point type according to its width.
5505 You may also specify a mode of @code{byte} or @code{__byte__} to
5506 indicate the mode corresponding to a one-byte integer, @code{word} or
5507 @code{__word__} for the mode of a one-word integer, and @code{pointer}
5508 or @code{__pointer__} for the mode used to represent pointers.
5511 @cindex @code{packed} variable attribute
5512 The @code{packed} attribute specifies that a variable or structure field
5513 should have the smallest possible alignment---one byte for a variable,
5514 and one bit for a field, unless you specify a larger value with the
5515 @code{aligned} attribute.
5517 Here is a structure in which the field @code{x} is packed, so that it
5518 immediately follows @code{a}:
5524 int x[2] __attribute__ ((packed));
5528 @emph{Note:} The 4.1, 4.2 and 4.3 series of GCC ignore the
5529 @code{packed} attribute on bit-fields of type @code{char}. This has
5530 been fixed in GCC 4.4 but the change can lead to differences in the
5531 structure layout. See the documentation of
5532 @option{-Wpacked-bitfield-compat} for more information.
5534 @item section ("@var{section-name}")
5535 @cindex @code{section} variable attribute
5536 Normally, the compiler places the objects it generates in sections like
5537 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
5538 or you need certain particular variables to appear in special sections,
5539 for example to map to special hardware. The @code{section}
5540 attribute specifies that a variable (or function) lives in a particular
5541 section. For example, this small program uses several specific section names:
5544 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
5545 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
5546 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
5547 int init_data __attribute__ ((section ("INITDATA")));
5551 /* @r{Initialize stack pointer} */
5552 init_sp (stack + sizeof (stack));
5554 /* @r{Initialize initialized data} */
5555 memcpy (&init_data, &data, &edata - &data);
5557 /* @r{Turn on the serial ports} */
5564 Use the @code{section} attribute with
5565 @emph{global} variables and not @emph{local} variables,
5566 as shown in the example.
5568 You may use the @code{section} attribute with initialized or
5569 uninitialized global variables but the linker requires
5570 each object be defined once, with the exception that uninitialized
5571 variables tentatively go in the @code{common} (or @code{bss}) section
5572 and can be multiply ``defined''. Using the @code{section} attribute
5573 changes what section the variable goes into and may cause the
5574 linker to issue an error if an uninitialized variable has multiple
5575 definitions. You can force a variable to be initialized with the
5576 @option{-fno-common} flag or the @code{nocommon} attribute.
5578 Some file formats do not support arbitrary sections so the @code{section}
5579 attribute is not available on all platforms.
5580 If you need to map the entire contents of a module to a particular
5581 section, consider using the facilities of the linker instead.
5583 @item tls_model ("@var{tls_model}")
5584 @cindex @code{tls_model} variable attribute
5585 The @code{tls_model} attribute sets thread-local storage model
5586 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
5587 overriding @option{-ftls-model=} command-line switch on a per-variable
5589 The @var{tls_model} argument should be one of @code{global-dynamic},
5590 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
5592 Not all targets support this attribute.
5595 @cindex @code{unused} variable attribute
5596 This attribute, attached to a variable, means that the variable is meant
5597 to be possibly unused. GCC does not produce a warning for this
5601 @cindex @code{used} variable attribute
5602 This attribute, attached to a variable with static storage, means that
5603 the variable must be emitted even if it appears that the variable is not
5606 When applied to a static data member of a C++ class template, the
5607 attribute also means that the member is instantiated if the
5608 class itself is instantiated.
5610 @item vector_size (@var{bytes})
5611 @cindex @code{vector_size} variable attribute
5612 This attribute specifies the vector size for the variable, measured in
5613 bytes. For example, the declaration:
5616 int foo __attribute__ ((vector_size (16)));
5620 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
5621 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
5622 4 units of 4 bytes), the corresponding mode of @code{foo} is V4SI@.
5624 This attribute is only applicable to integral and float scalars,
5625 although arrays, pointers, and function return values are allowed in
5626 conjunction with this construct.
5628 Aggregates with this attribute are invalid, even if they are of the same
5629 size as a corresponding scalar. For example, the declaration:
5632 struct S @{ int a; @};
5633 struct S __attribute__ ((vector_size (16))) foo;
5637 is invalid even if the size of the structure is the same as the size of
5641 @cindex @code{weak} variable attribute
5642 The @code{weak} attribute is described in
5643 @ref{Common Function Attributes}.
5647 @node AVR Variable Attributes
5648 @subsection AVR Variable Attributes
5652 @cindex @code{progmem} variable attribute, AVR
5653 The @code{progmem} attribute is used on the AVR to place read-only
5654 data in the non-volatile program memory (flash). The @code{progmem}
5655 attribute accomplishes this by putting respective variables into a
5656 section whose name starts with @code{.progmem}.
5658 This attribute works similar to the @code{section} attribute
5659 but adds additional checking. Notice that just like the
5660 @code{section} attribute, @code{progmem} affects the location
5661 of the data but not how this data is accessed.
5663 In order to read data located with the @code{progmem} attribute
5664 (inline) assembler must be used.
5666 /* Use custom macros from @w{@uref{http://nongnu.org/avr-libc/user-manual/,AVR-LibC}} */
5667 #include <avr/pgmspace.h>
5669 /* Locate var in flash memory */
5670 const int var[2] PROGMEM = @{ 1, 2 @};
5672 int read_var (int i)
5674 /* Access var[] by accessor macro from avr/pgmspace.h */
5675 return (int) pgm_read_word (& var[i]);
5679 AVR is a Harvard architecture processor and data and read-only data
5680 normally resides in the data memory (RAM).
5682 See also the @ref{AVR Named Address Spaces} section for
5683 an alternate way to locate and access data in flash memory.
5686 @itemx io (@var{addr})
5687 @cindex @code{io} variable attribute, AVR
5688 Variables with the @code{io} attribute are used to address
5689 memory-mapped peripherals in the io address range.
5690 If an address is specified, the variable
5691 is assigned that address, and the value is interpreted as an
5692 address in the data address space.
5696 volatile int porta __attribute__((io (0x22)));
5699 The address specified in the address in the data address range.
5701 Otherwise, the variable it is not assigned an address, but the
5702 compiler will still use in/out instructions where applicable,
5703 assuming some other module assigns an address in the io address range.
5707 extern volatile int porta __attribute__((io));
5711 @itemx io_low (@var{addr})
5712 @cindex @code{io_low} variable attribute, AVR
5713 This is like the @code{io} attribute, but additionally it informs the
5714 compiler that the object lies in the lower half of the I/O area,
5715 allowing the use of @code{cbi}, @code{sbi}, @code{sbic} and @code{sbis}
5719 @itemx address (@var{addr})
5720 @cindex @code{address} variable attribute, AVR
5721 Variables with the @code{address} attribute are used to address
5722 memory-mapped peripherals that may lie outside the io address range.
5725 volatile int porta __attribute__((address (0x600)));
5730 @node Blackfin Variable Attributes
5731 @subsection Blackfin Variable Attributes
5733 Three attributes are currently defined for the Blackfin.
5739 @cindex @code{l1_data} variable attribute, Blackfin
5740 @cindex @code{l1_data_A} variable attribute, Blackfin
5741 @cindex @code{l1_data_B} variable attribute, Blackfin
5742 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
5743 Variables with @code{l1_data} attribute are put into the specific section
5744 named @code{.l1.data}. Those with @code{l1_data_A} attribute are put into
5745 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
5746 attribute are put into the specific section named @code{.l1.data.B}.
5749 @cindex @code{l2} variable attribute, Blackfin
5750 Use this attribute on the Blackfin to place the variable into L2 SRAM.
5751 Variables with @code{l2} attribute are put into the specific section
5752 named @code{.l2.data}.
5755 @node H8/300 Variable Attributes
5756 @subsection H8/300 Variable Attributes
5758 These variable attributes are available for H8/300 targets:
5762 @cindex @code{eightbit_data} variable attribute, H8/300
5763 @cindex eight-bit data on the H8/300, H8/300H, and H8S
5764 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
5765 variable should be placed into the eight-bit data section.
5766 The compiler generates more efficient code for certain operations
5767 on data in the eight-bit data area. Note the eight-bit data area is limited to
5770 You must use GAS and GLD from GNU binutils version 2.7 or later for
5771 this attribute to work correctly.
5774 @cindex @code{tiny_data} variable attribute, H8/300
5775 @cindex tiny data section on the H8/300H and H8S
5776 Use this attribute on the H8/300H and H8S to indicate that the specified
5777 variable should be placed into the tiny data section.
5778 The compiler generates more efficient code for loads and stores
5779 on data in the tiny data section. Note the tiny data area is limited to
5780 slightly under 32KB of data.
5784 @node IA-64 Variable Attributes
5785 @subsection IA-64 Variable Attributes
5787 The IA-64 back end supports the following variable attribute:
5790 @item model (@var{model-name})
5791 @cindex @code{model} variable attribute, IA-64
5793 On IA-64, use this attribute to set the addressability of an object.
5794 At present, the only supported identifier for @var{model-name} is
5795 @code{small}, indicating addressability via ``small'' (22-bit)
5796 addresses (so that their addresses can be loaded with the @code{addl}
5797 instruction). Caveat: such addressing is by definition not position
5798 independent and hence this attribute must not be used for objects
5799 defined by shared libraries.
5803 @node M32R/D Variable Attributes
5804 @subsection M32R/D Variable Attributes
5806 One attribute is currently defined for the M32R/D@.
5809 @item model (@var{model-name})
5810 @cindex @code{model-name} variable attribute, M32R/D
5811 @cindex variable addressability on the M32R/D
5812 Use this attribute on the M32R/D to set the addressability of an object.
5813 The identifier @var{model-name} is one of @code{small}, @code{medium},
5814 or @code{large}, representing each of the code models.
5816 Small model objects live in the lower 16MB of memory (so that their
5817 addresses can be loaded with the @code{ld24} instruction).
5819 Medium and large model objects may live anywhere in the 32-bit address space
5820 (the compiler generates @code{seth/add3} instructions to load their
5824 @node MeP Variable Attributes
5825 @subsection MeP Variable Attributes
5827 The MeP target has a number of addressing modes and busses. The
5828 @code{near} space spans the standard memory space's first 16 megabytes
5829 (24 bits). The @code{far} space spans the entire 32-bit memory space.
5830 The @code{based} space is a 128-byte region in the memory space that
5831 is addressed relative to the @code{$tp} register. The @code{tiny}
5832 space is a 65536-byte region relative to the @code{$gp} register. In
5833 addition to these memory regions, the MeP target has a separate 16-bit
5834 control bus which is specified with @code{cb} attributes.
5839 @cindex @code{based} variable attribute, MeP
5840 Any variable with the @code{based} attribute is assigned to the
5841 @code{.based} section, and is accessed with relative to the
5842 @code{$tp} register.
5845 @cindex @code{tiny} variable attribute, MeP
5846 Likewise, the @code{tiny} attribute assigned variables to the
5847 @code{.tiny} section, relative to the @code{$gp} register.
5850 @cindex @code{near} variable attribute, MeP
5851 Variables with the @code{near} attribute are assumed to have addresses
5852 that fit in a 24-bit addressing mode. This is the default for large
5853 variables (@code{-mtiny=4} is the default) but this attribute can
5854 override @code{-mtiny=} for small variables, or override @code{-ml}.
5857 @cindex @code{far} variable attribute, MeP
5858 Variables with the @code{far} attribute are addressed using a full
5859 32-bit address. Since this covers the entire memory space, this
5860 allows modules to make no assumptions about where variables might be
5864 @cindex @code{io} variable attribute, MeP
5865 @itemx io (@var{addr})
5866 Variables with the @code{io} attribute are used to address
5867 memory-mapped peripherals. If an address is specified, the variable
5868 is assigned that address, else it is not assigned an address (it is
5869 assumed some other module assigns an address). Example:
5872 int timer_count __attribute__((io(0x123)));
5876 @itemx cb (@var{addr})
5877 @cindex @code{cb} variable attribute, MeP
5878 Variables with the @code{cb} attribute are used to access the control
5879 bus, using special instructions. @code{addr} indicates the control bus
5883 int cpu_clock __attribute__((cb(0x123)));
5888 @node Microsoft Windows Variable Attributes
5889 @subsection Microsoft Windows Variable Attributes
5891 You can use these attributes on Microsoft Windows targets.
5892 @ref{x86 Variable Attributes} for additional Windows compatibility
5893 attributes available on all x86 targets.
5898 @cindex @code{dllimport} variable attribute
5899 @cindex @code{dllexport} variable attribute
5900 The @code{dllimport} and @code{dllexport} attributes are described in
5901 @ref{Microsoft Windows Function Attributes}.
5904 @cindex @code{selectany} variable attribute
5905 The @code{selectany} attribute causes an initialized global variable to
5906 have link-once semantics. When multiple definitions of the variable are
5907 encountered by the linker, the first is selected and the remainder are
5908 discarded. Following usage by the Microsoft compiler, the linker is told
5909 @emph{not} to warn about size or content differences of the multiple
5912 Although the primary usage of this attribute is for POD types, the
5913 attribute can also be applied to global C++ objects that are initialized
5914 by a constructor. In this case, the static initialization and destruction
5915 code for the object is emitted in each translation defining the object,
5916 but the calls to the constructor and destructor are protected by a
5917 link-once guard variable.
5919 The @code{selectany} attribute is only available on Microsoft Windows
5920 targets. You can use @code{__declspec (selectany)} as a synonym for
5921 @code{__attribute__ ((selectany))} for compatibility with other
5925 @cindex @code{shared} variable attribute
5926 On Microsoft Windows, in addition to putting variable definitions in a named
5927 section, the section can also be shared among all running copies of an
5928 executable or DLL@. For example, this small program defines shared data
5929 by putting it in a named section @code{shared} and marking the section
5933 int foo __attribute__((section ("shared"), shared)) = 0;
5938 /* @r{Read and write foo. All running
5939 copies see the same value.} */
5945 You may only use the @code{shared} attribute along with @code{section}
5946 attribute with a fully-initialized global definition because of the way
5947 linkers work. See @code{section} attribute for more information.
5949 The @code{shared} attribute is only available on Microsoft Windows@.
5953 @node MSP430 Variable Attributes
5954 @subsection MSP430 Variable Attributes
5958 @cindex @code{noinit} MSP430 variable attribute
5959 Any data with the @code{noinit} attribute will not be initialised by
5960 the C runtime startup code, or the program loader. Not initialising
5961 data in this way can reduce program startup times.
5964 @cindex @code{persistent} MSP430 variable attribute
5965 Any variable with the @code{persistent} attribute will not be
5966 initialised by the C runtime startup code. Instead its value will be
5967 set once, when the application is loaded, and then never initialised
5968 again, even if the processor is reset or the program restarts.
5969 Persistent data is intended to be placed into FLASH RAM, where its
5970 value will be retained across resets. The linker script being used to
5971 create the application should ensure that persistent data is correctly
5977 @cindex @code{lower} memory region on the MSP430
5978 @cindex @code{upper} memory region on the MSP430
5979 @cindex @code{either} memory region on the MSP430
5980 These attributes are the same as the MSP430 function attributes of the
5981 same name. These attributes can be applied to both functions and
5985 @node PowerPC Variable Attributes
5986 @subsection PowerPC Variable Attributes
5988 Three attributes currently are defined for PowerPC configurations:
5989 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
5991 @cindex @code{ms_struct} variable attribute, PowerPC
5992 @cindex @code{gcc_struct} variable attribute, PowerPC
5993 For full documentation of the struct attributes please see the
5994 documentation in @ref{x86 Variable Attributes}.
5996 @cindex @code{altivec} variable attribute, PowerPC
5997 For documentation of @code{altivec} attribute please see the
5998 documentation in @ref{PowerPC Type Attributes}.
6000 @node SPU Variable Attributes
6001 @subsection SPU Variable Attributes
6003 @cindex @code{spu_vector} variable attribute, SPU
6004 The SPU supports the @code{spu_vector} attribute for variables. For
6005 documentation of this attribute please see the documentation in
6006 @ref{SPU Type Attributes}.
6008 @node x86 Variable Attributes
6009 @subsection x86 Variable Attributes
6011 Two attributes are currently defined for x86 configurations:
6012 @code{ms_struct} and @code{gcc_struct}.
6017 @cindex @code{ms_struct} variable attribute, x86
6018 @cindex @code{gcc_struct} variable attribute, x86
6020 If @code{packed} is used on a structure, or if bit-fields are used,
6021 it may be that the Microsoft ABI lays out the structure differently
6022 than the way GCC normally does. Particularly when moving packed
6023 data between functions compiled with GCC and the native Microsoft compiler
6024 (either via function call or as data in a file), it may be necessary to access
6027 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows x86
6028 compilers to match the native Microsoft compiler.
6030 The Microsoft structure layout algorithm is fairly simple with the exception
6031 of the bit-field packing.
6032 The padding and alignment of members of structures and whether a bit-field
6033 can straddle a storage-unit boundary are determine by these rules:
6036 @item Structure members are stored sequentially in the order in which they are
6037 declared: the first member has the lowest memory address and the last member
6040 @item Every data object has an alignment requirement. The alignment requirement
6041 for all data except structures, unions, and arrays is either the size of the
6042 object or the current packing size (specified with either the
6043 @code{aligned} attribute or the @code{pack} pragma),
6044 whichever is less. For structures, unions, and arrays,
6045 the alignment requirement is the largest alignment requirement of its members.
6046 Every object is allocated an offset so that:
6049 offset % alignment_requirement == 0
6052 @item Adjacent bit-fields are packed into the same 1-, 2-, or 4-byte allocation
6053 unit if the integral types are the same size and if the next bit-field fits
6054 into the current allocation unit without crossing the boundary imposed by the
6055 common alignment requirements of the bit-fields.
6058 MSVC interprets zero-length bit-fields in the following ways:
6061 @item If a zero-length bit-field is inserted between two bit-fields that
6062 are normally coalesced, the bit-fields are not coalesced.
6069 unsigned long bf_1 : 12;
6071 unsigned long bf_2 : 12;
6076 The size of @code{t1} is 8 bytes with the zero-length bit-field. If the
6077 zero-length bit-field were removed, @code{t1}'s size would be 4 bytes.
6079 @item If a zero-length bit-field is inserted after a bit-field, @code{foo}, and the
6080 alignment of the zero-length bit-field is greater than the member that follows it,
6081 @code{bar}, @code{bar} is aligned as the type of the zero-length bit-field.
6102 For @code{t2}, @code{bar} is placed at offset 2, rather than offset 1.
6103 Accordingly, the size of @code{t2} is 4. For @code{t3}, the zero-length
6104 bit-field does not affect the alignment of @code{bar} or, as a result, the size
6107 Taking this into account, it is important to note the following:
6110 @item If a zero-length bit-field follows a normal bit-field, the type of the
6111 zero-length bit-field may affect the alignment of the structure as whole. For
6112 example, @code{t2} has a size of 4 bytes, since the zero-length bit-field follows a
6113 normal bit-field, and is of type short.
6115 @item Even if a zero-length bit-field is not followed by a normal bit-field, it may
6116 still affect the alignment of the structure:
6127 Here, @code{t4} takes up 4 bytes.
6130 @item Zero-length bit-fields following non-bit-field members are ignored:
6142 Here, @code{t5} takes up 2 bytes.
6146 @node Xstormy16 Variable Attributes
6147 @subsection Xstormy16 Variable Attributes
6149 One attribute is currently defined for xstormy16 configurations:
6154 @cindex @code{below100} variable attribute, Xstormy16
6156 If a variable has the @code{below100} attribute (@code{BELOW100} is
6157 allowed also), GCC places the variable in the first 0x100 bytes of
6158 memory and use special opcodes to access it. Such variables are
6159 placed in either the @code{.bss_below100} section or the
6160 @code{.data_below100} section.
6164 @node Type Attributes
6165 @section Specifying Attributes of Types
6166 @cindex attribute of types
6167 @cindex type attributes
6169 The keyword @code{__attribute__} allows you to specify special
6170 attributes of types. Some type attributes apply only to @code{struct}
6171 and @code{union} types, while others can apply to any type defined
6172 via a @code{typedef} declaration. Other attributes are defined for
6173 functions (@pxref{Function Attributes}), labels (@pxref{Label
6174 Attributes}), enumerators (@pxref{Enumerator Attributes}), and for
6175 variables (@pxref{Variable Attributes}).
6177 The @code{__attribute__} keyword is followed by an attribute specification
6178 inside double parentheses.
6180 You may specify type attributes in an enum, struct or union type
6181 declaration or definition by placing them immediately after the
6182 @code{struct}, @code{union} or @code{enum} keyword. A less preferred
6183 syntax is to place them just past the closing curly brace of the
6186 You can also include type attributes in a @code{typedef} declaration.
6187 @xref{Attribute Syntax}, for details of the exact syntax for using
6191 * Common Type Attributes::
6192 * ARM Type Attributes::
6193 * MeP Type Attributes::
6194 * PowerPC Type Attributes::
6195 * SPU Type Attributes::
6196 * x86 Type Attributes::
6199 @node Common Type Attributes
6200 @subsection Common Type Attributes
6202 The following type attributes are supported on most targets.
6205 @cindex @code{aligned} type attribute
6206 @item aligned (@var{alignment})
6207 This attribute specifies a minimum alignment (in bytes) for variables
6208 of the specified type. For example, the declarations:
6211 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
6212 typedef int more_aligned_int __attribute__ ((aligned (8)));
6216 force the compiler to ensure (as far as it can) that each variable whose
6217 type is @code{struct S} or @code{more_aligned_int} is allocated and
6218 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
6219 variables of type @code{struct S} aligned to 8-byte boundaries allows
6220 the compiler to use the @code{ldd} and @code{std} (doubleword load and
6221 store) instructions when copying one variable of type @code{struct S} to
6222 another, thus improving run-time efficiency.
6224 Note that the alignment of any given @code{struct} or @code{union} type
6225 is required by the ISO C standard to be at least a perfect multiple of
6226 the lowest common multiple of the alignments of all of the members of
6227 the @code{struct} or @code{union} in question. This means that you @emph{can}
6228 effectively adjust the alignment of a @code{struct} or @code{union}
6229 type by attaching an @code{aligned} attribute to any one of the members
6230 of such a type, but the notation illustrated in the example above is a
6231 more obvious, intuitive, and readable way to request the compiler to
6232 adjust the alignment of an entire @code{struct} or @code{union} type.
6234 As in the preceding example, you can explicitly specify the alignment
6235 (in bytes) that you wish the compiler to use for a given @code{struct}
6236 or @code{union} type. Alternatively, you can leave out the alignment factor
6237 and just ask the compiler to align a type to the maximum
6238 useful alignment for the target machine you are compiling for. For
6239 example, you could write:
6242 struct S @{ short f[3]; @} __attribute__ ((aligned));
6245 Whenever you leave out the alignment factor in an @code{aligned}
6246 attribute specification, the compiler automatically sets the alignment
6247 for the type to the largest alignment that is ever used for any data
6248 type on the target machine you are compiling for. Doing this can often
6249 make copy operations more efficient, because the compiler can use
6250 whatever instructions copy the biggest chunks of memory when performing
6251 copies to or from the variables that have types that you have aligned
6254 In the example above, if the size of each @code{short} is 2 bytes, then
6255 the size of the entire @code{struct S} type is 6 bytes. The smallest
6256 power of two that is greater than or equal to that is 8, so the
6257 compiler sets the alignment for the entire @code{struct S} type to 8
6260 Note that although you can ask the compiler to select a time-efficient
6261 alignment for a given type and then declare only individual stand-alone
6262 objects of that type, the compiler's ability to select a time-efficient
6263 alignment is primarily useful only when you plan to create arrays of
6264 variables having the relevant (efficiently aligned) type. If you
6265 declare or use arrays of variables of an efficiently-aligned type, then
6266 it is likely that your program also does pointer arithmetic (or
6267 subscripting, which amounts to the same thing) on pointers to the
6268 relevant type, and the code that the compiler generates for these
6269 pointer arithmetic operations is often more efficient for
6270 efficiently-aligned types than for other types.
6272 The @code{aligned} attribute can only increase the alignment; but you
6273 can decrease it by specifying @code{packed} as well. See below.
6275 Note that the effectiveness of @code{aligned} attributes may be limited
6276 by inherent limitations in your linker. On many systems, the linker is
6277 only able to arrange for variables to be aligned up to a certain maximum
6278 alignment. (For some linkers, the maximum supported alignment may
6279 be very very small.) If your linker is only able to align variables
6280 up to a maximum of 8-byte alignment, then specifying @code{aligned(16)}
6281 in an @code{__attribute__} still only provides you with 8-byte
6282 alignment. See your linker documentation for further information.
6284 @opindex fshort-enums
6285 Specifying this attribute for @code{struct} and @code{union} types is
6286 equivalent to specifying the @code{packed} attribute on each of the
6287 structure or union members. Specifying the @option{-fshort-enums}
6288 flag on the line is equivalent to specifying the @code{packed}
6289 attribute on all @code{enum} definitions.
6291 In the following example @code{struct my_packed_struct}'s members are
6292 packed closely together, but the internal layout of its @code{s} member
6293 is not packed---to do that, @code{struct my_unpacked_struct} needs to
6297 struct my_unpacked_struct
6303 struct __attribute__ ((__packed__)) my_packed_struct
6307 struct my_unpacked_struct s;
6311 You may only specify this attribute on the definition of an @code{enum},
6312 @code{struct} or @code{union}, not on a @code{typedef} that does not
6313 also define the enumerated type, structure or union.
6315 @item bnd_variable_size
6316 @cindex @code{bnd_variable_size} type attribute
6317 @cindex Pointer Bounds Checker attributes
6318 When applied to a structure field, this attribute tells Pointer
6319 Bounds Checker that the size of this field should not be computed
6320 using static type information. It may be used to mark variably-sized
6321 static array fields placed at the end of a structure.
6329 S *p = (S *)malloc (sizeof(S) + 100);
6330 p->data[10] = 0; //Bounds violation
6334 By using an attribute for the field we may avoid unwanted bound
6341 char data[1] __attribute__((bnd_variable_size));
6343 S *p = (S *)malloc (sizeof(S) + 100);
6344 p->data[10] = 0; //OK
6348 @itemx deprecated (@var{msg})
6349 @cindex @code{deprecated} type attribute
6350 The @code{deprecated} attribute results in a warning if the type
6351 is used anywhere in the source file. This is useful when identifying
6352 types that are expected to be removed in a future version of a program.
6353 If possible, the warning also includes the location of the declaration
6354 of the deprecated type, to enable users to easily find further
6355 information about why the type is deprecated, or what they should do
6356 instead. Note that the warnings only occur for uses and then only
6357 if the type is being applied to an identifier that itself is not being
6358 declared as deprecated.
6361 typedef int T1 __attribute__ ((deprecated));
6365 typedef T1 T3 __attribute__ ((deprecated));
6366 T3 z __attribute__ ((deprecated));
6370 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
6371 warning is issued for line 4 because T2 is not explicitly
6372 deprecated. Line 5 has no warning because T3 is explicitly
6373 deprecated. Similarly for line 6. The optional @var{msg}
6374 argument, which must be a string, is printed in the warning if
6377 The @code{deprecated} attribute can also be used for functions and
6378 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
6380 @item designated_init
6381 @cindex @code{designated_init} type attribute
6382 This attribute may only be applied to structure types. It indicates
6383 that any initialization of an object of this type must use designated
6384 initializers rather than positional initializers. The intent of this
6385 attribute is to allow the programmer to indicate that a structure's
6386 layout may change, and that therefore relying on positional
6387 initialization will result in future breakage.
6389 GCC emits warnings based on this attribute by default; use
6390 @option{-Wno-designated-init} to suppress them.
6393 @cindex @code{may_alias} type attribute
6394 Accesses through pointers to types with this attribute are not subject
6395 to type-based alias analysis, but are instead assumed to be able to alias
6396 any other type of objects.
6397 In the context of section 6.5 paragraph 7 of the C99 standard,
6398 an lvalue expression
6399 dereferencing such a pointer is treated like having a character type.
6400 See @option{-fstrict-aliasing} for more information on aliasing issues.
6401 This extension exists to support some vector APIs, in which pointers to
6402 one vector type are permitted to alias pointers to a different vector type.
6404 Note that an object of a type with this attribute does not have any
6410 typedef short __attribute__((__may_alias__)) short_a;
6416 short_a *b = (short_a *) &a;
6420 if (a == 0x12345678)
6428 If you replaced @code{short_a} with @code{short} in the variable
6429 declaration, the above program would abort when compiled with
6430 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
6434 @cindex @code{packed} type attribute
6435 This attribute, attached to @code{struct} or @code{union} type
6436 definition, specifies that each member (other than zero-width bit-fields)
6437 of the structure or union is placed to minimize the memory required. When
6438 attached to an @code{enum} definition, it indicates that the smallest
6439 integral type should be used.
6441 @item scalar_storage_order ("@var{endianness}")
6442 @cindex @code{scalar_storage_order} type attribute
6443 When attached to a @code{union} or a @code{struct}, this attribute sets
6444 the storage order, aka endianness, of the scalar fields of the type, as
6445 well as the array fields whose component is scalar. The supported
6446 endianness are @code{big-endian} and @code{little-endian}. The attribute
6447 has no effects on fields which are themselves a @code{union}, a @code{struct}
6448 or an array whose component is a @code{union} or a @code{struct}, and it is
6449 possible to have fields with a different scalar storage order than the
6452 This attribute is supported only for targets that use a uniform default
6453 scalar storage order (fortunately, most of them), i.e. targets that store
6454 the scalars either all in big-endian or all in little-endian.
6456 Additional restrictions are enforced for types with the reverse scalar
6457 storage order with regard to the scalar storage order of the target:
6460 @item Taking the address of a scalar field of a @code{union} or a
6461 @code{struct} with reverse scalar storage order is not permitted and will
6463 @item Taking the address of an array field, whose component is scalar, of
6464 a @code{union} or a @code{struct} with reverse scalar storage order is
6465 permitted but will yield a warning, unless @option{-Wno-scalar-storage-order}
6467 @item Taking the address of a @code{union} or a @code{struct} with reverse
6468 scalar storage order is permitted.
6471 These restrictions exist because the storage order attribute is lost when
6472 the address of a scalar or the address of an array with scalar component
6473 is taken, so storing indirectly through this address will generally not work.
6474 The second case is nevertheless allowed to be able to perform a block copy
6475 from or to the array.
6477 @item transparent_union
6478 @cindex @code{transparent_union} type attribute
6480 This attribute, attached to a @code{union} type definition, indicates
6481 that any function parameter having that union type causes calls to that
6482 function to be treated in a special way.
6484 First, the argument corresponding to a transparent union type can be of
6485 any type in the union; no cast is required. Also, if the union contains
6486 a pointer type, the corresponding argument can be a null pointer
6487 constant or a void pointer expression; and if the union contains a void
6488 pointer type, the corresponding argument can be any pointer expression.
6489 If the union member type is a pointer, qualifiers like @code{const} on
6490 the referenced type must be respected, just as with normal pointer
6493 Second, the argument is passed to the function using the calling
6494 conventions of the first member of the transparent union, not the calling
6495 conventions of the union itself. All members of the union must have the
6496 same machine representation; this is necessary for this argument passing
6499 Transparent unions are designed for library functions that have multiple
6500 interfaces for compatibility reasons. For example, suppose the
6501 @code{wait} function must accept either a value of type @code{int *} to
6502 comply with POSIX, or a value of type @code{union wait *} to comply with
6503 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
6504 @code{wait} would accept both kinds of arguments, but it would also
6505 accept any other pointer type and this would make argument type checking
6506 less useful. Instead, @code{<sys/wait.h>} might define the interface
6510 typedef union __attribute__ ((__transparent_union__))
6514 @} wait_status_ptr_t;
6516 pid_t wait (wait_status_ptr_t);
6520 This interface allows either @code{int *} or @code{union wait *}
6521 arguments to be passed, using the @code{int *} calling convention.
6522 The program can call @code{wait} with arguments of either type:
6525 int w1 () @{ int w; return wait (&w); @}
6526 int w2 () @{ union wait w; return wait (&w); @}
6530 With this interface, @code{wait}'s implementation might look like this:
6533 pid_t wait (wait_status_ptr_t p)
6535 return waitpid (-1, p.__ip, 0);
6540 @cindex @code{unused} type attribute
6541 When attached to a type (including a @code{union} or a @code{struct}),
6542 this attribute means that variables of that type are meant to appear
6543 possibly unused. GCC does not produce a warning for any variables of
6544 that type, even if the variable appears to do nothing. This is often
6545 the case with lock or thread classes, which are usually defined and then
6546 not referenced, but contain constructors and destructors that have
6547 nontrivial bookkeeping functions.
6550 @cindex @code{visibility} type attribute
6551 In C++, attribute visibility (@pxref{Function Attributes}) can also be
6552 applied to class, struct, union and enum types. Unlike other type
6553 attributes, the attribute must appear between the initial keyword and
6554 the name of the type; it cannot appear after the body of the type.
6556 Note that the type visibility is applied to vague linkage entities
6557 associated with the class (vtable, typeinfo node, etc.). In
6558 particular, if a class is thrown as an exception in one shared object
6559 and caught in another, the class must have default visibility.
6560 Otherwise the two shared objects are unable to use the same
6561 typeinfo node and exception handling will break.
6565 To specify multiple attributes, separate them by commas within the
6566 double parentheses: for example, @samp{__attribute__ ((aligned (16),
6569 @node ARM Type Attributes
6570 @subsection ARM Type Attributes
6572 @cindex @code{notshared} type attribute, ARM
6573 On those ARM targets that support @code{dllimport} (such as Symbian
6574 OS), you can use the @code{notshared} attribute to indicate that the
6575 virtual table and other similar data for a class should not be
6576 exported from a DLL@. For example:
6579 class __declspec(notshared) C @{
6581 __declspec(dllimport) C();
6585 __declspec(dllexport)
6590 In this code, @code{C::C} is exported from the current DLL, but the
6591 virtual table for @code{C} is not exported. (You can use
6592 @code{__attribute__} instead of @code{__declspec} if you prefer, but
6593 most Symbian OS code uses @code{__declspec}.)
6595 @node MeP Type Attributes
6596 @subsection MeP Type Attributes
6598 @cindex @code{based} type attribute, MeP
6599 @cindex @code{tiny} type attribute, MeP
6600 @cindex @code{near} type attribute, MeP
6601 @cindex @code{far} type attribute, MeP
6602 Many of the MeP variable attributes may be applied to types as well.
6603 Specifically, the @code{based}, @code{tiny}, @code{near}, and
6604 @code{far} attributes may be applied to either. The @code{io} and
6605 @code{cb} attributes may not be applied to types.
6607 @node PowerPC Type Attributes
6608 @subsection PowerPC Type Attributes
6610 Three attributes currently are defined for PowerPC configurations:
6611 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
6613 @cindex @code{ms_struct} type attribute, PowerPC
6614 @cindex @code{gcc_struct} type attribute, PowerPC
6615 For full documentation of the @code{ms_struct} and @code{gcc_struct}
6616 attributes please see the documentation in @ref{x86 Type Attributes}.
6618 @cindex @code{altivec} type attribute, PowerPC
6619 The @code{altivec} attribute allows one to declare AltiVec vector data
6620 types supported by the AltiVec Programming Interface Manual. The
6621 attribute requires an argument to specify one of three vector types:
6622 @code{vector__}, @code{pixel__} (always followed by unsigned short),
6623 and @code{bool__} (always followed by unsigned).
6626 __attribute__((altivec(vector__)))
6627 __attribute__((altivec(pixel__))) unsigned short
6628 __attribute__((altivec(bool__))) unsigned
6631 These attributes mainly are intended to support the @code{__vector},
6632 @code{__pixel}, and @code{__bool} AltiVec keywords.
6634 @node SPU Type Attributes
6635 @subsection SPU Type Attributes
6637 @cindex @code{spu_vector} type attribute, SPU
6638 The SPU supports the @code{spu_vector} attribute for types. This attribute
6639 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
6640 Language Extensions Specification. It is intended to support the
6641 @code{__vector} keyword.
6643 @node x86 Type Attributes
6644 @subsection x86 Type Attributes
6646 Two attributes are currently defined for x86 configurations:
6647 @code{ms_struct} and @code{gcc_struct}.
6653 @cindex @code{ms_struct} type attribute, x86
6654 @cindex @code{gcc_struct} type attribute, x86
6656 If @code{packed} is used on a structure, or if bit-fields are used
6657 it may be that the Microsoft ABI packs them differently
6658 than GCC normally packs them. Particularly when moving packed
6659 data between functions compiled with GCC and the native Microsoft compiler
6660 (either via function call or as data in a file), it may be necessary to access
6663 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows x86
6664 compilers to match the native Microsoft compiler.
6667 @node Label Attributes
6668 @section Label Attributes
6669 @cindex Label Attributes
6671 GCC allows attributes to be set on C labels. @xref{Attribute Syntax}, for
6672 details of the exact syntax for using attributes. Other attributes are
6673 available for functions (@pxref{Function Attributes}), variables
6674 (@pxref{Variable Attributes}), enumerators (@pxref{Enumerator Attributes}),
6675 and for types (@pxref{Type Attributes}).
6677 This example uses the @code{cold} label attribute to indicate the
6678 @code{ErrorHandling} branch is unlikely to be taken and that the
6679 @code{ErrorHandling} label is unused:
6683 asm goto ("some asm" : : : : NoError);
6685 /* This branch (the fall-through from the asm) is less commonly used */
6687 __attribute__((cold, unused)); /* Semi-colon is required here */
6692 printf("no error\n");
6698 @cindex @code{unused} label attribute
6699 This feature is intended for program-generated code that may contain
6700 unused labels, but which is compiled with @option{-Wall}. It is
6701 not normally appropriate to use in it human-written code, though it
6702 could be useful in cases where the code that jumps to the label is
6703 contained within an @code{#ifdef} conditional.
6706 @cindex @code{hot} label attribute
6707 The @code{hot} attribute on a label is used to inform the compiler that
6708 the path following the label is more likely than paths that are not so
6709 annotated. This attribute is used in cases where @code{__builtin_expect}
6710 cannot be used, for instance with computed goto or @code{asm goto}.
6713 @cindex @code{cold} label attribute
6714 The @code{cold} attribute on labels is used to inform the compiler that
6715 the path following the label is unlikely to be executed. This attribute
6716 is used in cases where @code{__builtin_expect} cannot be used, for instance
6717 with computed goto or @code{asm goto}.
6721 @node Enumerator Attributes
6722 @section Enumerator Attributes
6723 @cindex Enumerator Attributes
6725 GCC allows attributes to be set on enumerators. @xref{Attribute Syntax}, for
6726 details of the exact syntax for using attributes. Other attributes are
6727 available for functions (@pxref{Function Attributes}), variables
6728 (@pxref{Variable Attributes}), labels (@pxref{Label Attributes}),
6729 and for types (@pxref{Type Attributes}).
6731 This example uses the @code{deprecated} enumerator attribute to indicate the
6732 @code{oldval} enumerator is deprecated:
6736 oldval __attribute__((deprecated)),
6749 @cindex @code{deprecated} enumerator attribute
6750 The @code{deprecated} attribute results in a warning if the enumerator
6751 is used anywhere in the source file. This is useful when identifying
6752 enumerators that are expected to be removed in a future version of a
6753 program. The warning also includes the location of the declaration
6754 of the deprecated enumerator, to enable users to easily find further
6755 information about why the enumerator is deprecated, or what they should
6756 do instead. Note that the warnings only occurs for uses.
6760 @node Attribute Syntax
6761 @section Attribute Syntax
6762 @cindex attribute syntax
6764 This section describes the syntax with which @code{__attribute__} may be
6765 used, and the constructs to which attribute specifiers bind, for the C
6766 language. Some details may vary for C++ and Objective-C@. Because of
6767 infelicities in the grammar for attributes, some forms described here
6768 may not be successfully parsed in all cases.
6770 There are some problems with the semantics of attributes in C++. For
6771 example, there are no manglings for attributes, although they may affect
6772 code generation, so problems may arise when attributed types are used in
6773 conjunction with templates or overloading. Similarly, @code{typeid}
6774 does not distinguish between types with different attributes. Support
6775 for attributes in C++ may be restricted in future to attributes on
6776 declarations only, but not on nested declarators.
6778 @xref{Function Attributes}, for details of the semantics of attributes
6779 applying to functions. @xref{Variable Attributes}, for details of the
6780 semantics of attributes applying to variables. @xref{Type Attributes},
6781 for details of the semantics of attributes applying to structure, union
6782 and enumerated types.
6783 @xref{Label Attributes}, for details of the semantics of attributes
6785 @xref{Enumerator Attributes}, for details of the semantics of attributes
6786 applying to enumerators.
6788 An @dfn{attribute specifier} is of the form
6789 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
6790 is a possibly empty comma-separated sequence of @dfn{attributes}, where
6791 each attribute is one of the following:
6795 Empty. Empty attributes are ignored.
6799 (which may be an identifier such as @code{unused}, or a reserved
6800 word such as @code{const}).
6803 An attribute name followed by a parenthesized list of
6804 parameters for the attribute.
6805 These parameters take one of the following forms:
6809 An identifier. For example, @code{mode} attributes use this form.
6812 An identifier followed by a comma and a non-empty comma-separated list
6813 of expressions. For example, @code{format} attributes use this form.
6816 A possibly empty comma-separated list of expressions. For example,
6817 @code{format_arg} attributes use this form with the list being a single
6818 integer constant expression, and @code{alias} attributes use this form
6819 with the list being a single string constant.
6823 An @dfn{attribute specifier list} is a sequence of one or more attribute
6824 specifiers, not separated by any other tokens.
6826 You may optionally specify attribute names with @samp{__}
6827 preceding and following the name.
6828 This allows you to use them in header files without
6829 being concerned about a possible macro of the same name. For example,
6830 you may use the attribute name @code{__noreturn__} instead of @code{noreturn}.
6833 @subsubheading Label Attributes
6835 In GNU C, an attribute specifier list may appear after the colon following a
6836 label, other than a @code{case} or @code{default} label. GNU C++ only permits
6837 attributes on labels if the attribute specifier is immediately
6838 followed by a semicolon (i.e., the label applies to an empty
6839 statement). If the semicolon is missing, C++ label attributes are
6840 ambiguous, as it is permissible for a declaration, which could begin
6841 with an attribute list, to be labelled in C++. Declarations cannot be
6842 labelled in C90 or C99, so the ambiguity does not arise there.
6844 @subsubheading Enumerator Attributes
6846 In GNU C, an attribute specifier list may appear as part of an enumerator.
6847 The attribute goes after the enumeration constant, before @code{=}, if
6848 present. The optional attribute in the enumerator appertains to the
6849 enumeration constant. It is not possible to place the attribute after
6850 the constant expression, if present.
6852 @subsubheading Type Attributes
6854 An attribute specifier list may appear as part of a @code{struct},
6855 @code{union} or @code{enum} specifier. It may go either immediately
6856 after the @code{struct}, @code{union} or @code{enum} keyword, or after
6857 the closing brace. The former syntax is preferred.
6858 Where attribute specifiers follow the closing brace, they are considered
6859 to relate to the structure, union or enumerated type defined, not to any
6860 enclosing declaration the type specifier appears in, and the type
6861 defined is not complete until after the attribute specifiers.
6862 @c Otherwise, there would be the following problems: a shift/reduce
6863 @c conflict between attributes binding the struct/union/enum and
6864 @c binding to the list of specifiers/qualifiers; and "aligned"
6865 @c attributes could use sizeof for the structure, but the size could be
6866 @c changed later by "packed" attributes.
6869 @subsubheading All other attributes
6871 Otherwise, an attribute specifier appears as part of a declaration,
6872 counting declarations of unnamed parameters and type names, and relates
6873 to that declaration (which may be nested in another declaration, for
6874 example in the case of a parameter declaration), or to a particular declarator
6875 within a declaration. Where an
6876 attribute specifier is applied to a parameter declared as a function or
6877 an array, it should apply to the function or array rather than the
6878 pointer to which the parameter is implicitly converted, but this is not
6879 yet correctly implemented.
6881 Any list of specifiers and qualifiers at the start of a declaration may
6882 contain attribute specifiers, whether or not such a list may in that
6883 context contain storage class specifiers. (Some attributes, however,
6884 are essentially in the nature of storage class specifiers, and only make
6885 sense where storage class specifiers may be used; for example,
6886 @code{section}.) There is one necessary limitation to this syntax: the
6887 first old-style parameter declaration in a function definition cannot
6888 begin with an attribute specifier, because such an attribute applies to
6889 the function instead by syntax described below (which, however, is not
6890 yet implemented in this case). In some other cases, attribute
6891 specifiers are permitted by this grammar but not yet supported by the
6892 compiler. All attribute specifiers in this place relate to the
6893 declaration as a whole. In the obsolescent usage where a type of
6894 @code{int} is implied by the absence of type specifiers, such a list of
6895 specifiers and qualifiers may be an attribute specifier list with no
6896 other specifiers or qualifiers.
6898 At present, the first parameter in a function prototype must have some
6899 type specifier that is not an attribute specifier; this resolves an
6900 ambiguity in the interpretation of @code{void f(int
6901 (__attribute__((foo)) x))}, but is subject to change. At present, if
6902 the parentheses of a function declarator contain only attributes then
6903 those attributes are ignored, rather than yielding an error or warning
6904 or implying a single parameter of type int, but this is subject to
6907 An attribute specifier list may appear immediately before a declarator
6908 (other than the first) in a comma-separated list of declarators in a
6909 declaration of more than one identifier using a single list of
6910 specifiers and qualifiers. Such attribute specifiers apply
6911 only to the identifier before whose declarator they appear. For
6915 __attribute__((noreturn)) void d0 (void),
6916 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
6921 the @code{noreturn} attribute applies to all the functions
6922 declared; the @code{format} attribute only applies to @code{d1}.
6924 An attribute specifier list may appear immediately before the comma,
6925 @code{=} or semicolon terminating the declaration of an identifier other
6926 than a function definition. Such attribute specifiers apply
6927 to the declared object or function. Where an
6928 assembler name for an object or function is specified (@pxref{Asm
6929 Labels}), the attribute must follow the @code{asm}
6932 An attribute specifier list may, in future, be permitted to appear after
6933 the declarator in a function definition (before any old-style parameter
6934 declarations or the function body).
6936 Attribute specifiers may be mixed with type qualifiers appearing inside
6937 the @code{[]} of a parameter array declarator, in the C99 construct by
6938 which such qualifiers are applied to the pointer to which the array is
6939 implicitly converted. Such attribute specifiers apply to the pointer,
6940 not to the array, but at present this is not implemented and they are
6943 An attribute specifier list may appear at the start of a nested
6944 declarator. At present, there are some limitations in this usage: the
6945 attributes correctly apply to the declarator, but for most individual
6946 attributes the semantics this implies are not implemented.
6947 When attribute specifiers follow the @code{*} of a pointer
6948 declarator, they may be mixed with any type qualifiers present.
6949 The following describes the formal semantics of this syntax. It makes the
6950 most sense if you are familiar with the formal specification of
6951 declarators in the ISO C standard.
6953 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
6954 D1}, where @code{T} contains declaration specifiers that specify a type
6955 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
6956 contains an identifier @var{ident}. The type specified for @var{ident}
6957 for derived declarators whose type does not include an attribute
6958 specifier is as in the ISO C standard.
6960 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
6961 and the declaration @code{T D} specifies the type
6962 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
6963 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
6964 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
6966 If @code{D1} has the form @code{*
6967 @var{type-qualifier-and-attribute-specifier-list} D}, and the
6968 declaration @code{T D} specifies the type
6969 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
6970 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
6971 @var{type-qualifier-and-attribute-specifier-list} pointer to @var{Type}'' for
6977 void (__attribute__((noreturn)) ****f) (void);
6981 specifies the type ``pointer to pointer to pointer to pointer to
6982 non-returning function returning @code{void}''. As another example,
6985 char *__attribute__((aligned(8))) *f;
6989 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
6990 Note again that this does not work with most attributes; for example,
6991 the usage of @samp{aligned} and @samp{noreturn} attributes given above
6992 is not yet supported.
6994 For compatibility with existing code written for compiler versions that
6995 did not implement attributes on nested declarators, some laxity is
6996 allowed in the placing of attributes. If an attribute that only applies
6997 to types is applied to a declaration, it is treated as applying to
6998 the type of that declaration. If an attribute that only applies to
6999 declarations is applied to the type of a declaration, it is treated
7000 as applying to that declaration; and, for compatibility with code
7001 placing the attributes immediately before the identifier declared, such
7002 an attribute applied to a function return type is treated as
7003 applying to the function type, and such an attribute applied to an array
7004 element type is treated as applying to the array type. If an
7005 attribute that only applies to function types is applied to a
7006 pointer-to-function type, it is treated as applying to the pointer
7007 target type; if such an attribute is applied to a function return type
7008 that is not a pointer-to-function type, it is treated as applying
7009 to the function type.
7011 @node Function Prototypes
7012 @section Prototypes and Old-Style Function Definitions
7013 @cindex function prototype declarations
7014 @cindex old-style function definitions
7015 @cindex promotion of formal parameters
7017 GNU C extends ISO C to allow a function prototype to override a later
7018 old-style non-prototype definition. Consider the following example:
7021 /* @r{Use prototypes unless the compiler is old-fashioned.} */
7028 /* @r{Prototype function declaration.} */
7029 int isroot P((uid_t));
7031 /* @r{Old-style function definition.} */
7033 isroot (x) /* @r{??? lossage here ???} */
7040 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
7041 not allow this example, because subword arguments in old-style
7042 non-prototype definitions are promoted. Therefore in this example the
7043 function definition's argument is really an @code{int}, which does not
7044 match the prototype argument type of @code{short}.
7046 This restriction of ISO C makes it hard to write code that is portable
7047 to traditional C compilers, because the programmer does not know
7048 whether the @code{uid_t} type is @code{short}, @code{int}, or
7049 @code{long}. Therefore, in cases like these GNU C allows a prototype
7050 to override a later old-style definition. More precisely, in GNU C, a
7051 function prototype argument type overrides the argument type specified
7052 by a later old-style definition if the former type is the same as the
7053 latter type before promotion. Thus in GNU C the above example is
7054 equivalent to the following:
7067 GNU C++ does not support old-style function definitions, so this
7068 extension is irrelevant.
7071 @section C++ Style Comments
7073 @cindex C++ comments
7074 @cindex comments, C++ style
7076 In GNU C, you may use C++ style comments, which start with @samp{//} and
7077 continue until the end of the line. Many other C implementations allow
7078 such comments, and they are included in the 1999 C standard. However,
7079 C++ style comments are not recognized if you specify an @option{-std}
7080 option specifying a version of ISO C before C99, or @option{-ansi}
7081 (equivalent to @option{-std=c90}).
7084 @section Dollar Signs in Identifier Names
7086 @cindex dollar signs in identifier names
7087 @cindex identifier names, dollar signs in
7089 In GNU C, you may normally use dollar signs in identifier names.
7090 This is because many traditional C implementations allow such identifiers.
7091 However, dollar signs in identifiers are not supported on a few target
7092 machines, typically because the target assembler does not allow them.
7094 @node Character Escapes
7095 @section The Character @key{ESC} in Constants
7097 You can use the sequence @samp{\e} in a string or character constant to
7098 stand for the ASCII character @key{ESC}.
7101 @section Inquiring on Alignment of Types or Variables
7103 @cindex type alignment
7104 @cindex variable alignment
7106 The keyword @code{__alignof__} allows you to inquire about how an object
7107 is aligned, or the minimum alignment usually required by a type. Its
7108 syntax is just like @code{sizeof}.
7110 For example, if the target machine requires a @code{double} value to be
7111 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
7112 This is true on many RISC machines. On more traditional machine
7113 designs, @code{__alignof__ (double)} is 4 or even 2.
7115 Some machines never actually require alignment; they allow reference to any
7116 data type even at an odd address. For these machines, @code{__alignof__}
7117 reports the smallest alignment that GCC gives the data type, usually as
7118 mandated by the target ABI.
7120 If the operand of @code{__alignof__} is an lvalue rather than a type,
7121 its value is the required alignment for its type, taking into account
7122 any minimum alignment specified with GCC's @code{__attribute__}
7123 extension (@pxref{Variable Attributes}). For example, after this
7127 struct foo @{ int x; char y; @} foo1;
7131 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
7132 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
7134 It is an error to ask for the alignment of an incomplete type.
7138 @section An Inline Function is As Fast As a Macro
7139 @cindex inline functions
7140 @cindex integrating function code
7142 @cindex macros, inline alternative
7144 By declaring a function inline, you can direct GCC to make
7145 calls to that function faster. One way GCC can achieve this is to
7146 integrate that function's code into the code for its callers. This
7147 makes execution faster by eliminating the function-call overhead; in
7148 addition, if any of the actual argument values are constant, their
7149 known values may permit simplifications at compile time so that not
7150 all of the inline function's code needs to be included. The effect on
7151 code size is less predictable; object code may be larger or smaller
7152 with function inlining, depending on the particular case. You can
7153 also direct GCC to try to integrate all ``simple enough'' functions
7154 into their callers with the option @option{-finline-functions}.
7156 GCC implements three different semantics of declaring a function
7157 inline. One is available with @option{-std=gnu89} or
7158 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
7159 on all inline declarations, another when
7160 @option{-std=c99}, @option{-std=c11},
7161 @option{-std=gnu99} or @option{-std=gnu11}
7162 (without @option{-fgnu89-inline}), and the third
7163 is used when compiling C++.
7165 To declare a function inline, use the @code{inline} keyword in its
7166 declaration, like this:
7176 If you are writing a header file to be included in ISO C90 programs, write
7177 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
7179 The three types of inlining behave similarly in two important cases:
7180 when the @code{inline} keyword is used on a @code{static} function,
7181 like the example above, and when a function is first declared without
7182 using the @code{inline} keyword and then is defined with
7183 @code{inline}, like this:
7186 extern int inc (int *a);
7194 In both of these common cases, the program behaves the same as if you
7195 had not used the @code{inline} keyword, except for its speed.
7197 @cindex inline functions, omission of
7198 @opindex fkeep-inline-functions
7199 When a function is both inline and @code{static}, if all calls to the
7200 function are integrated into the caller, and the function's address is
7201 never used, then the function's own assembler code is never referenced.
7202 In this case, GCC does not actually output assembler code for the
7203 function, unless you specify the option @option{-fkeep-inline-functions}.
7204 If there is a nonintegrated call, then the function is compiled to
7205 assembler code as usual. The function must also be compiled as usual if
7206 the program refers to its address, because that can't be inlined.
7209 Note that certain usages in a function definition can make it unsuitable
7210 for inline substitution. Among these usages are: variadic functions,
7211 use of @code{alloca}, use of computed goto (@pxref{Labels as Values}),
7212 use of nonlocal goto, use of nested functions, use of @code{setjmp}, use
7213 of @code{__builtin_longjmp} and use of @code{__builtin_return} or
7214 @code{__builtin_apply_args}. Using @option{-Winline} warns when a
7215 function marked @code{inline} could not be substituted, and gives the
7216 reason for the failure.
7218 @cindex automatic @code{inline} for C++ member fns
7219 @cindex @code{inline} automatic for C++ member fns
7220 @cindex member fns, automatically @code{inline}
7221 @cindex C++ member fns, automatically @code{inline}
7222 @opindex fno-default-inline
7223 As required by ISO C++, GCC considers member functions defined within
7224 the body of a class to be marked inline even if they are
7225 not explicitly declared with the @code{inline} keyword. You can
7226 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
7227 Options,,Options Controlling C++ Dialect}.
7229 GCC does not inline any functions when not optimizing unless you specify
7230 the @samp{always_inline} attribute for the function, like this:
7233 /* @r{Prototype.} */
7234 inline void foo (const char) __attribute__((always_inline));
7237 The remainder of this section is specific to GNU C90 inlining.
7239 @cindex non-static inline function
7240 When an inline function is not @code{static}, then the compiler must assume
7241 that there may be calls from other source files; since a global symbol can
7242 be defined only once in any program, the function must not be defined in
7243 the other source files, so the calls therein cannot be integrated.
7244 Therefore, a non-@code{static} inline function is always compiled on its
7245 own in the usual fashion.
7247 If you specify both @code{inline} and @code{extern} in the function
7248 definition, then the definition is used only for inlining. In no case
7249 is the function compiled on its own, not even if you refer to its
7250 address explicitly. Such an address becomes an external reference, as
7251 if you had only declared the function, and had not defined it.
7253 This combination of @code{inline} and @code{extern} has almost the
7254 effect of a macro. The way to use it is to put a function definition in
7255 a header file with these keywords, and put another copy of the
7256 definition (lacking @code{inline} and @code{extern}) in a library file.
7257 The definition in the header file causes most calls to the function
7258 to be inlined. If any uses of the function remain, they refer to
7259 the single copy in the library.
7262 @section When is a Volatile Object Accessed?
7263 @cindex accessing volatiles
7264 @cindex volatile read
7265 @cindex volatile write
7266 @cindex volatile access
7268 C has the concept of volatile objects. These are normally accessed by
7269 pointers and used for accessing hardware or inter-thread
7270 communication. The standard encourages compilers to refrain from
7271 optimizations concerning accesses to volatile objects, but leaves it
7272 implementation defined as to what constitutes a volatile access. The
7273 minimum requirement is that at a sequence point all previous accesses
7274 to volatile objects have stabilized and no subsequent accesses have
7275 occurred. Thus an implementation is free to reorder and combine
7276 volatile accesses that occur between sequence points, but cannot do
7277 so for accesses across a sequence point. The use of volatile does
7278 not allow you to violate the restriction on updating objects multiple
7279 times between two sequence points.
7281 Accesses to non-volatile objects are not ordered with respect to
7282 volatile accesses. You cannot use a volatile object as a memory
7283 barrier to order a sequence of writes to non-volatile memory. For
7287 int *ptr = @var{something};
7289 *ptr = @var{something};
7294 Unless @var{*ptr} and @var{vobj} can be aliased, it is not guaranteed
7295 that the write to @var{*ptr} occurs by the time the update
7296 of @var{vobj} happens. If you need this guarantee, you must use
7297 a stronger memory barrier such as:
7300 int *ptr = @var{something};
7302 *ptr = @var{something};
7303 asm volatile ("" : : : "memory");
7307 A scalar volatile object is read when it is accessed in a void context:
7310 volatile int *src = @var{somevalue};
7314 Such expressions are rvalues, and GCC implements this as a
7315 read of the volatile object being pointed to.
7317 Assignments are also expressions and have an rvalue. However when
7318 assigning to a scalar volatile, the volatile object is not reread,
7319 regardless of whether the assignment expression's rvalue is used or
7320 not. If the assignment's rvalue is used, the value is that assigned
7321 to the volatile object. For instance, there is no read of @var{vobj}
7322 in all the following cases:
7327 vobj = @var{something};
7328 obj = vobj = @var{something};
7329 obj ? vobj = @var{onething} : vobj = @var{anotherthing};
7330 obj = (@var{something}, vobj = @var{anotherthing});
7333 If you need to read the volatile object after an assignment has
7334 occurred, you must use a separate expression with an intervening
7337 As bit-fields are not individually addressable, volatile bit-fields may
7338 be implicitly read when written to, or when adjacent bit-fields are
7339 accessed. Bit-field operations may be optimized such that adjacent
7340 bit-fields are only partially accessed, if they straddle a storage unit
7341 boundary. For these reasons it is unwise to use volatile bit-fields to
7344 @node Using Assembly Language with C
7345 @section How to Use Inline Assembly Language in C Code
7346 @cindex @code{asm} keyword
7347 @cindex assembly language in C
7348 @cindex inline assembly language
7349 @cindex mixing assembly language and C
7351 The @code{asm} keyword allows you to embed assembler instructions
7352 within C code. GCC provides two forms of inline @code{asm}
7353 statements. A @dfn{basic @code{asm}} statement is one with no
7354 operands (@pxref{Basic Asm}), while an @dfn{extended @code{asm}}
7355 statement (@pxref{Extended Asm}) includes one or more operands.
7356 The extended form is preferred for mixing C and assembly language
7357 within a function, but to include assembly language at
7358 top level you must use basic @code{asm}.
7360 You can also use the @code{asm} keyword to override the assembler name
7361 for a C symbol, or to place a C variable in a specific register.
7364 * Basic Asm:: Inline assembler without operands.
7365 * Extended Asm:: Inline assembler with operands.
7366 * Constraints:: Constraints for @code{asm} operands
7367 * Asm Labels:: Specifying the assembler name to use for a C symbol.
7368 * Explicit Register Variables:: Defining variables residing in specified
7370 * Size of an asm:: How GCC calculates the size of an @code{asm} block.
7374 @subsection Basic Asm --- Assembler Instructions Without Operands
7375 @cindex basic @code{asm}
7376 @cindex assembly language in C, basic
7378 A basic @code{asm} statement has the following syntax:
7381 asm @r{[} volatile @r{]} ( @var{AssemblerInstructions} )
7384 The @code{asm} keyword is a GNU extension.
7385 When writing code that can be compiled with @option{-ansi} and the
7386 various @option{-std} options, use @code{__asm__} instead of
7387 @code{asm} (@pxref{Alternate Keywords}).
7389 @subsubheading Qualifiers
7392 The optional @code{volatile} qualifier has no effect.
7393 All basic @code{asm} blocks are implicitly volatile.
7396 @subsubheading Parameters
7399 @item AssemblerInstructions
7400 This is a literal string that specifies the assembler code. The string can
7401 contain any instructions recognized by the assembler, including directives.
7402 GCC does not parse the assembler instructions themselves and
7403 does not know what they mean or even whether they are valid assembler input.
7405 You may place multiple assembler instructions together in a single @code{asm}
7406 string, separated by the characters normally used in assembly code for the
7407 system. A combination that works in most places is a newline to break the
7408 line, plus a tab character (written as @samp{\n\t}).
7409 Some assemblers allow semicolons as a line separator. However,
7410 note that some assembler dialects use semicolons to start a comment.
7413 @subsubheading Remarks
7414 Using extended @code{asm} typically produces smaller, safer, and more
7415 efficient code, and in most cases it is a better solution than basic
7416 @code{asm}. However, there are two situations where only basic @code{asm}
7421 Extended @code{asm} statements have to be inside a C
7422 function, so to write inline assembly language at file scope (``top-level''),
7423 outside of C functions, you must use basic @code{asm}.
7424 You can use this technique to emit assembler directives,
7425 define assembly language macros that can be invoked elsewhere in the file,
7426 or write entire functions in assembly language.
7430 with the @code{naked} attribute also require basic @code{asm}
7431 (@pxref{Function Attributes}).
7434 Safely accessing C data and calling functions from basic @code{asm} is more
7435 complex than it may appear. To access C data, it is better to use extended
7438 Do not expect a sequence of @code{asm} statements to remain perfectly
7439 consecutive after compilation. If certain instructions need to remain
7440 consecutive in the output, put them in a single multi-instruction @code{asm}
7441 statement. Note that GCC's optimizers can move @code{asm} statements
7442 relative to other code, including across jumps.
7444 @code{asm} statements may not perform jumps into other @code{asm} statements.
7445 GCC does not know about these jumps, and therefore cannot take
7446 account of them when deciding how to optimize. Jumps from @code{asm} to C
7447 labels are only supported in extended @code{asm}.
7449 Under certain circumstances, GCC may duplicate (or remove duplicates of) your
7450 assembly code when optimizing. This can lead to unexpected duplicate
7451 symbol errors during compilation if your assembly code defines symbols or
7454 Since GCC does not parse the @var{AssemblerInstructions}, it has no
7455 visibility of any symbols it references. This may result in GCC discarding
7456 those symbols as unreferenced.
7458 The compiler copies the assembler instructions in a basic @code{asm}
7459 verbatim to the assembly language output file, without
7460 processing dialects or any of the @samp{%} operators that are available with
7461 extended @code{asm}. This results in minor differences between basic
7462 @code{asm} strings and extended @code{asm} templates. For example, to refer to
7463 registers you might use @samp{%eax} in basic @code{asm} and
7464 @samp{%%eax} in extended @code{asm}.
7466 On targets such as x86 that support multiple assembler dialects,
7467 all basic @code{asm} blocks use the assembler dialect specified by the
7468 @option{-masm} command-line option (@pxref{x86 Options}).
7469 Basic @code{asm} provides no
7470 mechanism to provide different assembler strings for different dialects.
7472 Here is an example of basic @code{asm} for i386:
7475 /* Note that this code will not compile with -masm=intel */
7476 #define DebugBreak() asm("int $3")
7480 @subsection Extended Asm - Assembler Instructions with C Expression Operands
7481 @cindex extended @code{asm}
7482 @cindex assembly language in C, extended
7484 With extended @code{asm} you can read and write C variables from
7485 assembler and perform jumps from assembler code to C labels.
7486 Extended @code{asm} syntax uses colons (@samp{:}) to delimit
7487 the operand parameters after the assembler template:
7490 asm @r{[}volatile@r{]} ( @var{AssemblerTemplate}
7491 : @var{OutputOperands}
7492 @r{[} : @var{InputOperands}
7493 @r{[} : @var{Clobbers} @r{]} @r{]})
7495 asm @r{[}volatile@r{]} goto ( @var{AssemblerTemplate}
7497 : @var{InputOperands}
7502 The @code{asm} keyword is a GNU extension.
7503 When writing code that can be compiled with @option{-ansi} and the
7504 various @option{-std} options, use @code{__asm__} instead of
7505 @code{asm} (@pxref{Alternate Keywords}).
7507 @subsubheading Qualifiers
7511 The typical use of extended @code{asm} statements is to manipulate input
7512 values to produce output values. However, your @code{asm} statements may
7513 also produce side effects. If so, you may need to use the @code{volatile}
7514 qualifier to disable certain optimizations. @xref{Volatile}.
7517 This qualifier informs the compiler that the @code{asm} statement may
7518 perform a jump to one of the labels listed in the @var{GotoLabels}.
7522 @subsubheading Parameters
7524 @item AssemblerTemplate
7525 This is a literal string that is the template for the assembler code. It is a
7526 combination of fixed text and tokens that refer to the input, output,
7527 and goto parameters. @xref{AssemblerTemplate}.
7529 @item OutputOperands
7530 A comma-separated list of the C variables modified by the instructions in the
7531 @var{AssemblerTemplate}. An empty list is permitted. @xref{OutputOperands}.
7534 A comma-separated list of C expressions read by the instructions in the
7535 @var{AssemblerTemplate}. An empty list is permitted. @xref{InputOperands}.
7538 A comma-separated list of registers or other values changed by the
7539 @var{AssemblerTemplate}, beyond those listed as outputs.
7540 An empty list is permitted. @xref{Clobbers}.
7543 When you are using the @code{goto} form of @code{asm}, this section contains
7544 the list of all C labels to which the code in the
7545 @var{AssemblerTemplate} may jump.
7548 @code{asm} statements may not perform jumps into other @code{asm} statements,
7549 only to the listed @var{GotoLabels}.
7550 GCC's optimizers do not know about other jumps; therefore they cannot take
7551 account of them when deciding how to optimize.
7554 The total number of input + output + goto operands is limited to 30.
7556 @subsubheading Remarks
7557 The @code{asm} statement allows you to include assembly instructions directly
7558 within C code. This may help you to maximize performance in time-sensitive
7559 code or to access assembly instructions that are not readily available to C
7562 Note that extended @code{asm} statements must be inside a function. Only
7563 basic @code{asm} may be outside functions (@pxref{Basic Asm}).
7564 Functions declared with the @code{naked} attribute also require basic
7565 @code{asm} (@pxref{Function Attributes}).
7567 While the uses of @code{asm} are many and varied, it may help to think of an
7568 @code{asm} statement as a series of low-level instructions that convert input
7569 parameters to output parameters. So a simple (if not particularly useful)
7570 example for i386 using @code{asm} might look like this:
7576 asm ("mov %1, %0\n\t"
7581 printf("%d\n", dst);
7584 This code copies @code{src} to @code{dst} and add 1 to @code{dst}.
7587 @subsubsection Volatile
7588 @cindex volatile @code{asm}
7589 @cindex @code{asm} volatile
7591 GCC's optimizers sometimes discard @code{asm} statements if they determine
7592 there is no need for the output variables. Also, the optimizers may move
7593 code out of loops if they believe that the code will always return the same
7594 result (i.e. none of its input values change between calls). Using the
7595 @code{volatile} qualifier disables these optimizations. @code{asm} statements
7596 that have no output operands, including @code{asm goto} statements,
7597 are implicitly volatile.
7599 This i386 code demonstrates a case that does not use (or require) the
7600 @code{volatile} qualifier. If it is performing assertion checking, this code
7601 uses @code{asm} to perform the validation. Otherwise, @code{dwRes} is
7602 unreferenced by any code. As a result, the optimizers can discard the
7603 @code{asm} statement, which in turn removes the need for the entire
7604 @code{DoCheck} routine. By omitting the @code{volatile} qualifier when it
7605 isn't needed you allow the optimizers to produce the most efficient code
7609 void DoCheck(uint32_t dwSomeValue)
7613 // Assumes dwSomeValue is not zero.
7623 The next example shows a case where the optimizers can recognize that the input
7624 (@code{dwSomeValue}) never changes during the execution of the function and can
7625 therefore move the @code{asm} outside the loop to produce more efficient code.
7626 Again, using @code{volatile} disables this type of optimization.
7629 void do_print(uint32_t dwSomeValue)
7633 for (uint32_t x=0; x < 5; x++)
7635 // Assumes dwSomeValue is not zero.
7641 printf("%u: %u %u\n", x, dwSomeValue, dwRes);
7646 The following example demonstrates a case where you need to use the
7647 @code{volatile} qualifier.
7648 It uses the x86 @code{rdtsc} instruction, which reads
7649 the computer's time-stamp counter. Without the @code{volatile} qualifier,
7650 the optimizers might assume that the @code{asm} block will always return the
7651 same value and therefore optimize away the second call.
7656 asm volatile ( "rdtsc\n\t" // Returns the time in EDX:EAX.
7657 "shl $32, %%rdx\n\t" // Shift the upper bits left.
7658 "or %%rdx, %0" // 'Or' in the lower bits.
7663 printf("msr: %llx\n", msr);
7667 // Reprint the timestamp
7668 asm volatile ( "rdtsc\n\t" // Returns the time in EDX:EAX.
7669 "shl $32, %%rdx\n\t" // Shift the upper bits left.
7670 "or %%rdx, %0" // 'Or' in the lower bits.
7675 printf("msr: %llx\n", msr);
7678 GCC's optimizers do not treat this code like the non-volatile code in the
7679 earlier examples. They do not move it out of loops or omit it on the
7680 assumption that the result from a previous call is still valid.
7682 Note that the compiler can move even volatile @code{asm} instructions relative
7683 to other code, including across jump instructions. For example, on many
7684 targets there is a system register that controls the rounding mode of
7685 floating-point operations. Setting it with a volatile @code{asm}, as in the
7686 following PowerPC example, does not work reliably.
7689 asm volatile("mtfsf 255, %0" : : "f" (fpenv));
7693 The compiler may move the addition back before the volatile @code{asm}. To
7694 make it work as expected, add an artificial dependency to the @code{asm} by
7695 referencing a variable in the subsequent code, for example:
7698 asm volatile ("mtfsf 255,%1" : "=X" (sum) : "f" (fpenv));
7702 Under certain circumstances, GCC may duplicate (or remove duplicates of) your
7703 assembly code when optimizing. This can lead to unexpected duplicate symbol
7704 errors during compilation if your asm code defines symbols or labels.
7706 (@pxref{AssemblerTemplate}) may help resolve this problem.
7708 @anchor{AssemblerTemplate}
7709 @subsubsection Assembler Template
7710 @cindex @code{asm} assembler template
7712 An assembler template is a literal string containing assembler instructions.
7713 The compiler replaces tokens in the template that refer
7714 to inputs, outputs, and goto labels,
7715 and then outputs the resulting string to the assembler. The
7716 string can contain any instructions recognized by the assembler, including
7717 directives. GCC does not parse the assembler instructions
7718 themselves and does not know what they mean or even whether they are valid
7719 assembler input. However, it does count the statements
7720 (@pxref{Size of an asm}).
7722 You may place multiple assembler instructions together in a single @code{asm}
7723 string, separated by the characters normally used in assembly code for the
7724 system. A combination that works in most places is a newline to break the
7725 line, plus a tab character to move to the instruction field (written as
7727 Some assemblers allow semicolons as a line separator. However, note
7728 that some assembler dialects use semicolons to start a comment.
7730 Do not expect a sequence of @code{asm} statements to remain perfectly
7731 consecutive after compilation, even when you are using the @code{volatile}
7732 qualifier. If certain instructions need to remain consecutive in the output,
7733 put them in a single multi-instruction asm statement.
7735 Accessing data from C programs without using input/output operands (such as
7736 by using global symbols directly from the assembler template) may not work as
7737 expected. Similarly, calling functions directly from an assembler template
7738 requires a detailed understanding of the target assembler and ABI.
7740 Since GCC does not parse the assembler template,
7741 it has no visibility of any
7742 symbols it references. This may result in GCC discarding those symbols as
7743 unreferenced unless they are also listed as input, output, or goto operands.
7745 @subsubheading Special format strings
7747 In addition to the tokens described by the input, output, and goto operands,
7748 these tokens have special meanings in the assembler template:
7752 Outputs a single @samp{%} into the assembler code.
7755 Outputs a number that is unique to each instance of the @code{asm}
7756 statement in the entire compilation. This option is useful when creating local
7757 labels and referring to them multiple times in a single template that
7758 generates multiple assembler instructions.
7763 Outputs @samp{@{}, @samp{|}, and @samp{@}} characters (respectively)
7764 into the assembler code. When unescaped, these characters have special
7765 meaning to indicate multiple assembler dialects, as described below.
7768 @subsubheading Multiple assembler dialects in @code{asm} templates
7770 On targets such as x86, GCC supports multiple assembler dialects.
7771 The @option{-masm} option controls which dialect GCC uses as its
7772 default for inline assembler. The target-specific documentation for the
7773 @option{-masm} option contains the list of supported dialects, as well as the
7774 default dialect if the option is not specified. This information may be
7775 important to understand, since assembler code that works correctly when
7776 compiled using one dialect will likely fail if compiled using another.
7779 If your code needs to support multiple assembler dialects (for example, if
7780 you are writing public headers that need to support a variety of compilation
7781 options), use constructs of this form:
7784 @{ dialect0 | dialect1 | dialect2... @}
7787 This construct outputs @code{dialect0}
7788 when using dialect #0 to compile the code,
7789 @code{dialect1} for dialect #1, etc. If there are fewer alternatives within the
7790 braces than the number of dialects the compiler supports, the construct
7793 For example, if an x86 compiler supports two dialects
7794 (@samp{att}, @samp{intel}), an
7795 assembler template such as this:
7798 "bt@{l %[Offset],%[Base] | %[Base],%[Offset]@}; jc %l2"
7802 is equivalent to one of
7805 "btl %[Offset],%[Base] ; jc %l2" @r{/* att dialect */}
7806 "bt %[Base],%[Offset]; jc %l2" @r{/* intel dialect */}
7809 Using that same compiler, this code:
7812 "xchg@{l@}\t@{%%@}ebx, %1"
7816 corresponds to either
7819 "xchgl\t%%ebx, %1" @r{/* att dialect */}
7820 "xchg\tebx, %1" @r{/* intel dialect */}
7823 There is no support for nesting dialect alternatives.
7825 @anchor{OutputOperands}
7826 @subsubsection Output Operands
7827 @cindex @code{asm} output operands
7829 An @code{asm} statement has zero or more output operands indicating the names
7830 of C variables modified by the assembler code.
7832 In this i386 example, @code{old} (referred to in the template string as
7833 @code{%0}) and @code{*Base} (as @code{%1}) are outputs and @code{Offset}
7834 (@code{%2}) is an input:
7839 __asm__ ("btsl %2,%1\n\t" // Turn on zero-based bit #Offset in Base.
7840 "sbb %0,%0" // Use the CF to calculate old.
7841 : "=r" (old), "+rm" (*Base)
7848 Operands are separated by commas. Each operand has this format:
7851 @r{[} [@var{asmSymbolicName}] @r{]} @var{constraint} (@var{cvariablename})
7855 @item asmSymbolicName
7856 Specifies a symbolic name for the operand.
7857 Reference the name in the assembler template
7858 by enclosing it in square brackets
7859 (i.e. @samp{%[Value]}). The scope of the name is the @code{asm} statement
7860 that contains the definition. Any valid C variable name is acceptable,
7861 including names already defined in the surrounding code. No two operands
7862 within the same @code{asm} statement can use the same symbolic name.
7864 When not using an @var{asmSymbolicName}, use the (zero-based) position
7866 in the list of operands in the assembler template. For example if there are
7867 three output operands, use @samp{%0} in the template to refer to the first,
7868 @samp{%1} for the second, and @samp{%2} for the third.
7871 A string constant specifying constraints on the placement of the operand;
7872 @xref{Constraints}, for details.
7874 Output constraints must begin with either @samp{=} (a variable overwriting an
7875 existing value) or @samp{+} (when reading and writing). When using
7876 @samp{=}, do not assume the location contains the existing value
7877 on entry to the @code{asm}, except
7878 when the operand is tied to an input; @pxref{InputOperands,,Input Operands}.
7880 After the prefix, there must be one or more additional constraints
7881 (@pxref{Constraints}) that describe where the value resides. Common
7882 constraints include @samp{r} for register and @samp{m} for memory.
7883 When you list more than one possible location (for example, @code{"=rm"}),
7884 the compiler chooses the most efficient one based on the current context.
7885 If you list as many alternates as the @code{asm} statement allows, you permit
7886 the optimizers to produce the best possible code.
7887 If you must use a specific register, but your Machine Constraints do not
7888 provide sufficient control to select the specific register you want,
7889 local register variables may provide a solution (@pxref{Local Register
7893 Specifies a C lvalue expression to hold the output, typically a variable name.
7894 The enclosing parentheses are a required part of the syntax.
7898 When the compiler selects the registers to use to
7899 represent the output operands, it does not use any of the clobbered registers
7902 Output operand expressions must be lvalues. The compiler cannot check whether
7903 the operands have data types that are reasonable for the instruction being
7904 executed. For output expressions that are not directly addressable (for
7905 example a bit-field), the constraint must allow a register. In that case, GCC
7906 uses the register as the output of the @code{asm}, and then stores that
7907 register into the output.
7909 Operands using the @samp{+} constraint modifier count as two operands
7910 (that is, both as input and output) towards the total maximum of 30 operands
7911 per @code{asm} statement.
7913 Use the @samp{&} constraint modifier (@pxref{Modifiers}) on all output
7914 operands that must not overlap an input. Otherwise,
7915 GCC may allocate the output operand in the same register as an unrelated
7916 input operand, on the assumption that the assembler code consumes its
7917 inputs before producing outputs. This assumption may be false if the assembler
7918 code actually consists of more than one instruction.
7920 The same problem can occur if one output parameter (@var{a}) allows a register
7921 constraint and another output parameter (@var{b}) allows a memory constraint.
7922 The code generated by GCC to access the memory address in @var{b} can contain
7923 registers which @emph{might} be shared by @var{a}, and GCC considers those
7924 registers to be inputs to the asm. As above, GCC assumes that such input
7925 registers are consumed before any outputs are written. This assumption may
7926 result in incorrect behavior if the asm writes to @var{a} before using
7927 @var{b}. Combining the @samp{&} modifier with the register constraint on @var{a}
7928 ensures that modifying @var{a} does not affect the address referenced by
7929 @var{b}. Otherwise, the location of @var{b}
7930 is undefined if @var{a} is modified before using @var{b}.
7932 @code{asm} supports operand modifiers on operands (for example @samp{%k2}
7933 instead of simply @samp{%2}). Typically these qualifiers are hardware
7934 dependent. The list of supported modifiers for x86 is found at
7935 @ref{x86Operandmodifiers,x86 Operand modifiers}.
7937 If the C code that follows the @code{asm} makes no use of any of the output
7938 operands, use @code{volatile} for the @code{asm} statement to prevent the
7939 optimizers from discarding the @code{asm} statement as unneeded
7940 (see @ref{Volatile}).
7942 This code makes no use of the optional @var{asmSymbolicName}. Therefore it
7943 references the first output operand as @code{%0} (were there a second, it
7944 would be @code{%1}, etc). The number of the first input operand is one greater
7945 than that of the last output operand. In this i386 example, that makes
7946 @code{Mask} referenced as @code{%1}:
7949 uint32_t Mask = 1234;
7958 That code overwrites the variable @code{Index} (@samp{=}),
7959 placing the value in a register (@samp{r}).
7960 Using the generic @samp{r} constraint instead of a constraint for a specific
7961 register allows the compiler to pick the register to use, which can result
7962 in more efficient code. This may not be possible if an assembler instruction
7963 requires a specific register.
7965 The following i386 example uses the @var{asmSymbolicName} syntax.
7967 same result as the code above, but some may consider it more readable or more
7968 maintainable since reordering index numbers is not necessary when adding or
7969 removing operands. The names @code{aIndex} and @code{aMask}
7970 are only used in this example to emphasize which
7971 names get used where.
7972 It is acceptable to reuse the names @code{Index} and @code{Mask}.
7975 uint32_t Mask = 1234;
7978 asm ("bsfl %[aMask], %[aIndex]"
7979 : [aIndex] "=r" (Index)
7980 : [aMask] "r" (Mask)
7984 Here are some more examples of output operands.
7991 asm ("mov %[e], %[d]"
7996 Here, @code{d} may either be in a register or in memory. Since the compiler
7997 might already have the current value of the @code{uint32_t} location
7998 pointed to by @code{e}
7999 in a register, you can enable it to choose the best location
8000 for @code{d} by specifying both constraints.
8002 @anchor{FlagOutputOperands}
8003 @subsection Flag Output Operands
8004 @cindex @code{asm} flag output operands
8006 Some targets have a special register that holds the ``flags'' for the
8007 result of an operation or comparison. Normally, the contents of that
8008 register are either unmodifed by the asm, or the asm is considered to
8009 clobber the contents.
8011 On some targets, a special form of output operand exists by which
8012 conditions in the flags register may be outputs of the asm. The set of
8013 conditions supported are target specific, but the general rule is that
8014 the output variable must be a scalar integer, and the value will be boolean.
8015 When supported, the target will define the preprocessor symbol
8016 @code{__GCC_ASM_FLAG_OUTPUTS__}.
8018 Because of the special nature of the flag output operands, the constraint
8019 may not include alternatives.
8021 Most often, the target has only one flags register, and thus is an implied
8022 operand of many instructions. In this case, the operand should not be
8023 referenced within the assembler template via @code{%0} etc, as there's
8024 no corresponding text in the assembly language.
8028 The flag output constraints for the x86 family are of the form
8029 @samp{=@@cc@var{cond}} where @var{cond} is one of the standard
8030 conditions defined in the ISA manual for @code{j@var{cc}} or
8035 ``above'' or unsigned greater than
8037 ``above or equal'' or unsigned greater than or equal
8039 ``below'' or unsigned less than
8041 ``below or equal'' or unsigned less than or equal
8046 ``equal'' or zero flag set
8050 signed greater than or equal
8054 signed less than or equal
8075 ``not'' @var{flag}, or inverted versions of those above
8080 @anchor{InputOperands}
8081 @subsubsection Input Operands
8082 @cindex @code{asm} input operands
8083 @cindex @code{asm} expressions
8085 Input operands make values from C variables and expressions available to the
8088 Operands are separated by commas. Each operand has this format:
8091 @r{[} [@var{asmSymbolicName}] @r{]} @var{constraint} (@var{cexpression})
8095 @item asmSymbolicName
8096 Specifies a symbolic name for the operand.
8097 Reference the name in the assembler template
8098 by enclosing it in square brackets
8099 (i.e. @samp{%[Value]}). The scope of the name is the @code{asm} statement
8100 that contains the definition. Any valid C variable name is acceptable,
8101 including names already defined in the surrounding code. No two operands
8102 within the same @code{asm} statement can use the same symbolic name.
8104 When not using an @var{asmSymbolicName}, use the (zero-based) position
8106 in the list of operands in the assembler template. For example if there are
8107 two output operands and three inputs,
8108 use @samp{%2} in the template to refer to the first input operand,
8109 @samp{%3} for the second, and @samp{%4} for the third.
8112 A string constant specifying constraints on the placement of the operand;
8113 @xref{Constraints}, for details.
8115 Input constraint strings may not begin with either @samp{=} or @samp{+}.
8116 When you list more than one possible location (for example, @samp{"irm"}),
8117 the compiler chooses the most efficient one based on the current context.
8118 If you must use a specific register, but your Machine Constraints do not
8119 provide sufficient control to select the specific register you want,
8120 local register variables may provide a solution (@pxref{Local Register
8123 Input constraints can also be digits (for example, @code{"0"}). This indicates
8124 that the specified input must be in the same place as the output constraint
8125 at the (zero-based) index in the output constraint list.
8126 When using @var{asmSymbolicName} syntax for the output operands,
8127 you may use these names (enclosed in brackets @samp{[]}) instead of digits.
8130 This is the C variable or expression being passed to the @code{asm} statement
8131 as input. The enclosing parentheses are a required part of the syntax.
8135 When the compiler selects the registers to use to represent the input
8136 operands, it does not use any of the clobbered registers (@pxref{Clobbers}).
8138 If there are no output operands but there are input operands, place two
8139 consecutive colons where the output operands would go:
8142 __asm__ ("some instructions"
8144 : "r" (Offset / 8));
8147 @strong{Warning:} Do @emph{not} modify the contents of input-only operands
8148 (except for inputs tied to outputs). The compiler assumes that on exit from
8149 the @code{asm} statement these operands contain the same values as they
8150 had before executing the statement.
8151 It is @emph{not} possible to use clobbers
8152 to inform the compiler that the values in these inputs are changing. One
8153 common work-around is to tie the changing input variable to an output variable
8154 that never gets used. Note, however, that if the code that follows the
8155 @code{asm} statement makes no use of any of the output operands, the GCC
8156 optimizers may discard the @code{asm} statement as unneeded
8157 (see @ref{Volatile}).
8159 @code{asm} supports operand modifiers on operands (for example @samp{%k2}
8160 instead of simply @samp{%2}). Typically these qualifiers are hardware
8161 dependent. The list of supported modifiers for x86 is found at
8162 @ref{x86Operandmodifiers,x86 Operand modifiers}.
8164 In this example using the fictitious @code{combine} instruction, the
8165 constraint @code{"0"} for input operand 1 says that it must occupy the same
8166 location as output operand 0. Only input operands may use numbers in
8167 constraints, and they must each refer to an output operand. Only a number (or
8168 the symbolic assembler name) in the constraint can guarantee that one operand
8169 is in the same place as another. The mere fact that @code{foo} is the value of
8170 both operands is not enough to guarantee that they are in the same place in
8171 the generated assembler code.
8174 asm ("combine %2, %0"
8176 : "0" (foo), "g" (bar));
8179 Here is an example using symbolic names.
8182 asm ("cmoveq %1, %2, %[result]"
8183 : [result] "=r"(result)
8184 : "r" (test), "r" (new), "[result]" (old));
8188 @subsubsection Clobbers
8189 @cindex @code{asm} clobbers
8191 While the compiler is aware of changes to entries listed in the output
8192 operands, the inline @code{asm} code may modify more than just the outputs. For
8193 example, calculations may require additional registers, or the processor may
8194 overwrite a register as a side effect of a particular assembler instruction.
8195 In order to inform the compiler of these changes, list them in the clobber
8196 list. Clobber list items are either register names or the special clobbers
8197 (listed below). Each clobber list item is a string constant
8198 enclosed in double quotes and separated by commas.
8200 Clobber descriptions may not in any way overlap with an input or output
8201 operand. For example, you may not have an operand describing a register class
8202 with one member when listing that register in the clobber list. Variables
8203 declared to live in specific registers (@pxref{Explicit Register
8204 Variables}) and used
8205 as @code{asm} input or output operands must have no part mentioned in the
8206 clobber description. In particular, there is no way to specify that input
8207 operands get modified without also specifying them as output operands.
8209 When the compiler selects which registers to use to represent input and output
8210 operands, it does not use any of the clobbered registers. As a result,
8211 clobbered registers are available for any use in the assembler code.
8213 Here is a realistic example for the VAX showing the use of clobbered
8217 asm volatile ("movc3 %0, %1, %2"
8219 : "g" (from), "g" (to), "g" (count)
8220 : "r0", "r1", "r2", "r3", "r4", "r5");
8223 Also, there are two special clobber arguments:
8227 The @code{"cc"} clobber indicates that the assembler code modifies the flags
8228 register. On some machines, GCC represents the condition codes as a specific
8229 hardware register; @code{"cc"} serves to name this register.
8230 On other machines, condition code handling is different,
8231 and specifying @code{"cc"} has no effect. But
8232 it is valid no matter what the target.
8235 The @code{"memory"} clobber tells the compiler that the assembly code
8237 reads or writes to items other than those listed in the input and output
8238 operands (for example, accessing the memory pointed to by one of the input
8239 parameters). To ensure memory contains correct values, GCC may need to flush
8240 specific register values to memory before executing the @code{asm}. Further,
8241 the compiler does not assume that any values read from memory before an
8242 @code{asm} remain unchanged after that @code{asm}; it reloads them as
8244 Using the @code{"memory"} clobber effectively forms a read/write
8245 memory barrier for the compiler.
8247 Note that this clobber does not prevent the @emph{processor} from doing
8248 speculative reads past the @code{asm} statement. To prevent that, you need
8249 processor-specific fence instructions.
8251 Flushing registers to memory has performance implications and may be an issue
8252 for time-sensitive code. You can use a trick to avoid this if the size of
8253 the memory being accessed is known at compile time. For example, if accessing
8254 ten bytes of a string, use a memory input like:
8256 @code{@{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}}.
8261 @subsubsection Goto Labels
8262 @cindex @code{asm} goto labels
8264 @code{asm goto} allows assembly code to jump to one or more C labels. The
8265 @var{GotoLabels} section in an @code{asm goto} statement contains
8267 list of all C labels to which the assembler code may jump. GCC assumes that
8268 @code{asm} execution falls through to the next statement (if this is not the
8269 case, consider using the @code{__builtin_unreachable} intrinsic after the
8270 @code{asm} statement). Optimization of @code{asm goto} may be improved by
8271 using the @code{hot} and @code{cold} label attributes (@pxref{Label
8274 An @code{asm goto} statement cannot have outputs.
8275 This is due to an internal restriction of
8276 the compiler: control transfer instructions cannot have outputs.
8277 If the assembler code does modify anything, use the @code{"memory"} clobber
8279 optimizers to flush all register values to memory and reload them if
8280 necessary after the @code{asm} statement.
8282 Also note that an @code{asm goto} statement is always implicitly
8283 considered volatile.
8285 To reference a label in the assembler template,
8286 prefix it with @samp{%l} (lowercase @samp{L}) followed
8287 by its (zero-based) position in @var{GotoLabels} plus the number of input
8288 operands. For example, if the @code{asm} has three inputs and references two
8289 labels, refer to the first label as @samp{%l3} and the second as @samp{%l4}).
8291 Alternately, you can reference labels using the actual C label name enclosed
8292 in brackets. For example, to reference a label named @code{carry}, you can
8293 use @samp{%l[carry]}. The label must still be listed in the @var{GotoLabels}
8294 section when using this approach.
8296 Here is an example of @code{asm goto} for i386:
8303 : "r" (p1), "r" (p2)
8313 The following example shows an @code{asm goto} that uses a memory clobber.
8319 asm goto ("frob %%r5, %1; jc %l[error]; mov (%2), %%r5"
8330 @anchor{x86Operandmodifiers}
8331 @subsubsection x86 Operand Modifiers
8333 References to input, output, and goto operands in the assembler template
8334 of extended @code{asm} statements can use
8335 modifiers to affect the way the operands are formatted in
8336 the code output to the assembler. For example, the
8337 following code uses the @samp{h} and @samp{b} modifiers for x86:
8341 asm volatile ("xchg %h0, %b0" : "+a" (num) );
8345 These modifiers generate this assembler code:
8351 The rest of this discussion uses the following code for illustrative purposes.
8360 asm volatile goto ("some assembler instructions here"
8362 : "q" (iInt), "X" (sizeof(unsigned char) + 1)
8363 : /* No clobbers. */
8368 With no modifiers, this is what the output from the operands would be for the
8369 @samp{att} and @samp{intel} dialects of assembler:
8371 @multitable {Operand} {masm=att} {OFFSET FLAT:.L2}
8372 @headitem Operand @tab masm=att @tab masm=intel
8381 @tab @code{OFFSET FLAT:.L2}
8384 The table below shows the list of supported modifiers and their effects.
8386 @multitable {Modifier} {Print the opcode suffix for the size of th} {Operand} {masm=att} {masm=intel}
8387 @headitem Modifier @tab Description @tab Operand @tab @option{masm=att} @tab @option{masm=intel}
8389 @tab Print the opcode suffix for the size of the current integer operand (one of @code{b}/@code{w}/@code{l}/@code{q}).
8394 @tab Print the QImode name of the register.
8399 @tab Print the QImode name for a ``high'' register.
8404 @tab Print the HImode name of the register.
8409 @tab Print the SImode name of the register.
8414 @tab Print the DImode name of the register.
8419 @tab Print the label name with no punctuation.
8424 @tab Require a constant operand and print the constant expression with no punctuation.
8430 @anchor{x86floatingpointasmoperands}
8431 @subsubsection x86 Floating-Point @code{asm} Operands
8433 On x86 targets, there are several rules on the usage of stack-like registers
8434 in the operands of an @code{asm}. These rules apply only to the operands
8435 that are stack-like registers:
8439 Given a set of input registers that die in an @code{asm}, it is
8440 necessary to know which are implicitly popped by the @code{asm}, and
8441 which must be explicitly popped by GCC@.
8443 An input register that is implicitly popped by the @code{asm} must be
8444 explicitly clobbered, unless it is constrained to match an
8448 For any input register that is implicitly popped by an @code{asm}, it is
8449 necessary to know how to adjust the stack to compensate for the pop.
8450 If any non-popped input is closer to the top of the reg-stack than
8451 the implicitly popped register, it would not be possible to know what the
8452 stack looked like---it's not clear how the rest of the stack ``slides
8455 All implicitly popped input registers must be closer to the top of
8456 the reg-stack than any input that is not implicitly popped.
8458 It is possible that if an input dies in an @code{asm}, the compiler might
8459 use the input register for an output reload. Consider this example:
8462 asm ("foo" : "=t" (a) : "f" (b));
8466 This code says that input @code{b} is not popped by the @code{asm}, and that
8467 the @code{asm} pushes a result onto the reg-stack, i.e., the stack is one
8468 deeper after the @code{asm} than it was before. But, it is possible that
8469 reload may think that it can use the same register for both the input and
8472 To prevent this from happening,
8473 if any input operand uses the @samp{f} constraint, all output register
8474 constraints must use the @samp{&} early-clobber modifier.
8476 The example above is correctly written as:
8479 asm ("foo" : "=&t" (a) : "f" (b));
8483 Some operands need to be in particular places on the stack. All
8484 output operands fall in this category---GCC has no other way to
8485 know which registers the outputs appear in unless you indicate
8486 this in the constraints.
8488 Output operands must specifically indicate which register an output
8489 appears in after an @code{asm}. @samp{=f} is not allowed: the operand
8490 constraints must select a class with a single register.
8493 Output operands may not be ``inserted'' between existing stack registers.
8494 Since no 387 opcode uses a read/write operand, all output operands
8495 are dead before the @code{asm}, and are pushed by the @code{asm}.
8496 It makes no sense to push anywhere but the top of the reg-stack.
8498 Output operands must start at the top of the reg-stack: output
8499 operands may not ``skip'' a register.
8502 Some @code{asm} statements may need extra stack space for internal
8503 calculations. This can be guaranteed by clobbering stack registers
8504 unrelated to the inputs and outputs.
8509 takes one input, which is internally popped, and produces two outputs.
8512 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
8516 This @code{asm} takes two inputs, which are popped by the @code{fyl2xp1} opcode,
8517 and replaces them with one output. The @code{st(1)} clobber is necessary
8518 for the compiler to know that @code{fyl2xp1} pops both inputs.
8521 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
8529 @subsection Controlling Names Used in Assembler Code
8530 @cindex assembler names for identifiers
8531 @cindex names used in assembler code
8532 @cindex identifiers, names in assembler code
8534 You can specify the name to be used in the assembler code for a C
8535 function or variable by writing the @code{asm} (or @code{__asm__})
8536 keyword after the declarator.
8537 It is up to you to make sure that the assembler names you choose do not
8538 conflict with any other assembler symbols, or reference registers.
8540 @subsubheading Assembler names for data:
8542 This sample shows how to specify the assembler name for data:
8545 int foo asm ("myfoo") = 2;
8549 This specifies that the name to be used for the variable @code{foo} in
8550 the assembler code should be @samp{myfoo} rather than the usual
8553 On systems where an underscore is normally prepended to the name of a C
8554 variable, this feature allows you to define names for the
8555 linker that do not start with an underscore.
8557 GCC does not support using this feature with a non-static local variable
8558 since such variables do not have assembler names. If you are
8559 trying to put the variable in a particular register, see
8560 @ref{Explicit Register Variables}.
8562 @subsubheading Assembler names for functions:
8564 To specify the assembler name for functions, write a declaration for the
8565 function before its definition and put @code{asm} there, like this:
8568 int func (int x, int y) asm ("MYFUNC");
8570 int func (int x, int y)
8576 This specifies that the name to be used for the function @code{func} in
8577 the assembler code should be @code{MYFUNC}.
8579 @node Explicit Register Variables
8580 @subsection Variables in Specified Registers
8581 @anchor{Explicit Reg Vars}
8582 @cindex explicit register variables
8583 @cindex variables in specified registers
8584 @cindex specified registers
8586 GNU C allows you to associate specific hardware registers with C
8587 variables. In almost all cases, allowing the compiler to assign
8588 registers produces the best code. However under certain unusual
8589 circumstances, more precise control over the variable storage is
8592 Both global and local variables can be associated with a register. The
8593 consequences of performing this association are very different between
8594 the two, as explained in the sections below.
8597 * Global Register Variables:: Variables declared at global scope.
8598 * Local Register Variables:: Variables declared within a function.
8601 @node Global Register Variables
8602 @subsubsection Defining Global Register Variables
8603 @anchor{Global Reg Vars}
8604 @cindex global register variables
8605 @cindex registers, global variables in
8606 @cindex registers, global allocation
8608 You can define a global register variable and associate it with a specified
8612 register int *foo asm ("r12");
8616 Here @code{r12} is the name of the register that should be used. Note that
8617 this is the same syntax used for defining local register variables, but for
8618 a global variable the declaration appears outside a function. The
8619 @code{register} keyword is required, and cannot be combined with
8620 @code{static}. The register name must be a valid register name for the
8623 Registers are a scarce resource on most systems and allowing the
8624 compiler to manage their usage usually results in the best code. However,
8625 under special circumstances it can make sense to reserve some globally.
8626 For example this may be useful in programs such as programming language
8627 interpreters that have a couple of global variables that are accessed
8630 After defining a global register variable, for the current compilation
8634 @item The register is reserved entirely for this use, and will not be
8635 allocated for any other purpose.
8636 @item The register is not saved and restored by any functions.
8637 @item Stores into this register are never deleted even if they appear to be
8638 dead, but references may be deleted, moved or simplified.
8641 Note that these points @emph{only} apply to code that is compiled with the
8642 definition. The behavior of code that is merely linked in (for example
8643 code from libraries) is not affected.
8645 If you want to recompile source files that do not actually use your global
8646 register variable so they do not use the specified register for any other
8647 purpose, you need not actually add the global register declaration to
8648 their source code. It suffices to specify the compiler option
8649 @option{-ffixed-@var{reg}} (@pxref{Code Gen Options}) to reserve the
8652 @subsubheading Declaring the variable
8654 Global register variables can not have initial values, because an
8655 executable file has no means to supply initial contents for a register.
8657 When selecting a register, choose one that is normally saved and
8658 restored by function calls on your machine. This ensures that code
8659 which is unaware of this reservation (such as library routines) will
8660 restore it before returning.
8662 On machines with register windows, be sure to choose a global
8663 register that is not affected magically by the function call mechanism.
8665 @subsubheading Using the variable
8667 @cindex @code{qsort}, and global register variables
8668 When calling routines that are not aware of the reservation, be
8669 cautious if those routines call back into code which uses them. As an
8670 example, if you call the system library version of @code{qsort}, it may
8671 clobber your registers during execution, but (if you have selected
8672 appropriate registers) it will restore them before returning. However
8673 it will @emph{not} restore them before calling @code{qsort}'s comparison
8674 function. As a result, global values will not reliably be available to
8675 the comparison function unless the @code{qsort} function itself is rebuilt.
8677 Similarly, it is not safe to access the global register variables from signal
8678 handlers or from more than one thread of control. Unless you recompile
8679 them specially for the task at hand, the system library routines may
8680 temporarily use the register for other things.
8682 @cindex register variable after @code{longjmp}
8683 @cindex global register after @code{longjmp}
8684 @cindex value after @code{longjmp}
8687 On most machines, @code{longjmp} restores to each global register
8688 variable the value it had at the time of the @code{setjmp}. On some
8689 machines, however, @code{longjmp} does not change the value of global
8690 register variables. To be portable, the function that called @code{setjmp}
8691 should make other arrangements to save the values of the global register
8692 variables, and to restore them in a @code{longjmp}. This way, the same
8693 thing happens regardless of what @code{longjmp} does.
8695 Eventually there may be a way of asking the compiler to choose a register
8696 automatically, but first we need to figure out how it should choose and
8697 how to enable you to guide the choice. No solution is evident.
8699 @node Local Register Variables
8700 @subsubsection Specifying Registers for Local Variables
8701 @anchor{Local Reg Vars}
8702 @cindex local variables, specifying registers
8703 @cindex specifying registers for local variables
8704 @cindex registers for local variables
8706 You can define a local register variable and associate it with a specified
8710 register int *foo asm ("r12");
8714 Here @code{r12} is the name of the register that should be used. Note
8715 that this is the same syntax used for defining global register variables,
8716 but for a local variable the declaration appears within a function. The
8717 @code{register} keyword is required, and cannot be combined with
8718 @code{static}. The register name must be a valid register name for the
8721 As with global register variables, it is recommended that you choose
8722 a register that is normally saved and restored by function calls on your
8723 machine, so that calls to library routines will not clobber it.
8725 The only supported use for this feature is to specify registers
8726 for input and output operands when calling Extended @code{asm}
8727 (@pxref{Extended Asm}). This may be necessary if the constraints for a
8728 particular machine don't provide sufficient control to select the desired
8729 register. To force an operand into a register, create a local variable
8730 and specify the register name after the variable's declaration. Then use
8731 the local variable for the @code{asm} operand and specify any constraint
8732 letter that matches the register:
8735 register int *p1 asm ("r0") = @dots{};
8736 register int *p2 asm ("r1") = @dots{};
8737 register int *result asm ("r0");
8738 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
8741 @emph{Warning:} In the above example, be aware that a register (for example
8742 @code{r0}) can be call-clobbered by subsequent code, including function
8743 calls and library calls for arithmetic operators on other variables (for
8744 example the initialization of @code{p2}). In this case, use temporary
8745 variables for expressions between the register assignments:
8749 register int *p1 asm ("r0") = @dots{};
8750 register int *p2 asm ("r1") = t1;
8751 register int *result asm ("r0");
8752 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
8755 Defining a register variable does not reserve the register. Other than
8756 when invoking the Extended @code{asm}, the contents of the specified
8757 register are not guaranteed. For this reason, the following uses
8758 are explicitly @emph{not} supported. If they appear to work, it is only
8759 happenstance, and may stop working as intended due to (seemingly)
8760 unrelated changes in surrounding code, or even minor changes in the
8761 optimization of a future version of gcc:
8764 @item Passing parameters to or from Basic @code{asm}
8765 @item Passing parameters to or from Extended @code{asm} without using input
8767 @item Passing parameters to or from routines written in assembler (or
8768 other languages) using non-standard calling conventions.
8771 Some developers use Local Register Variables in an attempt to improve
8772 gcc's allocation of registers, especially in large functions. In this
8773 case the register name is essentially a hint to the register allocator.
8774 While in some instances this can generate better code, improvements are
8775 subject to the whims of the allocator/optimizers. Since there are no
8776 guarantees that your improvements won't be lost, this usage of Local
8777 Register Variables is discouraged.
8779 On the MIPS platform, there is related use for local register variables
8780 with slightly different characteristics (@pxref{MIPS Coprocessors,,
8781 Defining coprocessor specifics for MIPS targets, gccint,
8782 GNU Compiler Collection (GCC) Internals}).
8784 @node Size of an asm
8785 @subsection Size of an @code{asm}
8787 Some targets require that GCC track the size of each instruction used
8788 in order to generate correct code. Because the final length of the
8789 code produced by an @code{asm} statement is only known by the
8790 assembler, GCC must make an estimate as to how big it will be. It
8791 does this by counting the number of instructions in the pattern of the
8792 @code{asm} and multiplying that by the length of the longest
8793 instruction supported by that processor. (When working out the number
8794 of instructions, it assumes that any occurrence of a newline or of
8795 whatever statement separator character is supported by the assembler --
8796 typically @samp{;} --- indicates the end of an instruction.)
8798 Normally, GCC's estimate is adequate to ensure that correct
8799 code is generated, but it is possible to confuse the compiler if you use
8800 pseudo instructions or assembler macros that expand into multiple real
8801 instructions, or if you use assembler directives that expand to more
8802 space in the object file than is needed for a single instruction.
8803 If this happens then the assembler may produce a diagnostic saying that
8804 a label is unreachable.
8806 @node Alternate Keywords
8807 @section Alternate Keywords
8808 @cindex alternate keywords
8809 @cindex keywords, alternate
8811 @option{-ansi} and the various @option{-std} options disable certain
8812 keywords. This causes trouble when you want to use GNU C extensions, or
8813 a general-purpose header file that should be usable by all programs,
8814 including ISO C programs. The keywords @code{asm}, @code{typeof} and
8815 @code{inline} are not available in programs compiled with
8816 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
8817 program compiled with @option{-std=c99} or @option{-std=c11}). The
8819 @code{restrict} is only available when @option{-std=gnu99} (which will
8820 eventually be the default) or @option{-std=c99} (or the equivalent
8821 @option{-std=iso9899:1999}), or an option for a later standard
8824 The way to solve these problems is to put @samp{__} at the beginning and
8825 end of each problematical keyword. For example, use @code{__asm__}
8826 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
8828 Other C compilers won't accept these alternative keywords; if you want to
8829 compile with another compiler, you can define the alternate keywords as
8830 macros to replace them with the customary keywords. It looks like this:
8838 @findex __extension__
8840 @option{-pedantic} and other options cause warnings for many GNU C extensions.
8842 prevent such warnings within one expression by writing
8843 @code{__extension__} before the expression. @code{__extension__} has no
8844 effect aside from this.
8846 @node Incomplete Enums
8847 @section Incomplete @code{enum} Types
8849 You can define an @code{enum} tag without specifying its possible values.
8850 This results in an incomplete type, much like what you get if you write
8851 @code{struct foo} without describing the elements. A later declaration
8852 that does specify the possible values completes the type.
8854 You can't allocate variables or storage using the type while it is
8855 incomplete. However, you can work with pointers to that type.
8857 This extension may not be very useful, but it makes the handling of
8858 @code{enum} more consistent with the way @code{struct} and @code{union}
8861 This extension is not supported by GNU C++.
8863 @node Function Names
8864 @section Function Names as Strings
8865 @cindex @code{__func__} identifier
8866 @cindex @code{__FUNCTION__} identifier
8867 @cindex @code{__PRETTY_FUNCTION__} identifier
8869 GCC provides three magic variables that hold the name of the current
8870 function, as a string. The first of these is @code{__func__}, which
8871 is part of the C99 standard:
8873 The identifier @code{__func__} is implicitly declared by the translator
8874 as if, immediately following the opening brace of each function
8875 definition, the declaration
8878 static const char __func__[] = "function-name";
8882 appeared, where function-name is the name of the lexically-enclosing
8883 function. This name is the unadorned name of the function.
8885 @code{__FUNCTION__} is another name for @code{__func__}, provided for
8886 backward compatibility with old versions of GCC.
8888 In C, @code{__PRETTY_FUNCTION__} is yet another name for
8889 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
8890 the type signature of the function as well as its bare name. For
8891 example, this program:
8895 extern int printf (char *, ...);
8902 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
8903 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
8921 __PRETTY_FUNCTION__ = void a::sub(int)
8924 These identifiers are variables, not preprocessor macros, and may not
8925 be used to initialize @code{char} arrays or be concatenated with other string
8928 @node Return Address
8929 @section Getting the Return or Frame Address of a Function
8931 These functions may be used to get information about the callers of a
8934 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
8935 This function returns the return address of the current function, or of
8936 one of its callers. The @var{level} argument is number of frames to
8937 scan up the call stack. A value of @code{0} yields the return address
8938 of the current function, a value of @code{1} yields the return address
8939 of the caller of the current function, and so forth. When inlining
8940 the expected behavior is that the function returns the address of
8941 the function that is returned to. To work around this behavior use
8942 the @code{noinline} function attribute.
8944 The @var{level} argument must be a constant integer.
8946 On some machines it may be impossible to determine the return address of
8947 any function other than the current one; in such cases, or when the top
8948 of the stack has been reached, this function returns @code{0} or a
8949 random value. In addition, @code{__builtin_frame_address} may be used
8950 to determine if the top of the stack has been reached.
8952 Additional post-processing of the returned value may be needed, see
8953 @code{__builtin_extract_return_addr}.
8955 Calling this function with a nonzero argument can have unpredictable
8956 effects, including crashing the calling program. As a result, calls
8957 that are considered unsafe are diagnosed when the @option{-Wframe-address}
8958 option is in effect. Such calls should only be made in debugging
8962 @deftypefn {Built-in Function} {void *} __builtin_extract_return_addr (void *@var{addr})
8963 The address as returned by @code{__builtin_return_address} may have to be fed
8964 through this function to get the actual encoded address. For example, on the
8965 31-bit S/390 platform the highest bit has to be masked out, or on SPARC
8966 platforms an offset has to be added for the true next instruction to be
8969 If no fixup is needed, this function simply passes through @var{addr}.
8972 @deftypefn {Built-in Function} {void *} __builtin_frob_return_address (void *@var{addr})
8973 This function does the reverse of @code{__builtin_extract_return_addr}.
8976 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
8977 This function is similar to @code{__builtin_return_address}, but it
8978 returns the address of the function frame rather than the return address
8979 of the function. Calling @code{__builtin_frame_address} with a value of
8980 @code{0} yields the frame address of the current function, a value of
8981 @code{1} yields the frame address of the caller of the current function,
8984 The frame is the area on the stack that holds local variables and saved
8985 registers. The frame address is normally the address of the first word
8986 pushed on to the stack by the function. However, the exact definition
8987 depends upon the processor and the calling convention. If the processor
8988 has a dedicated frame pointer register, and the function has a frame,
8989 then @code{__builtin_frame_address} returns the value of the frame
8992 On some machines it may be impossible to determine the frame address of
8993 any function other than the current one; in such cases, or when the top
8994 of the stack has been reached, this function returns @code{0} if
8995 the first frame pointer is properly initialized by the startup code.
8997 Calling this function with a nonzero argument can have unpredictable
8998 effects, including crashing the calling program. As a result, calls
8999 that are considered unsafe are diagnosed when the @option{-Wframe-address}
9000 option is in effect. Such calls should only be made in debugging
9004 @node Vector Extensions
9005 @section Using Vector Instructions through Built-in Functions
9007 On some targets, the instruction set contains SIMD vector instructions which
9008 operate on multiple values contained in one large register at the same time.
9009 For example, on the x86 the MMX, 3DNow!@: and SSE extensions can be used
9012 The first step in using these extensions is to provide the necessary data
9013 types. This should be done using an appropriate @code{typedef}:
9016 typedef int v4si __attribute__ ((vector_size (16)));
9020 The @code{int} type specifies the base type, while the attribute specifies
9021 the vector size for the variable, measured in bytes. For example, the
9022 declaration above causes the compiler to set the mode for the @code{v4si}
9023 type to be 16 bytes wide and divided into @code{int} sized units. For
9024 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
9025 corresponding mode of @code{foo} is @acronym{V4SI}.
9027 The @code{vector_size} attribute is only applicable to integral and
9028 float scalars, although arrays, pointers, and function return values
9029 are allowed in conjunction with this construct. Only sizes that are
9030 a power of two are currently allowed.
9032 All the basic integer types can be used as base types, both as signed
9033 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
9034 @code{long long}. In addition, @code{float} and @code{double} can be
9035 used to build floating-point vector types.
9037 Specifying a combination that is not valid for the current architecture
9038 causes GCC to synthesize the instructions using a narrower mode.
9039 For example, if you specify a variable of type @code{V4SI} and your
9040 architecture does not allow for this specific SIMD type, GCC
9041 produces code that uses 4 @code{SIs}.
9043 The types defined in this manner can be used with a subset of normal C
9044 operations. Currently, GCC allows using the following operators
9045 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~, %}@.
9047 The operations behave like C++ @code{valarrays}. Addition is defined as
9048 the addition of the corresponding elements of the operands. For
9049 example, in the code below, each of the 4 elements in @var{a} is
9050 added to the corresponding 4 elements in @var{b} and the resulting
9051 vector is stored in @var{c}.
9054 typedef int v4si __attribute__ ((vector_size (16)));
9061 Subtraction, multiplication, division, and the logical operations
9062 operate in a similar manner. Likewise, the result of using the unary
9063 minus or complement operators on a vector type is a vector whose
9064 elements are the negative or complemented values of the corresponding
9065 elements in the operand.
9067 It is possible to use shifting operators @code{<<}, @code{>>} on
9068 integer-type vectors. The operation is defined as following: @code{@{a0,
9069 a1, @dots{}, an@} >> @{b0, b1, @dots{}, bn@} == @{a0 >> b0, a1 >> b1,
9070 @dots{}, an >> bn@}}@. Vector operands must have the same number of
9073 For convenience, it is allowed to use a binary vector operation
9074 where one operand is a scalar. In that case the compiler transforms
9075 the scalar operand into a vector where each element is the scalar from
9076 the operation. The transformation happens only if the scalar could be
9077 safely converted to the vector-element type.
9078 Consider the following code.
9081 typedef int v4si __attribute__ ((vector_size (16)));
9086 a = b + 1; /* a = b + @{1,1,1,1@}; */
9087 a = 2 * b; /* a = @{2,2,2,2@} * b; */
9089 a = l + a; /* Error, cannot convert long to int. */
9092 Vectors can be subscripted as if the vector were an array with
9093 the same number of elements and base type. Out of bound accesses
9094 invoke undefined behavior at run time. Warnings for out of bound
9095 accesses for vector subscription can be enabled with
9096 @option{-Warray-bounds}.
9098 Vector comparison is supported with standard comparison
9099 operators: @code{==, !=, <, <=, >, >=}. Comparison operands can be
9100 vector expressions of integer-type or real-type. Comparison between
9101 integer-type vectors and real-type vectors are not supported. The
9102 result of the comparison is a vector of the same width and number of
9103 elements as the comparison operands with a signed integral element
9106 Vectors are compared element-wise producing 0 when comparison is false
9107 and -1 (constant of the appropriate type where all bits are set)
9108 otherwise. Consider the following example.
9111 typedef int v4si __attribute__ ((vector_size (16)));
9113 v4si a = @{1,2,3,4@};
9114 v4si b = @{3,2,1,4@};
9117 c = a > b; /* The result would be @{0, 0,-1, 0@} */
9118 c = a == b; /* The result would be @{0,-1, 0,-1@} */
9121 In C++, the ternary operator @code{?:} is available. @code{a?b:c}, where
9122 @code{b} and @code{c} are vectors of the same type and @code{a} is an
9123 integer vector with the same number of elements of the same size as @code{b}
9124 and @code{c}, computes all three arguments and creates a vector
9125 @code{@{a[0]?b[0]:c[0], a[1]?b[1]:c[1], @dots{}@}}. Note that unlike in
9126 OpenCL, @code{a} is thus interpreted as @code{a != 0} and not @code{a < 0}.
9127 As in the case of binary operations, this syntax is also accepted when
9128 one of @code{b} or @code{c} is a scalar that is then transformed into a
9129 vector. If both @code{b} and @code{c} are scalars and the type of
9130 @code{true?b:c} has the same size as the element type of @code{a}, then
9131 @code{b} and @code{c} are converted to a vector type whose elements have
9132 this type and with the same number of elements as @code{a}.
9134 In C++, the logic operators @code{!, &&, ||} are available for vectors.
9135 @code{!v} is equivalent to @code{v == 0}, @code{a && b} is equivalent to
9136 @code{a!=0 & b!=0} and @code{a || b} is equivalent to @code{a!=0 | b!=0}.
9137 For mixed operations between a scalar @code{s} and a vector @code{v},
9138 @code{s && v} is equivalent to @code{s?v!=0:0} (the evaluation is
9139 short-circuit) and @code{v && s} is equivalent to @code{v!=0 & (s?-1:0)}.
9141 Vector shuffling is available using functions
9142 @code{__builtin_shuffle (vec, mask)} and
9143 @code{__builtin_shuffle (vec0, vec1, mask)}.
9144 Both functions construct a permutation of elements from one or two
9145 vectors and return a vector of the same type as the input vector(s).
9146 The @var{mask} is an integral vector with the same width (@var{W})
9147 and element count (@var{N}) as the output vector.
9149 The elements of the input vectors are numbered in memory ordering of
9150 @var{vec0} beginning at 0 and @var{vec1} beginning at @var{N}. The
9151 elements of @var{mask} are considered modulo @var{N} in the single-operand
9152 case and modulo @math{2*@var{N}} in the two-operand case.
9154 Consider the following example,
9157 typedef int v4si __attribute__ ((vector_size (16)));
9159 v4si a = @{1,2,3,4@};
9160 v4si b = @{5,6,7,8@};
9161 v4si mask1 = @{0,1,1,3@};
9162 v4si mask2 = @{0,4,2,5@};
9165 res = __builtin_shuffle (a, mask1); /* res is @{1,2,2,4@} */
9166 res = __builtin_shuffle (a, b, mask2); /* res is @{1,5,3,6@} */
9169 Note that @code{__builtin_shuffle} is intentionally semantically
9170 compatible with the OpenCL @code{shuffle} and @code{shuffle2} functions.
9172 You can declare variables and use them in function calls and returns, as
9173 well as in assignments and some casts. You can specify a vector type as
9174 a return type for a function. Vector types can also be used as function
9175 arguments. It is possible to cast from one vector type to another,
9176 provided they are of the same size (in fact, you can also cast vectors
9177 to and from other datatypes of the same size).
9179 You cannot operate between vectors of different lengths or different
9180 signedness without a cast.
9183 @section Support for @code{offsetof}
9184 @findex __builtin_offsetof
9186 GCC implements for both C and C++ a syntactic extension to implement
9187 the @code{offsetof} macro.
9191 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
9193 offsetof_member_designator:
9195 | offsetof_member_designator "." @code{identifier}
9196 | offsetof_member_designator "[" @code{expr} "]"
9199 This extension is sufficient such that
9202 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
9206 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
9207 may be dependent. In either case, @var{member} may consist of a single
9208 identifier, or a sequence of member accesses and array references.
9210 @node __sync Builtins
9211 @section Legacy @code{__sync} Built-in Functions for Atomic Memory Access
9213 The following built-in functions
9214 are intended to be compatible with those described
9215 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
9216 section 7.4. As such, they depart from normal GCC practice by not using
9217 the @samp{__builtin_} prefix and also by being overloaded so that they
9218 work on multiple types.
9220 The definition given in the Intel documentation allows only for the use of
9221 the types @code{int}, @code{long}, @code{long long} or their unsigned
9222 counterparts. GCC allows any integral scalar or pointer type that is
9223 1, 2, 4 or 8 bytes in length.
9225 These functions are implemented in terms of the @samp{__atomic}
9226 builtins (@pxref{__atomic Builtins}). They should not be used for new
9227 code which should use the @samp{__atomic} builtins instead.
9229 Not all operations are supported by all target processors. If a particular
9230 operation cannot be implemented on the target processor, a warning is
9231 generated and a call to an external function is generated. The external
9232 function carries the same name as the built-in version,
9233 with an additional suffix
9234 @samp{_@var{n}} where @var{n} is the size of the data type.
9236 @c ??? Should we have a mechanism to suppress this warning? This is almost
9237 @c useful for implementing the operation under the control of an external
9240 In most cases, these built-in functions are considered a @dfn{full barrier}.
9242 no memory operand is moved across the operation, either forward or
9243 backward. Further, instructions are issued as necessary to prevent the
9244 processor from speculating loads across the operation and from queuing stores
9245 after the operation.
9247 All of the routines are described in the Intel documentation to take
9248 ``an optional list of variables protected by the memory barrier''. It's
9249 not clear what is meant by that; it could mean that @emph{only} the
9250 listed variables are protected, or it could mean a list of additional
9251 variables to be protected. The list is ignored by GCC which treats it as
9252 empty. GCC interprets an empty list as meaning that all globally
9253 accessible variables should be protected.
9256 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
9257 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
9258 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
9259 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
9260 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
9261 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
9262 @findex __sync_fetch_and_add
9263 @findex __sync_fetch_and_sub
9264 @findex __sync_fetch_and_or
9265 @findex __sync_fetch_and_and
9266 @findex __sync_fetch_and_xor
9267 @findex __sync_fetch_and_nand
9268 These built-in functions perform the operation suggested by the name, and
9269 returns the value that had previously been in memory. That is,
9272 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
9273 @{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; @} // nand
9276 @emph{Note:} GCC 4.4 and later implement @code{__sync_fetch_and_nand}
9277 as @code{*ptr = ~(tmp & value)} instead of @code{*ptr = ~tmp & value}.
9279 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
9280 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
9281 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
9282 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
9283 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
9284 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
9285 @findex __sync_add_and_fetch
9286 @findex __sync_sub_and_fetch
9287 @findex __sync_or_and_fetch
9288 @findex __sync_and_and_fetch
9289 @findex __sync_xor_and_fetch
9290 @findex __sync_nand_and_fetch
9291 These built-in functions perform the operation suggested by the name, and
9292 return the new value. That is,
9295 @{ *ptr @var{op}= value; return *ptr; @}
9296 @{ *ptr = ~(*ptr & value); return *ptr; @} // nand
9299 @emph{Note:} GCC 4.4 and later implement @code{__sync_nand_and_fetch}
9300 as @code{*ptr = ~(*ptr & value)} instead of
9301 @code{*ptr = ~*ptr & value}.
9303 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval, @var{type} newval, ...)
9304 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval, @var{type} newval, ...)
9305 @findex __sync_bool_compare_and_swap
9306 @findex __sync_val_compare_and_swap
9307 These built-in functions perform an atomic compare and swap.
9308 That is, if the current
9309 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
9312 The ``bool'' version returns true if the comparison is successful and
9313 @var{newval} is written. The ``val'' version returns the contents
9314 of @code{*@var{ptr}} before the operation.
9316 @item __sync_synchronize (...)
9317 @findex __sync_synchronize
9318 This built-in function issues a full memory barrier.
9320 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
9321 @findex __sync_lock_test_and_set
9322 This built-in function, as described by Intel, is not a traditional test-and-set
9323 operation, but rather an atomic exchange operation. It writes @var{value}
9324 into @code{*@var{ptr}}, and returns the previous contents of
9327 Many targets have only minimal support for such locks, and do not support
9328 a full exchange operation. In this case, a target may support reduced
9329 functionality here by which the @emph{only} valid value to store is the
9330 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
9331 is implementation defined.
9333 This built-in function is not a full barrier,
9334 but rather an @dfn{acquire barrier}.
9335 This means that references after the operation cannot move to (or be
9336 speculated to) before the operation, but previous memory stores may not
9337 be globally visible yet, and previous memory loads may not yet be
9340 @item void __sync_lock_release (@var{type} *ptr, ...)
9341 @findex __sync_lock_release
9342 This built-in function releases the lock acquired by
9343 @code{__sync_lock_test_and_set}.
9344 Normally this means writing the constant 0 to @code{*@var{ptr}}.
9346 This built-in function is not a full barrier,
9347 but rather a @dfn{release barrier}.
9348 This means that all previous memory stores are globally visible, and all
9349 previous memory loads have been satisfied, but following memory reads
9350 are not prevented from being speculated to before the barrier.
9353 @node __atomic Builtins
9354 @section Built-in Functions for Memory Model Aware Atomic Operations
9356 The following built-in functions approximately match the requirements
9357 for the C++11 memory model. They are all
9358 identified by being prefixed with @samp{__atomic} and most are
9359 overloaded so that they work with multiple types.
9361 These functions are intended to replace the legacy @samp{__sync}
9362 builtins. The main difference is that the memory order that is requested
9363 is a parameter to the functions. New code should always use the
9364 @samp{__atomic} builtins rather than the @samp{__sync} builtins.
9366 Note that the @samp{__atomic} builtins assume that programs will
9367 conform to the C++11 memory model. In particular, they assume
9368 that programs are free of data races. See the C++11 standard for
9369 detailed requirements.
9371 The @samp{__atomic} builtins can be used with any integral scalar or
9372 pointer type that is 1, 2, 4, or 8 bytes in length. 16-byte integral
9373 types are also allowed if @samp{__int128} (@pxref{__int128}) is
9374 supported by the architecture.
9376 The four non-arithmetic functions (load, store, exchange, and
9377 compare_exchange) all have a generic version as well. This generic
9378 version works on any data type. It uses the lock-free built-in function
9379 if the specific data type size makes that possible; otherwise, an
9380 external call is left to be resolved at run time. This external call is
9381 the same format with the addition of a @samp{size_t} parameter inserted
9382 as the first parameter indicating the size of the object being pointed to.
9383 All objects must be the same size.
9385 There are 6 different memory orders that can be specified. These map
9386 to the C++11 memory orders with the same names, see the C++11 standard
9387 or the @uref{http://gcc.gnu.org/wiki/Atomic/GCCMM/AtomicSync,GCC wiki
9388 on atomic synchronization} for detailed definitions. Individual
9389 targets may also support additional memory orders for use on specific
9390 architectures. Refer to the target documentation for details of
9393 An atomic operation can both constrain code motion and
9394 be mapped to hardware instructions for synchronization between threads
9395 (e.g., a fence). To which extent this happens is controlled by the
9396 memory orders, which are listed here in approximately ascending order of
9397 strength. The description of each memory order is only meant to roughly
9398 illustrate the effects and is not a specification; see the C++11
9399 memory model for precise semantics.
9402 @item __ATOMIC_RELAXED
9403 Implies no inter-thread ordering constraints.
9404 @item __ATOMIC_CONSUME
9405 This is currently implemented using the stronger @code{__ATOMIC_ACQUIRE}
9406 memory order because of a deficiency in C++11's semantics for
9407 @code{memory_order_consume}.
9408 @item __ATOMIC_ACQUIRE
9409 Creates an inter-thread happens-before constraint from the release (or
9410 stronger) semantic store to this acquire load. Can prevent hoisting
9411 of code to before the operation.
9412 @item __ATOMIC_RELEASE
9413 Creates an inter-thread happens-before constraint to acquire (or stronger)
9414 semantic loads that read from this release store. Can prevent sinking
9415 of code to after the operation.
9416 @item __ATOMIC_ACQ_REL
9417 Combines the effects of both @code{__ATOMIC_ACQUIRE} and
9418 @code{__ATOMIC_RELEASE}.
9419 @item __ATOMIC_SEQ_CST
9420 Enforces total ordering with all other @code{__ATOMIC_SEQ_CST} operations.
9423 Note that in the C++11 memory model, @emph{fences} (e.g.,
9424 @samp{__atomic_thread_fence}) take effect in combination with other
9425 atomic operations on specific memory locations (e.g., atomic loads);
9426 operations on specific memory locations do not necessarily affect other
9427 operations in the same way.
9429 Target architectures are encouraged to provide their own patterns for
9430 each of the atomic built-in functions. If no target is provided, the original
9431 non-memory model set of @samp{__sync} atomic built-in functions are
9432 used, along with any required synchronization fences surrounding it in
9433 order to achieve the proper behavior. Execution in this case is subject
9434 to the same restrictions as those built-in functions.
9436 If there is no pattern or mechanism to provide a lock-free instruction
9437 sequence, a call is made to an external routine with the same parameters
9438 to be resolved at run time.
9440 When implementing patterns for these built-in functions, the memory order
9441 parameter can be ignored as long as the pattern implements the most
9442 restrictive @code{__ATOMIC_SEQ_CST} memory order. Any of the other memory
9443 orders execute correctly with this memory order but they may not execute as
9444 efficiently as they could with a more appropriate implementation of the
9445 relaxed requirements.
9447 Note that the C++11 standard allows for the memory order parameter to be
9448 determined at run time rather than at compile time. These built-in
9449 functions map any run-time value to @code{__ATOMIC_SEQ_CST} rather
9450 than invoke a runtime library call or inline a switch statement. This is
9451 standard compliant, safe, and the simplest approach for now.
9453 The memory order parameter is a signed int, but only the lower 16 bits are
9454 reserved for the memory order. The remainder of the signed int is reserved
9455 for target use and should be 0. Use of the predefined atomic values
9456 ensures proper usage.
9458 @deftypefn {Built-in Function} @var{type} __atomic_load_n (@var{type} *ptr, int memorder)
9459 This built-in function implements an atomic load operation. It returns the
9460 contents of @code{*@var{ptr}}.
9462 The valid memory order variants are
9463 @code{__ATOMIC_RELAXED}, @code{__ATOMIC_SEQ_CST}, @code{__ATOMIC_ACQUIRE},
9464 and @code{__ATOMIC_CONSUME}.
9468 @deftypefn {Built-in Function} void __atomic_load (@var{type} *ptr, @var{type} *ret, int memorder)
9469 This is the generic version of an atomic load. It returns the
9470 contents of @code{*@var{ptr}} in @code{*@var{ret}}.
9474 @deftypefn {Built-in Function} void __atomic_store_n (@var{type} *ptr, @var{type} val, int memorder)
9475 This built-in function implements an atomic store operation. It writes
9476 @code{@var{val}} into @code{*@var{ptr}}.
9478 The valid memory order variants are
9479 @code{__ATOMIC_RELAXED}, @code{__ATOMIC_SEQ_CST}, and @code{__ATOMIC_RELEASE}.
9483 @deftypefn {Built-in Function} void __atomic_store (@var{type} *ptr, @var{type} *val, int memorder)
9484 This is the generic version of an atomic store. It stores the value
9485 of @code{*@var{val}} into @code{*@var{ptr}}.
9489 @deftypefn {Built-in Function} @var{type} __atomic_exchange_n (@var{type} *ptr, @var{type} val, int memorder)
9490 This built-in function implements an atomic exchange operation. It writes
9491 @var{val} into @code{*@var{ptr}}, and returns the previous contents of
9494 The valid memory order variants are
9495 @code{__ATOMIC_RELAXED}, @code{__ATOMIC_SEQ_CST}, @code{__ATOMIC_ACQUIRE},
9496 @code{__ATOMIC_RELEASE}, and @code{__ATOMIC_ACQ_REL}.
9500 @deftypefn {Built-in Function} void __atomic_exchange (@var{type} *ptr, @var{type} *val, @var{type} *ret, int memorder)
9501 This is the generic version of an atomic exchange. It stores the
9502 contents of @code{*@var{val}} into @code{*@var{ptr}}. The original value
9503 of @code{*@var{ptr}} is copied into @code{*@var{ret}}.
9507 @deftypefn {Built-in Function} bool __atomic_compare_exchange_n (@var{type} *ptr, @var{type} *expected, @var{type} desired, bool weak, int success_memorder, int failure_memorder)
9508 This built-in function implements an atomic compare and exchange operation.
9509 This compares the contents of @code{*@var{ptr}} with the contents of
9510 @code{*@var{expected}}. If equal, the operation is a @emph{read-modify-write}
9511 operation that writes @var{desired} into @code{*@var{ptr}}. If they are not
9512 equal, the operation is a @emph{read} and the current contents of
9513 @code{*@var{ptr}} is written into @code{*@var{expected}}. @var{weak} is true
9514 for weak compare_exchange, and false for the strong variation. Many targets
9515 only offer the strong variation and ignore the parameter. When in doubt, use
9516 the strong variation.
9518 True is returned if @var{desired} is written into
9519 @code{*@var{ptr}} and the operation is considered to conform to the
9520 memory order specified by @var{success_memorder}. There are no
9521 restrictions on what memory order can be used here.
9523 False is returned otherwise, and the operation is considered to conform
9524 to @var{failure_memorder}. This memory order cannot be
9525 @code{__ATOMIC_RELEASE} nor @code{__ATOMIC_ACQ_REL}. It also cannot be a
9526 stronger order than that specified by @var{success_memorder}.
9530 @deftypefn {Built-in Function} bool __atomic_compare_exchange (@var{type} *ptr, @var{type} *expected, @var{type} *desired, bool weak, int success_memorder, int failure_memorder)
9531 This built-in function implements the generic version of
9532 @code{__atomic_compare_exchange}. The function is virtually identical to
9533 @code{__atomic_compare_exchange_n}, except the desired value is also a
9538 @deftypefn {Built-in Function} @var{type} __atomic_add_fetch (@var{type} *ptr, @var{type} val, int memorder)
9539 @deftypefnx {Built-in Function} @var{type} __atomic_sub_fetch (@var{type} *ptr, @var{type} val, int memorder)
9540 @deftypefnx {Built-in Function} @var{type} __atomic_and_fetch (@var{type} *ptr, @var{type} val, int memorder)
9541 @deftypefnx {Built-in Function} @var{type} __atomic_xor_fetch (@var{type} *ptr, @var{type} val, int memorder)
9542 @deftypefnx {Built-in Function} @var{type} __atomic_or_fetch (@var{type} *ptr, @var{type} val, int memorder)
9543 @deftypefnx {Built-in Function} @var{type} __atomic_nand_fetch (@var{type} *ptr, @var{type} val, int memorder)
9544 These built-in functions perform the operation suggested by the name, and
9545 return the result of the operation. That is,
9548 @{ *ptr @var{op}= val; return *ptr; @}
9551 All memory orders are valid.
9555 @deftypefn {Built-in Function} @var{type} __atomic_fetch_add (@var{type} *ptr, @var{type} val, int memorder)
9556 @deftypefnx {Built-in Function} @var{type} __atomic_fetch_sub (@var{type} *ptr, @var{type} val, int memorder)
9557 @deftypefnx {Built-in Function} @var{type} __atomic_fetch_and (@var{type} *ptr, @var{type} val, int memorder)
9558 @deftypefnx {Built-in Function} @var{type} __atomic_fetch_xor (@var{type} *ptr, @var{type} val, int memorder)
9559 @deftypefnx {Built-in Function} @var{type} __atomic_fetch_or (@var{type} *ptr, @var{type} val, int memorder)
9560 @deftypefnx {Built-in Function} @var{type} __atomic_fetch_nand (@var{type} *ptr, @var{type} val, int memorder)
9561 These built-in functions perform the operation suggested by the name, and
9562 return the value that had previously been in @code{*@var{ptr}}. That is,
9565 @{ tmp = *ptr; *ptr @var{op}= val; return tmp; @}
9568 All memory orders are valid.
9572 @deftypefn {Built-in Function} bool __atomic_test_and_set (void *ptr, int memorder)
9574 This built-in function performs an atomic test-and-set operation on
9575 the byte at @code{*@var{ptr}}. The byte is set to some implementation
9576 defined nonzero ``set'' value and the return value is @code{true} if and only
9577 if the previous contents were ``set''.
9578 It should be only used for operands of type @code{bool} or @code{char}. For
9579 other types only part of the value may be set.
9581 All memory orders are valid.
9585 @deftypefn {Built-in Function} void __atomic_clear (bool *ptr, int memorder)
9587 This built-in function performs an atomic clear operation on
9588 @code{*@var{ptr}}. After the operation, @code{*@var{ptr}} contains 0.
9589 It should be only used for operands of type @code{bool} or @code{char} and
9590 in conjunction with @code{__atomic_test_and_set}.
9591 For other types it may only clear partially. If the type is not @code{bool}
9592 prefer using @code{__atomic_store}.
9594 The valid memory order variants are
9595 @code{__ATOMIC_RELAXED}, @code{__ATOMIC_SEQ_CST}, and
9596 @code{__ATOMIC_RELEASE}.
9600 @deftypefn {Built-in Function} void __atomic_thread_fence (int memorder)
9602 This built-in function acts as a synchronization fence between threads
9603 based on the specified memory order.
9605 All memory orders are valid.
9609 @deftypefn {Built-in Function} void __atomic_signal_fence (int memorder)
9611 This built-in function acts as a synchronization fence between a thread
9612 and signal handlers based in the same thread.
9614 All memory orders are valid.
9618 @deftypefn {Built-in Function} bool __atomic_always_lock_free (size_t size, void *ptr)
9620 This built-in function returns true if objects of @var{size} bytes always
9621 generate lock-free atomic instructions for the target architecture.
9622 @var{size} must resolve to a compile-time constant and the result also
9623 resolves to a compile-time constant.
9625 @var{ptr} is an optional pointer to the object that may be used to determine
9626 alignment. A value of 0 indicates typical alignment should be used. The
9627 compiler may also ignore this parameter.
9630 if (_atomic_always_lock_free (sizeof (long long), 0))
9635 @deftypefn {Built-in Function} bool __atomic_is_lock_free (size_t size, void *ptr)
9637 This built-in function returns true if objects of @var{size} bytes always
9638 generate lock-free atomic instructions for the target architecture. If
9639 the built-in function is not known to be lock-free, a call is made to a
9640 runtime routine named @code{__atomic_is_lock_free}.
9642 @var{ptr} is an optional pointer to the object that may be used to determine
9643 alignment. A value of 0 indicates typical alignment should be used. The
9644 compiler may also ignore this parameter.
9647 @node Integer Overflow Builtins
9648 @section Built-in Functions to Perform Arithmetic with Overflow Checking
9650 The following built-in functions allow performing simple arithmetic operations
9651 together with checking whether the operations overflowed.
9653 @deftypefn {Built-in Function} bool __builtin_add_overflow (@var{type1} a, @var{type2} b, @var{type3} *res)
9654 @deftypefnx {Built-in Function} bool __builtin_sadd_overflow (int a, int b, int *res)
9655 @deftypefnx {Built-in Function} bool __builtin_saddl_overflow (long int a, long int b, long int *res)
9656 @deftypefnx {Built-in Function} bool __builtin_saddll_overflow (long long int a, long long int b, long int *res)
9657 @deftypefnx {Built-in Function} bool __builtin_uadd_overflow (unsigned int a, unsigned int b, unsigned int *res)
9658 @deftypefnx {Built-in Function} bool __builtin_uaddl_overflow (unsigned long int a, unsigned long int b, unsigned long int *res)
9659 @deftypefnx {Built-in Function} bool __builtin_uaddll_overflow (unsigned long long int a, unsigned long long int b, unsigned long int *res)
9661 These built-in functions promote the first two operands into infinite precision signed
9662 type and perform addition on those promoted operands. The result is then
9663 cast to the type the third pointer argument points to and stored there.
9664 If the stored result is equal to the infinite precision result, the built-in
9665 functions return false, otherwise they return true. As the addition is
9666 performed in infinite signed precision, these built-in functions have fully defined
9667 behavior for all argument values.
9669 The first built-in function allows arbitrary integral types for operands and
9670 the result type must be pointer to some integer type, the rest of the built-in
9671 functions have explicit integer types.
9673 The compiler will attempt to use hardware instructions to implement
9674 these built-in functions where possible, like conditional jump on overflow
9675 after addition, conditional jump on carry etc.
9679 @deftypefn {Built-in Function} bool __builtin_sub_overflow (@var{type1} a, @var{type2} b, @var{type3} *res)
9680 @deftypefnx {Built-in Function} bool __builtin_ssub_overflow (int a, int b, int *res)
9681 @deftypefnx {Built-in Function} bool __builtin_ssubl_overflow (long int a, long int b, long int *res)
9682 @deftypefnx {Built-in Function} bool __builtin_ssubll_overflow (long long int a, long long int b, long int *res)
9683 @deftypefnx {Built-in Function} bool __builtin_usub_overflow (unsigned int a, unsigned int b, unsigned int *res)
9684 @deftypefnx {Built-in Function} bool __builtin_usubl_overflow (unsigned long int a, unsigned long int b, unsigned long int *res)
9685 @deftypefnx {Built-in Function} bool __builtin_usubll_overflow (unsigned long long int a, unsigned long long int b, unsigned long int *res)
9687 These built-in functions are similar to the add overflow checking built-in
9688 functions above, except they perform subtraction, subtract the second argument
9689 from the first one, instead of addition.
9693 @deftypefn {Built-in Function} bool __builtin_mul_overflow (@var{type1} a, @var{type2} b, @var{type3} *res)
9694 @deftypefnx {Built-in Function} bool __builtin_smul_overflow (int a, int b, int *res)
9695 @deftypefnx {Built-in Function} bool __builtin_smull_overflow (long int a, long int b, long int *res)
9696 @deftypefnx {Built-in Function} bool __builtin_smulll_overflow (long long int a, long long int b, long int *res)
9697 @deftypefnx {Built-in Function} bool __builtin_umul_overflow (unsigned int a, unsigned int b, unsigned int *res)
9698 @deftypefnx {Built-in Function} bool __builtin_umull_overflow (unsigned long int a, unsigned long int b, unsigned long int *res)
9699 @deftypefnx {Built-in Function} bool __builtin_umulll_overflow (unsigned long long int a, unsigned long long int b, unsigned long int *res)
9701 These built-in functions are similar to the add overflow checking built-in
9702 functions above, except they perform multiplication, instead of addition.
9706 @node x86 specific memory model extensions for transactional memory
9707 @section x86-Specific Memory Model Extensions for Transactional Memory
9709 The x86 architecture supports additional memory ordering flags
9710 to mark lock critical sections for hardware lock elision.
9711 These must be specified in addition to an existing memory order to
9715 @item __ATOMIC_HLE_ACQUIRE
9716 Start lock elision on a lock variable.
9717 Memory order must be @code{__ATOMIC_ACQUIRE} or stronger.
9718 @item __ATOMIC_HLE_RELEASE
9719 End lock elision on a lock variable.
9720 Memory order must be @code{__ATOMIC_RELEASE} or stronger.
9723 When a lock acquire fails, it is required for good performance to abort
9724 the transaction quickly. This can be done with a @code{_mm_pause}.
9727 #include <immintrin.h> // For _mm_pause
9731 /* Acquire lock with lock elision */
9732 while (__atomic_exchange_n(&lockvar, 1, __ATOMIC_ACQUIRE|__ATOMIC_HLE_ACQUIRE))
9733 _mm_pause(); /* Abort failed transaction */
9735 /* Free lock with lock elision */
9736 __atomic_store_n(&lockvar, 0, __ATOMIC_RELEASE|__ATOMIC_HLE_RELEASE);
9739 @node Object Size Checking
9740 @section Object Size Checking Built-in Functions
9741 @findex __builtin_object_size
9742 @findex __builtin___memcpy_chk
9743 @findex __builtin___mempcpy_chk
9744 @findex __builtin___memmove_chk
9745 @findex __builtin___memset_chk
9746 @findex __builtin___strcpy_chk
9747 @findex __builtin___stpcpy_chk
9748 @findex __builtin___strncpy_chk
9749 @findex __builtin___strcat_chk
9750 @findex __builtin___strncat_chk
9751 @findex __builtin___sprintf_chk
9752 @findex __builtin___snprintf_chk
9753 @findex __builtin___vsprintf_chk
9754 @findex __builtin___vsnprintf_chk
9755 @findex __builtin___printf_chk
9756 @findex __builtin___vprintf_chk
9757 @findex __builtin___fprintf_chk
9758 @findex __builtin___vfprintf_chk
9760 GCC implements a limited buffer overflow protection mechanism
9761 that can prevent some buffer overflow attacks.
9763 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
9764 is a built-in construct that returns a constant number of bytes from
9765 @var{ptr} to the end of the object @var{ptr} pointer points to
9766 (if known at compile time). @code{__builtin_object_size} never evaluates
9767 its arguments for side-effects. If there are any side-effects in them, it
9768 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
9769 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
9770 point to and all of them are known at compile time, the returned number
9771 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
9772 0 and minimum if nonzero. If it is not possible to determine which objects
9773 @var{ptr} points to at compile time, @code{__builtin_object_size} should
9774 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
9775 for @var{type} 2 or 3.
9777 @var{type} is an integer constant from 0 to 3. If the least significant
9778 bit is clear, objects are whole variables, if it is set, a closest
9779 surrounding subobject is considered the object a pointer points to.
9780 The second bit determines if maximum or minimum of remaining bytes
9784 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
9785 char *p = &var.buf1[1], *q = &var.b;
9787 /* Here the object p points to is var. */
9788 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
9789 /* The subobject p points to is var.buf1. */
9790 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
9791 /* The object q points to is var. */
9792 assert (__builtin_object_size (q, 0)
9793 == (char *) (&var + 1) - (char *) &var.b);
9794 /* The subobject q points to is var.b. */
9795 assert (__builtin_object_size (q, 1) == sizeof (var.b));
9799 There are built-in functions added for many common string operation
9800 functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk}
9801 built-in is provided. This built-in has an additional last argument,
9802 which is the number of bytes remaining in object the @var{dest}
9803 argument points to or @code{(size_t) -1} if the size is not known.
9805 The built-in functions are optimized into the normal string functions
9806 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
9807 it is known at compile time that the destination object will not
9808 be overflown. If the compiler can determine at compile time the
9809 object will be always overflown, it issues a warning.
9811 The intended use can be e.g.@:
9815 #define bos0(dest) __builtin_object_size (dest, 0)
9816 #define memcpy(dest, src, n) \
9817 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
9821 /* It is unknown what object p points to, so this is optimized
9822 into plain memcpy - no checking is possible. */
9823 memcpy (p, "abcde", n);
9824 /* Destination is known and length too. It is known at compile
9825 time there will be no overflow. */
9826 memcpy (&buf[5], "abcde", 5);
9827 /* Destination is known, but the length is not known at compile time.
9828 This will result in __memcpy_chk call that can check for overflow
9830 memcpy (&buf[5], "abcde", n);
9831 /* Destination is known and it is known at compile time there will
9832 be overflow. There will be a warning and __memcpy_chk call that
9833 will abort the program at run time. */
9834 memcpy (&buf[6], "abcde", 5);
9837 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
9838 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
9839 @code{strcat} and @code{strncat}.
9841 There are also checking built-in functions for formatted output functions.
9843 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
9844 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
9845 const char *fmt, ...);
9846 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
9848 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
9849 const char *fmt, va_list ap);
9852 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
9853 etc.@: functions and can contain implementation specific flags on what
9854 additional security measures the checking function might take, such as
9855 handling @code{%n} differently.
9857 The @var{os} argument is the object size @var{s} points to, like in the
9858 other built-in functions. There is a small difference in the behavior
9859 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
9860 optimized into the non-checking functions only if @var{flag} is 0, otherwise
9861 the checking function is called with @var{os} argument set to
9864 In addition to this, there are checking built-in functions
9865 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
9866 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
9867 These have just one additional argument, @var{flag}, right before
9868 format string @var{fmt}. If the compiler is able to optimize them to
9869 @code{fputc} etc.@: functions, it does, otherwise the checking function
9870 is called and the @var{flag} argument passed to it.
9872 @node Pointer Bounds Checker builtins
9873 @section Pointer Bounds Checker Built-in Functions
9874 @cindex Pointer Bounds Checker builtins
9875 @findex __builtin___bnd_set_ptr_bounds
9876 @findex __builtin___bnd_narrow_ptr_bounds
9877 @findex __builtin___bnd_copy_ptr_bounds
9878 @findex __builtin___bnd_init_ptr_bounds
9879 @findex __builtin___bnd_null_ptr_bounds
9880 @findex __builtin___bnd_store_ptr_bounds
9881 @findex __builtin___bnd_chk_ptr_lbounds
9882 @findex __builtin___bnd_chk_ptr_ubounds
9883 @findex __builtin___bnd_chk_ptr_bounds
9884 @findex __builtin___bnd_get_ptr_lbound
9885 @findex __builtin___bnd_get_ptr_ubound
9887 GCC provides a set of built-in functions to control Pointer Bounds Checker
9888 instrumentation. Note that all Pointer Bounds Checker builtins can be used
9889 even if you compile with Pointer Bounds Checker off
9890 (@option{-fno-check-pointer-bounds}).
9891 The behavior may differ in such case as documented below.
9893 @deftypefn {Built-in Function} {void *} __builtin___bnd_set_ptr_bounds (const void *@var{q}, size_t @var{size})
9895 This built-in function returns a new pointer with the value of @var{q}, and
9896 associate it with the bounds [@var{q}, @var{q}+@var{size}-1]. With Pointer
9897 Bounds Checker off, the built-in function just returns the first argument.
9900 extern void *__wrap_malloc (size_t n)
9902 void *p = (void *)__real_malloc (n);
9903 if (!p) return __builtin___bnd_null_ptr_bounds (p);
9904 return __builtin___bnd_set_ptr_bounds (p, n);
9910 @deftypefn {Built-in Function} {void *} __builtin___bnd_narrow_ptr_bounds (const void *@var{p}, const void *@var{q}, size_t @var{size})
9912 This built-in function returns a new pointer with the value of @var{p}
9913 and associates it with the narrowed bounds formed by the intersection
9914 of bounds associated with @var{q} and the bounds
9915 [@var{p}, @var{p} + @var{size} - 1].
9916 With Pointer Bounds Checker off, the built-in function just returns the first
9920 void init_objects (object *objs, size_t size)
9923 /* Initialize objects one-by-one passing pointers with bounds of
9924 an object, not the full array of objects. */
9925 for (i = 0; i < size; i++)
9926 init_object (__builtin___bnd_narrow_ptr_bounds (objs + i, objs,
9933 @deftypefn {Built-in Function} {void *} __builtin___bnd_copy_ptr_bounds (const void *@var{q}, const void *@var{r})
9935 This built-in function returns a new pointer with the value of @var{q},
9936 and associates it with the bounds already associated with pointer @var{r}.
9937 With Pointer Bounds Checker off, the built-in function just returns the first
9941 /* Here is a way to get pointer to object's field but
9942 still with the full object's bounds. */
9943 int *field_ptr = __builtin___bnd_copy_ptr_bounds (&objptr->int_field,
9949 @deftypefn {Built-in Function} {void *} __builtin___bnd_init_ptr_bounds (const void *@var{q})
9951 This built-in function returns a new pointer with the value of @var{q}, and
9952 associates it with INIT (allowing full memory access) bounds. With Pointer
9953 Bounds Checker off, the built-in function just returns the first argument.
9957 @deftypefn {Built-in Function} {void *} __builtin___bnd_null_ptr_bounds (const void *@var{q})
9959 This built-in function returns a new pointer with the value of @var{q}, and
9960 associates it with NULL (allowing no memory access) bounds. With Pointer
9961 Bounds Checker off, the built-in function just returns the first argument.
9965 @deftypefn {Built-in Function} void __builtin___bnd_store_ptr_bounds (const void **@var{ptr_addr}, const void *@var{ptr_val})
9967 This built-in function stores the bounds associated with pointer @var{ptr_val}
9968 and location @var{ptr_addr} into Bounds Table. This can be useful to propagate
9969 bounds from legacy code without touching the associated pointer's memory when
9970 pointers are copied as integers. With Pointer Bounds Checker off, the built-in
9971 function call is ignored.
9975 @deftypefn {Built-in Function} void __builtin___bnd_chk_ptr_lbounds (const void *@var{q})
9977 This built-in function checks if the pointer @var{q} is within the lower
9978 bound of its associated bounds. With Pointer Bounds Checker off, the built-in
9979 function call is ignored.
9982 extern void *__wrap_memset (void *dst, int c, size_t len)
9986 __builtin___bnd_chk_ptr_lbounds (dst);
9987 __builtin___bnd_chk_ptr_ubounds ((char *)dst + len - 1);
9988 __real_memset (dst, c, len);
9996 @deftypefn {Built-in Function} void __builtin___bnd_chk_ptr_ubounds (const void *@var{q})
9998 This built-in function checks if the pointer @var{q} is within the upper
9999 bound of its associated bounds. With Pointer Bounds Checker off, the built-in
10000 function call is ignored.
10004 @deftypefn {Built-in Function} void __builtin___bnd_chk_ptr_bounds (const void *@var{q}, size_t @var{size})
10006 This built-in function checks if [@var{q}, @var{q} + @var{size} - 1] is within
10007 the lower and upper bounds associated with @var{q}. With Pointer Bounds Checker
10008 off, the built-in function call is ignored.
10011 extern void *__wrap_memcpy (void *dst, const void *src, size_t n)
10015 __bnd_chk_ptr_bounds (dst, n);
10016 __bnd_chk_ptr_bounds (src, n);
10017 __real_memcpy (dst, src, n);
10025 @deftypefn {Built-in Function} {const void *} __builtin___bnd_get_ptr_lbound (const void *@var{q})
10027 This built-in function returns the lower bound associated
10028 with the pointer @var{q}, as a pointer value.
10029 This is useful for debugging using @code{printf}.
10030 With Pointer Bounds Checker off, the built-in function returns 0.
10033 void *lb = __builtin___bnd_get_ptr_lbound (q);
10034 void *ub = __builtin___bnd_get_ptr_ubound (q);
10035 printf ("q = %p lb(q) = %p ub(q) = %p", q, lb, ub);
10040 @deftypefn {Built-in Function} {const void *} __builtin___bnd_get_ptr_ubound (const void *@var{q})
10042 This built-in function returns the upper bound (which is a pointer) associated
10043 with the pointer @var{q}. With Pointer Bounds Checker off,
10044 the built-in function returns -1.
10048 @node Cilk Plus Builtins
10049 @section Cilk Plus C/C++ Language Extension Built-in Functions
10051 GCC provides support for the following built-in reduction functions if Cilk Plus
10052 is enabled. Cilk Plus can be enabled using the @option{-fcilkplus} flag.
10055 @item @code{__sec_implicit_index}
10056 @item @code{__sec_reduce}
10057 @item @code{__sec_reduce_add}
10058 @item @code{__sec_reduce_all_nonzero}
10059 @item @code{__sec_reduce_all_zero}
10060 @item @code{__sec_reduce_any_nonzero}
10061 @item @code{__sec_reduce_any_zero}
10062 @item @code{__sec_reduce_max}
10063 @item @code{__sec_reduce_min}
10064 @item @code{__sec_reduce_max_ind}
10065 @item @code{__sec_reduce_min_ind}
10066 @item @code{__sec_reduce_mul}
10067 @item @code{__sec_reduce_mutating}
10070 Further details and examples about these built-in functions are described
10071 in the Cilk Plus language manual which can be found at
10072 @uref{http://www.cilkplus.org}.
10074 @node Other Builtins
10075 @section Other Built-in Functions Provided by GCC
10076 @cindex built-in functions
10077 @findex __builtin_call_with_static_chain
10078 @findex __builtin_fpclassify
10079 @findex __builtin_isfinite
10080 @findex __builtin_isnormal
10081 @findex __builtin_isgreater
10082 @findex __builtin_isgreaterequal
10083 @findex __builtin_isinf_sign
10084 @findex __builtin_isless
10085 @findex __builtin_islessequal
10086 @findex __builtin_islessgreater
10087 @findex __builtin_isunordered
10088 @findex __builtin_powi
10089 @findex __builtin_powif
10090 @findex __builtin_powil
10248 @findex fprintf_unlocked
10250 @findex fputs_unlocked
10358 @findex nexttowardf
10359 @findex nexttowardl
10367 @findex printf_unlocked
10397 @findex signbitd128
10398 @findex significand
10399 @findex significandf
10400 @findex significandl
10428 @findex strncasecmp
10471 GCC provides a large number of built-in functions other than the ones
10472 mentioned above. Some of these are for internal use in the processing
10473 of exceptions or variable-length argument lists and are not
10474 documented here because they may change from time to time; we do not
10475 recommend general use of these functions.
10477 The remaining functions are provided for optimization purposes.
10479 With the exception of built-ins that have library equivalents such as
10480 the standard C library functions discussed below, or that expand to
10481 library calls, GCC built-in functions are always expanded inline and
10482 thus do not have corresponding entry points and their address cannot
10483 be obtained. Attempting to use them in an expression other than
10484 a function call results in a compile-time error.
10486 @opindex fno-builtin
10487 GCC includes built-in versions of many of the functions in the standard
10488 C library. These functions come in two forms: one whose names start with
10489 the @code{__builtin_} prefix, and the other without. Both forms have the
10490 same type (including prototype), the same address (when their address is
10491 taken), and the same meaning as the C library functions even if you specify
10492 the @option{-fno-builtin} option @pxref{C Dialect Options}). Many of these
10493 functions are only optimized in certain cases; if they are not optimized in
10494 a particular case, a call to the library function is emitted.
10498 Outside strict ISO C mode (@option{-ansi}, @option{-std=c90},
10499 @option{-std=c99} or @option{-std=c11}), the functions
10500 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
10501 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
10502 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
10503 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
10504 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
10505 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
10506 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
10507 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
10508 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
10509 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
10510 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
10511 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
10512 @code{signbitd64}, @code{signbitd128}, @code{significandf},
10513 @code{significandl}, @code{significand}, @code{sincosf},
10514 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
10515 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
10516 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
10517 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
10519 may be handled as built-in functions.
10520 All these functions have corresponding versions
10521 prefixed with @code{__builtin_}, which may be used even in strict C90
10524 The ISO C99 functions
10525 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
10526 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
10527 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
10528 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
10529 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
10530 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
10531 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
10532 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
10533 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
10534 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
10535 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
10536 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
10537 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
10538 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
10539 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
10540 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
10541 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
10542 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
10543 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
10544 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
10545 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
10546 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
10547 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
10548 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
10549 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
10550 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
10551 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
10552 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
10553 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
10554 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
10555 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
10556 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
10557 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
10558 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
10559 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
10560 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
10561 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
10562 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
10563 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
10564 are handled as built-in functions
10565 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
10567 There are also built-in versions of the ISO C99 functions
10568 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
10569 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
10570 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
10571 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
10572 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
10573 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
10574 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
10575 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
10576 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
10577 that are recognized in any mode since ISO C90 reserves these names for
10578 the purpose to which ISO C99 puts them. All these functions have
10579 corresponding versions prefixed with @code{__builtin_}.
10581 The ISO C94 functions
10582 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
10583 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
10584 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
10586 are handled as built-in functions
10587 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c90}).
10589 The ISO C90 functions
10590 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
10591 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
10592 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
10593 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
10594 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
10595 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
10596 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
10597 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
10598 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
10599 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
10600 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
10601 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
10602 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
10603 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
10604 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
10605 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
10606 are all recognized as built-in functions unless
10607 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
10608 is specified for an individual function). All of these functions have
10609 corresponding versions prefixed with @code{__builtin_}.
10611 GCC provides built-in versions of the ISO C99 floating-point comparison
10612 macros that avoid raising exceptions for unordered operands. They have
10613 the same names as the standard macros ( @code{isgreater},
10614 @code{isgreaterequal}, @code{isless}, @code{islessequal},
10615 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
10616 prefixed. We intend for a library implementor to be able to simply
10617 @code{#define} each standard macro to its built-in equivalent.
10618 In the same fashion, GCC provides @code{fpclassify}, @code{isfinite},
10619 @code{isinf_sign}, @code{isnormal} and @code{signbit} built-ins used with
10620 @code{__builtin_} prefixed. The @code{isinf} and @code{isnan}
10621 built-in functions appear both with and without the @code{__builtin_} prefix.
10623 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
10625 You can use the built-in function @code{__builtin_types_compatible_p} to
10626 determine whether two types are the same.
10628 This built-in function returns 1 if the unqualified versions of the
10629 types @var{type1} and @var{type2} (which are types, not expressions) are
10630 compatible, 0 otherwise. The result of this built-in function can be
10631 used in integer constant expressions.
10633 This built-in function ignores top level qualifiers (e.g., @code{const},
10634 @code{volatile}). For example, @code{int} is equivalent to @code{const
10637 The type @code{int[]} and @code{int[5]} are compatible. On the other
10638 hand, @code{int} and @code{char *} are not compatible, even if the size
10639 of their types, on the particular architecture are the same. Also, the
10640 amount of pointer indirection is taken into account when determining
10641 similarity. Consequently, @code{short *} is not similar to
10642 @code{short **}. Furthermore, two types that are typedefed are
10643 considered compatible if their underlying types are compatible.
10645 An @code{enum} type is not considered to be compatible with another
10646 @code{enum} type even if both are compatible with the same integer
10647 type; this is what the C standard specifies.
10648 For example, @code{enum @{foo, bar@}} is not similar to
10649 @code{enum @{hot, dog@}}.
10651 You typically use this function in code whose execution varies
10652 depending on the arguments' types. For example:
10657 typeof (x) tmp = (x); \
10658 if (__builtin_types_compatible_p (typeof (x), long double)) \
10659 tmp = foo_long_double (tmp); \
10660 else if (__builtin_types_compatible_p (typeof (x), double)) \
10661 tmp = foo_double (tmp); \
10662 else if (__builtin_types_compatible_p (typeof (x), float)) \
10663 tmp = foo_float (tmp); \
10670 @emph{Note:} This construct is only available for C@.
10674 @deftypefn {Built-in Function} @var{type} __builtin_call_with_static_chain (@var{call_exp}, @var{pointer_exp})
10676 The @var{call_exp} expression must be a function call, and the
10677 @var{pointer_exp} expression must be a pointer. The @var{pointer_exp}
10678 is passed to the function call in the target's static chain location.
10679 The result of builtin is the result of the function call.
10681 @emph{Note:} This builtin is only available for C@.
10682 This builtin can be used to call Go closures from C.
10686 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
10688 You can use the built-in function @code{__builtin_choose_expr} to
10689 evaluate code depending on the value of a constant expression. This
10690 built-in function returns @var{exp1} if @var{const_exp}, which is an
10691 integer constant expression, is nonzero. Otherwise it returns @var{exp2}.
10693 This built-in function is analogous to the @samp{? :} operator in C,
10694 except that the expression returned has its type unaltered by promotion
10695 rules. Also, the built-in function does not evaluate the expression
10696 that is not chosen. For example, if @var{const_exp} evaluates to true,
10697 @var{exp2} is not evaluated even if it has side-effects.
10699 This built-in function can return an lvalue if the chosen argument is an
10702 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
10703 type. Similarly, if @var{exp2} is returned, its return type is the same
10710 __builtin_choose_expr ( \
10711 __builtin_types_compatible_p (typeof (x), double), \
10713 __builtin_choose_expr ( \
10714 __builtin_types_compatible_p (typeof (x), float), \
10716 /* @r{The void expression results in a compile-time error} \
10717 @r{when assigning the result to something.} */ \
10721 @emph{Note:} This construct is only available for C@. Furthermore, the
10722 unused expression (@var{exp1} or @var{exp2} depending on the value of
10723 @var{const_exp}) may still generate syntax errors. This may change in
10728 @deftypefn {Built-in Function} @var{type} __builtin_complex (@var{real}, @var{imag})
10730 The built-in function @code{__builtin_complex} is provided for use in
10731 implementing the ISO C11 macros @code{CMPLXF}, @code{CMPLX} and
10732 @code{CMPLXL}. @var{real} and @var{imag} must have the same type, a
10733 real binary floating-point type, and the result has the corresponding
10734 complex type with real and imaginary parts @var{real} and @var{imag}.
10735 Unlike @samp{@var{real} + I * @var{imag}}, this works even when
10736 infinities, NaNs and negative zeros are involved.
10740 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
10741 You can use the built-in function @code{__builtin_constant_p} to
10742 determine if a value is known to be constant at compile time and hence
10743 that GCC can perform constant-folding on expressions involving that
10744 value. The argument of the function is the value to test. The function
10745 returns the integer 1 if the argument is known to be a compile-time
10746 constant and 0 if it is not known to be a compile-time constant. A
10747 return of 0 does not indicate that the value is @emph{not} a constant,
10748 but merely that GCC cannot prove it is a constant with the specified
10749 value of the @option{-O} option.
10751 You typically use this function in an embedded application where
10752 memory is a critical resource. If you have some complex calculation,
10753 you may want it to be folded if it involves constants, but need to call
10754 a function if it does not. For example:
10757 #define Scale_Value(X) \
10758 (__builtin_constant_p (X) \
10759 ? ((X) * SCALE + OFFSET) : Scale (X))
10762 You may use this built-in function in either a macro or an inline
10763 function. However, if you use it in an inlined function and pass an
10764 argument of the function as the argument to the built-in, GCC
10765 never returns 1 when you call the inline function with a string constant
10766 or compound literal (@pxref{Compound Literals}) and does not return 1
10767 when you pass a constant numeric value to the inline function unless you
10768 specify the @option{-O} option.
10770 You may also use @code{__builtin_constant_p} in initializers for static
10771 data. For instance, you can write
10774 static const int table[] = @{
10775 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
10781 This is an acceptable initializer even if @var{EXPRESSION} is not a
10782 constant expression, including the case where
10783 @code{__builtin_constant_p} returns 1 because @var{EXPRESSION} can be
10784 folded to a constant but @var{EXPRESSION} contains operands that are
10785 not otherwise permitted in a static initializer (for example,
10786 @code{0 && foo ()}). GCC must be more conservative about evaluating the
10787 built-in in this case, because it has no opportunity to perform
10791 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
10792 @opindex fprofile-arcs
10793 You may use @code{__builtin_expect} to provide the compiler with
10794 branch prediction information. In general, you should prefer to
10795 use actual profile feedback for this (@option{-fprofile-arcs}), as
10796 programmers are notoriously bad at predicting how their programs
10797 actually perform. However, there are applications in which this
10798 data is hard to collect.
10800 The return value is the value of @var{exp}, which should be an integral
10801 expression. The semantics of the built-in are that it is expected that
10802 @var{exp} == @var{c}. For example:
10805 if (__builtin_expect (x, 0))
10810 indicates that we do not expect to call @code{foo}, since
10811 we expect @code{x} to be zero. Since you are limited to integral
10812 expressions for @var{exp}, you should use constructions such as
10815 if (__builtin_expect (ptr != NULL, 1))
10820 when testing pointer or floating-point values.
10823 @deftypefn {Built-in Function} void __builtin_trap (void)
10824 This function causes the program to exit abnormally. GCC implements
10825 this function by using a target-dependent mechanism (such as
10826 intentionally executing an illegal instruction) or by calling
10827 @code{abort}. The mechanism used may vary from release to release so
10828 you should not rely on any particular implementation.
10831 @deftypefn {Built-in Function} void __builtin_unreachable (void)
10832 If control flow reaches the point of the @code{__builtin_unreachable},
10833 the program is undefined. It is useful in situations where the
10834 compiler cannot deduce the unreachability of the code.
10836 One such case is immediately following an @code{asm} statement that
10837 either never terminates, or one that transfers control elsewhere
10838 and never returns. In this example, without the
10839 @code{__builtin_unreachable}, GCC issues a warning that control
10840 reaches the end of a non-void function. It also generates code
10841 to return after the @code{asm}.
10844 int f (int c, int v)
10852 asm("jmp error_handler");
10853 __builtin_unreachable ();
10859 Because the @code{asm} statement unconditionally transfers control out
10860 of the function, control never reaches the end of the function
10861 body. The @code{__builtin_unreachable} is in fact unreachable and
10862 communicates this fact to the compiler.
10864 Another use for @code{__builtin_unreachable} is following a call a
10865 function that never returns but that is not declared
10866 @code{__attribute__((noreturn))}, as in this example:
10869 void function_that_never_returns (void);
10879 function_that_never_returns ();
10880 __builtin_unreachable ();
10887 @deftypefn {Built-in Function} void *__builtin_assume_aligned (const void *@var{exp}, size_t @var{align}, ...)
10888 This function returns its first argument, and allows the compiler
10889 to assume that the returned pointer is at least @var{align} bytes
10890 aligned. This built-in can have either two or three arguments,
10891 if it has three, the third argument should have integer type, and
10892 if it is nonzero means misalignment offset. For example:
10895 void *x = __builtin_assume_aligned (arg, 16);
10899 means that the compiler can assume @code{x}, set to @code{arg}, is at least
10900 16-byte aligned, while:
10903 void *x = __builtin_assume_aligned (arg, 32, 8);
10907 means that the compiler can assume for @code{x}, set to @code{arg}, that
10908 @code{(char *) x - 8} is 32-byte aligned.
10911 @deftypefn {Built-in Function} int __builtin_LINE ()
10912 This function is the equivalent to the preprocessor @code{__LINE__}
10913 macro and returns the line number of the invocation of the built-in.
10914 In a C++ default argument for a function @var{F}, it gets the line number of
10915 the call to @var{F}.
10918 @deftypefn {Built-in Function} {const char *} __builtin_FUNCTION ()
10919 This function is the equivalent to the preprocessor @code{__FUNCTION__}
10920 macro and returns the function name the invocation of the built-in is in.
10923 @deftypefn {Built-in Function} {const char *} __builtin_FILE ()
10924 This function is the equivalent to the preprocessor @code{__FILE__}
10925 macro and returns the file name the invocation of the built-in is in.
10926 In a C++ default argument for a function @var{F}, it gets the file name of
10927 the call to @var{F}.
10930 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
10931 This function is used to flush the processor's instruction cache for
10932 the region of memory between @var{begin} inclusive and @var{end}
10933 exclusive. Some targets require that the instruction cache be
10934 flushed, after modifying memory containing code, in order to obtain
10935 deterministic behavior.
10937 If the target does not require instruction cache flushes,
10938 @code{__builtin___clear_cache} has no effect. Otherwise either
10939 instructions are emitted in-line to clear the instruction cache or a
10940 call to the @code{__clear_cache} function in libgcc is made.
10943 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
10944 This function is used to minimize cache-miss latency by moving data into
10945 a cache before it is accessed.
10946 You can insert calls to @code{__builtin_prefetch} into code for which
10947 you know addresses of data in memory that is likely to be accessed soon.
10948 If the target supports them, data prefetch instructions are generated.
10949 If the prefetch is done early enough before the access then the data will
10950 be in the cache by the time it is accessed.
10952 The value of @var{addr} is the address of the memory to prefetch.
10953 There are two optional arguments, @var{rw} and @var{locality}.
10954 The value of @var{rw} is a compile-time constant one or zero; one
10955 means that the prefetch is preparing for a write to the memory address
10956 and zero, the default, means that the prefetch is preparing for a read.
10957 The value @var{locality} must be a compile-time constant integer between
10958 zero and three. A value of zero means that the data has no temporal
10959 locality, so it need not be left in the cache after the access. A value
10960 of three means that the data has a high degree of temporal locality and
10961 should be left in all levels of cache possible. Values of one and two
10962 mean, respectively, a low or moderate degree of temporal locality. The
10966 for (i = 0; i < n; i++)
10968 a[i] = a[i] + b[i];
10969 __builtin_prefetch (&a[i+j], 1, 1);
10970 __builtin_prefetch (&b[i+j], 0, 1);
10975 Data prefetch does not generate faults if @var{addr} is invalid, but
10976 the address expression itself must be valid. For example, a prefetch
10977 of @code{p->next} does not fault if @code{p->next} is not a valid
10978 address, but evaluation faults if @code{p} is not a valid address.
10980 If the target does not support data prefetch, the address expression
10981 is evaluated if it includes side effects but no other code is generated
10982 and GCC does not issue a warning.
10985 @deftypefn {Built-in Function} double __builtin_huge_val (void)
10986 Returns a positive infinity, if supported by the floating-point format,
10987 else @code{DBL_MAX}. This function is suitable for implementing the
10988 ISO C macro @code{HUGE_VAL}.
10991 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
10992 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
10995 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
10996 Similar to @code{__builtin_huge_val}, except the return
10997 type is @code{long double}.
11000 @deftypefn {Built-in Function} int __builtin_fpclassify (int, int, int, int, int, ...)
11001 This built-in implements the C99 fpclassify functionality. The first
11002 five int arguments should be the target library's notion of the
11003 possible FP classes and are used for return values. They must be
11004 constant values and they must appear in this order: @code{FP_NAN},
11005 @code{FP_INFINITE}, @code{FP_NORMAL}, @code{FP_SUBNORMAL} and
11006 @code{FP_ZERO}. The ellipsis is for exactly one floating-point value
11007 to classify. GCC treats the last argument as type-generic, which
11008 means it does not do default promotion from float to double.
11011 @deftypefn {Built-in Function} double __builtin_inf (void)
11012 Similar to @code{__builtin_huge_val}, except a warning is generated
11013 if the target floating-point format does not support infinities.
11016 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
11017 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
11020 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
11021 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
11024 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
11025 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
11028 @deftypefn {Built-in Function} float __builtin_inff (void)
11029 Similar to @code{__builtin_inf}, except the return type is @code{float}.
11030 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
11033 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
11034 Similar to @code{__builtin_inf}, except the return
11035 type is @code{long double}.
11038 @deftypefn {Built-in Function} int __builtin_isinf_sign (...)
11039 Similar to @code{isinf}, except the return value is -1 for
11040 an argument of @code{-Inf} and 1 for an argument of @code{+Inf}.
11041 Note while the parameter list is an
11042 ellipsis, this function only accepts exactly one floating-point
11043 argument. GCC treats this parameter as type-generic, which means it
11044 does not do default promotion from float to double.
11047 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
11048 This is an implementation of the ISO C99 function @code{nan}.
11050 Since ISO C99 defines this function in terms of @code{strtod}, which we
11051 do not implement, a description of the parsing is in order. The string
11052 is parsed as by @code{strtol}; that is, the base is recognized by
11053 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
11054 in the significand such that the least significant bit of the number
11055 is at the least significant bit of the significand. The number is
11056 truncated to fit the significand field provided. The significand is
11057 forced to be a quiet NaN@.
11059 This function, if given a string literal all of which would have been
11060 consumed by @code{strtol}, is evaluated early enough that it is considered a
11061 compile-time constant.
11064 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
11065 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
11068 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
11069 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
11072 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
11073 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
11076 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
11077 Similar to @code{__builtin_nan}, except the return type is @code{float}.
11080 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
11081 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
11084 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
11085 Similar to @code{__builtin_nan}, except the significand is forced
11086 to be a signaling NaN@. The @code{nans} function is proposed by
11087 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
11090 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
11091 Similar to @code{__builtin_nans}, except the return type is @code{float}.
11094 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
11095 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
11098 @deftypefn {Built-in Function} int __builtin_ffs (int x)
11099 Returns one plus the index of the least significant 1-bit of @var{x}, or
11100 if @var{x} is zero, returns zero.
11103 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
11104 Returns the number of leading 0-bits in @var{x}, starting at the most
11105 significant bit position. If @var{x} is 0, the result is undefined.
11108 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
11109 Returns the number of trailing 0-bits in @var{x}, starting at the least
11110 significant bit position. If @var{x} is 0, the result is undefined.
11113 @deftypefn {Built-in Function} int __builtin_clrsb (int x)
11114 Returns the number of leading redundant sign bits in @var{x}, i.e.@: the
11115 number of bits following the most significant bit that are identical
11116 to it. There are no special cases for 0 or other values.
11119 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
11120 Returns the number of 1-bits in @var{x}.
11123 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
11124 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
11128 @deftypefn {Built-in Function} int __builtin_ffsl (long)
11129 Similar to @code{__builtin_ffs}, except the argument type is
11133 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
11134 Similar to @code{__builtin_clz}, except the argument type is
11135 @code{unsigned long}.
11138 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
11139 Similar to @code{__builtin_ctz}, except the argument type is
11140 @code{unsigned long}.
11143 @deftypefn {Built-in Function} int __builtin_clrsbl (long)
11144 Similar to @code{__builtin_clrsb}, except the argument type is
11148 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
11149 Similar to @code{__builtin_popcount}, except the argument type is
11150 @code{unsigned long}.
11153 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
11154 Similar to @code{__builtin_parity}, except the argument type is
11155 @code{unsigned long}.
11158 @deftypefn {Built-in Function} int __builtin_ffsll (long long)
11159 Similar to @code{__builtin_ffs}, except the argument type is
11163 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
11164 Similar to @code{__builtin_clz}, except the argument type is
11165 @code{unsigned long long}.
11168 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
11169 Similar to @code{__builtin_ctz}, except the argument type is
11170 @code{unsigned long long}.
11173 @deftypefn {Built-in Function} int __builtin_clrsbll (long long)
11174 Similar to @code{__builtin_clrsb}, except the argument type is
11178 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
11179 Similar to @code{__builtin_popcount}, except the argument type is
11180 @code{unsigned long long}.
11183 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
11184 Similar to @code{__builtin_parity}, except the argument type is
11185 @code{unsigned long long}.
11188 @deftypefn {Built-in Function} double __builtin_powi (double, int)
11189 Returns the first argument raised to the power of the second. Unlike the
11190 @code{pow} function no guarantees about precision and rounding are made.
11193 @deftypefn {Built-in Function} float __builtin_powif (float, int)
11194 Similar to @code{__builtin_powi}, except the argument and return types
11198 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
11199 Similar to @code{__builtin_powi}, except the argument and return types
11200 are @code{long double}.
11203 @deftypefn {Built-in Function} uint16_t __builtin_bswap16 (uint16_t x)
11204 Returns @var{x} with the order of the bytes reversed; for example,
11205 @code{0xaabb} becomes @code{0xbbaa}. Byte here always means
11209 @deftypefn {Built-in Function} uint32_t __builtin_bswap32 (uint32_t x)
11210 Similar to @code{__builtin_bswap16}, except the argument and return types
11214 @deftypefn {Built-in Function} uint64_t __builtin_bswap64 (uint64_t x)
11215 Similar to @code{__builtin_bswap32}, except the argument and return types
11219 @node Target Builtins
11220 @section Built-in Functions Specific to Particular Target Machines
11222 On some target machines, GCC supports many built-in functions specific
11223 to those machines. Generally these generate calls to specific machine
11224 instructions, but allow the compiler to schedule those calls.
11227 * AArch64 Built-in Functions::
11228 * Alpha Built-in Functions::
11229 * Altera Nios II Built-in Functions::
11230 * ARC Built-in Functions::
11231 * ARC SIMD Built-in Functions::
11232 * ARM iWMMXt Built-in Functions::
11233 * ARM C Language Extensions (ACLE)::
11234 * ARM Floating Point Status and Control Intrinsics::
11235 * AVR Built-in Functions::
11236 * Blackfin Built-in Functions::
11237 * FR-V Built-in Functions::
11238 * MIPS DSP Built-in Functions::
11239 * MIPS Paired-Single Support::
11240 * MIPS Loongson Built-in Functions::
11241 * Other MIPS Built-in Functions::
11242 * MSP430 Built-in Functions::
11243 * NDS32 Built-in Functions::
11244 * picoChip Built-in Functions::
11245 * PowerPC Built-in Functions::
11246 * PowerPC AltiVec/VSX Built-in Functions::
11247 * PowerPC Hardware Transactional Memory Built-in Functions::
11248 * RX Built-in Functions::
11249 * S/390 System z Built-in Functions::
11250 * SH Built-in Functions::
11251 * SPARC VIS Built-in Functions::
11252 * SPU Built-in Functions::
11253 * TI C6X Built-in Functions::
11254 * TILE-Gx Built-in Functions::
11255 * TILEPro Built-in Functions::
11256 * x86 Built-in Functions::
11257 * x86 transactional memory intrinsics::
11260 @node AArch64 Built-in Functions
11261 @subsection AArch64 Built-in Functions
11263 These built-in functions are available for the AArch64 family of
11266 unsigned int __builtin_aarch64_get_fpcr ()
11267 void __builtin_aarch64_set_fpcr (unsigned int)
11268 unsigned int __builtin_aarch64_get_fpsr ()
11269 void __builtin_aarch64_set_fpsr (unsigned int)
11272 @node Alpha Built-in Functions
11273 @subsection Alpha Built-in Functions
11275 These built-in functions are available for the Alpha family of
11276 processors, depending on the command-line switches used.
11278 The following built-in functions are always available. They
11279 all generate the machine instruction that is part of the name.
11282 long __builtin_alpha_implver (void)
11283 long __builtin_alpha_rpcc (void)
11284 long __builtin_alpha_amask (long)
11285 long __builtin_alpha_cmpbge (long, long)
11286 long __builtin_alpha_extbl (long, long)
11287 long __builtin_alpha_extwl (long, long)
11288 long __builtin_alpha_extll (long, long)
11289 long __builtin_alpha_extql (long, long)
11290 long __builtin_alpha_extwh (long, long)
11291 long __builtin_alpha_extlh (long, long)
11292 long __builtin_alpha_extqh (long, long)
11293 long __builtin_alpha_insbl (long, long)
11294 long __builtin_alpha_inswl (long, long)
11295 long __builtin_alpha_insll (long, long)
11296 long __builtin_alpha_insql (long, long)
11297 long __builtin_alpha_inswh (long, long)
11298 long __builtin_alpha_inslh (long, long)
11299 long __builtin_alpha_insqh (long, long)
11300 long __builtin_alpha_mskbl (long, long)
11301 long __builtin_alpha_mskwl (long, long)
11302 long __builtin_alpha_mskll (long, long)
11303 long __builtin_alpha_mskql (long, long)
11304 long __builtin_alpha_mskwh (long, long)
11305 long __builtin_alpha_msklh (long, long)
11306 long __builtin_alpha_mskqh (long, long)
11307 long __builtin_alpha_umulh (long, long)
11308 long __builtin_alpha_zap (long, long)
11309 long __builtin_alpha_zapnot (long, long)
11312 The following built-in functions are always with @option{-mmax}
11313 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
11314 later. They all generate the machine instruction that is part
11318 long __builtin_alpha_pklb (long)
11319 long __builtin_alpha_pkwb (long)
11320 long __builtin_alpha_unpkbl (long)
11321 long __builtin_alpha_unpkbw (long)
11322 long __builtin_alpha_minub8 (long, long)
11323 long __builtin_alpha_minsb8 (long, long)
11324 long __builtin_alpha_minuw4 (long, long)
11325 long __builtin_alpha_minsw4 (long, long)
11326 long __builtin_alpha_maxub8 (long, long)
11327 long __builtin_alpha_maxsb8 (long, long)
11328 long __builtin_alpha_maxuw4 (long, long)
11329 long __builtin_alpha_maxsw4 (long, long)
11330 long __builtin_alpha_perr (long, long)
11333 The following built-in functions are always with @option{-mcix}
11334 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
11335 later. They all generate the machine instruction that is part
11339 long __builtin_alpha_cttz (long)
11340 long __builtin_alpha_ctlz (long)
11341 long __builtin_alpha_ctpop (long)
11344 The following built-in functions are available on systems that use the OSF/1
11345 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
11346 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
11347 @code{rdval} and @code{wrval}.
11350 void *__builtin_thread_pointer (void)
11351 void __builtin_set_thread_pointer (void *)
11354 @node Altera Nios II Built-in Functions
11355 @subsection Altera Nios II Built-in Functions
11357 These built-in functions are available for the Altera Nios II
11358 family of processors.
11360 The following built-in functions are always available. They
11361 all generate the machine instruction that is part of the name.
11364 int __builtin_ldbio (volatile const void *)
11365 int __builtin_ldbuio (volatile const void *)
11366 int __builtin_ldhio (volatile const void *)
11367 int __builtin_ldhuio (volatile const void *)
11368 int __builtin_ldwio (volatile const void *)
11369 void __builtin_stbio (volatile void *, int)
11370 void __builtin_sthio (volatile void *, int)
11371 void __builtin_stwio (volatile void *, int)
11372 void __builtin_sync (void)
11373 int __builtin_rdctl (int)
11374 int __builtin_rdprs (int, int)
11375 void __builtin_wrctl (int, int)
11376 void __builtin_flushd (volatile void *)
11377 void __builtin_flushda (volatile void *)
11378 int __builtin_wrpie (int);
11379 void __builtin_eni (int);
11380 int __builtin_ldex (volatile const void *)
11381 int __builtin_stex (volatile void *, int)
11382 int __builtin_ldsex (volatile const void *)
11383 int __builtin_stsex (volatile void *, int)
11386 The following built-in functions are always available. They
11387 all generate a Nios II Custom Instruction. The name of the
11388 function represents the types that the function takes and
11389 returns. The letter before the @code{n} is the return type
11390 or void if absent. The @code{n} represents the first parameter
11391 to all the custom instructions, the custom instruction number.
11392 The two letters after the @code{n} represent the up to two
11393 parameters to the function.
11395 The letters represent the following data types:
11398 @code{void} for return type and no parameter for parameter types.
11401 @code{int} for return type and parameter type
11404 @code{float} for return type and parameter type
11407 @code{void *} for return type and parameter type
11411 And the function names are:
11413 void __builtin_custom_n (void)
11414 void __builtin_custom_ni (int)
11415 void __builtin_custom_nf (float)
11416 void __builtin_custom_np (void *)
11417 void __builtin_custom_nii (int, int)
11418 void __builtin_custom_nif (int, float)
11419 void __builtin_custom_nip (int, void *)
11420 void __builtin_custom_nfi (float, int)
11421 void __builtin_custom_nff (float, float)
11422 void __builtin_custom_nfp (float, void *)
11423 void __builtin_custom_npi (void *, int)
11424 void __builtin_custom_npf (void *, float)
11425 void __builtin_custom_npp (void *, void *)
11426 int __builtin_custom_in (void)
11427 int __builtin_custom_ini (int)
11428 int __builtin_custom_inf (float)
11429 int __builtin_custom_inp (void *)
11430 int __builtin_custom_inii (int, int)
11431 int __builtin_custom_inif (int, float)
11432 int __builtin_custom_inip (int, void *)
11433 int __builtin_custom_infi (float, int)
11434 int __builtin_custom_inff (float, float)
11435 int __builtin_custom_infp (float, void *)
11436 int __builtin_custom_inpi (void *, int)
11437 int __builtin_custom_inpf (void *, float)
11438 int __builtin_custom_inpp (void *, void *)
11439 float __builtin_custom_fn (void)
11440 float __builtin_custom_fni (int)
11441 float __builtin_custom_fnf (float)
11442 float __builtin_custom_fnp (void *)
11443 float __builtin_custom_fnii (int, int)
11444 float __builtin_custom_fnif (int, float)
11445 float __builtin_custom_fnip (int, void *)
11446 float __builtin_custom_fnfi (float, int)
11447 float __builtin_custom_fnff (float, float)
11448 float __builtin_custom_fnfp (float, void *)
11449 float __builtin_custom_fnpi (void *, int)
11450 float __builtin_custom_fnpf (void *, float)
11451 float __builtin_custom_fnpp (void *, void *)
11452 void * __builtin_custom_pn (void)
11453 void * __builtin_custom_pni (int)
11454 void * __builtin_custom_pnf (float)
11455 void * __builtin_custom_pnp (void *)
11456 void * __builtin_custom_pnii (int, int)
11457 void * __builtin_custom_pnif (int, float)
11458 void * __builtin_custom_pnip (int, void *)
11459 void * __builtin_custom_pnfi (float, int)
11460 void * __builtin_custom_pnff (float, float)
11461 void * __builtin_custom_pnfp (float, void *)
11462 void * __builtin_custom_pnpi (void *, int)
11463 void * __builtin_custom_pnpf (void *, float)
11464 void * __builtin_custom_pnpp (void *, void *)
11467 @node ARC Built-in Functions
11468 @subsection ARC Built-in Functions
11470 The following built-in functions are provided for ARC targets. The
11471 built-ins generate the corresponding assembly instructions. In the
11472 examples given below, the generated code often requires an operand or
11473 result to be in a register. Where necessary further code will be
11474 generated to ensure this is true, but for brevity this is not
11475 described in each case.
11477 @emph{Note:} Using a built-in to generate an instruction not supported
11478 by a target may cause problems. At present the compiler is not
11479 guaranteed to detect such misuse, and as a result an internal compiler
11480 error may be generated.
11482 @deftypefn {Built-in Function} int __builtin_arc_aligned (void *@var{val}, int @var{alignval})
11483 Return 1 if @var{val} is known to have the byte alignment given
11484 by @var{alignval}, otherwise return 0.
11485 Note that this is different from
11487 __alignof__(*(char *)@var{val}) >= alignval
11489 because __alignof__ sees only the type of the dereference, whereas
11490 __builtin_arc_align uses alignment information from the pointer
11491 as well as from the pointed-to type.
11492 The information available will depend on optimization level.
11495 @deftypefn {Built-in Function} void __builtin_arc_brk (void)
11502 @deftypefn {Built-in Function} {unsigned int} __builtin_arc_core_read (unsigned int @var{regno})
11503 The operand is the number of a register to be read. Generates:
11505 mov @var{dest}, r@var{regno}
11507 where the value in @var{dest} will be the result returned from the
11511 @deftypefn {Built-in Function} void __builtin_arc_core_write (unsigned int @var{regno}, unsigned int @var{val})
11512 The first operand is the number of a register to be written, the
11513 second operand is a compile time constant to write into that
11514 register. Generates:
11516 mov r@var{regno}, @var{val}
11520 @deftypefn {Built-in Function} int __builtin_arc_divaw (int @var{a}, int @var{b})
11521 Only available if either @option{-mcpu=ARC700} or @option{-meA} is set.
11524 divaw @var{dest}, @var{a}, @var{b}
11526 where the value in @var{dest} will be the result returned from the
11530 @deftypefn {Built-in Function} void __builtin_arc_flag (unsigned int @var{a})
11537 @deftypefn {Built-in Function} {unsigned int} __builtin_arc_lr (unsigned int @var{auxr})
11538 The operand, @var{auxv}, is the address of an auxiliary register and
11539 must be a compile time constant. Generates:
11541 lr @var{dest}, [@var{auxr}]
11543 Where the value in @var{dest} will be the result returned from the
11547 @deftypefn {Built-in Function} void __builtin_arc_mul64 (int @var{a}, int @var{b})
11548 Only available with @option{-mmul64}. Generates:
11550 mul64 @var{a}, @var{b}
11554 @deftypefn {Built-in Function} void __builtin_arc_mulu64 (unsigned int @var{a}, unsigned int @var{b})
11555 Only available with @option{-mmul64}. Generates:
11557 mulu64 @var{a}, @var{b}
11561 @deftypefn {Built-in Function} void __builtin_arc_nop (void)
11568 @deftypefn {Built-in Function} int __builtin_arc_norm (int @var{src})
11569 Only valid if the @samp{norm} instruction is available through the
11570 @option{-mnorm} option or by default with @option{-mcpu=ARC700}.
11573 norm @var{dest}, @var{src}
11575 Where the value in @var{dest} will be the result returned from the
11579 @deftypefn {Built-in Function} {short int} __builtin_arc_normw (short int @var{src})
11580 Only valid if the @samp{normw} instruction is available through the
11581 @option{-mnorm} option or by default with @option{-mcpu=ARC700}.
11584 normw @var{dest}, @var{src}
11586 Where the value in @var{dest} will be the result returned from the
11590 @deftypefn {Built-in Function} void __builtin_arc_rtie (void)
11597 @deftypefn {Built-in Function} void __builtin_arc_sleep (int @var{a}
11604 @deftypefn {Built-in Function} void __builtin_arc_sr (unsigned int @var{auxr}, unsigned int @var{val})
11605 The first argument, @var{auxv}, is the address of an auxiliary
11606 register, the second argument, @var{val}, is a compile time constant
11607 to be written to the register. Generates:
11609 sr @var{auxr}, [@var{val}]
11613 @deftypefn {Built-in Function} int __builtin_arc_swap (int @var{src})
11614 Only valid with @option{-mswap}. Generates:
11616 swap @var{dest}, @var{src}
11618 Where the value in @var{dest} will be the result returned from the
11622 @deftypefn {Built-in Function} void __builtin_arc_swi (void)
11629 @deftypefn {Built-in Function} void __builtin_arc_sync (void)
11630 Only available with @option{-mcpu=ARC700}. Generates:
11636 @deftypefn {Built-in Function} void __builtin_arc_trap_s (unsigned int @var{c})
11637 Only available with @option{-mcpu=ARC700}. Generates:
11643 @deftypefn {Built-in Function} void __builtin_arc_unimp_s (void)
11644 Only available with @option{-mcpu=ARC700}. Generates:
11650 The instructions generated by the following builtins are not
11651 considered as candidates for scheduling. They are not moved around by
11652 the compiler during scheduling, and thus can be expected to appear
11653 where they are put in the C code:
11655 __builtin_arc_brk()
11656 __builtin_arc_core_read()
11657 __builtin_arc_core_write()
11658 __builtin_arc_flag()
11660 __builtin_arc_sleep()
11662 __builtin_arc_swi()
11665 @node ARC SIMD Built-in Functions
11666 @subsection ARC SIMD Built-in Functions
11668 SIMD builtins provided by the compiler can be used to generate the
11669 vector instructions. This section describes the available builtins
11670 and their usage in programs. With the @option{-msimd} option, the
11671 compiler provides 128-bit vector types, which can be specified using
11672 the @code{vector_size} attribute. The header file @file{arc-simd.h}
11673 can be included to use the following predefined types:
11675 typedef int __v4si __attribute__((vector_size(16)));
11676 typedef short __v8hi __attribute__((vector_size(16)));
11679 These types can be used to define 128-bit variables. The built-in
11680 functions listed in the following section can be used on these
11681 variables to generate the vector operations.
11683 For all builtins, @code{__builtin_arc_@var{someinsn}}, the header file
11684 @file{arc-simd.h} also provides equivalent macros called
11685 @code{_@var{someinsn}} that can be used for programming ease and
11686 improved readability. The following macros for DMA control are also
11689 #define _setup_dma_in_channel_reg _vdiwr
11690 #define _setup_dma_out_channel_reg _vdowr
11693 The following is a complete list of all the SIMD built-ins provided
11694 for ARC, grouped by calling signature.
11696 The following take two @code{__v8hi} arguments and return a
11697 @code{__v8hi} result:
11699 __v8hi __builtin_arc_vaddaw (__v8hi, __v8hi)
11700 __v8hi __builtin_arc_vaddw (__v8hi, __v8hi)
11701 __v8hi __builtin_arc_vand (__v8hi, __v8hi)
11702 __v8hi __builtin_arc_vandaw (__v8hi, __v8hi)
11703 __v8hi __builtin_arc_vavb (__v8hi, __v8hi)
11704 __v8hi __builtin_arc_vavrb (__v8hi, __v8hi)
11705 __v8hi __builtin_arc_vbic (__v8hi, __v8hi)
11706 __v8hi __builtin_arc_vbicaw (__v8hi, __v8hi)
11707 __v8hi __builtin_arc_vdifaw (__v8hi, __v8hi)
11708 __v8hi __builtin_arc_vdifw (__v8hi, __v8hi)
11709 __v8hi __builtin_arc_veqw (__v8hi, __v8hi)
11710 __v8hi __builtin_arc_vh264f (__v8hi, __v8hi)
11711 __v8hi __builtin_arc_vh264ft (__v8hi, __v8hi)
11712 __v8hi __builtin_arc_vh264fw (__v8hi, __v8hi)
11713 __v8hi __builtin_arc_vlew (__v8hi, __v8hi)
11714 __v8hi __builtin_arc_vltw (__v8hi, __v8hi)
11715 __v8hi __builtin_arc_vmaxaw (__v8hi, __v8hi)
11716 __v8hi __builtin_arc_vmaxw (__v8hi, __v8hi)
11717 __v8hi __builtin_arc_vminaw (__v8hi, __v8hi)
11718 __v8hi __builtin_arc_vminw (__v8hi, __v8hi)
11719 __v8hi __builtin_arc_vmr1aw (__v8hi, __v8hi)
11720 __v8hi __builtin_arc_vmr1w (__v8hi, __v8hi)
11721 __v8hi __builtin_arc_vmr2aw (__v8hi, __v8hi)
11722 __v8hi __builtin_arc_vmr2w (__v8hi, __v8hi)
11723 __v8hi __builtin_arc_vmr3aw (__v8hi, __v8hi)
11724 __v8hi __builtin_arc_vmr3w (__v8hi, __v8hi)
11725 __v8hi __builtin_arc_vmr4aw (__v8hi, __v8hi)
11726 __v8hi __builtin_arc_vmr4w (__v8hi, __v8hi)
11727 __v8hi __builtin_arc_vmr5aw (__v8hi, __v8hi)
11728 __v8hi __builtin_arc_vmr5w (__v8hi, __v8hi)
11729 __v8hi __builtin_arc_vmr6aw (__v8hi, __v8hi)
11730 __v8hi __builtin_arc_vmr6w (__v8hi, __v8hi)
11731 __v8hi __builtin_arc_vmr7aw (__v8hi, __v8hi)
11732 __v8hi __builtin_arc_vmr7w (__v8hi, __v8hi)
11733 __v8hi __builtin_arc_vmrb (__v8hi, __v8hi)
11734 __v8hi __builtin_arc_vmulaw (__v8hi, __v8hi)
11735 __v8hi __builtin_arc_vmulfaw (__v8hi, __v8hi)
11736 __v8hi __builtin_arc_vmulfw (__v8hi, __v8hi)
11737 __v8hi __builtin_arc_vmulw (__v8hi, __v8hi)
11738 __v8hi __builtin_arc_vnew (__v8hi, __v8hi)
11739 __v8hi __builtin_arc_vor (__v8hi, __v8hi)
11740 __v8hi __builtin_arc_vsubaw (__v8hi, __v8hi)
11741 __v8hi __builtin_arc_vsubw (__v8hi, __v8hi)
11742 __v8hi __builtin_arc_vsummw (__v8hi, __v8hi)
11743 __v8hi __builtin_arc_vvc1f (__v8hi, __v8hi)
11744 __v8hi __builtin_arc_vvc1ft (__v8hi, __v8hi)
11745 __v8hi __builtin_arc_vxor (__v8hi, __v8hi)
11746 __v8hi __builtin_arc_vxoraw (__v8hi, __v8hi)
11749 The following take one @code{__v8hi} and one @code{int} argument and return a
11750 @code{__v8hi} result:
11753 __v8hi __builtin_arc_vbaddw (__v8hi, int)
11754 __v8hi __builtin_arc_vbmaxw (__v8hi, int)
11755 __v8hi __builtin_arc_vbminw (__v8hi, int)
11756 __v8hi __builtin_arc_vbmulaw (__v8hi, int)
11757 __v8hi __builtin_arc_vbmulfw (__v8hi, int)
11758 __v8hi __builtin_arc_vbmulw (__v8hi, int)
11759 __v8hi __builtin_arc_vbrsubw (__v8hi, int)
11760 __v8hi __builtin_arc_vbsubw (__v8hi, int)
11763 The following take one @code{__v8hi} argument and one @code{int} argument which
11764 must be a 3-bit compile time constant indicating a register number
11765 I0-I7. They return a @code{__v8hi} result.
11767 __v8hi __builtin_arc_vasrw (__v8hi, const int)
11768 __v8hi __builtin_arc_vsr8 (__v8hi, const int)
11769 __v8hi __builtin_arc_vsr8aw (__v8hi, const int)
11772 The following take one @code{__v8hi} argument and one @code{int}
11773 argument which must be a 6-bit compile time constant. They return a
11774 @code{__v8hi} result.
11776 __v8hi __builtin_arc_vasrpwbi (__v8hi, const int)
11777 __v8hi __builtin_arc_vasrrpwbi (__v8hi, const int)
11778 __v8hi __builtin_arc_vasrrwi (__v8hi, const int)
11779 __v8hi __builtin_arc_vasrsrwi (__v8hi, const int)
11780 __v8hi __builtin_arc_vasrwi (__v8hi, const int)
11781 __v8hi __builtin_arc_vsr8awi (__v8hi, const int)
11782 __v8hi __builtin_arc_vsr8i (__v8hi, const int)
11785 The following take one @code{__v8hi} argument and one @code{int} argument which
11786 must be a 8-bit compile time constant. They return a @code{__v8hi}
11789 __v8hi __builtin_arc_vd6tapf (__v8hi, const int)
11790 __v8hi __builtin_arc_vmvaw (__v8hi, const int)
11791 __v8hi __builtin_arc_vmvw (__v8hi, const int)
11792 __v8hi __builtin_arc_vmvzw (__v8hi, const int)
11795 The following take two @code{int} arguments, the second of which which
11796 must be a 8-bit compile time constant. They return a @code{__v8hi}
11799 __v8hi __builtin_arc_vmovaw (int, const int)
11800 __v8hi __builtin_arc_vmovw (int, const int)
11801 __v8hi __builtin_arc_vmovzw (int, const int)
11804 The following take a single @code{__v8hi} argument and return a
11805 @code{__v8hi} result:
11807 __v8hi __builtin_arc_vabsaw (__v8hi)
11808 __v8hi __builtin_arc_vabsw (__v8hi)
11809 __v8hi __builtin_arc_vaddsuw (__v8hi)
11810 __v8hi __builtin_arc_vexch1 (__v8hi)
11811 __v8hi __builtin_arc_vexch2 (__v8hi)
11812 __v8hi __builtin_arc_vexch4 (__v8hi)
11813 __v8hi __builtin_arc_vsignw (__v8hi)
11814 __v8hi __builtin_arc_vupbaw (__v8hi)
11815 __v8hi __builtin_arc_vupbw (__v8hi)
11816 __v8hi __builtin_arc_vupsbaw (__v8hi)
11817 __v8hi __builtin_arc_vupsbw (__v8hi)
11820 The following take two @code{int} arguments and return no result:
11822 void __builtin_arc_vdirun (int, int)
11823 void __builtin_arc_vdorun (int, int)
11826 The following take two @code{int} arguments and return no result. The
11827 first argument must a 3-bit compile time constant indicating one of
11828 the DR0-DR7 DMA setup channels:
11830 void __builtin_arc_vdiwr (const int, int)
11831 void __builtin_arc_vdowr (const int, int)
11834 The following take an @code{int} argument and return no result:
11836 void __builtin_arc_vendrec (int)
11837 void __builtin_arc_vrec (int)
11838 void __builtin_arc_vrecrun (int)
11839 void __builtin_arc_vrun (int)
11842 The following take a @code{__v8hi} argument and two @code{int}
11843 arguments and return a @code{__v8hi} result. The second argument must
11844 be a 3-bit compile time constants, indicating one the registers I0-I7,
11845 and the third argument must be an 8-bit compile time constant.
11847 @emph{Note:} Although the equivalent hardware instructions do not take
11848 an SIMD register as an operand, these builtins overwrite the relevant
11849 bits of the @code{__v8hi} register provided as the first argument with
11850 the value loaded from the @code{[Ib, u8]} location in the SDM.
11853 __v8hi __builtin_arc_vld32 (__v8hi, const int, const int)
11854 __v8hi __builtin_arc_vld32wh (__v8hi, const int, const int)
11855 __v8hi __builtin_arc_vld32wl (__v8hi, const int, const int)
11856 __v8hi __builtin_arc_vld64 (__v8hi, const int, const int)
11859 The following take two @code{int} arguments and return a @code{__v8hi}
11860 result. The first argument must be a 3-bit compile time constants,
11861 indicating one the registers I0-I7, and the second argument must be an
11862 8-bit compile time constant.
11865 __v8hi __builtin_arc_vld128 (const int, const int)
11866 __v8hi __builtin_arc_vld64w (const int, const int)
11869 The following take a @code{__v8hi} argument and two @code{int}
11870 arguments and return no result. The second argument must be a 3-bit
11871 compile time constants, indicating one the registers I0-I7, and the
11872 third argument must be an 8-bit compile time constant.
11875 void __builtin_arc_vst128 (__v8hi, const int, const int)
11876 void __builtin_arc_vst64 (__v8hi, const int, const int)
11879 The following take a @code{__v8hi} argument and three @code{int}
11880 arguments and return no result. The second argument must be a 3-bit
11881 compile-time constant, identifying the 16-bit sub-register to be
11882 stored, the third argument must be a 3-bit compile time constants,
11883 indicating one the registers I0-I7, and the fourth argument must be an
11884 8-bit compile time constant.
11887 void __builtin_arc_vst16_n (__v8hi, const int, const int, const int)
11888 void __builtin_arc_vst32_n (__v8hi, const int, const int, const int)
11891 @node ARM iWMMXt Built-in Functions
11892 @subsection ARM iWMMXt Built-in Functions
11894 These built-in functions are available for the ARM family of
11895 processors when the @option{-mcpu=iwmmxt} switch is used:
11898 typedef int v2si __attribute__ ((vector_size (8)));
11899 typedef short v4hi __attribute__ ((vector_size (8)));
11900 typedef char v8qi __attribute__ ((vector_size (8)));
11902 int __builtin_arm_getwcgr0 (void)
11903 void __builtin_arm_setwcgr0 (int)
11904 int __builtin_arm_getwcgr1 (void)
11905 void __builtin_arm_setwcgr1 (int)
11906 int __builtin_arm_getwcgr2 (void)
11907 void __builtin_arm_setwcgr2 (int)
11908 int __builtin_arm_getwcgr3 (void)
11909 void __builtin_arm_setwcgr3 (int)
11910 int __builtin_arm_textrmsb (v8qi, int)
11911 int __builtin_arm_textrmsh (v4hi, int)
11912 int __builtin_arm_textrmsw (v2si, int)
11913 int __builtin_arm_textrmub (v8qi, int)
11914 int __builtin_arm_textrmuh (v4hi, int)
11915 int __builtin_arm_textrmuw (v2si, int)
11916 v8qi __builtin_arm_tinsrb (v8qi, int, int)
11917 v4hi __builtin_arm_tinsrh (v4hi, int, int)
11918 v2si __builtin_arm_tinsrw (v2si, int, int)
11919 long long __builtin_arm_tmia (long long, int, int)
11920 long long __builtin_arm_tmiabb (long long, int, int)
11921 long long __builtin_arm_tmiabt (long long, int, int)
11922 long long __builtin_arm_tmiaph (long long, int, int)
11923 long long __builtin_arm_tmiatb (long long, int, int)
11924 long long __builtin_arm_tmiatt (long long, int, int)
11925 int __builtin_arm_tmovmskb (v8qi)
11926 int __builtin_arm_tmovmskh (v4hi)
11927 int __builtin_arm_tmovmskw (v2si)
11928 long long __builtin_arm_waccb (v8qi)
11929 long long __builtin_arm_wacch (v4hi)
11930 long long __builtin_arm_waccw (v2si)
11931 v8qi __builtin_arm_waddb (v8qi, v8qi)
11932 v8qi __builtin_arm_waddbss (v8qi, v8qi)
11933 v8qi __builtin_arm_waddbus (v8qi, v8qi)
11934 v4hi __builtin_arm_waddh (v4hi, v4hi)
11935 v4hi __builtin_arm_waddhss (v4hi, v4hi)
11936 v4hi __builtin_arm_waddhus (v4hi, v4hi)
11937 v2si __builtin_arm_waddw (v2si, v2si)
11938 v2si __builtin_arm_waddwss (v2si, v2si)
11939 v2si __builtin_arm_waddwus (v2si, v2si)
11940 v8qi __builtin_arm_walign (v8qi, v8qi, int)
11941 long long __builtin_arm_wand(long long, long long)
11942 long long __builtin_arm_wandn (long long, long long)
11943 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
11944 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
11945 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
11946 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
11947 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
11948 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
11949 v2si __builtin_arm_wcmpeqw (v2si, v2si)
11950 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
11951 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
11952 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
11953 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
11954 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
11955 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
11956 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
11957 long long __builtin_arm_wmacsz (v4hi, v4hi)
11958 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
11959 long long __builtin_arm_wmacuz (v4hi, v4hi)
11960 v4hi __builtin_arm_wmadds (v4hi, v4hi)
11961 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
11962 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
11963 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
11964 v2si __builtin_arm_wmaxsw (v2si, v2si)
11965 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
11966 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
11967 v2si __builtin_arm_wmaxuw (v2si, v2si)
11968 v8qi __builtin_arm_wminsb (v8qi, v8qi)
11969 v4hi __builtin_arm_wminsh (v4hi, v4hi)
11970 v2si __builtin_arm_wminsw (v2si, v2si)
11971 v8qi __builtin_arm_wminub (v8qi, v8qi)
11972 v4hi __builtin_arm_wminuh (v4hi, v4hi)
11973 v2si __builtin_arm_wminuw (v2si, v2si)
11974 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
11975 v4hi __builtin_arm_wmulul (v4hi, v4hi)
11976 v4hi __builtin_arm_wmulum (v4hi, v4hi)
11977 long long __builtin_arm_wor (long long, long long)
11978 v2si __builtin_arm_wpackdss (long long, long long)
11979 v2si __builtin_arm_wpackdus (long long, long long)
11980 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
11981 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
11982 v4hi __builtin_arm_wpackwss (v2si, v2si)
11983 v4hi __builtin_arm_wpackwus (v2si, v2si)
11984 long long __builtin_arm_wrord (long long, long long)
11985 long long __builtin_arm_wrordi (long long, int)
11986 v4hi __builtin_arm_wrorh (v4hi, long long)
11987 v4hi __builtin_arm_wrorhi (v4hi, int)
11988 v2si __builtin_arm_wrorw (v2si, long long)
11989 v2si __builtin_arm_wrorwi (v2si, int)
11990 v2si __builtin_arm_wsadb (v2si, v8qi, v8qi)
11991 v2si __builtin_arm_wsadbz (v8qi, v8qi)
11992 v2si __builtin_arm_wsadh (v2si, v4hi, v4hi)
11993 v2si __builtin_arm_wsadhz (v4hi, v4hi)
11994 v4hi __builtin_arm_wshufh (v4hi, int)
11995 long long __builtin_arm_wslld (long long, long long)
11996 long long __builtin_arm_wslldi (long long, int)
11997 v4hi __builtin_arm_wsllh (v4hi, long long)
11998 v4hi __builtin_arm_wsllhi (v4hi, int)
11999 v2si __builtin_arm_wsllw (v2si, long long)
12000 v2si __builtin_arm_wsllwi (v2si, int)
12001 long long __builtin_arm_wsrad (long long, long long)
12002 long long __builtin_arm_wsradi (long long, int)
12003 v4hi __builtin_arm_wsrah (v4hi, long long)
12004 v4hi __builtin_arm_wsrahi (v4hi, int)
12005 v2si __builtin_arm_wsraw (v2si, long long)
12006 v2si __builtin_arm_wsrawi (v2si, int)
12007 long long __builtin_arm_wsrld (long long, long long)
12008 long long __builtin_arm_wsrldi (long long, int)
12009 v4hi __builtin_arm_wsrlh (v4hi, long long)
12010 v4hi __builtin_arm_wsrlhi (v4hi, int)
12011 v2si __builtin_arm_wsrlw (v2si, long long)
12012 v2si __builtin_arm_wsrlwi (v2si, int)
12013 v8qi __builtin_arm_wsubb (v8qi, v8qi)
12014 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
12015 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
12016 v4hi __builtin_arm_wsubh (v4hi, v4hi)
12017 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
12018 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
12019 v2si __builtin_arm_wsubw (v2si, v2si)
12020 v2si __builtin_arm_wsubwss (v2si, v2si)
12021 v2si __builtin_arm_wsubwus (v2si, v2si)
12022 v4hi __builtin_arm_wunpckehsb (v8qi)
12023 v2si __builtin_arm_wunpckehsh (v4hi)
12024 long long __builtin_arm_wunpckehsw (v2si)
12025 v4hi __builtin_arm_wunpckehub (v8qi)
12026 v2si __builtin_arm_wunpckehuh (v4hi)
12027 long long __builtin_arm_wunpckehuw (v2si)
12028 v4hi __builtin_arm_wunpckelsb (v8qi)
12029 v2si __builtin_arm_wunpckelsh (v4hi)
12030 long long __builtin_arm_wunpckelsw (v2si)
12031 v4hi __builtin_arm_wunpckelub (v8qi)
12032 v2si __builtin_arm_wunpckeluh (v4hi)
12033 long long __builtin_arm_wunpckeluw (v2si)
12034 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
12035 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
12036 v2si __builtin_arm_wunpckihw (v2si, v2si)
12037 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
12038 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
12039 v2si __builtin_arm_wunpckilw (v2si, v2si)
12040 long long __builtin_arm_wxor (long long, long long)
12041 long long __builtin_arm_wzero ()
12045 @node ARM C Language Extensions (ACLE)
12046 @subsection ARM C Language Extensions (ACLE)
12048 GCC implements extensions for C as described in the ARM C Language
12049 Extensions (ACLE) specification, which can be found at
12050 @uref{http://infocenter.arm.com/help/topic/com.arm.doc.ihi0053c/IHI0053C_acle_2_0.pdf}.
12052 As a part of ACLE, GCC implements extensions for Advanced SIMD as described in
12053 the ARM C Language Extensions Specification. The complete list of Advanced SIMD
12054 intrinsics can be found at
12055 @uref{http://infocenter.arm.com/help/topic/com.arm.doc.ihi0073a/IHI0073A_arm_neon_intrinsics_ref.pdf}.
12056 The built-in intrinsics for the Advanced SIMD extension are available when
12059 Currently, ARM and AArch64 back ends do not support ACLE 2.0 fully. Both
12060 back ends support CRC32 intrinsics from @file{arm_acle.h}. The ARM back end's
12061 16-bit floating-point Advanced SIMD intrinsics currently comply to ACLE v1.1.
12062 AArch64's back end does not have support for 16-bit floating point Advanced SIMD
12065 See @ref{ARM Options} and @ref{AArch64 Options} for more information on the
12066 availability of extensions.
12068 @node ARM Floating Point Status and Control Intrinsics
12069 @subsection ARM Floating Point Status and Control Intrinsics
12071 These built-in functions are available for the ARM family of
12072 processors with floating-point unit.
12075 unsigned int __builtin_arm_get_fpscr ()
12076 void __builtin_arm_set_fpscr (unsigned int)
12079 @node AVR Built-in Functions
12080 @subsection AVR Built-in Functions
12082 For each built-in function for AVR, there is an equally named,
12083 uppercase built-in macro defined. That way users can easily query if
12084 or if not a specific built-in is implemented or not. For example, if
12085 @code{__builtin_avr_nop} is available the macro
12086 @code{__BUILTIN_AVR_NOP} is defined to @code{1} and undefined otherwise.
12088 The following built-in functions map to the respective machine
12089 instruction, i.e.@: @code{nop}, @code{sei}, @code{cli}, @code{sleep},
12090 @code{wdr}, @code{swap}, @code{fmul}, @code{fmuls}
12091 resp. @code{fmulsu}. The three @code{fmul*} built-ins are implemented
12092 as library call if no hardware multiplier is available.
12095 void __builtin_avr_nop (void)
12096 void __builtin_avr_sei (void)
12097 void __builtin_avr_cli (void)
12098 void __builtin_avr_sleep (void)
12099 void __builtin_avr_wdr (void)
12100 unsigned char __builtin_avr_swap (unsigned char)
12101 unsigned int __builtin_avr_fmul (unsigned char, unsigned char)
12102 int __builtin_avr_fmuls (char, char)
12103 int __builtin_avr_fmulsu (char, unsigned char)
12106 In order to delay execution for a specific number of cycles, GCC
12109 void __builtin_avr_delay_cycles (unsigned long ticks)
12113 @code{ticks} is the number of ticks to delay execution. Note that this
12114 built-in does not take into account the effect of interrupts that
12115 might increase delay time. @code{ticks} must be a compile-time
12116 integer constant; delays with a variable number of cycles are not supported.
12119 char __builtin_avr_flash_segment (const __memx void*)
12123 This built-in takes a byte address to the 24-bit
12124 @ref{AVR Named Address Spaces,address space} @code{__memx} and returns
12125 the number of the flash segment (the 64 KiB chunk) where the address
12126 points to. Counting starts at @code{0}.
12127 If the address does not point to flash memory, return @code{-1}.
12130 unsigned char __builtin_avr_insert_bits (unsigned long map, unsigned char bits, unsigned char val)
12134 Insert bits from @var{bits} into @var{val} and return the resulting
12135 value. The nibbles of @var{map} determine how the insertion is
12136 performed: Let @var{X} be the @var{n}-th nibble of @var{map}
12138 @item If @var{X} is @code{0xf},
12139 then the @var{n}-th bit of @var{val} is returned unaltered.
12141 @item If X is in the range 0@dots{}7,
12142 then the @var{n}-th result bit is set to the @var{X}-th bit of @var{bits}
12144 @item If X is in the range 8@dots{}@code{0xe},
12145 then the @var{n}-th result bit is undefined.
12149 One typical use case for this built-in is adjusting input and
12150 output values to non-contiguous port layouts. Some examples:
12153 // same as val, bits is unused
12154 __builtin_avr_insert_bits (0xffffffff, bits, val)
12158 // same as bits, val is unused
12159 __builtin_avr_insert_bits (0x76543210, bits, val)
12163 // same as rotating bits by 4
12164 __builtin_avr_insert_bits (0x32107654, bits, 0)
12168 // high nibble of result is the high nibble of val
12169 // low nibble of result is the low nibble of bits
12170 __builtin_avr_insert_bits (0xffff3210, bits, val)
12174 // reverse the bit order of bits
12175 __builtin_avr_insert_bits (0x01234567, bits, 0)
12178 @node Blackfin Built-in Functions
12179 @subsection Blackfin Built-in Functions
12181 Currently, there are two Blackfin-specific built-in functions. These are
12182 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
12183 using inline assembly; by using these built-in functions the compiler can
12184 automatically add workarounds for hardware errata involving these
12185 instructions. These functions are named as follows:
12188 void __builtin_bfin_csync (void)
12189 void __builtin_bfin_ssync (void)
12192 @node FR-V Built-in Functions
12193 @subsection FR-V Built-in Functions
12195 GCC provides many FR-V-specific built-in functions. In general,
12196 these functions are intended to be compatible with those described
12197 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
12198 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
12199 @code{__MBTOHE}, the GCC forms of which pass 128-bit values by
12200 pointer rather than by value.
12202 Most of the functions are named after specific FR-V instructions.
12203 Such functions are said to be ``directly mapped'' and are summarized
12204 here in tabular form.
12208 * Directly-mapped Integer Functions::
12209 * Directly-mapped Media Functions::
12210 * Raw read/write Functions::
12211 * Other Built-in Functions::
12214 @node Argument Types
12215 @subsubsection Argument Types
12217 The arguments to the built-in functions can be divided into three groups:
12218 register numbers, compile-time constants and run-time values. In order
12219 to make this classification clear at a glance, the arguments and return
12220 values are given the following pseudo types:
12222 @multitable @columnfractions .20 .30 .15 .35
12223 @item Pseudo type @tab Real C type @tab Constant? @tab Description
12224 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
12225 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
12226 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
12227 @item @code{uw2} @tab @code{unsigned long long} @tab No
12228 @tab an unsigned doubleword
12229 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
12230 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
12231 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
12232 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
12235 These pseudo types are not defined by GCC, they are simply a notational
12236 convenience used in this manual.
12238 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
12239 and @code{sw2} are evaluated at run time. They correspond to
12240 register operands in the underlying FR-V instructions.
12242 @code{const} arguments represent immediate operands in the underlying
12243 FR-V instructions. They must be compile-time constants.
12245 @code{acc} arguments are evaluated at compile time and specify the number
12246 of an accumulator register. For example, an @code{acc} argument of 2
12247 selects the ACC2 register.
12249 @code{iacc} arguments are similar to @code{acc} arguments but specify the
12250 number of an IACC register. See @pxref{Other Built-in Functions}
12253 @node Directly-mapped Integer Functions
12254 @subsubsection Directly-Mapped Integer Functions
12256 The functions listed below map directly to FR-V I-type instructions.
12258 @multitable @columnfractions .45 .32 .23
12259 @item Function prototype @tab Example usage @tab Assembly output
12260 @item @code{sw1 __ADDSS (sw1, sw1)}
12261 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
12262 @tab @code{ADDSS @var{a},@var{b},@var{c}}
12263 @item @code{sw1 __SCAN (sw1, sw1)}
12264 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
12265 @tab @code{SCAN @var{a},@var{b},@var{c}}
12266 @item @code{sw1 __SCUTSS (sw1)}
12267 @tab @code{@var{b} = __SCUTSS (@var{a})}
12268 @tab @code{SCUTSS @var{a},@var{b}}
12269 @item @code{sw1 __SLASS (sw1, sw1)}
12270 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
12271 @tab @code{SLASS @var{a},@var{b},@var{c}}
12272 @item @code{void __SMASS (sw1, sw1)}
12273 @tab @code{__SMASS (@var{a}, @var{b})}
12274 @tab @code{SMASS @var{a},@var{b}}
12275 @item @code{void __SMSSS (sw1, sw1)}
12276 @tab @code{__SMSSS (@var{a}, @var{b})}
12277 @tab @code{SMSSS @var{a},@var{b}}
12278 @item @code{void __SMU (sw1, sw1)}
12279 @tab @code{__SMU (@var{a}, @var{b})}
12280 @tab @code{SMU @var{a},@var{b}}
12281 @item @code{sw2 __SMUL (sw1, sw1)}
12282 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
12283 @tab @code{SMUL @var{a},@var{b},@var{c}}
12284 @item @code{sw1 __SUBSS (sw1, sw1)}
12285 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
12286 @tab @code{SUBSS @var{a},@var{b},@var{c}}
12287 @item @code{uw2 __UMUL (uw1, uw1)}
12288 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
12289 @tab @code{UMUL @var{a},@var{b},@var{c}}
12292 @node Directly-mapped Media Functions
12293 @subsubsection Directly-Mapped Media Functions
12295 The functions listed below map directly to FR-V M-type instructions.
12297 @multitable @columnfractions .45 .32 .23
12298 @item Function prototype @tab Example usage @tab Assembly output
12299 @item @code{uw1 __MABSHS (sw1)}
12300 @tab @code{@var{b} = __MABSHS (@var{a})}
12301 @tab @code{MABSHS @var{a},@var{b}}
12302 @item @code{void __MADDACCS (acc, acc)}
12303 @tab @code{__MADDACCS (@var{b}, @var{a})}
12304 @tab @code{MADDACCS @var{a},@var{b}}
12305 @item @code{sw1 __MADDHSS (sw1, sw1)}
12306 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
12307 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
12308 @item @code{uw1 __MADDHUS (uw1, uw1)}
12309 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
12310 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
12311 @item @code{uw1 __MAND (uw1, uw1)}
12312 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
12313 @tab @code{MAND @var{a},@var{b},@var{c}}
12314 @item @code{void __MASACCS (acc, acc)}
12315 @tab @code{__MASACCS (@var{b}, @var{a})}
12316 @tab @code{MASACCS @var{a},@var{b}}
12317 @item @code{uw1 __MAVEH (uw1, uw1)}
12318 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
12319 @tab @code{MAVEH @var{a},@var{b},@var{c}}
12320 @item @code{uw2 __MBTOH (uw1)}
12321 @tab @code{@var{b} = __MBTOH (@var{a})}
12322 @tab @code{MBTOH @var{a},@var{b}}
12323 @item @code{void __MBTOHE (uw1 *, uw1)}
12324 @tab @code{__MBTOHE (&@var{b}, @var{a})}
12325 @tab @code{MBTOHE @var{a},@var{b}}
12326 @item @code{void __MCLRACC (acc)}
12327 @tab @code{__MCLRACC (@var{a})}
12328 @tab @code{MCLRACC @var{a}}
12329 @item @code{void __MCLRACCA (void)}
12330 @tab @code{__MCLRACCA ()}
12331 @tab @code{MCLRACCA}
12332 @item @code{uw1 __Mcop1 (uw1, uw1)}
12333 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
12334 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
12335 @item @code{uw1 __Mcop2 (uw1, uw1)}
12336 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
12337 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
12338 @item @code{uw1 __MCPLHI (uw2, const)}
12339 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
12340 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
12341 @item @code{uw1 __MCPLI (uw2, const)}
12342 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
12343 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
12344 @item @code{void __MCPXIS (acc, sw1, sw1)}
12345 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
12346 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
12347 @item @code{void __MCPXIU (acc, uw1, uw1)}
12348 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
12349 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
12350 @item @code{void __MCPXRS (acc, sw1, sw1)}
12351 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
12352 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
12353 @item @code{void __MCPXRU (acc, uw1, uw1)}
12354 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
12355 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
12356 @item @code{uw1 __MCUT (acc, uw1)}
12357 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
12358 @tab @code{MCUT @var{a},@var{b},@var{c}}
12359 @item @code{uw1 __MCUTSS (acc, sw1)}
12360 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
12361 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
12362 @item @code{void __MDADDACCS (acc, acc)}
12363 @tab @code{__MDADDACCS (@var{b}, @var{a})}
12364 @tab @code{MDADDACCS @var{a},@var{b}}
12365 @item @code{void __MDASACCS (acc, acc)}
12366 @tab @code{__MDASACCS (@var{b}, @var{a})}
12367 @tab @code{MDASACCS @var{a},@var{b}}
12368 @item @code{uw2 __MDCUTSSI (acc, const)}
12369 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
12370 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
12371 @item @code{uw2 __MDPACKH (uw2, uw2)}
12372 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
12373 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
12374 @item @code{uw2 __MDROTLI (uw2, const)}
12375 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
12376 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
12377 @item @code{void __MDSUBACCS (acc, acc)}
12378 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
12379 @tab @code{MDSUBACCS @var{a},@var{b}}
12380 @item @code{void __MDUNPACKH (uw1 *, uw2)}
12381 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
12382 @tab @code{MDUNPACKH @var{a},@var{b}}
12383 @item @code{uw2 __MEXPDHD (uw1, const)}
12384 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
12385 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
12386 @item @code{uw1 __MEXPDHW (uw1, const)}
12387 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
12388 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
12389 @item @code{uw1 __MHDSETH (uw1, const)}
12390 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
12391 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
12392 @item @code{sw1 __MHDSETS (const)}
12393 @tab @code{@var{b} = __MHDSETS (@var{a})}
12394 @tab @code{MHDSETS #@var{a},@var{b}}
12395 @item @code{uw1 __MHSETHIH (uw1, const)}
12396 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
12397 @tab @code{MHSETHIH #@var{a},@var{b}}
12398 @item @code{sw1 __MHSETHIS (sw1, const)}
12399 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
12400 @tab @code{MHSETHIS #@var{a},@var{b}}
12401 @item @code{uw1 __MHSETLOH (uw1, const)}
12402 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
12403 @tab @code{MHSETLOH #@var{a},@var{b}}
12404 @item @code{sw1 __MHSETLOS (sw1, const)}
12405 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
12406 @tab @code{MHSETLOS #@var{a},@var{b}}
12407 @item @code{uw1 __MHTOB (uw2)}
12408 @tab @code{@var{b} = __MHTOB (@var{a})}
12409 @tab @code{MHTOB @var{a},@var{b}}
12410 @item @code{void __MMACHS (acc, sw1, sw1)}
12411 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
12412 @tab @code{MMACHS @var{a},@var{b},@var{c}}
12413 @item @code{void __MMACHU (acc, uw1, uw1)}
12414 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
12415 @tab @code{MMACHU @var{a},@var{b},@var{c}}
12416 @item @code{void __MMRDHS (acc, sw1, sw1)}
12417 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
12418 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
12419 @item @code{void __MMRDHU (acc, uw1, uw1)}
12420 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
12421 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
12422 @item @code{void __MMULHS (acc, sw1, sw1)}
12423 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
12424 @tab @code{MMULHS @var{a},@var{b},@var{c}}
12425 @item @code{void __MMULHU (acc, uw1, uw1)}
12426 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
12427 @tab @code{MMULHU @var{a},@var{b},@var{c}}
12428 @item @code{void __MMULXHS (acc, sw1, sw1)}
12429 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
12430 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
12431 @item @code{void __MMULXHU (acc, uw1, uw1)}
12432 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
12433 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
12434 @item @code{uw1 __MNOT (uw1)}
12435 @tab @code{@var{b} = __MNOT (@var{a})}
12436 @tab @code{MNOT @var{a},@var{b}}
12437 @item @code{uw1 __MOR (uw1, uw1)}
12438 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
12439 @tab @code{MOR @var{a},@var{b},@var{c}}
12440 @item @code{uw1 __MPACKH (uh, uh)}
12441 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
12442 @tab @code{MPACKH @var{a},@var{b},@var{c}}
12443 @item @code{sw2 __MQADDHSS (sw2, sw2)}
12444 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
12445 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
12446 @item @code{uw2 __MQADDHUS (uw2, uw2)}
12447 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
12448 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
12449 @item @code{void __MQCPXIS (acc, sw2, sw2)}
12450 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
12451 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
12452 @item @code{void __MQCPXIU (acc, uw2, uw2)}
12453 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
12454 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
12455 @item @code{void __MQCPXRS (acc, sw2, sw2)}
12456 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
12457 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
12458 @item @code{void __MQCPXRU (acc, uw2, uw2)}
12459 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
12460 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
12461 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
12462 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
12463 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
12464 @item @code{sw2 __MQLMTHS (sw2, sw2)}
12465 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
12466 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
12467 @item @code{void __MQMACHS (acc, sw2, sw2)}
12468 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
12469 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
12470 @item @code{void __MQMACHU (acc, uw2, uw2)}
12471 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
12472 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
12473 @item @code{void __MQMACXHS (acc, sw2, sw2)}
12474 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
12475 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
12476 @item @code{void __MQMULHS (acc, sw2, sw2)}
12477 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
12478 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
12479 @item @code{void __MQMULHU (acc, uw2, uw2)}
12480 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
12481 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
12482 @item @code{void __MQMULXHS (acc, sw2, sw2)}
12483 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
12484 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
12485 @item @code{void __MQMULXHU (acc, uw2, uw2)}
12486 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
12487 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
12488 @item @code{sw2 __MQSATHS (sw2, sw2)}
12489 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
12490 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
12491 @item @code{uw2 __MQSLLHI (uw2, int)}
12492 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
12493 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
12494 @item @code{sw2 __MQSRAHI (sw2, int)}
12495 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
12496 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
12497 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
12498 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
12499 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
12500 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
12501 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
12502 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
12503 @item @code{void __MQXMACHS (acc, sw2, sw2)}
12504 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
12505 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
12506 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
12507 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
12508 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
12509 @item @code{uw1 __MRDACC (acc)}
12510 @tab @code{@var{b} = __MRDACC (@var{a})}
12511 @tab @code{MRDACC @var{a},@var{b}}
12512 @item @code{uw1 __MRDACCG (acc)}
12513 @tab @code{@var{b} = __MRDACCG (@var{a})}
12514 @tab @code{MRDACCG @var{a},@var{b}}
12515 @item @code{uw1 __MROTLI (uw1, const)}
12516 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
12517 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
12518 @item @code{uw1 __MROTRI (uw1, const)}
12519 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
12520 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
12521 @item @code{sw1 __MSATHS (sw1, sw1)}
12522 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
12523 @tab @code{MSATHS @var{a},@var{b},@var{c}}
12524 @item @code{uw1 __MSATHU (uw1, uw1)}
12525 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
12526 @tab @code{MSATHU @var{a},@var{b},@var{c}}
12527 @item @code{uw1 __MSLLHI (uw1, const)}
12528 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
12529 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
12530 @item @code{sw1 __MSRAHI (sw1, const)}
12531 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
12532 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
12533 @item @code{uw1 __MSRLHI (uw1, const)}
12534 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
12535 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
12536 @item @code{void __MSUBACCS (acc, acc)}
12537 @tab @code{__MSUBACCS (@var{b}, @var{a})}
12538 @tab @code{MSUBACCS @var{a},@var{b}}
12539 @item @code{sw1 __MSUBHSS (sw1, sw1)}
12540 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
12541 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
12542 @item @code{uw1 __MSUBHUS (uw1, uw1)}
12543 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
12544 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
12545 @item @code{void __MTRAP (void)}
12546 @tab @code{__MTRAP ()}
12548 @item @code{uw2 __MUNPACKH (uw1)}
12549 @tab @code{@var{b} = __MUNPACKH (@var{a})}
12550 @tab @code{MUNPACKH @var{a},@var{b}}
12551 @item @code{uw1 __MWCUT (uw2, uw1)}
12552 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
12553 @tab @code{MWCUT @var{a},@var{b},@var{c}}
12554 @item @code{void __MWTACC (acc, uw1)}
12555 @tab @code{__MWTACC (@var{b}, @var{a})}
12556 @tab @code{MWTACC @var{a},@var{b}}
12557 @item @code{void __MWTACCG (acc, uw1)}
12558 @tab @code{__MWTACCG (@var{b}, @var{a})}
12559 @tab @code{MWTACCG @var{a},@var{b}}
12560 @item @code{uw1 __MXOR (uw1, uw1)}
12561 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
12562 @tab @code{MXOR @var{a},@var{b},@var{c}}
12565 @node Raw read/write Functions
12566 @subsubsection Raw Read/Write Functions
12568 This sections describes built-in functions related to read and write
12569 instructions to access memory. These functions generate
12570 @code{membar} instructions to flush the I/O load and stores where
12571 appropriate, as described in Fujitsu's manual described above.
12575 @item unsigned char __builtin_read8 (void *@var{data})
12576 @item unsigned short __builtin_read16 (void *@var{data})
12577 @item unsigned long __builtin_read32 (void *@var{data})
12578 @item unsigned long long __builtin_read64 (void *@var{data})
12580 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
12581 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
12582 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
12583 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
12586 @node Other Built-in Functions
12587 @subsubsection Other Built-in Functions
12589 This section describes built-in functions that are not named after
12590 a specific FR-V instruction.
12593 @item sw2 __IACCreadll (iacc @var{reg})
12594 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
12595 for future expansion and must be 0.
12597 @item sw1 __IACCreadl (iacc @var{reg})
12598 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
12599 Other values of @var{reg} are rejected as invalid.
12601 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
12602 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
12603 is reserved for future expansion and must be 0.
12605 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
12606 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
12607 is 1. Other values of @var{reg} are rejected as invalid.
12609 @item void __data_prefetch0 (const void *@var{x})
12610 Use the @code{dcpl} instruction to load the contents of address @var{x}
12611 into the data cache.
12613 @item void __data_prefetch (const void *@var{x})
12614 Use the @code{nldub} instruction to load the contents of address @var{x}
12615 into the data cache. The instruction is issued in slot I1@.
12618 @node MIPS DSP Built-in Functions
12619 @subsection MIPS DSP Built-in Functions
12621 The MIPS DSP Application-Specific Extension (ASE) includes new
12622 instructions that are designed to improve the performance of DSP and
12623 media applications. It provides instructions that operate on packed
12624 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
12626 GCC supports MIPS DSP operations using both the generic
12627 vector extensions (@pxref{Vector Extensions}) and a collection of
12628 MIPS-specific built-in functions. Both kinds of support are
12629 enabled by the @option{-mdsp} command-line option.
12631 Revision 2 of the ASE was introduced in the second half of 2006.
12632 This revision adds extra instructions to the original ASE, but is
12633 otherwise backwards-compatible with it. You can select revision 2
12634 using the command-line option @option{-mdspr2}; this option implies
12637 The SCOUNT and POS bits of the DSP control register are global. The
12638 WRDSP, EXTPDP, EXTPDPV and MTHLIP instructions modify the SCOUNT and
12639 POS bits. During optimization, the compiler does not delete these
12640 instructions and it does not delete calls to functions containing
12641 these instructions.
12643 At present, GCC only provides support for operations on 32-bit
12644 vectors. The vector type associated with 8-bit integer data is
12645 usually called @code{v4i8}, the vector type associated with Q7
12646 is usually called @code{v4q7}, the vector type associated with 16-bit
12647 integer data is usually called @code{v2i16}, and the vector type
12648 associated with Q15 is usually called @code{v2q15}. They can be
12649 defined in C as follows:
12652 typedef signed char v4i8 __attribute__ ((vector_size(4)));
12653 typedef signed char v4q7 __attribute__ ((vector_size(4)));
12654 typedef short v2i16 __attribute__ ((vector_size(4)));
12655 typedef short v2q15 __attribute__ ((vector_size(4)));
12658 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
12659 initialized in the same way as aggregates. For example:
12662 v4i8 a = @{1, 2, 3, 4@};
12664 b = (v4i8) @{5, 6, 7, 8@};
12666 v2q15 c = @{0x0fcb, 0x3a75@};
12668 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
12671 @emph{Note:} The CPU's endianness determines the order in which values
12672 are packed. On little-endian targets, the first value is the least
12673 significant and the last value is the most significant. The opposite
12674 order applies to big-endian targets. For example, the code above
12675 sets the lowest byte of @code{a} to @code{1} on little-endian targets
12676 and @code{4} on big-endian targets.
12678 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
12679 representation. As shown in this example, the integer representation
12680 of a Q7 value can be obtained by multiplying the fractional value by
12681 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
12682 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
12685 The table below lists the @code{v4i8} and @code{v2q15} operations for which
12686 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
12687 and @code{c} and @code{d} are @code{v2q15} values.
12689 @multitable @columnfractions .50 .50
12690 @item C code @tab MIPS instruction
12691 @item @code{a + b} @tab @code{addu.qb}
12692 @item @code{c + d} @tab @code{addq.ph}
12693 @item @code{a - b} @tab @code{subu.qb}
12694 @item @code{c - d} @tab @code{subq.ph}
12697 The table below lists the @code{v2i16} operation for which
12698 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
12699 @code{v2i16} values.
12701 @multitable @columnfractions .50 .50
12702 @item C code @tab MIPS instruction
12703 @item @code{e * f} @tab @code{mul.ph}
12706 It is easier to describe the DSP built-in functions if we first define
12707 the following types:
12712 typedef unsigned int ui32;
12713 typedef long long a64;
12716 @code{q31} and @code{i32} are actually the same as @code{int}, but we
12717 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
12718 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
12719 @code{long long}, but we use @code{a64} to indicate values that are
12720 placed in one of the four DSP accumulators (@code{$ac0},
12721 @code{$ac1}, @code{$ac2} or @code{$ac3}).
12723 Also, some built-in functions prefer or require immediate numbers as
12724 parameters, because the corresponding DSP instructions accept both immediate
12725 numbers and register operands, or accept immediate numbers only. The
12726 immediate parameters are listed as follows.
12734 imm0_255: 0 to 255.
12735 imm_n32_31: -32 to 31.
12736 imm_n512_511: -512 to 511.
12739 The following built-in functions map directly to a particular MIPS DSP
12740 instruction. Please refer to the architecture specification
12741 for details on what each instruction does.
12744 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
12745 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
12746 q31 __builtin_mips_addq_s_w (q31, q31)
12747 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
12748 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
12749 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
12750 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
12751 q31 __builtin_mips_subq_s_w (q31, q31)
12752 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
12753 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
12754 i32 __builtin_mips_addsc (i32, i32)
12755 i32 __builtin_mips_addwc (i32, i32)
12756 i32 __builtin_mips_modsub (i32, i32)
12757 i32 __builtin_mips_raddu_w_qb (v4i8)
12758 v2q15 __builtin_mips_absq_s_ph (v2q15)
12759 q31 __builtin_mips_absq_s_w (q31)
12760 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
12761 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
12762 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
12763 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
12764 q31 __builtin_mips_preceq_w_phl (v2q15)
12765 q31 __builtin_mips_preceq_w_phr (v2q15)
12766 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
12767 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
12768 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
12769 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
12770 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
12771 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
12772 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
12773 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
12774 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
12775 v4i8 __builtin_mips_shll_qb (v4i8, i32)
12776 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
12777 v2q15 __builtin_mips_shll_ph (v2q15, i32)
12778 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
12779 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
12780 q31 __builtin_mips_shll_s_w (q31, imm0_31)
12781 q31 __builtin_mips_shll_s_w (q31, i32)
12782 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
12783 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
12784 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
12785 v2q15 __builtin_mips_shra_ph (v2q15, i32)
12786 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
12787 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
12788 q31 __builtin_mips_shra_r_w (q31, imm0_31)
12789 q31 __builtin_mips_shra_r_w (q31, i32)
12790 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
12791 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
12792 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
12793 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
12794 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
12795 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
12796 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
12797 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
12798 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
12799 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
12800 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
12801 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
12802 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
12803 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
12804 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
12805 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
12806 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
12807 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
12808 i32 __builtin_mips_bitrev (i32)
12809 i32 __builtin_mips_insv (i32, i32)
12810 v4i8 __builtin_mips_repl_qb (imm0_255)
12811 v4i8 __builtin_mips_repl_qb (i32)
12812 v2q15 __builtin_mips_repl_ph (imm_n512_511)
12813 v2q15 __builtin_mips_repl_ph (i32)
12814 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
12815 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
12816 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
12817 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
12818 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
12819 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
12820 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
12821 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
12822 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
12823 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
12824 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
12825 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
12826 i32 __builtin_mips_extr_w (a64, imm0_31)
12827 i32 __builtin_mips_extr_w (a64, i32)
12828 i32 __builtin_mips_extr_r_w (a64, imm0_31)
12829 i32 __builtin_mips_extr_s_h (a64, i32)
12830 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
12831 i32 __builtin_mips_extr_rs_w (a64, i32)
12832 i32 __builtin_mips_extr_s_h (a64, imm0_31)
12833 i32 __builtin_mips_extr_r_w (a64, i32)
12834 i32 __builtin_mips_extp (a64, imm0_31)
12835 i32 __builtin_mips_extp (a64, i32)
12836 i32 __builtin_mips_extpdp (a64, imm0_31)
12837 i32 __builtin_mips_extpdp (a64, i32)
12838 a64 __builtin_mips_shilo (a64, imm_n32_31)
12839 a64 __builtin_mips_shilo (a64, i32)
12840 a64 __builtin_mips_mthlip (a64, i32)
12841 void __builtin_mips_wrdsp (i32, imm0_63)
12842 i32 __builtin_mips_rddsp (imm0_63)
12843 i32 __builtin_mips_lbux (void *, i32)
12844 i32 __builtin_mips_lhx (void *, i32)
12845 i32 __builtin_mips_lwx (void *, i32)
12846 a64 __builtin_mips_ldx (void *, i32) [MIPS64 only]
12847 i32 __builtin_mips_bposge32 (void)
12848 a64 __builtin_mips_madd (a64, i32, i32);
12849 a64 __builtin_mips_maddu (a64, ui32, ui32);
12850 a64 __builtin_mips_msub (a64, i32, i32);
12851 a64 __builtin_mips_msubu (a64, ui32, ui32);
12852 a64 __builtin_mips_mult (i32, i32);
12853 a64 __builtin_mips_multu (ui32, ui32);
12856 The following built-in functions map directly to a particular MIPS DSP REV 2
12857 instruction. Please refer to the architecture specification
12858 for details on what each instruction does.
12861 v4q7 __builtin_mips_absq_s_qb (v4q7);
12862 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
12863 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
12864 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
12865 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
12866 i32 __builtin_mips_append (i32, i32, imm0_31);
12867 i32 __builtin_mips_balign (i32, i32, imm0_3);
12868 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
12869 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
12870 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
12871 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
12872 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
12873 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
12874 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
12875 q31 __builtin_mips_mulq_rs_w (q31, q31);
12876 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
12877 q31 __builtin_mips_mulq_s_w (q31, q31);
12878 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
12879 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
12880 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
12881 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
12882 i32 __builtin_mips_prepend (i32, i32, imm0_31);
12883 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
12884 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
12885 v4i8 __builtin_mips_shra_qb (v4i8, i32);
12886 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
12887 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
12888 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
12889 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
12890 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
12891 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
12892 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
12893 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
12894 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
12895 q31 __builtin_mips_addqh_w (q31, q31);
12896 q31 __builtin_mips_addqh_r_w (q31, q31);
12897 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
12898 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
12899 q31 __builtin_mips_subqh_w (q31, q31);
12900 q31 __builtin_mips_subqh_r_w (q31, q31);
12901 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
12902 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
12903 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
12904 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
12905 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
12906 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
12910 @node MIPS Paired-Single Support
12911 @subsection MIPS Paired-Single Support
12913 The MIPS64 architecture includes a number of instructions that
12914 operate on pairs of single-precision floating-point values.
12915 Each pair is packed into a 64-bit floating-point register,
12916 with one element being designated the ``upper half'' and
12917 the other being designated the ``lower half''.
12919 GCC supports paired-single operations using both the generic
12920 vector extensions (@pxref{Vector Extensions}) and a collection of
12921 MIPS-specific built-in functions. Both kinds of support are
12922 enabled by the @option{-mpaired-single} command-line option.
12924 The vector type associated with paired-single values is usually
12925 called @code{v2sf}. It can be defined in C as follows:
12928 typedef float v2sf __attribute__ ((vector_size (8)));
12931 @code{v2sf} values are initialized in the same way as aggregates.
12935 v2sf a = @{1.5, 9.1@};
12938 b = (v2sf) @{e, f@};
12941 @emph{Note:} The CPU's endianness determines which value is stored in
12942 the upper half of a register and which value is stored in the lower half.
12943 On little-endian targets, the first value is the lower one and the second
12944 value is the upper one. The opposite order applies to big-endian targets.
12945 For example, the code above sets the lower half of @code{a} to
12946 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
12948 @node MIPS Loongson Built-in Functions
12949 @subsection MIPS Loongson Built-in Functions
12951 GCC provides intrinsics to access the SIMD instructions provided by the
12952 ST Microelectronics Loongson-2E and -2F processors. These intrinsics,
12953 available after inclusion of the @code{loongson.h} header file,
12954 operate on the following 64-bit vector types:
12957 @item @code{uint8x8_t}, a vector of eight unsigned 8-bit integers;
12958 @item @code{uint16x4_t}, a vector of four unsigned 16-bit integers;
12959 @item @code{uint32x2_t}, a vector of two unsigned 32-bit integers;
12960 @item @code{int8x8_t}, a vector of eight signed 8-bit integers;
12961 @item @code{int16x4_t}, a vector of four signed 16-bit integers;
12962 @item @code{int32x2_t}, a vector of two signed 32-bit integers.
12965 The intrinsics provided are listed below; each is named after the
12966 machine instruction to which it corresponds, with suffixes added as
12967 appropriate to distinguish intrinsics that expand to the same machine
12968 instruction yet have different argument types. Refer to the architecture
12969 documentation for a description of the functionality of each
12973 int16x4_t packsswh (int32x2_t s, int32x2_t t);
12974 int8x8_t packsshb (int16x4_t s, int16x4_t t);
12975 uint8x8_t packushb (uint16x4_t s, uint16x4_t t);
12976 uint32x2_t paddw_u (uint32x2_t s, uint32x2_t t);
12977 uint16x4_t paddh_u (uint16x4_t s, uint16x4_t t);
12978 uint8x8_t paddb_u (uint8x8_t s, uint8x8_t t);
12979 int32x2_t paddw_s (int32x2_t s, int32x2_t t);
12980 int16x4_t paddh_s (int16x4_t s, int16x4_t t);
12981 int8x8_t paddb_s (int8x8_t s, int8x8_t t);
12982 uint64_t paddd_u (uint64_t s, uint64_t t);
12983 int64_t paddd_s (int64_t s, int64_t t);
12984 int16x4_t paddsh (int16x4_t s, int16x4_t t);
12985 int8x8_t paddsb (int8x8_t s, int8x8_t t);
12986 uint16x4_t paddush (uint16x4_t s, uint16x4_t t);
12987 uint8x8_t paddusb (uint8x8_t s, uint8x8_t t);
12988 uint64_t pandn_ud (uint64_t s, uint64_t t);
12989 uint32x2_t pandn_uw (uint32x2_t s, uint32x2_t t);
12990 uint16x4_t pandn_uh (uint16x4_t s, uint16x4_t t);
12991 uint8x8_t pandn_ub (uint8x8_t s, uint8x8_t t);
12992 int64_t pandn_sd (int64_t s, int64_t t);
12993 int32x2_t pandn_sw (int32x2_t s, int32x2_t t);
12994 int16x4_t pandn_sh (int16x4_t s, int16x4_t t);
12995 int8x8_t pandn_sb (int8x8_t s, int8x8_t t);
12996 uint16x4_t pavgh (uint16x4_t s, uint16x4_t t);
12997 uint8x8_t pavgb (uint8x8_t s, uint8x8_t t);
12998 uint32x2_t pcmpeqw_u (uint32x2_t s, uint32x2_t t);
12999 uint16x4_t pcmpeqh_u (uint16x4_t s, uint16x4_t t);
13000 uint8x8_t pcmpeqb_u (uint8x8_t s, uint8x8_t t);
13001 int32x2_t pcmpeqw_s (int32x2_t s, int32x2_t t);
13002 int16x4_t pcmpeqh_s (int16x4_t s, int16x4_t t);
13003 int8x8_t pcmpeqb_s (int8x8_t s, int8x8_t t);
13004 uint32x2_t pcmpgtw_u (uint32x2_t s, uint32x2_t t);
13005 uint16x4_t pcmpgth_u (uint16x4_t s, uint16x4_t t);
13006 uint8x8_t pcmpgtb_u (uint8x8_t s, uint8x8_t t);
13007 int32x2_t pcmpgtw_s (int32x2_t s, int32x2_t t);
13008 int16x4_t pcmpgth_s (int16x4_t s, int16x4_t t);
13009 int8x8_t pcmpgtb_s (int8x8_t s, int8x8_t t);
13010 uint16x4_t pextrh_u (uint16x4_t s, int field);
13011 int16x4_t pextrh_s (int16x4_t s, int field);
13012 uint16x4_t pinsrh_0_u (uint16x4_t s, uint16x4_t t);
13013 uint16x4_t pinsrh_1_u (uint16x4_t s, uint16x4_t t);
13014 uint16x4_t pinsrh_2_u (uint16x4_t s, uint16x4_t t);
13015 uint16x4_t pinsrh_3_u (uint16x4_t s, uint16x4_t t);
13016 int16x4_t pinsrh_0_s (int16x4_t s, int16x4_t t);
13017 int16x4_t pinsrh_1_s (int16x4_t s, int16x4_t t);
13018 int16x4_t pinsrh_2_s (int16x4_t s, int16x4_t t);
13019 int16x4_t pinsrh_3_s (int16x4_t s, int16x4_t t);
13020 int32x2_t pmaddhw (int16x4_t s, int16x4_t t);
13021 int16x4_t pmaxsh (int16x4_t s, int16x4_t t);
13022 uint8x8_t pmaxub (uint8x8_t s, uint8x8_t t);
13023 int16x4_t pminsh (int16x4_t s, int16x4_t t);
13024 uint8x8_t pminub (uint8x8_t s, uint8x8_t t);
13025 uint8x8_t pmovmskb_u (uint8x8_t s);
13026 int8x8_t pmovmskb_s (int8x8_t s);
13027 uint16x4_t pmulhuh (uint16x4_t s, uint16x4_t t);
13028 int16x4_t pmulhh (int16x4_t s, int16x4_t t);
13029 int16x4_t pmullh (int16x4_t s, int16x4_t t);
13030 int64_t pmuluw (uint32x2_t s, uint32x2_t t);
13031 uint8x8_t pasubub (uint8x8_t s, uint8x8_t t);
13032 uint16x4_t biadd (uint8x8_t s);
13033 uint16x4_t psadbh (uint8x8_t s, uint8x8_t t);
13034 uint16x4_t pshufh_u (uint16x4_t dest, uint16x4_t s, uint8_t order);
13035 int16x4_t pshufh_s (int16x4_t dest, int16x4_t s, uint8_t order);
13036 uint16x4_t psllh_u (uint16x4_t s, uint8_t amount);
13037 int16x4_t psllh_s (int16x4_t s, uint8_t amount);
13038 uint32x2_t psllw_u (uint32x2_t s, uint8_t amount);
13039 int32x2_t psllw_s (int32x2_t s, uint8_t amount);
13040 uint16x4_t psrlh_u (uint16x4_t s, uint8_t amount);
13041 int16x4_t psrlh_s (int16x4_t s, uint8_t amount);
13042 uint32x2_t psrlw_u (uint32x2_t s, uint8_t amount);
13043 int32x2_t psrlw_s (int32x2_t s, uint8_t amount);
13044 uint16x4_t psrah_u (uint16x4_t s, uint8_t amount);
13045 int16x4_t psrah_s (int16x4_t s, uint8_t amount);
13046 uint32x2_t psraw_u (uint32x2_t s, uint8_t amount);
13047 int32x2_t psraw_s (int32x2_t s, uint8_t amount);
13048 uint32x2_t psubw_u (uint32x2_t s, uint32x2_t t);
13049 uint16x4_t psubh_u (uint16x4_t s, uint16x4_t t);
13050 uint8x8_t psubb_u (uint8x8_t s, uint8x8_t t);
13051 int32x2_t psubw_s (int32x2_t s, int32x2_t t);
13052 int16x4_t psubh_s (int16x4_t s, int16x4_t t);
13053 int8x8_t psubb_s (int8x8_t s, int8x8_t t);
13054 uint64_t psubd_u (uint64_t s, uint64_t t);
13055 int64_t psubd_s (int64_t s, int64_t t);
13056 int16x4_t psubsh (int16x4_t s, int16x4_t t);
13057 int8x8_t psubsb (int8x8_t s, int8x8_t t);
13058 uint16x4_t psubush (uint16x4_t s, uint16x4_t t);
13059 uint8x8_t psubusb (uint8x8_t s, uint8x8_t t);
13060 uint32x2_t punpckhwd_u (uint32x2_t s, uint32x2_t t);
13061 uint16x4_t punpckhhw_u (uint16x4_t s, uint16x4_t t);
13062 uint8x8_t punpckhbh_u (uint8x8_t s, uint8x8_t t);
13063 int32x2_t punpckhwd_s (int32x2_t s, int32x2_t t);
13064 int16x4_t punpckhhw_s (int16x4_t s, int16x4_t t);
13065 int8x8_t punpckhbh_s (int8x8_t s, int8x8_t t);
13066 uint32x2_t punpcklwd_u (uint32x2_t s, uint32x2_t t);
13067 uint16x4_t punpcklhw_u (uint16x4_t s, uint16x4_t t);
13068 uint8x8_t punpcklbh_u (uint8x8_t s, uint8x8_t t);
13069 int32x2_t punpcklwd_s (int32x2_t s, int32x2_t t);
13070 int16x4_t punpcklhw_s (int16x4_t s, int16x4_t t);
13071 int8x8_t punpcklbh_s (int8x8_t s, int8x8_t t);
13075 * Paired-Single Arithmetic::
13076 * Paired-Single Built-in Functions::
13077 * MIPS-3D Built-in Functions::
13080 @node Paired-Single Arithmetic
13081 @subsubsection Paired-Single Arithmetic
13083 The table below lists the @code{v2sf} operations for which hardware
13084 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
13085 values and @code{x} is an integral value.
13087 @multitable @columnfractions .50 .50
13088 @item C code @tab MIPS instruction
13089 @item @code{a + b} @tab @code{add.ps}
13090 @item @code{a - b} @tab @code{sub.ps}
13091 @item @code{-a} @tab @code{neg.ps}
13092 @item @code{a * b} @tab @code{mul.ps}
13093 @item @code{a * b + c} @tab @code{madd.ps}
13094 @item @code{a * b - c} @tab @code{msub.ps}
13095 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
13096 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
13097 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
13100 Note that the multiply-accumulate instructions can be disabled
13101 using the command-line option @code{-mno-fused-madd}.
13103 @node Paired-Single Built-in Functions
13104 @subsubsection Paired-Single Built-in Functions
13106 The following paired-single functions map directly to a particular
13107 MIPS instruction. Please refer to the architecture specification
13108 for details on what each instruction does.
13111 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
13112 Pair lower lower (@code{pll.ps}).
13114 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
13115 Pair upper lower (@code{pul.ps}).
13117 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
13118 Pair lower upper (@code{plu.ps}).
13120 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
13121 Pair upper upper (@code{puu.ps}).
13123 @item v2sf __builtin_mips_cvt_ps_s (float, float)
13124 Convert pair to paired single (@code{cvt.ps.s}).
13126 @item float __builtin_mips_cvt_s_pl (v2sf)
13127 Convert pair lower to single (@code{cvt.s.pl}).
13129 @item float __builtin_mips_cvt_s_pu (v2sf)
13130 Convert pair upper to single (@code{cvt.s.pu}).
13132 @item v2sf __builtin_mips_abs_ps (v2sf)
13133 Absolute value (@code{abs.ps}).
13135 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
13136 Align variable (@code{alnv.ps}).
13138 @emph{Note:} The value of the third parameter must be 0 or 4
13139 modulo 8, otherwise the result is unpredictable. Please read the
13140 instruction description for details.
13143 The following multi-instruction functions are also available.
13144 In each case, @var{cond} can be any of the 16 floating-point conditions:
13145 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
13146 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
13147 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
13150 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
13151 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
13152 Conditional move based on floating-point comparison (@code{c.@var{cond}.ps},
13153 @code{movt.ps}/@code{movf.ps}).
13155 The @code{movt} functions return the value @var{x} computed by:
13158 c.@var{cond}.ps @var{cc},@var{a},@var{b}
13159 mov.ps @var{x},@var{c}
13160 movt.ps @var{x},@var{d},@var{cc}
13163 The @code{movf} functions are similar but use @code{movf.ps} instead
13166 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
13167 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
13168 Comparison of two paired-single values (@code{c.@var{cond}.ps},
13169 @code{bc1t}/@code{bc1f}).
13171 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
13172 and return either the upper or lower half of the result. For example:
13176 if (__builtin_mips_upper_c_eq_ps (a, b))
13177 upper_halves_are_equal ();
13179 upper_halves_are_unequal ();
13181 if (__builtin_mips_lower_c_eq_ps (a, b))
13182 lower_halves_are_equal ();
13184 lower_halves_are_unequal ();
13188 @node MIPS-3D Built-in Functions
13189 @subsubsection MIPS-3D Built-in Functions
13191 The MIPS-3D Application-Specific Extension (ASE) includes additional
13192 paired-single instructions that are designed to improve the performance
13193 of 3D graphics operations. Support for these instructions is controlled
13194 by the @option{-mips3d} command-line option.
13196 The functions listed below map directly to a particular MIPS-3D
13197 instruction. Please refer to the architecture specification for
13198 more details on what each instruction does.
13201 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
13202 Reduction add (@code{addr.ps}).
13204 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
13205 Reduction multiply (@code{mulr.ps}).
13207 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
13208 Convert paired single to paired word (@code{cvt.pw.ps}).
13210 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
13211 Convert paired word to paired single (@code{cvt.ps.pw}).
13213 @item float __builtin_mips_recip1_s (float)
13214 @itemx double __builtin_mips_recip1_d (double)
13215 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
13216 Reduced-precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
13218 @item float __builtin_mips_recip2_s (float, float)
13219 @itemx double __builtin_mips_recip2_d (double, double)
13220 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
13221 Reduced-precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
13223 @item float __builtin_mips_rsqrt1_s (float)
13224 @itemx double __builtin_mips_rsqrt1_d (double)
13225 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
13226 Reduced-precision reciprocal square root (sequence step 1)
13227 (@code{rsqrt1.@var{fmt}}).
13229 @item float __builtin_mips_rsqrt2_s (float, float)
13230 @itemx double __builtin_mips_rsqrt2_d (double, double)
13231 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
13232 Reduced-precision reciprocal square root (sequence step 2)
13233 (@code{rsqrt2.@var{fmt}}).
13236 The following multi-instruction functions are also available.
13237 In each case, @var{cond} can be any of the 16 floating-point conditions:
13238 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
13239 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
13240 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
13243 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
13244 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
13245 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
13246 @code{bc1t}/@code{bc1f}).
13248 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
13249 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
13254 if (__builtin_mips_cabs_eq_s (a, b))
13260 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
13261 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
13262 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
13263 @code{bc1t}/@code{bc1f}).
13265 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
13266 and return either the upper or lower half of the result. For example:
13270 if (__builtin_mips_upper_cabs_eq_ps (a, b))
13271 upper_halves_are_equal ();
13273 upper_halves_are_unequal ();
13275 if (__builtin_mips_lower_cabs_eq_ps (a, b))
13276 lower_halves_are_equal ();
13278 lower_halves_are_unequal ();
13281 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
13282 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
13283 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
13284 @code{movt.ps}/@code{movf.ps}).
13286 The @code{movt} functions return the value @var{x} computed by:
13289 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
13290 mov.ps @var{x},@var{c}
13291 movt.ps @var{x},@var{d},@var{cc}
13294 The @code{movf} functions are similar but use @code{movf.ps} instead
13297 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
13298 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
13299 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
13300 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
13301 Comparison of two paired-single values
13302 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
13303 @code{bc1any2t}/@code{bc1any2f}).
13305 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
13306 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
13307 result is true and the @code{all} forms return true if both results are true.
13312 if (__builtin_mips_any_c_eq_ps (a, b))
13317 if (__builtin_mips_all_c_eq_ps (a, b))
13323 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
13324 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
13325 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
13326 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
13327 Comparison of four paired-single values
13328 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
13329 @code{bc1any4t}/@code{bc1any4f}).
13331 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
13332 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
13333 The @code{any} forms return true if any of the four results are true
13334 and the @code{all} forms return true if all four results are true.
13339 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
13344 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
13351 @node Other MIPS Built-in Functions
13352 @subsection Other MIPS Built-in Functions
13354 GCC provides other MIPS-specific built-in functions:
13357 @item void __builtin_mips_cache (int @var{op}, const volatile void *@var{addr})
13358 Insert a @samp{cache} instruction with operands @var{op} and @var{addr}.
13359 GCC defines the preprocessor macro @code{___GCC_HAVE_BUILTIN_MIPS_CACHE}
13360 when this function is available.
13362 @item unsigned int __builtin_mips_get_fcsr (void)
13363 @itemx void __builtin_mips_set_fcsr (unsigned int @var{value})
13364 Get and set the contents of the floating-point control and status register
13365 (FPU control register 31). These functions are only available in hard-float
13366 code but can be called in both MIPS16 and non-MIPS16 contexts.
13368 @code{__builtin_mips_set_fcsr} can be used to change any bit of the
13369 register except the condition codes, which GCC assumes are preserved.
13372 @node MSP430 Built-in Functions
13373 @subsection MSP430 Built-in Functions
13375 GCC provides a couple of special builtin functions to aid in the
13376 writing of interrupt handlers in C.
13379 @item __bic_SR_register_on_exit (int @var{mask})
13380 This clears the indicated bits in the saved copy of the status register
13381 currently residing on the stack. This only works inside interrupt
13382 handlers and the changes to the status register will only take affect
13383 once the handler returns.
13385 @item __bis_SR_register_on_exit (int @var{mask})
13386 This sets the indicated bits in the saved copy of the status register
13387 currently residing on the stack. This only works inside interrupt
13388 handlers and the changes to the status register will only take affect
13389 once the handler returns.
13391 @item __delay_cycles (long long @var{cycles})
13392 This inserts an instruction sequence that takes exactly @var{cycles}
13393 cycles (between 0 and about 17E9) to complete. The inserted sequence
13394 may use jumps, loops, or no-ops, and does not interfere with any other
13395 instructions. Note that @var{cycles} must be a compile-time constant
13396 integer - that is, you must pass a number, not a variable that may be
13397 optimized to a constant later. The number of cycles delayed by this
13401 @node NDS32 Built-in Functions
13402 @subsection NDS32 Built-in Functions
13404 These built-in functions are available for the NDS32 target:
13406 @deftypefn {Built-in Function} void __builtin_nds32_isync (int *@var{addr})
13407 Insert an ISYNC instruction into the instruction stream where
13408 @var{addr} is an instruction address for serialization.
13411 @deftypefn {Built-in Function} void __builtin_nds32_isb (void)
13412 Insert an ISB instruction into the instruction stream.
13415 @deftypefn {Built-in Function} int __builtin_nds32_mfsr (int @var{sr})
13416 Return the content of a system register which is mapped by @var{sr}.
13419 @deftypefn {Built-in Function} int __builtin_nds32_mfusr (int @var{usr})
13420 Return the content of a user space register which is mapped by @var{usr}.
13423 @deftypefn {Built-in Function} void __builtin_nds32_mtsr (int @var{value}, int @var{sr})
13424 Move the @var{value} to a system register which is mapped by @var{sr}.
13427 @deftypefn {Built-in Function} void __builtin_nds32_mtusr (int @var{value}, int @var{usr})
13428 Move the @var{value} to a user space register which is mapped by @var{usr}.
13431 @deftypefn {Built-in Function} void __builtin_nds32_setgie_en (void)
13432 Enable global interrupt.
13435 @deftypefn {Built-in Function} void __builtin_nds32_setgie_dis (void)
13436 Disable global interrupt.
13439 @node picoChip Built-in Functions
13440 @subsection picoChip Built-in Functions
13442 GCC provides an interface to selected machine instructions from the
13443 picoChip instruction set.
13446 @item int __builtin_sbc (int @var{value})
13447 Sign bit count. Return the number of consecutive bits in @var{value}
13448 that have the same value as the sign bit. The result is the number of
13449 leading sign bits minus one, giving the number of redundant sign bits in
13452 @item int __builtin_byteswap (int @var{value})
13453 Byte swap. Return the result of swapping the upper and lower bytes of
13456 @item int __builtin_brev (int @var{value})
13457 Bit reversal. Return the result of reversing the bits in
13458 @var{value}. Bit 15 is swapped with bit 0, bit 14 is swapped with bit 1,
13461 @item int __builtin_adds (int @var{x}, int @var{y})
13462 Saturating addition. Return the result of adding @var{x} and @var{y},
13463 storing the value 32767 if the result overflows.
13465 @item int __builtin_subs (int @var{x}, int @var{y})
13466 Saturating subtraction. Return the result of subtracting @var{y} from
13467 @var{x}, storing the value @minus{}32768 if the result overflows.
13469 @item void __builtin_halt (void)
13470 Halt. The processor stops execution. This built-in is useful for
13471 implementing assertions.
13475 @node PowerPC Built-in Functions
13476 @subsection PowerPC Built-in Functions
13478 These built-in functions are available for the PowerPC family of
13481 float __builtin_recipdivf (float, float);
13482 float __builtin_rsqrtf (float);
13483 double __builtin_recipdiv (double, double);
13484 double __builtin_rsqrt (double);
13485 uint64_t __builtin_ppc_get_timebase ();
13486 unsigned long __builtin_ppc_mftb ();
13487 double __builtin_unpack_longdouble (long double, int);
13488 long double __builtin_pack_longdouble (double, double);
13491 The @code{vec_rsqrt}, @code{__builtin_rsqrt}, and
13492 @code{__builtin_rsqrtf} functions generate multiple instructions to
13493 implement the reciprocal sqrt functionality using reciprocal sqrt
13494 estimate instructions.
13496 The @code{__builtin_recipdiv}, and @code{__builtin_recipdivf}
13497 functions generate multiple instructions to implement division using
13498 the reciprocal estimate instructions.
13500 The @code{__builtin_ppc_get_timebase} and @code{__builtin_ppc_mftb}
13501 functions generate instructions to read the Time Base Register. The
13502 @code{__builtin_ppc_get_timebase} function may generate multiple
13503 instructions and always returns the 64 bits of the Time Base Register.
13504 The @code{__builtin_ppc_mftb} function always generates one instruction and
13505 returns the Time Base Register value as an unsigned long, throwing away
13506 the most significant word on 32-bit environments.
13508 The following built-in functions are available for the PowerPC family
13509 of processors, starting with ISA 2.06 or later (@option{-mcpu=power7}
13510 or @option{-mpopcntd}):
13512 long __builtin_bpermd (long, long);
13513 int __builtin_divwe (int, int);
13514 int __builtin_divweo (int, int);
13515 unsigned int __builtin_divweu (unsigned int, unsigned int);
13516 unsigned int __builtin_divweuo (unsigned int, unsigned int);
13517 long __builtin_divde (long, long);
13518 long __builtin_divdeo (long, long);
13519 unsigned long __builtin_divdeu (unsigned long, unsigned long);
13520 unsigned long __builtin_divdeuo (unsigned long, unsigned long);
13521 unsigned int cdtbcd (unsigned int);
13522 unsigned int cbcdtd (unsigned int);
13523 unsigned int addg6s (unsigned int, unsigned int);
13526 The @code{__builtin_divde}, @code{__builtin_divdeo},
13527 @code{__builtin_divdeu}, @code{__builtin_divdeou} functions require a
13528 64-bit environment support ISA 2.06 or later.
13530 The following built-in functions are available for the PowerPC family
13531 of processors when hardware decimal floating point
13532 (@option{-mhard-dfp}) is available:
13534 _Decimal64 __builtin_dxex (_Decimal64);
13535 _Decimal128 __builtin_dxexq (_Decimal128);
13536 _Decimal64 __builtin_ddedpd (int, _Decimal64);
13537 _Decimal128 __builtin_ddedpdq (int, _Decimal128);
13538 _Decimal64 __builtin_denbcd (int, _Decimal64);
13539 _Decimal128 __builtin_denbcdq (int, _Decimal128);
13540 _Decimal64 __builtin_diex (_Decimal64, _Decimal64);
13541 _Decimal128 _builtin_diexq (_Decimal128, _Decimal128);
13542 _Decimal64 __builtin_dscli (_Decimal64, int);
13543 _Decimal128 __builtin_dscliq (_Decimal128, int);
13544 _Decimal64 __builtin_dscri (_Decimal64, int);
13545 _Decimal128 __builtin_dscriq (_Decimal128, int);
13546 unsigned long long __builtin_unpack_dec128 (_Decimal128, int);
13547 _Decimal128 __builtin_pack_dec128 (unsigned long long, unsigned long long);
13550 The following built-in functions are available for the PowerPC family
13551 of processors when the Vector Scalar (vsx) instruction set is
13554 unsigned long long __builtin_unpack_vector_int128 (vector __int128_t, int);
13555 vector __int128_t __builtin_pack_vector_int128 (unsigned long long,
13556 unsigned long long);
13559 @node PowerPC AltiVec/VSX Built-in Functions
13560 @subsection PowerPC AltiVec Built-in Functions
13562 GCC provides an interface for the PowerPC family of processors to access
13563 the AltiVec operations described in Motorola's AltiVec Programming
13564 Interface Manual. The interface is made available by including
13565 @code{<altivec.h>} and using @option{-maltivec} and
13566 @option{-mabi=altivec}. The interface supports the following vector
13570 vector unsigned char
13574 vector unsigned short
13575 vector signed short
13579 vector unsigned int
13585 If @option{-mvsx} is used the following additional vector types are
13589 vector unsigned long
13594 The long types are only implemented for 64-bit code generation, and
13595 the long type is only used in the floating point/integer conversion
13598 GCC's implementation of the high-level language interface available from
13599 C and C++ code differs from Motorola's documentation in several ways.
13604 A vector constant is a list of constant expressions within curly braces.
13607 A vector initializer requires no cast if the vector constant is of the
13608 same type as the variable it is initializing.
13611 If @code{signed} or @code{unsigned} is omitted, the signedness of the
13612 vector type is the default signedness of the base type. The default
13613 varies depending on the operating system, so a portable program should
13614 always specify the signedness.
13617 Compiling with @option{-maltivec} adds keywords @code{__vector},
13618 @code{vector}, @code{__pixel}, @code{pixel}, @code{__bool} and
13619 @code{bool}. When compiling ISO C, the context-sensitive substitution
13620 of the keywords @code{vector}, @code{pixel} and @code{bool} is
13621 disabled. To use them, you must include @code{<altivec.h>} instead.
13624 GCC allows using a @code{typedef} name as the type specifier for a
13628 For C, overloaded functions are implemented with macros so the following
13632 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
13636 Since @code{vec_add} is a macro, the vector constant in the example
13637 is treated as four separate arguments. Wrap the entire argument in
13638 parentheses for this to work.
13641 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
13642 Internally, GCC uses built-in functions to achieve the functionality in
13643 the aforementioned header file, but they are not supported and are
13644 subject to change without notice.
13646 The following interfaces are supported for the generic and specific
13647 AltiVec operations and the AltiVec predicates. In cases where there
13648 is a direct mapping between generic and specific operations, only the
13649 generic names are shown here, although the specific operations can also
13652 Arguments that are documented as @code{const int} require literal
13653 integral values within the range required for that operation.
13656 vector signed char vec_abs (vector signed char);
13657 vector signed short vec_abs (vector signed short);
13658 vector signed int vec_abs (vector signed int);
13659 vector float vec_abs (vector float);
13661 vector signed char vec_abss (vector signed char);
13662 vector signed short vec_abss (vector signed short);
13663 vector signed int vec_abss (vector signed int);
13665 vector signed char vec_add (vector bool char, vector signed char);
13666 vector signed char vec_add (vector signed char, vector bool char);
13667 vector signed char vec_add (vector signed char, vector signed char);
13668 vector unsigned char vec_add (vector bool char, vector unsigned char);
13669 vector unsigned char vec_add (vector unsigned char, vector bool char);
13670 vector unsigned char vec_add (vector unsigned char,
13671 vector unsigned char);
13672 vector signed short vec_add (vector bool short, vector signed short);
13673 vector signed short vec_add (vector signed short, vector bool short);
13674 vector signed short vec_add (vector signed short, vector signed short);
13675 vector unsigned short vec_add (vector bool short,
13676 vector unsigned short);
13677 vector unsigned short vec_add (vector unsigned short,
13678 vector bool short);
13679 vector unsigned short vec_add (vector unsigned short,
13680 vector unsigned short);
13681 vector signed int vec_add (vector bool int, vector signed int);
13682 vector signed int vec_add (vector signed int, vector bool int);
13683 vector signed int vec_add (vector signed int, vector signed int);
13684 vector unsigned int vec_add (vector bool int, vector unsigned int);
13685 vector unsigned int vec_add (vector unsigned int, vector bool int);
13686 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
13687 vector float vec_add (vector float, vector float);
13689 vector float vec_vaddfp (vector float, vector float);
13691 vector signed int vec_vadduwm (vector bool int, vector signed int);
13692 vector signed int vec_vadduwm (vector signed int, vector bool int);
13693 vector signed int vec_vadduwm (vector signed int, vector signed int);
13694 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
13695 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
13696 vector unsigned int vec_vadduwm (vector unsigned int,
13697 vector unsigned int);
13699 vector signed short vec_vadduhm (vector bool short,
13700 vector signed short);
13701 vector signed short vec_vadduhm (vector signed short,
13702 vector bool short);
13703 vector signed short vec_vadduhm (vector signed short,
13704 vector signed short);
13705 vector unsigned short vec_vadduhm (vector bool short,
13706 vector unsigned short);
13707 vector unsigned short vec_vadduhm (vector unsigned short,
13708 vector bool short);
13709 vector unsigned short vec_vadduhm (vector unsigned short,
13710 vector unsigned short);
13712 vector signed char vec_vaddubm (vector bool char, vector signed char);
13713 vector signed char vec_vaddubm (vector signed char, vector bool char);
13714 vector signed char vec_vaddubm (vector signed char, vector signed char);
13715 vector unsigned char vec_vaddubm (vector bool char,
13716 vector unsigned char);
13717 vector unsigned char vec_vaddubm (vector unsigned char,
13719 vector unsigned char vec_vaddubm (vector unsigned char,
13720 vector unsigned char);
13722 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
13724 vector unsigned char vec_adds (vector bool char, vector unsigned char);
13725 vector unsigned char vec_adds (vector unsigned char, vector bool char);
13726 vector unsigned char vec_adds (vector unsigned char,
13727 vector unsigned char);
13728 vector signed char vec_adds (vector bool char, vector signed char);
13729 vector signed char vec_adds (vector signed char, vector bool char);
13730 vector signed char vec_adds (vector signed char, vector signed char);
13731 vector unsigned short vec_adds (vector bool short,
13732 vector unsigned short);
13733 vector unsigned short vec_adds (vector unsigned short,
13734 vector bool short);
13735 vector unsigned short vec_adds (vector unsigned short,
13736 vector unsigned short);
13737 vector signed short vec_adds (vector bool short, vector signed short);
13738 vector signed short vec_adds (vector signed short, vector bool short);
13739 vector signed short vec_adds (vector signed short, vector signed short);
13740 vector unsigned int vec_adds (vector bool int, vector unsigned int);
13741 vector unsigned int vec_adds (vector unsigned int, vector bool int);
13742 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
13743 vector signed int vec_adds (vector bool int, vector signed int);
13744 vector signed int vec_adds (vector signed int, vector bool int);
13745 vector signed int vec_adds (vector signed int, vector signed int);
13747 vector signed int vec_vaddsws (vector bool int, vector signed int);
13748 vector signed int vec_vaddsws (vector signed int, vector bool int);
13749 vector signed int vec_vaddsws (vector signed int, vector signed int);
13751 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
13752 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
13753 vector unsigned int vec_vadduws (vector unsigned int,
13754 vector unsigned int);
13756 vector signed short vec_vaddshs (vector bool short,
13757 vector signed short);
13758 vector signed short vec_vaddshs (vector signed short,
13759 vector bool short);
13760 vector signed short vec_vaddshs (vector signed short,
13761 vector signed short);
13763 vector unsigned short vec_vadduhs (vector bool short,
13764 vector unsigned short);
13765 vector unsigned short vec_vadduhs (vector unsigned short,
13766 vector bool short);
13767 vector unsigned short vec_vadduhs (vector unsigned short,
13768 vector unsigned short);
13770 vector signed char vec_vaddsbs (vector bool char, vector signed char);
13771 vector signed char vec_vaddsbs (vector signed char, vector bool char);
13772 vector signed char vec_vaddsbs (vector signed char, vector signed char);
13774 vector unsigned char vec_vaddubs (vector bool char,
13775 vector unsigned char);
13776 vector unsigned char vec_vaddubs (vector unsigned char,
13778 vector unsigned char vec_vaddubs (vector unsigned char,
13779 vector unsigned char);
13781 vector float vec_and (vector float, vector float);
13782 vector float vec_and (vector float, vector bool int);
13783 vector float vec_and (vector bool int, vector float);
13784 vector bool int vec_and (vector bool int, vector bool int);
13785 vector signed int vec_and (vector bool int, vector signed int);
13786 vector signed int vec_and (vector signed int, vector bool int);
13787 vector signed int vec_and (vector signed int, vector signed int);
13788 vector unsigned int vec_and (vector bool int, vector unsigned int);
13789 vector unsigned int vec_and (vector unsigned int, vector bool int);
13790 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
13791 vector bool short vec_and (vector bool short, vector bool short);
13792 vector signed short vec_and (vector bool short, vector signed short);
13793 vector signed short vec_and (vector signed short, vector bool short);
13794 vector signed short vec_and (vector signed short, vector signed short);
13795 vector unsigned short vec_and (vector bool short,
13796 vector unsigned short);
13797 vector unsigned short vec_and (vector unsigned short,
13798 vector bool short);
13799 vector unsigned short vec_and (vector unsigned short,
13800 vector unsigned short);
13801 vector signed char vec_and (vector bool char, vector signed char);
13802 vector bool char vec_and (vector bool char, vector bool char);
13803 vector signed char vec_and (vector signed char, vector bool char);
13804 vector signed char vec_and (vector signed char, vector signed char);
13805 vector unsigned char vec_and (vector bool char, vector unsigned char);
13806 vector unsigned char vec_and (vector unsigned char, vector bool char);
13807 vector unsigned char vec_and (vector unsigned char,
13808 vector unsigned char);
13810 vector float vec_andc (vector float, vector float);
13811 vector float vec_andc (vector float, vector bool int);
13812 vector float vec_andc (vector bool int, vector float);
13813 vector bool int vec_andc (vector bool int, vector bool int);
13814 vector signed int vec_andc (vector bool int, vector signed int);
13815 vector signed int vec_andc (vector signed int, vector bool int);
13816 vector signed int vec_andc (vector signed int, vector signed int);
13817 vector unsigned int vec_andc (vector bool int, vector unsigned int);
13818 vector unsigned int vec_andc (vector unsigned int, vector bool int);
13819 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
13820 vector bool short vec_andc (vector bool short, vector bool short);
13821 vector signed short vec_andc (vector bool short, vector signed short);
13822 vector signed short vec_andc (vector signed short, vector bool short);
13823 vector signed short vec_andc (vector signed short, vector signed short);
13824 vector unsigned short vec_andc (vector bool short,
13825 vector unsigned short);
13826 vector unsigned short vec_andc (vector unsigned short,
13827 vector bool short);
13828 vector unsigned short vec_andc (vector unsigned short,
13829 vector unsigned short);
13830 vector signed char vec_andc (vector bool char, vector signed char);
13831 vector bool char vec_andc (vector bool char, vector bool char);
13832 vector signed char vec_andc (vector signed char, vector bool char);
13833 vector signed char vec_andc (vector signed char, vector signed char);
13834 vector unsigned char vec_andc (vector bool char, vector unsigned char);
13835 vector unsigned char vec_andc (vector unsigned char, vector bool char);
13836 vector unsigned char vec_andc (vector unsigned char,
13837 vector unsigned char);
13839 vector unsigned char vec_avg (vector unsigned char,
13840 vector unsigned char);
13841 vector signed char vec_avg (vector signed char, vector signed char);
13842 vector unsigned short vec_avg (vector unsigned short,
13843 vector unsigned short);
13844 vector signed short vec_avg (vector signed short, vector signed short);
13845 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
13846 vector signed int vec_avg (vector signed int, vector signed int);
13848 vector signed int vec_vavgsw (vector signed int, vector signed int);
13850 vector unsigned int vec_vavguw (vector unsigned int,
13851 vector unsigned int);
13853 vector signed short vec_vavgsh (vector signed short,
13854 vector signed short);
13856 vector unsigned short vec_vavguh (vector unsigned short,
13857 vector unsigned short);
13859 vector signed char vec_vavgsb (vector signed char, vector signed char);
13861 vector unsigned char vec_vavgub (vector unsigned char,
13862 vector unsigned char);
13864 vector float vec_copysign (vector float);
13866 vector float vec_ceil (vector float);
13868 vector signed int vec_cmpb (vector float, vector float);
13870 vector bool char vec_cmpeq (vector signed char, vector signed char);
13871 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
13872 vector bool short vec_cmpeq (vector signed short, vector signed short);
13873 vector bool short vec_cmpeq (vector unsigned short,
13874 vector unsigned short);
13875 vector bool int vec_cmpeq (vector signed int, vector signed int);
13876 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
13877 vector bool int vec_cmpeq (vector float, vector float);
13879 vector bool int vec_vcmpeqfp (vector float, vector float);
13881 vector bool int vec_vcmpequw (vector signed int, vector signed int);
13882 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
13884 vector bool short vec_vcmpequh (vector signed short,
13885 vector signed short);
13886 vector bool short vec_vcmpequh (vector unsigned short,
13887 vector unsigned short);
13889 vector bool char vec_vcmpequb (vector signed char, vector signed char);
13890 vector bool char vec_vcmpequb (vector unsigned char,
13891 vector unsigned char);
13893 vector bool int vec_cmpge (vector float, vector float);
13895 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
13896 vector bool char vec_cmpgt (vector signed char, vector signed char);
13897 vector bool short vec_cmpgt (vector unsigned short,
13898 vector unsigned short);
13899 vector bool short vec_cmpgt (vector signed short, vector signed short);
13900 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
13901 vector bool int vec_cmpgt (vector signed int, vector signed int);
13902 vector bool int vec_cmpgt (vector float, vector float);
13904 vector bool int vec_vcmpgtfp (vector float, vector float);
13906 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
13908 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
13910 vector bool short vec_vcmpgtsh (vector signed short,
13911 vector signed short);
13913 vector bool short vec_vcmpgtuh (vector unsigned short,
13914 vector unsigned short);
13916 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
13918 vector bool char vec_vcmpgtub (vector unsigned char,
13919 vector unsigned char);
13921 vector bool int vec_cmple (vector float, vector float);
13923 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
13924 vector bool char vec_cmplt (vector signed char, vector signed char);
13925 vector bool short vec_cmplt (vector unsigned short,
13926 vector unsigned short);
13927 vector bool short vec_cmplt (vector signed short, vector signed short);
13928 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
13929 vector bool int vec_cmplt (vector signed int, vector signed int);
13930 vector bool int vec_cmplt (vector float, vector float);
13932 vector float vec_cpsgn (vector float, vector float);
13934 vector float vec_ctf (vector unsigned int, const int);
13935 vector float vec_ctf (vector signed int, const int);
13936 vector double vec_ctf (vector unsigned long, const int);
13937 vector double vec_ctf (vector signed long, const int);
13939 vector float vec_vcfsx (vector signed int, const int);
13941 vector float vec_vcfux (vector unsigned int, const int);
13943 vector signed int vec_cts (vector float, const int);
13944 vector signed long vec_cts (vector double, const int);
13946 vector unsigned int vec_ctu (vector float, const int);
13947 vector unsigned long vec_ctu (vector double, const int);
13949 void vec_dss (const int);
13951 void vec_dssall (void);
13953 void vec_dst (const vector unsigned char *, int, const int);
13954 void vec_dst (const vector signed char *, int, const int);
13955 void vec_dst (const vector bool char *, int, const int);
13956 void vec_dst (const vector unsigned short *, int, const int);
13957 void vec_dst (const vector signed short *, int, const int);
13958 void vec_dst (const vector bool short *, int, const int);
13959 void vec_dst (const vector pixel *, int, const int);
13960 void vec_dst (const vector unsigned int *, int, const int);
13961 void vec_dst (const vector signed int *, int, const int);
13962 void vec_dst (const vector bool int *, int, const int);
13963 void vec_dst (const vector float *, int, const int);
13964 void vec_dst (const unsigned char *, int, const int);
13965 void vec_dst (const signed char *, int, const int);
13966 void vec_dst (const unsigned short *, int, const int);
13967 void vec_dst (const short *, int, const int);
13968 void vec_dst (const unsigned int *, int, const int);
13969 void vec_dst (const int *, int, const int);
13970 void vec_dst (const unsigned long *, int, const int);
13971 void vec_dst (const long *, int, const int);
13972 void vec_dst (const float *, int, const int);
13974 void vec_dstst (const vector unsigned char *, int, const int);
13975 void vec_dstst (const vector signed char *, int, const int);
13976 void vec_dstst (const vector bool char *, int, const int);
13977 void vec_dstst (const vector unsigned short *, int, const int);
13978 void vec_dstst (const vector signed short *, int, const int);
13979 void vec_dstst (const vector bool short *, int, const int);
13980 void vec_dstst (const vector pixel *, int, const int);
13981 void vec_dstst (const vector unsigned int *, int, const int);
13982 void vec_dstst (const vector signed int *, int, const int);
13983 void vec_dstst (const vector bool int *, int, const int);
13984 void vec_dstst (const vector float *, int, const int);
13985 void vec_dstst (const unsigned char *, int, const int);
13986 void vec_dstst (const signed char *, int, const int);
13987 void vec_dstst (const unsigned short *, int, const int);
13988 void vec_dstst (const short *, int, const int);
13989 void vec_dstst (const unsigned int *, int, const int);
13990 void vec_dstst (const int *, int, const int);
13991 void vec_dstst (const unsigned long *, int, const int);
13992 void vec_dstst (const long *, int, const int);
13993 void vec_dstst (const float *, int, const int);
13995 void vec_dststt (const vector unsigned char *, int, const int);
13996 void vec_dststt (const vector signed char *, int, const int);
13997 void vec_dststt (const vector bool char *, int, const int);
13998 void vec_dststt (const vector unsigned short *, int, const int);
13999 void vec_dststt (const vector signed short *, int, const int);
14000 void vec_dststt (const vector bool short *, int, const int);
14001 void vec_dststt (const vector pixel *, int, const int);
14002 void vec_dststt (const vector unsigned int *, int, const int);
14003 void vec_dststt (const vector signed int *, int, const int);
14004 void vec_dststt (const vector bool int *, int, const int);
14005 void vec_dststt (const vector float *, int, const int);
14006 void vec_dststt (const unsigned char *, int, const int);
14007 void vec_dststt (const signed char *, int, const int);
14008 void vec_dststt (const unsigned short *, int, const int);
14009 void vec_dststt (const short *, int, const int);
14010 void vec_dststt (const unsigned int *, int, const int);
14011 void vec_dststt (const int *, int, const int);
14012 void vec_dststt (const unsigned long *, int, const int);
14013 void vec_dststt (const long *, int, const int);
14014 void vec_dststt (const float *, int, const int);
14016 void vec_dstt (const vector unsigned char *, int, const int);
14017 void vec_dstt (const vector signed char *, int, const int);
14018 void vec_dstt (const vector bool char *, int, const int);
14019 void vec_dstt (const vector unsigned short *, int, const int);
14020 void vec_dstt (const vector signed short *, int, const int);
14021 void vec_dstt (const vector bool short *, int, const int);
14022 void vec_dstt (const vector pixel *, int, const int);
14023 void vec_dstt (const vector unsigned int *, int, const int);
14024 void vec_dstt (const vector signed int *, int, const int);
14025 void vec_dstt (const vector bool int *, int, const int);
14026 void vec_dstt (const vector float *, int, const int);
14027 void vec_dstt (const unsigned char *, int, const int);
14028 void vec_dstt (const signed char *, int, const int);
14029 void vec_dstt (const unsigned short *, int, const int);
14030 void vec_dstt (const short *, int, const int);
14031 void vec_dstt (const unsigned int *, int, const int);
14032 void vec_dstt (const int *, int, const int);
14033 void vec_dstt (const unsigned long *, int, const int);
14034 void vec_dstt (const long *, int, const int);
14035 void vec_dstt (const float *, int, const int);
14037 vector float vec_expte (vector float);
14039 vector float vec_floor (vector float);
14041 vector float vec_ld (int, const vector float *);
14042 vector float vec_ld (int, const float *);
14043 vector bool int vec_ld (int, const vector bool int *);
14044 vector signed int vec_ld (int, const vector signed int *);
14045 vector signed int vec_ld (int, const int *);
14046 vector signed int vec_ld (int, const long *);
14047 vector unsigned int vec_ld (int, const vector unsigned int *);
14048 vector unsigned int vec_ld (int, const unsigned int *);
14049 vector unsigned int vec_ld (int, const unsigned long *);
14050 vector bool short vec_ld (int, const vector bool short *);
14051 vector pixel vec_ld (int, const vector pixel *);
14052 vector signed short vec_ld (int, const vector signed short *);
14053 vector signed short vec_ld (int, const short *);
14054 vector unsigned short vec_ld (int, const vector unsigned short *);
14055 vector unsigned short vec_ld (int, const unsigned short *);
14056 vector bool char vec_ld (int, const vector bool char *);
14057 vector signed char vec_ld (int, const vector signed char *);
14058 vector signed char vec_ld (int, const signed char *);
14059 vector unsigned char vec_ld (int, const vector unsigned char *);
14060 vector unsigned char vec_ld (int, const unsigned char *);
14062 vector signed char vec_lde (int, const signed char *);
14063 vector unsigned char vec_lde (int, const unsigned char *);
14064 vector signed short vec_lde (int, const short *);
14065 vector unsigned short vec_lde (int, const unsigned short *);
14066 vector float vec_lde (int, const float *);
14067 vector signed int vec_lde (int, const int *);
14068 vector unsigned int vec_lde (int, const unsigned int *);
14069 vector signed int vec_lde (int, const long *);
14070 vector unsigned int vec_lde (int, const unsigned long *);
14072 vector float vec_lvewx (int, float *);
14073 vector signed int vec_lvewx (int, int *);
14074 vector unsigned int vec_lvewx (int, unsigned int *);
14075 vector signed int vec_lvewx (int, long *);
14076 vector unsigned int vec_lvewx (int, unsigned long *);
14078 vector signed short vec_lvehx (int, short *);
14079 vector unsigned short vec_lvehx (int, unsigned short *);
14081 vector signed char vec_lvebx (int, char *);
14082 vector unsigned char vec_lvebx (int, unsigned char *);
14084 vector float vec_ldl (int, const vector float *);
14085 vector float vec_ldl (int, const float *);
14086 vector bool int vec_ldl (int, const vector bool int *);
14087 vector signed int vec_ldl (int, const vector signed int *);
14088 vector signed int vec_ldl (int, const int *);
14089 vector signed int vec_ldl (int, const long *);
14090 vector unsigned int vec_ldl (int, const vector unsigned int *);
14091 vector unsigned int vec_ldl (int, const unsigned int *);
14092 vector unsigned int vec_ldl (int, const unsigned long *);
14093 vector bool short vec_ldl (int, const vector bool short *);
14094 vector pixel vec_ldl (int, const vector pixel *);
14095 vector signed short vec_ldl (int, const vector signed short *);
14096 vector signed short vec_ldl (int, const short *);
14097 vector unsigned short vec_ldl (int, const vector unsigned short *);
14098 vector unsigned short vec_ldl (int, const unsigned short *);
14099 vector bool char vec_ldl (int, const vector bool char *);
14100 vector signed char vec_ldl (int, const vector signed char *);
14101 vector signed char vec_ldl (int, const signed char *);
14102 vector unsigned char vec_ldl (int, const vector unsigned char *);
14103 vector unsigned char vec_ldl (int, const unsigned char *);
14105 vector float vec_loge (vector float);
14107 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
14108 vector unsigned char vec_lvsl (int, const volatile signed char *);
14109 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
14110 vector unsigned char vec_lvsl (int, const volatile short *);
14111 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
14112 vector unsigned char vec_lvsl (int, const volatile int *);
14113 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
14114 vector unsigned char vec_lvsl (int, const volatile long *);
14115 vector unsigned char vec_lvsl (int, const volatile float *);
14117 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
14118 vector unsigned char vec_lvsr (int, const volatile signed char *);
14119 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
14120 vector unsigned char vec_lvsr (int, const volatile short *);
14121 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
14122 vector unsigned char vec_lvsr (int, const volatile int *);
14123 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
14124 vector unsigned char vec_lvsr (int, const volatile long *);
14125 vector unsigned char vec_lvsr (int, const volatile float *);
14127 vector float vec_madd (vector float, vector float, vector float);
14129 vector signed short vec_madds (vector signed short,
14130 vector signed short,
14131 vector signed short);
14133 vector unsigned char vec_max (vector bool char, vector unsigned char);
14134 vector unsigned char vec_max (vector unsigned char, vector bool char);
14135 vector unsigned char vec_max (vector unsigned char,
14136 vector unsigned char);
14137 vector signed char vec_max (vector bool char, vector signed char);
14138 vector signed char vec_max (vector signed char, vector bool char);
14139 vector signed char vec_max (vector signed char, vector signed char);
14140 vector unsigned short vec_max (vector bool short,
14141 vector unsigned short);
14142 vector unsigned short vec_max (vector unsigned short,
14143 vector bool short);
14144 vector unsigned short vec_max (vector unsigned short,
14145 vector unsigned short);
14146 vector signed short vec_max (vector bool short, vector signed short);
14147 vector signed short vec_max (vector signed short, vector bool short);
14148 vector signed short vec_max (vector signed short, vector signed short);
14149 vector unsigned int vec_max (vector bool int, vector unsigned int);
14150 vector unsigned int vec_max (vector unsigned int, vector bool int);
14151 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
14152 vector signed int vec_max (vector bool int, vector signed int);
14153 vector signed int vec_max (vector signed int, vector bool int);
14154 vector signed int vec_max (vector signed int, vector signed int);
14155 vector float vec_max (vector float, vector float);
14157 vector float vec_vmaxfp (vector float, vector float);
14159 vector signed int vec_vmaxsw (vector bool int, vector signed int);
14160 vector signed int vec_vmaxsw (vector signed int, vector bool int);
14161 vector signed int vec_vmaxsw (vector signed int, vector signed int);
14163 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
14164 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
14165 vector unsigned int vec_vmaxuw (vector unsigned int,
14166 vector unsigned int);
14168 vector signed short vec_vmaxsh (vector bool short, vector signed short);
14169 vector signed short vec_vmaxsh (vector signed short, vector bool short);
14170 vector signed short vec_vmaxsh (vector signed short,
14171 vector signed short);
14173 vector unsigned short vec_vmaxuh (vector bool short,
14174 vector unsigned short);
14175 vector unsigned short vec_vmaxuh (vector unsigned short,
14176 vector bool short);
14177 vector unsigned short vec_vmaxuh (vector unsigned short,
14178 vector unsigned short);
14180 vector signed char vec_vmaxsb (vector bool char, vector signed char);
14181 vector signed char vec_vmaxsb (vector signed char, vector bool char);
14182 vector signed char vec_vmaxsb (vector signed char, vector signed char);
14184 vector unsigned char vec_vmaxub (vector bool char,
14185 vector unsigned char);
14186 vector unsigned char vec_vmaxub (vector unsigned char,
14188 vector unsigned char vec_vmaxub (vector unsigned char,
14189 vector unsigned char);
14191 vector bool char vec_mergeh (vector bool char, vector bool char);
14192 vector signed char vec_mergeh (vector signed char, vector signed char);
14193 vector unsigned char vec_mergeh (vector unsigned char,
14194 vector unsigned char);
14195 vector bool short vec_mergeh (vector bool short, vector bool short);
14196 vector pixel vec_mergeh (vector pixel, vector pixel);
14197 vector signed short vec_mergeh (vector signed short,
14198 vector signed short);
14199 vector unsigned short vec_mergeh (vector unsigned short,
14200 vector unsigned short);
14201 vector float vec_mergeh (vector float, vector float);
14202 vector bool int vec_mergeh (vector bool int, vector bool int);
14203 vector signed int vec_mergeh (vector signed int, vector signed int);
14204 vector unsigned int vec_mergeh (vector unsigned int,
14205 vector unsigned int);
14207 vector float vec_vmrghw (vector float, vector float);
14208 vector bool int vec_vmrghw (vector bool int, vector bool int);
14209 vector signed int vec_vmrghw (vector signed int, vector signed int);
14210 vector unsigned int vec_vmrghw (vector unsigned int,
14211 vector unsigned int);
14213 vector bool short vec_vmrghh (vector bool short, vector bool short);
14214 vector signed short vec_vmrghh (vector signed short,
14215 vector signed short);
14216 vector unsigned short vec_vmrghh (vector unsigned short,
14217 vector unsigned short);
14218 vector pixel vec_vmrghh (vector pixel, vector pixel);
14220 vector bool char vec_vmrghb (vector bool char, vector bool char);
14221 vector signed char vec_vmrghb (vector signed char, vector signed char);
14222 vector unsigned char vec_vmrghb (vector unsigned char,
14223 vector unsigned char);
14225 vector bool char vec_mergel (vector bool char, vector bool char);
14226 vector signed char vec_mergel (vector signed char, vector signed char);
14227 vector unsigned char vec_mergel (vector unsigned char,
14228 vector unsigned char);
14229 vector bool short vec_mergel (vector bool short, vector bool short);
14230 vector pixel vec_mergel (vector pixel, vector pixel);
14231 vector signed short vec_mergel (vector signed short,
14232 vector signed short);
14233 vector unsigned short vec_mergel (vector unsigned short,
14234 vector unsigned short);
14235 vector float vec_mergel (vector float, vector float);
14236 vector bool int vec_mergel (vector bool int, vector bool int);
14237 vector signed int vec_mergel (vector signed int, vector signed int);
14238 vector unsigned int vec_mergel (vector unsigned int,
14239 vector unsigned int);
14241 vector float vec_vmrglw (vector float, vector float);
14242 vector signed int vec_vmrglw (vector signed int, vector signed int);
14243 vector unsigned int vec_vmrglw (vector unsigned int,
14244 vector unsigned int);
14245 vector bool int vec_vmrglw (vector bool int, vector bool int);
14247 vector bool short vec_vmrglh (vector bool short, vector bool short);
14248 vector signed short vec_vmrglh (vector signed short,
14249 vector signed short);
14250 vector unsigned short vec_vmrglh (vector unsigned short,
14251 vector unsigned short);
14252 vector pixel vec_vmrglh (vector pixel, vector pixel);
14254 vector bool char vec_vmrglb (vector bool char, vector bool char);
14255 vector signed char vec_vmrglb (vector signed char, vector signed char);
14256 vector unsigned char vec_vmrglb (vector unsigned char,
14257 vector unsigned char);
14259 vector unsigned short vec_mfvscr (void);
14261 vector unsigned char vec_min (vector bool char, vector unsigned char);
14262 vector unsigned char vec_min (vector unsigned char, vector bool char);
14263 vector unsigned char vec_min (vector unsigned char,
14264 vector unsigned char);
14265 vector signed char vec_min (vector bool char, vector signed char);
14266 vector signed char vec_min (vector signed char, vector bool char);
14267 vector signed char vec_min (vector signed char, vector signed char);
14268 vector unsigned short vec_min (vector bool short,
14269 vector unsigned short);
14270 vector unsigned short vec_min (vector unsigned short,
14271 vector bool short);
14272 vector unsigned short vec_min (vector unsigned short,
14273 vector unsigned short);
14274 vector signed short vec_min (vector bool short, vector signed short);
14275 vector signed short vec_min (vector signed short, vector bool short);
14276 vector signed short vec_min (vector signed short, vector signed short);
14277 vector unsigned int vec_min (vector bool int, vector unsigned int);
14278 vector unsigned int vec_min (vector unsigned int, vector bool int);
14279 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
14280 vector signed int vec_min (vector bool int, vector signed int);
14281 vector signed int vec_min (vector signed int, vector bool int);
14282 vector signed int vec_min (vector signed int, vector signed int);
14283 vector float vec_min (vector float, vector float);
14285 vector float vec_vminfp (vector float, vector float);
14287 vector signed int vec_vminsw (vector bool int, vector signed int);
14288 vector signed int vec_vminsw (vector signed int, vector bool int);
14289 vector signed int vec_vminsw (vector signed int, vector signed int);
14291 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
14292 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
14293 vector unsigned int vec_vminuw (vector unsigned int,
14294 vector unsigned int);
14296 vector signed short vec_vminsh (vector bool short, vector signed short);
14297 vector signed short vec_vminsh (vector signed short, vector bool short);
14298 vector signed short vec_vminsh (vector signed short,
14299 vector signed short);
14301 vector unsigned short vec_vminuh (vector bool short,
14302 vector unsigned short);
14303 vector unsigned short vec_vminuh (vector unsigned short,
14304 vector bool short);
14305 vector unsigned short vec_vminuh (vector unsigned short,
14306 vector unsigned short);
14308 vector signed char vec_vminsb (vector bool char, vector signed char);
14309 vector signed char vec_vminsb (vector signed char, vector bool char);
14310 vector signed char vec_vminsb (vector signed char, vector signed char);
14312 vector unsigned char vec_vminub (vector bool char,
14313 vector unsigned char);
14314 vector unsigned char vec_vminub (vector unsigned char,
14316 vector unsigned char vec_vminub (vector unsigned char,
14317 vector unsigned char);
14319 vector signed short vec_mladd (vector signed short,
14320 vector signed short,
14321 vector signed short);
14322 vector signed short vec_mladd (vector signed short,
14323 vector unsigned short,
14324 vector unsigned short);
14325 vector signed short vec_mladd (vector unsigned short,
14326 vector signed short,
14327 vector signed short);
14328 vector unsigned short vec_mladd (vector unsigned short,
14329 vector unsigned short,
14330 vector unsigned short);
14332 vector signed short vec_mradds (vector signed short,
14333 vector signed short,
14334 vector signed short);
14336 vector unsigned int vec_msum (vector unsigned char,
14337 vector unsigned char,
14338 vector unsigned int);
14339 vector signed int vec_msum (vector signed char,
14340 vector unsigned char,
14341 vector signed int);
14342 vector unsigned int vec_msum (vector unsigned short,
14343 vector unsigned short,
14344 vector unsigned int);
14345 vector signed int vec_msum (vector signed short,
14346 vector signed short,
14347 vector signed int);
14349 vector signed int vec_vmsumshm (vector signed short,
14350 vector signed short,
14351 vector signed int);
14353 vector unsigned int vec_vmsumuhm (vector unsigned short,
14354 vector unsigned short,
14355 vector unsigned int);
14357 vector signed int vec_vmsummbm (vector signed char,
14358 vector unsigned char,
14359 vector signed int);
14361 vector unsigned int vec_vmsumubm (vector unsigned char,
14362 vector unsigned char,
14363 vector unsigned int);
14365 vector unsigned int vec_msums (vector unsigned short,
14366 vector unsigned short,
14367 vector unsigned int);
14368 vector signed int vec_msums (vector signed short,
14369 vector signed short,
14370 vector signed int);
14372 vector signed int vec_vmsumshs (vector signed short,
14373 vector signed short,
14374 vector signed int);
14376 vector unsigned int vec_vmsumuhs (vector unsigned short,
14377 vector unsigned short,
14378 vector unsigned int);
14380 void vec_mtvscr (vector signed int);
14381 void vec_mtvscr (vector unsigned int);
14382 void vec_mtvscr (vector bool int);
14383 void vec_mtvscr (vector signed short);
14384 void vec_mtvscr (vector unsigned short);
14385 void vec_mtvscr (vector bool short);
14386 void vec_mtvscr (vector pixel);
14387 void vec_mtvscr (vector signed char);
14388 void vec_mtvscr (vector unsigned char);
14389 void vec_mtvscr (vector bool char);
14391 vector unsigned short vec_mule (vector unsigned char,
14392 vector unsigned char);
14393 vector signed short vec_mule (vector signed char,
14394 vector signed char);
14395 vector unsigned int vec_mule (vector unsigned short,
14396 vector unsigned short);
14397 vector signed int vec_mule (vector signed short, vector signed short);
14399 vector signed int vec_vmulesh (vector signed short,
14400 vector signed short);
14402 vector unsigned int vec_vmuleuh (vector unsigned short,
14403 vector unsigned short);
14405 vector signed short vec_vmulesb (vector signed char,
14406 vector signed char);
14408 vector unsigned short vec_vmuleub (vector unsigned char,
14409 vector unsigned char);
14411 vector unsigned short vec_mulo (vector unsigned char,
14412 vector unsigned char);
14413 vector signed short vec_mulo (vector signed char, vector signed char);
14414 vector unsigned int vec_mulo (vector unsigned short,
14415 vector unsigned short);
14416 vector signed int vec_mulo (vector signed short, vector signed short);
14418 vector signed int vec_vmulosh (vector signed short,
14419 vector signed short);
14421 vector unsigned int vec_vmulouh (vector unsigned short,
14422 vector unsigned short);
14424 vector signed short vec_vmulosb (vector signed char,
14425 vector signed char);
14427 vector unsigned short vec_vmuloub (vector unsigned char,
14428 vector unsigned char);
14430 vector float vec_nmsub (vector float, vector float, vector float);
14432 vector float vec_nor (vector float, vector float);
14433 vector signed int vec_nor (vector signed int, vector signed int);
14434 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
14435 vector bool int vec_nor (vector bool int, vector bool int);
14436 vector signed short vec_nor (vector signed short, vector signed short);
14437 vector unsigned short vec_nor (vector unsigned short,
14438 vector unsigned short);
14439 vector bool short vec_nor (vector bool short, vector bool short);
14440 vector signed char vec_nor (vector signed char, vector signed char);
14441 vector unsigned char vec_nor (vector unsigned char,
14442 vector unsigned char);
14443 vector bool char vec_nor (vector bool char, vector bool char);
14445 vector float vec_or (vector float, vector float);
14446 vector float vec_or (vector float, vector bool int);
14447 vector float vec_or (vector bool int, vector float);
14448 vector bool int vec_or (vector bool int, vector bool int);
14449 vector signed int vec_or (vector bool int, vector signed int);
14450 vector signed int vec_or (vector signed int, vector bool int);
14451 vector signed int vec_or (vector signed int, vector signed int);
14452 vector unsigned int vec_or (vector bool int, vector unsigned int);
14453 vector unsigned int vec_or (vector unsigned int, vector bool int);
14454 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
14455 vector bool short vec_or (vector bool short, vector bool short);
14456 vector signed short vec_or (vector bool short, vector signed short);
14457 vector signed short vec_or (vector signed short, vector bool short);
14458 vector signed short vec_or (vector signed short, vector signed short);
14459 vector unsigned short vec_or (vector bool short, vector unsigned short);
14460 vector unsigned short vec_or (vector unsigned short, vector bool short);
14461 vector unsigned short vec_or (vector unsigned short,
14462 vector unsigned short);
14463 vector signed char vec_or (vector bool char, vector signed char);
14464 vector bool char vec_or (vector bool char, vector bool char);
14465 vector signed char vec_or (vector signed char, vector bool char);
14466 vector signed char vec_or (vector signed char, vector signed char);
14467 vector unsigned char vec_or (vector bool char, vector unsigned char);
14468 vector unsigned char vec_or (vector unsigned char, vector bool char);
14469 vector unsigned char vec_or (vector unsigned char,
14470 vector unsigned char);
14472 vector signed char vec_pack (vector signed short, vector signed short);
14473 vector unsigned char vec_pack (vector unsigned short,
14474 vector unsigned short);
14475 vector bool char vec_pack (vector bool short, vector bool short);
14476 vector signed short vec_pack (vector signed int, vector signed int);
14477 vector unsigned short vec_pack (vector unsigned int,
14478 vector unsigned int);
14479 vector bool short vec_pack (vector bool int, vector bool int);
14481 vector bool short vec_vpkuwum (vector bool int, vector bool int);
14482 vector signed short vec_vpkuwum (vector signed int, vector signed int);
14483 vector unsigned short vec_vpkuwum (vector unsigned int,
14484 vector unsigned int);
14486 vector bool char vec_vpkuhum (vector bool short, vector bool short);
14487 vector signed char vec_vpkuhum (vector signed short,
14488 vector signed short);
14489 vector unsigned char vec_vpkuhum (vector unsigned short,
14490 vector unsigned short);
14492 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
14494 vector unsigned char vec_packs (vector unsigned short,
14495 vector unsigned short);
14496 vector signed char vec_packs (vector signed short, vector signed short);
14497 vector unsigned short vec_packs (vector unsigned int,
14498 vector unsigned int);
14499 vector signed short vec_packs (vector signed int, vector signed int);
14501 vector signed short vec_vpkswss (vector signed int, vector signed int);
14503 vector unsigned short vec_vpkuwus (vector unsigned int,
14504 vector unsigned int);
14506 vector signed char vec_vpkshss (vector signed short,
14507 vector signed short);
14509 vector unsigned char vec_vpkuhus (vector unsigned short,
14510 vector unsigned short);
14512 vector unsigned char vec_packsu (vector unsigned short,
14513 vector unsigned short);
14514 vector unsigned char vec_packsu (vector signed short,
14515 vector signed short);
14516 vector unsigned short vec_packsu (vector unsigned int,
14517 vector unsigned int);
14518 vector unsigned short vec_packsu (vector signed int, vector signed int);
14520 vector unsigned short vec_vpkswus (vector signed int,
14521 vector signed int);
14523 vector unsigned char vec_vpkshus (vector signed short,
14524 vector signed short);
14526 vector float vec_perm (vector float,
14528 vector unsigned char);
14529 vector signed int vec_perm (vector signed int,
14531 vector unsigned char);
14532 vector unsigned int vec_perm (vector unsigned int,
14533 vector unsigned int,
14534 vector unsigned char);
14535 vector bool int vec_perm (vector bool int,
14537 vector unsigned char);
14538 vector signed short vec_perm (vector signed short,
14539 vector signed short,
14540 vector unsigned char);
14541 vector unsigned short vec_perm (vector unsigned short,
14542 vector unsigned short,
14543 vector unsigned char);
14544 vector bool short vec_perm (vector bool short,
14546 vector unsigned char);
14547 vector pixel vec_perm (vector pixel,
14549 vector unsigned char);
14550 vector signed char vec_perm (vector signed char,
14551 vector signed char,
14552 vector unsigned char);
14553 vector unsigned char vec_perm (vector unsigned char,
14554 vector unsigned char,
14555 vector unsigned char);
14556 vector bool char vec_perm (vector bool char,
14558 vector unsigned char);
14560 vector float vec_re (vector float);
14562 vector signed char vec_rl (vector signed char,
14563 vector unsigned char);
14564 vector unsigned char vec_rl (vector unsigned char,
14565 vector unsigned char);
14566 vector signed short vec_rl (vector signed short, vector unsigned short);
14567 vector unsigned short vec_rl (vector unsigned short,
14568 vector unsigned short);
14569 vector signed int vec_rl (vector signed int, vector unsigned int);
14570 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
14572 vector signed int vec_vrlw (vector signed int, vector unsigned int);
14573 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
14575 vector signed short vec_vrlh (vector signed short,
14576 vector unsigned short);
14577 vector unsigned short vec_vrlh (vector unsigned short,
14578 vector unsigned short);
14580 vector signed char vec_vrlb (vector signed char, vector unsigned char);
14581 vector unsigned char vec_vrlb (vector unsigned char,
14582 vector unsigned char);
14584 vector float vec_round (vector float);
14586 vector float vec_recip (vector float, vector float);
14588 vector float vec_rsqrt (vector float);
14590 vector float vec_rsqrte (vector float);
14592 vector float vec_sel (vector float, vector float, vector bool int);
14593 vector float vec_sel (vector float, vector float, vector unsigned int);
14594 vector signed int vec_sel (vector signed int,
14597 vector signed int vec_sel (vector signed int,
14599 vector unsigned int);
14600 vector unsigned int vec_sel (vector unsigned int,
14601 vector unsigned int,
14603 vector unsigned int vec_sel (vector unsigned int,
14604 vector unsigned int,
14605 vector unsigned int);
14606 vector bool int vec_sel (vector bool int,
14609 vector bool int vec_sel (vector bool int,
14611 vector unsigned int);
14612 vector signed short vec_sel (vector signed short,
14613 vector signed short,
14614 vector bool short);
14615 vector signed short vec_sel (vector signed short,
14616 vector signed short,
14617 vector unsigned short);
14618 vector unsigned short vec_sel (vector unsigned short,
14619 vector unsigned short,
14620 vector bool short);
14621 vector unsigned short vec_sel (vector unsigned short,
14622 vector unsigned short,
14623 vector unsigned short);
14624 vector bool short vec_sel (vector bool short,
14626 vector bool short);
14627 vector bool short vec_sel (vector bool short,
14629 vector unsigned short);
14630 vector signed char vec_sel (vector signed char,
14631 vector signed char,
14633 vector signed char vec_sel (vector signed char,
14634 vector signed char,
14635 vector unsigned char);
14636 vector unsigned char vec_sel (vector unsigned char,
14637 vector unsigned char,
14639 vector unsigned char vec_sel (vector unsigned char,
14640 vector unsigned char,
14641 vector unsigned char);
14642 vector bool char vec_sel (vector bool char,
14645 vector bool char vec_sel (vector bool char,
14647 vector unsigned char);
14649 vector signed char vec_sl (vector signed char,
14650 vector unsigned char);
14651 vector unsigned char vec_sl (vector unsigned char,
14652 vector unsigned char);
14653 vector signed short vec_sl (vector signed short, vector unsigned short);
14654 vector unsigned short vec_sl (vector unsigned short,
14655 vector unsigned short);
14656 vector signed int vec_sl (vector signed int, vector unsigned int);
14657 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
14659 vector signed int vec_vslw (vector signed int, vector unsigned int);
14660 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
14662 vector signed short vec_vslh (vector signed short,
14663 vector unsigned short);
14664 vector unsigned short vec_vslh (vector unsigned short,
14665 vector unsigned short);
14667 vector signed char vec_vslb (vector signed char, vector unsigned char);
14668 vector unsigned char vec_vslb (vector unsigned char,
14669 vector unsigned char);
14671 vector float vec_sld (vector float, vector float, const int);
14672 vector signed int vec_sld (vector signed int,
14675 vector unsigned int vec_sld (vector unsigned int,
14676 vector unsigned int,
14678 vector bool int vec_sld (vector bool int,
14681 vector signed short vec_sld (vector signed short,
14682 vector signed short,
14684 vector unsigned short vec_sld (vector unsigned short,
14685 vector unsigned short,
14687 vector bool short vec_sld (vector bool short,
14690 vector pixel vec_sld (vector pixel,
14693 vector signed char vec_sld (vector signed char,
14694 vector signed char,
14696 vector unsigned char vec_sld (vector unsigned char,
14697 vector unsigned char,
14699 vector bool char vec_sld (vector bool char,
14703 vector signed int vec_sll (vector signed int,
14704 vector unsigned int);
14705 vector signed int vec_sll (vector signed int,
14706 vector unsigned short);
14707 vector signed int vec_sll (vector signed int,
14708 vector unsigned char);
14709 vector unsigned int vec_sll (vector unsigned int,
14710 vector unsigned int);
14711 vector unsigned int vec_sll (vector unsigned int,
14712 vector unsigned short);
14713 vector unsigned int vec_sll (vector unsigned int,
14714 vector unsigned char);
14715 vector bool int vec_sll (vector bool int,
14716 vector unsigned int);
14717 vector bool int vec_sll (vector bool int,
14718 vector unsigned short);
14719 vector bool int vec_sll (vector bool int,
14720 vector unsigned char);
14721 vector signed short vec_sll (vector signed short,
14722 vector unsigned int);
14723 vector signed short vec_sll (vector signed short,
14724 vector unsigned short);
14725 vector signed short vec_sll (vector signed short,
14726 vector unsigned char);
14727 vector unsigned short vec_sll (vector unsigned short,
14728 vector unsigned int);
14729 vector unsigned short vec_sll (vector unsigned short,
14730 vector unsigned short);
14731 vector unsigned short vec_sll (vector unsigned short,
14732 vector unsigned char);
14733 vector bool short vec_sll (vector bool short, vector unsigned int);
14734 vector bool short vec_sll (vector bool short, vector unsigned short);
14735 vector bool short vec_sll (vector bool short, vector unsigned char);
14736 vector pixel vec_sll (vector pixel, vector unsigned int);
14737 vector pixel vec_sll (vector pixel, vector unsigned short);
14738 vector pixel vec_sll (vector pixel, vector unsigned char);
14739 vector signed char vec_sll (vector signed char, vector unsigned int);
14740 vector signed char vec_sll (vector signed char, vector unsigned short);
14741 vector signed char vec_sll (vector signed char, vector unsigned char);
14742 vector unsigned char vec_sll (vector unsigned char,
14743 vector unsigned int);
14744 vector unsigned char vec_sll (vector unsigned char,
14745 vector unsigned short);
14746 vector unsigned char vec_sll (vector unsigned char,
14747 vector unsigned char);
14748 vector bool char vec_sll (vector bool char, vector unsigned int);
14749 vector bool char vec_sll (vector bool char, vector unsigned short);
14750 vector bool char vec_sll (vector bool char, vector unsigned char);
14752 vector float vec_slo (vector float, vector signed char);
14753 vector float vec_slo (vector float, vector unsigned char);
14754 vector signed int vec_slo (vector signed int, vector signed char);
14755 vector signed int vec_slo (vector signed int, vector unsigned char);
14756 vector unsigned int vec_slo (vector unsigned int, vector signed char);
14757 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
14758 vector signed short vec_slo (vector signed short, vector signed char);
14759 vector signed short vec_slo (vector signed short, vector unsigned char);
14760 vector unsigned short vec_slo (vector unsigned short,
14761 vector signed char);
14762 vector unsigned short vec_slo (vector unsigned short,
14763 vector unsigned char);
14764 vector pixel vec_slo (vector pixel, vector signed char);
14765 vector pixel vec_slo (vector pixel, vector unsigned char);
14766 vector signed char vec_slo (vector signed char, vector signed char);
14767 vector signed char vec_slo (vector signed char, vector unsigned char);
14768 vector unsigned char vec_slo (vector unsigned char, vector signed char);
14769 vector unsigned char vec_slo (vector unsigned char,
14770 vector unsigned char);
14772 vector signed char vec_splat (vector signed char, const int);
14773 vector unsigned char vec_splat (vector unsigned char, const int);
14774 vector bool char vec_splat (vector bool char, const int);
14775 vector signed short vec_splat (vector signed short, const int);
14776 vector unsigned short vec_splat (vector unsigned short, const int);
14777 vector bool short vec_splat (vector bool short, const int);
14778 vector pixel vec_splat (vector pixel, const int);
14779 vector float vec_splat (vector float, const int);
14780 vector signed int vec_splat (vector signed int, const int);
14781 vector unsigned int vec_splat (vector unsigned int, const int);
14782 vector bool int vec_splat (vector bool int, const int);
14783 vector signed long vec_splat (vector signed long, const int);
14784 vector unsigned long vec_splat (vector unsigned long, const int);
14786 vector signed char vec_splats (signed char);
14787 vector unsigned char vec_splats (unsigned char);
14788 vector signed short vec_splats (signed short);
14789 vector unsigned short vec_splats (unsigned short);
14790 vector signed int vec_splats (signed int);
14791 vector unsigned int vec_splats (unsigned int);
14792 vector float vec_splats (float);
14794 vector float vec_vspltw (vector float, const int);
14795 vector signed int vec_vspltw (vector signed int, const int);
14796 vector unsigned int vec_vspltw (vector unsigned int, const int);
14797 vector bool int vec_vspltw (vector bool int, const int);
14799 vector bool short vec_vsplth (vector bool short, const int);
14800 vector signed short vec_vsplth (vector signed short, const int);
14801 vector unsigned short vec_vsplth (vector unsigned short, const int);
14802 vector pixel vec_vsplth (vector pixel, const int);
14804 vector signed char vec_vspltb (vector signed char, const int);
14805 vector unsigned char vec_vspltb (vector unsigned char, const int);
14806 vector bool char vec_vspltb (vector bool char, const int);
14808 vector signed char vec_splat_s8 (const int);
14810 vector signed short vec_splat_s16 (const int);
14812 vector signed int vec_splat_s32 (const int);
14814 vector unsigned char vec_splat_u8 (const int);
14816 vector unsigned short vec_splat_u16 (const int);
14818 vector unsigned int vec_splat_u32 (const int);
14820 vector signed char vec_sr (vector signed char, vector unsigned char);
14821 vector unsigned char vec_sr (vector unsigned char,
14822 vector unsigned char);
14823 vector signed short vec_sr (vector signed short,
14824 vector unsigned short);
14825 vector unsigned short vec_sr (vector unsigned short,
14826 vector unsigned short);
14827 vector signed int vec_sr (vector signed int, vector unsigned int);
14828 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
14830 vector signed int vec_vsrw (vector signed int, vector unsigned int);
14831 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
14833 vector signed short vec_vsrh (vector signed short,
14834 vector unsigned short);
14835 vector unsigned short vec_vsrh (vector unsigned short,
14836 vector unsigned short);
14838 vector signed char vec_vsrb (vector signed char, vector unsigned char);
14839 vector unsigned char vec_vsrb (vector unsigned char,
14840 vector unsigned char);
14842 vector signed char vec_sra (vector signed char, vector unsigned char);
14843 vector unsigned char vec_sra (vector unsigned char,
14844 vector unsigned char);
14845 vector signed short vec_sra (vector signed short,
14846 vector unsigned short);
14847 vector unsigned short vec_sra (vector unsigned short,
14848 vector unsigned short);
14849 vector signed int vec_sra (vector signed int, vector unsigned int);
14850 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
14852 vector signed int vec_vsraw (vector signed int, vector unsigned int);
14853 vector unsigned int vec_vsraw (vector unsigned int,
14854 vector unsigned int);
14856 vector signed short vec_vsrah (vector signed short,
14857 vector unsigned short);
14858 vector unsigned short vec_vsrah (vector unsigned short,
14859 vector unsigned short);
14861 vector signed char vec_vsrab (vector signed char, vector unsigned char);
14862 vector unsigned char vec_vsrab (vector unsigned char,
14863 vector unsigned char);
14865 vector signed int vec_srl (vector signed int, vector unsigned int);
14866 vector signed int vec_srl (vector signed int, vector unsigned short);
14867 vector signed int vec_srl (vector signed int, vector unsigned char);
14868 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
14869 vector unsigned int vec_srl (vector unsigned int,
14870 vector unsigned short);
14871 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
14872 vector bool int vec_srl (vector bool int, vector unsigned int);
14873 vector bool int vec_srl (vector bool int, vector unsigned short);
14874 vector bool int vec_srl (vector bool int, vector unsigned char);
14875 vector signed short vec_srl (vector signed short, vector unsigned int);
14876 vector signed short vec_srl (vector signed short,
14877 vector unsigned short);
14878 vector signed short vec_srl (vector signed short, vector unsigned char);
14879 vector unsigned short vec_srl (vector unsigned short,
14880 vector unsigned int);
14881 vector unsigned short vec_srl (vector unsigned short,
14882 vector unsigned short);
14883 vector unsigned short vec_srl (vector unsigned short,
14884 vector unsigned char);
14885 vector bool short vec_srl (vector bool short, vector unsigned int);
14886 vector bool short vec_srl (vector bool short, vector unsigned short);
14887 vector bool short vec_srl (vector bool short, vector unsigned char);
14888 vector pixel vec_srl (vector pixel, vector unsigned int);
14889 vector pixel vec_srl (vector pixel, vector unsigned short);
14890 vector pixel vec_srl (vector pixel, vector unsigned char);
14891 vector signed char vec_srl (vector signed char, vector unsigned int);
14892 vector signed char vec_srl (vector signed char, vector unsigned short);
14893 vector signed char vec_srl (vector signed char, vector unsigned char);
14894 vector unsigned char vec_srl (vector unsigned char,
14895 vector unsigned int);
14896 vector unsigned char vec_srl (vector unsigned char,
14897 vector unsigned short);
14898 vector unsigned char vec_srl (vector unsigned char,
14899 vector unsigned char);
14900 vector bool char vec_srl (vector bool char, vector unsigned int);
14901 vector bool char vec_srl (vector bool char, vector unsigned short);
14902 vector bool char vec_srl (vector bool char, vector unsigned char);
14904 vector float vec_sro (vector float, vector signed char);
14905 vector float vec_sro (vector float, vector unsigned char);
14906 vector signed int vec_sro (vector signed int, vector signed char);
14907 vector signed int vec_sro (vector signed int, vector unsigned char);
14908 vector unsigned int vec_sro (vector unsigned int, vector signed char);
14909 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
14910 vector signed short vec_sro (vector signed short, vector signed char);
14911 vector signed short vec_sro (vector signed short, vector unsigned char);
14912 vector unsigned short vec_sro (vector unsigned short,
14913 vector signed char);
14914 vector unsigned short vec_sro (vector unsigned short,
14915 vector unsigned char);
14916 vector pixel vec_sro (vector pixel, vector signed char);
14917 vector pixel vec_sro (vector pixel, vector unsigned char);
14918 vector signed char vec_sro (vector signed char, vector signed char);
14919 vector signed char vec_sro (vector signed char, vector unsigned char);
14920 vector unsigned char vec_sro (vector unsigned char, vector signed char);
14921 vector unsigned char vec_sro (vector unsigned char,
14922 vector unsigned char);
14924 void vec_st (vector float, int, vector float *);
14925 void vec_st (vector float, int, float *);
14926 void vec_st (vector signed int, int, vector signed int *);
14927 void vec_st (vector signed int, int, int *);
14928 void vec_st (vector unsigned int, int, vector unsigned int *);
14929 void vec_st (vector unsigned int, int, unsigned int *);
14930 void vec_st (vector bool int, int, vector bool int *);
14931 void vec_st (vector bool int, int, unsigned int *);
14932 void vec_st (vector bool int, int, int *);
14933 void vec_st (vector signed short, int, vector signed short *);
14934 void vec_st (vector signed short, int, short *);
14935 void vec_st (vector unsigned short, int, vector unsigned short *);
14936 void vec_st (vector unsigned short, int, unsigned short *);
14937 void vec_st (vector bool short, int, vector bool short *);
14938 void vec_st (vector bool short, int, unsigned short *);
14939 void vec_st (vector pixel, int, vector pixel *);
14940 void vec_st (vector pixel, int, unsigned short *);
14941 void vec_st (vector pixel, int, short *);
14942 void vec_st (vector bool short, int, short *);
14943 void vec_st (vector signed char, int, vector signed char *);
14944 void vec_st (vector signed char, int, signed char *);
14945 void vec_st (vector unsigned char, int, vector unsigned char *);
14946 void vec_st (vector unsigned char, int, unsigned char *);
14947 void vec_st (vector bool char, int, vector bool char *);
14948 void vec_st (vector bool char, int, unsigned char *);
14949 void vec_st (vector bool char, int, signed char *);
14951 void vec_ste (vector signed char, int, signed char *);
14952 void vec_ste (vector unsigned char, int, unsigned char *);
14953 void vec_ste (vector bool char, int, signed char *);
14954 void vec_ste (vector bool char, int, unsigned char *);
14955 void vec_ste (vector signed short, int, short *);
14956 void vec_ste (vector unsigned short, int, unsigned short *);
14957 void vec_ste (vector bool short, int, short *);
14958 void vec_ste (vector bool short, int, unsigned short *);
14959 void vec_ste (vector pixel, int, short *);
14960 void vec_ste (vector pixel, int, unsigned short *);
14961 void vec_ste (vector float, int, float *);
14962 void vec_ste (vector signed int, int, int *);
14963 void vec_ste (vector unsigned int, int, unsigned int *);
14964 void vec_ste (vector bool int, int, int *);
14965 void vec_ste (vector bool int, int, unsigned int *);
14967 void vec_stvewx (vector float, int, float *);
14968 void vec_stvewx (vector signed int, int, int *);
14969 void vec_stvewx (vector unsigned int, int, unsigned int *);
14970 void vec_stvewx (vector bool int, int, int *);
14971 void vec_stvewx (vector bool int, int, unsigned int *);
14973 void vec_stvehx (vector signed short, int, short *);
14974 void vec_stvehx (vector unsigned short, int, unsigned short *);
14975 void vec_stvehx (vector bool short, int, short *);
14976 void vec_stvehx (vector bool short, int, unsigned short *);
14977 void vec_stvehx (vector pixel, int, short *);
14978 void vec_stvehx (vector pixel, int, unsigned short *);
14980 void vec_stvebx (vector signed char, int, signed char *);
14981 void vec_stvebx (vector unsigned char, int, unsigned char *);
14982 void vec_stvebx (vector bool char, int, signed char *);
14983 void vec_stvebx (vector bool char, int, unsigned char *);
14985 void vec_stl (vector float, int, vector float *);
14986 void vec_stl (vector float, int, float *);
14987 void vec_stl (vector signed int, int, vector signed int *);
14988 void vec_stl (vector signed int, int, int *);
14989 void vec_stl (vector unsigned int, int, vector unsigned int *);
14990 void vec_stl (vector unsigned int, int, unsigned int *);
14991 void vec_stl (vector bool int, int, vector bool int *);
14992 void vec_stl (vector bool int, int, unsigned int *);
14993 void vec_stl (vector bool int, int, int *);
14994 void vec_stl (vector signed short, int, vector signed short *);
14995 void vec_stl (vector signed short, int, short *);
14996 void vec_stl (vector unsigned short, int, vector unsigned short *);
14997 void vec_stl (vector unsigned short, int, unsigned short *);
14998 void vec_stl (vector bool short, int, vector bool short *);
14999 void vec_stl (vector bool short, int, unsigned short *);
15000 void vec_stl (vector bool short, int, short *);
15001 void vec_stl (vector pixel, int, vector pixel *);
15002 void vec_stl (vector pixel, int, unsigned short *);
15003 void vec_stl (vector pixel, int, short *);
15004 void vec_stl (vector signed char, int, vector signed char *);
15005 void vec_stl (vector signed char, int, signed char *);
15006 void vec_stl (vector unsigned char, int, vector unsigned char *);
15007 void vec_stl (vector unsigned char, int, unsigned char *);
15008 void vec_stl (vector bool char, int, vector bool char *);
15009 void vec_stl (vector bool char, int, unsigned char *);
15010 void vec_stl (vector bool char, int, signed char *);
15012 vector signed char vec_sub (vector bool char, vector signed char);
15013 vector signed char vec_sub (vector signed char, vector bool char);
15014 vector signed char vec_sub (vector signed char, vector signed char);
15015 vector unsigned char vec_sub (vector bool char, vector unsigned char);
15016 vector unsigned char vec_sub (vector unsigned char, vector bool char);
15017 vector unsigned char vec_sub (vector unsigned char,
15018 vector unsigned char);
15019 vector signed short vec_sub (vector bool short, vector signed short);
15020 vector signed short vec_sub (vector signed short, vector bool short);
15021 vector signed short vec_sub (vector signed short, vector signed short);
15022 vector unsigned short vec_sub (vector bool short,
15023 vector unsigned short);
15024 vector unsigned short vec_sub (vector unsigned short,
15025 vector bool short);
15026 vector unsigned short vec_sub (vector unsigned short,
15027 vector unsigned short);
15028 vector signed int vec_sub (vector bool int, vector signed int);
15029 vector signed int vec_sub (vector signed int, vector bool int);
15030 vector signed int vec_sub (vector signed int, vector signed int);
15031 vector unsigned int vec_sub (vector bool int, vector unsigned int);
15032 vector unsigned int vec_sub (vector unsigned int, vector bool int);
15033 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
15034 vector float vec_sub (vector float, vector float);
15036 vector float vec_vsubfp (vector float, vector float);
15038 vector signed int vec_vsubuwm (vector bool int, vector signed int);
15039 vector signed int vec_vsubuwm (vector signed int, vector bool int);
15040 vector signed int vec_vsubuwm (vector signed int, vector signed int);
15041 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
15042 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
15043 vector unsigned int vec_vsubuwm (vector unsigned int,
15044 vector unsigned int);
15046 vector signed short vec_vsubuhm (vector bool short,
15047 vector signed short);
15048 vector signed short vec_vsubuhm (vector signed short,
15049 vector bool short);
15050 vector signed short vec_vsubuhm (vector signed short,
15051 vector signed short);
15052 vector unsigned short vec_vsubuhm (vector bool short,
15053 vector unsigned short);
15054 vector unsigned short vec_vsubuhm (vector unsigned short,
15055 vector bool short);
15056 vector unsigned short vec_vsubuhm (vector unsigned short,
15057 vector unsigned short);
15059 vector signed char vec_vsububm (vector bool char, vector signed char);
15060 vector signed char vec_vsububm (vector signed char, vector bool char);
15061 vector signed char vec_vsububm (vector signed char, vector signed char);
15062 vector unsigned char vec_vsububm (vector bool char,
15063 vector unsigned char);
15064 vector unsigned char vec_vsububm (vector unsigned char,
15066 vector unsigned char vec_vsububm (vector unsigned char,
15067 vector unsigned char);
15069 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
15071 vector unsigned char vec_subs (vector bool char, vector unsigned char);
15072 vector unsigned char vec_subs (vector unsigned char, vector bool char);
15073 vector unsigned char vec_subs (vector unsigned char,
15074 vector unsigned char);
15075 vector signed char vec_subs (vector bool char, vector signed char);
15076 vector signed char vec_subs (vector signed char, vector bool char);
15077 vector signed char vec_subs (vector signed char, vector signed char);
15078 vector unsigned short vec_subs (vector bool short,
15079 vector unsigned short);
15080 vector unsigned short vec_subs (vector unsigned short,
15081 vector bool short);
15082 vector unsigned short vec_subs (vector unsigned short,
15083 vector unsigned short);
15084 vector signed short vec_subs (vector bool short, vector signed short);
15085 vector signed short vec_subs (vector signed short, vector bool short);
15086 vector signed short vec_subs (vector signed short, vector signed short);
15087 vector unsigned int vec_subs (vector bool int, vector unsigned int);
15088 vector unsigned int vec_subs (vector unsigned int, vector bool int);
15089 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
15090 vector signed int vec_subs (vector bool int, vector signed int);
15091 vector signed int vec_subs (vector signed int, vector bool int);
15092 vector signed int vec_subs (vector signed int, vector signed int);
15094 vector signed int vec_vsubsws (vector bool int, vector signed int);
15095 vector signed int vec_vsubsws (vector signed int, vector bool int);
15096 vector signed int vec_vsubsws (vector signed int, vector signed int);
15098 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
15099 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
15100 vector unsigned int vec_vsubuws (vector unsigned int,
15101 vector unsigned int);
15103 vector signed short vec_vsubshs (vector bool short,
15104 vector signed short);
15105 vector signed short vec_vsubshs (vector signed short,
15106 vector bool short);
15107 vector signed short vec_vsubshs (vector signed short,
15108 vector signed short);
15110 vector unsigned short vec_vsubuhs (vector bool short,
15111 vector unsigned short);
15112 vector unsigned short vec_vsubuhs (vector unsigned short,
15113 vector bool short);
15114 vector unsigned short vec_vsubuhs (vector unsigned short,
15115 vector unsigned short);
15117 vector signed char vec_vsubsbs (vector bool char, vector signed char);
15118 vector signed char vec_vsubsbs (vector signed char, vector bool char);
15119 vector signed char vec_vsubsbs (vector signed char, vector signed char);
15121 vector unsigned char vec_vsububs (vector bool char,
15122 vector unsigned char);
15123 vector unsigned char vec_vsububs (vector unsigned char,
15125 vector unsigned char vec_vsububs (vector unsigned char,
15126 vector unsigned char);
15128 vector unsigned int vec_sum4s (vector unsigned char,
15129 vector unsigned int);
15130 vector signed int vec_sum4s (vector signed char, vector signed int);
15131 vector signed int vec_sum4s (vector signed short, vector signed int);
15133 vector signed int vec_vsum4shs (vector signed short, vector signed int);
15135 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
15137 vector unsigned int vec_vsum4ubs (vector unsigned char,
15138 vector unsigned int);
15140 vector signed int vec_sum2s (vector signed int, vector signed int);
15142 vector signed int vec_sums (vector signed int, vector signed int);
15144 vector float vec_trunc (vector float);
15146 vector signed short vec_unpackh (vector signed char);
15147 vector bool short vec_unpackh (vector bool char);
15148 vector signed int vec_unpackh (vector signed short);
15149 vector bool int vec_unpackh (vector bool short);
15150 vector unsigned int vec_unpackh (vector pixel);
15152 vector bool int vec_vupkhsh (vector bool short);
15153 vector signed int vec_vupkhsh (vector signed short);
15155 vector unsigned int vec_vupkhpx (vector pixel);
15157 vector bool short vec_vupkhsb (vector bool char);
15158 vector signed short vec_vupkhsb (vector signed char);
15160 vector signed short vec_unpackl (vector signed char);
15161 vector bool short vec_unpackl (vector bool char);
15162 vector unsigned int vec_unpackl (vector pixel);
15163 vector signed int vec_unpackl (vector signed short);
15164 vector bool int vec_unpackl (vector bool short);
15166 vector unsigned int vec_vupklpx (vector pixel);
15168 vector bool int vec_vupklsh (vector bool short);
15169 vector signed int vec_vupklsh (vector signed short);
15171 vector bool short vec_vupklsb (vector bool char);
15172 vector signed short vec_vupklsb (vector signed char);
15174 vector float vec_xor (vector float, vector float);
15175 vector float vec_xor (vector float, vector bool int);
15176 vector float vec_xor (vector bool int, vector float);
15177 vector bool int vec_xor (vector bool int, vector bool int);
15178 vector signed int vec_xor (vector bool int, vector signed int);
15179 vector signed int vec_xor (vector signed int, vector bool int);
15180 vector signed int vec_xor (vector signed int, vector signed int);
15181 vector unsigned int vec_xor (vector bool int, vector unsigned int);
15182 vector unsigned int vec_xor (vector unsigned int, vector bool int);
15183 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
15184 vector bool short vec_xor (vector bool short, vector bool short);
15185 vector signed short vec_xor (vector bool short, vector signed short);
15186 vector signed short vec_xor (vector signed short, vector bool short);
15187 vector signed short vec_xor (vector signed short, vector signed short);
15188 vector unsigned short vec_xor (vector bool short,
15189 vector unsigned short);
15190 vector unsigned short vec_xor (vector unsigned short,
15191 vector bool short);
15192 vector unsigned short vec_xor (vector unsigned short,
15193 vector unsigned short);
15194 vector signed char vec_xor (vector bool char, vector signed char);
15195 vector bool char vec_xor (vector bool char, vector bool char);
15196 vector signed char vec_xor (vector signed char, vector bool char);
15197 vector signed char vec_xor (vector signed char, vector signed char);
15198 vector unsigned char vec_xor (vector bool char, vector unsigned char);
15199 vector unsigned char vec_xor (vector unsigned char, vector bool char);
15200 vector unsigned char vec_xor (vector unsigned char,
15201 vector unsigned char);
15203 int vec_all_eq (vector signed char, vector bool char);
15204 int vec_all_eq (vector signed char, vector signed char);
15205 int vec_all_eq (vector unsigned char, vector bool char);
15206 int vec_all_eq (vector unsigned char, vector unsigned char);
15207 int vec_all_eq (vector bool char, vector bool char);
15208 int vec_all_eq (vector bool char, vector unsigned char);
15209 int vec_all_eq (vector bool char, vector signed char);
15210 int vec_all_eq (vector signed short, vector bool short);
15211 int vec_all_eq (vector signed short, vector signed short);
15212 int vec_all_eq (vector unsigned short, vector bool short);
15213 int vec_all_eq (vector unsigned short, vector unsigned short);
15214 int vec_all_eq (vector bool short, vector bool short);
15215 int vec_all_eq (vector bool short, vector unsigned short);
15216 int vec_all_eq (vector bool short, vector signed short);
15217 int vec_all_eq (vector pixel, vector pixel);
15218 int vec_all_eq (vector signed int, vector bool int);
15219 int vec_all_eq (vector signed int, vector signed int);
15220 int vec_all_eq (vector unsigned int, vector bool int);
15221 int vec_all_eq (vector unsigned int, vector unsigned int);
15222 int vec_all_eq (vector bool int, vector bool int);
15223 int vec_all_eq (vector bool int, vector unsigned int);
15224 int vec_all_eq (vector bool int, vector signed int);
15225 int vec_all_eq (vector float, vector float);
15227 int vec_all_ge (vector bool char, vector unsigned char);
15228 int vec_all_ge (vector unsigned char, vector bool char);
15229 int vec_all_ge (vector unsigned char, vector unsigned char);
15230 int vec_all_ge (vector bool char, vector signed char);
15231 int vec_all_ge (vector signed char, vector bool char);
15232 int vec_all_ge (vector signed char, vector signed char);
15233 int vec_all_ge (vector bool short, vector unsigned short);
15234 int vec_all_ge (vector unsigned short, vector bool short);
15235 int vec_all_ge (vector unsigned short, vector unsigned short);
15236 int vec_all_ge (vector signed short, vector signed short);
15237 int vec_all_ge (vector bool short, vector signed short);
15238 int vec_all_ge (vector signed short, vector bool short);
15239 int vec_all_ge (vector bool int, vector unsigned int);
15240 int vec_all_ge (vector unsigned int, vector bool int);
15241 int vec_all_ge (vector unsigned int, vector unsigned int);
15242 int vec_all_ge (vector bool int, vector signed int);
15243 int vec_all_ge (vector signed int, vector bool int);
15244 int vec_all_ge (vector signed int, vector signed int);
15245 int vec_all_ge (vector float, vector float);
15247 int vec_all_gt (vector bool char, vector unsigned char);
15248 int vec_all_gt (vector unsigned char, vector bool char);
15249 int vec_all_gt (vector unsigned char, vector unsigned char);
15250 int vec_all_gt (vector bool char, vector signed char);
15251 int vec_all_gt (vector signed char, vector bool char);
15252 int vec_all_gt (vector signed char, vector signed char);
15253 int vec_all_gt (vector bool short, vector unsigned short);
15254 int vec_all_gt (vector unsigned short, vector bool short);
15255 int vec_all_gt (vector unsigned short, vector unsigned short);
15256 int vec_all_gt (vector bool short, vector signed short);
15257 int vec_all_gt (vector signed short, vector bool short);
15258 int vec_all_gt (vector signed short, vector signed short);
15259 int vec_all_gt (vector bool int, vector unsigned int);
15260 int vec_all_gt (vector unsigned int, vector bool int);
15261 int vec_all_gt (vector unsigned int, vector unsigned int);
15262 int vec_all_gt (vector bool int, vector signed int);
15263 int vec_all_gt (vector signed int, vector bool int);
15264 int vec_all_gt (vector signed int, vector signed int);
15265 int vec_all_gt (vector float, vector float);
15267 int vec_all_in (vector float, vector float);
15269 int vec_all_le (vector bool char, vector unsigned char);
15270 int vec_all_le (vector unsigned char, vector bool char);
15271 int vec_all_le (vector unsigned char, vector unsigned char);
15272 int vec_all_le (vector bool char, vector signed char);
15273 int vec_all_le (vector signed char, vector bool char);
15274 int vec_all_le (vector signed char, vector signed char);
15275 int vec_all_le (vector bool short, vector unsigned short);
15276 int vec_all_le (vector unsigned short, vector bool short);
15277 int vec_all_le (vector unsigned short, vector unsigned short);
15278 int vec_all_le (vector bool short, vector signed short);
15279 int vec_all_le (vector signed short, vector bool short);
15280 int vec_all_le (vector signed short, vector signed short);
15281 int vec_all_le (vector bool int, vector unsigned int);
15282 int vec_all_le (vector unsigned int, vector bool int);
15283 int vec_all_le (vector unsigned int, vector unsigned int);
15284 int vec_all_le (vector bool int, vector signed int);
15285 int vec_all_le (vector signed int, vector bool int);
15286 int vec_all_le (vector signed int, vector signed int);
15287 int vec_all_le (vector float, vector float);
15289 int vec_all_lt (vector bool char, vector unsigned char);
15290 int vec_all_lt (vector unsigned char, vector bool char);
15291 int vec_all_lt (vector unsigned char, vector unsigned char);
15292 int vec_all_lt (vector bool char, vector signed char);
15293 int vec_all_lt (vector signed char, vector bool char);
15294 int vec_all_lt (vector signed char, vector signed char);
15295 int vec_all_lt (vector bool short, vector unsigned short);
15296 int vec_all_lt (vector unsigned short, vector bool short);
15297 int vec_all_lt (vector unsigned short, vector unsigned short);
15298 int vec_all_lt (vector bool short, vector signed short);
15299 int vec_all_lt (vector signed short, vector bool short);
15300 int vec_all_lt (vector signed short, vector signed short);
15301 int vec_all_lt (vector bool int, vector unsigned int);
15302 int vec_all_lt (vector unsigned int, vector bool int);
15303 int vec_all_lt (vector unsigned int, vector unsigned int);
15304 int vec_all_lt (vector bool int, vector signed int);
15305 int vec_all_lt (vector signed int, vector bool int);
15306 int vec_all_lt (vector signed int, vector signed int);
15307 int vec_all_lt (vector float, vector float);
15309 int vec_all_nan (vector float);
15311 int vec_all_ne (vector signed char, vector bool char);
15312 int vec_all_ne (vector signed char, vector signed char);
15313 int vec_all_ne (vector unsigned char, vector bool char);
15314 int vec_all_ne (vector unsigned char, vector unsigned char);
15315 int vec_all_ne (vector bool char, vector bool char);
15316 int vec_all_ne (vector bool char, vector unsigned char);
15317 int vec_all_ne (vector bool char, vector signed char);
15318 int vec_all_ne (vector signed short, vector bool short);
15319 int vec_all_ne (vector signed short, vector signed short);
15320 int vec_all_ne (vector unsigned short, vector bool short);
15321 int vec_all_ne (vector unsigned short, vector unsigned short);
15322 int vec_all_ne (vector bool short, vector bool short);
15323 int vec_all_ne (vector bool short, vector unsigned short);
15324 int vec_all_ne (vector bool short, vector signed short);
15325 int vec_all_ne (vector pixel, vector pixel);
15326 int vec_all_ne (vector signed int, vector bool int);
15327 int vec_all_ne (vector signed int, vector signed int);
15328 int vec_all_ne (vector unsigned int, vector bool int);
15329 int vec_all_ne (vector unsigned int, vector unsigned int);
15330 int vec_all_ne (vector bool int, vector bool int);
15331 int vec_all_ne (vector bool int, vector unsigned int);
15332 int vec_all_ne (vector bool int, vector signed int);
15333 int vec_all_ne (vector float, vector float);
15335 int vec_all_nge (vector float, vector float);
15337 int vec_all_ngt (vector float, vector float);
15339 int vec_all_nle (vector float, vector float);
15341 int vec_all_nlt (vector float, vector float);
15343 int vec_all_numeric (vector float);
15345 int vec_any_eq (vector signed char, vector bool char);
15346 int vec_any_eq (vector signed char, vector signed char);
15347 int vec_any_eq (vector unsigned char, vector bool char);
15348 int vec_any_eq (vector unsigned char, vector unsigned char);
15349 int vec_any_eq (vector bool char, vector bool char);
15350 int vec_any_eq (vector bool char, vector unsigned char);
15351 int vec_any_eq (vector bool char, vector signed char);
15352 int vec_any_eq (vector signed short, vector bool short);
15353 int vec_any_eq (vector signed short, vector signed short);
15354 int vec_any_eq (vector unsigned short, vector bool short);
15355 int vec_any_eq (vector unsigned short, vector unsigned short);
15356 int vec_any_eq (vector bool short, vector bool short);
15357 int vec_any_eq (vector bool short, vector unsigned short);
15358 int vec_any_eq (vector bool short, vector signed short);
15359 int vec_any_eq (vector pixel, vector pixel);
15360 int vec_any_eq (vector signed int, vector bool int);
15361 int vec_any_eq (vector signed int, vector signed int);
15362 int vec_any_eq (vector unsigned int, vector bool int);
15363 int vec_any_eq (vector unsigned int, vector unsigned int);
15364 int vec_any_eq (vector bool int, vector bool int);
15365 int vec_any_eq (vector bool int, vector unsigned int);
15366 int vec_any_eq (vector bool int, vector signed int);
15367 int vec_any_eq (vector float, vector float);
15369 int vec_any_ge (vector signed char, vector bool char);
15370 int vec_any_ge (vector unsigned char, vector bool char);
15371 int vec_any_ge (vector unsigned char, vector unsigned char);
15372 int vec_any_ge (vector signed char, vector signed char);
15373 int vec_any_ge (vector bool char, vector unsigned char);
15374 int vec_any_ge (vector bool char, vector signed char);
15375 int vec_any_ge (vector unsigned short, vector bool short);
15376 int vec_any_ge (vector unsigned short, vector unsigned short);
15377 int vec_any_ge (vector signed short, vector signed short);
15378 int vec_any_ge (vector signed short, vector bool short);
15379 int vec_any_ge (vector bool short, vector unsigned short);
15380 int vec_any_ge (vector bool short, vector signed short);
15381 int vec_any_ge (vector signed int, vector bool int);
15382 int vec_any_ge (vector unsigned int, vector bool int);
15383 int vec_any_ge (vector unsigned int, vector unsigned int);
15384 int vec_any_ge (vector signed int, vector signed int);
15385 int vec_any_ge (vector bool int, vector unsigned int);
15386 int vec_any_ge (vector bool int, vector signed int);
15387 int vec_any_ge (vector float, vector float);
15389 int vec_any_gt (vector bool char, vector unsigned char);
15390 int vec_any_gt (vector unsigned char, vector bool char);
15391 int vec_any_gt (vector unsigned char, vector unsigned char);
15392 int vec_any_gt (vector bool char, vector signed char);
15393 int vec_any_gt (vector signed char, vector bool char);
15394 int vec_any_gt (vector signed char, vector signed char);
15395 int vec_any_gt (vector bool short, vector unsigned short);
15396 int vec_any_gt (vector unsigned short, vector bool short);
15397 int vec_any_gt (vector unsigned short, vector unsigned short);
15398 int vec_any_gt (vector bool short, vector signed short);
15399 int vec_any_gt (vector signed short, vector bool short);
15400 int vec_any_gt (vector signed short, vector signed short);
15401 int vec_any_gt (vector bool int, vector unsigned int);
15402 int vec_any_gt (vector unsigned int, vector bool int);
15403 int vec_any_gt (vector unsigned int, vector unsigned int);
15404 int vec_any_gt (vector bool int, vector signed int);
15405 int vec_any_gt (vector signed int, vector bool int);
15406 int vec_any_gt (vector signed int, vector signed int);
15407 int vec_any_gt (vector float, vector float);
15409 int vec_any_le (vector bool char, vector unsigned char);
15410 int vec_any_le (vector unsigned char, vector bool char);
15411 int vec_any_le (vector unsigned char, vector unsigned char);
15412 int vec_any_le (vector bool char, vector signed char);
15413 int vec_any_le (vector signed char, vector bool char);
15414 int vec_any_le (vector signed char, vector signed char);
15415 int vec_any_le (vector bool short, vector unsigned short);
15416 int vec_any_le (vector unsigned short, vector bool short);
15417 int vec_any_le (vector unsigned short, vector unsigned short);
15418 int vec_any_le (vector bool short, vector signed short);
15419 int vec_any_le (vector signed short, vector bool short);
15420 int vec_any_le (vector signed short, vector signed short);
15421 int vec_any_le (vector bool int, vector unsigned int);
15422 int vec_any_le (vector unsigned int, vector bool int);
15423 int vec_any_le (vector unsigned int, vector unsigned int);
15424 int vec_any_le (vector bool int, vector signed int);
15425 int vec_any_le (vector signed int, vector bool int);
15426 int vec_any_le (vector signed int, vector signed int);
15427 int vec_any_le (vector float, vector float);
15429 int vec_any_lt (vector bool char, vector unsigned char);
15430 int vec_any_lt (vector unsigned char, vector bool char);
15431 int vec_any_lt (vector unsigned char, vector unsigned char);
15432 int vec_any_lt (vector bool char, vector signed char);
15433 int vec_any_lt (vector signed char, vector bool char);
15434 int vec_any_lt (vector signed char, vector signed char);
15435 int vec_any_lt (vector bool short, vector unsigned short);
15436 int vec_any_lt (vector unsigned short, vector bool short);
15437 int vec_any_lt (vector unsigned short, vector unsigned short);
15438 int vec_any_lt (vector bool short, vector signed short);
15439 int vec_any_lt (vector signed short, vector bool short);
15440 int vec_any_lt (vector signed short, vector signed short);
15441 int vec_any_lt (vector bool int, vector unsigned int);
15442 int vec_any_lt (vector unsigned int, vector bool int);
15443 int vec_any_lt (vector unsigned int, vector unsigned int);
15444 int vec_any_lt (vector bool int, vector signed int);
15445 int vec_any_lt (vector signed int, vector bool int);
15446 int vec_any_lt (vector signed int, vector signed int);
15447 int vec_any_lt (vector float, vector float);
15449 int vec_any_nan (vector float);
15451 int vec_any_ne (vector signed char, vector bool char);
15452 int vec_any_ne (vector signed char, vector signed char);
15453 int vec_any_ne (vector unsigned char, vector bool char);
15454 int vec_any_ne (vector unsigned char, vector unsigned char);
15455 int vec_any_ne (vector bool char, vector bool char);
15456 int vec_any_ne (vector bool char, vector unsigned char);
15457 int vec_any_ne (vector bool char, vector signed char);
15458 int vec_any_ne (vector signed short, vector bool short);
15459 int vec_any_ne (vector signed short, vector signed short);
15460 int vec_any_ne (vector unsigned short, vector bool short);
15461 int vec_any_ne (vector unsigned short, vector unsigned short);
15462 int vec_any_ne (vector bool short, vector bool short);
15463 int vec_any_ne (vector bool short, vector unsigned short);
15464 int vec_any_ne (vector bool short, vector signed short);
15465 int vec_any_ne (vector pixel, vector pixel);
15466 int vec_any_ne (vector signed int, vector bool int);
15467 int vec_any_ne (vector signed int, vector signed int);
15468 int vec_any_ne (vector unsigned int, vector bool int);
15469 int vec_any_ne (vector unsigned int, vector unsigned int);
15470 int vec_any_ne (vector bool int, vector bool int);
15471 int vec_any_ne (vector bool int, vector unsigned int);
15472 int vec_any_ne (vector bool int, vector signed int);
15473 int vec_any_ne (vector float, vector float);
15475 int vec_any_nge (vector float, vector float);
15477 int vec_any_ngt (vector float, vector float);
15479 int vec_any_nle (vector float, vector float);
15481 int vec_any_nlt (vector float, vector float);
15483 int vec_any_numeric (vector float);
15485 int vec_any_out (vector float, vector float);
15488 If the vector/scalar (VSX) instruction set is available, the following
15489 additional functions are available:
15492 vector double vec_abs (vector double);
15493 vector double vec_add (vector double, vector double);
15494 vector double vec_and (vector double, vector double);
15495 vector double vec_and (vector double, vector bool long);
15496 vector double vec_and (vector bool long, vector double);
15497 vector long vec_and (vector long, vector long);
15498 vector long vec_and (vector long, vector bool long);
15499 vector long vec_and (vector bool long, vector long);
15500 vector unsigned long vec_and (vector unsigned long, vector unsigned long);
15501 vector unsigned long vec_and (vector unsigned long, vector bool long);
15502 vector unsigned long vec_and (vector bool long, vector unsigned long);
15503 vector double vec_andc (vector double, vector double);
15504 vector double vec_andc (vector double, vector bool long);
15505 vector double vec_andc (vector bool long, vector double);
15506 vector long vec_andc (vector long, vector long);
15507 vector long vec_andc (vector long, vector bool long);
15508 vector long vec_andc (vector bool long, vector long);
15509 vector unsigned long vec_andc (vector unsigned long, vector unsigned long);
15510 vector unsigned long vec_andc (vector unsigned long, vector bool long);
15511 vector unsigned long vec_andc (vector bool long, vector unsigned long);
15512 vector double vec_ceil (vector double);
15513 vector bool long vec_cmpeq (vector double, vector double);
15514 vector bool long vec_cmpge (vector double, vector double);
15515 vector bool long vec_cmpgt (vector double, vector double);
15516 vector bool long vec_cmple (vector double, vector double);
15517 vector bool long vec_cmplt (vector double, vector double);
15518 vector double vec_cpsgn (vector double, vector double);
15519 vector float vec_div (vector float, vector float);
15520 vector double vec_div (vector double, vector double);
15521 vector long vec_div (vector long, vector long);
15522 vector unsigned long vec_div (vector unsigned long, vector unsigned long);
15523 vector double vec_floor (vector double);
15524 vector double vec_ld (int, const vector double *);
15525 vector double vec_ld (int, const double *);
15526 vector double vec_ldl (int, const vector double *);
15527 vector double vec_ldl (int, const double *);
15528 vector unsigned char vec_lvsl (int, const volatile double *);
15529 vector unsigned char vec_lvsr (int, const volatile double *);
15530 vector double vec_madd (vector double, vector double, vector double);
15531 vector double vec_max (vector double, vector double);
15532 vector signed long vec_mergeh (vector signed long, vector signed long);
15533 vector signed long vec_mergeh (vector signed long, vector bool long);
15534 vector signed long vec_mergeh (vector bool long, vector signed long);
15535 vector unsigned long vec_mergeh (vector unsigned long, vector unsigned long);
15536 vector unsigned long vec_mergeh (vector unsigned long, vector bool long);
15537 vector unsigned long vec_mergeh (vector bool long, vector unsigned long);
15538 vector signed long vec_mergel (vector signed long, vector signed long);
15539 vector signed long vec_mergel (vector signed long, vector bool long);
15540 vector signed long vec_mergel (vector bool long, vector signed long);
15541 vector unsigned long vec_mergel (vector unsigned long, vector unsigned long);
15542 vector unsigned long vec_mergel (vector unsigned long, vector bool long);
15543 vector unsigned long vec_mergel (vector bool long, vector unsigned long);
15544 vector double vec_min (vector double, vector double);
15545 vector float vec_msub (vector float, vector float, vector float);
15546 vector double vec_msub (vector double, vector double, vector double);
15547 vector float vec_mul (vector float, vector float);
15548 vector double vec_mul (vector double, vector double);
15549 vector long vec_mul (vector long, vector long);
15550 vector unsigned long vec_mul (vector unsigned long, vector unsigned long);
15551 vector float vec_nearbyint (vector float);
15552 vector double vec_nearbyint (vector double);
15553 vector float vec_nmadd (vector float, vector float, vector float);
15554 vector double vec_nmadd (vector double, vector double, vector double);
15555 vector double vec_nmsub (vector double, vector double, vector double);
15556 vector double vec_nor (vector double, vector double);
15557 vector long vec_nor (vector long, vector long);
15558 vector long vec_nor (vector long, vector bool long);
15559 vector long vec_nor (vector bool long, vector long);
15560 vector unsigned long vec_nor (vector unsigned long, vector unsigned long);
15561 vector unsigned long vec_nor (vector unsigned long, vector bool long);
15562 vector unsigned long vec_nor (vector bool long, vector unsigned long);
15563 vector double vec_or (vector double, vector double);
15564 vector double vec_or (vector double, vector bool long);
15565 vector double vec_or (vector bool long, vector double);
15566 vector long vec_or (vector long, vector long);
15567 vector long vec_or (vector long, vector bool long);
15568 vector long vec_or (vector bool long, vector long);
15569 vector unsigned long vec_or (vector unsigned long, vector unsigned long);
15570 vector unsigned long vec_or (vector unsigned long, vector bool long);
15571 vector unsigned long vec_or (vector bool long, vector unsigned long);
15572 vector double vec_perm (vector double, vector double, vector unsigned char);
15573 vector long vec_perm (vector long, vector long, vector unsigned char);
15574 vector unsigned long vec_perm (vector unsigned long, vector unsigned long,
15575 vector unsigned char);
15576 vector double vec_rint (vector double);
15577 vector double vec_recip (vector double, vector double);
15578 vector double vec_rsqrt (vector double);
15579 vector double vec_rsqrte (vector double);
15580 vector double vec_sel (vector double, vector double, vector bool long);
15581 vector double vec_sel (vector double, vector double, vector unsigned long);
15582 vector long vec_sel (vector long, vector long, vector long);
15583 vector long vec_sel (vector long, vector long, vector unsigned long);
15584 vector long vec_sel (vector long, vector long, vector bool long);
15585 vector unsigned long vec_sel (vector unsigned long, vector unsigned long,
15587 vector unsigned long vec_sel (vector unsigned long, vector unsigned long,
15588 vector unsigned long);
15589 vector unsigned long vec_sel (vector unsigned long, vector unsigned long,
15591 vector double vec_splats (double);
15592 vector signed long vec_splats (signed long);
15593 vector unsigned long vec_splats (unsigned long);
15594 vector float vec_sqrt (vector float);
15595 vector double vec_sqrt (vector double);
15596 void vec_st (vector double, int, vector double *);
15597 void vec_st (vector double, int, double *);
15598 vector double vec_sub (vector double, vector double);
15599 vector double vec_trunc (vector double);
15600 vector double vec_xor (vector double, vector double);
15601 vector double vec_xor (vector double, vector bool long);
15602 vector double vec_xor (vector bool long, vector double);
15603 vector long vec_xor (vector long, vector long);
15604 vector long vec_xor (vector long, vector bool long);
15605 vector long vec_xor (vector bool long, vector long);
15606 vector unsigned long vec_xor (vector unsigned long, vector unsigned long);
15607 vector unsigned long vec_xor (vector unsigned long, vector bool long);
15608 vector unsigned long vec_xor (vector bool long, vector unsigned long);
15609 int vec_all_eq (vector double, vector double);
15610 int vec_all_ge (vector double, vector double);
15611 int vec_all_gt (vector double, vector double);
15612 int vec_all_le (vector double, vector double);
15613 int vec_all_lt (vector double, vector double);
15614 int vec_all_nan (vector double);
15615 int vec_all_ne (vector double, vector double);
15616 int vec_all_nge (vector double, vector double);
15617 int vec_all_ngt (vector double, vector double);
15618 int vec_all_nle (vector double, vector double);
15619 int vec_all_nlt (vector double, vector double);
15620 int vec_all_numeric (vector double);
15621 int vec_any_eq (vector double, vector double);
15622 int vec_any_ge (vector double, vector double);
15623 int vec_any_gt (vector double, vector double);
15624 int vec_any_le (vector double, vector double);
15625 int vec_any_lt (vector double, vector double);
15626 int vec_any_nan (vector double);
15627 int vec_any_ne (vector double, vector double);
15628 int vec_any_nge (vector double, vector double);
15629 int vec_any_ngt (vector double, vector double);
15630 int vec_any_nle (vector double, vector double);
15631 int vec_any_nlt (vector double, vector double);
15632 int vec_any_numeric (vector double);
15634 vector double vec_vsx_ld (int, const vector double *);
15635 vector double vec_vsx_ld (int, const double *);
15636 vector float vec_vsx_ld (int, const vector float *);
15637 vector float vec_vsx_ld (int, const float *);
15638 vector bool int vec_vsx_ld (int, const vector bool int *);
15639 vector signed int vec_vsx_ld (int, const vector signed int *);
15640 vector signed int vec_vsx_ld (int, const int *);
15641 vector signed int vec_vsx_ld (int, const long *);
15642 vector unsigned int vec_vsx_ld (int, const vector unsigned int *);
15643 vector unsigned int vec_vsx_ld (int, const unsigned int *);
15644 vector unsigned int vec_vsx_ld (int, const unsigned long *);
15645 vector bool short vec_vsx_ld (int, const vector bool short *);
15646 vector pixel vec_vsx_ld (int, const vector pixel *);
15647 vector signed short vec_vsx_ld (int, const vector signed short *);
15648 vector signed short vec_vsx_ld (int, const short *);
15649 vector unsigned short vec_vsx_ld (int, const vector unsigned short *);
15650 vector unsigned short vec_vsx_ld (int, const unsigned short *);
15651 vector bool char vec_vsx_ld (int, const vector bool char *);
15652 vector signed char vec_vsx_ld (int, const vector signed char *);
15653 vector signed char vec_vsx_ld (int, const signed char *);
15654 vector unsigned char vec_vsx_ld (int, const vector unsigned char *);
15655 vector unsigned char vec_vsx_ld (int, const unsigned char *);
15657 void vec_vsx_st (vector double, int, vector double *);
15658 void vec_vsx_st (vector double, int, double *);
15659 void vec_vsx_st (vector float, int, vector float *);
15660 void vec_vsx_st (vector float, int, float *);
15661 void vec_vsx_st (vector signed int, int, vector signed int *);
15662 void vec_vsx_st (vector signed int, int, int *);
15663 void vec_vsx_st (vector unsigned int, int, vector unsigned int *);
15664 void vec_vsx_st (vector unsigned int, int, unsigned int *);
15665 void vec_vsx_st (vector bool int, int, vector bool int *);
15666 void vec_vsx_st (vector bool int, int, unsigned int *);
15667 void vec_vsx_st (vector bool int, int, int *);
15668 void vec_vsx_st (vector signed short, int, vector signed short *);
15669 void vec_vsx_st (vector signed short, int, short *);
15670 void vec_vsx_st (vector unsigned short, int, vector unsigned short *);
15671 void vec_vsx_st (vector unsigned short, int, unsigned short *);
15672 void vec_vsx_st (vector bool short, int, vector bool short *);
15673 void vec_vsx_st (vector bool short, int, unsigned short *);
15674 void vec_vsx_st (vector pixel, int, vector pixel *);
15675 void vec_vsx_st (vector pixel, int, unsigned short *);
15676 void vec_vsx_st (vector pixel, int, short *);
15677 void vec_vsx_st (vector bool short, int, short *);
15678 void vec_vsx_st (vector signed char, int, vector signed char *);
15679 void vec_vsx_st (vector signed char, int, signed char *);
15680 void vec_vsx_st (vector unsigned char, int, vector unsigned char *);
15681 void vec_vsx_st (vector unsigned char, int, unsigned char *);
15682 void vec_vsx_st (vector bool char, int, vector bool char *);
15683 void vec_vsx_st (vector bool char, int, unsigned char *);
15684 void vec_vsx_st (vector bool char, int, signed char *);
15686 vector double vec_xxpermdi (vector double, vector double, int);
15687 vector float vec_xxpermdi (vector float, vector float, int);
15688 vector long long vec_xxpermdi (vector long long, vector long long, int);
15689 vector unsigned long long vec_xxpermdi (vector unsigned long long,
15690 vector unsigned long long, int);
15691 vector int vec_xxpermdi (vector int, vector int, int);
15692 vector unsigned int vec_xxpermdi (vector unsigned int,
15693 vector unsigned int, int);
15694 vector short vec_xxpermdi (vector short, vector short, int);
15695 vector unsigned short vec_xxpermdi (vector unsigned short,
15696 vector unsigned short, int);
15697 vector signed char vec_xxpermdi (vector signed char, vector signed char, int);
15698 vector unsigned char vec_xxpermdi (vector unsigned char,
15699 vector unsigned char, int);
15701 vector double vec_xxsldi (vector double, vector double, int);
15702 vector float vec_xxsldi (vector float, vector float, int);
15703 vector long long vec_xxsldi (vector long long, vector long long, int);
15704 vector unsigned long long vec_xxsldi (vector unsigned long long,
15705 vector unsigned long long, int);
15706 vector int vec_xxsldi (vector int, vector int, int);
15707 vector unsigned int vec_xxsldi (vector unsigned int, vector unsigned int, int);
15708 vector short vec_xxsldi (vector short, vector short, int);
15709 vector unsigned short vec_xxsldi (vector unsigned short,
15710 vector unsigned short, int);
15711 vector signed char vec_xxsldi (vector signed char, vector signed char, int);
15712 vector unsigned char vec_xxsldi (vector unsigned char,
15713 vector unsigned char, int);
15716 Note that the @samp{vec_ld} and @samp{vec_st} built-in functions always
15717 generate the AltiVec @samp{LVX} and @samp{STVX} instructions even
15718 if the VSX instruction set is available. The @samp{vec_vsx_ld} and
15719 @samp{vec_vsx_st} built-in functions always generate the VSX @samp{LXVD2X},
15720 @samp{LXVW4X}, @samp{STXVD2X}, and @samp{STXVW4X} instructions.
15722 If the ISA 2.07 additions to the vector/scalar (power8-vector)
15723 instruction set is available, the following additional functions are
15724 available for both 32-bit and 64-bit targets. For 64-bit targets, you
15725 can use @var{vector long} instead of @var{vector long long},
15726 @var{vector bool long} instead of @var{vector bool long long}, and
15727 @var{vector unsigned long} instead of @var{vector unsigned long long}.
15730 vector long long vec_abs (vector long long);
15732 vector long long vec_add (vector long long, vector long long);
15733 vector unsigned long long vec_add (vector unsigned long long,
15734 vector unsigned long long);
15736 int vec_all_eq (vector long long, vector long long);
15737 int vec_all_eq (vector unsigned long long, vector unsigned long long);
15738 int vec_all_ge (vector long long, vector long long);
15739 int vec_all_ge (vector unsigned long long, vector unsigned long long);
15740 int vec_all_gt (vector long long, vector long long);
15741 int vec_all_gt (vector unsigned long long, vector unsigned long long);
15742 int vec_all_le (vector long long, vector long long);
15743 int vec_all_le (vector unsigned long long, vector unsigned long long);
15744 int vec_all_lt (vector long long, vector long long);
15745 int vec_all_lt (vector unsigned long long, vector unsigned long long);
15746 int vec_all_ne (vector long long, vector long long);
15747 int vec_all_ne (vector unsigned long long, vector unsigned long long);
15749 int vec_any_eq (vector long long, vector long long);
15750 int vec_any_eq (vector unsigned long long, vector unsigned long long);
15751 int vec_any_ge (vector long long, vector long long);
15752 int vec_any_ge (vector unsigned long long, vector unsigned long long);
15753 int vec_any_gt (vector long long, vector long long);
15754 int vec_any_gt (vector unsigned long long, vector unsigned long long);
15755 int vec_any_le (vector long long, vector long long);
15756 int vec_any_le (vector unsigned long long, vector unsigned long long);
15757 int vec_any_lt (vector long long, vector long long);
15758 int vec_any_lt (vector unsigned long long, vector unsigned long long);
15759 int vec_any_ne (vector long long, vector long long);
15760 int vec_any_ne (vector unsigned long long, vector unsigned long long);
15762 vector long long vec_eqv (vector long long, vector long long);
15763 vector long long vec_eqv (vector bool long long, vector long long);
15764 vector long long vec_eqv (vector long long, vector bool long long);
15765 vector unsigned long long vec_eqv (vector unsigned long long,
15766 vector unsigned long long);
15767 vector unsigned long long vec_eqv (vector bool long long,
15768 vector unsigned long long);
15769 vector unsigned long long vec_eqv (vector unsigned long long,
15770 vector bool long long);
15771 vector int vec_eqv (vector int, vector int);
15772 vector int vec_eqv (vector bool int, vector int);
15773 vector int vec_eqv (vector int, vector bool int);
15774 vector unsigned int vec_eqv (vector unsigned int, vector unsigned int);
15775 vector unsigned int vec_eqv (vector bool unsigned int,
15776 vector unsigned int);
15777 vector unsigned int vec_eqv (vector unsigned int,
15778 vector bool unsigned int);
15779 vector short vec_eqv (vector short, vector short);
15780 vector short vec_eqv (vector bool short, vector short);
15781 vector short vec_eqv (vector short, vector bool short);
15782 vector unsigned short vec_eqv (vector unsigned short, vector unsigned short);
15783 vector unsigned short vec_eqv (vector bool unsigned short,
15784 vector unsigned short);
15785 vector unsigned short vec_eqv (vector unsigned short,
15786 vector bool unsigned short);
15787 vector signed char vec_eqv (vector signed char, vector signed char);
15788 vector signed char vec_eqv (vector bool signed char, vector signed char);
15789 vector signed char vec_eqv (vector signed char, vector bool signed char);
15790 vector unsigned char vec_eqv (vector unsigned char, vector unsigned char);
15791 vector unsigned char vec_eqv (vector bool unsigned char, vector unsigned char);
15792 vector unsigned char vec_eqv (vector unsigned char, vector bool unsigned char);
15794 vector long long vec_max (vector long long, vector long long);
15795 vector unsigned long long vec_max (vector unsigned long long,
15796 vector unsigned long long);
15798 vector signed int vec_mergee (vector signed int, vector signed int);
15799 vector unsigned int vec_mergee (vector unsigned int, vector unsigned int);
15800 vector bool int vec_mergee (vector bool int, vector bool int);
15802 vector signed int vec_mergeo (vector signed int, vector signed int);
15803 vector unsigned int vec_mergeo (vector unsigned int, vector unsigned int);
15804 vector bool int vec_mergeo (vector bool int, vector bool int);
15806 vector long long vec_min (vector long long, vector long long);
15807 vector unsigned long long vec_min (vector unsigned long long,
15808 vector unsigned long long);
15810 vector long long vec_nand (vector long long, vector long long);
15811 vector long long vec_nand (vector bool long long, vector long long);
15812 vector long long vec_nand (vector long long, vector bool long long);
15813 vector unsigned long long vec_nand (vector unsigned long long,
15814 vector unsigned long long);
15815 vector unsigned long long vec_nand (vector bool long long,
15816 vector unsigned long long);
15817 vector unsigned long long vec_nand (vector unsigned long long,
15818 vector bool long long);
15819 vector int vec_nand (vector int, vector int);
15820 vector int vec_nand (vector bool int, vector int);
15821 vector int vec_nand (vector int, vector bool int);
15822 vector unsigned int vec_nand (vector unsigned int, vector unsigned int);
15823 vector unsigned int vec_nand (vector bool unsigned int,
15824 vector unsigned int);
15825 vector unsigned int vec_nand (vector unsigned int,
15826 vector bool unsigned int);
15827 vector short vec_nand (vector short, vector short);
15828 vector short vec_nand (vector bool short, vector short);
15829 vector short vec_nand (vector short, vector bool short);
15830 vector unsigned short vec_nand (vector unsigned short, vector unsigned short);
15831 vector unsigned short vec_nand (vector bool unsigned short,
15832 vector unsigned short);
15833 vector unsigned short vec_nand (vector unsigned short,
15834 vector bool unsigned short);
15835 vector signed char vec_nand (vector signed char, vector signed char);
15836 vector signed char vec_nand (vector bool signed char, vector signed char);
15837 vector signed char vec_nand (vector signed char, vector bool signed char);
15838 vector unsigned char vec_nand (vector unsigned char, vector unsigned char);
15839 vector unsigned char vec_nand (vector bool unsigned char, vector unsigned char);
15840 vector unsigned char vec_nand (vector unsigned char, vector bool unsigned char);
15842 vector long long vec_orc (vector long long, vector long long);
15843 vector long long vec_orc (vector bool long long, vector long long);
15844 vector long long vec_orc (vector long long, vector bool long long);
15845 vector unsigned long long vec_orc (vector unsigned long long,
15846 vector unsigned long long);
15847 vector unsigned long long vec_orc (vector bool long long,
15848 vector unsigned long long);
15849 vector unsigned long long vec_orc (vector unsigned long long,
15850 vector bool long long);
15851 vector int vec_orc (vector int, vector int);
15852 vector int vec_orc (vector bool int, vector int);
15853 vector int vec_orc (vector int, vector bool int);
15854 vector unsigned int vec_orc (vector unsigned int, vector unsigned int);
15855 vector unsigned int vec_orc (vector bool unsigned int,
15856 vector unsigned int);
15857 vector unsigned int vec_orc (vector unsigned int,
15858 vector bool unsigned int);
15859 vector short vec_orc (vector short, vector short);
15860 vector short vec_orc (vector bool short, vector short);
15861 vector short vec_orc (vector short, vector bool short);
15862 vector unsigned short vec_orc (vector unsigned short, vector unsigned short);
15863 vector unsigned short vec_orc (vector bool unsigned short,
15864 vector unsigned short);
15865 vector unsigned short vec_orc (vector unsigned short,
15866 vector bool unsigned short);
15867 vector signed char vec_orc (vector signed char, vector signed char);
15868 vector signed char vec_orc (vector bool signed char, vector signed char);
15869 vector signed char vec_orc (vector signed char, vector bool signed char);
15870 vector unsigned char vec_orc (vector unsigned char, vector unsigned char);
15871 vector unsigned char vec_orc (vector bool unsigned char, vector unsigned char);
15872 vector unsigned char vec_orc (vector unsigned char, vector bool unsigned char);
15874 vector int vec_pack (vector long long, vector long long);
15875 vector unsigned int vec_pack (vector unsigned long long,
15876 vector unsigned long long);
15877 vector bool int vec_pack (vector bool long long, vector bool long long);
15879 vector int vec_packs (vector long long, vector long long);
15880 vector unsigned int vec_packs (vector unsigned long long,
15881 vector unsigned long long);
15883 vector unsigned int vec_packsu (vector long long, vector long long);
15884 vector unsigned int vec_packsu (vector unsigned long long,
15885 vector unsigned long long);
15887 vector long long vec_rl (vector long long,
15888 vector unsigned long long);
15889 vector long long vec_rl (vector unsigned long long,
15890 vector unsigned long long);
15892 vector long long vec_sl (vector long long, vector unsigned long long);
15893 vector long long vec_sl (vector unsigned long long,
15894 vector unsigned long long);
15896 vector long long vec_sr (vector long long, vector unsigned long long);
15897 vector unsigned long long char vec_sr (vector unsigned long long,
15898 vector unsigned long long);
15900 vector long long vec_sra (vector long long, vector unsigned long long);
15901 vector unsigned long long vec_sra (vector unsigned long long,
15902 vector unsigned long long);
15904 vector long long vec_sub (vector long long, vector long long);
15905 vector unsigned long long vec_sub (vector unsigned long long,
15906 vector unsigned long long);
15908 vector long long vec_unpackh (vector int);
15909 vector unsigned long long vec_unpackh (vector unsigned int);
15911 vector long long vec_unpackl (vector int);
15912 vector unsigned long long vec_unpackl (vector unsigned int);
15914 vector long long vec_vaddudm (vector long long, vector long long);
15915 vector long long vec_vaddudm (vector bool long long, vector long long);
15916 vector long long vec_vaddudm (vector long long, vector bool long long);
15917 vector unsigned long long vec_vaddudm (vector unsigned long long,
15918 vector unsigned long long);
15919 vector unsigned long long vec_vaddudm (vector bool unsigned long long,
15920 vector unsigned long long);
15921 vector unsigned long long vec_vaddudm (vector unsigned long long,
15922 vector bool unsigned long long);
15924 vector long long vec_vbpermq (vector signed char, vector signed char);
15925 vector long long vec_vbpermq (vector unsigned char, vector unsigned char);
15927 vector long long vec_cntlz (vector long long);
15928 vector unsigned long long vec_cntlz (vector unsigned long long);
15929 vector int vec_cntlz (vector int);
15930 vector unsigned int vec_cntlz (vector int);
15931 vector short vec_cntlz (vector short);
15932 vector unsigned short vec_cntlz (vector unsigned short);
15933 vector signed char vec_cntlz (vector signed char);
15934 vector unsigned char vec_cntlz (vector unsigned char);
15936 vector long long vec_vclz (vector long long);
15937 vector unsigned long long vec_vclz (vector unsigned long long);
15938 vector int vec_vclz (vector int);
15939 vector unsigned int vec_vclz (vector int);
15940 vector short vec_vclz (vector short);
15941 vector unsigned short vec_vclz (vector unsigned short);
15942 vector signed char vec_vclz (vector signed char);
15943 vector unsigned char vec_vclz (vector unsigned char);
15945 vector signed char vec_vclzb (vector signed char);
15946 vector unsigned char vec_vclzb (vector unsigned char);
15948 vector long long vec_vclzd (vector long long);
15949 vector unsigned long long vec_vclzd (vector unsigned long long);
15951 vector short vec_vclzh (vector short);
15952 vector unsigned short vec_vclzh (vector unsigned short);
15954 vector int vec_vclzw (vector int);
15955 vector unsigned int vec_vclzw (vector int);
15957 vector signed char vec_vgbbd (vector signed char);
15958 vector unsigned char vec_vgbbd (vector unsigned char);
15960 vector long long vec_vmaxsd (vector long long, vector long long);
15962 vector unsigned long long vec_vmaxud (vector unsigned long long,
15963 unsigned vector long long);
15965 vector long long vec_vminsd (vector long long, vector long long);
15967 vector unsigned long long vec_vminud (vector long long,
15970 vector int vec_vpksdss (vector long long, vector long long);
15971 vector unsigned int vec_vpksdss (vector long long, vector long long);
15973 vector unsigned int vec_vpkudus (vector unsigned long long,
15974 vector unsigned long long);
15976 vector int vec_vpkudum (vector long long, vector long long);
15977 vector unsigned int vec_vpkudum (vector unsigned long long,
15978 vector unsigned long long);
15979 vector bool int vec_vpkudum (vector bool long long, vector bool long long);
15981 vector long long vec_vpopcnt (vector long long);
15982 vector unsigned long long vec_vpopcnt (vector unsigned long long);
15983 vector int vec_vpopcnt (vector int);
15984 vector unsigned int vec_vpopcnt (vector int);
15985 vector short vec_vpopcnt (vector short);
15986 vector unsigned short vec_vpopcnt (vector unsigned short);
15987 vector signed char vec_vpopcnt (vector signed char);
15988 vector unsigned char vec_vpopcnt (vector unsigned char);
15990 vector signed char vec_vpopcntb (vector signed char);
15991 vector unsigned char vec_vpopcntb (vector unsigned char);
15993 vector long long vec_vpopcntd (vector long long);
15994 vector unsigned long long vec_vpopcntd (vector unsigned long long);
15996 vector short vec_vpopcnth (vector short);
15997 vector unsigned short vec_vpopcnth (vector unsigned short);
15999 vector int vec_vpopcntw (vector int);
16000 vector unsigned int vec_vpopcntw (vector int);
16002 vector long long vec_vrld (vector long long, vector unsigned long long);
16003 vector unsigned long long vec_vrld (vector unsigned long long,
16004 vector unsigned long long);
16006 vector long long vec_vsld (vector long long, vector unsigned long long);
16007 vector long long vec_vsld (vector unsigned long long,
16008 vector unsigned long long);
16010 vector long long vec_vsrad (vector long long, vector unsigned long long);
16011 vector unsigned long long vec_vsrad (vector unsigned long long,
16012 vector unsigned long long);
16014 vector long long vec_vsrd (vector long long, vector unsigned long long);
16015 vector unsigned long long char vec_vsrd (vector unsigned long long,
16016 vector unsigned long long);
16018 vector long long vec_vsubudm (vector long long, vector long long);
16019 vector long long vec_vsubudm (vector bool long long, vector long long);
16020 vector long long vec_vsubudm (vector long long, vector bool long long);
16021 vector unsigned long long vec_vsubudm (vector unsigned long long,
16022 vector unsigned long long);
16023 vector unsigned long long vec_vsubudm (vector bool long long,
16024 vector unsigned long long);
16025 vector unsigned long long vec_vsubudm (vector unsigned long long,
16026 vector bool long long);
16028 vector long long vec_vupkhsw (vector int);
16029 vector unsigned long long vec_vupkhsw (vector unsigned int);
16031 vector long long vec_vupklsw (vector int);
16032 vector unsigned long long vec_vupklsw (vector int);
16035 If the ISA 2.07 additions to the vector/scalar (power8-vector)
16036 instruction set is available, the following additional functions are
16037 available for 64-bit targets. New vector types
16038 (@var{vector __int128_t} and @var{vector __uint128_t}) are available
16039 to hold the @var{__int128_t} and @var{__uint128_t} types to use these
16042 The normal vector extract, and set operations work on
16043 @var{vector __int128_t} and @var{vector __uint128_t} types,
16044 but the index value must be 0.
16047 vector __int128_t vec_vaddcuq (vector __int128_t, vector __int128_t);
16048 vector __uint128_t vec_vaddcuq (vector __uint128_t, vector __uint128_t);
16050 vector __int128_t vec_vadduqm (vector __int128_t, vector __int128_t);
16051 vector __uint128_t vec_vadduqm (vector __uint128_t, vector __uint128_t);
16053 vector __int128_t vec_vaddecuq (vector __int128_t, vector __int128_t,
16054 vector __int128_t);
16055 vector __uint128_t vec_vaddecuq (vector __uint128_t, vector __uint128_t,
16056 vector __uint128_t);
16058 vector __int128_t vec_vaddeuqm (vector __int128_t, vector __int128_t,
16059 vector __int128_t);
16060 vector __uint128_t vec_vaddeuqm (vector __uint128_t, vector __uint128_t,
16061 vector __uint128_t);
16063 vector __int128_t vec_vsubecuq (vector __int128_t, vector __int128_t,
16064 vector __int128_t);
16065 vector __uint128_t vec_vsubecuq (vector __uint128_t, vector __uint128_t,
16066 vector __uint128_t);
16068 vector __int128_t vec_vsubeuqm (vector __int128_t, vector __int128_t,
16069 vector __int128_t);
16070 vector __uint128_t vec_vsubeuqm (vector __uint128_t, vector __uint128_t,
16071 vector __uint128_t);
16073 vector __int128_t vec_vsubcuq (vector __int128_t, vector __int128_t);
16074 vector __uint128_t vec_vsubcuq (vector __uint128_t, vector __uint128_t);
16076 __int128_t vec_vsubuqm (__int128_t, __int128_t);
16077 __uint128_t vec_vsubuqm (__uint128_t, __uint128_t);
16079 vector __int128_t __builtin_bcdadd (vector __int128_t, vector__int128_t);
16080 int __builtin_bcdadd_lt (vector __int128_t, vector__int128_t);
16081 int __builtin_bcdadd_eq (vector __int128_t, vector__int128_t);
16082 int __builtin_bcdadd_gt (vector __int128_t, vector__int128_t);
16083 int __builtin_bcdadd_ov (vector __int128_t, vector__int128_t);
16084 vector __int128_t bcdsub (vector __int128_t, vector__int128_t);
16085 int __builtin_bcdsub_lt (vector __int128_t, vector__int128_t);
16086 int __builtin_bcdsub_eq (vector __int128_t, vector__int128_t);
16087 int __builtin_bcdsub_gt (vector __int128_t, vector__int128_t);
16088 int __builtin_bcdsub_ov (vector __int128_t, vector__int128_t);
16091 If the cryptographic instructions are enabled (@option{-mcrypto} or
16092 @option{-mcpu=power8}), the following builtins are enabled.
16095 vector unsigned long long __builtin_crypto_vsbox (vector unsigned long long);
16097 vector unsigned long long __builtin_crypto_vcipher (vector unsigned long long,
16098 vector unsigned long long);
16100 vector unsigned long long __builtin_crypto_vcipherlast
16101 (vector unsigned long long,
16102 vector unsigned long long);
16104 vector unsigned long long __builtin_crypto_vncipher (vector unsigned long long,
16105 vector unsigned long long);
16107 vector unsigned long long __builtin_crypto_vncipherlast
16108 (vector unsigned long long,
16109 vector unsigned long long);
16111 vector unsigned char __builtin_crypto_vpermxor (vector unsigned char,
16112 vector unsigned char,
16113 vector unsigned char);
16115 vector unsigned short __builtin_crypto_vpermxor (vector unsigned short,
16116 vector unsigned short,
16117 vector unsigned short);
16119 vector unsigned int __builtin_crypto_vpermxor (vector unsigned int,
16120 vector unsigned int,
16121 vector unsigned int);
16123 vector unsigned long long __builtin_crypto_vpermxor (vector unsigned long long,
16124 vector unsigned long long,
16125 vector unsigned long long);
16127 vector unsigned char __builtin_crypto_vpmsumb (vector unsigned char,
16128 vector unsigned char);
16130 vector unsigned short __builtin_crypto_vpmsumb (vector unsigned short,
16131 vector unsigned short);
16133 vector unsigned int __builtin_crypto_vpmsumb (vector unsigned int,
16134 vector unsigned int);
16136 vector unsigned long long __builtin_crypto_vpmsumb (vector unsigned long long,
16137 vector unsigned long long);
16139 vector unsigned long long __builtin_crypto_vshasigmad
16140 (vector unsigned long long, int, int);
16142 vector unsigned int __builtin_crypto_vshasigmaw (vector unsigned int,
16146 The second argument to the @var{__builtin_crypto_vshasigmad} and
16147 @var{__builtin_crypto_vshasigmaw} builtin functions must be a constant
16148 integer that is 0 or 1. The third argument to these builtin functions
16149 must be a constant integer in the range of 0 to 15.
16151 @node PowerPC Hardware Transactional Memory Built-in Functions
16152 @subsection PowerPC Hardware Transactional Memory Built-in Functions
16153 GCC provides two interfaces for accessing the Hardware Transactional
16154 Memory (HTM) instructions available on some of the PowerPC family
16155 of processors (eg, POWER8). The two interfaces come in a low level
16156 interface, consisting of built-in functions specific to PowerPC and a
16157 higher level interface consisting of inline functions that are common
16158 between PowerPC and S/390.
16160 @subsubsection PowerPC HTM Low Level Built-in Functions
16162 The following low level built-in functions are available with
16163 @option{-mhtm} or @option{-mcpu=CPU} where CPU is `power8' or later.
16164 They all generate the machine instruction that is part of the name.
16166 The HTM builtins (with the exception of @code{__builtin_tbegin}) return
16167 the full 4-bit condition register value set by their associated hardware
16168 instruction. The header file @code{htmintrin.h} defines some macros that can
16169 be used to decipher the return value. The @code{__builtin_tbegin} builtin
16170 returns a simple true or false value depending on whether a transaction was
16171 successfully started or not. The arguments of the builtins match exactly the
16172 type and order of the associated hardware instruction's operands, except for
16173 the @code{__builtin_tcheck} builtin, which does not take any input arguments.
16174 Refer to the ISA manual for a description of each instruction's operands.
16177 unsigned int __builtin_tbegin (unsigned int)
16178 unsigned int __builtin_tend (unsigned int)
16180 unsigned int __builtin_tabort (unsigned int)
16181 unsigned int __builtin_tabortdc (unsigned int, unsigned int, unsigned int)
16182 unsigned int __builtin_tabortdci (unsigned int, unsigned int, int)
16183 unsigned int __builtin_tabortwc (unsigned int, unsigned int, unsigned int)
16184 unsigned int __builtin_tabortwci (unsigned int, unsigned int, int)
16186 unsigned int __builtin_tcheck (void)
16187 unsigned int __builtin_treclaim (unsigned int)
16188 unsigned int __builtin_trechkpt (void)
16189 unsigned int __builtin_tsr (unsigned int)
16192 In addition to the above HTM built-ins, we have added built-ins for
16193 some common extended mnemonics of the HTM instructions:
16196 unsigned int __builtin_tendall (void)
16197 unsigned int __builtin_tresume (void)
16198 unsigned int __builtin_tsuspend (void)
16201 Note that the semantics of the above HTM builtins are required to mimic
16202 the locking semantics used for critical sections. Builtins that are used
16203 to create a new transaction or restart a suspended transaction must have
16204 lock acquisition like semantics while those builtins that end or suspend a
16205 transaction must have lock release like semantics. Specifically, this must
16206 mimic lock semantics as specified by C++11, for example: Lock acquisition is
16207 as-if an execution of __atomic_exchange_n(&globallock,1,__ATOMIC_ACQUIRE)
16208 that returns 0, and lock release is as-if an execution of
16209 __atomic_store(&globallock,0,__ATOMIC_RELEASE), with globallock being an
16210 implicit implementation-defined lock used for all transactions. The HTM
16211 instructions associated with with the builtins inherently provide the
16212 correct acquisition and release hardware barriers required. However,
16213 the compiler must also be prohibited from moving loads and stores across
16214 the builtins in a way that would violate their semantics. This has been
16215 accomplished by adding memory barriers to the associated HTM instructions
16216 (which is a conservative approach to provide acquire and release semantics).
16217 Earlier versions of the compiler did not treat the HTM instructions as
16218 memory barriers. A @code{__TM_FENCE__} macro has been added, which can
16219 be used to determine whether the current compiler treats HTM instructions
16220 as memory barriers or not. This allows the user to explicitly add memory
16221 barriers to their code when using an older version of the compiler.
16223 The following set of built-in functions are available to gain access
16224 to the HTM specific special purpose registers.
16227 unsigned long __builtin_get_texasr (void)
16228 unsigned long __builtin_get_texasru (void)
16229 unsigned long __builtin_get_tfhar (void)
16230 unsigned long __builtin_get_tfiar (void)
16232 void __builtin_set_texasr (unsigned long);
16233 void __builtin_set_texasru (unsigned long);
16234 void __builtin_set_tfhar (unsigned long);
16235 void __builtin_set_tfiar (unsigned long);
16238 Example usage of these low level built-in functions may look like:
16241 #include <htmintrin.h>
16243 int num_retries = 10;
16247 if (__builtin_tbegin (0))
16249 /* Transaction State Initiated. */
16250 if (is_locked (lock))
16251 __builtin_tabort (0);
16252 ... transaction code...
16253 __builtin_tend (0);
16258 /* Transaction State Failed. Use locks if the transaction
16259 failure is "persistent" or we've tried too many times. */
16260 if (num_retries-- <= 0
16261 || _TEXASRU_FAILURE_PERSISTENT (__builtin_get_texasru ()))
16263 acquire_lock (lock);
16264 ... non transactional fallback path...
16265 release_lock (lock);
16272 One final built-in function has been added that returns the value of
16273 the 2-bit Transaction State field of the Machine Status Register (MSR)
16274 as stored in @code{CR0}.
16277 unsigned long __builtin_ttest (void)
16280 This built-in can be used to determine the current transaction state
16281 using the following code example:
16284 #include <htmintrin.h>
16286 unsigned char tx_state = _HTM_STATE (__builtin_ttest ());
16288 if (tx_state == _HTM_TRANSACTIONAL)
16290 /* Code to use in transactional state. */
16292 else if (tx_state == _HTM_NONTRANSACTIONAL)
16294 /* Code to use in non-transactional state. */
16296 else if (tx_state == _HTM_SUSPENDED)
16298 /* Code to use in transaction suspended state. */
16302 @subsubsection PowerPC HTM High Level Inline Functions
16304 The following high level HTM interface is made available by including
16305 @code{<htmxlintrin.h>} and using @option{-mhtm} or @option{-mcpu=CPU}
16306 where CPU is `power8' or later. This interface is common between PowerPC
16307 and S/390, allowing users to write one HTM source implementation that
16308 can be compiled and executed on either system.
16311 long __TM_simple_begin (void)
16312 long __TM_begin (void* const TM_buff)
16313 long __TM_end (void)
16314 void __TM_abort (void)
16315 void __TM_named_abort (unsigned char const code)
16316 void __TM_resume (void)
16317 void __TM_suspend (void)
16319 long __TM_is_user_abort (void* const TM_buff)
16320 long __TM_is_named_user_abort (void* const TM_buff, unsigned char *code)
16321 long __TM_is_illegal (void* const TM_buff)
16322 long __TM_is_footprint_exceeded (void* const TM_buff)
16323 long __TM_nesting_depth (void* const TM_buff)
16324 long __TM_is_nested_too_deep(void* const TM_buff)
16325 long __TM_is_conflict(void* const TM_buff)
16326 long __TM_is_failure_persistent(void* const TM_buff)
16327 long __TM_failure_address(void* const TM_buff)
16328 long long __TM_failure_code(void* const TM_buff)
16331 Using these common set of HTM inline functions, we can create
16332 a more portable version of the HTM example in the previous
16333 section that will work on either PowerPC or S/390:
16336 #include <htmxlintrin.h>
16338 int num_retries = 10;
16339 TM_buff_type TM_buff;
16343 if (__TM_begin (TM_buff) == _HTM_TBEGIN_STARTED)
16345 /* Transaction State Initiated. */
16346 if (is_locked (lock))
16348 ... transaction code...
16354 /* Transaction State Failed. Use locks if the transaction
16355 failure is "persistent" or we've tried too many times. */
16356 if (num_retries-- <= 0
16357 || __TM_is_failure_persistent (TM_buff))
16359 acquire_lock (lock);
16360 ... non transactional fallback path...
16361 release_lock (lock);
16368 @node RX Built-in Functions
16369 @subsection RX Built-in Functions
16370 GCC supports some of the RX instructions which cannot be expressed in
16371 the C programming language via the use of built-in functions. The
16372 following functions are supported:
16374 @deftypefn {Built-in Function} void __builtin_rx_brk (void)
16375 Generates the @code{brk} machine instruction.
16378 @deftypefn {Built-in Function} void __builtin_rx_clrpsw (int)
16379 Generates the @code{clrpsw} machine instruction to clear the specified
16380 bit in the processor status word.
16383 @deftypefn {Built-in Function} void __builtin_rx_int (int)
16384 Generates the @code{int} machine instruction to generate an interrupt
16385 with the specified value.
16388 @deftypefn {Built-in Function} void __builtin_rx_machi (int, int)
16389 Generates the @code{machi} machine instruction to add the result of
16390 multiplying the top 16 bits of the two arguments into the
16394 @deftypefn {Built-in Function} void __builtin_rx_maclo (int, int)
16395 Generates the @code{maclo} machine instruction to add the result of
16396 multiplying the bottom 16 bits of the two arguments into the
16400 @deftypefn {Built-in Function} void __builtin_rx_mulhi (int, int)
16401 Generates the @code{mulhi} machine instruction to place the result of
16402 multiplying the top 16 bits of the two arguments into the
16406 @deftypefn {Built-in Function} void __builtin_rx_mullo (int, int)
16407 Generates the @code{mullo} machine instruction to place the result of
16408 multiplying the bottom 16 bits of the two arguments into the
16412 @deftypefn {Built-in Function} int __builtin_rx_mvfachi (void)
16413 Generates the @code{mvfachi} machine instruction to read the top
16414 32 bits of the accumulator.
16417 @deftypefn {Built-in Function} int __builtin_rx_mvfacmi (void)
16418 Generates the @code{mvfacmi} machine instruction to read the middle
16419 32 bits of the accumulator.
16422 @deftypefn {Built-in Function} int __builtin_rx_mvfc (int)
16423 Generates the @code{mvfc} machine instruction which reads the control
16424 register specified in its argument and returns its value.
16427 @deftypefn {Built-in Function} void __builtin_rx_mvtachi (int)
16428 Generates the @code{mvtachi} machine instruction to set the top
16429 32 bits of the accumulator.
16432 @deftypefn {Built-in Function} void __builtin_rx_mvtaclo (int)
16433 Generates the @code{mvtaclo} machine instruction to set the bottom
16434 32 bits of the accumulator.
16437 @deftypefn {Built-in Function} void __builtin_rx_mvtc (int reg, int val)
16438 Generates the @code{mvtc} machine instruction which sets control
16439 register number @code{reg} to @code{val}.
16442 @deftypefn {Built-in Function} void __builtin_rx_mvtipl (int)
16443 Generates the @code{mvtipl} machine instruction set the interrupt
16447 @deftypefn {Built-in Function} void __builtin_rx_racw (int)
16448 Generates the @code{racw} machine instruction to round the accumulator
16449 according to the specified mode.
16452 @deftypefn {Built-in Function} int __builtin_rx_revw (int)
16453 Generates the @code{revw} machine instruction which swaps the bytes in
16454 the argument so that bits 0--7 now occupy bits 8--15 and vice versa,
16455 and also bits 16--23 occupy bits 24--31 and vice versa.
16458 @deftypefn {Built-in Function} void __builtin_rx_rmpa (void)
16459 Generates the @code{rmpa} machine instruction which initiates a
16460 repeated multiply and accumulate sequence.
16463 @deftypefn {Built-in Function} void __builtin_rx_round (float)
16464 Generates the @code{round} machine instruction which returns the
16465 floating-point argument rounded according to the current rounding mode
16466 set in the floating-point status word register.
16469 @deftypefn {Built-in Function} int __builtin_rx_sat (int)
16470 Generates the @code{sat} machine instruction which returns the
16471 saturated value of the argument.
16474 @deftypefn {Built-in Function} void __builtin_rx_setpsw (int)
16475 Generates the @code{setpsw} machine instruction to set the specified
16476 bit in the processor status word.
16479 @deftypefn {Built-in Function} void __builtin_rx_wait (void)
16480 Generates the @code{wait} machine instruction.
16483 @node S/390 System z Built-in Functions
16484 @subsection S/390 System z Built-in Functions
16485 @deftypefn {Built-in Function} int __builtin_tbegin (void*)
16486 Generates the @code{tbegin} machine instruction starting a
16487 non-constraint hardware transaction. If the parameter is non-NULL the
16488 memory area is used to store the transaction diagnostic buffer and
16489 will be passed as first operand to @code{tbegin}. This buffer can be
16490 defined using the @code{struct __htm_tdb} C struct defined in
16491 @code{htmintrin.h} and must reside on a double-word boundary. The
16492 second tbegin operand is set to @code{0xff0c}. This enables
16493 save/restore of all GPRs and disables aborts for FPR and AR
16494 manipulations inside the transaction body. The condition code set by
16495 the tbegin instruction is returned as integer value. The tbegin
16496 instruction by definition overwrites the content of all FPRs. The
16497 compiler will generate code which saves and restores the FPRs. For
16498 soft-float code it is recommended to used the @code{*_nofloat}
16499 variant. In order to prevent a TDB from being written it is required
16500 to pass an constant zero value as parameter. Passing the zero value
16501 through a variable is not sufficient. Although modifications of
16502 access registers inside the transaction will not trigger an
16503 transaction abort it is not supported to actually modify them. Access
16504 registers do not get saved when entering a transaction. They will have
16505 undefined state when reaching the abort code.
16508 Macros for the possible return codes of tbegin are defined in the
16509 @code{htmintrin.h} header file:
16512 @item _HTM_TBEGIN_STARTED
16513 @code{tbegin} has been executed as part of normal processing. The
16514 transaction body is supposed to be executed.
16515 @item _HTM_TBEGIN_INDETERMINATE
16516 The transaction was aborted due to an indeterminate condition which
16517 might be persistent.
16518 @item _HTM_TBEGIN_TRANSIENT
16519 The transaction aborted due to a transient failure. The transaction
16520 should be re-executed in that case.
16521 @item _HTM_TBEGIN_PERSISTENT
16522 The transaction aborted due to a persistent failure. Re-execution
16523 under same circumstances will not be productive.
16526 @defmac _HTM_FIRST_USER_ABORT_CODE
16527 The @code{_HTM_FIRST_USER_ABORT_CODE} defined in @code{htmintrin.h}
16528 specifies the first abort code which can be used for
16529 @code{__builtin_tabort}. Values below this threshold are reserved for
16533 @deftp {Data type} {struct __htm_tdb}
16534 The @code{struct __htm_tdb} defined in @code{htmintrin.h} describes
16535 the structure of the transaction diagnostic block as specified in the
16536 Principles of Operation manual chapter 5-91.
16539 @deftypefn {Built-in Function} int __builtin_tbegin_nofloat (void*)
16540 Same as @code{__builtin_tbegin} but without FPR saves and restores.
16541 Using this variant in code making use of FPRs will leave the FPRs in
16542 undefined state when entering the transaction abort handler code.
16545 @deftypefn {Built-in Function} int __builtin_tbegin_retry (void*, int)
16546 In addition to @code{__builtin_tbegin} a loop for transient failures
16547 is generated. If tbegin returns a condition code of 2 the transaction
16548 will be retried as often as specified in the second argument. The
16549 perform processor assist instruction is used to tell the CPU about the
16550 number of fails so far.
16553 @deftypefn {Built-in Function} int __builtin_tbegin_retry_nofloat (void*, int)
16554 Same as @code{__builtin_tbegin_retry} but without FPR saves and
16555 restores. Using this variant in code making use of FPRs will leave
16556 the FPRs in undefined state when entering the transaction abort
16560 @deftypefn {Built-in Function} void __builtin_tbeginc (void)
16561 Generates the @code{tbeginc} machine instruction starting a constraint
16562 hardware transaction. The second operand is set to @code{0xff08}.
16565 @deftypefn {Built-in Function} int __builtin_tend (void)
16566 Generates the @code{tend} machine instruction finishing a transaction
16567 and making the changes visible to other threads. The condition code
16568 generated by tend is returned as integer value.
16571 @deftypefn {Built-in Function} void __builtin_tabort (int)
16572 Generates the @code{tabort} machine instruction with the specified
16573 abort code. Abort codes from 0 through 255 are reserved and will
16574 result in an error message.
16577 @deftypefn {Built-in Function} void __builtin_tx_assist (int)
16578 Generates the @code{ppa rX,rY,1} machine instruction. Where the
16579 integer parameter is loaded into rX and a value of zero is loaded into
16580 rY. The integer parameter specifies the number of times the
16581 transaction repeatedly aborted.
16584 @deftypefn {Built-in Function} int __builtin_tx_nesting_depth (void)
16585 Generates the @code{etnd} machine instruction. The current nesting
16586 depth is returned as integer value. For a nesting depth of 0 the code
16587 is not executed as part of an transaction.
16590 @deftypefn {Built-in Function} void __builtin_non_tx_store (uint64_t *, uint64_t)
16592 Generates the @code{ntstg} machine instruction. The second argument
16593 is written to the first arguments location. The store operation will
16594 not be rolled-back in case of an transaction abort.
16597 @node SH Built-in Functions
16598 @subsection SH Built-in Functions
16599 The following built-in functions are supported on the SH1, SH2, SH3 and SH4
16600 families of processors:
16602 @deftypefn {Built-in Function} {void} __builtin_set_thread_pointer (void *@var{ptr})
16603 Sets the @samp{GBR} register to the specified value @var{ptr}. This is usually
16604 used by system code that manages threads and execution contexts. The compiler
16605 normally does not generate code that modifies the contents of @samp{GBR} and
16606 thus the value is preserved across function calls. Changing the @samp{GBR}
16607 value in user code must be done with caution, since the compiler might use
16608 @samp{GBR} in order to access thread local variables.
16612 @deftypefn {Built-in Function} {void *} __builtin_thread_pointer (void)
16613 Returns the value that is currently set in the @samp{GBR} register.
16614 Memory loads and stores that use the thread pointer as a base address are
16615 turned into @samp{GBR} based displacement loads and stores, if possible.
16623 int get_tcb_value (void)
16625 // Generate @samp{mov.l @@(8,gbr),r0} instruction
16626 return ((my_tcb*)__builtin_thread_pointer ())->c;
16632 @deftypefn {Built-in Function} {unsigned int} __builtin_sh_get_fpscr (void)
16633 Returns the value that is currently set in the @samp{FPSCR} register.
16636 @deftypefn {Built-in Function} {void} __builtin_sh_set_fpscr (unsigned int @var{val})
16637 Sets the @samp{FPSCR} register to the specified value @var{val}, while
16638 preserving the current values of the FR, SZ and PR bits.
16641 @node SPARC VIS Built-in Functions
16642 @subsection SPARC VIS Built-in Functions
16644 GCC supports SIMD operations on the SPARC using both the generic vector
16645 extensions (@pxref{Vector Extensions}) as well as built-in functions for
16646 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
16647 switch, the VIS extension is exposed as the following built-in functions:
16650 typedef int v1si __attribute__ ((vector_size (4)));
16651 typedef int v2si __attribute__ ((vector_size (8)));
16652 typedef short v4hi __attribute__ ((vector_size (8)));
16653 typedef short v2hi __attribute__ ((vector_size (4)));
16654 typedef unsigned char v8qi __attribute__ ((vector_size (8)));
16655 typedef unsigned char v4qi __attribute__ ((vector_size (4)));
16657 void __builtin_vis_write_gsr (int64_t);
16658 int64_t __builtin_vis_read_gsr (void);
16660 void * __builtin_vis_alignaddr (void *, long);
16661 void * __builtin_vis_alignaddrl (void *, long);
16662 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
16663 v2si __builtin_vis_faligndatav2si (v2si, v2si);
16664 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
16665 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
16667 v4hi __builtin_vis_fexpand (v4qi);
16669 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
16670 v4hi __builtin_vis_fmul8x16au (v4qi, v2hi);
16671 v4hi __builtin_vis_fmul8x16al (v4qi, v2hi);
16672 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
16673 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
16674 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
16675 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
16677 v4qi __builtin_vis_fpack16 (v4hi);
16678 v8qi __builtin_vis_fpack32 (v2si, v8qi);
16679 v2hi __builtin_vis_fpackfix (v2si);
16680 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
16682 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
16684 long __builtin_vis_edge8 (void *, void *);
16685 long __builtin_vis_edge8l (void *, void *);
16686 long __builtin_vis_edge16 (void *, void *);
16687 long __builtin_vis_edge16l (void *, void *);
16688 long __builtin_vis_edge32 (void *, void *);
16689 long __builtin_vis_edge32l (void *, void *);
16691 long __builtin_vis_fcmple16 (v4hi, v4hi);
16692 long __builtin_vis_fcmple32 (v2si, v2si);
16693 long __builtin_vis_fcmpne16 (v4hi, v4hi);
16694 long __builtin_vis_fcmpne32 (v2si, v2si);
16695 long __builtin_vis_fcmpgt16 (v4hi, v4hi);
16696 long __builtin_vis_fcmpgt32 (v2si, v2si);
16697 long __builtin_vis_fcmpeq16 (v4hi, v4hi);
16698 long __builtin_vis_fcmpeq32 (v2si, v2si);
16700 v4hi __builtin_vis_fpadd16 (v4hi, v4hi);
16701 v2hi __builtin_vis_fpadd16s (v2hi, v2hi);
16702 v2si __builtin_vis_fpadd32 (v2si, v2si);
16703 v1si __builtin_vis_fpadd32s (v1si, v1si);
16704 v4hi __builtin_vis_fpsub16 (v4hi, v4hi);
16705 v2hi __builtin_vis_fpsub16s (v2hi, v2hi);
16706 v2si __builtin_vis_fpsub32 (v2si, v2si);
16707 v1si __builtin_vis_fpsub32s (v1si, v1si);
16709 long __builtin_vis_array8 (long, long);
16710 long __builtin_vis_array16 (long, long);
16711 long __builtin_vis_array32 (long, long);
16714 When you use the @option{-mvis2} switch, the VIS version 2.0 built-in
16715 functions also become available:
16718 long __builtin_vis_bmask (long, long);
16719 int64_t __builtin_vis_bshuffledi (int64_t, int64_t);
16720 v2si __builtin_vis_bshufflev2si (v2si, v2si);
16721 v4hi __builtin_vis_bshufflev2si (v4hi, v4hi);
16722 v8qi __builtin_vis_bshufflev2si (v8qi, v8qi);
16724 long __builtin_vis_edge8n (void *, void *);
16725 long __builtin_vis_edge8ln (void *, void *);
16726 long __builtin_vis_edge16n (void *, void *);
16727 long __builtin_vis_edge16ln (void *, void *);
16728 long __builtin_vis_edge32n (void *, void *);
16729 long __builtin_vis_edge32ln (void *, void *);
16732 When you use the @option{-mvis3} switch, the VIS version 3.0 built-in
16733 functions also become available:
16736 void __builtin_vis_cmask8 (long);
16737 void __builtin_vis_cmask16 (long);
16738 void __builtin_vis_cmask32 (long);
16740 v4hi __builtin_vis_fchksm16 (v4hi, v4hi);
16742 v4hi __builtin_vis_fsll16 (v4hi, v4hi);
16743 v4hi __builtin_vis_fslas16 (v4hi, v4hi);
16744 v4hi __builtin_vis_fsrl16 (v4hi, v4hi);
16745 v4hi __builtin_vis_fsra16 (v4hi, v4hi);
16746 v2si __builtin_vis_fsll16 (v2si, v2si);
16747 v2si __builtin_vis_fslas16 (v2si, v2si);
16748 v2si __builtin_vis_fsrl16 (v2si, v2si);
16749 v2si __builtin_vis_fsra16 (v2si, v2si);
16751 long __builtin_vis_pdistn (v8qi, v8qi);
16753 v4hi __builtin_vis_fmean16 (v4hi, v4hi);
16755 int64_t __builtin_vis_fpadd64 (int64_t, int64_t);
16756 int64_t __builtin_vis_fpsub64 (int64_t, int64_t);
16758 v4hi __builtin_vis_fpadds16 (v4hi, v4hi);
16759 v2hi __builtin_vis_fpadds16s (v2hi, v2hi);
16760 v4hi __builtin_vis_fpsubs16 (v4hi, v4hi);
16761 v2hi __builtin_vis_fpsubs16s (v2hi, v2hi);
16762 v2si __builtin_vis_fpadds32 (v2si, v2si);
16763 v1si __builtin_vis_fpadds32s (v1si, v1si);
16764 v2si __builtin_vis_fpsubs32 (v2si, v2si);
16765 v1si __builtin_vis_fpsubs32s (v1si, v1si);
16767 long __builtin_vis_fucmple8 (v8qi, v8qi);
16768 long __builtin_vis_fucmpne8 (v8qi, v8qi);
16769 long __builtin_vis_fucmpgt8 (v8qi, v8qi);
16770 long __builtin_vis_fucmpeq8 (v8qi, v8qi);
16772 float __builtin_vis_fhadds (float, float);
16773 double __builtin_vis_fhaddd (double, double);
16774 float __builtin_vis_fhsubs (float, float);
16775 double __builtin_vis_fhsubd (double, double);
16776 float __builtin_vis_fnhadds (float, float);
16777 double __builtin_vis_fnhaddd (double, double);
16779 int64_t __builtin_vis_umulxhi (int64_t, int64_t);
16780 int64_t __builtin_vis_xmulx (int64_t, int64_t);
16781 int64_t __builtin_vis_xmulxhi (int64_t, int64_t);
16784 @node SPU Built-in Functions
16785 @subsection SPU Built-in Functions
16787 GCC provides extensions for the SPU processor as described in the
16788 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
16789 found at @uref{http://cell.scei.co.jp/} or
16790 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
16791 implementation differs in several ways.
16796 The optional extension of specifying vector constants in parentheses is
16800 A vector initializer requires no cast if the vector constant is of the
16801 same type as the variable it is initializing.
16804 If @code{signed} or @code{unsigned} is omitted, the signedness of the
16805 vector type is the default signedness of the base type. The default
16806 varies depending on the operating system, so a portable program should
16807 always specify the signedness.
16810 By default, the keyword @code{__vector} is added. The macro
16811 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
16815 GCC allows using a @code{typedef} name as the type specifier for a
16819 For C, overloaded functions are implemented with macros so the following
16823 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
16827 Since @code{spu_add} is a macro, the vector constant in the example
16828 is treated as four separate arguments. Wrap the entire argument in
16829 parentheses for this to work.
16832 The extended version of @code{__builtin_expect} is not supported.
16836 @emph{Note:} Only the interface described in the aforementioned
16837 specification is supported. Internally, GCC uses built-in functions to
16838 implement the required functionality, but these are not supported and
16839 are subject to change without notice.
16841 @node TI C6X Built-in Functions
16842 @subsection TI C6X Built-in Functions
16844 GCC provides intrinsics to access certain instructions of the TI C6X
16845 processors. These intrinsics, listed below, are available after
16846 inclusion of the @code{c6x_intrinsics.h} header file. They map directly
16847 to C6X instructions.
16851 int _sadd (int, int)
16852 int _ssub (int, int)
16853 int _sadd2 (int, int)
16854 int _ssub2 (int, int)
16855 long long _mpy2 (int, int)
16856 long long _smpy2 (int, int)
16857 int _add4 (int, int)
16858 int _sub4 (int, int)
16859 int _saddu4 (int, int)
16861 int _smpy (int, int)
16862 int _smpyh (int, int)
16863 int _smpyhl (int, int)
16864 int _smpylh (int, int)
16866 int _sshl (int, int)
16867 int _subc (int, int)
16869 int _avg2 (int, int)
16870 int _avgu4 (int, int)
16872 int _clrr (int, int)
16873 int _extr (int, int)
16874 int _extru (int, int)
16880 @node TILE-Gx Built-in Functions
16881 @subsection TILE-Gx Built-in Functions
16883 GCC provides intrinsics to access every instruction of the TILE-Gx
16884 processor. The intrinsics are of the form:
16888 unsigned long long __insn_@var{op} (...)
16892 Where @var{op} is the name of the instruction. Refer to the ISA manual
16893 for the complete list of instructions.
16895 GCC also provides intrinsics to directly access the network registers.
16896 The intrinsics are:
16900 unsigned long long __tile_idn0_receive (void)
16901 unsigned long long __tile_idn1_receive (void)
16902 unsigned long long __tile_udn0_receive (void)
16903 unsigned long long __tile_udn1_receive (void)
16904 unsigned long long __tile_udn2_receive (void)
16905 unsigned long long __tile_udn3_receive (void)
16906 void __tile_idn_send (unsigned long long)
16907 void __tile_udn_send (unsigned long long)
16911 The intrinsic @code{void __tile_network_barrier (void)} is used to
16912 guarantee that no network operations before it are reordered with
16915 @node TILEPro Built-in Functions
16916 @subsection TILEPro Built-in Functions
16918 GCC provides intrinsics to access every instruction of the TILEPro
16919 processor. The intrinsics are of the form:
16923 unsigned __insn_@var{op} (...)
16928 where @var{op} is the name of the instruction. Refer to the ISA manual
16929 for the complete list of instructions.
16931 GCC also provides intrinsics to directly access the network registers.
16932 The intrinsics are:
16936 unsigned __tile_idn0_receive (void)
16937 unsigned __tile_idn1_receive (void)
16938 unsigned __tile_sn_receive (void)
16939 unsigned __tile_udn0_receive (void)
16940 unsigned __tile_udn1_receive (void)
16941 unsigned __tile_udn2_receive (void)
16942 unsigned __tile_udn3_receive (void)
16943 void __tile_idn_send (unsigned)
16944 void __tile_sn_send (unsigned)
16945 void __tile_udn_send (unsigned)
16949 The intrinsic @code{void __tile_network_barrier (void)} is used to
16950 guarantee that no network operations before it are reordered with
16953 @node x86 Built-in Functions
16954 @subsection x86 Built-in Functions
16956 These built-in functions are available for the x86-32 and x86-64 family
16957 of computers, depending on the command-line switches used.
16959 If you specify command-line switches such as @option{-msse},
16960 the compiler could use the extended instruction sets even if the built-ins
16961 are not used explicitly in the program. For this reason, applications
16962 that perform run-time CPU detection must compile separate files for each
16963 supported architecture, using the appropriate flags. In particular,
16964 the file containing the CPU detection code should be compiled without
16967 The following machine modes are available for use with MMX built-in functions
16968 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
16969 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
16970 vector of eight 8-bit integers. Some of the built-in functions operate on
16971 MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode.
16973 If 3DNow!@: extensions are enabled, @code{V2SF} is used as a mode for a vector
16974 of two 32-bit floating-point values.
16976 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
16977 floating-point values. Some instructions use a vector of four 32-bit
16978 integers, these use @code{V4SI}. Finally, some instructions operate on an
16979 entire vector register, interpreting it as a 128-bit integer, these use mode
16982 In 64-bit mode, the x86-64 family of processors uses additional built-in
16983 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
16984 floating point and @code{TC} 128-bit complex floating-point values.
16986 The following floating-point built-in functions are available in 64-bit
16987 mode. All of them implement the function that is part of the name.
16990 __float128 __builtin_fabsq (__float128)
16991 __float128 __builtin_copysignq (__float128, __float128)
16994 The following built-in function is always available.
16997 @item void __builtin_ia32_pause (void)
16998 Generates the @code{pause} machine instruction with a compiler memory
17002 The following floating-point built-in functions are made available in the
17006 @item __float128 __builtin_infq (void)
17007 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
17008 @findex __builtin_infq
17010 @item __float128 __builtin_huge_valq (void)
17011 Similar to @code{__builtin_huge_val}, except the return type is @code{__float128}.
17012 @findex __builtin_huge_valq
17015 The following built-in functions are always available and can be used to
17016 check the target platform type.
17018 @deftypefn {Built-in Function} void __builtin_cpu_init (void)
17019 This function runs the CPU detection code to check the type of CPU and the
17020 features supported. This built-in function needs to be invoked along with the built-in functions
17021 to check CPU type and features, @code{__builtin_cpu_is} and
17022 @code{__builtin_cpu_supports}, only when used in a function that is
17023 executed before any constructors are called. The CPU detection code is
17024 automatically executed in a very high priority constructor.
17026 For example, this function has to be used in @code{ifunc} resolvers that
17027 check for CPU type using the built-in functions @code{__builtin_cpu_is}
17028 and @code{__builtin_cpu_supports}, or in constructors on targets that
17029 don't support constructor priority.
17032 static void (*resolve_memcpy (void)) (void)
17034 // ifunc resolvers fire before constructors, explicitly call the init
17036 __builtin_cpu_init ();
17037 if (__builtin_cpu_supports ("ssse3"))
17038 return ssse3_memcpy; // super fast memcpy with ssse3 instructions.
17040 return default_memcpy;
17043 void *memcpy (void *, const void *, size_t)
17044 __attribute__ ((ifunc ("resolve_memcpy")));
17049 @deftypefn {Built-in Function} int __builtin_cpu_is (const char *@var{cpuname})
17050 This function returns a positive integer if the run-time CPU
17051 is of type @var{cpuname}
17052 and returns @code{0} otherwise. The following CPU names can be detected:
17068 Intel Core i7 Nehalem CPU.
17071 Intel Core i7 Westmere CPU.
17074 Intel Core i7 Sandy Bridge CPU.
17080 AMD Family 10h CPU.
17083 AMD Family 10h Barcelona CPU.
17086 AMD Family 10h Shanghai CPU.
17089 AMD Family 10h Istanbul CPU.
17092 AMD Family 14h CPU.
17095 AMD Family 15h CPU.
17098 AMD Family 15h Bulldozer version 1.
17101 AMD Family 15h Bulldozer version 2.
17104 AMD Family 15h Bulldozer version 3.
17107 AMD Family 15h Bulldozer version 4.
17110 AMD Family 16h CPU.
17113 AMD Family 17h CPU.
17116 Here is an example:
17118 if (__builtin_cpu_is ("corei7"))
17120 do_corei7 (); // Core i7 specific implementation.
17124 do_generic (); // Generic implementation.
17129 @deftypefn {Built-in Function} int __builtin_cpu_supports (const char *@var{feature})
17130 This function returns a positive integer if the run-time CPU
17131 supports @var{feature}
17132 and returns @code{0} otherwise. The following features can be detected:
17140 POPCNT instruction.
17148 SSSE3 instructions.
17150 SSE4.1 instructions.
17152 SSE4.2 instructions.
17158 AVX512F instructions.
17161 Here is an example:
17163 if (__builtin_cpu_supports ("popcnt"))
17165 asm("popcnt %1,%0" : "=r"(count) : "rm"(n) : "cc");
17169 count = generic_countbits (n); //generic implementation.
17175 The following built-in functions are made available by @option{-mmmx}.
17176 All of them generate the machine instruction that is part of the name.
17179 v8qi __builtin_ia32_paddb (v8qi, v8qi)
17180 v4hi __builtin_ia32_paddw (v4hi, v4hi)
17181 v2si __builtin_ia32_paddd (v2si, v2si)
17182 v8qi __builtin_ia32_psubb (v8qi, v8qi)
17183 v4hi __builtin_ia32_psubw (v4hi, v4hi)
17184 v2si __builtin_ia32_psubd (v2si, v2si)
17185 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
17186 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
17187 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
17188 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
17189 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
17190 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
17191 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
17192 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
17193 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
17194 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
17195 di __builtin_ia32_pand (di, di)
17196 di __builtin_ia32_pandn (di,di)
17197 di __builtin_ia32_por (di, di)
17198 di __builtin_ia32_pxor (di, di)
17199 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
17200 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
17201 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
17202 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
17203 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
17204 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
17205 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
17206 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
17207 v2si __builtin_ia32_punpckhdq (v2si, v2si)
17208 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
17209 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
17210 v2si __builtin_ia32_punpckldq (v2si, v2si)
17211 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
17212 v4hi __builtin_ia32_packssdw (v2si, v2si)
17213 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
17215 v4hi __builtin_ia32_psllw (v4hi, v4hi)
17216 v2si __builtin_ia32_pslld (v2si, v2si)
17217 v1di __builtin_ia32_psllq (v1di, v1di)
17218 v4hi __builtin_ia32_psrlw (v4hi, v4hi)
17219 v2si __builtin_ia32_psrld (v2si, v2si)
17220 v1di __builtin_ia32_psrlq (v1di, v1di)
17221 v4hi __builtin_ia32_psraw (v4hi, v4hi)
17222 v2si __builtin_ia32_psrad (v2si, v2si)
17223 v4hi __builtin_ia32_psllwi (v4hi, int)
17224 v2si __builtin_ia32_pslldi (v2si, int)
17225 v1di __builtin_ia32_psllqi (v1di, int)
17226 v4hi __builtin_ia32_psrlwi (v4hi, int)
17227 v2si __builtin_ia32_psrldi (v2si, int)
17228 v1di __builtin_ia32_psrlqi (v1di, int)
17229 v4hi __builtin_ia32_psrawi (v4hi, int)
17230 v2si __builtin_ia32_psradi (v2si, int)
17234 The following built-in functions are made available either with
17235 @option{-msse}, or with a combination of @option{-m3dnow} and
17236 @option{-march=athlon}. All of them generate the machine
17237 instruction that is part of the name.
17240 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
17241 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
17242 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
17243 v1di __builtin_ia32_psadbw (v8qi, v8qi)
17244 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
17245 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
17246 v8qi __builtin_ia32_pminub (v8qi, v8qi)
17247 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
17248 int __builtin_ia32_pmovmskb (v8qi)
17249 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
17250 void __builtin_ia32_movntq (di *, di)
17251 void __builtin_ia32_sfence (void)
17254 The following built-in functions are available when @option{-msse} is used.
17255 All of them generate the machine instruction that is part of the name.
17258 int __builtin_ia32_comieq (v4sf, v4sf)
17259 int __builtin_ia32_comineq (v4sf, v4sf)
17260 int __builtin_ia32_comilt (v4sf, v4sf)
17261 int __builtin_ia32_comile (v4sf, v4sf)
17262 int __builtin_ia32_comigt (v4sf, v4sf)
17263 int __builtin_ia32_comige (v4sf, v4sf)
17264 int __builtin_ia32_ucomieq (v4sf, v4sf)
17265 int __builtin_ia32_ucomineq (v4sf, v4sf)
17266 int __builtin_ia32_ucomilt (v4sf, v4sf)
17267 int __builtin_ia32_ucomile (v4sf, v4sf)
17268 int __builtin_ia32_ucomigt (v4sf, v4sf)
17269 int __builtin_ia32_ucomige (v4sf, v4sf)
17270 v4sf __builtin_ia32_addps (v4sf, v4sf)
17271 v4sf __builtin_ia32_subps (v4sf, v4sf)
17272 v4sf __builtin_ia32_mulps (v4sf, v4sf)
17273 v4sf __builtin_ia32_divps (v4sf, v4sf)
17274 v4sf __builtin_ia32_addss (v4sf, v4sf)
17275 v4sf __builtin_ia32_subss (v4sf, v4sf)
17276 v4sf __builtin_ia32_mulss (v4sf, v4sf)
17277 v4sf __builtin_ia32_divss (v4sf, v4sf)
17278 v4sf __builtin_ia32_cmpeqps (v4sf, v4sf)
17279 v4sf __builtin_ia32_cmpltps (v4sf, v4sf)
17280 v4sf __builtin_ia32_cmpleps (v4sf, v4sf)
17281 v4sf __builtin_ia32_cmpgtps (v4sf, v4sf)
17282 v4sf __builtin_ia32_cmpgeps (v4sf, v4sf)
17283 v4sf __builtin_ia32_cmpunordps (v4sf, v4sf)
17284 v4sf __builtin_ia32_cmpneqps (v4sf, v4sf)
17285 v4sf __builtin_ia32_cmpnltps (v4sf, v4sf)
17286 v4sf __builtin_ia32_cmpnleps (v4sf, v4sf)
17287 v4sf __builtin_ia32_cmpngtps (v4sf, v4sf)
17288 v4sf __builtin_ia32_cmpngeps (v4sf, v4sf)
17289 v4sf __builtin_ia32_cmpordps (v4sf, v4sf)
17290 v4sf __builtin_ia32_cmpeqss (v4sf, v4sf)
17291 v4sf __builtin_ia32_cmpltss (v4sf, v4sf)
17292 v4sf __builtin_ia32_cmpless (v4sf, v4sf)
17293 v4sf __builtin_ia32_cmpunordss (v4sf, v4sf)
17294 v4sf __builtin_ia32_cmpneqss (v4sf, v4sf)
17295 v4sf __builtin_ia32_cmpnltss (v4sf, v4sf)
17296 v4sf __builtin_ia32_cmpnless (v4sf, v4sf)
17297 v4sf __builtin_ia32_cmpordss (v4sf, v4sf)
17298 v4sf __builtin_ia32_maxps (v4sf, v4sf)
17299 v4sf __builtin_ia32_maxss (v4sf, v4sf)
17300 v4sf __builtin_ia32_minps (v4sf, v4sf)
17301 v4sf __builtin_ia32_minss (v4sf, v4sf)
17302 v4sf __builtin_ia32_andps (v4sf, v4sf)
17303 v4sf __builtin_ia32_andnps (v4sf, v4sf)
17304 v4sf __builtin_ia32_orps (v4sf, v4sf)
17305 v4sf __builtin_ia32_xorps (v4sf, v4sf)
17306 v4sf __builtin_ia32_movss (v4sf, v4sf)
17307 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
17308 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
17309 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
17310 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
17311 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
17312 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
17313 v2si __builtin_ia32_cvtps2pi (v4sf)
17314 int __builtin_ia32_cvtss2si (v4sf)
17315 v2si __builtin_ia32_cvttps2pi (v4sf)
17316 int __builtin_ia32_cvttss2si (v4sf)
17317 v4sf __builtin_ia32_rcpps (v4sf)
17318 v4sf __builtin_ia32_rsqrtps (v4sf)
17319 v4sf __builtin_ia32_sqrtps (v4sf)
17320 v4sf __builtin_ia32_rcpss (v4sf)
17321 v4sf __builtin_ia32_rsqrtss (v4sf)
17322 v4sf __builtin_ia32_sqrtss (v4sf)
17323 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
17324 void __builtin_ia32_movntps (float *, v4sf)
17325 int __builtin_ia32_movmskps (v4sf)
17328 The following built-in functions are available when @option{-msse} is used.
17331 @item v4sf __builtin_ia32_loadups (float *)
17332 Generates the @code{movups} machine instruction as a load from memory.
17333 @item void __builtin_ia32_storeups (float *, v4sf)
17334 Generates the @code{movups} machine instruction as a store to memory.
17335 @item v4sf __builtin_ia32_loadss (float *)
17336 Generates the @code{movss} machine instruction as a load from memory.
17337 @item v4sf __builtin_ia32_loadhps (v4sf, const v2sf *)
17338 Generates the @code{movhps} machine instruction as a load from memory.
17339 @item v4sf __builtin_ia32_loadlps (v4sf, const v2sf *)
17340 Generates the @code{movlps} machine instruction as a load from memory
17341 @item void __builtin_ia32_storehps (v2sf *, v4sf)
17342 Generates the @code{movhps} machine instruction as a store to memory.
17343 @item void __builtin_ia32_storelps (v2sf *, v4sf)
17344 Generates the @code{movlps} machine instruction as a store to memory.
17347 The following built-in functions are available when @option{-msse2} is used.
17348 All of them generate the machine instruction that is part of the name.
17351 int __builtin_ia32_comisdeq (v2df, v2df)
17352 int __builtin_ia32_comisdlt (v2df, v2df)
17353 int __builtin_ia32_comisdle (v2df, v2df)
17354 int __builtin_ia32_comisdgt (v2df, v2df)
17355 int __builtin_ia32_comisdge (v2df, v2df)
17356 int __builtin_ia32_comisdneq (v2df, v2df)
17357 int __builtin_ia32_ucomisdeq (v2df, v2df)
17358 int __builtin_ia32_ucomisdlt (v2df, v2df)
17359 int __builtin_ia32_ucomisdle (v2df, v2df)
17360 int __builtin_ia32_ucomisdgt (v2df, v2df)
17361 int __builtin_ia32_ucomisdge (v2df, v2df)
17362 int __builtin_ia32_ucomisdneq (v2df, v2df)
17363 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
17364 v2df __builtin_ia32_cmpltpd (v2df, v2df)
17365 v2df __builtin_ia32_cmplepd (v2df, v2df)
17366 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
17367 v2df __builtin_ia32_cmpgepd (v2df, v2df)
17368 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
17369 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
17370 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
17371 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
17372 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
17373 v2df __builtin_ia32_cmpngepd (v2df, v2df)
17374 v2df __builtin_ia32_cmpordpd (v2df, v2df)
17375 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
17376 v2df __builtin_ia32_cmpltsd (v2df, v2df)
17377 v2df __builtin_ia32_cmplesd (v2df, v2df)
17378 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
17379 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
17380 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
17381 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
17382 v2df __builtin_ia32_cmpordsd (v2df, v2df)
17383 v2di __builtin_ia32_paddq (v2di, v2di)
17384 v2di __builtin_ia32_psubq (v2di, v2di)
17385 v2df __builtin_ia32_addpd (v2df, v2df)
17386 v2df __builtin_ia32_subpd (v2df, v2df)
17387 v2df __builtin_ia32_mulpd (v2df, v2df)
17388 v2df __builtin_ia32_divpd (v2df, v2df)
17389 v2df __builtin_ia32_addsd (v2df, v2df)
17390 v2df __builtin_ia32_subsd (v2df, v2df)
17391 v2df __builtin_ia32_mulsd (v2df, v2df)
17392 v2df __builtin_ia32_divsd (v2df, v2df)
17393 v2df __builtin_ia32_minpd (v2df, v2df)
17394 v2df __builtin_ia32_maxpd (v2df, v2df)
17395 v2df __builtin_ia32_minsd (v2df, v2df)
17396 v2df __builtin_ia32_maxsd (v2df, v2df)
17397 v2df __builtin_ia32_andpd (v2df, v2df)
17398 v2df __builtin_ia32_andnpd (v2df, v2df)
17399 v2df __builtin_ia32_orpd (v2df, v2df)
17400 v2df __builtin_ia32_xorpd (v2df, v2df)
17401 v2df __builtin_ia32_movsd (v2df, v2df)
17402 v2df __builtin_ia32_unpckhpd (v2df, v2df)
17403 v2df __builtin_ia32_unpcklpd (v2df, v2df)
17404 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
17405 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
17406 v4si __builtin_ia32_paddd128 (v4si, v4si)
17407 v2di __builtin_ia32_paddq128 (v2di, v2di)
17408 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
17409 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
17410 v4si __builtin_ia32_psubd128 (v4si, v4si)
17411 v2di __builtin_ia32_psubq128 (v2di, v2di)
17412 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
17413 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
17414 v2di __builtin_ia32_pand128 (v2di, v2di)
17415 v2di __builtin_ia32_pandn128 (v2di, v2di)
17416 v2di __builtin_ia32_por128 (v2di, v2di)
17417 v2di __builtin_ia32_pxor128 (v2di, v2di)
17418 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
17419 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
17420 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
17421 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
17422 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
17423 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
17424 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
17425 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
17426 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
17427 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
17428 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
17429 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
17430 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
17431 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
17432 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
17433 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
17434 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
17435 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
17436 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
17437 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
17438 v16qi __builtin_ia32_packsswb128 (v8hi, v8hi)
17439 v8hi __builtin_ia32_packssdw128 (v4si, v4si)
17440 v16qi __builtin_ia32_packuswb128 (v8hi, v8hi)
17441 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
17442 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
17443 v2df __builtin_ia32_loadupd (double *)
17444 void __builtin_ia32_storeupd (double *, v2df)
17445 v2df __builtin_ia32_loadhpd (v2df, double const *)
17446 v2df __builtin_ia32_loadlpd (v2df, double const *)
17447 int __builtin_ia32_movmskpd (v2df)
17448 int __builtin_ia32_pmovmskb128 (v16qi)
17449 void __builtin_ia32_movnti (int *, int)
17450 void __builtin_ia32_movnti64 (long long int *, long long int)
17451 void __builtin_ia32_movntpd (double *, v2df)
17452 void __builtin_ia32_movntdq (v2df *, v2df)
17453 v4si __builtin_ia32_pshufd (v4si, int)
17454 v8hi __builtin_ia32_pshuflw (v8hi, int)
17455 v8hi __builtin_ia32_pshufhw (v8hi, int)
17456 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
17457 v2df __builtin_ia32_sqrtpd (v2df)
17458 v2df __builtin_ia32_sqrtsd (v2df)
17459 v2df __builtin_ia32_shufpd (v2df, v2df, int)
17460 v2df __builtin_ia32_cvtdq2pd (v4si)
17461 v4sf __builtin_ia32_cvtdq2ps (v4si)
17462 v4si __builtin_ia32_cvtpd2dq (v2df)
17463 v2si __builtin_ia32_cvtpd2pi (v2df)
17464 v4sf __builtin_ia32_cvtpd2ps (v2df)
17465 v4si __builtin_ia32_cvttpd2dq (v2df)
17466 v2si __builtin_ia32_cvttpd2pi (v2df)
17467 v2df __builtin_ia32_cvtpi2pd (v2si)
17468 int __builtin_ia32_cvtsd2si (v2df)
17469 int __builtin_ia32_cvttsd2si (v2df)
17470 long long __builtin_ia32_cvtsd2si64 (v2df)
17471 long long __builtin_ia32_cvttsd2si64 (v2df)
17472 v4si __builtin_ia32_cvtps2dq (v4sf)
17473 v2df __builtin_ia32_cvtps2pd (v4sf)
17474 v4si __builtin_ia32_cvttps2dq (v4sf)
17475 v2df __builtin_ia32_cvtsi2sd (v2df, int)
17476 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
17477 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
17478 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
17479 void __builtin_ia32_clflush (const void *)
17480 void __builtin_ia32_lfence (void)
17481 void __builtin_ia32_mfence (void)
17482 v16qi __builtin_ia32_loaddqu (const char *)
17483 void __builtin_ia32_storedqu (char *, v16qi)
17484 v1di __builtin_ia32_pmuludq (v2si, v2si)
17485 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
17486 v8hi __builtin_ia32_psllw128 (v8hi, v8hi)
17487 v4si __builtin_ia32_pslld128 (v4si, v4si)
17488 v2di __builtin_ia32_psllq128 (v2di, v2di)
17489 v8hi __builtin_ia32_psrlw128 (v8hi, v8hi)
17490 v4si __builtin_ia32_psrld128 (v4si, v4si)
17491 v2di __builtin_ia32_psrlq128 (v2di, v2di)
17492 v8hi __builtin_ia32_psraw128 (v8hi, v8hi)
17493 v4si __builtin_ia32_psrad128 (v4si, v4si)
17494 v2di __builtin_ia32_pslldqi128 (v2di, int)
17495 v8hi __builtin_ia32_psllwi128 (v8hi, int)
17496 v4si __builtin_ia32_pslldi128 (v4si, int)
17497 v2di __builtin_ia32_psllqi128 (v2di, int)
17498 v2di __builtin_ia32_psrldqi128 (v2di, int)
17499 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
17500 v4si __builtin_ia32_psrldi128 (v4si, int)
17501 v2di __builtin_ia32_psrlqi128 (v2di, int)
17502 v8hi __builtin_ia32_psrawi128 (v8hi, int)
17503 v4si __builtin_ia32_psradi128 (v4si, int)
17504 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
17505 v2di __builtin_ia32_movq128 (v2di)
17508 The following built-in functions are available when @option{-msse3} is used.
17509 All of them generate the machine instruction that is part of the name.
17512 v2df __builtin_ia32_addsubpd (v2df, v2df)
17513 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
17514 v2df __builtin_ia32_haddpd (v2df, v2df)
17515 v4sf __builtin_ia32_haddps (v4sf, v4sf)
17516 v2df __builtin_ia32_hsubpd (v2df, v2df)
17517 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
17518 v16qi __builtin_ia32_lddqu (char const *)
17519 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
17520 v4sf __builtin_ia32_movshdup (v4sf)
17521 v4sf __builtin_ia32_movsldup (v4sf)
17522 void __builtin_ia32_mwait (unsigned int, unsigned int)
17525 The following built-in functions are available when @option{-mssse3} is used.
17526 All of them generate the machine instruction that is part of the name.
17529 v2si __builtin_ia32_phaddd (v2si, v2si)
17530 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
17531 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
17532 v2si __builtin_ia32_phsubd (v2si, v2si)
17533 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
17534 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
17535 v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi)
17536 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
17537 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
17538 v8qi __builtin_ia32_psignb (v8qi, v8qi)
17539 v2si __builtin_ia32_psignd (v2si, v2si)
17540 v4hi __builtin_ia32_psignw (v4hi, v4hi)
17541 v1di __builtin_ia32_palignr (v1di, v1di, int)
17542 v8qi __builtin_ia32_pabsb (v8qi)
17543 v2si __builtin_ia32_pabsd (v2si)
17544 v4hi __builtin_ia32_pabsw (v4hi)
17547 The following built-in functions are available when @option{-mssse3} is used.
17548 All of them generate the machine instruction that is part of the name.
17551 v4si __builtin_ia32_phaddd128 (v4si, v4si)
17552 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
17553 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
17554 v4si __builtin_ia32_phsubd128 (v4si, v4si)
17555 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
17556 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
17557 v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
17558 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
17559 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
17560 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
17561 v4si __builtin_ia32_psignd128 (v4si, v4si)
17562 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
17563 v2di __builtin_ia32_palignr128 (v2di, v2di, int)
17564 v16qi __builtin_ia32_pabsb128 (v16qi)
17565 v4si __builtin_ia32_pabsd128 (v4si)
17566 v8hi __builtin_ia32_pabsw128 (v8hi)
17569 The following built-in functions are available when @option{-msse4.1} is
17570 used. All of them generate the machine instruction that is part of the
17574 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
17575 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
17576 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
17577 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
17578 v2df __builtin_ia32_dppd (v2df, v2df, const int)
17579 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
17580 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
17581 v2di __builtin_ia32_movntdqa (v2di *);
17582 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
17583 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
17584 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
17585 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
17586 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
17587 v8hi __builtin_ia32_phminposuw128 (v8hi)
17588 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
17589 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
17590 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
17591 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
17592 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
17593 v4si __builtin_ia32_pminsd128 (v4si, v4si)
17594 v4si __builtin_ia32_pminud128 (v4si, v4si)
17595 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
17596 v4si __builtin_ia32_pmovsxbd128 (v16qi)
17597 v2di __builtin_ia32_pmovsxbq128 (v16qi)
17598 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
17599 v2di __builtin_ia32_pmovsxdq128 (v4si)
17600 v4si __builtin_ia32_pmovsxwd128 (v8hi)
17601 v2di __builtin_ia32_pmovsxwq128 (v8hi)
17602 v4si __builtin_ia32_pmovzxbd128 (v16qi)
17603 v2di __builtin_ia32_pmovzxbq128 (v16qi)
17604 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
17605 v2di __builtin_ia32_pmovzxdq128 (v4si)
17606 v4si __builtin_ia32_pmovzxwd128 (v8hi)
17607 v2di __builtin_ia32_pmovzxwq128 (v8hi)
17608 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
17609 v4si __builtin_ia32_pmulld128 (v4si, v4si)
17610 int __builtin_ia32_ptestc128 (v2di, v2di)
17611 int __builtin_ia32_ptestnzc128 (v2di, v2di)
17612 int __builtin_ia32_ptestz128 (v2di, v2di)
17613 v2df __builtin_ia32_roundpd (v2df, const int)
17614 v4sf __builtin_ia32_roundps (v4sf, const int)
17615 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
17616 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
17619 The following built-in functions are available when @option{-msse4.1} is
17623 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
17624 Generates the @code{insertps} machine instruction.
17625 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
17626 Generates the @code{pextrb} machine instruction.
17627 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
17628 Generates the @code{pinsrb} machine instruction.
17629 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
17630 Generates the @code{pinsrd} machine instruction.
17631 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
17632 Generates the @code{pinsrq} machine instruction in 64bit mode.
17635 The following built-in functions are changed to generate new SSE4.1
17636 instructions when @option{-msse4.1} is used.
17639 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
17640 Generates the @code{extractps} machine instruction.
17641 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
17642 Generates the @code{pextrd} machine instruction.
17643 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
17644 Generates the @code{pextrq} machine instruction in 64bit mode.
17647 The following built-in functions are available when @option{-msse4.2} is
17648 used. All of them generate the machine instruction that is part of the
17652 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
17653 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
17654 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
17655 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
17656 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
17657 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
17658 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
17659 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
17660 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
17661 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
17662 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
17663 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
17664 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
17665 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
17666 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
17669 The following built-in functions are available when @option{-msse4.2} is
17673 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
17674 Generates the @code{crc32b} machine instruction.
17675 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
17676 Generates the @code{crc32w} machine instruction.
17677 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
17678 Generates the @code{crc32l} machine instruction.
17679 @item unsigned long long __builtin_ia32_crc32di (unsigned long long, unsigned long long)
17680 Generates the @code{crc32q} machine instruction.
17683 The following built-in functions are changed to generate new SSE4.2
17684 instructions when @option{-msse4.2} is used.
17687 @item int __builtin_popcount (unsigned int)
17688 Generates the @code{popcntl} machine instruction.
17689 @item int __builtin_popcountl (unsigned long)
17690 Generates the @code{popcntl} or @code{popcntq} machine instruction,
17691 depending on the size of @code{unsigned long}.
17692 @item int __builtin_popcountll (unsigned long long)
17693 Generates the @code{popcntq} machine instruction.
17696 The following built-in functions are available when @option{-mavx} is
17697 used. All of them generate the machine instruction that is part of the
17701 v4df __builtin_ia32_addpd256 (v4df,v4df)
17702 v8sf __builtin_ia32_addps256 (v8sf,v8sf)
17703 v4df __builtin_ia32_addsubpd256 (v4df,v4df)
17704 v8sf __builtin_ia32_addsubps256 (v8sf,v8sf)
17705 v4df __builtin_ia32_andnpd256 (v4df,v4df)
17706 v8sf __builtin_ia32_andnps256 (v8sf,v8sf)
17707 v4df __builtin_ia32_andpd256 (v4df,v4df)
17708 v8sf __builtin_ia32_andps256 (v8sf,v8sf)
17709 v4df __builtin_ia32_blendpd256 (v4df,v4df,int)
17710 v8sf __builtin_ia32_blendps256 (v8sf,v8sf,int)
17711 v4df __builtin_ia32_blendvpd256 (v4df,v4df,v4df)
17712 v8sf __builtin_ia32_blendvps256 (v8sf,v8sf,v8sf)
17713 v2df __builtin_ia32_cmppd (v2df,v2df,int)
17714 v4df __builtin_ia32_cmppd256 (v4df,v4df,int)
17715 v4sf __builtin_ia32_cmpps (v4sf,v4sf,int)
17716 v8sf __builtin_ia32_cmpps256 (v8sf,v8sf,int)
17717 v2df __builtin_ia32_cmpsd (v2df,v2df,int)
17718 v4sf __builtin_ia32_cmpss (v4sf,v4sf,int)
17719 v4df __builtin_ia32_cvtdq2pd256 (v4si)
17720 v8sf __builtin_ia32_cvtdq2ps256 (v8si)
17721 v4si __builtin_ia32_cvtpd2dq256 (v4df)
17722 v4sf __builtin_ia32_cvtpd2ps256 (v4df)
17723 v8si __builtin_ia32_cvtps2dq256 (v8sf)
17724 v4df __builtin_ia32_cvtps2pd256 (v4sf)
17725 v4si __builtin_ia32_cvttpd2dq256 (v4df)
17726 v8si __builtin_ia32_cvttps2dq256 (v8sf)
17727 v4df __builtin_ia32_divpd256 (v4df,v4df)
17728 v8sf __builtin_ia32_divps256 (v8sf,v8sf)
17729 v8sf __builtin_ia32_dpps256 (v8sf,v8sf,int)
17730 v4df __builtin_ia32_haddpd256 (v4df,v4df)
17731 v8sf __builtin_ia32_haddps256 (v8sf,v8sf)
17732 v4df __builtin_ia32_hsubpd256 (v4df,v4df)
17733 v8sf __builtin_ia32_hsubps256 (v8sf,v8sf)
17734 v32qi __builtin_ia32_lddqu256 (pcchar)
17735 v32qi __builtin_ia32_loaddqu256 (pcchar)
17736 v4df __builtin_ia32_loadupd256 (pcdouble)
17737 v8sf __builtin_ia32_loadups256 (pcfloat)
17738 v2df __builtin_ia32_maskloadpd (pcv2df,v2df)
17739 v4df __builtin_ia32_maskloadpd256 (pcv4df,v4df)
17740 v4sf __builtin_ia32_maskloadps (pcv4sf,v4sf)
17741 v8sf __builtin_ia32_maskloadps256 (pcv8sf,v8sf)
17742 void __builtin_ia32_maskstorepd (pv2df,v2df,v2df)
17743 void __builtin_ia32_maskstorepd256 (pv4df,v4df,v4df)
17744 void __builtin_ia32_maskstoreps (pv4sf,v4sf,v4sf)
17745 void __builtin_ia32_maskstoreps256 (pv8sf,v8sf,v8sf)
17746 v4df __builtin_ia32_maxpd256 (v4df,v4df)
17747 v8sf __builtin_ia32_maxps256 (v8sf,v8sf)
17748 v4df __builtin_ia32_minpd256 (v4df,v4df)
17749 v8sf __builtin_ia32_minps256 (v8sf,v8sf)
17750 v4df __builtin_ia32_movddup256 (v4df)
17751 int __builtin_ia32_movmskpd256 (v4df)
17752 int __builtin_ia32_movmskps256 (v8sf)
17753 v8sf __builtin_ia32_movshdup256 (v8sf)
17754 v8sf __builtin_ia32_movsldup256 (v8sf)
17755 v4df __builtin_ia32_mulpd256 (v4df,v4df)
17756 v8sf __builtin_ia32_mulps256 (v8sf,v8sf)
17757 v4df __builtin_ia32_orpd256 (v4df,v4df)
17758 v8sf __builtin_ia32_orps256 (v8sf,v8sf)
17759 v2df __builtin_ia32_pd_pd256 (v4df)
17760 v4df __builtin_ia32_pd256_pd (v2df)
17761 v4sf __builtin_ia32_ps_ps256 (v8sf)
17762 v8sf __builtin_ia32_ps256_ps (v4sf)
17763 int __builtin_ia32_ptestc256 (v4di,v4di,ptest)
17764 int __builtin_ia32_ptestnzc256 (v4di,v4di,ptest)
17765 int __builtin_ia32_ptestz256 (v4di,v4di,ptest)
17766 v8sf __builtin_ia32_rcpps256 (v8sf)
17767 v4df __builtin_ia32_roundpd256 (v4df,int)
17768 v8sf __builtin_ia32_roundps256 (v8sf,int)
17769 v8sf __builtin_ia32_rsqrtps_nr256 (v8sf)
17770 v8sf __builtin_ia32_rsqrtps256 (v8sf)
17771 v4df __builtin_ia32_shufpd256 (v4df,v4df,int)
17772 v8sf __builtin_ia32_shufps256 (v8sf,v8sf,int)
17773 v4si __builtin_ia32_si_si256 (v8si)
17774 v8si __builtin_ia32_si256_si (v4si)
17775 v4df __builtin_ia32_sqrtpd256 (v4df)
17776 v8sf __builtin_ia32_sqrtps_nr256 (v8sf)
17777 v8sf __builtin_ia32_sqrtps256 (v8sf)
17778 void __builtin_ia32_storedqu256 (pchar,v32qi)
17779 void __builtin_ia32_storeupd256 (pdouble,v4df)
17780 void __builtin_ia32_storeups256 (pfloat,v8sf)
17781 v4df __builtin_ia32_subpd256 (v4df,v4df)
17782 v8sf __builtin_ia32_subps256 (v8sf,v8sf)
17783 v4df __builtin_ia32_unpckhpd256 (v4df,v4df)
17784 v8sf __builtin_ia32_unpckhps256 (v8sf,v8sf)
17785 v4df __builtin_ia32_unpcklpd256 (v4df,v4df)
17786 v8sf __builtin_ia32_unpcklps256 (v8sf,v8sf)
17787 v4df __builtin_ia32_vbroadcastf128_pd256 (pcv2df)
17788 v8sf __builtin_ia32_vbroadcastf128_ps256 (pcv4sf)
17789 v4df __builtin_ia32_vbroadcastsd256 (pcdouble)
17790 v4sf __builtin_ia32_vbroadcastss (pcfloat)
17791 v8sf __builtin_ia32_vbroadcastss256 (pcfloat)
17792 v2df __builtin_ia32_vextractf128_pd256 (v4df,int)
17793 v4sf __builtin_ia32_vextractf128_ps256 (v8sf,int)
17794 v4si __builtin_ia32_vextractf128_si256 (v8si,int)
17795 v4df __builtin_ia32_vinsertf128_pd256 (v4df,v2df,int)
17796 v8sf __builtin_ia32_vinsertf128_ps256 (v8sf,v4sf,int)
17797 v8si __builtin_ia32_vinsertf128_si256 (v8si,v4si,int)
17798 v4df __builtin_ia32_vperm2f128_pd256 (v4df,v4df,int)
17799 v8sf __builtin_ia32_vperm2f128_ps256 (v8sf,v8sf,int)
17800 v8si __builtin_ia32_vperm2f128_si256 (v8si,v8si,int)
17801 v2df __builtin_ia32_vpermil2pd (v2df,v2df,v2di,int)
17802 v4df __builtin_ia32_vpermil2pd256 (v4df,v4df,v4di,int)
17803 v4sf __builtin_ia32_vpermil2ps (v4sf,v4sf,v4si,int)
17804 v8sf __builtin_ia32_vpermil2ps256 (v8sf,v8sf,v8si,int)
17805 v2df __builtin_ia32_vpermilpd (v2df,int)
17806 v4df __builtin_ia32_vpermilpd256 (v4df,int)
17807 v4sf __builtin_ia32_vpermilps (v4sf,int)
17808 v8sf __builtin_ia32_vpermilps256 (v8sf,int)
17809 v2df __builtin_ia32_vpermilvarpd (v2df,v2di)
17810 v4df __builtin_ia32_vpermilvarpd256 (v4df,v4di)
17811 v4sf __builtin_ia32_vpermilvarps (v4sf,v4si)
17812 v8sf __builtin_ia32_vpermilvarps256 (v8sf,v8si)
17813 int __builtin_ia32_vtestcpd (v2df,v2df,ptest)
17814 int __builtin_ia32_vtestcpd256 (v4df,v4df,ptest)
17815 int __builtin_ia32_vtestcps (v4sf,v4sf,ptest)
17816 int __builtin_ia32_vtestcps256 (v8sf,v8sf,ptest)
17817 int __builtin_ia32_vtestnzcpd (v2df,v2df,ptest)
17818 int __builtin_ia32_vtestnzcpd256 (v4df,v4df,ptest)
17819 int __builtin_ia32_vtestnzcps (v4sf,v4sf,ptest)
17820 int __builtin_ia32_vtestnzcps256 (v8sf,v8sf,ptest)
17821 int __builtin_ia32_vtestzpd (v2df,v2df,ptest)
17822 int __builtin_ia32_vtestzpd256 (v4df,v4df,ptest)
17823 int __builtin_ia32_vtestzps (v4sf,v4sf,ptest)
17824 int __builtin_ia32_vtestzps256 (v8sf,v8sf,ptest)
17825 void __builtin_ia32_vzeroall (void)
17826 void __builtin_ia32_vzeroupper (void)
17827 v4df __builtin_ia32_xorpd256 (v4df,v4df)
17828 v8sf __builtin_ia32_xorps256 (v8sf,v8sf)
17831 The following built-in functions are available when @option{-mavx2} is
17832 used. All of them generate the machine instruction that is part of the
17836 v32qi __builtin_ia32_mpsadbw256 (v32qi,v32qi,int)
17837 v32qi __builtin_ia32_pabsb256 (v32qi)
17838 v16hi __builtin_ia32_pabsw256 (v16hi)
17839 v8si __builtin_ia32_pabsd256 (v8si)
17840 v16hi __builtin_ia32_packssdw256 (v8si,v8si)
17841 v32qi __builtin_ia32_packsswb256 (v16hi,v16hi)
17842 v16hi __builtin_ia32_packusdw256 (v8si,v8si)
17843 v32qi __builtin_ia32_packuswb256 (v16hi,v16hi)
17844 v32qi __builtin_ia32_paddb256 (v32qi,v32qi)
17845 v16hi __builtin_ia32_paddw256 (v16hi,v16hi)
17846 v8si __builtin_ia32_paddd256 (v8si,v8si)
17847 v4di __builtin_ia32_paddq256 (v4di,v4di)
17848 v32qi __builtin_ia32_paddsb256 (v32qi,v32qi)
17849 v16hi __builtin_ia32_paddsw256 (v16hi,v16hi)
17850 v32qi __builtin_ia32_paddusb256 (v32qi,v32qi)
17851 v16hi __builtin_ia32_paddusw256 (v16hi,v16hi)
17852 v4di __builtin_ia32_palignr256 (v4di,v4di,int)
17853 v4di __builtin_ia32_andsi256 (v4di,v4di)
17854 v4di __builtin_ia32_andnotsi256 (v4di,v4di)
17855 v32qi __builtin_ia32_pavgb256 (v32qi,v32qi)
17856 v16hi __builtin_ia32_pavgw256 (v16hi,v16hi)
17857 v32qi __builtin_ia32_pblendvb256 (v32qi,v32qi,v32qi)
17858 v16hi __builtin_ia32_pblendw256 (v16hi,v16hi,int)
17859 v32qi __builtin_ia32_pcmpeqb256 (v32qi,v32qi)
17860 v16hi __builtin_ia32_pcmpeqw256 (v16hi,v16hi)
17861 v8si __builtin_ia32_pcmpeqd256 (c8si,v8si)
17862 v4di __builtin_ia32_pcmpeqq256 (v4di,v4di)
17863 v32qi __builtin_ia32_pcmpgtb256 (v32qi,v32qi)
17864 v16hi __builtin_ia32_pcmpgtw256 (16hi,v16hi)
17865 v8si __builtin_ia32_pcmpgtd256 (v8si,v8si)
17866 v4di __builtin_ia32_pcmpgtq256 (v4di,v4di)
17867 v16hi __builtin_ia32_phaddw256 (v16hi,v16hi)
17868 v8si __builtin_ia32_phaddd256 (v8si,v8si)
17869 v16hi __builtin_ia32_phaddsw256 (v16hi,v16hi)
17870 v16hi __builtin_ia32_phsubw256 (v16hi,v16hi)
17871 v8si __builtin_ia32_phsubd256 (v8si,v8si)
17872 v16hi __builtin_ia32_phsubsw256 (v16hi,v16hi)
17873 v32qi __builtin_ia32_pmaddubsw256 (v32qi,v32qi)
17874 v16hi __builtin_ia32_pmaddwd256 (v16hi,v16hi)
17875 v32qi __builtin_ia32_pmaxsb256 (v32qi,v32qi)
17876 v16hi __builtin_ia32_pmaxsw256 (v16hi,v16hi)
17877 v8si __builtin_ia32_pmaxsd256 (v8si,v8si)
17878 v32qi __builtin_ia32_pmaxub256 (v32qi,v32qi)
17879 v16hi __builtin_ia32_pmaxuw256 (v16hi,v16hi)
17880 v8si __builtin_ia32_pmaxud256 (v8si,v8si)
17881 v32qi __builtin_ia32_pminsb256 (v32qi,v32qi)
17882 v16hi __builtin_ia32_pminsw256 (v16hi,v16hi)
17883 v8si __builtin_ia32_pminsd256 (v8si,v8si)
17884 v32qi __builtin_ia32_pminub256 (v32qi,v32qi)
17885 v16hi __builtin_ia32_pminuw256 (v16hi,v16hi)
17886 v8si __builtin_ia32_pminud256 (v8si,v8si)
17887 int __builtin_ia32_pmovmskb256 (v32qi)
17888 v16hi __builtin_ia32_pmovsxbw256 (v16qi)
17889 v8si __builtin_ia32_pmovsxbd256 (v16qi)
17890 v4di __builtin_ia32_pmovsxbq256 (v16qi)
17891 v8si __builtin_ia32_pmovsxwd256 (v8hi)
17892 v4di __builtin_ia32_pmovsxwq256 (v8hi)
17893 v4di __builtin_ia32_pmovsxdq256 (v4si)
17894 v16hi __builtin_ia32_pmovzxbw256 (v16qi)
17895 v8si __builtin_ia32_pmovzxbd256 (v16qi)
17896 v4di __builtin_ia32_pmovzxbq256 (v16qi)
17897 v8si __builtin_ia32_pmovzxwd256 (v8hi)
17898 v4di __builtin_ia32_pmovzxwq256 (v8hi)
17899 v4di __builtin_ia32_pmovzxdq256 (v4si)
17900 v4di __builtin_ia32_pmuldq256 (v8si,v8si)
17901 v16hi __builtin_ia32_pmulhrsw256 (v16hi, v16hi)
17902 v16hi __builtin_ia32_pmulhuw256 (v16hi,v16hi)
17903 v16hi __builtin_ia32_pmulhw256 (v16hi,v16hi)
17904 v16hi __builtin_ia32_pmullw256 (v16hi,v16hi)
17905 v8si __builtin_ia32_pmulld256 (v8si,v8si)
17906 v4di __builtin_ia32_pmuludq256 (v8si,v8si)
17907 v4di __builtin_ia32_por256 (v4di,v4di)
17908 v16hi __builtin_ia32_psadbw256 (v32qi,v32qi)
17909 v32qi __builtin_ia32_pshufb256 (v32qi,v32qi)
17910 v8si __builtin_ia32_pshufd256 (v8si,int)
17911 v16hi __builtin_ia32_pshufhw256 (v16hi,int)
17912 v16hi __builtin_ia32_pshuflw256 (v16hi,int)
17913 v32qi __builtin_ia32_psignb256 (v32qi,v32qi)
17914 v16hi __builtin_ia32_psignw256 (v16hi,v16hi)
17915 v8si __builtin_ia32_psignd256 (v8si,v8si)
17916 v4di __builtin_ia32_pslldqi256 (v4di,int)
17917 v16hi __builtin_ia32_psllwi256 (16hi,int)
17918 v16hi __builtin_ia32_psllw256(v16hi,v8hi)
17919 v8si __builtin_ia32_pslldi256 (v8si,int)
17920 v8si __builtin_ia32_pslld256(v8si,v4si)
17921 v4di __builtin_ia32_psllqi256 (v4di,int)
17922 v4di __builtin_ia32_psllq256(v4di,v2di)
17923 v16hi __builtin_ia32_psrawi256 (v16hi,int)
17924 v16hi __builtin_ia32_psraw256 (v16hi,v8hi)
17925 v8si __builtin_ia32_psradi256 (v8si,int)
17926 v8si __builtin_ia32_psrad256 (v8si,v4si)
17927 v4di __builtin_ia32_psrldqi256 (v4di, int)
17928 v16hi __builtin_ia32_psrlwi256 (v16hi,int)
17929 v16hi __builtin_ia32_psrlw256 (v16hi,v8hi)
17930 v8si __builtin_ia32_psrldi256 (v8si,int)
17931 v8si __builtin_ia32_psrld256 (v8si,v4si)
17932 v4di __builtin_ia32_psrlqi256 (v4di,int)
17933 v4di __builtin_ia32_psrlq256(v4di,v2di)
17934 v32qi __builtin_ia32_psubb256 (v32qi,v32qi)
17935 v32hi __builtin_ia32_psubw256 (v16hi,v16hi)
17936 v8si __builtin_ia32_psubd256 (v8si,v8si)
17937 v4di __builtin_ia32_psubq256 (v4di,v4di)
17938 v32qi __builtin_ia32_psubsb256 (v32qi,v32qi)
17939 v16hi __builtin_ia32_psubsw256 (v16hi,v16hi)
17940 v32qi __builtin_ia32_psubusb256 (v32qi,v32qi)
17941 v16hi __builtin_ia32_psubusw256 (v16hi,v16hi)
17942 v32qi __builtin_ia32_punpckhbw256 (v32qi,v32qi)
17943 v16hi __builtin_ia32_punpckhwd256 (v16hi,v16hi)
17944 v8si __builtin_ia32_punpckhdq256 (v8si,v8si)
17945 v4di __builtin_ia32_punpckhqdq256 (v4di,v4di)
17946 v32qi __builtin_ia32_punpcklbw256 (v32qi,v32qi)
17947 v16hi __builtin_ia32_punpcklwd256 (v16hi,v16hi)
17948 v8si __builtin_ia32_punpckldq256 (v8si,v8si)
17949 v4di __builtin_ia32_punpcklqdq256 (v4di,v4di)
17950 v4di __builtin_ia32_pxor256 (v4di,v4di)
17951 v4di __builtin_ia32_movntdqa256 (pv4di)
17952 v4sf __builtin_ia32_vbroadcastss_ps (v4sf)
17953 v8sf __builtin_ia32_vbroadcastss_ps256 (v4sf)
17954 v4df __builtin_ia32_vbroadcastsd_pd256 (v2df)
17955 v4di __builtin_ia32_vbroadcastsi256 (v2di)
17956 v4si __builtin_ia32_pblendd128 (v4si,v4si)
17957 v8si __builtin_ia32_pblendd256 (v8si,v8si)
17958 v32qi __builtin_ia32_pbroadcastb256 (v16qi)
17959 v16hi __builtin_ia32_pbroadcastw256 (v8hi)
17960 v8si __builtin_ia32_pbroadcastd256 (v4si)
17961 v4di __builtin_ia32_pbroadcastq256 (v2di)
17962 v16qi __builtin_ia32_pbroadcastb128 (v16qi)
17963 v8hi __builtin_ia32_pbroadcastw128 (v8hi)
17964 v4si __builtin_ia32_pbroadcastd128 (v4si)
17965 v2di __builtin_ia32_pbroadcastq128 (v2di)
17966 v8si __builtin_ia32_permvarsi256 (v8si,v8si)
17967 v4df __builtin_ia32_permdf256 (v4df,int)
17968 v8sf __builtin_ia32_permvarsf256 (v8sf,v8sf)
17969 v4di __builtin_ia32_permdi256 (v4di,int)
17970 v4di __builtin_ia32_permti256 (v4di,v4di,int)
17971 v4di __builtin_ia32_extract128i256 (v4di,int)
17972 v4di __builtin_ia32_insert128i256 (v4di,v2di,int)
17973 v8si __builtin_ia32_maskloadd256 (pcv8si,v8si)
17974 v4di __builtin_ia32_maskloadq256 (pcv4di,v4di)
17975 v4si __builtin_ia32_maskloadd (pcv4si,v4si)
17976 v2di __builtin_ia32_maskloadq (pcv2di,v2di)
17977 void __builtin_ia32_maskstored256 (pv8si,v8si,v8si)
17978 void __builtin_ia32_maskstoreq256 (pv4di,v4di,v4di)
17979 void __builtin_ia32_maskstored (pv4si,v4si,v4si)
17980 void __builtin_ia32_maskstoreq (pv2di,v2di,v2di)
17981 v8si __builtin_ia32_psllv8si (v8si,v8si)
17982 v4si __builtin_ia32_psllv4si (v4si,v4si)
17983 v4di __builtin_ia32_psllv4di (v4di,v4di)
17984 v2di __builtin_ia32_psllv2di (v2di,v2di)
17985 v8si __builtin_ia32_psrav8si (v8si,v8si)
17986 v4si __builtin_ia32_psrav4si (v4si,v4si)
17987 v8si __builtin_ia32_psrlv8si (v8si,v8si)
17988 v4si __builtin_ia32_psrlv4si (v4si,v4si)
17989 v4di __builtin_ia32_psrlv4di (v4di,v4di)
17990 v2di __builtin_ia32_psrlv2di (v2di,v2di)
17991 v2df __builtin_ia32_gathersiv2df (v2df, pcdouble,v4si,v2df,int)
17992 v4df __builtin_ia32_gathersiv4df (v4df, pcdouble,v4si,v4df,int)
17993 v2df __builtin_ia32_gatherdiv2df (v2df, pcdouble,v2di,v2df,int)
17994 v4df __builtin_ia32_gatherdiv4df (v4df, pcdouble,v4di,v4df,int)
17995 v4sf __builtin_ia32_gathersiv4sf (v4sf, pcfloat,v4si,v4sf,int)
17996 v8sf __builtin_ia32_gathersiv8sf (v8sf, pcfloat,v8si,v8sf,int)
17997 v4sf __builtin_ia32_gatherdiv4sf (v4sf, pcfloat,v2di,v4sf,int)
17998 v4sf __builtin_ia32_gatherdiv4sf256 (v4sf, pcfloat,v4di,v4sf,int)
17999 v2di __builtin_ia32_gathersiv2di (v2di, pcint64,v4si,v2di,int)
18000 v4di __builtin_ia32_gathersiv4di (v4di, pcint64,v4si,v4di,int)
18001 v2di __builtin_ia32_gatherdiv2di (v2di, pcint64,v2di,v2di,int)
18002 v4di __builtin_ia32_gatherdiv4di (v4di, pcint64,v4di,v4di,int)
18003 v4si __builtin_ia32_gathersiv4si (v4si, pcint,v4si,v4si,int)
18004 v8si __builtin_ia32_gathersiv8si (v8si, pcint,v8si,v8si,int)
18005 v4si __builtin_ia32_gatherdiv4si (v4si, pcint,v2di,v4si,int)
18006 v4si __builtin_ia32_gatherdiv4si256 (v4si, pcint,v4di,v4si,int)
18009 The following built-in functions are available when @option{-maes} is
18010 used. All of them generate the machine instruction that is part of the
18014 v2di __builtin_ia32_aesenc128 (v2di, v2di)
18015 v2di __builtin_ia32_aesenclast128 (v2di, v2di)
18016 v2di __builtin_ia32_aesdec128 (v2di, v2di)
18017 v2di __builtin_ia32_aesdeclast128 (v2di, v2di)
18018 v2di __builtin_ia32_aeskeygenassist128 (v2di, const int)
18019 v2di __builtin_ia32_aesimc128 (v2di)
18022 The following built-in function is available when @option{-mpclmul} is
18026 @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int)
18027 Generates the @code{pclmulqdq} machine instruction.
18030 The following built-in function is available when @option{-mfsgsbase} is
18031 used. All of them generate the machine instruction that is part of the
18035 unsigned int __builtin_ia32_rdfsbase32 (void)
18036 unsigned long long __builtin_ia32_rdfsbase64 (void)
18037 unsigned int __builtin_ia32_rdgsbase32 (void)
18038 unsigned long long __builtin_ia32_rdgsbase64 (void)
18039 void _writefsbase_u32 (unsigned int)
18040 void _writefsbase_u64 (unsigned long long)
18041 void _writegsbase_u32 (unsigned int)
18042 void _writegsbase_u64 (unsigned long long)
18045 The following built-in function is available when @option{-mrdrnd} is
18046 used. All of them generate the machine instruction that is part of the
18050 unsigned int __builtin_ia32_rdrand16_step (unsigned short *)
18051 unsigned int __builtin_ia32_rdrand32_step (unsigned int *)
18052 unsigned int __builtin_ia32_rdrand64_step (unsigned long long *)
18055 The following built-in functions are available when @option{-msse4a} is used.
18056 All of them generate the machine instruction that is part of the name.
18059 void __builtin_ia32_movntsd (double *, v2df)
18060 void __builtin_ia32_movntss (float *, v4sf)
18061 v2di __builtin_ia32_extrq (v2di, v16qi)
18062 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
18063 v2di __builtin_ia32_insertq (v2di, v2di)
18064 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
18067 The following built-in functions are available when @option{-mxop} is used.
18069 v2df __builtin_ia32_vfrczpd (v2df)
18070 v4sf __builtin_ia32_vfrczps (v4sf)
18071 v2df __builtin_ia32_vfrczsd (v2df)
18072 v4sf __builtin_ia32_vfrczss (v4sf)
18073 v4df __builtin_ia32_vfrczpd256 (v4df)
18074 v8sf __builtin_ia32_vfrczps256 (v8sf)
18075 v2di __builtin_ia32_vpcmov (v2di, v2di, v2di)
18076 v2di __builtin_ia32_vpcmov_v2di (v2di, v2di, v2di)
18077 v4si __builtin_ia32_vpcmov_v4si (v4si, v4si, v4si)
18078 v8hi __builtin_ia32_vpcmov_v8hi (v8hi, v8hi, v8hi)
18079 v16qi __builtin_ia32_vpcmov_v16qi (v16qi, v16qi, v16qi)
18080 v2df __builtin_ia32_vpcmov_v2df (v2df, v2df, v2df)
18081 v4sf __builtin_ia32_vpcmov_v4sf (v4sf, v4sf, v4sf)
18082 v4di __builtin_ia32_vpcmov_v4di256 (v4di, v4di, v4di)
18083 v8si __builtin_ia32_vpcmov_v8si256 (v8si, v8si, v8si)
18084 v16hi __builtin_ia32_vpcmov_v16hi256 (v16hi, v16hi, v16hi)
18085 v32qi __builtin_ia32_vpcmov_v32qi256 (v32qi, v32qi, v32qi)
18086 v4df __builtin_ia32_vpcmov_v4df256 (v4df, v4df, v4df)
18087 v8sf __builtin_ia32_vpcmov_v8sf256 (v8sf, v8sf, v8sf)
18088 v16qi __builtin_ia32_vpcomeqb (v16qi, v16qi)
18089 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
18090 v4si __builtin_ia32_vpcomeqd (v4si, v4si)
18091 v2di __builtin_ia32_vpcomeqq (v2di, v2di)
18092 v16qi __builtin_ia32_vpcomequb (v16qi, v16qi)
18093 v4si __builtin_ia32_vpcomequd (v4si, v4si)
18094 v2di __builtin_ia32_vpcomequq (v2di, v2di)
18095 v8hi __builtin_ia32_vpcomequw (v8hi, v8hi)
18096 v8hi __builtin_ia32_vpcomeqw (v8hi, v8hi)
18097 v16qi __builtin_ia32_vpcomfalseb (v16qi, v16qi)
18098 v4si __builtin_ia32_vpcomfalsed (v4si, v4si)
18099 v2di __builtin_ia32_vpcomfalseq (v2di, v2di)
18100 v16qi __builtin_ia32_vpcomfalseub (v16qi, v16qi)
18101 v4si __builtin_ia32_vpcomfalseud (v4si, v4si)
18102 v2di __builtin_ia32_vpcomfalseuq (v2di, v2di)
18103 v8hi __builtin_ia32_vpcomfalseuw (v8hi, v8hi)
18104 v8hi __builtin_ia32_vpcomfalsew (v8hi, v8hi)
18105 v16qi __builtin_ia32_vpcomgeb (v16qi, v16qi)
18106 v4si __builtin_ia32_vpcomged (v4si, v4si)
18107 v2di __builtin_ia32_vpcomgeq (v2di, v2di)
18108 v16qi __builtin_ia32_vpcomgeub (v16qi, v16qi)
18109 v4si __builtin_ia32_vpcomgeud (v4si, v4si)
18110 v2di __builtin_ia32_vpcomgeuq (v2di, v2di)
18111 v8hi __builtin_ia32_vpcomgeuw (v8hi, v8hi)
18112 v8hi __builtin_ia32_vpcomgew (v8hi, v8hi)
18113 v16qi __builtin_ia32_vpcomgtb (v16qi, v16qi)
18114 v4si __builtin_ia32_vpcomgtd (v4si, v4si)
18115 v2di __builtin_ia32_vpcomgtq (v2di, v2di)
18116 v16qi __builtin_ia32_vpcomgtub (v16qi, v16qi)
18117 v4si __builtin_ia32_vpcomgtud (v4si, v4si)
18118 v2di __builtin_ia32_vpcomgtuq (v2di, v2di)
18119 v8hi __builtin_ia32_vpcomgtuw (v8hi, v8hi)
18120 v8hi __builtin_ia32_vpcomgtw (v8hi, v8hi)
18121 v16qi __builtin_ia32_vpcomleb (v16qi, v16qi)
18122 v4si __builtin_ia32_vpcomled (v4si, v4si)
18123 v2di __builtin_ia32_vpcomleq (v2di, v2di)
18124 v16qi __builtin_ia32_vpcomleub (v16qi, v16qi)
18125 v4si __builtin_ia32_vpcomleud (v4si, v4si)
18126 v2di __builtin_ia32_vpcomleuq (v2di, v2di)
18127 v8hi __builtin_ia32_vpcomleuw (v8hi, v8hi)
18128 v8hi __builtin_ia32_vpcomlew (v8hi, v8hi)
18129 v16qi __builtin_ia32_vpcomltb (v16qi, v16qi)
18130 v4si __builtin_ia32_vpcomltd (v4si, v4si)
18131 v2di __builtin_ia32_vpcomltq (v2di, v2di)
18132 v16qi __builtin_ia32_vpcomltub (v16qi, v16qi)
18133 v4si __builtin_ia32_vpcomltud (v4si, v4si)
18134 v2di __builtin_ia32_vpcomltuq (v2di, v2di)
18135 v8hi __builtin_ia32_vpcomltuw (v8hi, v8hi)
18136 v8hi __builtin_ia32_vpcomltw (v8hi, v8hi)
18137 v16qi __builtin_ia32_vpcomneb (v16qi, v16qi)
18138 v4si __builtin_ia32_vpcomned (v4si, v4si)
18139 v2di __builtin_ia32_vpcomneq (v2di, v2di)
18140 v16qi __builtin_ia32_vpcomneub (v16qi, v16qi)
18141 v4si __builtin_ia32_vpcomneud (v4si, v4si)
18142 v2di __builtin_ia32_vpcomneuq (v2di, v2di)
18143 v8hi __builtin_ia32_vpcomneuw (v8hi, v8hi)
18144 v8hi __builtin_ia32_vpcomnew (v8hi, v8hi)
18145 v16qi __builtin_ia32_vpcomtrueb (v16qi, v16qi)
18146 v4si __builtin_ia32_vpcomtrued (v4si, v4si)
18147 v2di __builtin_ia32_vpcomtrueq (v2di, v2di)
18148 v16qi __builtin_ia32_vpcomtrueub (v16qi, v16qi)
18149 v4si __builtin_ia32_vpcomtrueud (v4si, v4si)
18150 v2di __builtin_ia32_vpcomtrueuq (v2di, v2di)
18151 v8hi __builtin_ia32_vpcomtrueuw (v8hi, v8hi)
18152 v8hi __builtin_ia32_vpcomtruew (v8hi, v8hi)
18153 v4si __builtin_ia32_vphaddbd (v16qi)
18154 v2di __builtin_ia32_vphaddbq (v16qi)
18155 v8hi __builtin_ia32_vphaddbw (v16qi)
18156 v2di __builtin_ia32_vphadddq (v4si)
18157 v4si __builtin_ia32_vphaddubd (v16qi)
18158 v2di __builtin_ia32_vphaddubq (v16qi)
18159 v8hi __builtin_ia32_vphaddubw (v16qi)
18160 v2di __builtin_ia32_vphaddudq (v4si)
18161 v4si __builtin_ia32_vphadduwd (v8hi)
18162 v2di __builtin_ia32_vphadduwq (v8hi)
18163 v4si __builtin_ia32_vphaddwd (v8hi)
18164 v2di __builtin_ia32_vphaddwq (v8hi)
18165 v8hi __builtin_ia32_vphsubbw (v16qi)
18166 v2di __builtin_ia32_vphsubdq (v4si)
18167 v4si __builtin_ia32_vphsubwd (v8hi)
18168 v4si __builtin_ia32_vpmacsdd (v4si, v4si, v4si)
18169 v2di __builtin_ia32_vpmacsdqh (v4si, v4si, v2di)
18170 v2di __builtin_ia32_vpmacsdql (v4si, v4si, v2di)
18171 v4si __builtin_ia32_vpmacssdd (v4si, v4si, v4si)
18172 v2di __builtin_ia32_vpmacssdqh (v4si, v4si, v2di)
18173 v2di __builtin_ia32_vpmacssdql (v4si, v4si, v2di)
18174 v4si __builtin_ia32_vpmacsswd (v8hi, v8hi, v4si)
18175 v8hi __builtin_ia32_vpmacssww (v8hi, v8hi, v8hi)
18176 v4si __builtin_ia32_vpmacswd (v8hi, v8hi, v4si)
18177 v8hi __builtin_ia32_vpmacsww (v8hi, v8hi, v8hi)
18178 v4si __builtin_ia32_vpmadcsswd (v8hi, v8hi, v4si)
18179 v4si __builtin_ia32_vpmadcswd (v8hi, v8hi, v4si)
18180 v16qi __builtin_ia32_vpperm (v16qi, v16qi, v16qi)
18181 v16qi __builtin_ia32_vprotb (v16qi, v16qi)
18182 v4si __builtin_ia32_vprotd (v4si, v4si)
18183 v2di __builtin_ia32_vprotq (v2di, v2di)
18184 v8hi __builtin_ia32_vprotw (v8hi, v8hi)
18185 v16qi __builtin_ia32_vpshab (v16qi, v16qi)
18186 v4si __builtin_ia32_vpshad (v4si, v4si)
18187 v2di __builtin_ia32_vpshaq (v2di, v2di)
18188 v8hi __builtin_ia32_vpshaw (v8hi, v8hi)
18189 v16qi __builtin_ia32_vpshlb (v16qi, v16qi)
18190 v4si __builtin_ia32_vpshld (v4si, v4si)
18191 v2di __builtin_ia32_vpshlq (v2di, v2di)
18192 v8hi __builtin_ia32_vpshlw (v8hi, v8hi)
18195 The following built-in functions are available when @option{-mfma4} is used.
18196 All of them generate the machine instruction that is part of the name.
18199 v2df __builtin_ia32_vfmaddpd (v2df, v2df, v2df)
18200 v4sf __builtin_ia32_vfmaddps (v4sf, v4sf, v4sf)
18201 v2df __builtin_ia32_vfmaddsd (v2df, v2df, v2df)
18202 v4sf __builtin_ia32_vfmaddss (v4sf, v4sf, v4sf)
18203 v2df __builtin_ia32_vfmsubpd (v2df, v2df, v2df)
18204 v4sf __builtin_ia32_vfmsubps (v4sf, v4sf, v4sf)
18205 v2df __builtin_ia32_vfmsubsd (v2df, v2df, v2df)
18206 v4sf __builtin_ia32_vfmsubss (v4sf, v4sf, v4sf)
18207 v2df __builtin_ia32_vfnmaddpd (v2df, v2df, v2df)
18208 v4sf __builtin_ia32_vfnmaddps (v4sf, v4sf, v4sf)
18209 v2df __builtin_ia32_vfnmaddsd (v2df, v2df, v2df)
18210 v4sf __builtin_ia32_vfnmaddss (v4sf, v4sf, v4sf)
18211 v2df __builtin_ia32_vfnmsubpd (v2df, v2df, v2df)
18212 v4sf __builtin_ia32_vfnmsubps (v4sf, v4sf, v4sf)
18213 v2df __builtin_ia32_vfnmsubsd (v2df, v2df, v2df)
18214 v4sf __builtin_ia32_vfnmsubss (v4sf, v4sf, v4sf)
18215 v2df __builtin_ia32_vfmaddsubpd (v2df, v2df, v2df)
18216 v4sf __builtin_ia32_vfmaddsubps (v4sf, v4sf, v4sf)
18217 v2df __builtin_ia32_vfmsubaddpd (v2df, v2df, v2df)
18218 v4sf __builtin_ia32_vfmsubaddps (v4sf, v4sf, v4sf)
18219 v4df __builtin_ia32_vfmaddpd256 (v4df, v4df, v4df)
18220 v8sf __builtin_ia32_vfmaddps256 (v8sf, v8sf, v8sf)
18221 v4df __builtin_ia32_vfmsubpd256 (v4df, v4df, v4df)
18222 v8sf __builtin_ia32_vfmsubps256 (v8sf, v8sf, v8sf)
18223 v4df __builtin_ia32_vfnmaddpd256 (v4df, v4df, v4df)
18224 v8sf __builtin_ia32_vfnmaddps256 (v8sf, v8sf, v8sf)
18225 v4df __builtin_ia32_vfnmsubpd256 (v4df, v4df, v4df)
18226 v8sf __builtin_ia32_vfnmsubps256 (v8sf, v8sf, v8sf)
18227 v4df __builtin_ia32_vfmaddsubpd256 (v4df, v4df, v4df)
18228 v8sf __builtin_ia32_vfmaddsubps256 (v8sf, v8sf, v8sf)
18229 v4df __builtin_ia32_vfmsubaddpd256 (v4df, v4df, v4df)
18230 v8sf __builtin_ia32_vfmsubaddps256 (v8sf, v8sf, v8sf)
18234 The following built-in functions are available when @option{-mlwp} is used.
18237 void __builtin_ia32_llwpcb16 (void *);
18238 void __builtin_ia32_llwpcb32 (void *);
18239 void __builtin_ia32_llwpcb64 (void *);
18240 void * __builtin_ia32_llwpcb16 (void);
18241 void * __builtin_ia32_llwpcb32 (void);
18242 void * __builtin_ia32_llwpcb64 (void);
18243 void __builtin_ia32_lwpval16 (unsigned short, unsigned int, unsigned short)
18244 void __builtin_ia32_lwpval32 (unsigned int, unsigned int, unsigned int)
18245 void __builtin_ia32_lwpval64 (unsigned __int64, unsigned int, unsigned int)
18246 unsigned char __builtin_ia32_lwpins16 (unsigned short, unsigned int, unsigned short)
18247 unsigned char __builtin_ia32_lwpins32 (unsigned int, unsigned int, unsigned int)
18248 unsigned char __builtin_ia32_lwpins64 (unsigned __int64, unsigned int, unsigned int)
18251 The following built-in functions are available when @option{-mbmi} is used.
18252 All of them generate the machine instruction that is part of the name.
18254 unsigned int __builtin_ia32_bextr_u32(unsigned int, unsigned int);
18255 unsigned long long __builtin_ia32_bextr_u64 (unsigned long long, unsigned long long);
18258 The following built-in functions are available when @option{-mbmi2} is used.
18259 All of them generate the machine instruction that is part of the name.
18261 unsigned int _bzhi_u32 (unsigned int, unsigned int)
18262 unsigned int _pdep_u32 (unsigned int, unsigned int)
18263 unsigned int _pext_u32 (unsigned int, unsigned int)
18264 unsigned long long _bzhi_u64 (unsigned long long, unsigned long long)
18265 unsigned long long _pdep_u64 (unsigned long long, unsigned long long)
18266 unsigned long long _pext_u64 (unsigned long long, unsigned long long)
18269 The following built-in functions are available when @option{-mlzcnt} is used.
18270 All of them generate the machine instruction that is part of the name.
18272 unsigned short __builtin_ia32_lzcnt_16(unsigned short);
18273 unsigned int __builtin_ia32_lzcnt_u32(unsigned int);
18274 unsigned long long __builtin_ia32_lzcnt_u64 (unsigned long long);
18277 The following built-in functions are available when @option{-mfxsr} is used.
18278 All of them generate the machine instruction that is part of the name.
18280 void __builtin_ia32_fxsave (void *)
18281 void __builtin_ia32_fxrstor (void *)
18282 void __builtin_ia32_fxsave64 (void *)
18283 void __builtin_ia32_fxrstor64 (void *)
18286 The following built-in functions are available when @option{-mxsave} is used.
18287 All of them generate the machine instruction that is part of the name.
18289 void __builtin_ia32_xsave (void *, long long)
18290 void __builtin_ia32_xrstor (void *, long long)
18291 void __builtin_ia32_xsave64 (void *, long long)
18292 void __builtin_ia32_xrstor64 (void *, long long)
18295 The following built-in functions are available when @option{-mxsaveopt} is used.
18296 All of them generate the machine instruction that is part of the name.
18298 void __builtin_ia32_xsaveopt (void *, long long)
18299 void __builtin_ia32_xsaveopt64 (void *, long long)
18302 The following built-in functions are available when @option{-mtbm} is used.
18303 Both of them generate the immediate form of the bextr machine instruction.
18305 unsigned int __builtin_ia32_bextri_u32 (unsigned int, const unsigned int);
18306 unsigned long long __builtin_ia32_bextri_u64 (unsigned long long, const unsigned long long);
18310 The following built-in functions are available when @option{-m3dnow} is used.
18311 All of them generate the machine instruction that is part of the name.
18314 void __builtin_ia32_femms (void)
18315 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
18316 v2si __builtin_ia32_pf2id (v2sf)
18317 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
18318 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
18319 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
18320 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
18321 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
18322 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
18323 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
18324 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
18325 v2sf __builtin_ia32_pfrcp (v2sf)
18326 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
18327 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
18328 v2sf __builtin_ia32_pfrsqrt (v2sf)
18329 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
18330 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
18331 v2sf __builtin_ia32_pi2fd (v2si)
18332 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
18335 The following built-in functions are available when both @option{-m3dnow}
18336 and @option{-march=athlon} are used. All of them generate the machine
18337 instruction that is part of the name.
18340 v2si __builtin_ia32_pf2iw (v2sf)
18341 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
18342 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
18343 v2sf __builtin_ia32_pi2fw (v2si)
18344 v2sf __builtin_ia32_pswapdsf (v2sf)
18345 v2si __builtin_ia32_pswapdsi (v2si)
18348 The following built-in functions are available when @option{-mrtm} is used
18349 They are used for restricted transactional memory. These are the internal
18350 low level functions. Normally the functions in
18351 @ref{x86 transactional memory intrinsics} should be used instead.
18354 int __builtin_ia32_xbegin ()
18355 void __builtin_ia32_xend ()
18356 void __builtin_ia32_xabort (status)
18357 int __builtin_ia32_xtest ()
18360 The following built-in functions are available when @option{-mmwaitx} is used.
18361 All of them generate the machine instruction that is part of the name.
18363 void __builtin_ia32_monitorx (void *, unsigned int, unsigned int)
18364 void __builtin_ia32_mwaitx (unsigned int, unsigned int, unsigned int)
18367 @node x86 transactional memory intrinsics
18368 @subsection x86 Transactional Memory Intrinsics
18370 These hardware transactional memory intrinsics for x86 allow you to use
18371 memory transactions with RTM (Restricted Transactional Memory).
18372 This support is enabled with the @option{-mrtm} option.
18373 For using HLE (Hardware Lock Elision) see
18374 @ref{x86 specific memory model extensions for transactional memory} instead.
18376 A memory transaction commits all changes to memory in an atomic way,
18377 as visible to other threads. If the transaction fails it is rolled back
18378 and all side effects discarded.
18380 Generally there is no guarantee that a memory transaction ever succeeds
18381 and suitable fallback code always needs to be supplied.
18383 @deftypefn {RTM Function} {unsigned} _xbegin ()
18384 Start a RTM (Restricted Transactional Memory) transaction.
18385 Returns @code{_XBEGIN_STARTED} when the transaction
18386 started successfully (note this is not 0, so the constant has to be
18387 explicitly tested).
18389 If the transaction aborts, all side-effects
18390 are undone and an abort code encoded as a bit mask is returned.
18391 The following macros are defined:
18394 @item _XABORT_EXPLICIT
18395 Transaction was explicitly aborted with @code{_xabort}. The parameter passed
18396 to @code{_xabort} is available with @code{_XABORT_CODE(status)}.
18397 @item _XABORT_RETRY
18398 Transaction retry is possible.
18399 @item _XABORT_CONFLICT
18400 Transaction abort due to a memory conflict with another thread.
18401 @item _XABORT_CAPACITY
18402 Transaction abort due to the transaction using too much memory.
18403 @item _XABORT_DEBUG
18404 Transaction abort due to a debug trap.
18405 @item _XABORT_NESTED
18406 Transaction abort in an inner nested transaction.
18409 There is no guarantee
18410 any transaction ever succeeds, so there always needs to be a valid
18414 @deftypefn {RTM Function} {void} _xend ()
18415 Commit the current transaction. When no transaction is active this faults.
18416 All memory side-effects of the transaction become visible
18417 to other threads in an atomic manner.
18420 @deftypefn {RTM Function} {int} _xtest ()
18421 Return a nonzero value if a transaction is currently active, otherwise 0.
18424 @deftypefn {RTM Function} {void} _xabort (status)
18425 Abort the current transaction. When no transaction is active this is a no-op.
18426 The @var{status} is an 8-bit constant; its value is encoded in the return
18427 value from @code{_xbegin}.
18430 Here is an example showing handling for @code{_XABORT_RETRY}
18431 and a fallback path for other failures:
18434 #include <immintrin.h>
18436 int n_tries, max_tries;
18437 unsigned status = _XABORT_EXPLICIT;
18440 for (n_tries = 0; n_tries < max_tries; n_tries++)
18442 status = _xbegin ();
18443 if (status == _XBEGIN_STARTED || !(status & _XABORT_RETRY))
18446 if (status == _XBEGIN_STARTED)
18448 ... transaction code...
18453 ... non-transactional fallback path...
18458 Note that, in most cases, the transactional and non-transactional code
18459 must synchronize together to ensure consistency.
18461 @node Target Format Checks
18462 @section Format Checks Specific to Particular Target Machines
18464 For some target machines, GCC supports additional options to the
18466 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
18469 * Solaris Format Checks::
18470 * Darwin Format Checks::
18473 @node Solaris Format Checks
18474 @subsection Solaris Format Checks
18476 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
18477 check. @code{cmn_err} accepts a subset of the standard @code{printf}
18478 conversions, and the two-argument @code{%b} conversion for displaying
18479 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
18481 @node Darwin Format Checks
18482 @subsection Darwin Format Checks
18484 Darwin targets support the @code{CFString} (or @code{__CFString__}) in the format
18485 attribute context. Declarations made with such attribution are parsed for correct syntax
18486 and format argument types. However, parsing of the format string itself is currently undefined
18487 and is not carried out by this version of the compiler.
18489 Additionally, @code{CFStringRefs} (defined by the @code{CoreFoundation} headers) may
18490 also be used as format arguments. Note that the relevant headers are only likely to be
18491 available on Darwin (OSX) installations. On such installations, the XCode and system
18492 documentation provide descriptions of @code{CFString}, @code{CFStringRefs} and
18493 associated functions.
18496 @section Pragmas Accepted by GCC
18498 @cindex @code{#pragma}
18500 GCC supports several types of pragmas, primarily in order to compile
18501 code originally written for other compilers. Note that in general
18502 we do not recommend the use of pragmas; @xref{Function Attributes},
18503 for further explanation.
18506 * AArch64 Pragmas::
18510 * RS/6000 and PowerPC Pragmas::
18512 * Solaris Pragmas::
18513 * Symbol-Renaming Pragmas::
18514 * Structure-Layout Pragmas::
18516 * Diagnostic Pragmas::
18517 * Visibility Pragmas::
18518 * Push/Pop Macro Pragmas::
18519 * Function Specific Option Pragmas::
18520 * Loop-Specific Pragmas::
18523 @node AArch64 Pragmas
18524 @subsection AArch64 Pragmas
18526 The pragmas defined by the AArch64 target correspond to the AArch64
18527 target function attributes. They can be specified as below:
18529 #pragma GCC target("string")
18532 where @code{@var{string}} can be any string accepted as an AArch64 target
18533 attribute. @xref{AArch64 Function Attributes}, for more details
18534 on the permissible values of @code{string}.
18537 @subsection ARM Pragmas
18539 The ARM target defines pragmas for controlling the default addition of
18540 @code{long_call} and @code{short_call} attributes to functions.
18541 @xref{Function Attributes}, for information about the effects of these
18546 @cindex pragma, long_calls
18547 Set all subsequent functions to have the @code{long_call} attribute.
18549 @item no_long_calls
18550 @cindex pragma, no_long_calls
18551 Set all subsequent functions to have the @code{short_call} attribute.
18553 @item long_calls_off
18554 @cindex pragma, long_calls_off
18555 Do not affect the @code{long_call} or @code{short_call} attributes of
18556 subsequent functions.
18560 @subsection M32C Pragmas
18563 @item GCC memregs @var{number}
18564 @cindex pragma, memregs
18565 Overrides the command-line option @code{-memregs=} for the current
18566 file. Use with care! This pragma must be before any function in the
18567 file, and mixing different memregs values in different objects may
18568 make them incompatible. This pragma is useful when a
18569 performance-critical function uses a memreg for temporary values,
18570 as it may allow you to reduce the number of memregs used.
18572 @item ADDRESS @var{name} @var{address}
18573 @cindex pragma, address
18574 For any declared symbols matching @var{name}, this does three things
18575 to that symbol: it forces the symbol to be located at the given
18576 address (a number), it forces the symbol to be volatile, and it
18577 changes the symbol's scope to be static. This pragma exists for
18578 compatibility with other compilers, but note that the common
18579 @code{1234H} numeric syntax is not supported (use @code{0x1234}
18583 #pragma ADDRESS port3 0x103
18590 @subsection MeP Pragmas
18594 @item custom io_volatile (on|off)
18595 @cindex pragma, custom io_volatile
18596 Overrides the command-line option @code{-mio-volatile} for the current
18597 file. Note that for compatibility with future GCC releases, this
18598 option should only be used once before any @code{io} variables in each
18601 @item GCC coprocessor available @var{registers}
18602 @cindex pragma, coprocessor available
18603 Specifies which coprocessor registers are available to the register
18604 allocator. @var{registers} may be a single register, register range
18605 separated by ellipses, or comma-separated list of those. Example:
18608 #pragma GCC coprocessor available $c0...$c10, $c28
18611 @item GCC coprocessor call_saved @var{registers}
18612 @cindex pragma, coprocessor call_saved
18613 Specifies which coprocessor registers are to be saved and restored by
18614 any function using them. @var{registers} may be a single register,
18615 register range separated by ellipses, or comma-separated list of
18619 #pragma GCC coprocessor call_saved $c4...$c6, $c31
18622 @item GCC coprocessor subclass '(A|B|C|D)' = @var{registers}
18623 @cindex pragma, coprocessor subclass
18624 Creates and defines a register class. These register classes can be
18625 used by inline @code{asm} constructs. @var{registers} may be a single
18626 register, register range separated by ellipses, or comma-separated
18627 list of those. Example:
18630 #pragma GCC coprocessor subclass 'B' = $c2, $c4, $c6
18632 asm ("cpfoo %0" : "=B" (x));
18635 @item GCC disinterrupt @var{name} , @var{name} @dots{}
18636 @cindex pragma, disinterrupt
18637 For the named functions, the compiler adds code to disable interrupts
18638 for the duration of those functions. If any functions so named
18639 are not encountered in the source, a warning is emitted that the pragma is
18640 not used. Examples:
18643 #pragma disinterrupt foo
18644 #pragma disinterrupt bar, grill
18645 int foo () @{ @dots{} @}
18648 @item GCC call @var{name} , @var{name} @dots{}
18649 @cindex pragma, call
18650 For the named functions, the compiler always uses a register-indirect
18651 call model when calling the named functions. Examples:
18660 @node RS/6000 and PowerPC Pragmas
18661 @subsection RS/6000 and PowerPC Pragmas
18663 The RS/6000 and PowerPC targets define one pragma for controlling
18664 whether or not the @code{longcall} attribute is added to function
18665 declarations by default. This pragma overrides the @option{-mlongcall}
18666 option, but not the @code{longcall} and @code{shortcall} attributes.
18667 @xref{RS/6000 and PowerPC Options}, for more information about when long
18668 calls are and are not necessary.
18672 @cindex pragma, longcall
18673 Apply the @code{longcall} attribute to all subsequent function
18677 Do not apply the @code{longcall} attribute to subsequent function
18681 @c Describe h8300 pragmas here.
18682 @c Describe sh pragmas here.
18683 @c Describe v850 pragmas here.
18685 @node Darwin Pragmas
18686 @subsection Darwin Pragmas
18688 The following pragmas are available for all architectures running the
18689 Darwin operating system. These are useful for compatibility with other
18693 @item mark @var{tokens}@dots{}
18694 @cindex pragma, mark
18695 This pragma is accepted, but has no effect.
18697 @item options align=@var{alignment}
18698 @cindex pragma, options align
18699 This pragma sets the alignment of fields in structures. The values of
18700 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
18701 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
18702 properly; to restore the previous setting, use @code{reset} for the
18705 @item segment @var{tokens}@dots{}
18706 @cindex pragma, segment
18707 This pragma is accepted, but has no effect.
18709 @item unused (@var{var} [, @var{var}]@dots{})
18710 @cindex pragma, unused
18711 This pragma declares variables to be possibly unused. GCC does not
18712 produce warnings for the listed variables. The effect is similar to
18713 that of the @code{unused} attribute, except that this pragma may appear
18714 anywhere within the variables' scopes.
18717 @node Solaris Pragmas
18718 @subsection Solaris Pragmas
18720 The Solaris target supports @code{#pragma redefine_extname}
18721 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
18722 @code{#pragma} directives for compatibility with the system compiler.
18725 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
18726 @cindex pragma, align
18728 Increase the minimum alignment of each @var{variable} to @var{alignment}.
18729 This is the same as GCC's @code{aligned} attribute @pxref{Variable
18730 Attributes}). Macro expansion occurs on the arguments to this pragma
18731 when compiling C and Objective-C@. It does not currently occur when
18732 compiling C++, but this is a bug which may be fixed in a future
18735 @item fini (@var{function} [, @var{function}]...)
18736 @cindex pragma, fini
18738 This pragma causes each listed @var{function} to be called after
18739 main, or during shared module unloading, by adding a call to the
18740 @code{.fini} section.
18742 @item init (@var{function} [, @var{function}]...)
18743 @cindex pragma, init
18745 This pragma causes each listed @var{function} to be called during
18746 initialization (before @code{main}) or during shared module loading, by
18747 adding a call to the @code{.init} section.
18751 @node Symbol-Renaming Pragmas
18752 @subsection Symbol-Renaming Pragmas
18754 GCC supports a @code{#pragma} directive that changes the name used in
18755 assembly for a given declaration. While this pragma is supported on all
18756 platforms, it is intended primarily to provide compatibility with the
18757 Solaris system headers. This effect can also be achieved using the asm
18758 labels extension (@pxref{Asm Labels}).
18761 @item redefine_extname @var{oldname} @var{newname}
18762 @cindex pragma, redefine_extname
18764 This pragma gives the C function @var{oldname} the assembly symbol
18765 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
18766 is defined if this pragma is available (currently on all platforms).
18769 This pragma and the asm labels extension interact in a complicated
18770 manner. Here are some corner cases you may want to be aware of:
18773 @item This pragma silently applies only to declarations with external
18774 linkage. Asm labels do not have this restriction.
18776 @item In C++, this pragma silently applies only to declarations with
18777 ``C'' linkage. Again, asm labels do not have this restriction.
18779 @item If either of the ways of changing the assembly name of a
18780 declaration are applied to a declaration whose assembly name has
18781 already been determined (either by a previous use of one of these
18782 features, or because the compiler needed the assembly name in order to
18783 generate code), and the new name is different, a warning issues and
18784 the name does not change.
18786 @item The @var{oldname} used by @code{#pragma redefine_extname} is
18787 always the C-language name.
18790 @node Structure-Layout Pragmas
18791 @subsection Structure-Layout Pragmas
18793 For compatibility with Microsoft Windows compilers, GCC supports a
18794 set of @code{#pragma} directives that change the maximum alignment of
18795 members of structures (other than zero-width bit-fields), unions, and
18796 classes subsequently defined. The @var{n} value below always is required
18797 to be a small power of two and specifies the new alignment in bytes.
18800 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
18801 @item @code{#pragma pack()} sets the alignment to the one that was in
18802 effect when compilation started (see also command-line option
18803 @option{-fpack-struct[=@var{n}]} @pxref{Code Gen Options}).
18804 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
18805 setting on an internal stack and then optionally sets the new alignment.
18806 @item @code{#pragma pack(pop)} restores the alignment setting to the one
18807 saved at the top of the internal stack (and removes that stack entry).
18808 Note that @code{#pragma pack([@var{n}])} does not influence this internal
18809 stack; thus it is possible to have @code{#pragma pack(push)} followed by
18810 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
18811 @code{#pragma pack(pop)}.
18814 Some targets, e.g.@: x86 and PowerPC, support the @code{#pragma ms_struct}
18815 directive which lays out structures and unions subsequently defined as the
18816 documented @code{__attribute__ ((ms_struct))}.
18819 @item @code{#pragma ms_struct on} turns on the Microsoft layout.
18820 @item @code{#pragma ms_struct off} turns off the Microsoft layout.
18821 @item @code{#pragma ms_struct reset} goes back to the default layout.
18824 Most targets also support the @code{#pragma scalar_storage_order} directive
18825 which lays out structures and unions subsequently defined as the documented
18826 @code{__attribute__ ((scalar_storage_order))}.
18829 @item @code{#pragma scalar_storage_order big-endian} sets the storage order
18830 of the scalar fields to big-endian.
18831 @item @code{#pragma scalar_storage_order little-endian} sets the storage order
18832 of the scalar fields to little-endian.
18833 @item @code{#pragma scalar_storage_order default} goes back to the endianness
18834 that was in effect when compilation started (see also command-line option
18835 @option{-fsso-struct=@var{endianness}} @pxref{C Dialect Options}).
18839 @subsection Weak Pragmas
18841 For compatibility with SVR4, GCC supports a set of @code{#pragma}
18842 directives for declaring symbols to be weak, and defining weak
18846 @item #pragma weak @var{symbol}
18847 @cindex pragma, weak
18848 This pragma declares @var{symbol} to be weak, as if the declaration
18849 had the attribute of the same name. The pragma may appear before
18850 or after the declaration of @var{symbol}. It is not an error for
18851 @var{symbol} to never be defined at all.
18853 @item #pragma weak @var{symbol1} = @var{symbol2}
18854 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
18855 It is an error if @var{symbol2} is not defined in the current
18859 @node Diagnostic Pragmas
18860 @subsection Diagnostic Pragmas
18862 GCC allows the user to selectively enable or disable certain types of
18863 diagnostics, and change the kind of the diagnostic. For example, a
18864 project's policy might require that all sources compile with
18865 @option{-Werror} but certain files might have exceptions allowing
18866 specific types of warnings. Or, a project might selectively enable
18867 diagnostics and treat them as errors depending on which preprocessor
18868 macros are defined.
18871 @item #pragma GCC diagnostic @var{kind} @var{option}
18872 @cindex pragma, diagnostic
18874 Modifies the disposition of a diagnostic. Note that not all
18875 diagnostics are modifiable; at the moment only warnings (normally
18876 controlled by @samp{-W@dots{}}) can be controlled, and not all of them.
18877 Use @option{-fdiagnostics-show-option} to determine which diagnostics
18878 are controllable and which option controls them.
18880 @var{kind} is @samp{error} to treat this diagnostic as an error,
18881 @samp{warning} to treat it like a warning (even if @option{-Werror} is
18882 in effect), or @samp{ignored} if the diagnostic is to be ignored.
18883 @var{option} is a double quoted string that matches the command-line
18887 #pragma GCC diagnostic warning "-Wformat"
18888 #pragma GCC diagnostic error "-Wformat"
18889 #pragma GCC diagnostic ignored "-Wformat"
18892 Note that these pragmas override any command-line options. GCC keeps
18893 track of the location of each pragma, and issues diagnostics according
18894 to the state as of that point in the source file. Thus, pragmas occurring
18895 after a line do not affect diagnostics caused by that line.
18897 @item #pragma GCC diagnostic push
18898 @itemx #pragma GCC diagnostic pop
18900 Causes GCC to remember the state of the diagnostics as of each
18901 @code{push}, and restore to that point at each @code{pop}. If a
18902 @code{pop} has no matching @code{push}, the command-line options are
18906 #pragma GCC diagnostic error "-Wuninitialized"
18907 foo(a); /* error is given for this one */
18908 #pragma GCC diagnostic push
18909 #pragma GCC diagnostic ignored "-Wuninitialized"
18910 foo(b); /* no diagnostic for this one */
18911 #pragma GCC diagnostic pop
18912 foo(c); /* error is given for this one */
18913 #pragma GCC diagnostic pop
18914 foo(d); /* depends on command-line options */
18919 GCC also offers a simple mechanism for printing messages during
18923 @item #pragma message @var{string}
18924 @cindex pragma, diagnostic
18926 Prints @var{string} as a compiler message on compilation. The message
18927 is informational only, and is neither a compilation warning nor an error.
18930 #pragma message "Compiling " __FILE__ "..."
18933 @var{string} may be parenthesized, and is printed with location
18934 information. For example,
18937 #define DO_PRAGMA(x) _Pragma (#x)
18938 #define TODO(x) DO_PRAGMA(message ("TODO - " #x))
18940 TODO(Remember to fix this)
18944 prints @samp{/tmp/file.c:4: note: #pragma message:
18945 TODO - Remember to fix this}.
18949 @node Visibility Pragmas
18950 @subsection Visibility Pragmas
18953 @item #pragma GCC visibility push(@var{visibility})
18954 @itemx #pragma GCC visibility pop
18955 @cindex pragma, visibility
18957 This pragma allows the user to set the visibility for multiple
18958 declarations without having to give each a visibility attribute
18959 (@pxref{Function Attributes}).
18961 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
18962 declarations. Class members and template specializations are not
18963 affected; if you want to override the visibility for a particular
18964 member or instantiation, you must use an attribute.
18969 @node Push/Pop Macro Pragmas
18970 @subsection Push/Pop Macro Pragmas
18972 For compatibility with Microsoft Windows compilers, GCC supports
18973 @samp{#pragma push_macro(@var{"macro_name"})}
18974 and @samp{#pragma pop_macro(@var{"macro_name"})}.
18977 @item #pragma push_macro(@var{"macro_name"})
18978 @cindex pragma, push_macro
18979 This pragma saves the value of the macro named as @var{macro_name} to
18980 the top of the stack for this macro.
18982 @item #pragma pop_macro(@var{"macro_name"})
18983 @cindex pragma, pop_macro
18984 This pragma sets the value of the macro named as @var{macro_name} to
18985 the value on top of the stack for this macro. If the stack for
18986 @var{macro_name} is empty, the value of the macro remains unchanged.
18993 #pragma push_macro("X")
18996 #pragma pop_macro("X")
19001 In this example, the definition of X as 1 is saved by @code{#pragma
19002 push_macro} and restored by @code{#pragma pop_macro}.
19004 @node Function Specific Option Pragmas
19005 @subsection Function Specific Option Pragmas
19008 @item #pragma GCC target (@var{"string"}...)
19009 @cindex pragma GCC target
19011 This pragma allows you to set target specific options for functions
19012 defined later in the source file. One or more strings can be
19013 specified. Each function that is defined after this point is as
19014 if @code{attribute((target("STRING")))} was specified for that
19015 function. The parenthesis around the options is optional.
19016 @xref{Function Attributes}, for more information about the
19017 @code{target} attribute and the attribute syntax.
19019 The @code{#pragma GCC target} pragma is presently implemented for
19020 x86, PowerPC, and Nios II targets only.
19024 @item #pragma GCC optimize (@var{"string"}...)
19025 @cindex pragma GCC optimize
19027 This pragma allows you to set global optimization options for functions
19028 defined later in the source file. One or more strings can be
19029 specified. Each function that is defined after this point is as
19030 if @code{attribute((optimize("STRING")))} was specified for that
19031 function. The parenthesis around the options is optional.
19032 @xref{Function Attributes}, for more information about the
19033 @code{optimize} attribute and the attribute syntax.
19037 @item #pragma GCC push_options
19038 @itemx #pragma GCC pop_options
19039 @cindex pragma GCC push_options
19040 @cindex pragma GCC pop_options
19042 These pragmas maintain a stack of the current target and optimization
19043 options. It is intended for include files where you temporarily want
19044 to switch to using a different @samp{#pragma GCC target} or
19045 @samp{#pragma GCC optimize} and then to pop back to the previous
19050 @item #pragma GCC reset_options
19051 @cindex pragma GCC reset_options
19053 This pragma clears the current @code{#pragma GCC target} and
19054 @code{#pragma GCC optimize} to use the default switches as specified
19055 on the command line.
19058 @node Loop-Specific Pragmas
19059 @subsection Loop-Specific Pragmas
19062 @item #pragma GCC ivdep
19063 @cindex pragma GCC ivdep
19066 With this pragma, the programmer asserts that there are no loop-carried
19067 dependencies which would prevent consecutive iterations of
19068 the following loop from executing concurrently with SIMD
19069 (single instruction multiple data) instructions.
19071 For example, the compiler can only unconditionally vectorize the following
19072 loop with the pragma:
19075 void foo (int n, int *a, int *b, int *c)
19079 for (i = 0; i < n; ++i)
19080 a[i] = b[i] + c[i];
19085 In this example, using the @code{restrict} qualifier had the same
19086 effect. In the following example, that would not be possible. Assume
19087 @math{k < -m} or @math{k >= m}. Only with the pragma, the compiler knows
19088 that it can unconditionally vectorize the following loop:
19091 void ignore_vec_dep (int *a, int k, int c, int m)
19094 for (int i = 0; i < m; i++)
19095 a[i] = a[i + k] * c;
19100 @node Unnamed Fields
19101 @section Unnamed Structure and Union Fields
19102 @cindex @code{struct}
19103 @cindex @code{union}
19105 As permitted by ISO C11 and for compatibility with other compilers,
19106 GCC allows you to define
19107 a structure or union that contains, as fields, structures and unions
19108 without names. For example:
19122 In this example, you are able to access members of the unnamed
19123 union with code like @samp{foo.b}. Note that only unnamed structs and
19124 unions are allowed, you may not have, for example, an unnamed
19127 You must never create such structures that cause ambiguous field definitions.
19128 For example, in this structure:
19140 it is ambiguous which @code{a} is being referred to with @samp{foo.a}.
19141 The compiler gives errors for such constructs.
19143 @opindex fms-extensions
19144 Unless @option{-fms-extensions} is used, the unnamed field must be a
19145 structure or union definition without a tag (for example, @samp{struct
19146 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
19147 also be a definition with a tag such as @samp{struct foo @{ int a;
19148 @};}, a reference to a previously defined structure or union such as
19149 @samp{struct foo;}, or a reference to a @code{typedef} name for a
19150 previously defined structure or union type.
19152 @opindex fplan9-extensions
19153 The option @option{-fplan9-extensions} enables
19154 @option{-fms-extensions} as well as two other extensions. First, a
19155 pointer to a structure is automatically converted to a pointer to an
19156 anonymous field for assignments and function calls. For example:
19159 struct s1 @{ int a; @};
19160 struct s2 @{ struct s1; @};
19161 extern void f1 (struct s1 *);
19162 void f2 (struct s2 *p) @{ f1 (p); @}
19166 In the call to @code{f1} inside @code{f2}, the pointer @code{p} is
19167 converted into a pointer to the anonymous field.
19169 Second, when the type of an anonymous field is a @code{typedef} for a
19170 @code{struct} or @code{union}, code may refer to the field using the
19171 name of the @code{typedef}.
19174 typedef struct @{ int a; @} s1;
19175 struct s2 @{ s1; @};
19176 s1 f1 (struct s2 *p) @{ return p->s1; @}
19179 These usages are only permitted when they are not ambiguous.
19182 @section Thread-Local Storage
19183 @cindex Thread-Local Storage
19184 @cindex @acronym{TLS}
19185 @cindex @code{__thread}
19187 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
19188 are allocated such that there is one instance of the variable per extant
19189 thread. The runtime model GCC uses to implement this originates
19190 in the IA-64 processor-specific ABI, but has since been migrated
19191 to other processors as well. It requires significant support from
19192 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
19193 system libraries (@file{libc.so} and @file{libpthread.so}), so it
19194 is not available everywhere.
19196 At the user level, the extension is visible with a new storage
19197 class keyword: @code{__thread}. For example:
19201 extern __thread struct state s;
19202 static __thread char *p;
19205 The @code{__thread} specifier may be used alone, with the @code{extern}
19206 or @code{static} specifiers, but with no other storage class specifier.
19207 When used with @code{extern} or @code{static}, @code{__thread} must appear
19208 immediately after the other storage class specifier.
19210 The @code{__thread} specifier may be applied to any global, file-scoped
19211 static, function-scoped static, or static data member of a class. It may
19212 not be applied to block-scoped automatic or non-static data member.
19214 When the address-of operator is applied to a thread-local variable, it is
19215 evaluated at run time and returns the address of the current thread's
19216 instance of that variable. An address so obtained may be used by any
19217 thread. When a thread terminates, any pointers to thread-local variables
19218 in that thread become invalid.
19220 No static initialization may refer to the address of a thread-local variable.
19222 In C++, if an initializer is present for a thread-local variable, it must
19223 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
19226 See @uref{http://www.akkadia.org/drepper/tls.pdf,
19227 ELF Handling For Thread-Local Storage} for a detailed explanation of
19228 the four thread-local storage addressing models, and how the runtime
19229 is expected to function.
19232 * C99 Thread-Local Edits::
19233 * C++98 Thread-Local Edits::
19236 @node C99 Thread-Local Edits
19237 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
19239 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
19240 that document the exact semantics of the language extension.
19244 @cite{5.1.2 Execution environments}
19246 Add new text after paragraph 1
19249 Within either execution environment, a @dfn{thread} is a flow of
19250 control within a program. It is implementation defined whether
19251 or not there may be more than one thread associated with a program.
19252 It is implementation defined how threads beyond the first are
19253 created, the name and type of the function called at thread
19254 startup, and how threads may be terminated. However, objects
19255 with thread storage duration shall be initialized before thread
19260 @cite{6.2.4 Storage durations of objects}
19262 Add new text before paragraph 3
19265 An object whose identifier is declared with the storage-class
19266 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
19267 Its lifetime is the entire execution of the thread, and its
19268 stored value is initialized only once, prior to thread startup.
19272 @cite{6.4.1 Keywords}
19274 Add @code{__thread}.
19277 @cite{6.7.1 Storage-class specifiers}
19279 Add @code{__thread} to the list of storage class specifiers in
19282 Change paragraph 2 to
19285 With the exception of @code{__thread}, at most one storage-class
19286 specifier may be given [@dots{}]. The @code{__thread} specifier may
19287 be used alone, or immediately following @code{extern} or
19291 Add new text after paragraph 6
19294 The declaration of an identifier for a variable that has
19295 block scope that specifies @code{__thread} shall also
19296 specify either @code{extern} or @code{static}.
19298 The @code{__thread} specifier shall be used only with
19303 @node C++98 Thread-Local Edits
19304 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
19306 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
19307 that document the exact semantics of the language extension.
19311 @b{[intro.execution]}
19313 New text after paragraph 4
19316 A @dfn{thread} is a flow of control within the abstract machine.
19317 It is implementation defined whether or not there may be more than
19321 New text after paragraph 7
19324 It is unspecified whether additional action must be taken to
19325 ensure when and whether side effects are visible to other threads.
19331 Add @code{__thread}.
19334 @b{[basic.start.main]}
19336 Add after paragraph 5
19339 The thread that begins execution at the @code{main} function is called
19340 the @dfn{main thread}. It is implementation defined how functions
19341 beginning threads other than the main thread are designated or typed.
19342 A function so designated, as well as the @code{main} function, is called
19343 a @dfn{thread startup function}. It is implementation defined what
19344 happens if a thread startup function returns. It is implementation
19345 defined what happens to other threads when any thread calls @code{exit}.
19349 @b{[basic.start.init]}
19351 Add after paragraph 4
19354 The storage for an object of thread storage duration shall be
19355 statically initialized before the first statement of the thread startup
19356 function. An object of thread storage duration shall not require
19357 dynamic initialization.
19361 @b{[basic.start.term]}
19363 Add after paragraph 3
19366 The type of an object with thread storage duration shall not have a
19367 non-trivial destructor, nor shall it be an array type whose elements
19368 (directly or indirectly) have non-trivial destructors.
19374 Add ``thread storage duration'' to the list in paragraph 1.
19379 Thread, static, and automatic storage durations are associated with
19380 objects introduced by declarations [@dots{}].
19383 Add @code{__thread} to the list of specifiers in paragraph 3.
19386 @b{[basic.stc.thread]}
19388 New section before @b{[basic.stc.static]}
19391 The keyword @code{__thread} applied to a non-local object gives the
19392 object thread storage duration.
19394 A local variable or class data member declared both @code{static}
19395 and @code{__thread} gives the variable or member thread storage
19400 @b{[basic.stc.static]}
19405 All objects that have neither thread storage duration, dynamic
19406 storage duration nor are local [@dots{}].
19412 Add @code{__thread} to the list in paragraph 1.
19417 With the exception of @code{__thread}, at most one
19418 @var{storage-class-specifier} shall appear in a given
19419 @var{decl-specifier-seq}. The @code{__thread} specifier may
19420 be used alone, or immediately following the @code{extern} or
19421 @code{static} specifiers. [@dots{}]
19424 Add after paragraph 5
19427 The @code{__thread} specifier can be applied only to the names of objects
19428 and to anonymous unions.
19434 Add after paragraph 6
19437 Non-@code{static} members shall not be @code{__thread}.
19441 @node Binary constants
19442 @section Binary Constants using the @samp{0b} Prefix
19443 @cindex Binary constants using the @samp{0b} prefix
19445 Integer constants can be written as binary constants, consisting of a
19446 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
19447 @samp{0B}. This is particularly useful in environments that operate a
19448 lot on the bit level (like microcontrollers).
19450 The following statements are identical:
19459 The type of these constants follows the same rules as for octal or
19460 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
19463 @node C++ Extensions
19464 @chapter Extensions to the C++ Language
19465 @cindex extensions, C++ language
19466 @cindex C++ language extensions
19468 The GNU compiler provides these extensions to the C++ language (and you
19469 can also use most of the C language extensions in your C++ programs). If you
19470 want to write code that checks whether these features are available, you can
19471 test for the GNU compiler the same way as for C programs: check for a
19472 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
19473 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
19474 Predefined Macros,cpp,The GNU C Preprocessor}).
19477 * C++ Volatiles:: What constitutes an access to a volatile object.
19478 * Restricted Pointers:: C99 restricted pointers and references.
19479 * Vague Linkage:: Where G++ puts inlines, vtables and such.
19480 * C++ Interface:: You can use a single C++ header file for both
19481 declarations and definitions.
19482 * Template Instantiation:: Methods for ensuring that exactly one copy of
19483 each needed template instantiation is emitted.
19484 * Bound member functions:: You can extract a function pointer to the
19485 method denoted by a @samp{->*} or @samp{.*} expression.
19486 * C++ Attributes:: Variable, function, and type attributes for C++ only.
19487 * Function Multiversioning:: Declaring multiple function versions.
19488 * Namespace Association:: Strong using-directives for namespace association.
19489 * Type Traits:: Compiler support for type traits.
19490 * C++ Concepts:: Improved support for generic programming.
19491 * Java Exceptions:: Tweaking exception handling to work with Java.
19492 * Deprecated Features:: Things will disappear from G++.
19493 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
19496 @node C++ Volatiles
19497 @section When is a Volatile C++ Object Accessed?
19498 @cindex accessing volatiles
19499 @cindex volatile read
19500 @cindex volatile write
19501 @cindex volatile access
19503 The C++ standard differs from the C standard in its treatment of
19504 volatile objects. It fails to specify what constitutes a volatile
19505 access, except to say that C++ should behave in a similar manner to C
19506 with respect to volatiles, where possible. However, the different
19507 lvalueness of expressions between C and C++ complicate the behavior.
19508 G++ behaves the same as GCC for volatile access, @xref{C
19509 Extensions,,Volatiles}, for a description of GCC's behavior.
19511 The C and C++ language specifications differ when an object is
19512 accessed in a void context:
19515 volatile int *src = @var{somevalue};
19519 The C++ standard specifies that such expressions do not undergo lvalue
19520 to rvalue conversion, and that the type of the dereferenced object may
19521 be incomplete. The C++ standard does not specify explicitly that it
19522 is lvalue to rvalue conversion that is responsible for causing an
19523 access. There is reason to believe that it is, because otherwise
19524 certain simple expressions become undefined. However, because it
19525 would surprise most programmers, G++ treats dereferencing a pointer to
19526 volatile object of complete type as GCC would do for an equivalent
19527 type in C@. When the object has incomplete type, G++ issues a
19528 warning; if you wish to force an error, you must force a conversion to
19529 rvalue with, for instance, a static cast.
19531 When using a reference to volatile, G++ does not treat equivalent
19532 expressions as accesses to volatiles, but instead issues a warning that
19533 no volatile is accessed. The rationale for this is that otherwise it
19534 becomes difficult to determine where volatile access occur, and not
19535 possible to ignore the return value from functions returning volatile
19536 references. Again, if you wish to force a read, cast the reference to
19539 G++ implements the same behavior as GCC does when assigning to a
19540 volatile object---there is no reread of the assigned-to object, the
19541 assigned rvalue is reused. Note that in C++ assignment expressions
19542 are lvalues, and if used as an lvalue, the volatile object is
19543 referred to. For instance, @var{vref} refers to @var{vobj}, as
19544 expected, in the following example:
19548 volatile int &vref = vobj = @var{something};
19551 @node Restricted Pointers
19552 @section Restricting Pointer Aliasing
19553 @cindex restricted pointers
19554 @cindex restricted references
19555 @cindex restricted this pointer
19557 As with the C front end, G++ understands the C99 feature of restricted pointers,
19558 specified with the @code{__restrict__}, or @code{__restrict} type
19559 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
19560 language flag, @code{restrict} is not a keyword in C++.
19562 In addition to allowing restricted pointers, you can specify restricted
19563 references, which indicate that the reference is not aliased in the local
19567 void fn (int *__restrict__ rptr, int &__restrict__ rref)
19574 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
19575 @var{rref} refers to a (different) unaliased integer.
19577 You may also specify whether a member function's @var{this} pointer is
19578 unaliased by using @code{__restrict__} as a member function qualifier.
19581 void T::fn () __restrict__
19588 Within the body of @code{T::fn}, @var{this} has the effective
19589 definition @code{T *__restrict__ const this}. Notice that the
19590 interpretation of a @code{__restrict__} member function qualifier is
19591 different to that of @code{const} or @code{volatile} qualifier, in that it
19592 is applied to the pointer rather than the object. This is consistent with
19593 other compilers that implement restricted pointers.
19595 As with all outermost parameter qualifiers, @code{__restrict__} is
19596 ignored in function definition matching. This means you only need to
19597 specify @code{__restrict__} in a function definition, rather than
19598 in a function prototype as well.
19600 @node Vague Linkage
19601 @section Vague Linkage
19602 @cindex vague linkage
19604 There are several constructs in C++ that require space in the object
19605 file but are not clearly tied to a single translation unit. We say that
19606 these constructs have ``vague linkage''. Typically such constructs are
19607 emitted wherever they are needed, though sometimes we can be more
19611 @item Inline Functions
19612 Inline functions are typically defined in a header file which can be
19613 included in many different compilations. Hopefully they can usually be
19614 inlined, but sometimes an out-of-line copy is necessary, if the address
19615 of the function is taken or if inlining fails. In general, we emit an
19616 out-of-line copy in all translation units where one is needed. As an
19617 exception, we only emit inline virtual functions with the vtable, since
19618 it always requires a copy.
19620 Local static variables and string constants used in an inline function
19621 are also considered to have vague linkage, since they must be shared
19622 between all inlined and out-of-line instances of the function.
19626 C++ virtual functions are implemented in most compilers using a lookup
19627 table, known as a vtable. The vtable contains pointers to the virtual
19628 functions provided by a class, and each object of the class contains a
19629 pointer to its vtable (or vtables, in some multiple-inheritance
19630 situations). If the class declares any non-inline, non-pure virtual
19631 functions, the first one is chosen as the ``key method'' for the class,
19632 and the vtable is only emitted in the translation unit where the key
19635 @emph{Note:} If the chosen key method is later defined as inline, the
19636 vtable is still emitted in every translation unit that defines it.
19637 Make sure that any inline virtuals are declared inline in the class
19638 body, even if they are not defined there.
19640 @item @code{type_info} objects
19641 @cindex @code{type_info}
19643 C++ requires information about types to be written out in order to
19644 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
19645 For polymorphic classes (classes with virtual functions), the @samp{type_info}
19646 object is written out along with the vtable so that @samp{dynamic_cast}
19647 can determine the dynamic type of a class object at run time. For all
19648 other types, we write out the @samp{type_info} object when it is used: when
19649 applying @samp{typeid} to an expression, throwing an object, or
19650 referring to a type in a catch clause or exception specification.
19652 @item Template Instantiations
19653 Most everything in this section also applies to template instantiations,
19654 but there are other options as well.
19655 @xref{Template Instantiation,,Where's the Template?}.
19659 When used with GNU ld version 2.8 or later on an ELF system such as
19660 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
19661 these constructs will be discarded at link time. This is known as
19664 On targets that don't support COMDAT, but do support weak symbols, GCC
19665 uses them. This way one copy overrides all the others, but
19666 the unused copies still take up space in the executable.
19668 For targets that do not support either COMDAT or weak symbols,
19669 most entities with vague linkage are emitted as local symbols to
19670 avoid duplicate definition errors from the linker. This does not happen
19671 for local statics in inlines, however, as having multiple copies
19672 almost certainly breaks things.
19674 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
19675 another way to control placement of these constructs.
19677 @node C++ Interface
19678 @section C++ Interface and Implementation Pragmas
19680 @cindex interface and implementation headers, C++
19681 @cindex C++ interface and implementation headers
19682 @cindex pragmas, interface and implementation
19684 @code{#pragma interface} and @code{#pragma implementation} provide the
19685 user with a way of explicitly directing the compiler to emit entities
19686 with vague linkage (and debugging information) in a particular
19689 @emph{Note:} These @code{#pragma}s have been superceded as of GCC 2.7.2
19690 by COMDAT support and the ``key method'' heuristic
19691 mentioned in @ref{Vague Linkage}. Using them can actually cause your
19692 program to grow due to unnecessary out-of-line copies of inline
19696 @item #pragma interface
19697 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
19698 @kindex #pragma interface
19699 Use this directive in @emph{header files} that define object classes, to save
19700 space in most of the object files that use those classes. Normally,
19701 local copies of certain information (backup copies of inline member
19702 functions, debugging information, and the internal tables that implement
19703 virtual functions) must be kept in each object file that includes class
19704 definitions. You can use this pragma to avoid such duplication. When a
19705 header file containing @samp{#pragma interface} is included in a
19706 compilation, this auxiliary information is not generated (unless
19707 the main input source file itself uses @samp{#pragma implementation}).
19708 Instead, the object files contain references to be resolved at link
19711 The second form of this directive is useful for the case where you have
19712 multiple headers with the same name in different directories. If you
19713 use this form, you must specify the same string to @samp{#pragma
19716 @item #pragma implementation
19717 @itemx #pragma implementation "@var{objects}.h"
19718 @kindex #pragma implementation
19719 Use this pragma in a @emph{main input file}, when you want full output from
19720 included header files to be generated (and made globally visible). The
19721 included header file, in turn, should use @samp{#pragma interface}.
19722 Backup copies of inline member functions, debugging information, and the
19723 internal tables used to implement virtual functions are all generated in
19724 implementation files.
19726 @cindex implied @code{#pragma implementation}
19727 @cindex @code{#pragma implementation}, implied
19728 @cindex naming convention, implementation headers
19729 If you use @samp{#pragma implementation} with no argument, it applies to
19730 an include file with the same basename@footnote{A file's @dfn{basename}
19731 is the name stripped of all leading path information and of trailing
19732 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
19733 file. For example, in @file{allclass.cc}, giving just
19734 @samp{#pragma implementation}
19735 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
19737 Use the string argument if you want a single implementation file to
19738 include code from multiple header files. (You must also use
19739 @samp{#include} to include the header file; @samp{#pragma
19740 implementation} only specifies how to use the file---it doesn't actually
19743 There is no way to split up the contents of a single header file into
19744 multiple implementation files.
19747 @cindex inlining and C++ pragmas
19748 @cindex C++ pragmas, effect on inlining
19749 @cindex pragmas in C++, effect on inlining
19750 @samp{#pragma implementation} and @samp{#pragma interface} also have an
19751 effect on function inlining.
19753 If you define a class in a header file marked with @samp{#pragma
19754 interface}, the effect on an inline function defined in that class is
19755 similar to an explicit @code{extern} declaration---the compiler emits
19756 no code at all to define an independent version of the function. Its
19757 definition is used only for inlining with its callers.
19759 @opindex fno-implement-inlines
19760 Conversely, when you include the same header file in a main source file
19761 that declares it as @samp{#pragma implementation}, the compiler emits
19762 code for the function itself; this defines a version of the function
19763 that can be found via pointers (or by callers compiled without
19764 inlining). If all calls to the function can be inlined, you can avoid
19765 emitting the function by compiling with @option{-fno-implement-inlines}.
19766 If any calls are not inlined, you will get linker errors.
19768 @node Template Instantiation
19769 @section Where's the Template?
19770 @cindex template instantiation
19772 C++ templates were the first language feature to require more
19773 intelligence from the environment than was traditionally found on a UNIX
19774 system. Somehow the compiler and linker have to make sure that each
19775 template instance occurs exactly once in the executable if it is needed,
19776 and not at all otherwise. There are two basic approaches to this
19777 problem, which are referred to as the Borland model and the Cfront model.
19780 @item Borland model
19781 Borland C++ solved the template instantiation problem by adding the code
19782 equivalent of common blocks to their linker; the compiler emits template
19783 instances in each translation unit that uses them, and the linker
19784 collapses them together. The advantage of this model is that the linker
19785 only has to consider the object files themselves; there is no external
19786 complexity to worry about. The disadvantage is that compilation time
19787 is increased because the template code is being compiled repeatedly.
19788 Code written for this model tends to include definitions of all
19789 templates in the header file, since they must be seen to be
19793 The AT&T C++ translator, Cfront, solved the template instantiation
19794 problem by creating the notion of a template repository, an
19795 automatically maintained place where template instances are stored. A
19796 more modern version of the repository works as follows: As individual
19797 object files are built, the compiler places any template definitions and
19798 instantiations encountered in the repository. At link time, the link
19799 wrapper adds in the objects in the repository and compiles any needed
19800 instances that were not previously emitted. The advantages of this
19801 model are more optimal compilation speed and the ability to use the
19802 system linker; to implement the Borland model a compiler vendor also
19803 needs to replace the linker. The disadvantages are vastly increased
19804 complexity, and thus potential for error; for some code this can be
19805 just as transparent, but in practice it can been very difficult to build
19806 multiple programs in one directory and one program in multiple
19807 directories. Code written for this model tends to separate definitions
19808 of non-inline member templates into a separate file, which should be
19809 compiled separately.
19812 G++ implements the Borland model on targets where the linker supports it,
19813 including ELF targets (such as GNU/Linux), Mac OS X and Microsoft Windows.
19814 Otherwise G++ implements neither automatic model.
19816 You have the following options for dealing with template instantiations:
19820 Do nothing. Code written for the Borland model works fine, but
19821 each translation unit contains instances of each of the templates it
19822 uses. The duplicate instances will be discarded by the linker, but in
19823 a large program, this can lead to an unacceptable amount of code
19824 duplication in object files or shared libraries.
19826 Duplicate instances of a template can be avoided by defining an explicit
19827 instantiation in one object file, and preventing the compiler from doing
19828 implicit instantiations in any other object files by using an explicit
19829 instantiation declaration, using the @code{extern template} syntax:
19832 extern template int max (int, int);
19835 This syntax is defined in the C++ 2011 standard, but has been supported by
19836 G++ and other compilers since well before 2011.
19838 Explicit instantiations can be used for the largest or most frequently
19839 duplicated instances, without having to know exactly which other instances
19840 are used in the rest of the program. You can scatter the explicit
19841 instantiations throughout your program, perhaps putting them in the
19842 translation units where the instances are used or the translation units
19843 that define the templates themselves; you can put all of the explicit
19844 instantiations you need into one big file; or you can create small files
19851 template class Foo<int>;
19852 template ostream& operator <<
19853 (ostream&, const Foo<int>&);
19857 for each of the instances you need, and create a template instantiation
19858 library from those.
19860 This is the simplest option, but also offers flexibility and
19861 fine-grained control when necessary. It is also the most portable
19862 alternative and programs using this approach will work with most modern
19867 Compile your template-using code with @option{-frepo}. The compiler
19868 generates files with the extension @samp{.rpo} listing all of the
19869 template instantiations used in the corresponding object files that
19870 could be instantiated there; the link wrapper, @samp{collect2},
19871 then updates the @samp{.rpo} files to tell the compiler where to place
19872 those instantiations and rebuild any affected object files. The
19873 link-time overhead is negligible after the first pass, as the compiler
19874 continues to place the instantiations in the same files.
19876 This can be a suitable option for application code written for the Borland
19877 model, as it usually just works. Code written for the Cfront model
19878 needs to be modified so that the template definitions are available at
19879 one or more points of instantiation; usually this is as simple as adding
19880 @code{#include <tmethods.cc>} to the end of each template header.
19882 For library code, if you want the library to provide all of the template
19883 instantiations it needs, just try to link all of its object files
19884 together; the link will fail, but cause the instantiations to be
19885 generated as a side effect. Be warned, however, that this may cause
19886 conflicts if multiple libraries try to provide the same instantiations.
19887 For greater control, use explicit instantiation as described in the next
19891 @opindex fno-implicit-templates
19892 Compile your code with @option{-fno-implicit-templates} to disable the
19893 implicit generation of template instances, and explicitly instantiate
19894 all the ones you use. This approach requires more knowledge of exactly
19895 which instances you need than do the others, but it's less
19896 mysterious and allows greater control if you want to ensure that only
19897 the intended instances are used.
19899 If you are using Cfront-model code, you can probably get away with not
19900 using @option{-fno-implicit-templates} when compiling files that don't
19901 @samp{#include} the member template definitions.
19903 If you use one big file to do the instantiations, you may want to
19904 compile it without @option{-fno-implicit-templates} so you get all of the
19905 instances required by your explicit instantiations (but not by any
19906 other files) without having to specify them as well.
19908 In addition to forward declaration of explicit instantiations
19909 (with @code{extern}), G++ has extended the template instantiation
19910 syntax to support instantiation of the compiler support data for a
19911 template class (i.e.@: the vtable) without instantiating any of its
19912 members (with @code{inline}), and instantiation of only the static data
19913 members of a template class, without the support data or member
19914 functions (with @code{static}):
19917 inline template class Foo<int>;
19918 static template class Foo<int>;
19922 @node Bound member functions
19923 @section Extracting the Function Pointer from a Bound Pointer to Member Function
19925 @cindex pointer to member function
19926 @cindex bound pointer to member function
19928 In C++, pointer to member functions (PMFs) are implemented using a wide
19929 pointer of sorts to handle all the possible call mechanisms; the PMF
19930 needs to store information about how to adjust the @samp{this} pointer,
19931 and if the function pointed to is virtual, where to find the vtable, and
19932 where in the vtable to look for the member function. If you are using
19933 PMFs in an inner loop, you should really reconsider that decision. If
19934 that is not an option, you can extract the pointer to the function that
19935 would be called for a given object/PMF pair and call it directly inside
19936 the inner loop, to save a bit of time.
19938 Note that you still pay the penalty for the call through a
19939 function pointer; on most modern architectures, such a call defeats the
19940 branch prediction features of the CPU@. This is also true of normal
19941 virtual function calls.
19943 The syntax for this extension is
19947 extern int (A::*fp)();
19948 typedef int (*fptr)(A *);
19950 fptr p = (fptr)(a.*fp);
19953 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
19954 no object is needed to obtain the address of the function. They can be
19955 converted to function pointers directly:
19958 fptr p1 = (fptr)(&A::foo);
19961 @opindex Wno-pmf-conversions
19962 You must specify @option{-Wno-pmf-conversions} to use this extension.
19964 @node C++ Attributes
19965 @section C++-Specific Variable, Function, and Type Attributes
19967 Some attributes only make sense for C++ programs.
19970 @item abi_tag ("@var{tag}", ...)
19971 @cindex @code{abi_tag} function attribute
19972 @cindex @code{abi_tag} variable attribute
19973 @cindex @code{abi_tag} type attribute
19974 The @code{abi_tag} attribute can be applied to a function, variable, or class
19975 declaration. It modifies the mangled name of the entity to
19976 incorporate the tag name, in order to distinguish the function or
19977 class from an earlier version with a different ABI; perhaps the class
19978 has changed size, or the function has a different return type that is
19979 not encoded in the mangled name.
19981 The attribute can also be applied to an inline namespace, but does not
19982 affect the mangled name of the namespace; in this case it is only used
19983 for @option{-Wabi-tag} warnings and automatic tagging of functions and
19984 variables. Tagging inline namespaces is generally preferable to
19985 tagging individual declarations, but the latter is sometimes
19986 necessary, such as when only certain members of a class need to be
19989 The argument can be a list of strings of arbitrary length. The
19990 strings are sorted on output, so the order of the list is
19993 A redeclaration of an entity must not add new ABI tags,
19994 since doing so would change the mangled name.
19996 The ABI tags apply to a name, so all instantiations and
19997 specializations of a template have the same tags. The attribute will
19998 be ignored if applied to an explicit specialization or instantiation.
20000 The @option{-Wabi-tag} flag enables a warning about a class which does
20001 not have all the ABI tags used by its subobjects and virtual functions; for users with code
20002 that needs to coexist with an earlier ABI, using this option can help
20003 to find all affected types that need to be tagged.
20005 When a type involving an ABI tag is used as the type of a variable or
20006 return type of a function where that tag is not already present in the
20007 signature of the function, the tag is automatically applied to the
20008 variable or function. @option{-Wabi-tag} also warns about this
20009 situation; this warning can be avoided by explicitly tagging the
20010 variable or function or moving it into a tagged inline namespace.
20012 @item init_priority (@var{priority})
20013 @cindex @code{init_priority} variable attribute
20015 In Standard C++, objects defined at namespace scope are guaranteed to be
20016 initialized in an order in strict accordance with that of their definitions
20017 @emph{in a given translation unit}. No guarantee is made for initializations
20018 across translation units. However, GNU C++ allows users to control the
20019 order of initialization of objects defined at namespace scope with the
20020 @code{init_priority} attribute by specifying a relative @var{priority},
20021 a constant integral expression currently bounded between 101 and 65535
20022 inclusive. Lower numbers indicate a higher priority.
20024 In the following example, @code{A} would normally be created before
20025 @code{B}, but the @code{init_priority} attribute reverses that order:
20028 Some_Class A __attribute__ ((init_priority (2000)));
20029 Some_Class B __attribute__ ((init_priority (543)));
20033 Note that the particular values of @var{priority} do not matter; only their
20036 @item java_interface
20037 @cindex @code{java_interface} type attribute
20039 This type attribute informs C++ that the class is a Java interface. It may
20040 only be applied to classes declared within an @code{extern "Java"} block.
20041 Calls to methods declared in this interface are dispatched using GCJ's
20042 interface table mechanism, instead of regular virtual table dispatch.
20045 @cindex @code{warn_unused} type attribute
20047 For C++ types with non-trivial constructors and/or destructors it is
20048 impossible for the compiler to determine whether a variable of this
20049 type is truly unused if it is not referenced. This type attribute
20050 informs the compiler that variables of this type should be warned
20051 about if they appear to be unused, just like variables of fundamental
20054 This attribute is appropriate for types which just represent a value,
20055 such as @code{std::string}; it is not appropriate for types which
20056 control a resource, such as @code{std::mutex}.
20058 This attribute is also accepted in C, but it is unnecessary because C
20059 does not have constructors or destructors.
20063 See also @ref{Namespace Association}.
20065 @node Function Multiversioning
20066 @section Function Multiversioning
20067 @cindex function versions
20069 With the GNU C++ front end, for x86 targets, you may specify multiple
20070 versions of a function, where each function is specialized for a
20071 specific target feature. At runtime, the appropriate version of the
20072 function is automatically executed depending on the characteristics of
20073 the execution platform. Here is an example.
20076 __attribute__ ((target ("default")))
20079 // The default version of foo.
20083 __attribute__ ((target ("sse4.2")))
20086 // foo version for SSE4.2
20090 __attribute__ ((target ("arch=atom")))
20093 // foo version for the Intel ATOM processor
20097 __attribute__ ((target ("arch=amdfam10")))
20100 // foo version for the AMD Family 0x10 processors.
20107 assert ((*p) () == foo ());
20112 In the above example, four versions of function foo are created. The
20113 first version of foo with the target attribute "default" is the default
20114 version. This version gets executed when no other target specific
20115 version qualifies for execution on a particular platform. A new version
20116 of foo is created by using the same function signature but with a
20117 different target string. Function foo is called or a pointer to it is
20118 taken just like a regular function. GCC takes care of doing the
20119 dispatching to call the right version at runtime. Refer to the
20120 @uref{http://gcc.gnu.org/wiki/FunctionMultiVersioning, GCC wiki on
20121 Function Multiversioning} for more details.
20123 @node Namespace Association
20124 @section Namespace Association
20126 @strong{Caution:} The semantics of this extension are equivalent
20127 to C++ 2011 inline namespaces. Users should use inline namespaces
20128 instead as this extension will be removed in future versions of G++.
20130 A using-directive with @code{__attribute ((strong))} is stronger
20131 than a normal using-directive in two ways:
20135 Templates from the used namespace can be specialized and explicitly
20136 instantiated as though they were members of the using namespace.
20139 The using namespace is considered an associated namespace of all
20140 templates in the used namespace for purposes of argument-dependent
20144 The used namespace must be nested within the using namespace so that
20145 normal unqualified lookup works properly.
20147 This is useful for composing a namespace transparently from
20148 implementation namespaces. For example:
20153 template <class T> struct A @{ @};
20155 using namespace debug __attribute ((__strong__));
20156 template <> struct A<int> @{ @}; // @r{OK to specialize}
20158 template <class T> void f (A<T>);
20163 f (std::A<float>()); // @r{lookup finds} std::f
20169 @section Type Traits
20171 The C++ front end implements syntactic extensions that allow
20172 compile-time determination of
20173 various characteristics of a type (or of a
20177 @item __has_nothrow_assign (type)
20178 If @code{type} is const qualified or is a reference type then the trait is
20179 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
20180 is true, else if @code{type} is a cv class or union type with copy assignment
20181 operators that are known not to throw an exception then the trait is true,
20182 else it is false. Requires: @code{type} shall be a complete type,
20183 (possibly cv-qualified) @code{void}, or an array of unknown bound.
20185 @item __has_nothrow_copy (type)
20186 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
20187 @code{type} is a cv class or union type with copy constructors that
20188 are known not to throw an exception then the trait is true, else it is false.
20189 Requires: @code{type} shall be a complete type, (possibly cv-qualified)
20190 @code{void}, or an array of unknown bound.
20192 @item __has_nothrow_constructor (type)
20193 If @code{__has_trivial_constructor (type)} is true then the trait is
20194 true, else if @code{type} is a cv class or union type (or array
20195 thereof) with a default constructor that is known not to throw an
20196 exception then the trait is true, else it is false. Requires:
20197 @code{type} shall be a complete type, (possibly cv-qualified)
20198 @code{void}, or an array of unknown bound.
20200 @item __has_trivial_assign (type)
20201 If @code{type} is const qualified or is a reference type then the trait is
20202 false. Otherwise if @code{__is_pod (type)} is true then the trait is
20203 true, else if @code{type} is a cv class or union type with a trivial
20204 copy assignment ([class.copy]) then the trait is true, else it is
20205 false. Requires: @code{type} shall be a complete type, (possibly
20206 cv-qualified) @code{void}, or an array of unknown bound.
20208 @item __has_trivial_copy (type)
20209 If @code{__is_pod (type)} is true or @code{type} is a reference type
20210 then the trait is true, else if @code{type} is a cv class or union type
20211 with a trivial copy constructor ([class.copy]) then the trait
20212 is true, else it is false. Requires: @code{type} shall be a complete
20213 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
20215 @item __has_trivial_constructor (type)
20216 If @code{__is_pod (type)} is true then the trait is true, else if
20217 @code{type} is a cv class or union type (or array thereof) with a
20218 trivial default constructor ([class.ctor]) then the trait is true,
20219 else it is false. Requires: @code{type} shall be a complete
20220 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
20222 @item __has_trivial_destructor (type)
20223 If @code{__is_pod (type)} is true or @code{type} is a reference type then
20224 the trait is true, else if @code{type} is a cv class or union type (or
20225 array thereof) with a trivial destructor ([class.dtor]) then the trait
20226 is true, else it is false. Requires: @code{type} shall be a complete
20227 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
20229 @item __has_virtual_destructor (type)
20230 If @code{type} is a class type with a virtual destructor
20231 ([class.dtor]) then the trait is true, else it is false. Requires:
20232 @code{type} shall be a complete type, (possibly cv-qualified)
20233 @code{void}, or an array of unknown bound.
20235 @item __is_abstract (type)
20236 If @code{type} is an abstract class ([class.abstract]) then the trait
20237 is true, else it is false. Requires: @code{type} shall be a complete
20238 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
20240 @item __is_base_of (base_type, derived_type)
20241 If @code{base_type} is a base class of @code{derived_type}
20242 ([class.derived]) then the trait is true, otherwise it is false.
20243 Top-level cv qualifications of @code{base_type} and
20244 @code{derived_type} are ignored. For the purposes of this trait, a
20245 class type is considered is own base. Requires: if @code{__is_class
20246 (base_type)} and @code{__is_class (derived_type)} are true and
20247 @code{base_type} and @code{derived_type} are not the same type
20248 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
20249 type. Diagnostic is produced if this requirement is not met.
20251 @item __is_class (type)
20252 If @code{type} is a cv class type, and not a union type
20253 ([basic.compound]) the trait is true, else it is false.
20255 @item __is_empty (type)
20256 If @code{__is_class (type)} is false then the trait is false.
20257 Otherwise @code{type} is considered empty if and only if: @code{type}
20258 has no non-static data members, or all non-static data members, if
20259 any, are bit-fields of length 0, and @code{type} has no virtual
20260 members, and @code{type} has no virtual base classes, and @code{type}
20261 has no base classes @code{base_type} for which
20262 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
20263 be a complete type, (possibly cv-qualified) @code{void}, or an array
20266 @item __is_enum (type)
20267 If @code{type} is a cv enumeration type ([basic.compound]) the trait is
20268 true, else it is false.
20270 @item __is_literal_type (type)
20271 If @code{type} is a literal type ([basic.types]) the trait is
20272 true, else it is false. Requires: @code{type} shall be a complete type,
20273 (possibly cv-qualified) @code{void}, or an array of unknown bound.
20275 @item __is_pod (type)
20276 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
20277 else it is false. Requires: @code{type} shall be a complete type,
20278 (possibly cv-qualified) @code{void}, or an array of unknown bound.
20280 @item __is_polymorphic (type)
20281 If @code{type} is a polymorphic class ([class.virtual]) then the trait
20282 is true, else it is false. Requires: @code{type} shall be a complete
20283 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
20285 @item __is_standard_layout (type)
20286 If @code{type} is a standard-layout type ([basic.types]) the trait is
20287 true, else it is false. Requires: @code{type} shall be a complete
20288 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
20290 @item __is_trivial (type)
20291 If @code{type} is a trivial type ([basic.types]) the trait is
20292 true, else it is false. Requires: @code{type} shall be a complete
20293 type, (possibly cv-qualified) @code{void}, or an array of unknown bound.
20295 @item __is_union (type)
20296 If @code{type} is a cv union type ([basic.compound]) the trait is
20297 true, else it is false.
20299 @item __underlying_type (type)
20300 The underlying type of @code{type}. Requires: @code{type} shall be
20301 an enumeration type ([dcl.enum]).
20307 @section C++ Concepts
20309 C++ concepts provide much-improved support for generic programming. In
20310 particular, they allow the specification of constraints on template arguments.
20311 The constraints are used to extend the usual overloading and partial
20312 specialization capabilities of the language, allowing generic data structures
20313 and algorithms to be ``refined'' based on their properties rather than their
20316 The following keywords are reserved for concepts.
20320 States an expression as an assumption, and if possible, verifies that the
20321 assumption is valid. For example, @code{assume(n > 0)}.
20324 Introduces an axiom definition. Axioms introduce requirements on values.
20327 Introduces a universally quantified object in an axiom. For example,
20328 @code{forall (int n) n + 0 == n}).
20331 Introduces a concept definition. Concepts are sets of syntactic and semantic
20332 requirements on types and their values.
20335 Introduces constraints on template arguments or requirements for a member
20336 function of a class template.
20340 The front end also exposes a number of internal mechanism that can be used
20341 to simplify the writing of type traits. Note that some of these traits are
20342 likely to be removed in the future.
20345 @item __is_same (type1, type2)
20346 A binary type trait: true whenever the type arguments are the same.
20351 @node Java Exceptions
20352 @section Java Exceptions
20354 The Java language uses a slightly different exception handling model
20355 from C++. Normally, GNU C++ automatically detects when you are
20356 writing C++ code that uses Java exceptions, and handle them
20357 appropriately. However, if C++ code only needs to execute destructors
20358 when Java exceptions are thrown through it, GCC guesses incorrectly.
20359 Sample problematic code is:
20362 struct S @{ ~S(); @};
20363 extern void bar(); // @r{is written in Java, and may throw exceptions}
20372 The usual effect of an incorrect guess is a link failure, complaining of
20373 a missing routine called @samp{__gxx_personality_v0}.
20375 You can inform the compiler that Java exceptions are to be used in a
20376 translation unit, irrespective of what it might think, by writing
20377 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
20378 @samp{#pragma} must appear before any functions that throw or catch
20379 exceptions, or run destructors when exceptions are thrown through them.
20381 You cannot mix Java and C++ exceptions in the same translation unit. It
20382 is believed to be safe to throw a C++ exception from one file through
20383 another file compiled for the Java exception model, or vice versa, but
20384 there may be bugs in this area.
20386 @node Deprecated Features
20387 @section Deprecated Features
20389 In the past, the GNU C++ compiler was extended to experiment with new
20390 features, at a time when the C++ language was still evolving. Now that
20391 the C++ standard is complete, some of those features are superseded by
20392 superior alternatives. Using the old features might cause a warning in
20393 some cases that the feature will be dropped in the future. In other
20394 cases, the feature might be gone already.
20396 While the list below is not exhaustive, it documents some of the options
20397 that are now deprecated:
20400 @item -fexternal-templates
20401 @itemx -falt-external-templates
20402 These are two of the many ways for G++ to implement template
20403 instantiation. @xref{Template Instantiation}. The C++ standard clearly
20404 defines how template definitions have to be organized across
20405 implementation units. G++ has an implicit instantiation mechanism that
20406 should work just fine for standard-conforming code.
20408 @item -fstrict-prototype
20409 @itemx -fno-strict-prototype
20410 Previously it was possible to use an empty prototype parameter list to
20411 indicate an unspecified number of parameters (like C), rather than no
20412 parameters, as C++ demands. This feature has been removed, except where
20413 it is required for backwards compatibility. @xref{Backwards Compatibility}.
20416 G++ allows a virtual function returning @samp{void *} to be overridden
20417 by one returning a different pointer type. This extension to the
20418 covariant return type rules is now deprecated and will be removed from a
20421 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
20422 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
20423 and are now removed from G++. Code using these operators should be
20424 modified to use @code{std::min} and @code{std::max} instead.
20426 The named return value extension has been deprecated, and is now
20429 The use of initializer lists with new expressions has been deprecated,
20430 and is now removed from G++.
20432 Floating and complex non-type template parameters have been deprecated,
20433 and are now removed from G++.
20435 The implicit typename extension has been deprecated and is now
20438 The use of default arguments in function pointers, function typedefs
20439 and other places where they are not permitted by the standard is
20440 deprecated and will be removed from a future version of G++.
20442 G++ allows floating-point literals to appear in integral constant expressions,
20443 e.g.@: @samp{ enum E @{ e = int(2.2 * 3.7) @} }
20444 This extension is deprecated and will be removed from a future version.
20446 G++ allows static data members of const floating-point type to be declared
20447 with an initializer in a class definition. The standard only allows
20448 initializers for static members of const integral types and const
20449 enumeration types so this extension has been deprecated and will be removed
20450 from a future version.
20452 @node Backwards Compatibility
20453 @section Backwards Compatibility
20454 @cindex Backwards Compatibility
20455 @cindex ARM [Annotated C++ Reference Manual]
20457 Now that there is a definitive ISO standard C++, G++ has a specification
20458 to adhere to. The C++ language evolved over time, and features that
20459 used to be acceptable in previous drafts of the standard, such as the ARM
20460 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
20461 compilation of C++ written to such drafts, G++ contains some backwards
20462 compatibilities. @emph{All such backwards compatibility features are
20463 liable to disappear in future versions of G++.} They should be considered
20464 deprecated. @xref{Deprecated Features}.
20468 If a variable is declared at for scope, it used to remain in scope until
20469 the end of the scope that contained the for statement (rather than just
20470 within the for scope). G++ retains this, but issues a warning, if such a
20471 variable is accessed outside the for scope.
20473 @item Implicit C language
20474 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
20475 scope to set the language. On such systems, all header files are
20476 implicitly scoped inside a C language scope. Also, an empty prototype
20477 @code{()} is treated as an unspecified number of arguments, rather
20478 than no arguments, as C++ demands.
20481 @c LocalWords: emph deftypefn builtin ARCv2EM SIMD builtins msimd
20482 @c LocalWords: typedef v4si v8hi DMA dma vdiwr vdowr