1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1996, 1998, 1999, 2000,
2 @c 2001, 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
8 @chapter Extensions to the C Language Family
9 @cindex extensions, C language
10 @cindex C language extensions
13 GNU C provides several language features not found in ISO standard C@.
14 (The @option{-pedantic} option directs GCC to print a warning message if
15 any of these features is used.) To test for the availability of these
16 features in conditional compilation, check for a predefined macro
17 @code{__GNUC__}, which is always defined under GCC@.
19 These extensions are available in C and Objective-C@. Most of them are
20 also available in C++. @xref{C++ Extensions,,Extensions to the
21 C++ Language}, for extensions that apply @emph{only} to C++.
23 Some features that are in ISO C99 but not C89 or C++ are also, as
24 extensions, accepted by GCC in C89 mode and in C++.
27 * Statement Exprs:: Putting statements and declarations inside expressions.
28 * Local Labels:: Labels local to a block.
29 * Labels as Values:: Getting pointers to labels, and computed gotos.
30 * Nested Functions:: As in Algol and Pascal, lexical scoping of functions.
31 * Constructing Calls:: Dispatching a call to another function.
32 * Typeof:: @code{typeof}: referring to the type of an expression.
33 * Conditionals:: Omitting the middle operand of a @samp{?:} expression.
34 * Long Long:: Double-word integers---@code{long long int}.
35 * Complex:: Data types for complex numbers.
36 * Floating Types:: Additional Floating Types.
37 * Decimal Float:: Decimal Floating Types.
38 * Hex Floats:: Hexadecimal floating-point constants.
39 * Zero Length:: Zero-length arrays.
40 * Variable Length:: Arrays whose length is computed at run time.
41 * Empty Structures:: Structures with no members.
42 * Variadic Macros:: Macros with a variable number of arguments.
43 * Escaped Newlines:: Slightly looser rules for escaped newlines.
44 * Subscripting:: Any array can be subscripted, even if not an lvalue.
45 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
46 * Initializers:: Non-constant initializers.
47 * Compound Literals:: Compound literals give structures, unions
49 * Designated Inits:: Labeling elements of initializers.
50 * Cast to Union:: Casting to union type from any member of the union.
51 * Case Ranges:: `case 1 ... 9' and such.
52 * Mixed Declarations:: Mixing declarations and code.
53 * Function Attributes:: Declaring that functions have no side effects,
54 or that they can never return.
55 * Attribute Syntax:: Formal syntax for attributes.
56 * Function Prototypes:: Prototype declarations and old-style definitions.
57 * C++ Comments:: C++ comments are recognized.
58 * Dollar Signs:: Dollar sign is allowed in identifiers.
59 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
60 * Variable Attributes:: Specifying attributes of variables.
61 * Type Attributes:: Specifying attributes of types.
62 * Alignment:: Inquiring about the alignment of a type or variable.
63 * Inline:: Defining inline functions (as fast as macros).
64 * Extended Asm:: Assembler instructions with C expressions as operands.
65 (With them you can define ``built-in'' functions.)
66 * Constraints:: Constraints for asm operands
67 * Asm Labels:: Specifying the assembler name to use for a C symbol.
68 * Explicit Reg Vars:: Defining variables residing in specified registers.
69 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
70 * Incomplete Enums:: @code{enum foo;}, with details to follow.
71 * Function Names:: Printable strings which are the name of the current
73 * Return Address:: Getting the return or frame address of a function.
74 * Vector Extensions:: Using vector instructions through built-in functions.
75 * Offsetof:: Special syntax for implementing @code{offsetof}.
76 * Atomic Builtins:: Built-in functions for atomic memory access.
77 * Object Size Checking:: Built-in functions for limited buffer overflow
79 * Other Builtins:: Other built-in functions.
80 * Target Builtins:: Built-in functions specific to particular targets.
81 * Target Format Checks:: Format checks specific to particular targets.
82 * Pragmas:: Pragmas accepted by GCC.
83 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
84 * Thread-Local:: Per-thread variables.
85 * Binary constants:: Binary constants using the @samp{0b} prefix.
89 @section Statements and Declarations in Expressions
90 @cindex statements inside expressions
91 @cindex declarations inside expressions
92 @cindex expressions containing statements
93 @cindex macros, statements in expressions
95 @c the above section title wrapped and causes an underfull hbox.. i
96 @c changed it from "within" to "in". --mew 4feb93
97 A compound statement enclosed in parentheses may appear as an expression
98 in GNU C@. This allows you to use loops, switches, and local variables
101 Recall that a compound statement is a sequence of statements surrounded
102 by braces; in this construct, parentheses go around the braces. For
106 (@{ int y = foo (); int z;
113 is a valid (though slightly more complex than necessary) expression
114 for the absolute value of @code{foo ()}.
116 The last thing in the compound statement should be an expression
117 followed by a semicolon; the value of this subexpression serves as the
118 value of the entire construct. (If you use some other kind of statement
119 last within the braces, the construct has type @code{void}, and thus
120 effectively no value.)
122 This feature is especially useful in making macro definitions ``safe'' (so
123 that they evaluate each operand exactly once). For example, the
124 ``maximum'' function is commonly defined as a macro in standard C as
128 #define max(a,b) ((a) > (b) ? (a) : (b))
132 @cindex side effects, macro argument
133 But this definition computes either @var{a} or @var{b} twice, with bad
134 results if the operand has side effects. In GNU C, if you know the
135 type of the operands (here taken as @code{int}), you can define
136 the macro safely as follows:
139 #define maxint(a,b) \
140 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
143 Embedded statements are not allowed in constant expressions, such as
144 the value of an enumeration constant, the width of a bit-field, or
145 the initial value of a static variable.
147 If you don't know the type of the operand, you can still do this, but you
148 must use @code{typeof} (@pxref{Typeof}).
150 In G++, the result value of a statement expression undergoes array and
151 function pointer decay, and is returned by value to the enclosing
152 expression. For instance, if @code{A} is a class, then
161 will construct a temporary @code{A} object to hold the result of the
162 statement expression, and that will be used to invoke @code{Foo}.
163 Therefore the @code{this} pointer observed by @code{Foo} will not be the
166 Any temporaries created within a statement within a statement expression
167 will be destroyed at the statement's end. This makes statement
168 expressions inside macros slightly different from function calls. In
169 the latter case temporaries introduced during argument evaluation will
170 be destroyed at the end of the statement that includes the function
171 call. In the statement expression case they will be destroyed during
172 the statement expression. For instance,
175 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
176 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
186 will have different places where temporaries are destroyed. For the
187 @code{macro} case, the temporary @code{X} will be destroyed just after
188 the initialization of @code{b}. In the @code{function} case that
189 temporary will be destroyed when the function returns.
191 These considerations mean that it is probably a bad idea to use
192 statement-expressions of this form in header files that are designed to
193 work with C++. (Note that some versions of the GNU C Library contained
194 header files using statement-expression that lead to precisely this
197 Jumping into a statement expression with @code{goto} or using a
198 @code{switch} statement outside the statement expression with a
199 @code{case} or @code{default} label inside the statement expression is
200 not permitted. Jumping into a statement expression with a computed
201 @code{goto} (@pxref{Labels as Values}) yields undefined behavior.
202 Jumping out of a statement expression is permitted, but if the
203 statement expression is part of a larger expression then it is
204 unspecified which other subexpressions of that expression have been
205 evaluated except where the language definition requires certain
206 subexpressions to be evaluated before or after the statement
207 expression. In any case, as with a function call the evaluation of a
208 statement expression is not interleaved with the evaluation of other
209 parts of the containing expression. For example,
212 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
216 will call @code{foo} and @code{bar1} and will not call @code{baz} but
217 may or may not call @code{bar2}. If @code{bar2} is called, it will be
218 called after @code{foo} and before @code{bar1}
221 @section Locally Declared Labels
223 @cindex macros, local labels
225 GCC allows you to declare @dfn{local labels} in any nested block
226 scope. A local label is just like an ordinary label, but you can
227 only reference it (with a @code{goto} statement, or by taking its
228 address) within the block in which it was declared.
230 A local label declaration looks like this:
233 __label__ @var{label};
240 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
243 Local label declarations must come at the beginning of the block,
244 before any ordinary declarations or statements.
246 The label declaration defines the label @emph{name}, but does not define
247 the label itself. You must do this in the usual way, with
248 @code{@var{label}:}, within the statements of the statement expression.
250 The local label feature is useful for complex macros. If a macro
251 contains nested loops, a @code{goto} can be useful for breaking out of
252 them. However, an ordinary label whose scope is the whole function
253 cannot be used: if the macro can be expanded several times in one
254 function, the label will be multiply defined in that function. A
255 local label avoids this problem. For example:
258 #define SEARCH(value, array, target) \
261 typeof (target) _SEARCH_target = (target); \
262 typeof (*(array)) *_SEARCH_array = (array); \
265 for (i = 0; i < max; i++) \
266 for (j = 0; j < max; j++) \
267 if (_SEARCH_array[i][j] == _SEARCH_target) \
268 @{ (value) = i; goto found; @} \
274 This could also be written using a statement-expression:
277 #define SEARCH(array, target) \
280 typeof (target) _SEARCH_target = (target); \
281 typeof (*(array)) *_SEARCH_array = (array); \
284 for (i = 0; i < max; i++) \
285 for (j = 0; j < max; j++) \
286 if (_SEARCH_array[i][j] == _SEARCH_target) \
287 @{ value = i; goto found; @} \
294 Local label declarations also make the labels they declare visible to
295 nested functions, if there are any. @xref{Nested Functions}, for details.
297 @node Labels as Values
298 @section Labels as Values
299 @cindex labels as values
300 @cindex computed gotos
301 @cindex goto with computed label
302 @cindex address of a label
304 You can get the address of a label defined in the current function
305 (or a containing function) with the unary operator @samp{&&}. The
306 value has type @code{void *}. This value is a constant and can be used
307 wherever a constant of that type is valid. For example:
315 To use these values, you need to be able to jump to one. This is done
316 with the computed goto statement@footnote{The analogous feature in
317 Fortran is called an assigned goto, but that name seems inappropriate in
318 C, where one can do more than simply store label addresses in label
319 variables.}, @code{goto *@var{exp};}. For example,
326 Any expression of type @code{void *} is allowed.
328 One way of using these constants is in initializing a static array that
329 will serve as a jump table:
332 static void *array[] = @{ &&foo, &&bar, &&hack @};
335 Then you can select a label with indexing, like this:
342 Note that this does not check whether the subscript is in bounds---array
343 indexing in C never does that.
345 Such an array of label values serves a purpose much like that of the
346 @code{switch} statement. The @code{switch} statement is cleaner, so
347 use that rather than an array unless the problem does not fit a
348 @code{switch} statement very well.
350 Another use of label values is in an interpreter for threaded code.
351 The labels within the interpreter function can be stored in the
352 threaded code for super-fast dispatching.
354 You may not use this mechanism to jump to code in a different function.
355 If you do that, totally unpredictable things will happen. The best way to
356 avoid this is to store the label address only in automatic variables and
357 never pass it as an argument.
359 An alternate way to write the above example is
362 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
364 goto *(&&foo + array[i]);
368 This is more friendly to code living in shared libraries, as it reduces
369 the number of dynamic relocations that are needed, and by consequence,
370 allows the data to be read-only.
372 @node Nested Functions
373 @section Nested Functions
374 @cindex nested functions
375 @cindex downward funargs
378 A @dfn{nested function} is a function defined inside another function.
379 (Nested functions are not supported for GNU C++.) The nested function's
380 name is local to the block where it is defined. For example, here we
381 define a nested function named @code{square}, and call it twice:
385 foo (double a, double b)
387 double square (double z) @{ return z * z; @}
389 return square (a) + square (b);
394 The nested function can access all the variables of the containing
395 function that are visible at the point of its definition. This is
396 called @dfn{lexical scoping}. For example, here we show a nested
397 function which uses an inherited variable named @code{offset}:
401 bar (int *array, int offset, int size)
403 int access (int *array, int index)
404 @{ return array[index + offset]; @}
407 for (i = 0; i < size; i++)
408 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
413 Nested function definitions are permitted within functions in the places
414 where variable definitions are allowed; that is, in any block, mixed
415 with the other declarations and statements in the block.
417 It is possible to call the nested function from outside the scope of its
418 name by storing its address or passing the address to another function:
421 hack (int *array, int size)
423 void store (int index, int value)
424 @{ array[index] = value; @}
426 intermediate (store, size);
430 Here, the function @code{intermediate} receives the address of
431 @code{store} as an argument. If @code{intermediate} calls @code{store},
432 the arguments given to @code{store} are used to store into @code{array}.
433 But this technique works only so long as the containing function
434 (@code{hack}, in this example) does not exit.
436 If you try to call the nested function through its address after the
437 containing function has exited, all hell will break loose. If you try
438 to call it after a containing scope level has exited, and if it refers
439 to some of the variables that are no longer in scope, you may be lucky,
440 but it's not wise to take the risk. If, however, the nested function
441 does not refer to anything that has gone out of scope, you should be
444 GCC implements taking the address of a nested function using a technique
445 called @dfn{trampolines}. A paper describing them is available as
448 @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}.
450 A nested function can jump to a label inherited from a containing
451 function, provided the label was explicitly declared in the containing
452 function (@pxref{Local Labels}). Such a jump returns instantly to the
453 containing function, exiting the nested function which did the
454 @code{goto} and any intermediate functions as well. Here is an example:
458 bar (int *array, int offset, int size)
461 int access (int *array, int index)
465 return array[index + offset];
469 for (i = 0; i < size; i++)
470 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
474 /* @r{Control comes here from @code{access}
475 if it detects an error.} */
482 A nested function always has no linkage. Declaring one with
483 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
484 before its definition, use @code{auto} (which is otherwise meaningless
485 for function declarations).
488 bar (int *array, int offset, int size)
491 auto int access (int *, int);
493 int access (int *array, int index)
497 return array[index + offset];
503 @node Constructing Calls
504 @section Constructing Function Calls
505 @cindex constructing calls
506 @cindex forwarding calls
508 Using the built-in functions described below, you can record
509 the arguments a function received, and call another function
510 with the same arguments, without knowing the number or types
513 You can also record the return value of that function call,
514 and later return that value, without knowing what data type
515 the function tried to return (as long as your caller expects
518 However, these built-in functions may interact badly with some
519 sophisticated features or other extensions of the language. It
520 is, therefore, not recommended to use them outside very simple
521 functions acting as mere forwarders for their arguments.
523 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
524 This built-in function returns a pointer to data
525 describing how to perform a call with the same arguments as were passed
526 to the current function.
528 The function saves the arg pointer register, structure value address,
529 and all registers that might be used to pass arguments to a function
530 into a block of memory allocated on the stack. Then it returns the
531 address of that block.
534 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
535 This built-in function invokes @var{function}
536 with a copy of the parameters described by @var{arguments}
539 The value of @var{arguments} should be the value returned by
540 @code{__builtin_apply_args}. The argument @var{size} specifies the size
541 of the stack argument data, in bytes.
543 This function returns a pointer to data describing
544 how to return whatever value was returned by @var{function}. The data
545 is saved in a block of memory allocated on the stack.
547 It is not always simple to compute the proper value for @var{size}. The
548 value is used by @code{__builtin_apply} to compute the amount of data
549 that should be pushed on the stack and copied from the incoming argument
553 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
554 This built-in function returns the value described by @var{result} from
555 the containing function. You should specify, for @var{result}, a value
556 returned by @code{__builtin_apply}.
560 @section Referring to a Type with @code{typeof}
563 @cindex macros, types of arguments
565 Another way to refer to the type of an expression is with @code{typeof}.
566 The syntax of using of this keyword looks like @code{sizeof}, but the
567 construct acts semantically like a type name defined with @code{typedef}.
569 There are two ways of writing the argument to @code{typeof}: with an
570 expression or with a type. Here is an example with an expression:
577 This assumes that @code{x} is an array of pointers to functions;
578 the type described is that of the values of the functions.
580 Here is an example with a typename as the argument:
587 Here the type described is that of pointers to @code{int}.
589 If you are writing a header file that must work when included in ISO C
590 programs, write @code{__typeof__} instead of @code{typeof}.
591 @xref{Alternate Keywords}.
593 A @code{typeof}-construct can be used anywhere a typedef name could be
594 used. For example, you can use it in a declaration, in a cast, or inside
595 of @code{sizeof} or @code{typeof}.
597 @code{typeof} is often useful in conjunction with the
598 statements-within-expressions feature. Here is how the two together can
599 be used to define a safe ``maximum'' macro that operates on any
600 arithmetic type and evaluates each of its arguments exactly once:
604 (@{ typeof (a) _a = (a); \
605 typeof (b) _b = (b); \
606 _a > _b ? _a : _b; @})
609 @cindex underscores in variables in macros
610 @cindex @samp{_} in variables in macros
611 @cindex local variables in macros
612 @cindex variables, local, in macros
613 @cindex macros, local variables in
615 The reason for using names that start with underscores for the local
616 variables is to avoid conflicts with variable names that occur within the
617 expressions that are substituted for @code{a} and @code{b}. Eventually we
618 hope to design a new form of declaration syntax that allows you to declare
619 variables whose scopes start only after their initializers; this will be a
620 more reliable way to prevent such conflicts.
623 Some more examples of the use of @code{typeof}:
627 This declares @code{y} with the type of what @code{x} points to.
634 This declares @code{y} as an array of such values.
641 This declares @code{y} as an array of pointers to characters:
644 typeof (typeof (char *)[4]) y;
648 It is equivalent to the following traditional C declaration:
654 To see the meaning of the declaration using @code{typeof}, and why it
655 might be a useful way to write, rewrite it with these macros:
658 #define pointer(T) typeof(T *)
659 #define array(T, N) typeof(T [N])
663 Now the declaration can be rewritten this way:
666 array (pointer (char), 4) y;
670 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
671 pointers to @code{char}.
674 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
675 a more limited extension which permitted one to write
678 typedef @var{T} = @var{expr};
682 with the effect of declaring @var{T} to have the type of the expression
683 @var{expr}. This extension does not work with GCC 3 (versions between
684 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
685 relies on it should be rewritten to use @code{typeof}:
688 typedef typeof(@var{expr}) @var{T};
692 This will work with all versions of GCC@.
695 @section Conditionals with Omitted Operands
696 @cindex conditional expressions, extensions
697 @cindex omitted middle-operands
698 @cindex middle-operands, omitted
699 @cindex extensions, @code{?:}
700 @cindex @code{?:} extensions
702 The middle operand in a conditional expression may be omitted. Then
703 if the first operand is nonzero, its value is the value of the conditional
706 Therefore, the expression
713 has the value of @code{x} if that is nonzero; otherwise, the value of
716 This example is perfectly equivalent to
722 @cindex side effect in ?:
723 @cindex ?: side effect
725 In this simple case, the ability to omit the middle operand is not
726 especially useful. When it becomes useful is when the first operand does,
727 or may (if it is a macro argument), contain a side effect. Then repeating
728 the operand in the middle would perform the side effect twice. Omitting
729 the middle operand uses the value already computed without the undesirable
730 effects of recomputing it.
733 @section Double-Word Integers
734 @cindex @code{long long} data types
735 @cindex double-word arithmetic
736 @cindex multiprecision arithmetic
737 @cindex @code{LL} integer suffix
738 @cindex @code{ULL} integer suffix
740 ISO C99 supports data types for integers that are at least 64 bits wide,
741 and as an extension GCC supports them in C89 mode and in C++.
742 Simply write @code{long long int} for a signed integer, or
743 @code{unsigned long long int} for an unsigned integer. To make an
744 integer constant of type @code{long long int}, add the suffix @samp{LL}
745 to the integer. To make an integer constant of type @code{unsigned long
746 long int}, add the suffix @samp{ULL} to the integer.
748 You can use these types in arithmetic like any other integer types.
749 Addition, subtraction, and bitwise boolean operations on these types
750 are open-coded on all types of machines. Multiplication is open-coded
751 if the machine supports fullword-to-doubleword a widening multiply
752 instruction. Division and shifts are open-coded only on machines that
753 provide special support. The operations that are not open-coded use
754 special library routines that come with GCC@.
756 There may be pitfalls when you use @code{long long} types for function
757 arguments, unless you declare function prototypes. If a function
758 expects type @code{int} for its argument, and you pass a value of type
759 @code{long long int}, confusion will result because the caller and the
760 subroutine will disagree about the number of bytes for the argument.
761 Likewise, if the function expects @code{long long int} and you pass
762 @code{int}. The best way to avoid such problems is to use prototypes.
765 @section Complex Numbers
766 @cindex complex numbers
767 @cindex @code{_Complex} keyword
768 @cindex @code{__complex__} keyword
770 ISO C99 supports complex floating data types, and as an extension GCC
771 supports them in C89 mode and in C++, and supports complex integer data
772 types which are not part of ISO C99. You can declare complex types
773 using the keyword @code{_Complex}. As an extension, the older GNU
774 keyword @code{__complex__} is also supported.
776 For example, @samp{_Complex double x;} declares @code{x} as a
777 variable whose real part and imaginary part are both of type
778 @code{double}. @samp{_Complex short int y;} declares @code{y} to
779 have real and imaginary parts of type @code{short int}; this is not
780 likely to be useful, but it shows that the set of complex types is
783 To write a constant with a complex data type, use the suffix @samp{i} or
784 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
785 has type @code{_Complex float} and @code{3i} has type
786 @code{_Complex int}. Such a constant always has a pure imaginary
787 value, but you can form any complex value you like by adding one to a
788 real constant. This is a GNU extension; if you have an ISO C99
789 conforming C library (such as GNU libc), and want to construct complex
790 constants of floating type, you should include @code{<complex.h>} and
791 use the macros @code{I} or @code{_Complex_I} instead.
793 @cindex @code{__real__} keyword
794 @cindex @code{__imag__} keyword
795 To extract the real part of a complex-valued expression @var{exp}, write
796 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
797 extract the imaginary part. This is a GNU extension; for values of
798 floating type, you should use the ISO C99 functions @code{crealf},
799 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
800 @code{cimagl}, declared in @code{<complex.h>} and also provided as
801 built-in functions by GCC@.
803 @cindex complex conjugation
804 The operator @samp{~} performs complex conjugation when used on a value
805 with a complex type. This is a GNU extension; for values of
806 floating type, you should use the ISO C99 functions @code{conjf},
807 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
808 provided as built-in functions by GCC@.
810 GCC can allocate complex automatic variables in a noncontiguous
811 fashion; it's even possible for the real part to be in a register while
812 the imaginary part is on the stack (or vice-versa). Only the DWARF2
813 debug info format can represent this, so use of DWARF2 is recommended.
814 If you are using the stabs debug info format, GCC describes a noncontiguous
815 complex variable as if it were two separate variables of noncomplex type.
816 If the variable's actual name is @code{foo}, the two fictitious
817 variables are named @code{foo$real} and @code{foo$imag}. You can
818 examine and set these two fictitious variables with your debugger.
821 @section Additional Floating Types
822 @cindex additional floating types
823 @cindex @code{__float80} data type
824 @cindex @code{__float128} data type
825 @cindex @code{w} floating point suffix
826 @cindex @code{q} floating point suffix
827 @cindex @code{W} floating point suffix
828 @cindex @code{Q} floating point suffix
830 As an extension, the GNU C compiler supports additional floating
831 types, @code{__float80} and @code{__float128} to support 80bit
832 (@code{XFmode}) and 128 bit (@code{TFmode}) floating types.
833 Support for additional types includes the arithmetic operators:
834 add, subtract, multiply, divide; unary arithmetic operators;
835 relational operators; equality operators; and conversions to and from
836 integer and other floating types. Use a suffix @samp{w} or @samp{W}
837 in a literal constant of type @code{__float80} and @samp{q} or @samp{Q}
838 for @code{_float128}. You can declare complex types using the
839 corresponding internal complex type, @code{XCmode} for @code{__float80}
840 type and @code{TCmode} for @code{__float128} type:
843 typedef _Complex float __attribute__((mode(TC))) _Complex128;
844 typedef _Complex float __attribute__((mode(XC))) _Complex80;
847 Not all targets support additional floating point types. @code{__float80}
848 is supported on i386, x86_64 and ia64 targets and target @code{__float128}
849 is supported on x86_64 and ia64 targets.
852 @section Decimal Floating Types
853 @cindex decimal floating types
854 @cindex @code{_Decimal32} data type
855 @cindex @code{_Decimal64} data type
856 @cindex @code{_Decimal128} data type
857 @cindex @code{df} integer suffix
858 @cindex @code{dd} integer suffix
859 @cindex @code{dl} integer suffix
860 @cindex @code{DF} integer suffix
861 @cindex @code{DD} integer suffix
862 @cindex @code{DL} integer suffix
864 As an extension, the GNU C compiler supports decimal floating types as
865 defined in the N1176 draft of ISO/IEC WDTR24732. Support for decimal
866 floating types in GCC will evolve as the draft technical report changes.
867 Calling conventions for any target might also change. Not all targets
868 support decimal floating types.
870 The decimal floating types are @code{_Decimal32}, @code{_Decimal64}, and
871 @code{_Decimal128}. They use a radix of ten, unlike the floating types
872 @code{float}, @code{double}, and @code{long double} whose radix is not
873 specified by the C standard but is usually two.
875 Support for decimal floating types includes the arithmetic operators
876 add, subtract, multiply, divide; unary arithmetic operators;
877 relational operators; equality operators; and conversions to and from
878 integer and other floating types. Use a suffix @samp{df} or
879 @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd}
880 or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for
883 GCC support of decimal float as specified by the draft technical report
888 Translation time data type (TTDT) is not supported.
891 When the value of a decimal floating type cannot be represented in the
892 integer type to which it is being converted, the result is undefined
893 rather than the result value specified by the draft technical report.
896 Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128}
897 are supported by the DWARF2 debug information format.
903 ISO C99 supports floating-point numbers written not only in the usual
904 decimal notation, such as @code{1.55e1}, but also numbers such as
905 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
906 supports this in C89 mode (except in some cases when strictly
907 conforming) and in C++. In that format the
908 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
909 mandatory. The exponent is a decimal number that indicates the power of
910 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
917 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
918 is the same as @code{1.55e1}.
920 Unlike for floating-point numbers in the decimal notation the exponent
921 is always required in the hexadecimal notation. Otherwise the compiler
922 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
923 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
924 extension for floating-point constants of type @code{float}.
927 @section Arrays of Length Zero
928 @cindex arrays of length zero
929 @cindex zero-length arrays
930 @cindex length-zero arrays
931 @cindex flexible array members
933 Zero-length arrays are allowed in GNU C@. They are very useful as the
934 last element of a structure which is really a header for a variable-length
943 struct line *thisline = (struct line *)
944 malloc (sizeof (struct line) + this_length);
945 thisline->length = this_length;
948 In ISO C90, you would have to give @code{contents} a length of 1, which
949 means either you waste space or complicate the argument to @code{malloc}.
951 In ISO C99, you would use a @dfn{flexible array member}, which is
952 slightly different in syntax and semantics:
956 Flexible array members are written as @code{contents[]} without
960 Flexible array members have incomplete type, and so the @code{sizeof}
961 operator may not be applied. As a quirk of the original implementation
962 of zero-length arrays, @code{sizeof} evaluates to zero.
965 Flexible array members may only appear as the last member of a
966 @code{struct} that is otherwise non-empty.
969 A structure containing a flexible array member, or a union containing
970 such a structure (possibly recursively), may not be a member of a
971 structure or an element of an array. (However, these uses are
972 permitted by GCC as extensions.)
975 GCC versions before 3.0 allowed zero-length arrays to be statically
976 initialized, as if they were flexible arrays. In addition to those
977 cases that were useful, it also allowed initializations in situations
978 that would corrupt later data. Non-empty initialization of zero-length
979 arrays is now treated like any case where there are more initializer
980 elements than the array holds, in that a suitable warning about "excess
981 elements in array" is given, and the excess elements (all of them, in
982 this case) are ignored.
984 Instead GCC allows static initialization of flexible array members.
985 This is equivalent to defining a new structure containing the original
986 structure followed by an array of sufficient size to contain the data.
987 I.e.@: in the following, @code{f1} is constructed as if it were declared
993 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
996 struct f1 f1; int data[3];
997 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
1001 The convenience of this extension is that @code{f1} has the desired
1002 type, eliminating the need to consistently refer to @code{f2.f1}.
1004 This has symmetry with normal static arrays, in that an array of
1005 unknown size is also written with @code{[]}.
1007 Of course, this extension only makes sense if the extra data comes at
1008 the end of a top-level object, as otherwise we would be overwriting
1009 data at subsequent offsets. To avoid undue complication and confusion
1010 with initialization of deeply nested arrays, we simply disallow any
1011 non-empty initialization except when the structure is the top-level
1012 object. For example:
1015 struct foo @{ int x; int y[]; @};
1016 struct bar @{ struct foo z; @};
1018 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
1019 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1020 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
1021 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
1024 @node Empty Structures
1025 @section Structures With No Members
1026 @cindex empty structures
1027 @cindex zero-size structures
1029 GCC permits a C structure to have no members:
1036 The structure will have size zero. In C++, empty structures are part
1037 of the language. G++ treats empty structures as if they had a single
1038 member of type @code{char}.
1040 @node Variable Length
1041 @section Arrays of Variable Length
1042 @cindex variable-length arrays
1043 @cindex arrays of variable length
1046 Variable-length automatic arrays are allowed in ISO C99, and as an
1047 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1048 implementation of variable-length arrays does not yet conform in detail
1049 to the ISO C99 standard.) These arrays are
1050 declared like any other automatic arrays, but with a length that is not
1051 a constant expression. The storage is allocated at the point of
1052 declaration and deallocated when the brace-level is exited. For
1057 concat_fopen (char *s1, char *s2, char *mode)
1059 char str[strlen (s1) + strlen (s2) + 1];
1062 return fopen (str, mode);
1066 @cindex scope of a variable length array
1067 @cindex variable-length array scope
1068 @cindex deallocating variable length arrays
1069 Jumping or breaking out of the scope of the array name deallocates the
1070 storage. Jumping into the scope is not allowed; you get an error
1073 @cindex @code{alloca} vs variable-length arrays
1074 You can use the function @code{alloca} to get an effect much like
1075 variable-length arrays. The function @code{alloca} is available in
1076 many other C implementations (but not in all). On the other hand,
1077 variable-length arrays are more elegant.
1079 There are other differences between these two methods. Space allocated
1080 with @code{alloca} exists until the containing @emph{function} returns.
1081 The space for a variable-length array is deallocated as soon as the array
1082 name's scope ends. (If you use both variable-length arrays and
1083 @code{alloca} in the same function, deallocation of a variable-length array
1084 will also deallocate anything more recently allocated with @code{alloca}.)
1086 You can also use variable-length arrays as arguments to functions:
1090 tester (int len, char data[len][len])
1096 The length of an array is computed once when the storage is allocated
1097 and is remembered for the scope of the array in case you access it with
1100 If you want to pass the array first and the length afterward, you can
1101 use a forward declaration in the parameter list---another GNU extension.
1105 tester (int len; char data[len][len], int len)
1111 @cindex parameter forward declaration
1112 The @samp{int len} before the semicolon is a @dfn{parameter forward
1113 declaration}, and it serves the purpose of making the name @code{len}
1114 known when the declaration of @code{data} is parsed.
1116 You can write any number of such parameter forward declarations in the
1117 parameter list. They can be separated by commas or semicolons, but the
1118 last one must end with a semicolon, which is followed by the ``real''
1119 parameter declarations. Each forward declaration must match a ``real''
1120 declaration in parameter name and data type. ISO C99 does not support
1121 parameter forward declarations.
1123 @node Variadic Macros
1124 @section Macros with a Variable Number of Arguments.
1125 @cindex variable number of arguments
1126 @cindex macro with variable arguments
1127 @cindex rest argument (in macro)
1128 @cindex variadic macros
1130 In the ISO C standard of 1999, a macro can be declared to accept a
1131 variable number of arguments much as a function can. The syntax for
1132 defining the macro is similar to that of a function. Here is an
1136 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1139 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1140 such a macro, it represents the zero or more tokens until the closing
1141 parenthesis that ends the invocation, including any commas. This set of
1142 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1143 wherever it appears. See the CPP manual for more information.
1145 GCC has long supported variadic macros, and used a different syntax that
1146 allowed you to give a name to the variable arguments just like any other
1147 argument. Here is an example:
1150 #define debug(format, args...) fprintf (stderr, format, args)
1153 This is in all ways equivalent to the ISO C example above, but arguably
1154 more readable and descriptive.
1156 GNU CPP has two further variadic macro extensions, and permits them to
1157 be used with either of the above forms of macro definition.
1159 In standard C, you are not allowed to leave the variable argument out
1160 entirely; but you are allowed to pass an empty argument. For example,
1161 this invocation is invalid in ISO C, because there is no comma after
1168 GNU CPP permits you to completely omit the variable arguments in this
1169 way. In the above examples, the compiler would complain, though since
1170 the expansion of the macro still has the extra comma after the format
1173 To help solve this problem, CPP behaves specially for variable arguments
1174 used with the token paste operator, @samp{##}. If instead you write
1177 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1180 and if the variable arguments are omitted or empty, the @samp{##}
1181 operator causes the preprocessor to remove the comma before it. If you
1182 do provide some variable arguments in your macro invocation, GNU CPP
1183 does not complain about the paste operation and instead places the
1184 variable arguments after the comma. Just like any other pasted macro
1185 argument, these arguments are not macro expanded.
1187 @node Escaped Newlines
1188 @section Slightly Looser Rules for Escaped Newlines
1189 @cindex escaped newlines
1190 @cindex newlines (escaped)
1192 Recently, the preprocessor has relaxed its treatment of escaped
1193 newlines. Previously, the newline had to immediately follow a
1194 backslash. The current implementation allows whitespace in the form
1195 of spaces, horizontal and vertical tabs, and form feeds between the
1196 backslash and the subsequent newline. The preprocessor issues a
1197 warning, but treats it as a valid escaped newline and combines the two
1198 lines to form a single logical line. This works within comments and
1199 tokens, as well as between tokens. Comments are @emph{not} treated as
1200 whitespace for the purposes of this relaxation, since they have not
1201 yet been replaced with spaces.
1204 @section Non-Lvalue Arrays May Have Subscripts
1205 @cindex subscripting
1206 @cindex arrays, non-lvalue
1208 @cindex subscripting and function values
1209 In ISO C99, arrays that are not lvalues still decay to pointers, and
1210 may be subscripted, although they may not be modified or used after
1211 the next sequence point and the unary @samp{&} operator may not be
1212 applied to them. As an extension, GCC allows such arrays to be
1213 subscripted in C89 mode, though otherwise they do not decay to
1214 pointers outside C99 mode. For example,
1215 this is valid in GNU C though not valid in C89:
1219 struct foo @{int a[4];@};
1225 return f().a[index];
1231 @section Arithmetic on @code{void}- and Function-Pointers
1232 @cindex void pointers, arithmetic
1233 @cindex void, size of pointer to
1234 @cindex function pointers, arithmetic
1235 @cindex function, size of pointer to
1237 In GNU C, addition and subtraction operations are supported on pointers to
1238 @code{void} and on pointers to functions. This is done by treating the
1239 size of a @code{void} or of a function as 1.
1241 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1242 and on function types, and returns 1.
1244 @opindex Wpointer-arith
1245 The option @option{-Wpointer-arith} requests a warning if these extensions
1249 @section Non-Constant Initializers
1250 @cindex initializers, non-constant
1251 @cindex non-constant initializers
1253 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1254 automatic variable are not required to be constant expressions in GNU C@.
1255 Here is an example of an initializer with run-time varying elements:
1258 foo (float f, float g)
1260 float beat_freqs[2] = @{ f-g, f+g @};
1265 @node Compound Literals
1266 @section Compound Literals
1267 @cindex constructor expressions
1268 @cindex initializations in expressions
1269 @cindex structures, constructor expression
1270 @cindex expressions, constructor
1271 @cindex compound literals
1272 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1274 ISO C99 supports compound literals. A compound literal looks like
1275 a cast containing an initializer. Its value is an object of the
1276 type specified in the cast, containing the elements specified in
1277 the initializer; it is an lvalue. As an extension, GCC supports
1278 compound literals in C89 mode and in C++.
1280 Usually, the specified type is a structure. Assume that
1281 @code{struct foo} and @code{structure} are declared as shown:
1284 struct foo @{int a; char b[2];@} structure;
1288 Here is an example of constructing a @code{struct foo} with a compound literal:
1291 structure = ((struct foo) @{x + y, 'a', 0@});
1295 This is equivalent to writing the following:
1299 struct foo temp = @{x + y, 'a', 0@};
1304 You can also construct an array. If all the elements of the compound literal
1305 are (made up of) simple constant expressions, suitable for use in
1306 initializers of objects of static storage duration, then the compound
1307 literal can be coerced to a pointer to its first element and used in
1308 such an initializer, as shown here:
1311 char **foo = (char *[]) @{ "x", "y", "z" @};
1314 Compound literals for scalar types and union types are is
1315 also allowed, but then the compound literal is equivalent
1318 As a GNU extension, GCC allows initialization of objects with static storage
1319 duration by compound literals (which is not possible in ISO C99, because
1320 the initializer is not a constant).
1321 It is handled as if the object was initialized only with the bracket
1322 enclosed list if the types of the compound literal and the object match.
1323 The initializer list of the compound literal must be constant.
1324 If the object being initialized has array type of unknown size, the size is
1325 determined by compound literal size.
1328 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1329 static int y[] = (int []) @{1, 2, 3@};
1330 static int z[] = (int [3]) @{1@};
1334 The above lines are equivalent to the following:
1336 static struct foo x = @{1, 'a', 'b'@};
1337 static int y[] = @{1, 2, 3@};
1338 static int z[] = @{1, 0, 0@};
1341 @node Designated Inits
1342 @section Designated Initializers
1343 @cindex initializers with labeled elements
1344 @cindex labeled elements in initializers
1345 @cindex case labels in initializers
1346 @cindex designated initializers
1348 Standard C89 requires the elements of an initializer to appear in a fixed
1349 order, the same as the order of the elements in the array or structure
1352 In ISO C99 you can give the elements in any order, specifying the array
1353 indices or structure field names they apply to, and GNU C allows this as
1354 an extension in C89 mode as well. This extension is not
1355 implemented in GNU C++.
1357 To specify an array index, write
1358 @samp{[@var{index}] =} before the element value. For example,
1361 int a[6] = @{ [4] = 29, [2] = 15 @};
1368 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1372 The index values must be constant expressions, even if the array being
1373 initialized is automatic.
1375 An alternative syntax for this which has been obsolete since GCC 2.5 but
1376 GCC still accepts is to write @samp{[@var{index}]} before the element
1377 value, with no @samp{=}.
1379 To initialize a range of elements to the same value, write
1380 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1381 extension. For example,
1384 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1388 If the value in it has side-effects, the side-effects will happen only once,
1389 not for each initialized field by the range initializer.
1392 Note that the length of the array is the highest value specified
1395 In a structure initializer, specify the name of a field to initialize
1396 with @samp{.@var{fieldname} =} before the element value. For example,
1397 given the following structure,
1400 struct point @{ int x, y; @};
1404 the following initialization
1407 struct point p = @{ .y = yvalue, .x = xvalue @};
1414 struct point p = @{ xvalue, yvalue @};
1417 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1418 @samp{@var{fieldname}:}, as shown here:
1421 struct point p = @{ y: yvalue, x: xvalue @};
1425 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1426 @dfn{designator}. You can also use a designator (or the obsolete colon
1427 syntax) when initializing a union, to specify which element of the union
1428 should be used. For example,
1431 union foo @{ int i; double d; @};
1433 union foo f = @{ .d = 4 @};
1437 will convert 4 to a @code{double} to store it in the union using
1438 the second element. By contrast, casting 4 to type @code{union foo}
1439 would store it into the union as the integer @code{i}, since it is
1440 an integer. (@xref{Cast to Union}.)
1442 You can combine this technique of naming elements with ordinary C
1443 initialization of successive elements. Each initializer element that
1444 does not have a designator applies to the next consecutive element of the
1445 array or structure. For example,
1448 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1455 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1458 Labeling the elements of an array initializer is especially useful
1459 when the indices are characters or belong to an @code{enum} type.
1464 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1465 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1468 @cindex designator lists
1469 You can also write a series of @samp{.@var{fieldname}} and
1470 @samp{[@var{index}]} designators before an @samp{=} to specify a
1471 nested subobject to initialize; the list is taken relative to the
1472 subobject corresponding to the closest surrounding brace pair. For
1473 example, with the @samp{struct point} declaration above:
1476 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1480 If the same field is initialized multiple times, it will have value from
1481 the last initialization. If any such overridden initialization has
1482 side-effect, it is unspecified whether the side-effect happens or not.
1483 Currently, GCC will discard them and issue a warning.
1486 @section Case Ranges
1488 @cindex ranges in case statements
1490 You can specify a range of consecutive values in a single @code{case} label,
1494 case @var{low} ... @var{high}:
1498 This has the same effect as the proper number of individual @code{case}
1499 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1501 This feature is especially useful for ranges of ASCII character codes:
1507 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1508 it may be parsed wrong when you use it with integer values. For example,
1523 @section Cast to a Union Type
1524 @cindex cast to a union
1525 @cindex union, casting to a
1527 A cast to union type is similar to other casts, except that the type
1528 specified is a union type. You can specify the type either with
1529 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1530 a constructor though, not a cast, and hence does not yield an lvalue like
1531 normal casts. (@xref{Compound Literals}.)
1533 The types that may be cast to the union type are those of the members
1534 of the union. Thus, given the following union and variables:
1537 union foo @{ int i; double d; @};
1543 both @code{x} and @code{y} can be cast to type @code{union foo}.
1545 Using the cast as the right-hand side of an assignment to a variable of
1546 union type is equivalent to storing in a member of the union:
1551 u = (union foo) x @equiv{} u.i = x
1552 u = (union foo) y @equiv{} u.d = y
1555 You can also use the union cast as a function argument:
1558 void hack (union foo);
1560 hack ((union foo) x);
1563 @node Mixed Declarations
1564 @section Mixed Declarations and Code
1565 @cindex mixed declarations and code
1566 @cindex declarations, mixed with code
1567 @cindex code, mixed with declarations
1569 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1570 within compound statements. As an extension, GCC also allows this in
1571 C89 mode. For example, you could do:
1580 Each identifier is visible from where it is declared until the end of
1581 the enclosing block.
1583 @node Function Attributes
1584 @section Declaring Attributes of Functions
1585 @cindex function attributes
1586 @cindex declaring attributes of functions
1587 @cindex functions that never return
1588 @cindex functions that return more than once
1589 @cindex functions that have no side effects
1590 @cindex functions in arbitrary sections
1591 @cindex functions that behave like malloc
1592 @cindex @code{volatile} applied to function
1593 @cindex @code{const} applied to function
1594 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1595 @cindex functions with non-null pointer arguments
1596 @cindex functions that are passed arguments in registers on the 386
1597 @cindex functions that pop the argument stack on the 386
1598 @cindex functions that do not pop the argument stack on the 386
1600 In GNU C, you declare certain things about functions called in your program
1601 which help the compiler optimize function calls and check your code more
1604 The keyword @code{__attribute__} allows you to specify special
1605 attributes when making a declaration. This keyword is followed by an
1606 attribute specification inside double parentheses. The following
1607 attributes are currently defined for functions on all targets:
1608 @code{aligned}, @code{alloc_size}, @code{noreturn},
1609 @code{returns_twice}, @code{noinline}, @code{always_inline},
1610 @code{flatten}, @code{pure}, @code{const}, @code{nothrow},
1611 @code{sentinel}, @code{format}, @code{format_arg},
1612 @code{no_instrument_function}, @code{section}, @code{constructor},
1613 @code{destructor}, @code{used}, @code{unused}, @code{deprecated},
1614 @code{weak}, @code{malloc}, @code{alias}, @code{warn_unused_result},
1615 @code{nonnull}, @code{gnu_inline} and @code{externally_visible},
1616 @code{hot}, @code{cold}.
1617 Several other attributes are defined for functions on particular
1618 target systems. Other attributes, including @code{section} are
1619 supported for variables declarations (@pxref{Variable Attributes}) and
1620 for types (@pxref{Type Attributes}).
1622 You may also specify attributes with @samp{__} preceding and following
1623 each keyword. This allows you to use them in header files without
1624 being concerned about a possible macro of the same name. For example,
1625 you may use @code{__noreturn__} instead of @code{noreturn}.
1627 @xref{Attribute Syntax}, for details of the exact syntax for using
1631 @c Keep this table alphabetized by attribute name. Treat _ as space.
1633 @item alias ("@var{target}")
1634 @cindex @code{alias} attribute
1635 The @code{alias} attribute causes the declaration to be emitted as an
1636 alias for another symbol, which must be specified. For instance,
1639 void __f () @{ /* @r{Do something.} */; @}
1640 void f () __attribute__ ((weak, alias ("__f")));
1643 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
1644 mangled name for the target must be used. It is an error if @samp{__f}
1645 is not defined in the same translation unit.
1647 Not all target machines support this attribute.
1649 @item aligned (@var{alignment})
1650 @cindex @code{aligned} attribute
1651 This attribute specifies a minimum alignment for the function,
1654 You cannot use this attribute to decrease the alignment of a function,
1655 only to increase it. However, when you explicitly specify a function
1656 alignment this will override the effect of the
1657 @option{-falign-functions} (@pxref{Optimize Options}) option for this
1660 Note that the effectiveness of @code{aligned} attributes may be
1661 limited by inherent limitations in your linker. On many systems, the
1662 linker is only able to arrange for functions to be aligned up to a
1663 certain maximum alignment. (For some linkers, the maximum supported
1664 alignment may be very very small.) See your linker documentation for
1665 further information.
1667 The @code{aligned} attribute can also be used for variables and fields
1668 (@pxref{Variable Attributes}.)
1671 @cindex @code{alloc_size} attribute
1672 The @code{alloc_size} attribute is used to tell the compiler that the
1673 function return value points to memory, where the size is given by
1674 one or two of the functions parameters. GCC uses this
1675 information to improve the correctness of @code{__builtin_object_size}.
1677 The function parameter(s) denoting the allocated size are specified by
1678 one or two integer arguments supplied to the attribute. The allocated size
1679 is either the value of the single function argument specified or the product
1680 of the two function arguments specified. Argument numbering starts at
1686 void* my_calloc(size_t, size_t) __attribute__((alloc_size(1,2)))
1687 void my_realloc(void* size_t) __attribute__((alloc_size(2)))
1690 declares that my_calloc will return memory of the size given by
1691 the product of parameter 1 and 2 and that my_realloc will return memory
1692 of the size given by parameter 2.
1695 @cindex @code{always_inline} function attribute
1696 Generally, functions are not inlined unless optimization is specified.
1697 For functions declared inline, this attribute inlines the function even
1698 if no optimization level was specified.
1701 @cindex @code{gnu_inline} function attribute
1702 This attribute should be used with a function which is also declared
1703 with the @code{inline} keyword. It directs GCC to treat the function
1704 as if it were defined in gnu89 mode even when compiling in C99 or
1707 If the function is declared @code{extern}, then this definition of the
1708 function is used only for inlining. In no case is the function
1709 compiled as a standalone function, not even if you take its address
1710 explicitly. Such an address becomes an external reference, as if you
1711 had only declared the function, and had not defined it. This has
1712 almost the effect of a macro. The way to use this is to put a
1713 function definition in a header file with this attribute, and put
1714 another copy of the function, without @code{extern}, in a library
1715 file. The definition in the header file will cause most calls to the
1716 function to be inlined. If any uses of the function remain, they will
1717 refer to the single copy in the library. Note that the two
1718 definitions of the functions need not be precisely the same, although
1719 if they do not have the same effect your program may behave oddly.
1721 In C, if the function is neither @code{extern} nor @code{static}, then
1722 the function is compiled as a standalone function, as well as being
1723 inlined where possible.
1725 This is how GCC traditionally handled functions declared
1726 @code{inline}. Since ISO C99 specifies a different semantics for
1727 @code{inline}, this function attribute is provided as a transition
1728 measure and as a useful feature in its own right. This attribute is
1729 available in GCC 4.1.3 and later. It is available if either of the
1730 preprocessor macros @code{__GNUC_GNU_INLINE__} or
1731 @code{__GNUC_STDC_INLINE__} are defined. @xref{Inline,,An Inline
1732 Function is As Fast As a Macro}.
1734 In C++, this attribute does not depend on @code{extern} in any way,
1735 but it still requires the @code{inline} keyword to enable its special
1738 @cindex @code{flatten} function attribute
1740 Generally, inlining into a function is limited. For a function marked with
1741 this attribute, every call inside this function will be inlined, if possible.
1742 Whether the function itself is considered for inlining depends on its size and
1743 the current inlining parameters. The @code{flatten} attribute only works
1744 reliably in unit-at-a-time mode.
1747 @cindex functions that do pop the argument stack on the 386
1749 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1750 assume that the calling function will pop off the stack space used to
1751 pass arguments. This is
1752 useful to override the effects of the @option{-mrtd} switch.
1755 @cindex @code{const} function attribute
1756 Many functions do not examine any values except their arguments, and
1757 have no effects except the return value. Basically this is just slightly
1758 more strict class than the @code{pure} attribute below, since function is not
1759 allowed to read global memory.
1761 @cindex pointer arguments
1762 Note that a function that has pointer arguments and examines the data
1763 pointed to must @emph{not} be declared @code{const}. Likewise, a
1764 function that calls a non-@code{const} function usually must not be
1765 @code{const}. It does not make sense for a @code{const} function to
1768 The attribute @code{const} is not implemented in GCC versions earlier
1769 than 2.5. An alternative way to declare that a function has no side
1770 effects, which works in the current version and in some older versions,
1774 typedef int intfn ();
1776 extern const intfn square;
1779 This approach does not work in GNU C++ from 2.6.0 on, since the language
1780 specifies that the @samp{const} must be attached to the return value.
1784 @itemx constructor (@var{priority})
1785 @itemx destructor (@var{priority})
1786 @cindex @code{constructor} function attribute
1787 @cindex @code{destructor} function attribute
1788 The @code{constructor} attribute causes the function to be called
1789 automatically before execution enters @code{main ()}. Similarly, the
1790 @code{destructor} attribute causes the function to be called
1791 automatically after @code{main ()} has completed or @code{exit ()} has
1792 been called. Functions with these attributes are useful for
1793 initializing data that will be used implicitly during the execution of
1796 You may provide an optional integer priority to control the order in
1797 which constructor and destructor functions are run. A constructor
1798 with a smaller priority number runs before a constructor with a larger
1799 priority number; the opposite relationship holds for destructors. So,
1800 if you have a constructor that allocates a resource and a destructor
1801 that deallocates the same resource, both functions typically have the
1802 same priority. The priorities for constructor and destructor
1803 functions are the same as those specified for namespace-scope C++
1804 objects (@pxref{C++ Attributes}).
1806 These attributes are not currently implemented for Objective-C@.
1809 @cindex @code{deprecated} attribute.
1810 The @code{deprecated} attribute results in a warning if the function
1811 is used anywhere in the source file. This is useful when identifying
1812 functions that are expected to be removed in a future version of a
1813 program. The warning also includes the location of the declaration
1814 of the deprecated function, to enable users to easily find further
1815 information about why the function is deprecated, or what they should
1816 do instead. Note that the warnings only occurs for uses:
1819 int old_fn () __attribute__ ((deprecated));
1821 int (*fn_ptr)() = old_fn;
1824 results in a warning on line 3 but not line 2.
1826 The @code{deprecated} attribute can also be used for variables and
1827 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
1830 @cindex @code{__declspec(dllexport)}
1831 On Microsoft Windows targets and Symbian OS targets the
1832 @code{dllexport} attribute causes the compiler to provide a global
1833 pointer to a pointer in a DLL, so that it can be referenced with the
1834 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
1835 name is formed by combining @code{_imp__} and the function or variable
1838 You can use @code{__declspec(dllexport)} as a synonym for
1839 @code{__attribute__ ((dllexport))} for compatibility with other
1842 On systems that support the @code{visibility} attribute, this
1843 attribute also implies ``default'' visibility. It is an error to
1844 explicitly specify any other visibility.
1846 Currently, the @code{dllexport} attribute is ignored for inlined
1847 functions, unless the @option{-fkeep-inline-functions} flag has been
1848 used. The attribute is also ignored for undefined symbols.
1850 When applied to C++ classes, the attribute marks defined non-inlined
1851 member functions and static data members as exports. Static consts
1852 initialized in-class are not marked unless they are also defined
1855 For Microsoft Windows targets there are alternative methods for
1856 including the symbol in the DLL's export table such as using a
1857 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
1858 the @option{--export-all} linker flag.
1861 @cindex @code{__declspec(dllimport)}
1862 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
1863 attribute causes the compiler to reference a function or variable via
1864 a global pointer to a pointer that is set up by the DLL exporting the
1865 symbol. The attribute implies @code{extern}. On Microsoft Windows
1866 targets, the pointer name is formed by combining @code{_imp__} and the
1867 function or variable name.
1869 You can use @code{__declspec(dllimport)} as a synonym for
1870 @code{__attribute__ ((dllimport))} for compatibility with other
1873 On systems that support the @code{visibility} attribute, this
1874 attribute also implies ``default'' visibility. It is an error to
1875 explicitly specify any other visibility.
1877 Currently, the attribute is ignored for inlined functions. If the
1878 attribute is applied to a symbol @emph{definition}, an error is reported.
1879 If a symbol previously declared @code{dllimport} is later defined, the
1880 attribute is ignored in subsequent references, and a warning is emitted.
1881 The attribute is also overridden by a subsequent declaration as
1884 When applied to C++ classes, the attribute marks non-inlined
1885 member functions and static data members as imports. However, the
1886 attribute is ignored for virtual methods to allow creation of vtables
1889 On the SH Symbian OS target the @code{dllimport} attribute also has
1890 another affect---it can cause the vtable and run-time type information
1891 for a class to be exported. This happens when the class has a
1892 dllimport'ed constructor or a non-inline, non-pure virtual function
1893 and, for either of those two conditions, the class also has a inline
1894 constructor or destructor and has a key function that is defined in
1895 the current translation unit.
1897 For Microsoft Windows based targets the use of the @code{dllimport}
1898 attribute on functions is not necessary, but provides a small
1899 performance benefit by eliminating a thunk in the DLL@. The use of the
1900 @code{dllimport} attribute on imported variables was required on older
1901 versions of the GNU linker, but can now be avoided by passing the
1902 @option{--enable-auto-import} switch to the GNU linker. As with
1903 functions, using the attribute for a variable eliminates a thunk in
1906 One drawback to using this attribute is that a pointer to a function
1907 or variable marked as @code{dllimport} cannot be used as a constant
1908 address. On Microsoft Windows targets, the attribute can be disabled
1909 for functions by setting the @option{-mnop-fun-dllimport} flag.
1912 @cindex eight bit data on the H8/300, H8/300H, and H8S
1913 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1914 variable should be placed into the eight bit data section.
1915 The compiler will generate more efficient code for certain operations
1916 on data in the eight bit data area. Note the eight bit data area is limited to
1919 You must use GAS and GLD from GNU binutils version 2.7 or later for
1920 this attribute to work correctly.
1922 @item exception_handler
1923 @cindex exception handler functions on the Blackfin processor
1924 Use this attribute on the Blackfin to indicate that the specified function
1925 is an exception handler. The compiler will generate function entry and
1926 exit sequences suitable for use in an exception handler when this
1927 attribute is present.
1930 @cindex functions which handle memory bank switching
1931 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
1932 use a calling convention that takes care of switching memory banks when
1933 entering and leaving a function. This calling convention is also the
1934 default when using the @option{-mlong-calls} option.
1936 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
1937 to call and return from a function.
1939 On 68HC11 the compiler will generate a sequence of instructions
1940 to invoke a board-specific routine to switch the memory bank and call the
1941 real function. The board-specific routine simulates a @code{call}.
1942 At the end of a function, it will jump to a board-specific routine
1943 instead of using @code{rts}. The board-specific return routine simulates
1947 @cindex functions that pop the argument stack on the 386
1948 On the Intel 386, the @code{fastcall} attribute causes the compiler to
1949 pass the first argument (if of integral type) in the register ECX and
1950 the second argument (if of integral type) in the register EDX@. Subsequent
1951 and other typed arguments are passed on the stack. The called function will
1952 pop the arguments off the stack. If the number of arguments is variable all
1953 arguments are pushed on the stack.
1955 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
1956 @cindex @code{format} function attribute
1958 The @code{format} attribute specifies that a function takes @code{printf},
1959 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
1960 should be type-checked against a format string. For example, the
1965 my_printf (void *my_object, const char *my_format, ...)
1966 __attribute__ ((format (printf, 2, 3)));
1970 causes the compiler to check the arguments in calls to @code{my_printf}
1971 for consistency with the @code{printf} style format string argument
1974 The parameter @var{archetype} determines how the format string is
1975 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
1976 or @code{strfmon}. (You can also use @code{__printf__},
1977 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
1978 parameter @var{string-index} specifies which argument is the format
1979 string argument (starting from 1), while @var{first-to-check} is the
1980 number of the first argument to check against the format string. For
1981 functions where the arguments are not available to be checked (such as
1982 @code{vprintf}), specify the third parameter as zero. In this case the
1983 compiler only checks the format string for consistency. For
1984 @code{strftime} formats, the third parameter is required to be zero.
1985 Since non-static C++ methods have an implicit @code{this} argument, the
1986 arguments of such methods should be counted from two, not one, when
1987 giving values for @var{string-index} and @var{first-to-check}.
1989 In the example above, the format string (@code{my_format}) is the second
1990 argument of the function @code{my_print}, and the arguments to check
1991 start with the third argument, so the correct parameters for the format
1992 attribute are 2 and 3.
1994 @opindex ffreestanding
1995 @opindex fno-builtin
1996 The @code{format} attribute allows you to identify your own functions
1997 which take format strings as arguments, so that GCC can check the
1998 calls to these functions for errors. The compiler always (unless
1999 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
2000 for the standard library functions @code{printf}, @code{fprintf},
2001 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
2002 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
2003 warnings are requested (using @option{-Wformat}), so there is no need to
2004 modify the header file @file{stdio.h}. In C99 mode, the functions
2005 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
2006 @code{vsscanf} are also checked. Except in strictly conforming C
2007 standard modes, the X/Open function @code{strfmon} is also checked as
2008 are @code{printf_unlocked} and @code{fprintf_unlocked}.
2009 @xref{C Dialect Options,,Options Controlling C Dialect}.
2011 The target may provide additional types of format checks.
2012 @xref{Target Format Checks,,Format Checks Specific to Particular
2015 @item format_arg (@var{string-index})
2016 @cindex @code{format_arg} function attribute
2017 @opindex Wformat-nonliteral
2018 The @code{format_arg} attribute specifies that a function takes a format
2019 string for a @code{printf}, @code{scanf}, @code{strftime} or
2020 @code{strfmon} style function and modifies it (for example, to translate
2021 it into another language), so the result can be passed to a
2022 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
2023 function (with the remaining arguments to the format function the same
2024 as they would have been for the unmodified string). For example, the
2029 my_dgettext (char *my_domain, const char *my_format)
2030 __attribute__ ((format_arg (2)));
2034 causes the compiler to check the arguments in calls to a @code{printf},
2035 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
2036 format string argument is a call to the @code{my_dgettext} function, for
2037 consistency with the format string argument @code{my_format}. If the
2038 @code{format_arg} attribute had not been specified, all the compiler
2039 could tell in such calls to format functions would be that the format
2040 string argument is not constant; this would generate a warning when
2041 @option{-Wformat-nonliteral} is used, but the calls could not be checked
2042 without the attribute.
2044 The parameter @var{string-index} specifies which argument is the format
2045 string argument (starting from one). Since non-static C++ methods have
2046 an implicit @code{this} argument, the arguments of such methods should
2047 be counted from two.
2049 The @code{format-arg} attribute allows you to identify your own
2050 functions which modify format strings, so that GCC can check the
2051 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
2052 type function whose operands are a call to one of your own function.
2053 The compiler always treats @code{gettext}, @code{dgettext}, and
2054 @code{dcgettext} in this manner except when strict ISO C support is
2055 requested by @option{-ansi} or an appropriate @option{-std} option, or
2056 @option{-ffreestanding} or @option{-fno-builtin}
2057 is used. @xref{C Dialect Options,,Options
2058 Controlling C Dialect}.
2060 @item function_vector
2061 @cindex calling functions through the function vector on H8/300, M16C, and M32C processors
2062 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
2063 function should be called through the function vector. Calling a
2064 function through the function vector will reduce code size, however;
2065 the function vector has a limited size (maximum 128 entries on the H8/300
2066 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
2068 You must use GAS and GLD from GNU binutils version 2.7 or later for
2069 this attribute to work correctly.
2071 On M16C/M32C targets, the @code{function_vector} attribute declares a
2072 special page subroutine call function. Use of this attribute reduces
2073 the code size by 2 bytes for each call generated to the
2074 subroutine. The argument to the attribute is the vector number entry
2075 from the special page vector table which contains the 16 low-order
2076 bits of the subroutine's entry address. Each vector table has special
2077 page number (18 to 255) which are used in @code{jsrs} instruction.
2078 Jump addresses of the routines are generated by adding 0x0F0000 (in
2079 case of M16C targets) or 0xFF0000 (in case of M32C targets), to the 2
2080 byte addresses set in the vector table. Therefore you need to ensure
2081 that all the special page vector routines should get mapped within the
2082 address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF
2085 In the following example 2 bytes will be saved for each call to
2086 function @code{foo}.
2089 void foo (void) __attribute__((function_vector(0x18)));
2100 If functions are defined in one file and are called in another file,
2101 then be sure to write this declaration in both files.
2103 This attribute is ignored for R8C target.
2106 @cindex interrupt handler functions
2107 Use this attribute on the ARM, AVR, C4x, CRX, M32C, M32R/D, m68k, MS1,
2108 and Xstormy16 ports to indicate that the specified function is an
2109 interrupt handler. The compiler will generate function entry and exit
2110 sequences suitable for use in an interrupt handler when this attribute
2113 Note, interrupt handlers for the Blackfin, H8/300, H8/300H, H8S, and
2114 SH processors can be specified via the @code{interrupt_handler} attribute.
2116 Note, on the AVR, interrupts will be enabled inside the function.
2118 Note, for the ARM, you can specify the kind of interrupt to be handled by
2119 adding an optional parameter to the interrupt attribute like this:
2122 void f () __attribute__ ((interrupt ("IRQ")));
2125 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
2127 On ARMv7-M the interrupt type is ignored, and the attribute means the function
2128 may be called with a word aligned stack pointer.
2130 @item interrupt_handler
2131 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
2132 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
2133 indicate that the specified function is an interrupt handler. The compiler
2134 will generate function entry and exit sequences suitable for use in an
2135 interrupt handler when this attribute is present.
2137 @item interrupt_thread
2138 @cindex interrupt thread functions on fido
2139 Use this attribute on fido, a subarchitecture of the m68k, to indicate
2140 that the specified function is an interrupt handler that is designed
2141 to run as a thread. The compiler omits generate prologue/epilogue
2142 sequences and replaces the return instruction with a @code{sleep}
2143 instruction. This attribute is available only on fido.
2146 @cindex User stack pointer in interrupts on the Blackfin
2147 When used together with @code{interrupt_handler}, @code{exception_handler}
2148 or @code{nmi_handler}, code will be generated to load the stack pointer
2149 from the USP register in the function prologue.
2152 @cindex @code{l1_text} function attribute
2153 This attribute specifies a function to be placed into L1 Instruction
2154 SRAM. The function will be put into a specific section named @code{.l1.text}.
2155 With @option{-mfdpic}, function calls with a such function as the callee
2156 or caller will use inlined PLT.
2158 @item long_call/short_call
2159 @cindex indirect calls on ARM
2160 This attribute specifies how a particular function is called on
2161 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
2162 command line switch and @code{#pragma long_calls} settings. The
2163 @code{long_call} attribute indicates that the function might be far
2164 away from the call site and require a different (more expensive)
2165 calling sequence. The @code{short_call} attribute always places
2166 the offset to the function from the call site into the @samp{BL}
2167 instruction directly.
2169 @item longcall/shortcall
2170 @cindex functions called via pointer on the RS/6000 and PowerPC
2171 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
2172 indicates that the function might be far away from the call site and
2173 require a different (more expensive) calling sequence. The
2174 @code{shortcall} attribute indicates that the function is always close
2175 enough for the shorter calling sequence to be used. These attributes
2176 override both the @option{-mlongcall} switch and, on the RS/6000 and
2177 PowerPC, the @code{#pragma longcall} setting.
2179 @xref{RS/6000 and PowerPC Options}, for more information on whether long
2180 calls are necessary.
2182 @item long_call/near/far
2183 @cindex indirect calls on MIPS
2184 These attributes specify how a particular function is called on MIPS@.
2185 The attributes override the @option{-mlong-calls} (@pxref{MIPS Options})
2186 command-line switch. The @code{long_call} and @code{far} attributes are
2187 synonyms, and cause the compiler to always call
2188 the function by first loading its address into a register, and then using
2189 the contents of that register. The @code{near} attribute has the opposite
2190 effect; it specifies that non-PIC calls should be made using the more
2191 efficient @code{jal} instruction.
2194 @cindex @code{malloc} attribute
2195 The @code{malloc} attribute is used to tell the compiler that a function
2196 may be treated as if any non-@code{NULL} pointer it returns cannot
2197 alias any other pointer valid when the function returns.
2198 This will often improve optimization.
2199 Standard functions with this property include @code{malloc} and
2200 @code{calloc}. @code{realloc}-like functions have this property as
2201 long as the old pointer is never referred to (including comparing it
2202 to the new pointer) after the function returns a non-@code{NULL}
2205 @item model (@var{model-name})
2206 @cindex function addressability on the M32R/D
2207 @cindex variable addressability on the IA-64
2209 On the M32R/D, use this attribute to set the addressability of an
2210 object, and of the code generated for a function. The identifier
2211 @var{model-name} is one of @code{small}, @code{medium}, or
2212 @code{large}, representing each of the code models.
2214 Small model objects live in the lower 16MB of memory (so that their
2215 addresses can be loaded with the @code{ld24} instruction), and are
2216 callable with the @code{bl} instruction.
2218 Medium model objects may live anywhere in the 32-bit address space (the
2219 compiler will generate @code{seth/add3} instructions to load their addresses),
2220 and are callable with the @code{bl} instruction.
2222 Large model objects may live anywhere in the 32-bit address space (the
2223 compiler will generate @code{seth/add3} instructions to load their addresses),
2224 and may not be reachable with the @code{bl} instruction (the compiler will
2225 generate the much slower @code{seth/add3/jl} instruction sequence).
2227 On IA-64, use this attribute to set the addressability of an object.
2228 At present, the only supported identifier for @var{model-name} is
2229 @code{small}, indicating addressability via ``small'' (22-bit)
2230 addresses (so that their addresses can be loaded with the @code{addl}
2231 instruction). Caveat: such addressing is by definition not position
2232 independent and hence this attribute must not be used for objects
2233 defined by shared libraries.
2236 @cindex function without a prologue/epilogue code
2237 Use this attribute on the ARM, AVR, C4x, IP2K and SPU ports to indicate that
2238 the specified function does not need prologue/epilogue sequences generated by
2239 the compiler. It is up to the programmer to provide these sequences.
2242 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2243 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2244 use the normal calling convention based on @code{jsr} and @code{rts}.
2245 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2249 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2250 Use this attribute together with @code{interrupt_handler},
2251 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2252 entry code should enable nested interrupts or exceptions.
2255 @cindex NMI handler functions on the Blackfin processor
2256 Use this attribute on the Blackfin to indicate that the specified function
2257 is an NMI handler. The compiler will generate function entry and
2258 exit sequences suitable for use in an NMI handler when this
2259 attribute is present.
2261 @item no_instrument_function
2262 @cindex @code{no_instrument_function} function attribute
2263 @opindex finstrument-functions
2264 If @option{-finstrument-functions} is given, profiling function calls will
2265 be generated at entry and exit of most user-compiled functions.
2266 Functions with this attribute will not be so instrumented.
2269 @cindex @code{noinline} function attribute
2270 This function attribute prevents a function from being considered for
2273 @item nonnull (@var{arg-index}, @dots{})
2274 @cindex @code{nonnull} function attribute
2275 The @code{nonnull} attribute specifies that some function parameters should
2276 be non-null pointers. For instance, the declaration:
2280 my_memcpy (void *dest, const void *src, size_t len)
2281 __attribute__((nonnull (1, 2)));
2285 causes the compiler to check that, in calls to @code{my_memcpy},
2286 arguments @var{dest} and @var{src} are non-null. If the compiler
2287 determines that a null pointer is passed in an argument slot marked
2288 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2289 is issued. The compiler may also choose to make optimizations based
2290 on the knowledge that certain function arguments will not be null.
2292 If no argument index list is given to the @code{nonnull} attribute,
2293 all pointer arguments are marked as non-null. To illustrate, the
2294 following declaration is equivalent to the previous example:
2298 my_memcpy (void *dest, const void *src, size_t len)
2299 __attribute__((nonnull));
2303 @cindex @code{noreturn} function attribute
2304 A few standard library functions, such as @code{abort} and @code{exit},
2305 cannot return. GCC knows this automatically. Some programs define
2306 their own functions that never return. You can declare them
2307 @code{noreturn} to tell the compiler this fact. For example,
2311 void fatal () __attribute__ ((noreturn));
2314 fatal (/* @r{@dots{}} */)
2316 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2322 The @code{noreturn} keyword tells the compiler to assume that
2323 @code{fatal} cannot return. It can then optimize without regard to what
2324 would happen if @code{fatal} ever did return. This makes slightly
2325 better code. More importantly, it helps avoid spurious warnings of
2326 uninitialized variables.
2328 The @code{noreturn} keyword does not affect the exceptional path when that
2329 applies: a @code{noreturn}-marked function may still return to the caller
2330 by throwing an exception or calling @code{longjmp}.
2332 Do not assume that registers saved by the calling function are
2333 restored before calling the @code{noreturn} function.
2335 It does not make sense for a @code{noreturn} function to have a return
2336 type other than @code{void}.
2338 The attribute @code{noreturn} is not implemented in GCC versions
2339 earlier than 2.5. An alternative way to declare that a function does
2340 not return, which works in the current version and in some older
2341 versions, is as follows:
2344 typedef void voidfn ();
2346 volatile voidfn fatal;
2349 This approach does not work in GNU C++.
2352 @cindex @code{nothrow} function attribute
2353 The @code{nothrow} attribute is used to inform the compiler that a
2354 function cannot throw an exception. For example, most functions in
2355 the standard C library can be guaranteed not to throw an exception
2356 with the notable exceptions of @code{qsort} and @code{bsearch} that
2357 take function pointer arguments. The @code{nothrow} attribute is not
2358 implemented in GCC versions earlier than 3.3.
2361 @cindex @code{pure} function attribute
2362 Many functions have no effects except the return value and their
2363 return value depends only on the parameters and/or global variables.
2364 Such a function can be subject
2365 to common subexpression elimination and loop optimization just as an
2366 arithmetic operator would be. These functions should be declared
2367 with the attribute @code{pure}. For example,
2370 int square (int) __attribute__ ((pure));
2374 says that the hypothetical function @code{square} is safe to call
2375 fewer times than the program says.
2377 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2378 Interesting non-pure functions are functions with infinite loops or those
2379 depending on volatile memory or other system resource, that may change between
2380 two consecutive calls (such as @code{feof} in a multithreading environment).
2382 The attribute @code{pure} is not implemented in GCC versions earlier
2386 @cindex @code{hot} function attribute
2387 The @code{hot} attribute is used to inform the compiler that a function is a
2388 hot spot of the compiled program. The function is optimized more aggressively
2389 and on many target it is placed into special subsection of the text section so
2390 all hot functions appears close together improving locality.
2392 When profile feedback is available, via @option{-fprofile-use}, hot functions
2393 are automatically detected and this attribute is ignored.
2395 The @code{hot} attribute is not implemented in GCC versions earlier than 4.3.
2398 @cindex @code{cold} function attribute
2399 The @code{cold} attribute is used to inform the compiler that a function is
2400 unlikely executed. The function is optimized for size rather than speed and on
2401 many targets it is placed into special subsection of the text section so all
2402 cold functions appears close together improving code locality of non-cold parts
2403 of program. The paths leading to call of cold functions within code are marked
2404 as unlikely by the branch prediction mechanism. It is thus useful to mark
2405 functions used to handle unlikely conditions, such as @code{perror}, as cold to
2406 improve optimization of hot functions that do call marked functions in rare
2409 When profile feedback is available, via @option{-fprofile-use}, hot functions
2410 are automatically detected and this attribute is ignored.
2412 The @code{hot} attribute is not implemented in GCC versions earlier than 4.3.
2414 @item regparm (@var{number})
2415 @cindex @code{regparm} attribute
2416 @cindex functions that are passed arguments in registers on the 386
2417 On the Intel 386, the @code{regparm} attribute causes the compiler to
2418 pass arguments number one to @var{number} if they are of integral type
2419 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2420 take a variable number of arguments will continue to be passed all of their
2421 arguments on the stack.
2423 Beware that on some ELF systems this attribute is unsuitable for
2424 global functions in shared libraries with lazy binding (which is the
2425 default). Lazy binding will send the first call via resolving code in
2426 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2427 per the standard calling conventions. Solaris 8 is affected by this.
2428 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2429 safe since the loaders there save all registers. (Lazy binding can be
2430 disabled with the linker or the loader if desired, to avoid the
2434 @cindex @code{sseregparm} attribute
2435 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2436 causes the compiler to pass up to 3 floating point arguments in
2437 SSE registers instead of on the stack. Functions that take a
2438 variable number of arguments will continue to pass all of their
2439 floating point arguments on the stack.
2441 @item force_align_arg_pointer
2442 @cindex @code{force_align_arg_pointer} attribute
2443 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
2444 applied to individual function definitions, generating an alternate
2445 prologue and epilogue that realigns the runtime stack. This supports
2446 mixing legacy codes that run with a 4-byte aligned stack with modern
2447 codes that keep a 16-byte stack for SSE compatibility. The alternate
2448 prologue and epilogue are slower and bigger than the regular ones, and
2449 the alternate prologue requires a scratch register; this lowers the
2450 number of registers available if used in conjunction with the
2451 @code{regparm} attribute. The @code{force_align_arg_pointer}
2452 attribute is incompatible with nested functions; this is considered a
2456 @cindex @code{returns_twice} attribute
2457 The @code{returns_twice} attribute tells the compiler that a function may
2458 return more than one time. The compiler will ensure that all registers
2459 are dead before calling such a function and will emit a warning about
2460 the variables that may be clobbered after the second return from the
2461 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2462 The @code{longjmp}-like counterpart of such function, if any, might need
2463 to be marked with the @code{noreturn} attribute.
2466 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2467 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2468 all registers except the stack pointer should be saved in the prologue
2469 regardless of whether they are used or not.
2471 @item section ("@var{section-name}")
2472 @cindex @code{section} function attribute
2473 Normally, the compiler places the code it generates in the @code{text} section.
2474 Sometimes, however, you need additional sections, or you need certain
2475 particular functions to appear in special sections. The @code{section}
2476 attribute specifies that a function lives in a particular section.
2477 For example, the declaration:
2480 extern void foobar (void) __attribute__ ((section ("bar")));
2484 puts the function @code{foobar} in the @code{bar} section.
2486 Some file formats do not support arbitrary sections so the @code{section}
2487 attribute is not available on all platforms.
2488 If you need to map the entire contents of a module to a particular
2489 section, consider using the facilities of the linker instead.
2492 @cindex @code{sentinel} function attribute
2493 This function attribute ensures that a parameter in a function call is
2494 an explicit @code{NULL}. The attribute is only valid on variadic
2495 functions. By default, the sentinel is located at position zero, the
2496 last parameter of the function call. If an optional integer position
2497 argument P is supplied to the attribute, the sentinel must be located at
2498 position P counting backwards from the end of the argument list.
2501 __attribute__ ((sentinel))
2503 __attribute__ ((sentinel(0)))
2506 The attribute is automatically set with a position of 0 for the built-in
2507 functions @code{execl} and @code{execlp}. The built-in function
2508 @code{execle} has the attribute set with a position of 1.
2510 A valid @code{NULL} in this context is defined as zero with any pointer
2511 type. If your system defines the @code{NULL} macro with an integer type
2512 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2513 with a copy that redefines NULL appropriately.
2515 The warnings for missing or incorrect sentinels are enabled with
2519 See long_call/short_call.
2522 See longcall/shortcall.
2525 @cindex signal handler functions on the AVR processors
2526 Use this attribute on the AVR to indicate that the specified
2527 function is a signal handler. The compiler will generate function
2528 entry and exit sequences suitable for use in a signal handler when this
2529 attribute is present. Interrupts will be disabled inside the function.
2532 Use this attribute on the SH to indicate an @code{interrupt_handler}
2533 function should switch to an alternate stack. It expects a string
2534 argument that names a global variable holding the address of the
2539 void f () __attribute__ ((interrupt_handler,
2540 sp_switch ("alt_stack")));
2544 @cindex functions that pop the argument stack on the 386
2545 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2546 assume that the called function will pop off the stack space used to
2547 pass arguments, unless it takes a variable number of arguments.
2550 @cindex tiny data section on the H8/300H and H8S
2551 Use this attribute on the H8/300H and H8S to indicate that the specified
2552 variable should be placed into the tiny data section.
2553 The compiler will generate more efficient code for loads and stores
2554 on data in the tiny data section. Note the tiny data area is limited to
2555 slightly under 32kbytes of data.
2558 Use this attribute on the SH for an @code{interrupt_handler} to return using
2559 @code{trapa} instead of @code{rte}. This attribute expects an integer
2560 argument specifying the trap number to be used.
2563 @cindex @code{unused} attribute.
2564 This attribute, attached to a function, means that the function is meant
2565 to be possibly unused. GCC will not produce a warning for this
2569 @cindex @code{used} attribute.
2570 This attribute, attached to a function, means that code must be emitted
2571 for the function even if it appears that the function is not referenced.
2572 This is useful, for example, when the function is referenced only in
2576 @cindex @code{version_id} attribute on IA64 HP-UX
2577 This attribute, attached to a global variable or function, renames a
2578 symbol to contain a version string, thus allowing for function level
2579 versioning. HP-UX system header files may use version level functioning
2580 for some system calls.
2583 extern int foo () __attribute__((version_id ("20040821")));
2586 Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}.
2588 @item visibility ("@var{visibility_type}")
2589 @cindex @code{visibility} attribute
2590 This attribute affects the linkage of the declaration to which it is attached.
2591 There are four supported @var{visibility_type} values: default,
2592 hidden, protected or internal visibility.
2595 void __attribute__ ((visibility ("protected")))
2596 f () @{ /* @r{Do something.} */; @}
2597 int i __attribute__ ((visibility ("hidden")));
2600 The possible values of @var{visibility_type} correspond to the
2601 visibility settings in the ELF gABI.
2604 @c keep this list of visibilities in alphabetical order.
2607 Default visibility is the normal case for the object file format.
2608 This value is available for the visibility attribute to override other
2609 options that may change the assumed visibility of entities.
2611 On ELF, default visibility means that the declaration is visible to other
2612 modules and, in shared libraries, means that the declared entity may be
2615 On Darwin, default visibility means that the declaration is visible to
2618 Default visibility corresponds to ``external linkage'' in the language.
2621 Hidden visibility indicates that the entity declared will have a new
2622 form of linkage, which we'll call ``hidden linkage''. Two
2623 declarations of an object with hidden linkage refer to the same object
2624 if they are in the same shared object.
2627 Internal visibility is like hidden visibility, but with additional
2628 processor specific semantics. Unless otherwise specified by the
2629 psABI, GCC defines internal visibility to mean that a function is
2630 @emph{never} called from another module. Compare this with hidden
2631 functions which, while they cannot be referenced directly by other
2632 modules, can be referenced indirectly via function pointers. By
2633 indicating that a function cannot be called from outside the module,
2634 GCC may for instance omit the load of a PIC register since it is known
2635 that the calling function loaded the correct value.
2638 Protected visibility is like default visibility except that it
2639 indicates that references within the defining module will bind to the
2640 definition in that module. That is, the declared entity cannot be
2641 overridden by another module.
2645 All visibilities are supported on many, but not all, ELF targets
2646 (supported when the assembler supports the @samp{.visibility}
2647 pseudo-op). Default visibility is supported everywhere. Hidden
2648 visibility is supported on Darwin targets.
2650 The visibility attribute should be applied only to declarations which
2651 would otherwise have external linkage. The attribute should be applied
2652 consistently, so that the same entity should not be declared with
2653 different settings of the attribute.
2655 In C++, the visibility attribute applies to types as well as functions
2656 and objects, because in C++ types have linkage. A class must not have
2657 greater visibility than its non-static data member types and bases,
2658 and class members default to the visibility of their class. Also, a
2659 declaration without explicit visibility is limited to the visibility
2662 In C++, you can mark member functions and static member variables of a
2663 class with the visibility attribute. This is useful if if you know a
2664 particular method or static member variable should only be used from
2665 one shared object; then you can mark it hidden while the rest of the
2666 class has default visibility. Care must be taken to avoid breaking
2667 the One Definition Rule; for example, it is usually not useful to mark
2668 an inline method as hidden without marking the whole class as hidden.
2670 A C++ namespace declaration can also have the visibility attribute.
2671 This attribute applies only to the particular namespace body, not to
2672 other definitions of the same namespace; it is equivalent to using
2673 @samp{#pragma GCC visibility} before and after the namespace
2674 definition (@pxref{Visibility Pragmas}).
2676 In C++, if a template argument has limited visibility, this
2677 restriction is implicitly propagated to the template instantiation.
2678 Otherwise, template instantiations and specializations default to the
2679 visibility of their template.
2681 If both the template and enclosing class have explicit visibility, the
2682 visibility from the template is used.
2684 @item warn_unused_result
2685 @cindex @code{warn_unused_result} attribute
2686 The @code{warn_unused_result} attribute causes a warning to be emitted
2687 if a caller of the function with this attribute does not use its
2688 return value. This is useful for functions where not checking
2689 the result is either a security problem or always a bug, such as
2693 int fn () __attribute__ ((warn_unused_result));
2696 if (fn () < 0) return -1;
2702 results in warning on line 5.
2705 @cindex @code{weak} attribute
2706 The @code{weak} attribute causes the declaration to be emitted as a weak
2707 symbol rather than a global. This is primarily useful in defining
2708 library functions which can be overridden in user code, though it can
2709 also be used with non-function declarations. Weak symbols are supported
2710 for ELF targets, and also for a.out targets when using the GNU assembler
2714 @itemx weakref ("@var{target}")
2715 @cindex @code{weakref} attribute
2716 The @code{weakref} attribute marks a declaration as a weak reference.
2717 Without arguments, it should be accompanied by an @code{alias} attribute
2718 naming the target symbol. Optionally, the @var{target} may be given as
2719 an argument to @code{weakref} itself. In either case, @code{weakref}
2720 implicitly marks the declaration as @code{weak}. Without a
2721 @var{target}, given as an argument to @code{weakref} or to @code{alias},
2722 @code{weakref} is equivalent to @code{weak}.
2725 static int x() __attribute__ ((weakref ("y")));
2726 /* is equivalent to... */
2727 static int x() __attribute__ ((weak, weakref, alias ("y")));
2729 static int x() __attribute__ ((weakref));
2730 static int x() __attribute__ ((alias ("y")));
2733 A weak reference is an alias that does not by itself require a
2734 definition to be given for the target symbol. If the target symbol is
2735 only referenced through weak references, then the becomes a @code{weak}
2736 undefined symbol. If it is directly referenced, however, then such
2737 strong references prevail, and a definition will be required for the
2738 symbol, not necessarily in the same translation unit.
2740 The effect is equivalent to moving all references to the alias to a
2741 separate translation unit, renaming the alias to the aliased symbol,
2742 declaring it as weak, compiling the two separate translation units and
2743 performing a reloadable link on them.
2745 At present, a declaration to which @code{weakref} is attached can
2746 only be @code{static}.
2748 @item externally_visible
2749 @cindex @code{externally_visible} attribute.
2750 This attribute, attached to a global variable or function nullify
2751 effect of @option{-fwhole-program} command line option, so the object
2752 remain visible outside the current compilation unit
2756 You can specify multiple attributes in a declaration by separating them
2757 by commas within the double parentheses or by immediately following an
2758 attribute declaration with another attribute declaration.
2760 @cindex @code{#pragma}, reason for not using
2761 @cindex pragma, reason for not using
2762 Some people object to the @code{__attribute__} feature, suggesting that
2763 ISO C's @code{#pragma} should be used instead. At the time
2764 @code{__attribute__} was designed, there were two reasons for not doing
2769 It is impossible to generate @code{#pragma} commands from a macro.
2772 There is no telling what the same @code{#pragma} might mean in another
2776 These two reasons applied to almost any application that might have been
2777 proposed for @code{#pragma}. It was basically a mistake to use
2778 @code{#pragma} for @emph{anything}.
2780 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2781 to be generated from macros. In addition, a @code{#pragma GCC}
2782 namespace is now in use for GCC-specific pragmas. However, it has been
2783 found convenient to use @code{__attribute__} to achieve a natural
2784 attachment of attributes to their corresponding declarations, whereas
2785 @code{#pragma GCC} is of use for constructs that do not naturally form
2786 part of the grammar. @xref{Other Directives,,Miscellaneous
2787 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2789 @node Attribute Syntax
2790 @section Attribute Syntax
2791 @cindex attribute syntax
2793 This section describes the syntax with which @code{__attribute__} may be
2794 used, and the constructs to which attribute specifiers bind, for the C
2795 language. Some details may vary for C++ and Objective-C@. Because of
2796 infelicities in the grammar for attributes, some forms described here
2797 may not be successfully parsed in all cases.
2799 There are some problems with the semantics of attributes in C++. For
2800 example, there are no manglings for attributes, although they may affect
2801 code generation, so problems may arise when attributed types are used in
2802 conjunction with templates or overloading. Similarly, @code{typeid}
2803 does not distinguish between types with different attributes. Support
2804 for attributes in C++ may be restricted in future to attributes on
2805 declarations only, but not on nested declarators.
2807 @xref{Function Attributes}, for details of the semantics of attributes
2808 applying to functions. @xref{Variable Attributes}, for details of the
2809 semantics of attributes applying to variables. @xref{Type Attributes},
2810 for details of the semantics of attributes applying to structure, union
2811 and enumerated types.
2813 An @dfn{attribute specifier} is of the form
2814 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2815 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2816 each attribute is one of the following:
2820 Empty. Empty attributes are ignored.
2823 A word (which may be an identifier such as @code{unused}, or a reserved
2824 word such as @code{const}).
2827 A word, followed by, in parentheses, parameters for the attribute.
2828 These parameters take one of the following forms:
2832 An identifier. For example, @code{mode} attributes use this form.
2835 An identifier followed by a comma and a non-empty comma-separated list
2836 of expressions. For example, @code{format} attributes use this form.
2839 A possibly empty comma-separated list of expressions. For example,
2840 @code{format_arg} attributes use this form with the list being a single
2841 integer constant expression, and @code{alias} attributes use this form
2842 with the list being a single string constant.
2846 An @dfn{attribute specifier list} is a sequence of one or more attribute
2847 specifiers, not separated by any other tokens.
2849 In GNU C, an attribute specifier list may appear after the colon following a
2850 label, other than a @code{case} or @code{default} label. The only
2851 attribute it makes sense to use after a label is @code{unused}. This
2852 feature is intended for code generated by programs which contains labels
2853 that may be unused but which is compiled with @option{-Wall}. It would
2854 not normally be appropriate to use in it human-written code, though it
2855 could be useful in cases where the code that jumps to the label is
2856 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2857 such placement of attribute lists, as it is permissible for a
2858 declaration, which could begin with an attribute list, to be labelled in
2859 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2860 does not arise there.
2862 An attribute specifier list may appear as part of a @code{struct},
2863 @code{union} or @code{enum} specifier. It may go either immediately
2864 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2865 the closing brace. The former syntax is preferred.
2866 Where attribute specifiers follow the closing brace, they are considered
2867 to relate to the structure, union or enumerated type defined, not to any
2868 enclosing declaration the type specifier appears in, and the type
2869 defined is not complete until after the attribute specifiers.
2870 @c Otherwise, there would be the following problems: a shift/reduce
2871 @c conflict between attributes binding the struct/union/enum and
2872 @c binding to the list of specifiers/qualifiers; and "aligned"
2873 @c attributes could use sizeof for the structure, but the size could be
2874 @c changed later by "packed" attributes.
2876 Otherwise, an attribute specifier appears as part of a declaration,
2877 counting declarations of unnamed parameters and type names, and relates
2878 to that declaration (which may be nested in another declaration, for
2879 example in the case of a parameter declaration), or to a particular declarator
2880 within a declaration. Where an
2881 attribute specifier is applied to a parameter declared as a function or
2882 an array, it should apply to the function or array rather than the
2883 pointer to which the parameter is implicitly converted, but this is not
2884 yet correctly implemented.
2886 Any list of specifiers and qualifiers at the start of a declaration may
2887 contain attribute specifiers, whether or not such a list may in that
2888 context contain storage class specifiers. (Some attributes, however,
2889 are essentially in the nature of storage class specifiers, and only make
2890 sense where storage class specifiers may be used; for example,
2891 @code{section}.) There is one necessary limitation to this syntax: the
2892 first old-style parameter declaration in a function definition cannot
2893 begin with an attribute specifier, because such an attribute applies to
2894 the function instead by syntax described below (which, however, is not
2895 yet implemented in this case). In some other cases, attribute
2896 specifiers are permitted by this grammar but not yet supported by the
2897 compiler. All attribute specifiers in this place relate to the
2898 declaration as a whole. In the obsolescent usage where a type of
2899 @code{int} is implied by the absence of type specifiers, such a list of
2900 specifiers and qualifiers may be an attribute specifier list with no
2901 other specifiers or qualifiers.
2903 At present, the first parameter in a function prototype must have some
2904 type specifier which is not an attribute specifier; this resolves an
2905 ambiguity in the interpretation of @code{void f(int
2906 (__attribute__((foo)) x))}, but is subject to change. At present, if
2907 the parentheses of a function declarator contain only attributes then
2908 those attributes are ignored, rather than yielding an error or warning
2909 or implying a single parameter of type int, but this is subject to
2912 An attribute specifier list may appear immediately before a declarator
2913 (other than the first) in a comma-separated list of declarators in a
2914 declaration of more than one identifier using a single list of
2915 specifiers and qualifiers. Such attribute specifiers apply
2916 only to the identifier before whose declarator they appear. For
2920 __attribute__((noreturn)) void d0 (void),
2921 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2926 the @code{noreturn} attribute applies to all the functions
2927 declared; the @code{format} attribute only applies to @code{d1}.
2929 An attribute specifier list may appear immediately before the comma,
2930 @code{=} or semicolon terminating the declaration of an identifier other
2931 than a function definition. At present, such attribute specifiers apply
2932 to the declared object or function, but in future they may attach to the
2933 outermost adjacent declarator. In simple cases there is no difference,
2934 but, for example, in
2937 void (****f)(void) __attribute__((noreturn));
2941 at present the @code{noreturn} attribute applies to @code{f}, which
2942 causes a warning since @code{f} is not a function, but in future it may
2943 apply to the function @code{****f}. The precise semantics of what
2944 attributes in such cases will apply to are not yet specified. Where an
2945 assembler name for an object or function is specified (@pxref{Asm
2946 Labels}), at present the attribute must follow the @code{asm}
2947 specification; in future, attributes before the @code{asm} specification
2948 may apply to the adjacent declarator, and those after it to the declared
2951 An attribute specifier list may, in future, be permitted to appear after
2952 the declarator in a function definition (before any old-style parameter
2953 declarations or the function body).
2955 Attribute specifiers may be mixed with type qualifiers appearing inside
2956 the @code{[]} of a parameter array declarator, in the C99 construct by
2957 which such qualifiers are applied to the pointer to which the array is
2958 implicitly converted. Such attribute specifiers apply to the pointer,
2959 not to the array, but at present this is not implemented and they are
2962 An attribute specifier list may appear at the start of a nested
2963 declarator. At present, there are some limitations in this usage: the
2964 attributes correctly apply to the declarator, but for most individual
2965 attributes the semantics this implies are not implemented.
2966 When attribute specifiers follow the @code{*} of a pointer
2967 declarator, they may be mixed with any type qualifiers present.
2968 The following describes the formal semantics of this syntax. It will make the
2969 most sense if you are familiar with the formal specification of
2970 declarators in the ISO C standard.
2972 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2973 D1}, where @code{T} contains declaration specifiers that specify a type
2974 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2975 contains an identifier @var{ident}. The type specified for @var{ident}
2976 for derived declarators whose type does not include an attribute
2977 specifier is as in the ISO C standard.
2979 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2980 and the declaration @code{T D} specifies the type
2981 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2982 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2983 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2985 If @code{D1} has the form @code{*
2986 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2987 declaration @code{T D} specifies the type
2988 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2989 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2990 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2996 void (__attribute__((noreturn)) ****f) (void);
3000 specifies the type ``pointer to pointer to pointer to pointer to
3001 non-returning function returning @code{void}''. As another example,
3004 char *__attribute__((aligned(8))) *f;
3008 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
3009 Note again that this does not work with most attributes; for example,
3010 the usage of @samp{aligned} and @samp{noreturn} attributes given above
3011 is not yet supported.
3013 For compatibility with existing code written for compiler versions that
3014 did not implement attributes on nested declarators, some laxity is
3015 allowed in the placing of attributes. If an attribute that only applies
3016 to types is applied to a declaration, it will be treated as applying to
3017 the type of that declaration. If an attribute that only applies to
3018 declarations is applied to the type of a declaration, it will be treated
3019 as applying to that declaration; and, for compatibility with code
3020 placing the attributes immediately before the identifier declared, such
3021 an attribute applied to a function return type will be treated as
3022 applying to the function type, and such an attribute applied to an array
3023 element type will be treated as applying to the array type. If an
3024 attribute that only applies to function types is applied to a
3025 pointer-to-function type, it will be treated as applying to the pointer
3026 target type; if such an attribute is applied to a function return type
3027 that is not a pointer-to-function type, it will be treated as applying
3028 to the function type.
3030 @node Function Prototypes
3031 @section Prototypes and Old-Style Function Definitions
3032 @cindex function prototype declarations
3033 @cindex old-style function definitions
3034 @cindex promotion of formal parameters
3036 GNU C extends ISO C to allow a function prototype to override a later
3037 old-style non-prototype definition. Consider the following example:
3040 /* @r{Use prototypes unless the compiler is old-fashioned.} */
3047 /* @r{Prototype function declaration.} */
3048 int isroot P((uid_t));
3050 /* @r{Old-style function definition.} */
3052 isroot (x) /* @r{??? lossage here ???} */
3059 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
3060 not allow this example, because subword arguments in old-style
3061 non-prototype definitions are promoted. Therefore in this example the
3062 function definition's argument is really an @code{int}, which does not
3063 match the prototype argument type of @code{short}.
3065 This restriction of ISO C makes it hard to write code that is portable
3066 to traditional C compilers, because the programmer does not know
3067 whether the @code{uid_t} type is @code{short}, @code{int}, or
3068 @code{long}. Therefore, in cases like these GNU C allows a prototype
3069 to override a later old-style definition. More precisely, in GNU C, a
3070 function prototype argument type overrides the argument type specified
3071 by a later old-style definition if the former type is the same as the
3072 latter type before promotion. Thus in GNU C the above example is
3073 equivalent to the following:
3086 GNU C++ does not support old-style function definitions, so this
3087 extension is irrelevant.
3090 @section C++ Style Comments
3092 @cindex C++ comments
3093 @cindex comments, C++ style
3095 In GNU C, you may use C++ style comments, which start with @samp{//} and
3096 continue until the end of the line. Many other C implementations allow
3097 such comments, and they are included in the 1999 C standard. However,
3098 C++ style comments are not recognized if you specify an @option{-std}
3099 option specifying a version of ISO C before C99, or @option{-ansi}
3100 (equivalent to @option{-std=c89}).
3103 @section Dollar Signs in Identifier Names
3105 @cindex dollar signs in identifier names
3106 @cindex identifier names, dollar signs in
3108 In GNU C, you may normally use dollar signs in identifier names.
3109 This is because many traditional C implementations allow such identifiers.
3110 However, dollar signs in identifiers are not supported on a few target
3111 machines, typically because the target assembler does not allow them.
3113 @node Character Escapes
3114 @section The Character @key{ESC} in Constants
3116 You can use the sequence @samp{\e} in a string or character constant to
3117 stand for the ASCII character @key{ESC}.
3120 @section Inquiring on Alignment of Types or Variables
3122 @cindex type alignment
3123 @cindex variable alignment
3125 The keyword @code{__alignof__} allows you to inquire about how an object
3126 is aligned, or the minimum alignment usually required by a type. Its
3127 syntax is just like @code{sizeof}.
3129 For example, if the target machine requires a @code{double} value to be
3130 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
3131 This is true on many RISC machines. On more traditional machine
3132 designs, @code{__alignof__ (double)} is 4 or even 2.
3134 Some machines never actually require alignment; they allow reference to any
3135 data type even at an odd address. For these machines, @code{__alignof__}
3136 reports the @emph{recommended} alignment of a type.
3138 If the operand of @code{__alignof__} is an lvalue rather than a type,
3139 its value is the required alignment for its type, taking into account
3140 any minimum alignment specified with GCC's @code{__attribute__}
3141 extension (@pxref{Variable Attributes}). For example, after this
3145 struct foo @{ int x; char y; @} foo1;
3149 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
3150 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
3152 It is an error to ask for the alignment of an incomplete type.
3154 @node Variable Attributes
3155 @section Specifying Attributes of Variables
3156 @cindex attribute of variables
3157 @cindex variable attributes
3159 The keyword @code{__attribute__} allows you to specify special
3160 attributes of variables or structure fields. This keyword is followed
3161 by an attribute specification inside double parentheses. Some
3162 attributes are currently defined generically for variables.
3163 Other attributes are defined for variables on particular target
3164 systems. Other attributes are available for functions
3165 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
3166 Other front ends might define more attributes
3167 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
3169 You may also specify attributes with @samp{__} preceding and following
3170 each keyword. This allows you to use them in header files without
3171 being concerned about a possible macro of the same name. For example,
3172 you may use @code{__aligned__} instead of @code{aligned}.
3174 @xref{Attribute Syntax}, for details of the exact syntax for using
3178 @cindex @code{aligned} attribute
3179 @item aligned (@var{alignment})
3180 This attribute specifies a minimum alignment for the variable or
3181 structure field, measured in bytes. For example, the declaration:
3184 int x __attribute__ ((aligned (16))) = 0;
3188 causes the compiler to allocate the global variable @code{x} on a
3189 16-byte boundary. On a 68040, this could be used in conjunction with
3190 an @code{asm} expression to access the @code{move16} instruction which
3191 requires 16-byte aligned operands.
3193 You can also specify the alignment of structure fields. For example, to
3194 create a double-word aligned @code{int} pair, you could write:
3197 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
3201 This is an alternative to creating a union with a @code{double} member
3202 that forces the union to be double-word aligned.
3204 As in the preceding examples, you can explicitly specify the alignment
3205 (in bytes) that you wish the compiler to use for a given variable or
3206 structure field. Alternatively, you can leave out the alignment factor
3207 and just ask the compiler to align a variable or field to the maximum
3208 useful alignment for the target machine you are compiling for. For
3209 example, you could write:
3212 short array[3] __attribute__ ((aligned));
3215 Whenever you leave out the alignment factor in an @code{aligned} attribute
3216 specification, the compiler automatically sets the alignment for the declared
3217 variable or field to the largest alignment which is ever used for any data
3218 type on the target machine you are compiling for. Doing this can often make
3219 copy operations more efficient, because the compiler can use whatever
3220 instructions copy the biggest chunks of memory when performing copies to
3221 or from the variables or fields that you have aligned this way.
3223 When used on a struct, or struct member, the @code{aligned} attribute can
3224 only increase the alignment; in order to decrease it, the @code{packed}
3225 attribute must be specified as well. When used as part of a typedef, the
3226 @code{aligned} attribute can both increase and decrease alignment, and
3227 specifying the @code{packed} attribute will generate a warning.
3229 Note that the effectiveness of @code{aligned} attributes may be limited
3230 by inherent limitations in your linker. On many systems, the linker is
3231 only able to arrange for variables to be aligned up to a certain maximum
3232 alignment. (For some linkers, the maximum supported alignment may
3233 be very very small.) If your linker is only able to align variables
3234 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3235 in an @code{__attribute__} will still only provide you with 8 byte
3236 alignment. See your linker documentation for further information.
3238 The @code{aligned} attribute can also be used for functions
3239 (@pxref{Function Attributes}.)
3241 @item cleanup (@var{cleanup_function})
3242 @cindex @code{cleanup} attribute
3243 The @code{cleanup} attribute runs a function when the variable goes
3244 out of scope. This attribute can only be applied to auto function
3245 scope variables; it may not be applied to parameters or variables
3246 with static storage duration. The function must take one parameter,
3247 a pointer to a type compatible with the variable. The return value
3248 of the function (if any) is ignored.
3250 If @option{-fexceptions} is enabled, then @var{cleanup_function}
3251 will be run during the stack unwinding that happens during the
3252 processing of the exception. Note that the @code{cleanup} attribute
3253 does not allow the exception to be caught, only to perform an action.
3254 It is undefined what happens if @var{cleanup_function} does not
3259 @cindex @code{common} attribute
3260 @cindex @code{nocommon} attribute
3263 The @code{common} attribute requests GCC to place a variable in
3264 ``common'' storage. The @code{nocommon} attribute requests the
3265 opposite---to allocate space for it directly.
3267 These attributes override the default chosen by the
3268 @option{-fno-common} and @option{-fcommon} flags respectively.
3271 @cindex @code{deprecated} attribute
3272 The @code{deprecated} attribute results in a warning if the variable
3273 is used anywhere in the source file. This is useful when identifying
3274 variables that are expected to be removed in a future version of a
3275 program. The warning also includes the location of the declaration
3276 of the deprecated variable, to enable users to easily find further
3277 information about why the variable is deprecated, or what they should
3278 do instead. Note that the warning only occurs for uses:
3281 extern int old_var __attribute__ ((deprecated));
3283 int new_fn () @{ return old_var; @}
3286 results in a warning on line 3 but not line 2.
3288 The @code{deprecated} attribute can also be used for functions and
3289 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3291 @item mode (@var{mode})
3292 @cindex @code{mode} attribute
3293 This attribute specifies the data type for the declaration---whichever
3294 type corresponds to the mode @var{mode}. This in effect lets you
3295 request an integer or floating point type according to its width.
3297 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3298 indicate the mode corresponding to a one-byte integer, @samp{word} or
3299 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3300 or @samp{__pointer__} for the mode used to represent pointers.
3303 @cindex @code{packed} attribute
3304 The @code{packed} attribute specifies that a variable or structure field
3305 should have the smallest possible alignment---one byte for a variable,
3306 and one bit for a field, unless you specify a larger value with the
3307 @code{aligned} attribute.
3309 Here is a structure in which the field @code{x} is packed, so that it
3310 immediately follows @code{a}:
3316 int x[2] __attribute__ ((packed));
3320 @item section ("@var{section-name}")
3321 @cindex @code{section} variable attribute
3322 Normally, the compiler places the objects it generates in sections like
3323 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3324 or you need certain particular variables to appear in special sections,
3325 for example to map to special hardware. The @code{section}
3326 attribute specifies that a variable (or function) lives in a particular
3327 section. For example, this small program uses several specific section names:
3330 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3331 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3332 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3333 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3337 /* @r{Initialize stack pointer} */
3338 init_sp (stack + sizeof (stack));
3340 /* @r{Initialize initialized data} */
3341 memcpy (&init_data, &data, &edata - &data);
3343 /* @r{Turn on the serial ports} */
3350 Use the @code{section} attribute with an @emph{initialized} definition
3351 of a @emph{global} variable, as shown in the example. GCC issues
3352 a warning and otherwise ignores the @code{section} attribute in
3353 uninitialized variable declarations.
3355 You may only use the @code{section} attribute with a fully initialized
3356 global definition because of the way linkers work. The linker requires
3357 each object be defined once, with the exception that uninitialized
3358 variables tentatively go in the @code{common} (or @code{bss}) section
3359 and can be multiply ``defined''. You can force a variable to be
3360 initialized with the @option{-fno-common} flag or the @code{nocommon}
3363 Some file formats do not support arbitrary sections so the @code{section}
3364 attribute is not available on all platforms.
3365 If you need to map the entire contents of a module to a particular
3366 section, consider using the facilities of the linker instead.
3369 @cindex @code{shared} variable attribute
3370 On Microsoft Windows, in addition to putting variable definitions in a named
3371 section, the section can also be shared among all running copies of an
3372 executable or DLL@. For example, this small program defines shared data
3373 by putting it in a named section @code{shared} and marking the section
3377 int foo __attribute__((section ("shared"), shared)) = 0;
3382 /* @r{Read and write foo. All running
3383 copies see the same value.} */
3389 You may only use the @code{shared} attribute along with @code{section}
3390 attribute with a fully initialized global definition because of the way
3391 linkers work. See @code{section} attribute for more information.
3393 The @code{shared} attribute is only available on Microsoft Windows@.
3395 @item tls_model ("@var{tls_model}")
3396 @cindex @code{tls_model} attribute
3397 The @code{tls_model} attribute sets thread-local storage model
3398 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3399 overriding @option{-ftls-model=} command line switch on a per-variable
3401 The @var{tls_model} argument should be one of @code{global-dynamic},
3402 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3404 Not all targets support this attribute.
3407 This attribute, attached to a variable, means that the variable is meant
3408 to be possibly unused. GCC will not produce a warning for this
3412 This attribute, attached to a variable, means that the variable must be
3413 emitted even if it appears that the variable is not referenced.
3415 @item vector_size (@var{bytes})
3416 This attribute specifies the vector size for the variable, measured in
3417 bytes. For example, the declaration:
3420 int foo __attribute__ ((vector_size (16)));
3424 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3425 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3426 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3428 This attribute is only applicable to integral and float scalars,
3429 although arrays, pointers, and function return values are allowed in
3430 conjunction with this construct.
3432 Aggregates with this attribute are invalid, even if they are of the same
3433 size as a corresponding scalar. For example, the declaration:
3436 struct S @{ int a; @};
3437 struct S __attribute__ ((vector_size (16))) foo;
3441 is invalid even if the size of the structure is the same as the size of
3445 The @code{selectany} attribute causes an initialized global variable to
3446 have link-once semantics. When multiple definitions of the variable are
3447 encountered by the linker, the first is selected and the remainder are
3448 discarded. Following usage by the Microsoft compiler, the linker is told
3449 @emph{not} to warn about size or content differences of the multiple
3452 Although the primary usage of this attribute is for POD types, the
3453 attribute can also be applied to global C++ objects that are initialized
3454 by a constructor. In this case, the static initialization and destruction
3455 code for the object is emitted in each translation defining the object,
3456 but the calls to the constructor and destructor are protected by a
3457 link-once guard variable.
3459 The @code{selectany} attribute is only available on Microsoft Windows
3460 targets. You can use @code{__declspec (selectany)} as a synonym for
3461 @code{__attribute__ ((selectany))} for compatibility with other
3465 The @code{weak} attribute is described in @xref{Function Attributes}.
3468 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3471 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3475 @subsection Blackfin Variable Attributes
3477 Three attributes are currently defined for the Blackfin.
3483 @cindex @code{l1_data} variable attribute
3484 @cindex @code{l1_data_A} variable attribute
3485 @cindex @code{l1_data_B} variable attribute
3486 Use these attributes on the Blackfin to place the variable into L1 Data SRAM.
3487 Variables with @code{l1_data} attribute will be put into the specific section
3488 named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into
3489 the specific section named @code{.l1.data.A}. Those with @code{l1_data_B}
3490 attribute will be put into the specific section named @code{.l1.data.B}.
3493 @subsection M32R/D Variable Attributes
3495 One attribute is currently defined for the M32R/D@.
3498 @item model (@var{model-name})
3499 @cindex variable addressability on the M32R/D
3500 Use this attribute on the M32R/D to set the addressability of an object.
3501 The identifier @var{model-name} is one of @code{small}, @code{medium},
3502 or @code{large}, representing each of the code models.
3504 Small model objects live in the lower 16MB of memory (so that their
3505 addresses can be loaded with the @code{ld24} instruction).
3507 Medium and large model objects may live anywhere in the 32-bit address space
3508 (the compiler will generate @code{seth/add3} instructions to load their
3512 @anchor{i386 Variable Attributes}
3513 @subsection i386 Variable Attributes
3515 Two attributes are currently defined for i386 configurations:
3516 @code{ms_struct} and @code{gcc_struct}
3521 @cindex @code{ms_struct} attribute
3522 @cindex @code{gcc_struct} attribute
3524 If @code{packed} is used on a structure, or if bit-fields are used
3525 it may be that the Microsoft ABI packs them differently
3526 than GCC would normally pack them. Particularly when moving packed
3527 data between functions compiled with GCC and the native Microsoft compiler
3528 (either via function call or as data in a file), it may be necessary to access
3531 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3532 compilers to match the native Microsoft compiler.
3534 The Microsoft structure layout algorithm is fairly simple with the exception
3535 of the bitfield packing:
3537 The padding and alignment of members of structures and whether a bit field
3538 can straddle a storage-unit boundary
3541 @item Structure members are stored sequentially in the order in which they are
3542 declared: the first member has the lowest memory address and the last member
3545 @item Every data object has an alignment-requirement. The alignment-requirement
3546 for all data except structures, unions, and arrays is either the size of the
3547 object or the current packing size (specified with either the aligned attribute
3548 or the pack pragma), whichever is less. For structures, unions, and arrays,
3549 the alignment-requirement is the largest alignment-requirement of its members.
3550 Every object is allocated an offset so that:
3552 offset % alignment-requirement == 0
3554 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
3555 unit if the integral types are the same size and if the next bit field fits
3556 into the current allocation unit without crossing the boundary imposed by the
3557 common alignment requirements of the bit fields.
3560 Handling of zero-length bitfields:
3562 MSVC interprets zero-length bitfields in the following ways:
3565 @item If a zero-length bitfield is inserted between two bitfields that would
3566 normally be coalesced, the bitfields will not be coalesced.
3573 unsigned long bf_1 : 12;
3575 unsigned long bf_2 : 12;
3579 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
3580 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
3582 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
3583 alignment of the zero-length bitfield is greater than the member that follows it,
3584 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
3604 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
3605 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
3606 bitfield will not affect the alignment of @code{bar} or, as a result, the size
3609 Taking this into account, it is important to note the following:
3612 @item If a zero-length bitfield follows a normal bitfield, the type of the
3613 zero-length bitfield may affect the alignment of the structure as whole. For
3614 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
3615 normal bitfield, and is of type short.
3617 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
3618 still affect the alignment of the structure:
3628 Here, @code{t4} will take up 4 bytes.
3631 @item Zero-length bitfields following non-bitfield members are ignored:
3642 Here, @code{t5} will take up 2 bytes.
3646 @subsection PowerPC Variable Attributes
3648 Three attributes currently are defined for PowerPC configurations:
3649 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
3651 For full documentation of the struct attributes please see the
3652 documentation in the @xref{i386 Variable Attributes}, section.
3654 For documentation of @code{altivec} attribute please see the
3655 documentation in the @xref{PowerPC Type Attributes}, section.
3657 @subsection SPU Variable Attributes
3659 The SPU supports the @code{spu_vector} attribute for variables. For
3660 documentation of this attribute please see the documentation in the
3661 @xref{SPU Type Attributes}, section.
3663 @subsection Xstormy16 Variable Attributes
3665 One attribute is currently defined for xstormy16 configurations:
3670 @cindex @code{below100} attribute
3672 If a variable has the @code{below100} attribute (@code{BELOW100} is
3673 allowed also), GCC will place the variable in the first 0x100 bytes of
3674 memory and use special opcodes to access it. Such variables will be
3675 placed in either the @code{.bss_below100} section or the
3676 @code{.data_below100} section.
3680 @node Type Attributes
3681 @section Specifying Attributes of Types
3682 @cindex attribute of types
3683 @cindex type attributes
3685 The keyword @code{__attribute__} allows you to specify special
3686 attributes of @code{struct} and @code{union} types when you define
3687 such types. This keyword is followed by an attribute specification
3688 inside double parentheses. Seven attributes are currently defined for
3689 types: @code{aligned}, @code{packed}, @code{transparent_union},
3690 @code{unused}, @code{deprecated}, @code{visibility}, and
3691 @code{may_alias}. Other attributes are defined for functions
3692 (@pxref{Function Attributes}) and for variables (@pxref{Variable
3695 You may also specify any one of these attributes with @samp{__}
3696 preceding and following its keyword. This allows you to use these
3697 attributes in header files without being concerned about a possible
3698 macro of the same name. For example, you may use @code{__aligned__}
3699 instead of @code{aligned}.
3701 You may specify type attributes either in a @code{typedef} declaration
3702 or in an enum, struct or union type declaration or definition.
3704 For an enum, struct or union type, you may specify attributes either
3705 between the enum, struct or union tag and the name of the type, or
3706 just past the closing curly brace of the @emph{definition}. The
3707 former syntax is preferred.
3709 @xref{Attribute Syntax}, for details of the exact syntax for using
3713 @cindex @code{aligned} attribute
3714 @item aligned (@var{alignment})
3715 This attribute specifies a minimum alignment (in bytes) for variables
3716 of the specified type. For example, the declarations:
3719 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3720 typedef int more_aligned_int __attribute__ ((aligned (8)));
3724 force the compiler to insure (as far as it can) that each variable whose
3725 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3726 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3727 variables of type @code{struct S} aligned to 8-byte boundaries allows
3728 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3729 store) instructions when copying one variable of type @code{struct S} to
3730 another, thus improving run-time efficiency.
3732 Note that the alignment of any given @code{struct} or @code{union} type
3733 is required by the ISO C standard to be at least a perfect multiple of
3734 the lowest common multiple of the alignments of all of the members of
3735 the @code{struct} or @code{union} in question. This means that you @emph{can}
3736 effectively adjust the alignment of a @code{struct} or @code{union}
3737 type by attaching an @code{aligned} attribute to any one of the members
3738 of such a type, but the notation illustrated in the example above is a
3739 more obvious, intuitive, and readable way to request the compiler to
3740 adjust the alignment of an entire @code{struct} or @code{union} type.
3742 As in the preceding example, you can explicitly specify the alignment
3743 (in bytes) that you wish the compiler to use for a given @code{struct}
3744 or @code{union} type. Alternatively, you can leave out the alignment factor
3745 and just ask the compiler to align a type to the maximum
3746 useful alignment for the target machine you are compiling for. For
3747 example, you could write:
3750 struct S @{ short f[3]; @} __attribute__ ((aligned));
3753 Whenever you leave out the alignment factor in an @code{aligned}
3754 attribute specification, the compiler automatically sets the alignment
3755 for the type to the largest alignment which is ever used for any data
3756 type on the target machine you are compiling for. Doing this can often
3757 make copy operations more efficient, because the compiler can use
3758 whatever instructions copy the biggest chunks of memory when performing
3759 copies to or from the variables which have types that you have aligned
3762 In the example above, if the size of each @code{short} is 2 bytes, then
3763 the size of the entire @code{struct S} type is 6 bytes. The smallest
3764 power of two which is greater than or equal to that is 8, so the
3765 compiler sets the alignment for the entire @code{struct S} type to 8
3768 Note that although you can ask the compiler to select a time-efficient
3769 alignment for a given type and then declare only individual stand-alone
3770 objects of that type, the compiler's ability to select a time-efficient
3771 alignment is primarily useful only when you plan to create arrays of
3772 variables having the relevant (efficiently aligned) type. If you
3773 declare or use arrays of variables of an efficiently-aligned type, then
3774 it is likely that your program will also be doing pointer arithmetic (or
3775 subscripting, which amounts to the same thing) on pointers to the
3776 relevant type, and the code that the compiler generates for these
3777 pointer arithmetic operations will often be more efficient for
3778 efficiently-aligned types than for other types.
3780 The @code{aligned} attribute can only increase the alignment; but you
3781 can decrease it by specifying @code{packed} as well. See below.
3783 Note that the effectiveness of @code{aligned} attributes may be limited
3784 by inherent limitations in your linker. On many systems, the linker is
3785 only able to arrange for variables to be aligned up to a certain maximum
3786 alignment. (For some linkers, the maximum supported alignment may
3787 be very very small.) If your linker is only able to align variables
3788 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3789 in an @code{__attribute__} will still only provide you with 8 byte
3790 alignment. See your linker documentation for further information.
3793 This attribute, attached to @code{struct} or @code{union} type
3794 definition, specifies that each member (other than zero-width bitfields)
3795 of the structure or union is placed to minimize the memory required. When
3796 attached to an @code{enum} definition, it indicates that the smallest
3797 integral type should be used.
3799 @opindex fshort-enums
3800 Specifying this attribute for @code{struct} and @code{union} types is
3801 equivalent to specifying the @code{packed} attribute on each of the
3802 structure or union members. Specifying the @option{-fshort-enums}
3803 flag on the line is equivalent to specifying the @code{packed}
3804 attribute on all @code{enum} definitions.
3806 In the following example @code{struct my_packed_struct}'s members are
3807 packed closely together, but the internal layout of its @code{s} member
3808 is not packed---to do that, @code{struct my_unpacked_struct} would need to
3812 struct my_unpacked_struct
3818 struct __attribute__ ((__packed__)) my_packed_struct
3822 struct my_unpacked_struct s;
3826 You may only specify this attribute on the definition of a @code{enum},
3827 @code{struct} or @code{union}, not on a @code{typedef} which does not
3828 also define the enumerated type, structure or union.
3830 @item transparent_union
3831 This attribute, attached to a @code{union} type definition, indicates
3832 that any function parameter having that union type causes calls to that
3833 function to be treated in a special way.
3835 First, the argument corresponding to a transparent union type can be of
3836 any type in the union; no cast is required. Also, if the union contains
3837 a pointer type, the corresponding argument can be a null pointer
3838 constant or a void pointer expression; and if the union contains a void
3839 pointer type, the corresponding argument can be any pointer expression.
3840 If the union member type is a pointer, qualifiers like @code{const} on
3841 the referenced type must be respected, just as with normal pointer
3844 Second, the argument is passed to the function using the calling
3845 conventions of the first member of the transparent union, not the calling
3846 conventions of the union itself. All members of the union must have the
3847 same machine representation; this is necessary for this argument passing
3850 Transparent unions are designed for library functions that have multiple
3851 interfaces for compatibility reasons. For example, suppose the
3852 @code{wait} function must accept either a value of type @code{int *} to
3853 comply with Posix, or a value of type @code{union wait *} to comply with
3854 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3855 @code{wait} would accept both kinds of arguments, but it would also
3856 accept any other pointer type and this would make argument type checking
3857 less useful. Instead, @code{<sys/wait.h>} might define the interface
3865 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3867 pid_t wait (wait_status_ptr_t);
3870 This interface allows either @code{int *} or @code{union wait *}
3871 arguments to be passed, using the @code{int *} calling convention.
3872 The program can call @code{wait} with arguments of either type:
3875 int w1 () @{ int w; return wait (&w); @}
3876 int w2 () @{ union wait w; return wait (&w); @}
3879 With this interface, @code{wait}'s implementation might look like this:
3882 pid_t wait (wait_status_ptr_t p)
3884 return waitpid (-1, p.__ip, 0);
3889 When attached to a type (including a @code{union} or a @code{struct}),
3890 this attribute means that variables of that type are meant to appear
3891 possibly unused. GCC will not produce a warning for any variables of
3892 that type, even if the variable appears to do nothing. This is often
3893 the case with lock or thread classes, which are usually defined and then
3894 not referenced, but contain constructors and destructors that have
3895 nontrivial bookkeeping functions.
3898 The @code{deprecated} attribute results in a warning if the type
3899 is used anywhere in the source file. This is useful when identifying
3900 types that are expected to be removed in a future version of a program.
3901 If possible, the warning also includes the location of the declaration
3902 of the deprecated type, to enable users to easily find further
3903 information about why the type is deprecated, or what they should do
3904 instead. Note that the warnings only occur for uses and then only
3905 if the type is being applied to an identifier that itself is not being
3906 declared as deprecated.
3909 typedef int T1 __attribute__ ((deprecated));
3913 typedef T1 T3 __attribute__ ((deprecated));
3914 T3 z __attribute__ ((deprecated));
3917 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3918 warning is issued for line 4 because T2 is not explicitly
3919 deprecated. Line 5 has no warning because T3 is explicitly
3920 deprecated. Similarly for line 6.
3922 The @code{deprecated} attribute can also be used for functions and
3923 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3926 Accesses to objects with types with this attribute are not subjected to
3927 type-based alias analysis, but are instead assumed to be able to alias
3928 any other type of objects, just like the @code{char} type. See
3929 @option{-fstrict-aliasing} for more information on aliasing issues.
3934 typedef short __attribute__((__may_alias__)) short_a;
3940 short_a *b = (short_a *) &a;
3944 if (a == 0x12345678)
3951 If you replaced @code{short_a} with @code{short} in the variable
3952 declaration, the above program would abort when compiled with
3953 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3954 above in recent GCC versions.
3957 In C++, attribute visibility (@pxref{Function Attributes}) can also be
3958 applied to class, struct, union and enum types. Unlike other type
3959 attributes, the attribute must appear between the initial keyword and
3960 the name of the type; it cannot appear after the body of the type.
3962 Note that the type visibility is applied to vague linkage entities
3963 associated with the class (vtable, typeinfo node, etc.). In
3964 particular, if a class is thrown as an exception in one shared object
3965 and caught in another, the class must have default visibility.
3966 Otherwise the two shared objects will be unable to use the same
3967 typeinfo node and exception handling will break.
3969 @subsection ARM Type Attributes
3971 On those ARM targets that support @code{dllimport} (such as Symbian
3972 OS), you can use the @code{notshared} attribute to indicate that the
3973 virtual table and other similar data for a class should not be
3974 exported from a DLL@. For example:
3977 class __declspec(notshared) C @{
3979 __declspec(dllimport) C();
3983 __declspec(dllexport)
3987 In this code, @code{C::C} is exported from the current DLL, but the
3988 virtual table for @code{C} is not exported. (You can use
3989 @code{__attribute__} instead of @code{__declspec} if you prefer, but
3990 most Symbian OS code uses @code{__declspec}.)
3992 @anchor{i386 Type Attributes}
3993 @subsection i386 Type Attributes
3995 Two attributes are currently defined for i386 configurations:
3996 @code{ms_struct} and @code{gcc_struct}
4000 @cindex @code{ms_struct}
4001 @cindex @code{gcc_struct}
4003 If @code{packed} is used on a structure, or if bit-fields are used
4004 it may be that the Microsoft ABI packs them differently
4005 than GCC would normally pack them. Particularly when moving packed
4006 data between functions compiled with GCC and the native Microsoft compiler
4007 (either via function call or as data in a file), it may be necessary to access
4010 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
4011 compilers to match the native Microsoft compiler.
4014 To specify multiple attributes, separate them by commas within the
4015 double parentheses: for example, @samp{__attribute__ ((aligned (16),
4018 @anchor{PowerPC Type Attributes}
4019 @subsection PowerPC Type Attributes
4021 Three attributes currently are defined for PowerPC configurations:
4022 @code{altivec}, @code{ms_struct} and @code{gcc_struct}.
4024 For full documentation of the struct attributes please see the
4025 documentation in the @xref{i386 Type Attributes}, section.
4027 The @code{altivec} attribute allows one to declare AltiVec vector data
4028 types supported by the AltiVec Programming Interface Manual. The
4029 attribute requires an argument to specify one of three vector types:
4030 @code{vector__}, @code{pixel__} (always followed by unsigned short),
4031 and @code{bool__} (always followed by unsigned).
4034 __attribute__((altivec(vector__)))
4035 __attribute__((altivec(pixel__))) unsigned short
4036 __attribute__((altivec(bool__))) unsigned
4039 These attributes mainly are intended to support the @code{__vector},
4040 @code{__pixel}, and @code{__bool} AltiVec keywords.
4042 @anchor{SPU Type Attributes}
4043 @subsection SPU Type Attributes
4045 The SPU supports the @code{spu_vector} attribute for types. This attribute
4046 allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU
4047 Language Extensions Specification. It is intended to support the
4048 @code{__vector} keyword.
4052 @section An Inline Function is As Fast As a Macro
4053 @cindex inline functions
4054 @cindex integrating function code
4056 @cindex macros, inline alternative
4058 By declaring a function inline, you can direct GCC to make
4059 calls to that function faster. One way GCC can achieve this is to
4060 integrate that function's code into the code for its callers. This
4061 makes execution faster by eliminating the function-call overhead; in
4062 addition, if any of the actual argument values are constant, their
4063 known values may permit simplifications at compile time so that not
4064 all of the inline function's code needs to be included. The effect on
4065 code size is less predictable; object code may be larger or smaller
4066 with function inlining, depending on the particular case. You can
4067 also direct GCC to try to integrate all ``simple enough'' functions
4068 into their callers with the option @option{-finline-functions}.
4070 GCC implements three different semantics of declaring a function
4071 inline. One is available with @option{-std=gnu89} or
4072 @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present
4073 on all inline declarations, another when @option{-std=c99} or
4074 @option{-std=gnu99} (without @option{-fgnu89-inline}), and the third
4075 is used when compiling C++.
4077 To declare a function inline, use the @code{inline} keyword in its
4078 declaration, like this:
4088 If you are writing a header file to be included in ISO C89 programs, write
4089 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.
4091 The three types of inlining behave similarly in two important cases:
4092 when the @code{inline} keyword is used on a @code{static} function,
4093 like the example above, and when a function is first declared without
4094 using the @code{inline} keyword and then is defined with
4095 @code{inline}, like this:
4098 extern int inc (int *a);
4106 In both of these common cases, the program behaves the same as if you
4107 had not used the @code{inline} keyword, except for its speed.
4109 @cindex inline functions, omission of
4110 @opindex fkeep-inline-functions
4111 When a function is both inline and @code{static}, if all calls to the
4112 function are integrated into the caller, and the function's address is
4113 never used, then the function's own assembler code is never referenced.
4114 In this case, GCC does not actually output assembler code for the
4115 function, unless you specify the option @option{-fkeep-inline-functions}.
4116 Some calls cannot be integrated for various reasons (in particular,
4117 calls that precede the function's definition cannot be integrated, and
4118 neither can recursive calls within the definition). If there is a
4119 nonintegrated call, then the function is compiled to assembler code as
4120 usual. The function must also be compiled as usual if the program
4121 refers to its address, because that can't be inlined.
4124 Note that certain usages in a function definition can make it unsuitable
4125 for inline substitution. Among these usages are: use of varargs, use of
4126 alloca, use of variable sized data types (@pxref{Variable Length}),
4127 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
4128 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
4129 will warn when a function marked @code{inline} could not be substituted,
4130 and will give the reason for the failure.
4132 @cindex automatic @code{inline} for C++ member fns
4133 @cindex @code{inline} automatic for C++ member fns
4134 @cindex member fns, automatically @code{inline}
4135 @cindex C++ member fns, automatically @code{inline}
4136 @opindex fno-default-inline
4137 As required by ISO C++, GCC considers member functions defined within
4138 the body of a class to be marked inline even if they are
4139 not explicitly declared with the @code{inline} keyword. You can
4140 override this with @option{-fno-default-inline}; @pxref{C++ Dialect
4141 Options,,Options Controlling C++ Dialect}.
4143 GCC does not inline any functions when not optimizing unless you specify
4144 the @samp{always_inline} attribute for the function, like this:
4147 /* @r{Prototype.} */
4148 inline void foo (const char) __attribute__((always_inline));
4151 The remainder of this section is specific to GNU C89 inlining.
4153 @cindex non-static inline function
4154 When an inline function is not @code{static}, then the compiler must assume
4155 that there may be calls from other source files; since a global symbol can
4156 be defined only once in any program, the function must not be defined in
4157 the other source files, so the calls therein cannot be integrated.
4158 Therefore, a non-@code{static} inline function is always compiled on its
4159 own in the usual fashion.
4161 If you specify both @code{inline} and @code{extern} in the function
4162 definition, then the definition is used only for inlining. In no case
4163 is the function compiled on its own, not even if you refer to its
4164 address explicitly. Such an address becomes an external reference, as
4165 if you had only declared the function, and had not defined it.
4167 This combination of @code{inline} and @code{extern} has almost the
4168 effect of a macro. The way to use it is to put a function definition in
4169 a header file with these keywords, and put another copy of the
4170 definition (lacking @code{inline} and @code{extern}) in a library file.
4171 The definition in the header file will cause most calls to the function
4172 to be inlined. If any uses of the function remain, they will refer to
4173 the single copy in the library.
4176 @section Assembler Instructions with C Expression Operands
4177 @cindex extended @code{asm}
4178 @cindex @code{asm} expressions
4179 @cindex assembler instructions
4182 In an assembler instruction using @code{asm}, you can specify the
4183 operands of the instruction using C expressions. This means you need not
4184 guess which registers or memory locations will contain the data you want
4187 You must specify an assembler instruction template much like what
4188 appears in a machine description, plus an operand constraint string for
4191 For example, here is how to use the 68881's @code{fsinx} instruction:
4194 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
4198 Here @code{angle} is the C expression for the input operand while
4199 @code{result} is that of the output operand. Each has @samp{"f"} as its
4200 operand constraint, saying that a floating point register is required.
4201 The @samp{=} in @samp{=f} indicates that the operand is an output; all
4202 output operands' constraints must use @samp{=}. The constraints use the
4203 same language used in the machine description (@pxref{Constraints}).
4205 Each operand is described by an operand-constraint string followed by
4206 the C expression in parentheses. A colon separates the assembler
4207 template from the first output operand and another separates the last
4208 output operand from the first input, if any. Commas separate the
4209 operands within each group. The total number of operands is currently
4210 limited to 30; this limitation may be lifted in some future version of
4213 If there are no output operands but there are input operands, you must
4214 place two consecutive colons surrounding the place where the output
4217 As of GCC version 3.1, it is also possible to specify input and output
4218 operands using symbolic names which can be referenced within the
4219 assembler code. These names are specified inside square brackets
4220 preceding the constraint string, and can be referenced inside the
4221 assembler code using @code{%[@var{name}]} instead of a percentage sign
4222 followed by the operand number. Using named operands the above example
4226 asm ("fsinx %[angle],%[output]"
4227 : [output] "=f" (result)
4228 : [angle] "f" (angle));
4232 Note that the symbolic operand names have no relation whatsoever to
4233 other C identifiers. You may use any name you like, even those of
4234 existing C symbols, but you must ensure that no two operands within the same
4235 assembler construct use the same symbolic name.
4237 Output operand expressions must be lvalues; the compiler can check this.
4238 The input operands need not be lvalues. The compiler cannot check
4239 whether the operands have data types that are reasonable for the
4240 instruction being executed. It does not parse the assembler instruction
4241 template and does not know what it means or even whether it is valid
4242 assembler input. The extended @code{asm} feature is most often used for
4243 machine instructions the compiler itself does not know exist. If
4244 the output expression cannot be directly addressed (for example, it is a
4245 bit-field), your constraint must allow a register. In that case, GCC
4246 will use the register as the output of the @code{asm}, and then store
4247 that register into the output.
4249 The ordinary output operands must be write-only; GCC will assume that
4250 the values in these operands before the instruction are dead and need
4251 not be generated. Extended asm supports input-output or read-write
4252 operands. Use the constraint character @samp{+} to indicate such an
4253 operand and list it with the output operands. You should only use
4254 read-write operands when the constraints for the operand (or the
4255 operand in which only some of the bits are to be changed) allow a
4258 You may, as an alternative, logically split its function into two
4259 separate operands, one input operand and one write-only output
4260 operand. The connection between them is expressed by constraints
4261 which say they need to be in the same location when the instruction
4262 executes. You can use the same C expression for both operands, or
4263 different expressions. For example, here we write the (fictitious)
4264 @samp{combine} instruction with @code{bar} as its read-only source
4265 operand and @code{foo} as its read-write destination:
4268 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
4272 The constraint @samp{"0"} for operand 1 says that it must occupy the
4273 same location as operand 0. A number in constraint is allowed only in
4274 an input operand and it must refer to an output operand.
4276 Only a number in the constraint can guarantee that one operand will be in
4277 the same place as another. The mere fact that @code{foo} is the value
4278 of both operands is not enough to guarantee that they will be in the
4279 same place in the generated assembler code. The following would not
4283 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
4286 Various optimizations or reloading could cause operands 0 and 1 to be in
4287 different registers; GCC knows no reason not to do so. For example, the
4288 compiler might find a copy of the value of @code{foo} in one register and
4289 use it for operand 1, but generate the output operand 0 in a different
4290 register (copying it afterward to @code{foo}'s own address). Of course,
4291 since the register for operand 1 is not even mentioned in the assembler
4292 code, the result will not work, but GCC can't tell that.
4294 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
4295 the operand number for a matching constraint. For example:
4298 asm ("cmoveq %1,%2,%[result]"
4299 : [result] "=r"(result)
4300 : "r" (test), "r"(new), "[result]"(old));
4303 Sometimes you need to make an @code{asm} operand be a specific register,
4304 but there's no matching constraint letter for that register @emph{by
4305 itself}. To force the operand into that register, use a local variable
4306 for the operand and specify the register in the variable declaration.
4307 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
4308 register constraint letter that matches the register:
4311 register int *p1 asm ("r0") = @dots{};
4312 register int *p2 asm ("r1") = @dots{};
4313 register int *result asm ("r0");
4314 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4317 @anchor{Example of asm with clobbered asm reg}
4318 In the above example, beware that a register that is call-clobbered by
4319 the target ABI will be overwritten by any function call in the
4320 assignment, including library calls for arithmetic operators.
4321 Assuming it is a call-clobbered register, this may happen to @code{r0}
4322 above by the assignment to @code{p2}. If you have to use such a
4323 register, use temporary variables for expressions between the register
4328 register int *p1 asm ("r0") = @dots{};
4329 register int *p2 asm ("r1") = t1;
4330 register int *result asm ("r0");
4331 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
4334 Some instructions clobber specific hard registers. To describe this,
4335 write a third colon after the input operands, followed by the names of
4336 the clobbered hard registers (given as strings). Here is a realistic
4337 example for the VAX:
4340 asm volatile ("movc3 %0,%1,%2"
4341 : /* @r{no outputs} */
4342 : "g" (from), "g" (to), "g" (count)
4343 : "r0", "r1", "r2", "r3", "r4", "r5");
4346 You may not write a clobber description in a way that overlaps with an
4347 input or output operand. For example, you may not have an operand
4348 describing a register class with one member if you mention that register
4349 in the clobber list. Variables declared to live in specific registers
4350 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
4351 have no part mentioned in the clobber description.
4352 There is no way for you to specify that an input
4353 operand is modified without also specifying it as an output
4354 operand. Note that if all the output operands you specify are for this
4355 purpose (and hence unused), you will then also need to specify
4356 @code{volatile} for the @code{asm} construct, as described below, to
4357 prevent GCC from deleting the @code{asm} statement as unused.
4359 If you refer to a particular hardware register from the assembler code,
4360 you will probably have to list the register after the third colon to
4361 tell the compiler the register's value is modified. In some assemblers,
4362 the register names begin with @samp{%}; to produce one @samp{%} in the
4363 assembler code, you must write @samp{%%} in the input.
4365 If your assembler instruction can alter the condition code register, add
4366 @samp{cc} to the list of clobbered registers. GCC on some machines
4367 represents the condition codes as a specific hardware register;
4368 @samp{cc} serves to name this register. On other machines, the
4369 condition code is handled differently, and specifying @samp{cc} has no
4370 effect. But it is valid no matter what the machine.
4372 If your assembler instructions access memory in an unpredictable
4373 fashion, add @samp{memory} to the list of clobbered registers. This
4374 will cause GCC to not keep memory values cached in registers across the
4375 assembler instruction and not optimize stores or loads to that memory.
4376 You will also want to add the @code{volatile} keyword if the memory
4377 affected is not listed in the inputs or outputs of the @code{asm}, as
4378 the @samp{memory} clobber does not count as a side-effect of the
4379 @code{asm}. If you know how large the accessed memory is, you can add
4380 it as input or output but if this is not known, you should add
4381 @samp{memory}. As an example, if you access ten bytes of a string, you
4382 can use a memory input like:
4385 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
4388 Note that in the following example the memory input is necessary,
4389 otherwise GCC might optimize the store to @code{x} away:
4396 asm ("magic stuff accessing an 'int' pointed to by '%1'"
4397 "=&d" (r) : "a" (y), "m" (*y));
4402 You can put multiple assembler instructions together in a single
4403 @code{asm} template, separated by the characters normally used in assembly
4404 code for the system. A combination that works in most places is a newline
4405 to break the line, plus a tab character to move to the instruction field
4406 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
4407 assembler allows semicolons as a line-breaking character. Note that some
4408 assembler dialects use semicolons to start a comment.
4409 The input operands are guaranteed not to use any of the clobbered
4410 registers, and neither will the output operands' addresses, so you can
4411 read and write the clobbered registers as many times as you like. Here
4412 is an example of multiple instructions in a template; it assumes the
4413 subroutine @code{_foo} accepts arguments in registers 9 and 10:
4416 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
4418 : "g" (from), "g" (to)
4422 Unless an output operand has the @samp{&} constraint modifier, GCC
4423 may allocate it in the same register as an unrelated input operand, on
4424 the assumption the inputs are consumed before the outputs are produced.
4425 This assumption may be false if the assembler code actually consists of
4426 more than one instruction. In such a case, use @samp{&} for each output
4427 operand that may not overlap an input. @xref{Modifiers}.
4429 If you want to test the condition code produced by an assembler
4430 instruction, you must include a branch and a label in the @code{asm}
4431 construct, as follows:
4434 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
4440 This assumes your assembler supports local labels, as the GNU assembler
4441 and most Unix assemblers do.
4443 Speaking of labels, jumps from one @code{asm} to another are not
4444 supported. The compiler's optimizers do not know about these jumps, and
4445 therefore they cannot take account of them when deciding how to
4448 @cindex macros containing @code{asm}
4449 Usually the most convenient way to use these @code{asm} instructions is to
4450 encapsulate them in macros that look like functions. For example,
4454 (@{ double __value, __arg = (x); \
4455 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
4460 Here the variable @code{__arg} is used to make sure that the instruction
4461 operates on a proper @code{double} value, and to accept only those
4462 arguments @code{x} which can convert automatically to a @code{double}.
4464 Another way to make sure the instruction operates on the correct data
4465 type is to use a cast in the @code{asm}. This is different from using a
4466 variable @code{__arg} in that it converts more different types. For
4467 example, if the desired type were @code{int}, casting the argument to
4468 @code{int} would accept a pointer with no complaint, while assigning the
4469 argument to an @code{int} variable named @code{__arg} would warn about
4470 using a pointer unless the caller explicitly casts it.
4472 If an @code{asm} has output operands, GCC assumes for optimization
4473 purposes the instruction has no side effects except to change the output
4474 operands. This does not mean instructions with a side effect cannot be
4475 used, but you must be careful, because the compiler may eliminate them
4476 if the output operands aren't used, or move them out of loops, or
4477 replace two with one if they constitute a common subexpression. Also,
4478 if your instruction does have a side effect on a variable that otherwise
4479 appears not to change, the old value of the variable may be reused later
4480 if it happens to be found in a register.
4482 You can prevent an @code{asm} instruction from being deleted
4483 by writing the keyword @code{volatile} after
4484 the @code{asm}. For example:
4487 #define get_and_set_priority(new) \
4489 asm volatile ("get_and_set_priority %0, %1" \
4490 : "=g" (__old) : "g" (new)); \
4495 The @code{volatile} keyword indicates that the instruction has
4496 important side-effects. GCC will not delete a volatile @code{asm} if
4497 it is reachable. (The instruction can still be deleted if GCC can
4498 prove that control-flow will never reach the location of the
4499 instruction.) Note that even a volatile @code{asm} instruction
4500 can be moved relative to other code, including across jump
4501 instructions. For example, on many targets there is a system
4502 register which can be set to control the rounding mode of
4503 floating point operations. You might try
4504 setting it with a volatile @code{asm}, like this PowerPC example:
4507 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
4512 This will not work reliably, as the compiler may move the addition back
4513 before the volatile @code{asm}. To make it work you need to add an
4514 artificial dependency to the @code{asm} referencing a variable in the code
4515 you don't want moved, for example:
4518 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
4522 Similarly, you can't expect a
4523 sequence of volatile @code{asm} instructions to remain perfectly
4524 consecutive. If you want consecutive output, use a single @code{asm}.
4525 Also, GCC will perform some optimizations across a volatile @code{asm}
4526 instruction; GCC does not ``forget everything'' when it encounters
4527 a volatile @code{asm} instruction the way some other compilers do.
4529 An @code{asm} instruction without any output operands will be treated
4530 identically to a volatile @code{asm} instruction.
4532 It is a natural idea to look for a way to give access to the condition
4533 code left by the assembler instruction. However, when we attempted to
4534 implement this, we found no way to make it work reliably. The problem
4535 is that output operands might need reloading, which would result in
4536 additional following ``store'' instructions. On most machines, these
4537 instructions would alter the condition code before there was time to
4538 test it. This problem doesn't arise for ordinary ``test'' and
4539 ``compare'' instructions because they don't have any output operands.
4541 For reasons similar to those described above, it is not possible to give
4542 an assembler instruction access to the condition code left by previous
4545 If you are writing a header file that should be includable in ISO C
4546 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
4549 @subsection Size of an @code{asm}
4551 Some targets require that GCC track the size of each instruction used in
4552 order to generate correct code. Because the final length of an
4553 @code{asm} is only known by the assembler, GCC must make an estimate as
4554 to how big it will be. The estimate is formed by counting the number of
4555 statements in the pattern of the @code{asm} and multiplying that by the
4556 length of the longest instruction on that processor. Statements in the
4557 @code{asm} are identified by newline characters and whatever statement
4558 separator characters are supported by the assembler; on most processors
4559 this is the `@code{;}' character.
4561 Normally, GCC's estimate is perfectly adequate to ensure that correct
4562 code is generated, but it is possible to confuse the compiler if you use
4563 pseudo instructions or assembler macros that expand into multiple real
4564 instructions or if you use assembler directives that expand to more
4565 space in the object file than would be needed for a single instruction.
4566 If this happens then the assembler will produce a diagnostic saying that
4567 a label is unreachable.
4569 @subsection i386 floating point asm operands
4571 There are several rules on the usage of stack-like regs in
4572 asm_operands insns. These rules apply only to the operands that are
4577 Given a set of input regs that die in an asm_operands, it is
4578 necessary to know which are implicitly popped by the asm, and
4579 which must be explicitly popped by gcc.
4581 An input reg that is implicitly popped by the asm must be
4582 explicitly clobbered, unless it is constrained to match an
4586 For any input reg that is implicitly popped by an asm, it is
4587 necessary to know how to adjust the stack to compensate for the pop.
4588 If any non-popped input is closer to the top of the reg-stack than
4589 the implicitly popped reg, it would not be possible to know what the
4590 stack looked like---it's not clear how the rest of the stack ``slides
4593 All implicitly popped input regs must be closer to the top of
4594 the reg-stack than any input that is not implicitly popped.
4596 It is possible that if an input dies in an insn, reload might
4597 use the input reg for an output reload. Consider this example:
4600 asm ("foo" : "=t" (a) : "f" (b));
4603 This asm says that input B is not popped by the asm, and that
4604 the asm pushes a result onto the reg-stack, i.e., the stack is one
4605 deeper after the asm than it was before. But, it is possible that
4606 reload will think that it can use the same reg for both the input and
4607 the output, if input B dies in this insn.
4609 If any input operand uses the @code{f} constraint, all output reg
4610 constraints must use the @code{&} earlyclobber.
4612 The asm above would be written as
4615 asm ("foo" : "=&t" (a) : "f" (b));
4619 Some operands need to be in particular places on the stack. All
4620 output operands fall in this category---there is no other way to
4621 know which regs the outputs appear in unless the user indicates
4622 this in the constraints.
4624 Output operands must specifically indicate which reg an output
4625 appears in after an asm. @code{=f} is not allowed: the operand
4626 constraints must select a class with a single reg.
4629 Output operands may not be ``inserted'' between existing stack regs.
4630 Since no 387 opcode uses a read/write operand, all output operands
4631 are dead before the asm_operands, and are pushed by the asm_operands.
4632 It makes no sense to push anywhere but the top of the reg-stack.
4634 Output operands must start at the top of the reg-stack: output
4635 operands may not ``skip'' a reg.
4638 Some asm statements may need extra stack space for internal
4639 calculations. This can be guaranteed by clobbering stack registers
4640 unrelated to the inputs and outputs.
4644 Here are a couple of reasonable asms to want to write. This asm
4645 takes one input, which is internally popped, and produces two outputs.
4648 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4651 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4652 and replaces them with one output. The user must code the @code{st(1)}
4653 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4656 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4662 @section Controlling Names Used in Assembler Code
4663 @cindex assembler names for identifiers
4664 @cindex names used in assembler code
4665 @cindex identifiers, names in assembler code
4667 You can specify the name to be used in the assembler code for a C
4668 function or variable by writing the @code{asm} (or @code{__asm__})
4669 keyword after the declarator as follows:
4672 int foo asm ("myfoo") = 2;
4676 This specifies that the name to be used for the variable @code{foo} in
4677 the assembler code should be @samp{myfoo} rather than the usual
4680 On systems where an underscore is normally prepended to the name of a C
4681 function or variable, this feature allows you to define names for the
4682 linker that do not start with an underscore.
4684 It does not make sense to use this feature with a non-static local
4685 variable since such variables do not have assembler names. If you are
4686 trying to put the variable in a particular register, see @ref{Explicit
4687 Reg Vars}. GCC presently accepts such code with a warning, but will
4688 probably be changed to issue an error, rather than a warning, in the
4691 You cannot use @code{asm} in this way in a function @emph{definition}; but
4692 you can get the same effect by writing a declaration for the function
4693 before its definition and putting @code{asm} there, like this:
4696 extern func () asm ("FUNC");
4703 It is up to you to make sure that the assembler names you choose do not
4704 conflict with any other assembler symbols. Also, you must not use a
4705 register name; that would produce completely invalid assembler code. GCC
4706 does not as yet have the ability to store static variables in registers.
4707 Perhaps that will be added.
4709 @node Explicit Reg Vars
4710 @section Variables in Specified Registers
4711 @cindex explicit register variables
4712 @cindex variables in specified registers
4713 @cindex specified registers
4714 @cindex registers, global allocation
4716 GNU C allows you to put a few global variables into specified hardware
4717 registers. You can also specify the register in which an ordinary
4718 register variable should be allocated.
4722 Global register variables reserve registers throughout the program.
4723 This may be useful in programs such as programming language
4724 interpreters which have a couple of global variables that are accessed
4728 Local register variables in specific registers do not reserve the
4729 registers, except at the point where they are used as input or output
4730 operands in an @code{asm} statement and the @code{asm} statement itself is
4731 not deleted. The compiler's data flow analysis is capable of determining
4732 where the specified registers contain live values, and where they are
4733 available for other uses. Stores into local register variables may be deleted
4734 when they appear to be dead according to dataflow analysis. References
4735 to local register variables may be deleted or moved or simplified.
4737 These local variables are sometimes convenient for use with the extended
4738 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4739 output of the assembler instruction directly into a particular register.
4740 (This will work provided the register you specify fits the constraints
4741 specified for that operand in the @code{asm}.)
4749 @node Global Reg Vars
4750 @subsection Defining Global Register Variables
4751 @cindex global register variables
4752 @cindex registers, global variables in
4754 You can define a global register variable in GNU C like this:
4757 register int *foo asm ("a5");
4761 Here @code{a5} is the name of the register which should be used. Choose a
4762 register which is normally saved and restored by function calls on your
4763 machine, so that library routines will not clobber it.
4765 Naturally the register name is cpu-dependent, so you would need to
4766 conditionalize your program according to cpu type. The register
4767 @code{a5} would be a good choice on a 68000 for a variable of pointer
4768 type. On machines with register windows, be sure to choose a ``global''
4769 register that is not affected magically by the function call mechanism.
4771 In addition, operating systems on one type of cpu may differ in how they
4772 name the registers; then you would need additional conditionals. For
4773 example, some 68000 operating systems call this register @code{%a5}.
4775 Eventually there may be a way of asking the compiler to choose a register
4776 automatically, but first we need to figure out how it should choose and
4777 how to enable you to guide the choice. No solution is evident.
4779 Defining a global register variable in a certain register reserves that
4780 register entirely for this use, at least within the current compilation.
4781 The register will not be allocated for any other purpose in the functions
4782 in the current compilation. The register will not be saved and restored by
4783 these functions. Stores into this register are never deleted even if they
4784 would appear to be dead, but references may be deleted or moved or
4787 It is not safe to access the global register variables from signal
4788 handlers, or from more than one thread of control, because the system
4789 library routines may temporarily use the register for other things (unless
4790 you recompile them specially for the task at hand).
4792 @cindex @code{qsort}, and global register variables
4793 It is not safe for one function that uses a global register variable to
4794 call another such function @code{foo} by way of a third function
4795 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4796 different source file in which the variable wasn't declared). This is
4797 because @code{lose} might save the register and put some other value there.
4798 For example, you can't expect a global register variable to be available in
4799 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4800 might have put something else in that register. (If you are prepared to
4801 recompile @code{qsort} with the same global register variable, you can
4802 solve this problem.)
4804 If you want to recompile @code{qsort} or other source files which do not
4805 actually use your global register variable, so that they will not use that
4806 register for any other purpose, then it suffices to specify the compiler
4807 option @option{-ffixed-@var{reg}}. You need not actually add a global
4808 register declaration to their source code.
4810 A function which can alter the value of a global register variable cannot
4811 safely be called from a function compiled without this variable, because it
4812 could clobber the value the caller expects to find there on return.
4813 Therefore, the function which is the entry point into the part of the
4814 program that uses the global register variable must explicitly save and
4815 restore the value which belongs to its caller.
4817 @cindex register variable after @code{longjmp}
4818 @cindex global register after @code{longjmp}
4819 @cindex value after @code{longjmp}
4822 On most machines, @code{longjmp} will restore to each global register
4823 variable the value it had at the time of the @code{setjmp}. On some
4824 machines, however, @code{longjmp} will not change the value of global
4825 register variables. To be portable, the function that called @code{setjmp}
4826 should make other arrangements to save the values of the global register
4827 variables, and to restore them in a @code{longjmp}. This way, the same
4828 thing will happen regardless of what @code{longjmp} does.
4830 All global register variable declarations must precede all function
4831 definitions. If such a declaration could appear after function
4832 definitions, the declaration would be too late to prevent the register from
4833 being used for other purposes in the preceding functions.
4835 Global register variables may not have initial values, because an
4836 executable file has no means to supply initial contents for a register.
4838 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4839 registers, but certain library functions, such as @code{getwd}, as well
4840 as the subroutines for division and remainder, modify g3 and g4. g1 and
4841 g2 are local temporaries.
4843 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4844 Of course, it will not do to use more than a few of those.
4846 @node Local Reg Vars
4847 @subsection Specifying Registers for Local Variables
4848 @cindex local variables, specifying registers
4849 @cindex specifying registers for local variables
4850 @cindex registers for local variables
4852 You can define a local register variable with a specified register
4856 register int *foo asm ("a5");
4860 Here @code{a5} is the name of the register which should be used. Note
4861 that this is the same syntax used for defining global register
4862 variables, but for a local variable it would appear within a function.
4864 Naturally the register name is cpu-dependent, but this is not a
4865 problem, since specific registers are most often useful with explicit
4866 assembler instructions (@pxref{Extended Asm}). Both of these things
4867 generally require that you conditionalize your program according to
4870 In addition, operating systems on one type of cpu may differ in how they
4871 name the registers; then you would need additional conditionals. For
4872 example, some 68000 operating systems call this register @code{%a5}.
4874 Defining such a register variable does not reserve the register; it
4875 remains available for other uses in places where flow control determines
4876 the variable's value is not live.
4878 This option does not guarantee that GCC will generate code that has
4879 this variable in the register you specify at all times. You may not
4880 code an explicit reference to this register in the @emph{assembler
4881 instruction template} part of an @code{asm} statement and assume it will
4882 always refer to this variable. However, using the variable as an
4883 @code{asm} @emph{operand} guarantees that the specified register is used
4886 Stores into local register variables may be deleted when they appear to be dead
4887 according to dataflow analysis. References to local register variables may
4888 be deleted or moved or simplified.
4890 As for global register variables, it's recommended that you choose a
4891 register which is normally saved and restored by function calls on
4892 your machine, so that library routines will not clobber it. A common
4893 pitfall is to initialize multiple call-clobbered registers with
4894 arbitrary expressions, where a function call or library call for an
4895 arithmetic operator will overwrite a register value from a previous
4896 assignment, for example @code{r0} below:
4898 register int *p1 asm ("r0") = @dots{};
4899 register int *p2 asm ("r1") = @dots{};
4901 In those cases, a solution is to use a temporary variable for
4902 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
4904 @node Alternate Keywords
4905 @section Alternate Keywords
4906 @cindex alternate keywords
4907 @cindex keywords, alternate
4909 @option{-ansi} and the various @option{-std} options disable certain
4910 keywords. This causes trouble when you want to use GNU C extensions, or
4911 a general-purpose header file that should be usable by all programs,
4912 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4913 @code{inline} are not available in programs compiled with
4914 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4915 program compiled with @option{-std=c99}). The ISO C99 keyword
4916 @code{restrict} is only available when @option{-std=gnu99} (which will
4917 eventually be the default) or @option{-std=c99} (or the equivalent
4918 @option{-std=iso9899:1999}) is used.
4920 The way to solve these problems is to put @samp{__} at the beginning and
4921 end of each problematical keyword. For example, use @code{__asm__}
4922 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4924 Other C compilers won't accept these alternative keywords; if you want to
4925 compile with another compiler, you can define the alternate keywords as
4926 macros to replace them with the customary keywords. It looks like this:
4934 @findex __extension__
4936 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4938 prevent such warnings within one expression by writing
4939 @code{__extension__} before the expression. @code{__extension__} has no
4940 effect aside from this.
4942 @node Incomplete Enums
4943 @section Incomplete @code{enum} Types
4945 You can define an @code{enum} tag without specifying its possible values.
4946 This results in an incomplete type, much like what you get if you write
4947 @code{struct foo} without describing the elements. A later declaration
4948 which does specify the possible values completes the type.
4950 You can't allocate variables or storage using the type while it is
4951 incomplete. However, you can work with pointers to that type.
4953 This extension may not be very useful, but it makes the handling of
4954 @code{enum} more consistent with the way @code{struct} and @code{union}
4957 This extension is not supported by GNU C++.
4959 @node Function Names
4960 @section Function Names as Strings
4961 @cindex @code{__func__} identifier
4962 @cindex @code{__FUNCTION__} identifier
4963 @cindex @code{__PRETTY_FUNCTION__} identifier
4965 GCC provides three magic variables which hold the name of the current
4966 function, as a string. The first of these is @code{__func__}, which
4967 is part of the C99 standard:
4970 The identifier @code{__func__} is implicitly declared by the translator
4971 as if, immediately following the opening brace of each function
4972 definition, the declaration
4975 static const char __func__[] = "function-name";
4978 appeared, where function-name is the name of the lexically-enclosing
4979 function. This name is the unadorned name of the function.
4982 @code{__FUNCTION__} is another name for @code{__func__}. Older
4983 versions of GCC recognize only this name. However, it is not
4984 standardized. For maximum portability, we recommend you use
4985 @code{__func__}, but provide a fallback definition with the
4989 #if __STDC_VERSION__ < 199901L
4991 # define __func__ __FUNCTION__
4993 # define __func__ "<unknown>"
4998 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4999 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
5000 the type signature of the function as well as its bare name. For
5001 example, this program:
5005 extern int printf (char *, ...);
5012 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
5013 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
5031 __PRETTY_FUNCTION__ = void a::sub(int)
5034 These identifiers are not preprocessor macros. In GCC 3.3 and
5035 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
5036 were treated as string literals; they could be used to initialize
5037 @code{char} arrays, and they could be concatenated with other string
5038 literals. GCC 3.4 and later treat them as variables, like
5039 @code{__func__}. In C++, @code{__FUNCTION__} and
5040 @code{__PRETTY_FUNCTION__} have always been variables.
5042 @node Return Address
5043 @section Getting the Return or Frame Address of a Function
5045 These functions may be used to get information about the callers of a
5048 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
5049 This function returns the return address of the current function, or of
5050 one of its callers. The @var{level} argument is number of frames to
5051 scan up the call stack. A value of @code{0} yields the return address
5052 of the current function, a value of @code{1} yields the return address
5053 of the caller of the current function, and so forth. When inlining
5054 the expected behavior is that the function will return the address of
5055 the function that will be returned to. To work around this behavior use
5056 the @code{noinline} function attribute.
5058 The @var{level} argument must be a constant integer.
5060 On some machines it may be impossible to determine the return address of
5061 any function other than the current one; in such cases, or when the top
5062 of the stack has been reached, this function will return @code{0} or a
5063 random value. In addition, @code{__builtin_frame_address} may be used
5064 to determine if the top of the stack has been reached.
5066 This function should only be used with a nonzero argument for debugging
5070 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
5071 This function is similar to @code{__builtin_return_address}, but it
5072 returns the address of the function frame rather than the return address
5073 of the function. Calling @code{__builtin_frame_address} with a value of
5074 @code{0} yields the frame address of the current function, a value of
5075 @code{1} yields the frame address of the caller of the current function,
5078 The frame is the area on the stack which holds local variables and saved
5079 registers. The frame address is normally the address of the first word
5080 pushed on to the stack by the function. However, the exact definition
5081 depends upon the processor and the calling convention. If the processor
5082 has a dedicated frame pointer register, and the function has a frame,
5083 then @code{__builtin_frame_address} will return the value of the frame
5086 On some machines it may be impossible to determine the frame address of
5087 any function other than the current one; in such cases, or when the top
5088 of the stack has been reached, this function will return @code{0} if
5089 the first frame pointer is properly initialized by the startup code.
5091 This function should only be used with a nonzero argument for debugging
5095 @node Vector Extensions
5096 @section Using vector instructions through built-in functions
5098 On some targets, the instruction set contains SIMD vector instructions that
5099 operate on multiple values contained in one large register at the same time.
5100 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
5103 The first step in using these extensions is to provide the necessary data
5104 types. This should be done using an appropriate @code{typedef}:
5107 typedef int v4si __attribute__ ((vector_size (16)));
5110 The @code{int} type specifies the base type, while the attribute specifies
5111 the vector size for the variable, measured in bytes. For example, the
5112 declaration above causes the compiler to set the mode for the @code{v4si}
5113 type to be 16 bytes wide and divided into @code{int} sized units. For
5114 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
5115 corresponding mode of @code{foo} will be @acronym{V4SI}.
5117 The @code{vector_size} attribute is only applicable to integral and
5118 float scalars, although arrays, pointers, and function return values
5119 are allowed in conjunction with this construct.
5121 All the basic integer types can be used as base types, both as signed
5122 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
5123 @code{long long}. In addition, @code{float} and @code{double} can be
5124 used to build floating-point vector types.
5126 Specifying a combination that is not valid for the current architecture
5127 will cause GCC to synthesize the instructions using a narrower mode.
5128 For example, if you specify a variable of type @code{V4SI} and your
5129 architecture does not allow for this specific SIMD type, GCC will
5130 produce code that uses 4 @code{SIs}.
5132 The types defined in this manner can be used with a subset of normal C
5133 operations. Currently, GCC will allow using the following operators
5134 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
5136 The operations behave like C++ @code{valarrays}. Addition is defined as
5137 the addition of the corresponding elements of the operands. For
5138 example, in the code below, each of the 4 elements in @var{a} will be
5139 added to the corresponding 4 elements in @var{b} and the resulting
5140 vector will be stored in @var{c}.
5143 typedef int v4si __attribute__ ((vector_size (16)));
5150 Subtraction, multiplication, division, and the logical operations
5151 operate in a similar manner. Likewise, the result of using the unary
5152 minus or complement operators on a vector type is a vector whose
5153 elements are the negative or complemented values of the corresponding
5154 elements in the operand.
5156 You can declare variables and use them in function calls and returns, as
5157 well as in assignments and some casts. You can specify a vector type as
5158 a return type for a function. Vector types can also be used as function
5159 arguments. It is possible to cast from one vector type to another,
5160 provided they are of the same size (in fact, you can also cast vectors
5161 to and from other datatypes of the same size).
5163 You cannot operate between vectors of different lengths or different
5164 signedness without a cast.
5166 A port that supports hardware vector operations, usually provides a set
5167 of built-in functions that can be used to operate on vectors. For
5168 example, a function to add two vectors and multiply the result by a
5169 third could look like this:
5172 v4si f (v4si a, v4si b, v4si c)
5174 v4si tmp = __builtin_addv4si (a, b);
5175 return __builtin_mulv4si (tmp, c);
5182 @findex __builtin_offsetof
5184 GCC implements for both C and C++ a syntactic extension to implement
5185 the @code{offsetof} macro.
5189 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
5191 offsetof_member_designator:
5193 | offsetof_member_designator "." @code{identifier}
5194 | offsetof_member_designator "[" @code{expr} "]"
5197 This extension is sufficient such that
5200 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
5203 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
5204 may be dependent. In either case, @var{member} may consist of a single
5205 identifier, or a sequence of member accesses and array references.
5207 @node Atomic Builtins
5208 @section Built-in functions for atomic memory access
5210 The following builtins are intended to be compatible with those described
5211 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
5212 section 7.4. As such, they depart from the normal GCC practice of using
5213 the ``__builtin_'' prefix, and further that they are overloaded such that
5214 they work on multiple types.
5216 The definition given in the Intel documentation allows only for the use of
5217 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
5218 counterparts. GCC will allow any integral scalar or pointer type that is
5219 1, 2, 4 or 8 bytes in length.
5221 Not all operations are supported by all target processors. If a particular
5222 operation cannot be implemented on the target processor, a warning will be
5223 generated and a call an external function will be generated. The external
5224 function will carry the same name as the builtin, with an additional suffix
5225 @samp{_@var{n}} where @var{n} is the size of the data type.
5227 @c ??? Should we have a mechanism to suppress this warning? This is almost
5228 @c useful for implementing the operation under the control of an external
5231 In most cases, these builtins are considered a @dfn{full barrier}. That is,
5232 no memory operand will be moved across the operation, either forward or
5233 backward. Further, instructions will be issued as necessary to prevent the
5234 processor from speculating loads across the operation and from queuing stores
5235 after the operation.
5237 All of the routines are are described in the Intel documentation to take
5238 ``an optional list of variables protected by the memory barrier''. It's
5239 not clear what is meant by that; it could mean that @emph{only} the
5240 following variables are protected, or it could mean that these variables
5241 should in addition be protected. At present GCC ignores this list and
5242 protects all variables which are globally accessible. If in the future
5243 we make some use of this list, an empty list will continue to mean all
5244 globally accessible variables.
5247 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
5248 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
5249 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
5250 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
5251 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
5252 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
5253 @findex __sync_fetch_and_add
5254 @findex __sync_fetch_and_sub
5255 @findex __sync_fetch_and_or
5256 @findex __sync_fetch_and_and
5257 @findex __sync_fetch_and_xor
5258 @findex __sync_fetch_and_nand
5259 These builtins perform the operation suggested by the name, and
5260 returns the value that had previously been in memory. That is,
5263 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
5264 @{ tmp = *ptr; *ptr = ~tmp & value; return tmp; @} // nand
5267 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
5268 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
5269 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
5270 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
5271 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
5272 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
5273 @findex __sync_add_and_fetch
5274 @findex __sync_sub_and_fetch
5275 @findex __sync_or_and_fetch
5276 @findex __sync_and_and_fetch
5277 @findex __sync_xor_and_fetch
5278 @findex __sync_nand_and_fetch
5279 These builtins perform the operation suggested by the name, and
5280 return the new value. That is,
5283 @{ *ptr @var{op}= value; return *ptr; @}
5284 @{ *ptr = ~*ptr & value; return *ptr; @} // nand
5287 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5288 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
5289 @findex __sync_bool_compare_and_swap
5290 @findex __sync_val_compare_and_swap
5291 These builtins perform an atomic compare and swap. That is, if the current
5292 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
5295 The ``bool'' version returns true if the comparison is successful and
5296 @var{newval} was written. The ``val'' version returns the contents
5297 of @code{*@var{ptr}} before the operation.
5299 @item __sync_synchronize (...)
5300 @findex __sync_synchronize
5301 This builtin issues a full memory barrier.
5303 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
5304 @findex __sync_lock_test_and_set
5305 This builtin, as described by Intel, is not a traditional test-and-set
5306 operation, but rather an atomic exchange operation. It writes @var{value}
5307 into @code{*@var{ptr}}, and returns the previous contents of
5310 Many targets have only minimal support for such locks, and do not support
5311 a full exchange operation. In this case, a target may support reduced
5312 functionality here by which the @emph{only} valid value to store is the
5313 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
5314 is implementation defined.
5316 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
5317 This means that references after the builtin cannot move to (or be
5318 speculated to) before the builtin, but previous memory stores may not
5319 be globally visible yet, and previous memory loads may not yet be
5322 @item void __sync_lock_release (@var{type} *ptr, ...)
5323 @findex __sync_lock_release
5324 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
5325 Normally this means writing the constant 0 to @code{*@var{ptr}}.
5327 This builtin is not a full barrier, but rather a @dfn{release barrier}.
5328 This means that all previous memory stores are globally visible, and all
5329 previous memory loads have been satisfied, but following memory reads
5330 are not prevented from being speculated to before the barrier.
5333 @node Object Size Checking
5334 @section Object Size Checking Builtins
5335 @findex __builtin_object_size
5336 @findex __builtin___memcpy_chk
5337 @findex __builtin___mempcpy_chk
5338 @findex __builtin___memmove_chk
5339 @findex __builtin___memset_chk
5340 @findex __builtin___strcpy_chk
5341 @findex __builtin___stpcpy_chk
5342 @findex __builtin___strncpy_chk
5343 @findex __builtin___strcat_chk
5344 @findex __builtin___strncat_chk
5345 @findex __builtin___sprintf_chk
5346 @findex __builtin___snprintf_chk
5347 @findex __builtin___vsprintf_chk
5348 @findex __builtin___vsnprintf_chk
5349 @findex __builtin___printf_chk
5350 @findex __builtin___vprintf_chk
5351 @findex __builtin___fprintf_chk
5352 @findex __builtin___vfprintf_chk
5354 GCC implements a limited buffer overflow protection mechanism
5355 that can prevent some buffer overflow attacks.
5357 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
5358 is a built-in construct that returns a constant number of bytes from
5359 @var{ptr} to the end of the object @var{ptr} pointer points to
5360 (if known at compile time). @code{__builtin_object_size} never evaluates
5361 its arguments for side-effects. If there are any side-effects in them, it
5362 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5363 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
5364 point to and all of them are known at compile time, the returned number
5365 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
5366 0 and minimum if nonzero. If it is not possible to determine which objects
5367 @var{ptr} points to at compile time, @code{__builtin_object_size} should
5368 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5369 for @var{type} 2 or 3.
5371 @var{type} is an integer constant from 0 to 3. If the least significant
5372 bit is clear, objects are whole variables, if it is set, a closest
5373 surrounding subobject is considered the object a pointer points to.
5374 The second bit determines if maximum or minimum of remaining bytes
5378 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
5379 char *p = &var.buf1[1], *q = &var.b;
5381 /* Here the object p points to is var. */
5382 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
5383 /* The subobject p points to is var.buf1. */
5384 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
5385 /* The object q points to is var. */
5386 assert (__builtin_object_size (q, 0)
5387 == (char *) (&var + 1) - (char *) &var.b);
5388 /* The subobject q points to is var.b. */
5389 assert (__builtin_object_size (q, 1) == sizeof (var.b));
5393 There are built-in functions added for many common string operation
5394 functions, e.g. for @code{memcpy} @code{__builtin___memcpy_chk}
5395 built-in is provided. This built-in has an additional last argument,
5396 which is the number of bytes remaining in object the @var{dest}
5397 argument points to or @code{(size_t) -1} if the size is not known.
5399 The built-in functions are optimized into the normal string functions
5400 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
5401 it is known at compile time that the destination object will not
5402 be overflown. If the compiler can determine at compile time the
5403 object will be always overflown, it issues a warning.
5405 The intended use can be e.g.
5409 #define bos0(dest) __builtin_object_size (dest, 0)
5410 #define memcpy(dest, src, n) \
5411 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
5415 /* It is unknown what object p points to, so this is optimized
5416 into plain memcpy - no checking is possible. */
5417 memcpy (p, "abcde", n);
5418 /* Destination is known and length too. It is known at compile
5419 time there will be no overflow. */
5420 memcpy (&buf[5], "abcde", 5);
5421 /* Destination is known, but the length is not known at compile time.
5422 This will result in __memcpy_chk call that can check for overflow
5424 memcpy (&buf[5], "abcde", n);
5425 /* Destination is known and it is known at compile time there will
5426 be overflow. There will be a warning and __memcpy_chk call that
5427 will abort the program at runtime. */
5428 memcpy (&buf[6], "abcde", 5);
5431 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
5432 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
5433 @code{strcat} and @code{strncat}.
5435 There are also checking built-in functions for formatted output functions.
5437 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
5438 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5439 const char *fmt, ...);
5440 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
5442 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5443 const char *fmt, va_list ap);
5446 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
5447 etc. functions and can contain implementation specific flags on what
5448 additional security measures the checking function might take, such as
5449 handling @code{%n} differently.
5451 The @var{os} argument is the object size @var{s} points to, like in the
5452 other built-in functions. There is a small difference in the behavior
5453 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
5454 optimized into the non-checking functions only if @var{flag} is 0, otherwise
5455 the checking function is called with @var{os} argument set to
5458 In addition to this, there are checking built-in functions
5459 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
5460 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
5461 These have just one additional argument, @var{flag}, right before
5462 format string @var{fmt}. If the compiler is able to optimize them to
5463 @code{fputc} etc. functions, it will, otherwise the checking function
5464 should be called and the @var{flag} argument passed to it.
5466 @node Other Builtins
5467 @section Other built-in functions provided by GCC
5468 @cindex built-in functions
5469 @findex __builtin_isfinite
5470 @findex __builtin_isnormal
5471 @findex __builtin_isgreater
5472 @findex __builtin_isgreaterequal
5473 @findex __builtin_isless
5474 @findex __builtin_islessequal
5475 @findex __builtin_islessgreater
5476 @findex __builtin_isunordered
5477 @findex __builtin_powi
5478 @findex __builtin_powif
5479 @findex __builtin_powil
5637 @findex fprintf_unlocked
5639 @findex fputs_unlocked
5756 @findex printf_unlocked
5788 @findex significandf
5789 @findex significandl
5860 GCC provides a large number of built-in functions other than the ones
5861 mentioned above. Some of these are for internal use in the processing
5862 of exceptions or variable-length argument lists and will not be
5863 documented here because they may change from time to time; we do not
5864 recommend general use of these functions.
5866 The remaining functions are provided for optimization purposes.
5868 @opindex fno-builtin
5869 GCC includes built-in versions of many of the functions in the standard
5870 C library. The versions prefixed with @code{__builtin_} will always be
5871 treated as having the same meaning as the C library function even if you
5872 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
5873 Many of these functions are only optimized in certain cases; if they are
5874 not optimized in a particular case, a call to the library function will
5879 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
5880 @option{-std=c99}), the functions
5881 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
5882 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
5883 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
5884 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked},
5885 @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma},
5886 @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext},
5887 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
5888 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
5889 @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy},
5890 @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked},
5891 @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb},
5892 @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32},
5893 @code{signbitd64}, @code{signbitd128}, @code{significandf},
5894 @code{significandl}, @code{significand}, @code{sincosf},
5895 @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy},
5896 @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp},
5897 @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0},
5898 @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and
5900 may be handled as built-in functions.
5901 All these functions have corresponding versions
5902 prefixed with @code{__builtin_}, which may be used even in strict C89
5905 The ISO C99 functions
5906 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
5907 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
5908 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
5909 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
5910 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
5911 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
5912 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
5913 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
5914 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
5915 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
5916 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
5917 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
5918 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
5919 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
5920 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
5921 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
5922 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
5923 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
5924 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
5925 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
5926 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
5927 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
5928 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
5929 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
5930 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
5931 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
5932 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
5933 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
5934 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
5935 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
5936 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
5937 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
5938 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
5939 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
5940 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
5941 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
5942 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
5943 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
5944 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
5945 are handled as built-in functions
5946 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5948 There are also built-in versions of the ISO C99 functions
5949 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5950 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5951 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5952 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5953 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5954 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5955 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5956 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5957 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5958 that are recognized in any mode since ISO C90 reserves these names for
5959 the purpose to which ISO C99 puts them. All these functions have
5960 corresponding versions prefixed with @code{__builtin_}.
5962 The ISO C94 functions
5963 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
5964 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
5965 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
5967 are handled as built-in functions
5968 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5970 The ISO C90 functions
5971 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5972 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5973 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5974 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
5975 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
5976 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
5977 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
5978 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
5979 @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy},
5980 @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar},
5981 @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf},
5982 @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat},
5983 @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn},
5984 @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy},
5985 @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr},
5986 @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf}
5987 are all recognized as built-in functions unless
5988 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5989 is specified for an individual function). All of these functions have
5990 corresponding versions prefixed with @code{__builtin_}.
5992 GCC provides built-in versions of the ISO C99 floating point comparison
5993 macros that avoid raising exceptions for unordered operands. They have
5994 the same names as the standard macros ( @code{isgreater},
5995 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5996 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5997 prefixed. We intend for a library implementor to be able to simply
5998 @code{#define} each standard macro to its built-in equivalent.
5999 In the same fashion, GCC provides @code{isfinite} and @code{isnormal}
6000 built-ins used with @code{__builtin_} prefixed.
6002 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
6004 You can use the built-in function @code{__builtin_types_compatible_p} to
6005 determine whether two types are the same.
6007 This built-in function returns 1 if the unqualified versions of the
6008 types @var{type1} and @var{type2} (which are types, not expressions) are
6009 compatible, 0 otherwise. The result of this built-in function can be
6010 used in integer constant expressions.
6012 This built-in function ignores top level qualifiers (e.g., @code{const},
6013 @code{volatile}). For example, @code{int} is equivalent to @code{const
6016 The type @code{int[]} and @code{int[5]} are compatible. On the other
6017 hand, @code{int} and @code{char *} are not compatible, even if the size
6018 of their types, on the particular architecture are the same. Also, the
6019 amount of pointer indirection is taken into account when determining
6020 similarity. Consequently, @code{short *} is not similar to
6021 @code{short **}. Furthermore, two types that are typedefed are
6022 considered compatible if their underlying types are compatible.
6024 An @code{enum} type is not considered to be compatible with another
6025 @code{enum} type even if both are compatible with the same integer
6026 type; this is what the C standard specifies.
6027 For example, @code{enum @{foo, bar@}} is not similar to
6028 @code{enum @{hot, dog@}}.
6030 You would typically use this function in code whose execution varies
6031 depending on the arguments' types. For example:
6036 typeof (x) tmp = (x); \
6037 if (__builtin_types_compatible_p (typeof (x), long double)) \
6038 tmp = foo_long_double (tmp); \
6039 else if (__builtin_types_compatible_p (typeof (x), double)) \
6040 tmp = foo_double (tmp); \
6041 else if (__builtin_types_compatible_p (typeof (x), float)) \
6042 tmp = foo_float (tmp); \
6049 @emph{Note:} This construct is only available for C@.
6053 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
6055 You can use the built-in function @code{__builtin_choose_expr} to
6056 evaluate code depending on the value of a constant expression. This
6057 built-in function returns @var{exp1} if @var{const_exp}, which is a
6058 constant expression that must be able to be determined at compile time,
6059 is nonzero. Otherwise it returns 0.
6061 This built-in function is analogous to the @samp{? :} operator in C,
6062 except that the expression returned has its type unaltered by promotion
6063 rules. Also, the built-in function does not evaluate the expression
6064 that was not chosen. For example, if @var{const_exp} evaluates to true,
6065 @var{exp2} is not evaluated even if it has side-effects.
6067 This built-in function can return an lvalue if the chosen argument is an
6070 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
6071 type. Similarly, if @var{exp2} is returned, its return type is the same
6078 __builtin_choose_expr ( \
6079 __builtin_types_compatible_p (typeof (x), double), \
6081 __builtin_choose_expr ( \
6082 __builtin_types_compatible_p (typeof (x), float), \
6084 /* @r{The void expression results in a compile-time error} \
6085 @r{when assigning the result to something.} */ \
6089 @emph{Note:} This construct is only available for C@. Furthermore, the
6090 unused expression (@var{exp1} or @var{exp2} depending on the value of
6091 @var{const_exp}) may still generate syntax errors. This may change in
6096 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
6097 You can use the built-in function @code{__builtin_constant_p} to
6098 determine if a value is known to be constant at compile-time and hence
6099 that GCC can perform constant-folding on expressions involving that
6100 value. The argument of the function is the value to test. The function
6101 returns the integer 1 if the argument is known to be a compile-time
6102 constant and 0 if it is not known to be a compile-time constant. A
6103 return of 0 does not indicate that the value is @emph{not} a constant,
6104 but merely that GCC cannot prove it is a constant with the specified
6105 value of the @option{-O} option.
6107 You would typically use this function in an embedded application where
6108 memory was a critical resource. If you have some complex calculation,
6109 you may want it to be folded if it involves constants, but need to call
6110 a function if it does not. For example:
6113 #define Scale_Value(X) \
6114 (__builtin_constant_p (X) \
6115 ? ((X) * SCALE + OFFSET) : Scale (X))
6118 You may use this built-in function in either a macro or an inline
6119 function. However, if you use it in an inlined function and pass an
6120 argument of the function as the argument to the built-in, GCC will
6121 never return 1 when you call the inline function with a string constant
6122 or compound literal (@pxref{Compound Literals}) and will not return 1
6123 when you pass a constant numeric value to the inline function unless you
6124 specify the @option{-O} option.
6126 You may also use @code{__builtin_constant_p} in initializers for static
6127 data. For instance, you can write
6130 static const int table[] = @{
6131 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
6137 This is an acceptable initializer even if @var{EXPRESSION} is not a
6138 constant expression. GCC must be more conservative about evaluating the
6139 built-in in this case, because it has no opportunity to perform
6142 Previous versions of GCC did not accept this built-in in data
6143 initializers. The earliest version where it is completely safe is
6147 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
6148 @opindex fprofile-arcs
6149 You may use @code{__builtin_expect} to provide the compiler with
6150 branch prediction information. In general, you should prefer to
6151 use actual profile feedback for this (@option{-fprofile-arcs}), as
6152 programmers are notoriously bad at predicting how their programs
6153 actually perform. However, there are applications in which this
6154 data is hard to collect.
6156 The return value is the value of @var{exp}, which should be an integral
6157 expression. The semantics of the built-in are that it is expected that
6158 @var{exp} == @var{c}. For example:
6161 if (__builtin_expect (x, 0))
6166 would indicate that we do not expect to call @code{foo}, since
6167 we expect @code{x} to be zero. Since you are limited to integral
6168 expressions for @var{exp}, you should use constructions such as
6171 if (__builtin_expect (ptr != NULL, 1))
6176 when testing pointer or floating-point values.
6179 @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end})
6180 This function is used to flush the processor's instruction cache for
6181 the region of memory between @var{begin} inclusive and @var{end}
6182 exclusive. Some targets require that the instruction cache be
6183 flushed, after modifying memory containing code, in order to obtain
6184 deterministic behavior.
6186 If the target does not require instruction cache flushes,
6187 @code{__builtin___clear_cache} has no effect. Otherwise either
6188 instructions are emitted in-line to clear the instruction cache or a
6189 call to the @code{__clear_cache} function in libgcc is made.
6192 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
6193 This function is used to minimize cache-miss latency by moving data into
6194 a cache before it is accessed.
6195 You can insert calls to @code{__builtin_prefetch} into code for which
6196 you know addresses of data in memory that is likely to be accessed soon.
6197 If the target supports them, data prefetch instructions will be generated.
6198 If the prefetch is done early enough before the access then the data will
6199 be in the cache by the time it is accessed.
6201 The value of @var{addr} is the address of the memory to prefetch.
6202 There are two optional arguments, @var{rw} and @var{locality}.
6203 The value of @var{rw} is a compile-time constant one or zero; one
6204 means that the prefetch is preparing for a write to the memory address
6205 and zero, the default, means that the prefetch is preparing for a read.
6206 The value @var{locality} must be a compile-time constant integer between
6207 zero and three. A value of zero means that the data has no temporal
6208 locality, so it need not be left in the cache after the access. A value
6209 of three means that the data has a high degree of temporal locality and
6210 should be left in all levels of cache possible. Values of one and two
6211 mean, respectively, a low or moderate degree of temporal locality. The
6215 for (i = 0; i < n; i++)
6218 __builtin_prefetch (&a[i+j], 1, 1);
6219 __builtin_prefetch (&b[i+j], 0, 1);
6224 Data prefetch does not generate faults if @var{addr} is invalid, but
6225 the address expression itself must be valid. For example, a prefetch
6226 of @code{p->next} will not fault if @code{p->next} is not a valid
6227 address, but evaluation will fault if @code{p} is not a valid address.
6229 If the target does not support data prefetch, the address expression
6230 is evaluated if it includes side effects but no other code is generated
6231 and GCC does not issue a warning.
6234 @deftypefn {Built-in Function} double __builtin_huge_val (void)
6235 Returns a positive infinity, if supported by the floating-point format,
6236 else @code{DBL_MAX}. This function is suitable for implementing the
6237 ISO C macro @code{HUGE_VAL}.
6240 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
6241 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
6244 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
6245 Similar to @code{__builtin_huge_val}, except the return
6246 type is @code{long double}.
6249 @deftypefn {Built-in Function} double __builtin_inf (void)
6250 Similar to @code{__builtin_huge_val}, except a warning is generated
6251 if the target floating-point format does not support infinities.
6254 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
6255 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
6258 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
6259 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
6262 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
6263 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
6266 @deftypefn {Built-in Function} float __builtin_inff (void)
6267 Similar to @code{__builtin_inf}, except the return type is @code{float}.
6268 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
6271 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
6272 Similar to @code{__builtin_inf}, except the return
6273 type is @code{long double}.
6276 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
6277 This is an implementation of the ISO C99 function @code{nan}.
6279 Since ISO C99 defines this function in terms of @code{strtod}, which we
6280 do not implement, a description of the parsing is in order. The string
6281 is parsed as by @code{strtol}; that is, the base is recognized by
6282 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
6283 in the significand such that the least significant bit of the number
6284 is at the least significant bit of the significand. The number is
6285 truncated to fit the significand field provided. The significand is
6286 forced to be a quiet NaN@.
6288 This function, if given a string literal all of which would have been
6289 consumed by strtol, is evaluated early enough that it is considered a
6290 compile-time constant.
6293 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
6294 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
6297 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
6298 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
6301 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
6302 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
6305 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
6306 Similar to @code{__builtin_nan}, except the return type is @code{float}.
6309 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
6310 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
6313 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
6314 Similar to @code{__builtin_nan}, except the significand is forced
6315 to be a signaling NaN@. The @code{nans} function is proposed by
6316 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
6319 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
6320 Similar to @code{__builtin_nans}, except the return type is @code{float}.
6323 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
6324 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
6327 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
6328 Returns one plus the index of the least significant 1-bit of @var{x}, or
6329 if @var{x} is zero, returns zero.
6332 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
6333 Returns the number of leading 0-bits in @var{x}, starting at the most
6334 significant bit position. If @var{x} is 0, the result is undefined.
6337 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
6338 Returns the number of trailing 0-bits in @var{x}, starting at the least
6339 significant bit position. If @var{x} is 0, the result is undefined.
6342 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
6343 Returns the number of 1-bits in @var{x}.
6346 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
6347 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
6351 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
6352 Similar to @code{__builtin_ffs}, except the argument type is
6353 @code{unsigned long}.
6356 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
6357 Similar to @code{__builtin_clz}, except the argument type is
6358 @code{unsigned long}.
6361 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
6362 Similar to @code{__builtin_ctz}, except the argument type is
6363 @code{unsigned long}.
6366 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
6367 Similar to @code{__builtin_popcount}, except the argument type is
6368 @code{unsigned long}.
6371 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
6372 Similar to @code{__builtin_parity}, except the argument type is
6373 @code{unsigned long}.
6376 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
6377 Similar to @code{__builtin_ffs}, except the argument type is
6378 @code{unsigned long long}.
6381 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
6382 Similar to @code{__builtin_clz}, except the argument type is
6383 @code{unsigned long long}.
6386 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
6387 Similar to @code{__builtin_ctz}, except the argument type is
6388 @code{unsigned long long}.
6391 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
6392 Similar to @code{__builtin_popcount}, except the argument type is
6393 @code{unsigned long long}.
6396 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
6397 Similar to @code{__builtin_parity}, except the argument type is
6398 @code{unsigned long long}.
6401 @deftypefn {Built-in Function} double __builtin_powi (double, int)
6402 Returns the first argument raised to the power of the second. Unlike the
6403 @code{pow} function no guarantees about precision and rounding are made.
6406 @deftypefn {Built-in Function} float __builtin_powif (float, int)
6407 Similar to @code{__builtin_powi}, except the argument and return types
6411 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
6412 Similar to @code{__builtin_powi}, except the argument and return types
6413 are @code{long double}.
6416 @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x)
6417 Returns @var{x} with the order of the bytes reversed; for example,
6418 @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means
6422 @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x)
6423 Similar to @code{__builtin_bswap32}, except the argument and return types
6427 @node Target Builtins
6428 @section Built-in Functions Specific to Particular Target Machines
6430 On some target machines, GCC supports many built-in functions specific
6431 to those machines. Generally these generate calls to specific machine
6432 instructions, but allow the compiler to schedule those calls.
6435 * Alpha Built-in Functions::
6436 * ARM iWMMXt Built-in Functions::
6437 * ARM NEON Intrinsics::
6438 * Blackfin Built-in Functions::
6439 * FR-V Built-in Functions::
6440 * X86 Built-in Functions::
6441 * MIPS DSP Built-in Functions::
6442 * MIPS Paired-Single Support::
6443 * PowerPC AltiVec Built-in Functions::
6444 * SPARC VIS Built-in Functions::
6445 * SPU Built-in Functions::
6448 @node Alpha Built-in Functions
6449 @subsection Alpha Built-in Functions
6451 These built-in functions are available for the Alpha family of
6452 processors, depending on the command-line switches used.
6454 The following built-in functions are always available. They
6455 all generate the machine instruction that is part of the name.
6458 long __builtin_alpha_implver (void)
6459 long __builtin_alpha_rpcc (void)
6460 long __builtin_alpha_amask (long)
6461 long __builtin_alpha_cmpbge (long, long)
6462 long __builtin_alpha_extbl (long, long)
6463 long __builtin_alpha_extwl (long, long)
6464 long __builtin_alpha_extll (long, long)
6465 long __builtin_alpha_extql (long, long)
6466 long __builtin_alpha_extwh (long, long)
6467 long __builtin_alpha_extlh (long, long)
6468 long __builtin_alpha_extqh (long, long)
6469 long __builtin_alpha_insbl (long, long)
6470 long __builtin_alpha_inswl (long, long)
6471 long __builtin_alpha_insll (long, long)
6472 long __builtin_alpha_insql (long, long)
6473 long __builtin_alpha_inswh (long, long)
6474 long __builtin_alpha_inslh (long, long)
6475 long __builtin_alpha_insqh (long, long)
6476 long __builtin_alpha_mskbl (long, long)
6477 long __builtin_alpha_mskwl (long, long)
6478 long __builtin_alpha_mskll (long, long)
6479 long __builtin_alpha_mskql (long, long)
6480 long __builtin_alpha_mskwh (long, long)
6481 long __builtin_alpha_msklh (long, long)
6482 long __builtin_alpha_mskqh (long, long)
6483 long __builtin_alpha_umulh (long, long)
6484 long __builtin_alpha_zap (long, long)
6485 long __builtin_alpha_zapnot (long, long)
6488 The following built-in functions are always with @option{-mmax}
6489 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
6490 later. They all generate the machine instruction that is part
6494 long __builtin_alpha_pklb (long)
6495 long __builtin_alpha_pkwb (long)
6496 long __builtin_alpha_unpkbl (long)
6497 long __builtin_alpha_unpkbw (long)
6498 long __builtin_alpha_minub8 (long, long)
6499 long __builtin_alpha_minsb8 (long, long)
6500 long __builtin_alpha_minuw4 (long, long)
6501 long __builtin_alpha_minsw4 (long, long)
6502 long __builtin_alpha_maxub8 (long, long)
6503 long __builtin_alpha_maxsb8 (long, long)
6504 long __builtin_alpha_maxuw4 (long, long)
6505 long __builtin_alpha_maxsw4 (long, long)
6506 long __builtin_alpha_perr (long, long)
6509 The following built-in functions are always with @option{-mcix}
6510 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
6511 later. They all generate the machine instruction that is part
6515 long __builtin_alpha_cttz (long)
6516 long __builtin_alpha_ctlz (long)
6517 long __builtin_alpha_ctpop (long)
6520 The following builtins are available on systems that use the OSF/1
6521 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
6522 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
6523 @code{rdval} and @code{wrval}.
6526 void *__builtin_thread_pointer (void)
6527 void __builtin_set_thread_pointer (void *)
6530 @node ARM iWMMXt Built-in Functions
6531 @subsection ARM iWMMXt Built-in Functions
6533 These built-in functions are available for the ARM family of
6534 processors when the @option{-mcpu=iwmmxt} switch is used:
6537 typedef int v2si __attribute__ ((vector_size (8)));
6538 typedef short v4hi __attribute__ ((vector_size (8)));
6539 typedef char v8qi __attribute__ ((vector_size (8)));
6541 int __builtin_arm_getwcx (int)
6542 void __builtin_arm_setwcx (int, int)
6543 int __builtin_arm_textrmsb (v8qi, int)
6544 int __builtin_arm_textrmsh (v4hi, int)
6545 int __builtin_arm_textrmsw (v2si, int)
6546 int __builtin_arm_textrmub (v8qi, int)
6547 int __builtin_arm_textrmuh (v4hi, int)
6548 int __builtin_arm_textrmuw (v2si, int)
6549 v8qi __builtin_arm_tinsrb (v8qi, int)
6550 v4hi __builtin_arm_tinsrh (v4hi, int)
6551 v2si __builtin_arm_tinsrw (v2si, int)
6552 long long __builtin_arm_tmia (long long, int, int)
6553 long long __builtin_arm_tmiabb (long long, int, int)
6554 long long __builtin_arm_tmiabt (long long, int, int)
6555 long long __builtin_arm_tmiaph (long long, int, int)
6556 long long __builtin_arm_tmiatb (long long, int, int)
6557 long long __builtin_arm_tmiatt (long long, int, int)
6558 int __builtin_arm_tmovmskb (v8qi)
6559 int __builtin_arm_tmovmskh (v4hi)
6560 int __builtin_arm_tmovmskw (v2si)
6561 long long __builtin_arm_waccb (v8qi)
6562 long long __builtin_arm_wacch (v4hi)
6563 long long __builtin_arm_waccw (v2si)
6564 v8qi __builtin_arm_waddb (v8qi, v8qi)
6565 v8qi __builtin_arm_waddbss (v8qi, v8qi)
6566 v8qi __builtin_arm_waddbus (v8qi, v8qi)
6567 v4hi __builtin_arm_waddh (v4hi, v4hi)
6568 v4hi __builtin_arm_waddhss (v4hi, v4hi)
6569 v4hi __builtin_arm_waddhus (v4hi, v4hi)
6570 v2si __builtin_arm_waddw (v2si, v2si)
6571 v2si __builtin_arm_waddwss (v2si, v2si)
6572 v2si __builtin_arm_waddwus (v2si, v2si)
6573 v8qi __builtin_arm_walign (v8qi, v8qi, int)
6574 long long __builtin_arm_wand(long long, long long)
6575 long long __builtin_arm_wandn (long long, long long)
6576 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
6577 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
6578 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
6579 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
6580 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
6581 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
6582 v2si __builtin_arm_wcmpeqw (v2si, v2si)
6583 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
6584 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
6585 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
6586 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
6587 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
6588 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
6589 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
6590 long long __builtin_arm_wmacsz (v4hi, v4hi)
6591 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
6592 long long __builtin_arm_wmacuz (v4hi, v4hi)
6593 v4hi __builtin_arm_wmadds (v4hi, v4hi)
6594 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
6595 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
6596 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
6597 v2si __builtin_arm_wmaxsw (v2si, v2si)
6598 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
6599 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
6600 v2si __builtin_arm_wmaxuw (v2si, v2si)
6601 v8qi __builtin_arm_wminsb (v8qi, v8qi)
6602 v4hi __builtin_arm_wminsh (v4hi, v4hi)
6603 v2si __builtin_arm_wminsw (v2si, v2si)
6604 v8qi __builtin_arm_wminub (v8qi, v8qi)
6605 v4hi __builtin_arm_wminuh (v4hi, v4hi)
6606 v2si __builtin_arm_wminuw (v2si, v2si)
6607 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
6608 v4hi __builtin_arm_wmulul (v4hi, v4hi)
6609 v4hi __builtin_arm_wmulum (v4hi, v4hi)
6610 long long __builtin_arm_wor (long long, long long)
6611 v2si __builtin_arm_wpackdss (long long, long long)
6612 v2si __builtin_arm_wpackdus (long long, long long)
6613 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
6614 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
6615 v4hi __builtin_arm_wpackwss (v2si, v2si)
6616 v4hi __builtin_arm_wpackwus (v2si, v2si)
6617 long long __builtin_arm_wrord (long long, long long)
6618 long long __builtin_arm_wrordi (long long, int)
6619 v4hi __builtin_arm_wrorh (v4hi, long long)
6620 v4hi __builtin_arm_wrorhi (v4hi, int)
6621 v2si __builtin_arm_wrorw (v2si, long long)
6622 v2si __builtin_arm_wrorwi (v2si, int)
6623 v2si __builtin_arm_wsadb (v8qi, v8qi)
6624 v2si __builtin_arm_wsadbz (v8qi, v8qi)
6625 v2si __builtin_arm_wsadh (v4hi, v4hi)
6626 v2si __builtin_arm_wsadhz (v4hi, v4hi)
6627 v4hi __builtin_arm_wshufh (v4hi, int)
6628 long long __builtin_arm_wslld (long long, long long)
6629 long long __builtin_arm_wslldi (long long, int)
6630 v4hi __builtin_arm_wsllh (v4hi, long long)
6631 v4hi __builtin_arm_wsllhi (v4hi, int)
6632 v2si __builtin_arm_wsllw (v2si, long long)
6633 v2si __builtin_arm_wsllwi (v2si, int)
6634 long long __builtin_arm_wsrad (long long, long long)
6635 long long __builtin_arm_wsradi (long long, int)
6636 v4hi __builtin_arm_wsrah (v4hi, long long)
6637 v4hi __builtin_arm_wsrahi (v4hi, int)
6638 v2si __builtin_arm_wsraw (v2si, long long)
6639 v2si __builtin_arm_wsrawi (v2si, int)
6640 long long __builtin_arm_wsrld (long long, long long)
6641 long long __builtin_arm_wsrldi (long long, int)
6642 v4hi __builtin_arm_wsrlh (v4hi, long long)
6643 v4hi __builtin_arm_wsrlhi (v4hi, int)
6644 v2si __builtin_arm_wsrlw (v2si, long long)
6645 v2si __builtin_arm_wsrlwi (v2si, int)
6646 v8qi __builtin_arm_wsubb (v8qi, v8qi)
6647 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
6648 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
6649 v4hi __builtin_arm_wsubh (v4hi, v4hi)
6650 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
6651 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
6652 v2si __builtin_arm_wsubw (v2si, v2si)
6653 v2si __builtin_arm_wsubwss (v2si, v2si)
6654 v2si __builtin_arm_wsubwus (v2si, v2si)
6655 v4hi __builtin_arm_wunpckehsb (v8qi)
6656 v2si __builtin_arm_wunpckehsh (v4hi)
6657 long long __builtin_arm_wunpckehsw (v2si)
6658 v4hi __builtin_arm_wunpckehub (v8qi)
6659 v2si __builtin_arm_wunpckehuh (v4hi)
6660 long long __builtin_arm_wunpckehuw (v2si)
6661 v4hi __builtin_arm_wunpckelsb (v8qi)
6662 v2si __builtin_arm_wunpckelsh (v4hi)
6663 long long __builtin_arm_wunpckelsw (v2si)
6664 v4hi __builtin_arm_wunpckelub (v8qi)
6665 v2si __builtin_arm_wunpckeluh (v4hi)
6666 long long __builtin_arm_wunpckeluw (v2si)
6667 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
6668 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
6669 v2si __builtin_arm_wunpckihw (v2si, v2si)
6670 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
6671 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
6672 v2si __builtin_arm_wunpckilw (v2si, v2si)
6673 long long __builtin_arm_wxor (long long, long long)
6674 long long __builtin_arm_wzero ()
6677 @node ARM NEON Intrinsics
6678 @subsection ARM NEON Intrinsics
6680 These built-in intrinsics for the ARM Advanced SIMD extension are available
6681 when the @option{-mfpu=neon} switch is used:
6683 @include arm-neon-intrinsics.texi
6685 @node Blackfin Built-in Functions
6686 @subsection Blackfin Built-in Functions
6688 Currently, there are two Blackfin-specific built-in functions. These are
6689 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
6690 using inline assembly; by using these built-in functions the compiler can
6691 automatically add workarounds for hardware errata involving these
6692 instructions. These functions are named as follows:
6695 void __builtin_bfin_csync (void)
6696 void __builtin_bfin_ssync (void)
6699 @node FR-V Built-in Functions
6700 @subsection FR-V Built-in Functions
6702 GCC provides many FR-V-specific built-in functions. In general,
6703 these functions are intended to be compatible with those described
6704 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
6705 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
6706 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
6707 pointer rather than by value.
6709 Most of the functions are named after specific FR-V instructions.
6710 Such functions are said to be ``directly mapped'' and are summarized
6711 here in tabular form.
6715 * Directly-mapped Integer Functions::
6716 * Directly-mapped Media Functions::
6717 * Raw read/write Functions::
6718 * Other Built-in Functions::
6721 @node Argument Types
6722 @subsubsection Argument Types
6724 The arguments to the built-in functions can be divided into three groups:
6725 register numbers, compile-time constants and run-time values. In order
6726 to make this classification clear at a glance, the arguments and return
6727 values are given the following pseudo types:
6729 @multitable @columnfractions .20 .30 .15 .35
6730 @item Pseudo type @tab Real C type @tab Constant? @tab Description
6731 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
6732 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
6733 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
6734 @item @code{uw2} @tab @code{unsigned long long} @tab No
6735 @tab an unsigned doubleword
6736 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
6737 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
6738 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
6739 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
6742 These pseudo types are not defined by GCC, they are simply a notational
6743 convenience used in this manual.
6745 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
6746 and @code{sw2} are evaluated at run time. They correspond to
6747 register operands in the underlying FR-V instructions.
6749 @code{const} arguments represent immediate operands in the underlying
6750 FR-V instructions. They must be compile-time constants.
6752 @code{acc} arguments are evaluated at compile time and specify the number
6753 of an accumulator register. For example, an @code{acc} argument of 2
6754 will select the ACC2 register.
6756 @code{iacc} arguments are similar to @code{acc} arguments but specify the
6757 number of an IACC register. See @pxref{Other Built-in Functions}
6760 @node Directly-mapped Integer Functions
6761 @subsubsection Directly-mapped Integer Functions
6763 The functions listed below map directly to FR-V I-type instructions.
6765 @multitable @columnfractions .45 .32 .23
6766 @item Function prototype @tab Example usage @tab Assembly output
6767 @item @code{sw1 __ADDSS (sw1, sw1)}
6768 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
6769 @tab @code{ADDSS @var{a},@var{b},@var{c}}
6770 @item @code{sw1 __SCAN (sw1, sw1)}
6771 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
6772 @tab @code{SCAN @var{a},@var{b},@var{c}}
6773 @item @code{sw1 __SCUTSS (sw1)}
6774 @tab @code{@var{b} = __SCUTSS (@var{a})}
6775 @tab @code{SCUTSS @var{a},@var{b}}
6776 @item @code{sw1 __SLASS (sw1, sw1)}
6777 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
6778 @tab @code{SLASS @var{a},@var{b},@var{c}}
6779 @item @code{void __SMASS (sw1, sw1)}
6780 @tab @code{__SMASS (@var{a}, @var{b})}
6781 @tab @code{SMASS @var{a},@var{b}}
6782 @item @code{void __SMSSS (sw1, sw1)}
6783 @tab @code{__SMSSS (@var{a}, @var{b})}
6784 @tab @code{SMSSS @var{a},@var{b}}
6785 @item @code{void __SMU (sw1, sw1)}
6786 @tab @code{__SMU (@var{a}, @var{b})}
6787 @tab @code{SMU @var{a},@var{b}}
6788 @item @code{sw2 __SMUL (sw1, sw1)}
6789 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
6790 @tab @code{SMUL @var{a},@var{b},@var{c}}
6791 @item @code{sw1 __SUBSS (sw1, sw1)}
6792 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
6793 @tab @code{SUBSS @var{a},@var{b},@var{c}}
6794 @item @code{uw2 __UMUL (uw1, uw1)}
6795 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
6796 @tab @code{UMUL @var{a},@var{b},@var{c}}
6799 @node Directly-mapped Media Functions
6800 @subsubsection Directly-mapped Media Functions
6802 The functions listed below map directly to FR-V M-type instructions.
6804 @multitable @columnfractions .45 .32 .23
6805 @item Function prototype @tab Example usage @tab Assembly output
6806 @item @code{uw1 __MABSHS (sw1)}
6807 @tab @code{@var{b} = __MABSHS (@var{a})}
6808 @tab @code{MABSHS @var{a},@var{b}}
6809 @item @code{void __MADDACCS (acc, acc)}
6810 @tab @code{__MADDACCS (@var{b}, @var{a})}
6811 @tab @code{MADDACCS @var{a},@var{b}}
6812 @item @code{sw1 __MADDHSS (sw1, sw1)}
6813 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
6814 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
6815 @item @code{uw1 __MADDHUS (uw1, uw1)}
6816 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
6817 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
6818 @item @code{uw1 __MAND (uw1, uw1)}
6819 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
6820 @tab @code{MAND @var{a},@var{b},@var{c}}
6821 @item @code{void __MASACCS (acc, acc)}
6822 @tab @code{__MASACCS (@var{b}, @var{a})}
6823 @tab @code{MASACCS @var{a},@var{b}}
6824 @item @code{uw1 __MAVEH (uw1, uw1)}
6825 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
6826 @tab @code{MAVEH @var{a},@var{b},@var{c}}
6827 @item @code{uw2 __MBTOH (uw1)}
6828 @tab @code{@var{b} = __MBTOH (@var{a})}
6829 @tab @code{MBTOH @var{a},@var{b}}
6830 @item @code{void __MBTOHE (uw1 *, uw1)}
6831 @tab @code{__MBTOHE (&@var{b}, @var{a})}
6832 @tab @code{MBTOHE @var{a},@var{b}}
6833 @item @code{void __MCLRACC (acc)}
6834 @tab @code{__MCLRACC (@var{a})}
6835 @tab @code{MCLRACC @var{a}}
6836 @item @code{void __MCLRACCA (void)}
6837 @tab @code{__MCLRACCA ()}
6838 @tab @code{MCLRACCA}
6839 @item @code{uw1 __Mcop1 (uw1, uw1)}
6840 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
6841 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
6842 @item @code{uw1 __Mcop2 (uw1, uw1)}
6843 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
6844 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
6845 @item @code{uw1 __MCPLHI (uw2, const)}
6846 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
6847 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
6848 @item @code{uw1 __MCPLI (uw2, const)}
6849 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
6850 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
6851 @item @code{void __MCPXIS (acc, sw1, sw1)}
6852 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
6853 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
6854 @item @code{void __MCPXIU (acc, uw1, uw1)}
6855 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
6856 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
6857 @item @code{void __MCPXRS (acc, sw1, sw1)}
6858 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
6859 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
6860 @item @code{void __MCPXRU (acc, uw1, uw1)}
6861 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
6862 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
6863 @item @code{uw1 __MCUT (acc, uw1)}
6864 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
6865 @tab @code{MCUT @var{a},@var{b},@var{c}}
6866 @item @code{uw1 __MCUTSS (acc, sw1)}
6867 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
6868 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
6869 @item @code{void __MDADDACCS (acc, acc)}
6870 @tab @code{__MDADDACCS (@var{b}, @var{a})}
6871 @tab @code{MDADDACCS @var{a},@var{b}}
6872 @item @code{void __MDASACCS (acc, acc)}
6873 @tab @code{__MDASACCS (@var{b}, @var{a})}
6874 @tab @code{MDASACCS @var{a},@var{b}}
6875 @item @code{uw2 __MDCUTSSI (acc, const)}
6876 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
6877 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
6878 @item @code{uw2 __MDPACKH (uw2, uw2)}
6879 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
6880 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
6881 @item @code{uw2 __MDROTLI (uw2, const)}
6882 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
6883 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
6884 @item @code{void __MDSUBACCS (acc, acc)}
6885 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
6886 @tab @code{MDSUBACCS @var{a},@var{b}}
6887 @item @code{void __MDUNPACKH (uw1 *, uw2)}
6888 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
6889 @tab @code{MDUNPACKH @var{a},@var{b}}
6890 @item @code{uw2 __MEXPDHD (uw1, const)}
6891 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
6892 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
6893 @item @code{uw1 __MEXPDHW (uw1, const)}
6894 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
6895 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
6896 @item @code{uw1 __MHDSETH (uw1, const)}
6897 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
6898 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
6899 @item @code{sw1 __MHDSETS (const)}
6900 @tab @code{@var{b} = __MHDSETS (@var{a})}
6901 @tab @code{MHDSETS #@var{a},@var{b}}
6902 @item @code{uw1 __MHSETHIH (uw1, const)}
6903 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
6904 @tab @code{MHSETHIH #@var{a},@var{b}}
6905 @item @code{sw1 __MHSETHIS (sw1, const)}
6906 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
6907 @tab @code{MHSETHIS #@var{a},@var{b}}
6908 @item @code{uw1 __MHSETLOH (uw1, const)}
6909 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
6910 @tab @code{MHSETLOH #@var{a},@var{b}}
6911 @item @code{sw1 __MHSETLOS (sw1, const)}
6912 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
6913 @tab @code{MHSETLOS #@var{a},@var{b}}
6914 @item @code{uw1 __MHTOB (uw2)}
6915 @tab @code{@var{b} = __MHTOB (@var{a})}
6916 @tab @code{MHTOB @var{a},@var{b}}
6917 @item @code{void __MMACHS (acc, sw1, sw1)}
6918 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
6919 @tab @code{MMACHS @var{a},@var{b},@var{c}}
6920 @item @code{void __MMACHU (acc, uw1, uw1)}
6921 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
6922 @tab @code{MMACHU @var{a},@var{b},@var{c}}
6923 @item @code{void __MMRDHS (acc, sw1, sw1)}
6924 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
6925 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
6926 @item @code{void __MMRDHU (acc, uw1, uw1)}
6927 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
6928 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
6929 @item @code{void __MMULHS (acc, sw1, sw1)}
6930 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
6931 @tab @code{MMULHS @var{a},@var{b},@var{c}}
6932 @item @code{void __MMULHU (acc, uw1, uw1)}
6933 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
6934 @tab @code{MMULHU @var{a},@var{b},@var{c}}
6935 @item @code{void __MMULXHS (acc, sw1, sw1)}
6936 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
6937 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
6938 @item @code{void __MMULXHU (acc, uw1, uw1)}
6939 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
6940 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
6941 @item @code{uw1 __MNOT (uw1)}
6942 @tab @code{@var{b} = __MNOT (@var{a})}
6943 @tab @code{MNOT @var{a},@var{b}}
6944 @item @code{uw1 __MOR (uw1, uw1)}
6945 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
6946 @tab @code{MOR @var{a},@var{b},@var{c}}
6947 @item @code{uw1 __MPACKH (uh, uh)}
6948 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
6949 @tab @code{MPACKH @var{a},@var{b},@var{c}}
6950 @item @code{sw2 __MQADDHSS (sw2, sw2)}
6951 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
6952 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
6953 @item @code{uw2 __MQADDHUS (uw2, uw2)}
6954 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
6955 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
6956 @item @code{void __MQCPXIS (acc, sw2, sw2)}
6957 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
6958 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
6959 @item @code{void __MQCPXIU (acc, uw2, uw2)}
6960 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
6961 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
6962 @item @code{void __MQCPXRS (acc, sw2, sw2)}
6963 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
6964 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
6965 @item @code{void __MQCPXRU (acc, uw2, uw2)}
6966 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
6967 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
6968 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
6969 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
6970 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
6971 @item @code{sw2 __MQLMTHS (sw2, sw2)}
6972 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
6973 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
6974 @item @code{void __MQMACHS (acc, sw2, sw2)}
6975 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
6976 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
6977 @item @code{void __MQMACHU (acc, uw2, uw2)}
6978 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
6979 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
6980 @item @code{void __MQMACXHS (acc, sw2, sw2)}
6981 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
6982 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
6983 @item @code{void __MQMULHS (acc, sw2, sw2)}
6984 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
6985 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
6986 @item @code{void __MQMULHU (acc, uw2, uw2)}
6987 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
6988 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
6989 @item @code{void __MQMULXHS (acc, sw2, sw2)}
6990 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
6991 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
6992 @item @code{void __MQMULXHU (acc, uw2, uw2)}
6993 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
6994 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
6995 @item @code{sw2 __MQSATHS (sw2, sw2)}
6996 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
6997 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
6998 @item @code{uw2 __MQSLLHI (uw2, int)}
6999 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
7000 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
7001 @item @code{sw2 __MQSRAHI (sw2, int)}
7002 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
7003 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
7004 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
7005 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
7006 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
7007 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
7008 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
7009 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
7010 @item @code{void __MQXMACHS (acc, sw2, sw2)}
7011 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
7012 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
7013 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
7014 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
7015 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
7016 @item @code{uw1 __MRDACC (acc)}
7017 @tab @code{@var{b} = __MRDACC (@var{a})}
7018 @tab @code{MRDACC @var{a},@var{b}}
7019 @item @code{uw1 __MRDACCG (acc)}
7020 @tab @code{@var{b} = __MRDACCG (@var{a})}
7021 @tab @code{MRDACCG @var{a},@var{b}}
7022 @item @code{uw1 __MROTLI (uw1, const)}
7023 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
7024 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
7025 @item @code{uw1 __MROTRI (uw1, const)}
7026 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
7027 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
7028 @item @code{sw1 __MSATHS (sw1, sw1)}
7029 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
7030 @tab @code{MSATHS @var{a},@var{b},@var{c}}
7031 @item @code{uw1 __MSATHU (uw1, uw1)}
7032 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
7033 @tab @code{MSATHU @var{a},@var{b},@var{c}}
7034 @item @code{uw1 __MSLLHI (uw1, const)}
7035 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
7036 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
7037 @item @code{sw1 __MSRAHI (sw1, const)}
7038 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
7039 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
7040 @item @code{uw1 __MSRLHI (uw1, const)}
7041 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
7042 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
7043 @item @code{void __MSUBACCS (acc, acc)}
7044 @tab @code{__MSUBACCS (@var{b}, @var{a})}
7045 @tab @code{MSUBACCS @var{a},@var{b}}
7046 @item @code{sw1 __MSUBHSS (sw1, sw1)}
7047 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
7048 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
7049 @item @code{uw1 __MSUBHUS (uw1, uw1)}
7050 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
7051 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
7052 @item @code{void __MTRAP (void)}
7053 @tab @code{__MTRAP ()}
7055 @item @code{uw2 __MUNPACKH (uw1)}
7056 @tab @code{@var{b} = __MUNPACKH (@var{a})}
7057 @tab @code{MUNPACKH @var{a},@var{b}}
7058 @item @code{uw1 __MWCUT (uw2, uw1)}
7059 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
7060 @tab @code{MWCUT @var{a},@var{b},@var{c}}
7061 @item @code{void __MWTACC (acc, uw1)}
7062 @tab @code{__MWTACC (@var{b}, @var{a})}
7063 @tab @code{MWTACC @var{a},@var{b}}
7064 @item @code{void __MWTACCG (acc, uw1)}
7065 @tab @code{__MWTACCG (@var{b}, @var{a})}
7066 @tab @code{MWTACCG @var{a},@var{b}}
7067 @item @code{uw1 __MXOR (uw1, uw1)}
7068 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
7069 @tab @code{MXOR @var{a},@var{b},@var{c}}
7072 @node Raw read/write Functions
7073 @subsubsection Raw read/write Functions
7075 This sections describes built-in functions related to read and write
7076 instructions to access memory. These functions generate
7077 @code{membar} instructions to flush the I/O load and stores where
7078 appropriate, as described in Fujitsu's manual described above.
7082 @item unsigned char __builtin_read8 (void *@var{data})
7083 @item unsigned short __builtin_read16 (void *@var{data})
7084 @item unsigned long __builtin_read32 (void *@var{data})
7085 @item unsigned long long __builtin_read64 (void *@var{data})
7087 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
7088 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
7089 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
7090 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
7093 @node Other Built-in Functions
7094 @subsubsection Other Built-in Functions
7096 This section describes built-in functions that are not named after
7097 a specific FR-V instruction.
7100 @item sw2 __IACCreadll (iacc @var{reg})
7101 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
7102 for future expansion and must be 0.
7104 @item sw1 __IACCreadl (iacc @var{reg})
7105 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
7106 Other values of @var{reg} are rejected as invalid.
7108 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
7109 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
7110 is reserved for future expansion and must be 0.
7112 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
7113 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
7114 is 1. Other values of @var{reg} are rejected as invalid.
7116 @item void __data_prefetch0 (const void *@var{x})
7117 Use the @code{dcpl} instruction to load the contents of address @var{x}
7118 into the data cache.
7120 @item void __data_prefetch (const void *@var{x})
7121 Use the @code{nldub} instruction to load the contents of address @var{x}
7122 into the data cache. The instruction will be issued in slot I1@.
7125 @node X86 Built-in Functions
7126 @subsection X86 Built-in Functions
7128 These built-in functions are available for the i386 and x86-64 family
7129 of computers, depending on the command-line switches used.
7131 Note that, if you specify command-line switches such as @option{-msse},
7132 the compiler could use the extended instruction sets even if the built-ins
7133 are not used explicitly in the program. For this reason, applications
7134 which perform runtime CPU detection must compile separate files for each
7135 supported architecture, using the appropriate flags. In particular,
7136 the file containing the CPU detection code should be compiled without
7139 The following machine modes are available for use with MMX built-in functions
7140 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
7141 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
7142 vector of eight 8-bit integers. Some of the built-in functions operate on
7143 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
7145 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
7146 of two 32-bit floating point values.
7148 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
7149 floating point values. Some instructions use a vector of four 32-bit
7150 integers, these use @code{V4SI}. Finally, some instructions operate on an
7151 entire vector register, interpreting it as a 128-bit integer, these use mode
7154 In the 64-bit mode, x86-64 family of processors uses additional built-in
7155 functions for efficient use of @code{TF} (@code{__float128}) 128-bit
7156 floating point and @code{TC} 128-bit complex floating point values.
7158 The following floating point built-in functions are made available in the
7159 64-bit mode. All of them implement the function that is part of the name.
7162 __float128 __builtin_fabsq (__float128)
7163 __float128 __builtin_copysignq (__float128, __float128)
7166 The following floating point built-in functions are made available in the
7170 @item __float128 __builtin_infq (void)
7171 Similar to @code{__builtin_inf}, except the return type is @code{__float128}.
7174 The following built-in functions are made available by @option{-mmmx}.
7175 All of them generate the machine instruction that is part of the name.
7178 v8qi __builtin_ia32_paddb (v8qi, v8qi)
7179 v4hi __builtin_ia32_paddw (v4hi, v4hi)
7180 v2si __builtin_ia32_paddd (v2si, v2si)
7181 v8qi __builtin_ia32_psubb (v8qi, v8qi)
7182 v4hi __builtin_ia32_psubw (v4hi, v4hi)
7183 v2si __builtin_ia32_psubd (v2si, v2si)
7184 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
7185 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
7186 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
7187 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
7188 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
7189 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
7190 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
7191 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
7192 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
7193 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
7194 di __builtin_ia32_pand (di, di)
7195 di __builtin_ia32_pandn (di,di)
7196 di __builtin_ia32_por (di, di)
7197 di __builtin_ia32_pxor (di, di)
7198 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7199 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7200 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7201 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7202 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7203 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7204 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7205 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7206 v2si __builtin_ia32_punpckhdq (v2si, v2si)
7207 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7208 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7209 v2si __builtin_ia32_punpckldq (v2si, v2si)
7210 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7211 v4hi __builtin_ia32_packssdw (v2si, v2si)
7212 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7215 The following built-in functions are made available either with
7216 @option{-msse}, or with a combination of @option{-m3dnow} and
7217 @option{-march=athlon}. All of them generate the machine
7218 instruction that is part of the name.
7221 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7222 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7223 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7224 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
7225 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7226 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7227 v8qi __builtin_ia32_pminub (v8qi, v8qi)
7228 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7229 int __builtin_ia32_pextrw (v4hi, int)
7230 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7231 int __builtin_ia32_pmovmskb (v8qi)
7232 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7233 void __builtin_ia32_movntq (di *, di)
7234 void __builtin_ia32_sfence (void)
7237 The following built-in functions are available when @option{-msse} is used.
7238 All of them generate the machine instruction that is part of the name.
7241 int __builtin_ia32_comieq (v4sf, v4sf)
7242 int __builtin_ia32_comineq (v4sf, v4sf)
7243 int __builtin_ia32_comilt (v4sf, v4sf)
7244 int __builtin_ia32_comile (v4sf, v4sf)
7245 int __builtin_ia32_comigt (v4sf, v4sf)
7246 int __builtin_ia32_comige (v4sf, v4sf)
7247 int __builtin_ia32_ucomieq (v4sf, v4sf)
7248 int __builtin_ia32_ucomineq (v4sf, v4sf)
7249 int __builtin_ia32_ucomilt (v4sf, v4sf)
7250 int __builtin_ia32_ucomile (v4sf, v4sf)
7251 int __builtin_ia32_ucomigt (v4sf, v4sf)
7252 int __builtin_ia32_ucomige (v4sf, v4sf)
7253 v4sf __builtin_ia32_addps (v4sf, v4sf)
7254 v4sf __builtin_ia32_subps (v4sf, v4sf)
7255 v4sf __builtin_ia32_mulps (v4sf, v4sf)
7256 v4sf __builtin_ia32_divps (v4sf, v4sf)
7257 v4sf __builtin_ia32_addss (v4sf, v4sf)
7258 v4sf __builtin_ia32_subss (v4sf, v4sf)
7259 v4sf __builtin_ia32_mulss (v4sf, v4sf)
7260 v4sf __builtin_ia32_divss (v4sf, v4sf)
7261 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
7262 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
7263 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
7264 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
7265 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
7266 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
7267 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
7268 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
7269 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
7270 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
7271 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
7272 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
7273 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
7274 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
7275 v4si __builtin_ia32_cmpless (v4sf, v4sf)
7276 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
7277 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
7278 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
7279 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
7280 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
7281 v4sf __builtin_ia32_maxps (v4sf, v4sf)
7282 v4sf __builtin_ia32_maxss (v4sf, v4sf)
7283 v4sf __builtin_ia32_minps (v4sf, v4sf)
7284 v4sf __builtin_ia32_minss (v4sf, v4sf)
7285 v4sf __builtin_ia32_andps (v4sf, v4sf)
7286 v4sf __builtin_ia32_andnps (v4sf, v4sf)
7287 v4sf __builtin_ia32_orps (v4sf, v4sf)
7288 v4sf __builtin_ia32_xorps (v4sf, v4sf)
7289 v4sf __builtin_ia32_movss (v4sf, v4sf)
7290 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
7291 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
7292 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
7293 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
7294 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
7295 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
7296 v2si __builtin_ia32_cvtps2pi (v4sf)
7297 int __builtin_ia32_cvtss2si (v4sf)
7298 v2si __builtin_ia32_cvttps2pi (v4sf)
7299 int __builtin_ia32_cvttss2si (v4sf)
7300 v4sf __builtin_ia32_rcpps (v4sf)
7301 v4sf __builtin_ia32_rsqrtps (v4sf)
7302 v4sf __builtin_ia32_sqrtps (v4sf)
7303 v4sf __builtin_ia32_rcpss (v4sf)
7304 v4sf __builtin_ia32_rsqrtss (v4sf)
7305 v4sf __builtin_ia32_sqrtss (v4sf)
7306 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
7307 void __builtin_ia32_movntps (float *, v4sf)
7308 int __builtin_ia32_movmskps (v4sf)
7311 The following built-in functions are available when @option{-msse} is used.
7314 @item v4sf __builtin_ia32_loadaps (float *)
7315 Generates the @code{movaps} machine instruction as a load from memory.
7316 @item void __builtin_ia32_storeaps (float *, v4sf)
7317 Generates the @code{movaps} machine instruction as a store to memory.
7318 @item v4sf __builtin_ia32_loadups (float *)
7319 Generates the @code{movups} machine instruction as a load from memory.
7320 @item void __builtin_ia32_storeups (float *, v4sf)
7321 Generates the @code{movups} machine instruction as a store to memory.
7322 @item v4sf __builtin_ia32_loadsss (float *)
7323 Generates the @code{movss} machine instruction as a load from memory.
7324 @item void __builtin_ia32_storess (float *, v4sf)
7325 Generates the @code{movss} machine instruction as a store to memory.
7326 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
7327 Generates the @code{movhps} machine instruction as a load from memory.
7328 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
7329 Generates the @code{movlps} machine instruction as a load from memory
7330 @item void __builtin_ia32_storehps (v4sf, v2si *)
7331 Generates the @code{movhps} machine instruction as a store to memory.
7332 @item void __builtin_ia32_storelps (v4sf, v2si *)
7333 Generates the @code{movlps} machine instruction as a store to memory.
7336 The following built-in functions are available when @option{-msse2} is used.
7337 All of them generate the machine instruction that is part of the name.
7340 int __builtin_ia32_comisdeq (v2df, v2df)
7341 int __builtin_ia32_comisdlt (v2df, v2df)
7342 int __builtin_ia32_comisdle (v2df, v2df)
7343 int __builtin_ia32_comisdgt (v2df, v2df)
7344 int __builtin_ia32_comisdge (v2df, v2df)
7345 int __builtin_ia32_comisdneq (v2df, v2df)
7346 int __builtin_ia32_ucomisdeq (v2df, v2df)
7347 int __builtin_ia32_ucomisdlt (v2df, v2df)
7348 int __builtin_ia32_ucomisdle (v2df, v2df)
7349 int __builtin_ia32_ucomisdgt (v2df, v2df)
7350 int __builtin_ia32_ucomisdge (v2df, v2df)
7351 int __builtin_ia32_ucomisdneq (v2df, v2df)
7352 v2df __builtin_ia32_cmpeqpd (v2df, v2df)
7353 v2df __builtin_ia32_cmpltpd (v2df, v2df)
7354 v2df __builtin_ia32_cmplepd (v2df, v2df)
7355 v2df __builtin_ia32_cmpgtpd (v2df, v2df)
7356 v2df __builtin_ia32_cmpgepd (v2df, v2df)
7357 v2df __builtin_ia32_cmpunordpd (v2df, v2df)
7358 v2df __builtin_ia32_cmpneqpd (v2df, v2df)
7359 v2df __builtin_ia32_cmpnltpd (v2df, v2df)
7360 v2df __builtin_ia32_cmpnlepd (v2df, v2df)
7361 v2df __builtin_ia32_cmpngtpd (v2df, v2df)
7362 v2df __builtin_ia32_cmpngepd (v2df, v2df)
7363 v2df __builtin_ia32_cmpordpd (v2df, v2df)
7364 v2df __builtin_ia32_cmpeqsd (v2df, v2df)
7365 v2df __builtin_ia32_cmpltsd (v2df, v2df)
7366 v2df __builtin_ia32_cmplesd (v2df, v2df)
7367 v2df __builtin_ia32_cmpunordsd (v2df, v2df)
7368 v2df __builtin_ia32_cmpneqsd (v2df, v2df)
7369 v2df __builtin_ia32_cmpnltsd (v2df, v2df)
7370 v2df __builtin_ia32_cmpnlesd (v2df, v2df)
7371 v2df __builtin_ia32_cmpordsd (v2df, v2df)
7372 v2di __builtin_ia32_paddq (v2di, v2di)
7373 v2di __builtin_ia32_psubq (v2di, v2di)
7374 v2df __builtin_ia32_addpd (v2df, v2df)
7375 v2df __builtin_ia32_subpd (v2df, v2df)
7376 v2df __builtin_ia32_mulpd (v2df, v2df)
7377 v2df __builtin_ia32_divpd (v2df, v2df)
7378 v2df __builtin_ia32_addsd (v2df, v2df)
7379 v2df __builtin_ia32_subsd (v2df, v2df)
7380 v2df __builtin_ia32_mulsd (v2df, v2df)
7381 v2df __builtin_ia32_divsd (v2df, v2df)
7382 v2df __builtin_ia32_minpd (v2df, v2df)
7383 v2df __builtin_ia32_maxpd (v2df, v2df)
7384 v2df __builtin_ia32_minsd (v2df, v2df)
7385 v2df __builtin_ia32_maxsd (v2df, v2df)
7386 v2df __builtin_ia32_andpd (v2df, v2df)
7387 v2df __builtin_ia32_andnpd (v2df, v2df)
7388 v2df __builtin_ia32_orpd (v2df, v2df)
7389 v2df __builtin_ia32_xorpd (v2df, v2df)
7390 v2df __builtin_ia32_movsd (v2df, v2df)
7391 v2df __builtin_ia32_unpckhpd (v2df, v2df)
7392 v2df __builtin_ia32_unpcklpd (v2df, v2df)
7393 v16qi __builtin_ia32_paddb128 (v16qi, v16qi)
7394 v8hi __builtin_ia32_paddw128 (v8hi, v8hi)
7395 v4si __builtin_ia32_paddd128 (v4si, v4si)
7396 v2di __builtin_ia32_paddq128 (v2di, v2di)
7397 v16qi __builtin_ia32_psubb128 (v16qi, v16qi)
7398 v8hi __builtin_ia32_psubw128 (v8hi, v8hi)
7399 v4si __builtin_ia32_psubd128 (v4si, v4si)
7400 v2di __builtin_ia32_psubq128 (v2di, v2di)
7401 v8hi __builtin_ia32_pmullw128 (v8hi, v8hi)
7402 v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi)
7403 v2di __builtin_ia32_pand128 (v2di, v2di)
7404 v2di __builtin_ia32_pandn128 (v2di, v2di)
7405 v2di __builtin_ia32_por128 (v2di, v2di)
7406 v2di __builtin_ia32_pxor128 (v2di, v2di)
7407 v16qi __builtin_ia32_pavgb128 (v16qi, v16qi)
7408 v8hi __builtin_ia32_pavgw128 (v8hi, v8hi)
7409 v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi)
7410 v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi)
7411 v4si __builtin_ia32_pcmpeqd128 (v4si, v4si)
7412 v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi)
7413 v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi)
7414 v4si __builtin_ia32_pcmpgtd128 (v4si, v4si)
7415 v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi)
7416 v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi)
7417 v16qi __builtin_ia32_pminub128 (v16qi, v16qi)
7418 v8hi __builtin_ia32_pminsw128 (v8hi, v8hi)
7419 v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi)
7420 v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi)
7421 v4si __builtin_ia32_punpckhdq128 (v4si, v4si)
7422 v2di __builtin_ia32_punpckhqdq128 (v2di, v2di)
7423 v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi)
7424 v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi)
7425 v4si __builtin_ia32_punpckldq128 (v4si, v4si)
7426 v2di __builtin_ia32_punpcklqdq128 (v2di, v2di)
7427 v16qi __builtin_ia32_packsswb128 (v16qi, v16qi)
7428 v8hi __builtin_ia32_packssdw128 (v8hi, v8hi)
7429 v16qi __builtin_ia32_packuswb128 (v16qi, v16qi)
7430 v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi)
7431 void __builtin_ia32_maskmovdqu (v16qi, v16qi)
7432 v2df __builtin_ia32_loadupd (double *)
7433 void __builtin_ia32_storeupd (double *, v2df)
7434 v2df __builtin_ia32_loadhpd (v2df, double *)
7435 v2df __builtin_ia32_loadlpd (v2df, double *)
7436 int __builtin_ia32_movmskpd (v2df)
7437 int __builtin_ia32_pmovmskb128 (v16qi)
7438 void __builtin_ia32_movnti (int *, int)
7439 void __builtin_ia32_movntpd (double *, v2df)
7440 void __builtin_ia32_movntdq (v2df *, v2df)
7441 v4si __builtin_ia32_pshufd (v4si, int)
7442 v8hi __builtin_ia32_pshuflw (v8hi, int)
7443 v8hi __builtin_ia32_pshufhw (v8hi, int)
7444 v2di __builtin_ia32_psadbw128 (v16qi, v16qi)
7445 v2df __builtin_ia32_sqrtpd (v2df)
7446 v2df __builtin_ia32_sqrtsd (v2df)
7447 v2df __builtin_ia32_shufpd (v2df, v2df, int)
7448 v2df __builtin_ia32_cvtdq2pd (v4si)
7449 v4sf __builtin_ia32_cvtdq2ps (v4si)
7450 v4si __builtin_ia32_cvtpd2dq (v2df)
7451 v2si __builtin_ia32_cvtpd2pi (v2df)
7452 v4sf __builtin_ia32_cvtpd2ps (v2df)
7453 v4si __builtin_ia32_cvttpd2dq (v2df)
7454 v2si __builtin_ia32_cvttpd2pi (v2df)
7455 v2df __builtin_ia32_cvtpi2pd (v2si)
7456 int __builtin_ia32_cvtsd2si (v2df)
7457 int __builtin_ia32_cvttsd2si (v2df)
7458 long long __builtin_ia32_cvtsd2si64 (v2df)
7459 long long __builtin_ia32_cvttsd2si64 (v2df)
7460 v4si __builtin_ia32_cvtps2dq (v4sf)
7461 v2df __builtin_ia32_cvtps2pd (v4sf)
7462 v4si __builtin_ia32_cvttps2dq (v4sf)
7463 v2df __builtin_ia32_cvtsi2sd (v2df, int)
7464 v2df __builtin_ia32_cvtsi642sd (v2df, long long)
7465 v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df)
7466 v2df __builtin_ia32_cvtss2sd (v2df, v4sf)
7467 void __builtin_ia32_clflush (const void *)
7468 void __builtin_ia32_lfence (void)
7469 void __builtin_ia32_mfence (void)
7470 v16qi __builtin_ia32_loaddqu (const char *)
7471 void __builtin_ia32_storedqu (char *, v16qi)
7472 unsigned long long __builtin_ia32_pmuludq (v2si, v2si)
7473 v2di __builtin_ia32_pmuludq128 (v4si, v4si)
7474 v8hi __builtin_ia32_psllw128 (v8hi, v2di)
7475 v4si __builtin_ia32_pslld128 (v4si, v2di)
7476 v2di __builtin_ia32_psllq128 (v4si, v2di)
7477 v8hi __builtin_ia32_psrlw128 (v8hi, v2di)
7478 v4si __builtin_ia32_psrld128 (v4si, v2di)
7479 v2di __builtin_ia32_psrlq128 (v2di, v2di)
7480 v8hi __builtin_ia32_psraw128 (v8hi, v2di)
7481 v4si __builtin_ia32_psrad128 (v4si, v2di)
7482 v2di __builtin_ia32_pslldqi128 (v2di, int)
7483 v8hi __builtin_ia32_psllwi128 (v8hi, int)
7484 v4si __builtin_ia32_pslldi128 (v4si, int)
7485 v2di __builtin_ia32_psllqi128 (v2di, int)
7486 v2di __builtin_ia32_psrldqi128 (v2di, int)
7487 v8hi __builtin_ia32_psrlwi128 (v8hi, int)
7488 v4si __builtin_ia32_psrldi128 (v4si, int)
7489 v2di __builtin_ia32_psrlqi128 (v2di, int)
7490 v8hi __builtin_ia32_psrawi128 (v8hi, int)
7491 v4si __builtin_ia32_psradi128 (v4si, int)
7492 v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi)
7495 The following built-in functions are available when @option{-msse3} is used.
7496 All of them generate the machine instruction that is part of the name.
7499 v2df __builtin_ia32_addsubpd (v2df, v2df)
7500 v4sf __builtin_ia32_addsubps (v4sf, v4sf)
7501 v2df __builtin_ia32_haddpd (v2df, v2df)
7502 v4sf __builtin_ia32_haddps (v4sf, v4sf)
7503 v2df __builtin_ia32_hsubpd (v2df, v2df)
7504 v4sf __builtin_ia32_hsubps (v4sf, v4sf)
7505 v16qi __builtin_ia32_lddqu (char const *)
7506 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
7507 v2df __builtin_ia32_movddup (v2df)
7508 v4sf __builtin_ia32_movshdup (v4sf)
7509 v4sf __builtin_ia32_movsldup (v4sf)
7510 void __builtin_ia32_mwait (unsigned int, unsigned int)
7513 The following built-in functions are available when @option{-msse3} is used.
7516 @item v2df __builtin_ia32_loadddup (double const *)
7517 Generates the @code{movddup} machine instruction as a load from memory.
7520 The following built-in functions are available when @option{-mssse3} is used.
7521 All of them generate the machine instruction that is part of the name
7525 v2si __builtin_ia32_phaddd (v2si, v2si)
7526 v4hi __builtin_ia32_phaddw (v4hi, v4hi)
7527 v4hi __builtin_ia32_phaddsw (v4hi, v4hi)
7528 v2si __builtin_ia32_phsubd (v2si, v2si)
7529 v4hi __builtin_ia32_phsubw (v4hi, v4hi)
7530 v4hi __builtin_ia32_phsubsw (v4hi, v4hi)
7531 v8qi __builtin_ia32_pmaddubsw (v8qi, v8qi)
7532 v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi)
7533 v8qi __builtin_ia32_pshufb (v8qi, v8qi)
7534 v8qi __builtin_ia32_psignb (v8qi, v8qi)
7535 v2si __builtin_ia32_psignd (v2si, v2si)
7536 v4hi __builtin_ia32_psignw (v4hi, v4hi)
7537 long long __builtin_ia32_palignr (long long, long long, int)
7538 v8qi __builtin_ia32_pabsb (v8qi)
7539 v2si __builtin_ia32_pabsd (v2si)
7540 v4hi __builtin_ia32_pabsw (v4hi)
7543 The following built-in functions are available when @option{-mssse3} is used.
7544 All of them generate the machine instruction that is part of the name
7548 v4si __builtin_ia32_phaddd128 (v4si, v4si)
7549 v8hi __builtin_ia32_phaddw128 (v8hi, v8hi)
7550 v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi)
7551 v4si __builtin_ia32_phsubd128 (v4si, v4si)
7552 v8hi __builtin_ia32_phsubw128 (v8hi, v8hi)
7553 v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi)
7554 v16qi __builtin_ia32_pmaddubsw128 (v16qi, v16qi)
7555 v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi)
7556 v16qi __builtin_ia32_pshufb128 (v16qi, v16qi)
7557 v16qi __builtin_ia32_psignb128 (v16qi, v16qi)
7558 v4si __builtin_ia32_psignd128 (v4si, v4si)
7559 v8hi __builtin_ia32_psignw128 (v8hi, v8hi)
7560 v2di __builtin_ia32_palignr (v2di, v2di, int)
7561 v16qi __builtin_ia32_pabsb128 (v16qi)
7562 v4si __builtin_ia32_pabsd128 (v4si)
7563 v8hi __builtin_ia32_pabsw128 (v8hi)
7566 The following built-in functions are available when @option{-msse4.1} is
7567 used. All of them generate the machine instruction that is part of the
7571 v2df __builtin_ia32_blendpd (v2df, v2df, const int)
7572 v4sf __builtin_ia32_blendps (v4sf, v4sf, const int)
7573 v2df __builtin_ia32_blendvpd (v2df, v2df, v2df)
7574 v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf)
7575 v2df __builtin_ia32_dppd (v2df, v2df, const int)
7576 v4sf __builtin_ia32_dpps (v4sf, v4sf, const int)
7577 v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int)
7578 v2di __builtin_ia32_movntdqa (v2di *);
7579 v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int)
7580 v8hi __builtin_ia32_packusdw128 (v4si, v4si)
7581 v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi)
7582 v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int)
7583 v2di __builtin_ia32_pcmpeqq (v2di, v2di)
7584 v8hi __builtin_ia32_phminposuw128 (v8hi)
7585 v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi)
7586 v4si __builtin_ia32_pmaxsd128 (v4si, v4si)
7587 v4si __builtin_ia32_pmaxud128 (v4si, v4si)
7588 v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi)
7589 v16qi __builtin_ia32_pminsb128 (v16qi, v16qi)
7590 v4si __builtin_ia32_pminsd128 (v4si, v4si)
7591 v4si __builtin_ia32_pminud128 (v4si, v4si)
7592 v8hi __builtin_ia32_pminuw128 (v8hi, v8hi)
7593 v4si __builtin_ia32_pmovsxbd128 (v16qi)
7594 v2di __builtin_ia32_pmovsxbq128 (v16qi)
7595 v8hi __builtin_ia32_pmovsxbw128 (v16qi)
7596 v2di __builtin_ia32_pmovsxdq128 (v4si)
7597 v4si __builtin_ia32_pmovsxwd128 (v8hi)
7598 v2di __builtin_ia32_pmovsxwq128 (v8hi)
7599 v4si __builtin_ia32_pmovzxbd128 (v16qi)
7600 v2di __builtin_ia32_pmovzxbq128 (v16qi)
7601 v8hi __builtin_ia32_pmovzxbw128 (v16qi)
7602 v2di __builtin_ia32_pmovzxdq128 (v4si)
7603 v4si __builtin_ia32_pmovzxwd128 (v8hi)
7604 v2di __builtin_ia32_pmovzxwq128 (v8hi)
7605 v2di __builtin_ia32_pmuldq128 (v4si, v4si)
7606 v4si __builtin_ia32_pmulld128 (v4si, v4si)
7607 int __builtin_ia32_ptestc128 (v2di, v2di)
7608 int __builtin_ia32_ptestnzc128 (v2di, v2di)
7609 int __builtin_ia32_ptestz128 (v2di, v2di)
7610 v2df __builtin_ia32_roundpd (v2df, const int)
7611 v4sf __builtin_ia32_roundps (v4sf, const int)
7612 v2df __builtin_ia32_roundsd (v2df, v2df, const int)
7613 v4sf __builtin_ia32_roundss (v4sf, v4sf, const int)
7616 The following built-in functions are available when @option{-msse4.1} is
7620 @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int)
7621 Generates the @code{insertps} machine instruction.
7622 @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int)
7623 Generates the @code{pextrb} machine instruction.
7624 @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int)
7625 Generates the @code{pinsrb} machine instruction.
7626 @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int)
7627 Generates the @code{pinsrd} machine instruction.
7628 @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int)
7629 Generates the @code{pinsrq} machine instruction in 64bit mode.
7632 The following built-in functions are changed to generate new SSE4.1
7633 instructions when @option{-msse4.1} is used.
7636 @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int)
7637 Generates the @code{extractps} machine instruction.
7638 @item int __builtin_ia32_vec_ext_v4si (v4si, const int)
7639 Generates the @code{pextrd} machine instruction.
7640 @item long long __builtin_ia32_vec_ext_v2di (v2di, const int)
7641 Generates the @code{pextrq} machine instruction in 64bit mode.
7644 The following built-in functions are available when @option{-msse4.2} is
7645 used. All of them generate the machine instruction that is part of the
7649 v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int)
7650 int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int)
7651 int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int)
7652 int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int)
7653 int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int)
7654 int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int)
7655 int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int)
7656 v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int)
7657 int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int)
7658 int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int)
7659 int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int)
7660 int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int)
7661 int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int)
7662 int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int)
7663 v2di __builtin_ia32_pcmpgtq (v2di, v2di)
7666 The following built-in functions are available when @option{-msse4.2} is
7670 @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char)
7671 Generates the @code{crc32b} machine instruction.
7672 @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short)
7673 Generates the @code{crc32w} machine instruction.
7674 @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int)
7675 Generates the @code{crc32l} machine instruction.
7676 @item unsigned long long __builtin_ia32_crc32di (unsigned int, unsigned long long)
7679 The following built-in functions are changed to generate new SSE4.2
7680 instructions when @option{-msse4.2} is used.
7683 @item int __builtin_popcount (unsigned int)
7684 Generates the @code{popcntl} machine instruction.
7685 @item int __builtin_popcountl (unsigned long)
7686 Generates the @code{popcntl} or @code{popcntq} machine instruction,
7687 depending on the size of @code{unsigned long}.
7688 @item int __builtin_popcountll (unsigned long long)
7689 Generates the @code{popcntq} machine instruction.
7692 The following built-in functions are available when @option{-msse4a} is used.
7693 All of them generate the machine instruction that is part of the name.
7696 void __builtin_ia32_movntsd (double *, v2df)
7697 void __builtin_ia32_movntss (float *, v4sf)
7698 v2di __builtin_ia32_extrq (v2di, v16qi)
7699 v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int)
7700 v2di __builtin_ia32_insertq (v2di, v2di)
7701 v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int)
7704 The following built-in functions are available when @option{-m3dnow} is used.
7705 All of them generate the machine instruction that is part of the name.
7708 void __builtin_ia32_femms (void)
7709 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
7710 v2si __builtin_ia32_pf2id (v2sf)
7711 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
7712 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
7713 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
7714 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
7715 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
7716 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
7717 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
7718 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
7719 v2sf __builtin_ia32_pfrcp (v2sf)
7720 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
7721 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
7722 v2sf __builtin_ia32_pfrsqrt (v2sf)
7723 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
7724 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
7725 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
7726 v2sf __builtin_ia32_pi2fd (v2si)
7727 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
7730 The following built-in functions are available when both @option{-m3dnow}
7731 and @option{-march=athlon} are used. All of them generate the machine
7732 instruction that is part of the name.
7735 v2si __builtin_ia32_pf2iw (v2sf)
7736 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
7737 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
7738 v2sf __builtin_ia32_pi2fw (v2si)
7739 v2sf __builtin_ia32_pswapdsf (v2sf)
7740 v2si __builtin_ia32_pswapdsi (v2si)
7743 @node MIPS DSP Built-in Functions
7744 @subsection MIPS DSP Built-in Functions
7746 The MIPS DSP Application-Specific Extension (ASE) includes new
7747 instructions that are designed to improve the performance of DSP and
7748 media applications. It provides instructions that operate on packed
7749 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data.
7751 GCC supports MIPS DSP operations using both the generic
7752 vector extensions (@pxref{Vector Extensions}) and a collection of
7753 MIPS-specific built-in functions. Both kinds of support are
7754 enabled by the @option{-mdsp} command-line option.
7756 Revision 2 of the ASE was introduced in the second half of 2006.
7757 This revision adds extra instructions to the original ASE, but is
7758 otherwise backwards-compatible with it. You can select revision 2
7759 using the command-line option @option{-mdspr2}; this option implies
7762 At present, GCC only provides support for operations on 32-bit
7763 vectors. The vector type associated with 8-bit integer data is
7764 usually called @code{v4i8}, the vector type associated with Q7
7765 is usually called @code{v4q7}, the vector type associated with 16-bit
7766 integer data is usually called @code{v2i16}, and the vector type
7767 associated with Q15 is usually called @code{v2q15}. They can be
7768 defined in C as follows:
7771 typedef signed char v4i8 __attribute__ ((vector_size(4)));
7772 typedef signed char v4q7 __attribute__ ((vector_size(4)));
7773 typedef short v2i16 __attribute__ ((vector_size(4)));
7774 typedef short v2q15 __attribute__ ((vector_size(4)));
7777 @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are
7778 initialized in the same way as aggregates. For example:
7781 v4i8 a = @{1, 2, 3, 4@};
7783 b = (v4i8) @{5, 6, 7, 8@};
7785 v2q15 c = @{0x0fcb, 0x3a75@};
7787 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
7790 @emph{Note:} The CPU's endianness determines the order in which values
7791 are packed. On little-endian targets, the first value is the least
7792 significant and the last value is the most significant. The opposite
7793 order applies to big-endian targets. For example, the code above will
7794 set the lowest byte of @code{a} to @code{1} on little-endian targets
7795 and @code{4} on big-endian targets.
7797 @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer
7798 representation. As shown in this example, the integer representation
7799 of a Q7 value can be obtained by multiplying the fractional value by
7800 @code{0x1.0p7}. The equivalent for Q15 values is to multiply by
7801 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
7804 The table below lists the @code{v4i8} and @code{v2q15} operations for which
7805 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
7806 and @code{c} and @code{d} are @code{v2q15} values.
7808 @multitable @columnfractions .50 .50
7809 @item C code @tab MIPS instruction
7810 @item @code{a + b} @tab @code{addu.qb}
7811 @item @code{c + d} @tab @code{addq.ph}
7812 @item @code{a - b} @tab @code{subu.qb}
7813 @item @code{c - d} @tab @code{subq.ph}
7816 The table below lists the @code{v2i16} operation for which
7817 hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are
7818 @code{v2i16} values.
7820 @multitable @columnfractions .50 .50
7821 @item C code @tab MIPS instruction
7822 @item @code{e * f} @tab @code{mul.ph}
7825 It is easier to describe the DSP built-in functions if we first define
7826 the following types:
7831 typedef unsigned int ui32;
7832 typedef long long a64;
7835 @code{q31} and @code{i32} are actually the same as @code{int}, but we
7836 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
7837 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
7838 @code{long long}, but we use @code{a64} to indicate values that will
7839 be placed in one of the four DSP accumulators (@code{$ac0},
7840 @code{$ac1}, @code{$ac2} or @code{$ac3}).
7842 Also, some built-in functions prefer or require immediate numbers as
7843 parameters, because the corresponding DSP instructions accept both immediate
7844 numbers and register operands, or accept immediate numbers only. The
7845 immediate parameters are listed as follows.
7854 imm_n32_31: -32 to 31.
7855 imm_n512_511: -512 to 511.
7858 The following built-in functions map directly to a particular MIPS DSP
7859 instruction. Please refer to the architecture specification
7860 for details on what each instruction does.
7863 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
7864 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
7865 q31 __builtin_mips_addq_s_w (q31, q31)
7866 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
7867 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
7868 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
7869 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
7870 q31 __builtin_mips_subq_s_w (q31, q31)
7871 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
7872 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
7873 i32 __builtin_mips_addsc (i32, i32)
7874 i32 __builtin_mips_addwc (i32, i32)
7875 i32 __builtin_mips_modsub (i32, i32)
7876 i32 __builtin_mips_raddu_w_qb (v4i8)
7877 v2q15 __builtin_mips_absq_s_ph (v2q15)
7878 q31 __builtin_mips_absq_s_w (q31)
7879 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
7880 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
7881 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
7882 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
7883 q31 __builtin_mips_preceq_w_phl (v2q15)
7884 q31 __builtin_mips_preceq_w_phr (v2q15)
7885 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
7886 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
7887 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
7888 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
7889 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
7890 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
7891 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
7892 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
7893 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
7894 v4i8 __builtin_mips_shll_qb (v4i8, i32)
7895 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
7896 v2q15 __builtin_mips_shll_ph (v2q15, i32)
7897 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
7898 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
7899 q31 __builtin_mips_shll_s_w (q31, imm0_31)
7900 q31 __builtin_mips_shll_s_w (q31, i32)
7901 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
7902 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
7903 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
7904 v2q15 __builtin_mips_shra_ph (v2q15, i32)
7905 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
7906 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
7907 q31 __builtin_mips_shra_r_w (q31, imm0_31)
7908 q31 __builtin_mips_shra_r_w (q31, i32)
7909 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
7910 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
7911 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
7912 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
7913 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
7914 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
7915 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
7916 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
7917 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
7918 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
7919 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
7920 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
7921 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
7922 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
7923 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
7924 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
7925 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
7926 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
7927 i32 __builtin_mips_bitrev (i32)
7928 i32 __builtin_mips_insv (i32, i32)
7929 v4i8 __builtin_mips_repl_qb (imm0_255)
7930 v4i8 __builtin_mips_repl_qb (i32)
7931 v2q15 __builtin_mips_repl_ph (imm_n512_511)
7932 v2q15 __builtin_mips_repl_ph (i32)
7933 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
7934 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
7935 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
7936 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
7937 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
7938 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
7939 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
7940 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
7941 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
7942 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
7943 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
7944 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
7945 i32 __builtin_mips_extr_w (a64, imm0_31)
7946 i32 __builtin_mips_extr_w (a64, i32)
7947 i32 __builtin_mips_extr_r_w (a64, imm0_31)
7948 i32 __builtin_mips_extr_s_h (a64, i32)
7949 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
7950 i32 __builtin_mips_extr_rs_w (a64, i32)
7951 i32 __builtin_mips_extr_s_h (a64, imm0_31)
7952 i32 __builtin_mips_extr_r_w (a64, i32)
7953 i32 __builtin_mips_extp (a64, imm0_31)
7954 i32 __builtin_mips_extp (a64, i32)
7955 i32 __builtin_mips_extpdp (a64, imm0_31)
7956 i32 __builtin_mips_extpdp (a64, i32)
7957 a64 __builtin_mips_shilo (a64, imm_n32_31)
7958 a64 __builtin_mips_shilo (a64, i32)
7959 a64 __builtin_mips_mthlip (a64, i32)
7960 void __builtin_mips_wrdsp (i32, imm0_63)
7961 i32 __builtin_mips_rddsp (imm0_63)
7962 i32 __builtin_mips_lbux (void *, i32)
7963 i32 __builtin_mips_lhx (void *, i32)
7964 i32 __builtin_mips_lwx (void *, i32)
7965 i32 __builtin_mips_bposge32 (void)
7968 The following built-in functions map directly to a particular MIPS DSP REV 2
7969 instruction. Please refer to the architecture specification
7970 for details on what each instruction does.
7973 v4q7 __builtin_mips_absq_s_qb (v4q7);
7974 v2i16 __builtin_mips_addu_ph (v2i16, v2i16);
7975 v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16);
7976 v4i8 __builtin_mips_adduh_qb (v4i8, v4i8);
7977 v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8);
7978 i32 __builtin_mips_append (i32, i32, imm0_31);
7979 i32 __builtin_mips_balign (i32, i32, imm0_3);
7980 i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8);
7981 i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8);
7982 i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8);
7983 a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16);
7984 a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16);
7985 a64 __builtin_mips_madd (a64, i32, i32);
7986 a64 __builtin_mips_maddu (a64, ui32, ui32);
7987 a64 __builtin_mips_msub (a64, i32, i32);
7988 a64 __builtin_mips_msubu (a64, ui32, ui32);
7989 v2i16 __builtin_mips_mul_ph (v2i16, v2i16);
7990 v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16);
7991 q31 __builtin_mips_mulq_rs_w (q31, q31);
7992 v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15);
7993 q31 __builtin_mips_mulq_s_w (q31, q31);
7994 a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16);
7995 a64 __builtin_mips_mult (i32, i32);
7996 a64 __builtin_mips_multu (ui32, ui32);
7997 v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16);
7998 v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31);
7999 v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31);
8000 i32 __builtin_mips_prepend (i32, i32, imm0_31);
8001 v4i8 __builtin_mips_shra_qb (v4i8, imm0_7);
8002 v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7);
8003 v4i8 __builtin_mips_shra_qb (v4i8, i32);
8004 v4i8 __builtin_mips_shra_r_qb (v4i8, i32);
8005 v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15);
8006 v2i16 __builtin_mips_shrl_ph (v2i16, i32);
8007 v2i16 __builtin_mips_subu_ph (v2i16, v2i16);
8008 v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16);
8009 v4i8 __builtin_mips_subuh_qb (v4i8, v4i8);
8010 v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8);
8011 v2q15 __builtin_mips_addqh_ph (v2q15, v2q15);
8012 v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15);
8013 q31 __builtin_mips_addqh_w (q31, q31);
8014 q31 __builtin_mips_addqh_r_w (q31, q31);
8015 v2q15 __builtin_mips_subqh_ph (v2q15, v2q15);
8016 v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15);
8017 q31 __builtin_mips_subqh_w (q31, q31);
8018 q31 __builtin_mips_subqh_r_w (q31, q31);
8019 a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16);
8020 a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16);
8021 a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15);
8022 a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15);
8023 a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15);
8024 a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15);
8028 @node MIPS Paired-Single Support
8029 @subsection MIPS Paired-Single Support
8031 The MIPS64 architecture includes a number of instructions that
8032 operate on pairs of single-precision floating-point values.
8033 Each pair is packed into a 64-bit floating-point register,
8034 with one element being designated the ``upper half'' and
8035 the other being designated the ``lower half''.
8037 GCC supports paired-single operations using both the generic
8038 vector extensions (@pxref{Vector Extensions}) and a collection of
8039 MIPS-specific built-in functions. Both kinds of support are
8040 enabled by the @option{-mpaired-single} command-line option.
8042 The vector type associated with paired-single values is usually
8043 called @code{v2sf}. It can be defined in C as follows:
8046 typedef float v2sf __attribute__ ((vector_size (8)));
8049 @code{v2sf} values are initialized in the same way as aggregates.
8053 v2sf a = @{1.5, 9.1@};
8056 b = (v2sf) @{e, f@};
8059 @emph{Note:} The CPU's endianness determines which value is stored in
8060 the upper half of a register and which value is stored in the lower half.
8061 On little-endian targets, the first value is the lower one and the second
8062 value is the upper one. The opposite order applies to big-endian targets.
8063 For example, the code above will set the lower half of @code{a} to
8064 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
8067 * Paired-Single Arithmetic::
8068 * Paired-Single Built-in Functions::
8069 * MIPS-3D Built-in Functions::
8072 @node Paired-Single Arithmetic
8073 @subsubsection Paired-Single Arithmetic
8075 The table below lists the @code{v2sf} operations for which hardware
8076 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
8077 values and @code{x} is an integral value.
8079 @multitable @columnfractions .50 .50
8080 @item C code @tab MIPS instruction
8081 @item @code{a + b} @tab @code{add.ps}
8082 @item @code{a - b} @tab @code{sub.ps}
8083 @item @code{-a} @tab @code{neg.ps}
8084 @item @code{a * b} @tab @code{mul.ps}
8085 @item @code{a * b + c} @tab @code{madd.ps}
8086 @item @code{a * b - c} @tab @code{msub.ps}
8087 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
8088 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
8089 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
8092 Note that the multiply-accumulate instructions can be disabled
8093 using the command-line option @code{-mno-fused-madd}.
8095 @node Paired-Single Built-in Functions
8096 @subsubsection Paired-Single Built-in Functions
8098 The following paired-single functions map directly to a particular
8099 MIPS instruction. Please refer to the architecture specification
8100 for details on what each instruction does.
8103 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
8104 Pair lower lower (@code{pll.ps}).
8106 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
8107 Pair upper lower (@code{pul.ps}).
8109 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
8110 Pair lower upper (@code{plu.ps}).
8112 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
8113 Pair upper upper (@code{puu.ps}).
8115 @item v2sf __builtin_mips_cvt_ps_s (float, float)
8116 Convert pair to paired single (@code{cvt.ps.s}).
8118 @item float __builtin_mips_cvt_s_pl (v2sf)
8119 Convert pair lower to single (@code{cvt.s.pl}).
8121 @item float __builtin_mips_cvt_s_pu (v2sf)
8122 Convert pair upper to single (@code{cvt.s.pu}).
8124 @item v2sf __builtin_mips_abs_ps (v2sf)
8125 Absolute value (@code{abs.ps}).
8127 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
8128 Align variable (@code{alnv.ps}).
8130 @emph{Note:} The value of the third parameter must be 0 or 4
8131 modulo 8, otherwise the result will be unpredictable. Please read the
8132 instruction description for details.
8135 The following multi-instruction functions are also available.
8136 In each case, @var{cond} can be any of the 16 floating-point conditions:
8137 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
8138 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
8139 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
8142 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8143 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8144 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
8145 @code{movt.ps}/@code{movf.ps}).
8147 The @code{movt} functions return the value @var{x} computed by:
8150 c.@var{cond}.ps @var{cc},@var{a},@var{b}
8151 mov.ps @var{x},@var{c}
8152 movt.ps @var{x},@var{d},@var{cc}
8155 The @code{movf} functions are similar but use @code{movf.ps} instead
8158 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8159 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8160 Comparison of two paired-single values (@code{c.@var{cond}.ps},
8161 @code{bc1t}/@code{bc1f}).
8163 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
8164 and return either the upper or lower half of the result. For example:
8168 if (__builtin_mips_upper_c_eq_ps (a, b))
8169 upper_halves_are_equal ();
8171 upper_halves_are_unequal ();
8173 if (__builtin_mips_lower_c_eq_ps (a, b))
8174 lower_halves_are_equal ();
8176 lower_halves_are_unequal ();
8180 @node MIPS-3D Built-in Functions
8181 @subsubsection MIPS-3D Built-in Functions
8183 The MIPS-3D Application-Specific Extension (ASE) includes additional
8184 paired-single instructions that are designed to improve the performance
8185 of 3D graphics operations. Support for these instructions is controlled
8186 by the @option{-mips3d} command-line option.
8188 The functions listed below map directly to a particular MIPS-3D
8189 instruction. Please refer to the architecture specification for
8190 more details on what each instruction does.
8193 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
8194 Reduction add (@code{addr.ps}).
8196 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
8197 Reduction multiply (@code{mulr.ps}).
8199 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
8200 Convert paired single to paired word (@code{cvt.pw.ps}).
8202 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
8203 Convert paired word to paired single (@code{cvt.ps.pw}).
8205 @item float __builtin_mips_recip1_s (float)
8206 @itemx double __builtin_mips_recip1_d (double)
8207 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
8208 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
8210 @item float __builtin_mips_recip2_s (float, float)
8211 @itemx double __builtin_mips_recip2_d (double, double)
8212 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
8213 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
8215 @item float __builtin_mips_rsqrt1_s (float)
8216 @itemx double __builtin_mips_rsqrt1_d (double)
8217 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
8218 Reduced precision reciprocal square root (sequence step 1)
8219 (@code{rsqrt1.@var{fmt}}).
8221 @item float __builtin_mips_rsqrt2_s (float, float)
8222 @itemx double __builtin_mips_rsqrt2_d (double, double)
8223 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
8224 Reduced precision reciprocal square root (sequence step 2)
8225 (@code{rsqrt2.@var{fmt}}).
8228 The following multi-instruction functions are also available.
8229 In each case, @var{cond} can be any of the 16 floating-point conditions:
8230 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
8231 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
8232 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
8235 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
8236 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
8237 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
8238 @code{bc1t}/@code{bc1f}).
8240 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
8241 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
8246 if (__builtin_mips_cabs_eq_s (a, b))
8252 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8253 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8254 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
8255 @code{bc1t}/@code{bc1f}).
8257 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
8258 and return either the upper or lower half of the result. For example:
8262 if (__builtin_mips_upper_cabs_eq_ps (a, b))
8263 upper_halves_are_equal ();
8265 upper_halves_are_unequal ();
8267 if (__builtin_mips_lower_cabs_eq_ps (a, b))
8268 lower_halves_are_equal ();
8270 lower_halves_are_unequal ();
8273 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8274 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8275 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
8276 @code{movt.ps}/@code{movf.ps}).
8278 The @code{movt} functions return the value @var{x} computed by:
8281 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
8282 mov.ps @var{x},@var{c}
8283 movt.ps @var{x},@var{d},@var{cc}
8286 The @code{movf} functions are similar but use @code{movf.ps} instead
8289 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8290 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8291 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8292 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
8293 Comparison of two paired-single values
8294 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
8295 @code{bc1any2t}/@code{bc1any2f}).
8297 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
8298 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
8299 result is true and the @code{all} forms return true if both results are true.
8304 if (__builtin_mips_any_c_eq_ps (a, b))
8309 if (__builtin_mips_all_c_eq_ps (a, b))
8315 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8316 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8317 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8318 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
8319 Comparison of four paired-single values
8320 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
8321 @code{bc1any4t}/@code{bc1any4f}).
8323 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
8324 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
8325 The @code{any} forms return true if any of the four results are true
8326 and the @code{all} forms return true if all four results are true.
8331 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
8336 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
8343 @node PowerPC AltiVec Built-in Functions
8344 @subsection PowerPC AltiVec Built-in Functions
8346 GCC provides an interface for the PowerPC family of processors to access
8347 the AltiVec operations described in Motorola's AltiVec Programming
8348 Interface Manual. The interface is made available by including
8349 @code{<altivec.h>} and using @option{-maltivec} and
8350 @option{-mabi=altivec}. The interface supports the following vector
8354 vector unsigned char
8358 vector unsigned short
8369 GCC's implementation of the high-level language interface available from
8370 C and C++ code differs from Motorola's documentation in several ways.
8375 A vector constant is a list of constant expressions within curly braces.
8378 A vector initializer requires no cast if the vector constant is of the
8379 same type as the variable it is initializing.
8382 If @code{signed} or @code{unsigned} is omitted, the signedness of the
8383 vector type is the default signedness of the base type. The default
8384 varies depending on the operating system, so a portable program should
8385 always specify the signedness.
8388 Compiling with @option{-maltivec} adds keywords @code{__vector},
8389 @code{__pixel}, and @code{__bool}. Macros @option{vector},
8390 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
8394 GCC allows using a @code{typedef} name as the type specifier for a
8398 For C, overloaded functions are implemented with macros so the following
8402 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
8405 Since @code{vec_add} is a macro, the vector constant in the example
8406 is treated as four separate arguments. Wrap the entire argument in
8407 parentheses for this to work.
8410 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
8411 Internally, GCC uses built-in functions to achieve the functionality in
8412 the aforementioned header file, but they are not supported and are
8413 subject to change without notice.
8415 The following interfaces are supported for the generic and specific
8416 AltiVec operations and the AltiVec predicates. In cases where there
8417 is a direct mapping between generic and specific operations, only the
8418 generic names are shown here, although the specific operations can also
8421 Arguments that are documented as @code{const int} require literal
8422 integral values within the range required for that operation.
8425 vector signed char vec_abs (vector signed char);
8426 vector signed short vec_abs (vector signed short);
8427 vector signed int vec_abs (vector signed int);
8428 vector float vec_abs (vector float);
8430 vector signed char vec_abss (vector signed char);
8431 vector signed short vec_abss (vector signed short);
8432 vector signed int vec_abss (vector signed int);
8434 vector signed char vec_add (vector bool char, vector signed char);
8435 vector signed char vec_add (vector signed char, vector bool char);
8436 vector signed char vec_add (vector signed char, vector signed char);
8437 vector unsigned char vec_add (vector bool char, vector unsigned char);
8438 vector unsigned char vec_add (vector unsigned char, vector bool char);
8439 vector unsigned char vec_add (vector unsigned char,
8440 vector unsigned char);
8441 vector signed short vec_add (vector bool short, vector signed short);
8442 vector signed short vec_add (vector signed short, vector bool short);
8443 vector signed short vec_add (vector signed short, vector signed short);
8444 vector unsigned short vec_add (vector bool short,
8445 vector unsigned short);
8446 vector unsigned short vec_add (vector unsigned short,
8448 vector unsigned short vec_add (vector unsigned short,
8449 vector unsigned short);
8450 vector signed int vec_add (vector bool int, vector signed int);
8451 vector signed int vec_add (vector signed int, vector bool int);
8452 vector signed int vec_add (vector signed int, vector signed int);
8453 vector unsigned int vec_add (vector bool int, vector unsigned int);
8454 vector unsigned int vec_add (vector unsigned int, vector bool int);
8455 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
8456 vector float vec_add (vector float, vector float);
8458 vector float vec_vaddfp (vector float, vector float);
8460 vector signed int vec_vadduwm (vector bool int, vector signed int);
8461 vector signed int vec_vadduwm (vector signed int, vector bool int);
8462 vector signed int vec_vadduwm (vector signed int, vector signed int);
8463 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
8464 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
8465 vector unsigned int vec_vadduwm (vector unsigned int,
8466 vector unsigned int);
8468 vector signed short vec_vadduhm (vector bool short,
8469 vector signed short);
8470 vector signed short vec_vadduhm (vector signed short,
8472 vector signed short vec_vadduhm (vector signed short,
8473 vector signed short);
8474 vector unsigned short vec_vadduhm (vector bool short,
8475 vector unsigned short);
8476 vector unsigned short vec_vadduhm (vector unsigned short,
8478 vector unsigned short vec_vadduhm (vector unsigned short,
8479 vector unsigned short);
8481 vector signed char vec_vaddubm (vector bool char, vector signed char);
8482 vector signed char vec_vaddubm (vector signed char, vector bool char);
8483 vector signed char vec_vaddubm (vector signed char, vector signed char);
8484 vector unsigned char vec_vaddubm (vector bool char,
8485 vector unsigned char);
8486 vector unsigned char vec_vaddubm (vector unsigned char,
8488 vector unsigned char vec_vaddubm (vector unsigned char,
8489 vector unsigned char);
8491 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
8493 vector unsigned char vec_adds (vector bool char, vector unsigned char);
8494 vector unsigned char vec_adds (vector unsigned char, vector bool char);
8495 vector unsigned char vec_adds (vector unsigned char,
8496 vector unsigned char);
8497 vector signed char vec_adds (vector bool char, vector signed char);
8498 vector signed char vec_adds (vector signed char, vector bool char);
8499 vector signed char vec_adds (vector signed char, vector signed char);
8500 vector unsigned short vec_adds (vector bool short,
8501 vector unsigned short);
8502 vector unsigned short vec_adds (vector unsigned short,
8504 vector unsigned short vec_adds (vector unsigned short,
8505 vector unsigned short);
8506 vector signed short vec_adds (vector bool short, vector signed short);
8507 vector signed short vec_adds (vector signed short, vector bool short);
8508 vector signed short vec_adds (vector signed short, vector signed short);
8509 vector unsigned int vec_adds (vector bool int, vector unsigned int);
8510 vector unsigned int vec_adds (vector unsigned int, vector bool int);
8511 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
8512 vector signed int vec_adds (vector bool int, vector signed int);
8513 vector signed int vec_adds (vector signed int, vector bool int);
8514 vector signed int vec_adds (vector signed int, vector signed int);
8516 vector signed int vec_vaddsws (vector bool int, vector signed int);
8517 vector signed int vec_vaddsws (vector signed int, vector bool int);
8518 vector signed int vec_vaddsws (vector signed int, vector signed int);
8520 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
8521 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
8522 vector unsigned int vec_vadduws (vector unsigned int,
8523 vector unsigned int);
8525 vector signed short vec_vaddshs (vector bool short,
8526 vector signed short);
8527 vector signed short vec_vaddshs (vector signed short,
8529 vector signed short vec_vaddshs (vector signed short,
8530 vector signed short);
8532 vector unsigned short vec_vadduhs (vector bool short,
8533 vector unsigned short);
8534 vector unsigned short vec_vadduhs (vector unsigned short,
8536 vector unsigned short vec_vadduhs (vector unsigned short,
8537 vector unsigned short);
8539 vector signed char vec_vaddsbs (vector bool char, vector signed char);
8540 vector signed char vec_vaddsbs (vector signed char, vector bool char);
8541 vector signed char vec_vaddsbs (vector signed char, vector signed char);
8543 vector unsigned char vec_vaddubs (vector bool char,
8544 vector unsigned char);
8545 vector unsigned char vec_vaddubs (vector unsigned char,
8547 vector unsigned char vec_vaddubs (vector unsigned char,
8548 vector unsigned char);
8550 vector float vec_and (vector float, vector float);
8551 vector float vec_and (vector float, vector bool int);
8552 vector float vec_and (vector bool int, vector float);
8553 vector bool int vec_and (vector bool int, vector bool int);
8554 vector signed int vec_and (vector bool int, vector signed int);
8555 vector signed int vec_and (vector signed int, vector bool int);
8556 vector signed int vec_and (vector signed int, vector signed int);
8557 vector unsigned int vec_and (vector bool int, vector unsigned int);
8558 vector unsigned int vec_and (vector unsigned int, vector bool int);
8559 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
8560 vector bool short vec_and (vector bool short, vector bool short);
8561 vector signed short vec_and (vector bool short, vector signed short);
8562 vector signed short vec_and (vector signed short, vector bool short);
8563 vector signed short vec_and (vector signed short, vector signed short);
8564 vector unsigned short vec_and (vector bool short,
8565 vector unsigned short);
8566 vector unsigned short vec_and (vector unsigned short,
8568 vector unsigned short vec_and (vector unsigned short,
8569 vector unsigned short);
8570 vector signed char vec_and (vector bool char, vector signed char);
8571 vector bool char vec_and (vector bool char, vector bool char);
8572 vector signed char vec_and (vector signed char, vector bool char);
8573 vector signed char vec_and (vector signed char, vector signed char);
8574 vector unsigned char vec_and (vector bool char, vector unsigned char);
8575 vector unsigned char vec_and (vector unsigned char, vector bool char);
8576 vector unsigned char vec_and (vector unsigned char,
8577 vector unsigned char);
8579 vector float vec_andc (vector float, vector float);
8580 vector float vec_andc (vector float, vector bool int);
8581 vector float vec_andc (vector bool int, vector float);
8582 vector bool int vec_andc (vector bool int, vector bool int);
8583 vector signed int vec_andc (vector bool int, vector signed int);
8584 vector signed int vec_andc (vector signed int, vector bool int);
8585 vector signed int vec_andc (vector signed int, vector signed int);
8586 vector unsigned int vec_andc (vector bool int, vector unsigned int);
8587 vector unsigned int vec_andc (vector unsigned int, vector bool int);
8588 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
8589 vector bool short vec_andc (vector bool short, vector bool short);
8590 vector signed short vec_andc (vector bool short, vector signed short);
8591 vector signed short vec_andc (vector signed short, vector bool short);
8592 vector signed short vec_andc (vector signed short, vector signed short);
8593 vector unsigned short vec_andc (vector bool short,
8594 vector unsigned short);
8595 vector unsigned short vec_andc (vector unsigned short,
8597 vector unsigned short vec_andc (vector unsigned short,
8598 vector unsigned short);
8599 vector signed char vec_andc (vector bool char, vector signed char);
8600 vector bool char vec_andc (vector bool char, vector bool char);
8601 vector signed char vec_andc (vector signed char, vector bool char);
8602 vector signed char vec_andc (vector signed char, vector signed char);
8603 vector unsigned char vec_andc (vector bool char, vector unsigned char);
8604 vector unsigned char vec_andc (vector unsigned char, vector bool char);
8605 vector unsigned char vec_andc (vector unsigned char,
8606 vector unsigned char);
8608 vector unsigned char vec_avg (vector unsigned char,
8609 vector unsigned char);
8610 vector signed char vec_avg (vector signed char, vector signed char);
8611 vector unsigned short vec_avg (vector unsigned short,
8612 vector unsigned short);
8613 vector signed short vec_avg (vector signed short, vector signed short);
8614 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
8615 vector signed int vec_avg (vector signed int, vector signed int);
8617 vector signed int vec_vavgsw (vector signed int, vector signed int);
8619 vector unsigned int vec_vavguw (vector unsigned int,
8620 vector unsigned int);
8622 vector signed short vec_vavgsh (vector signed short,
8623 vector signed short);
8625 vector unsigned short vec_vavguh (vector unsigned short,
8626 vector unsigned short);
8628 vector signed char vec_vavgsb (vector signed char, vector signed char);
8630 vector unsigned char vec_vavgub (vector unsigned char,
8631 vector unsigned char);
8633 vector float vec_ceil (vector float);
8635 vector signed int vec_cmpb (vector float, vector float);
8637 vector bool char vec_cmpeq (vector signed char, vector signed char);
8638 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
8639 vector bool short vec_cmpeq (vector signed short, vector signed short);
8640 vector bool short vec_cmpeq (vector unsigned short,
8641 vector unsigned short);
8642 vector bool int vec_cmpeq (vector signed int, vector signed int);
8643 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
8644 vector bool int vec_cmpeq (vector float, vector float);
8646 vector bool int vec_vcmpeqfp (vector float, vector float);
8648 vector bool int vec_vcmpequw (vector signed int, vector signed int);
8649 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
8651 vector bool short vec_vcmpequh (vector signed short,
8652 vector signed short);
8653 vector bool short vec_vcmpequh (vector unsigned short,
8654 vector unsigned short);
8656 vector bool char vec_vcmpequb (vector signed char, vector signed char);
8657 vector bool char vec_vcmpequb (vector unsigned char,
8658 vector unsigned char);
8660 vector bool int vec_cmpge (vector float, vector float);
8662 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
8663 vector bool char vec_cmpgt (vector signed char, vector signed char);
8664 vector bool short vec_cmpgt (vector unsigned short,
8665 vector unsigned short);
8666 vector bool short vec_cmpgt (vector signed short, vector signed short);
8667 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
8668 vector bool int vec_cmpgt (vector signed int, vector signed int);
8669 vector bool int vec_cmpgt (vector float, vector float);
8671 vector bool int vec_vcmpgtfp (vector float, vector float);
8673 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
8675 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
8677 vector bool short vec_vcmpgtsh (vector signed short,
8678 vector signed short);
8680 vector bool short vec_vcmpgtuh (vector unsigned short,
8681 vector unsigned short);
8683 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
8685 vector bool char vec_vcmpgtub (vector unsigned char,
8686 vector unsigned char);
8688 vector bool int vec_cmple (vector float, vector float);
8690 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
8691 vector bool char vec_cmplt (vector signed char, vector signed char);
8692 vector bool short vec_cmplt (vector unsigned short,
8693 vector unsigned short);
8694 vector bool short vec_cmplt (vector signed short, vector signed short);
8695 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
8696 vector bool int vec_cmplt (vector signed int, vector signed int);
8697 vector bool int vec_cmplt (vector float, vector float);
8699 vector float vec_ctf (vector unsigned int, const int);
8700 vector float vec_ctf (vector signed int, const int);
8702 vector float vec_vcfsx (vector signed int, const int);
8704 vector float vec_vcfux (vector unsigned int, const int);
8706 vector signed int vec_cts (vector float, const int);
8708 vector unsigned int vec_ctu (vector float, const int);
8710 void vec_dss (const int);
8712 void vec_dssall (void);
8714 void vec_dst (const vector unsigned char *, int, const int);
8715 void vec_dst (const vector signed char *, int, const int);
8716 void vec_dst (const vector bool char *, int, const int);
8717 void vec_dst (const vector unsigned short *, int, const int);
8718 void vec_dst (const vector signed short *, int, const int);
8719 void vec_dst (const vector bool short *, int, const int);
8720 void vec_dst (const vector pixel *, int, const int);
8721 void vec_dst (const vector unsigned int *, int, const int);
8722 void vec_dst (const vector signed int *, int, const int);
8723 void vec_dst (const vector bool int *, int, const int);
8724 void vec_dst (const vector float *, int, const int);
8725 void vec_dst (const unsigned char *, int, const int);
8726 void vec_dst (const signed char *, int, const int);
8727 void vec_dst (const unsigned short *, int, const int);
8728 void vec_dst (const short *, int, const int);
8729 void vec_dst (const unsigned int *, int, const int);
8730 void vec_dst (const int *, int, const int);
8731 void vec_dst (const unsigned long *, int, const int);
8732 void vec_dst (const long *, int, const int);
8733 void vec_dst (const float *, int, const int);
8735 void vec_dstst (const vector unsigned char *, int, const int);
8736 void vec_dstst (const vector signed char *, int, const int);
8737 void vec_dstst (const vector bool char *, int, const int);
8738 void vec_dstst (const vector unsigned short *, int, const int);
8739 void vec_dstst (const vector signed short *, int, const int);
8740 void vec_dstst (const vector bool short *, int, const int);
8741 void vec_dstst (const vector pixel *, int, const int);
8742 void vec_dstst (const vector unsigned int *, int, const int);
8743 void vec_dstst (const vector signed int *, int, const int);
8744 void vec_dstst (const vector bool int *, int, const int);
8745 void vec_dstst (const vector float *, int, const int);
8746 void vec_dstst (const unsigned char *, int, const int);
8747 void vec_dstst (const signed char *, int, const int);
8748 void vec_dstst (const unsigned short *, int, const int);
8749 void vec_dstst (const short *, int, const int);
8750 void vec_dstst (const unsigned int *, int, const int);
8751 void vec_dstst (const int *, int, const int);
8752 void vec_dstst (const unsigned long *, int, const int);
8753 void vec_dstst (const long *, int, const int);
8754 void vec_dstst (const float *, int, const int);
8756 void vec_dststt (const vector unsigned char *, int, const int);
8757 void vec_dststt (const vector signed char *, int, const int);
8758 void vec_dststt (const vector bool char *, int, const int);
8759 void vec_dststt (const vector unsigned short *, int, const int);
8760 void vec_dststt (const vector signed short *, int, const int);
8761 void vec_dststt (const vector bool short *, int, const int);
8762 void vec_dststt (const vector pixel *, int, const int);
8763 void vec_dststt (const vector unsigned int *, int, const int);
8764 void vec_dststt (const vector signed int *, int, const int);
8765 void vec_dststt (const vector bool int *, int, const int);
8766 void vec_dststt (const vector float *, int, const int);
8767 void vec_dststt (const unsigned char *, int, const int);
8768 void vec_dststt (const signed char *, int, const int);
8769 void vec_dststt (const unsigned short *, int, const int);
8770 void vec_dststt (const short *, int, const int);
8771 void vec_dststt (const unsigned int *, int, const int);
8772 void vec_dststt (const int *, int, const int);
8773 void vec_dststt (const unsigned long *, int, const int);
8774 void vec_dststt (const long *, int, const int);
8775 void vec_dststt (const float *, int, const int);
8777 void vec_dstt (const vector unsigned char *, int, const int);
8778 void vec_dstt (const vector signed char *, int, const int);
8779 void vec_dstt (const vector bool char *, int, const int);
8780 void vec_dstt (const vector unsigned short *, int, const int);
8781 void vec_dstt (const vector signed short *, int, const int);
8782 void vec_dstt (const vector bool short *, int, const int);
8783 void vec_dstt (const vector pixel *, int, const int);
8784 void vec_dstt (const vector unsigned int *, int, const int);
8785 void vec_dstt (const vector signed int *, int, const int);
8786 void vec_dstt (const vector bool int *, int, const int);
8787 void vec_dstt (const vector float *, int, const int);
8788 void vec_dstt (const unsigned char *, int, const int);
8789 void vec_dstt (const signed char *, int, const int);
8790 void vec_dstt (const unsigned short *, int, const int);
8791 void vec_dstt (const short *, int, const int);
8792 void vec_dstt (const unsigned int *, int, const int);
8793 void vec_dstt (const int *, int, const int);
8794 void vec_dstt (const unsigned long *, int, const int);
8795 void vec_dstt (const long *, int, const int);
8796 void vec_dstt (const float *, int, const int);
8798 vector float vec_expte (vector float);
8800 vector float vec_floor (vector float);
8802 vector float vec_ld (int, const vector float *);
8803 vector float vec_ld (int, const float *);
8804 vector bool int vec_ld (int, const vector bool int *);
8805 vector signed int vec_ld (int, const vector signed int *);
8806 vector signed int vec_ld (int, const int *);
8807 vector signed int vec_ld (int, const long *);
8808 vector unsigned int vec_ld (int, const vector unsigned int *);
8809 vector unsigned int vec_ld (int, const unsigned int *);
8810 vector unsigned int vec_ld (int, const unsigned long *);
8811 vector bool short vec_ld (int, const vector bool short *);
8812 vector pixel vec_ld (int, const vector pixel *);
8813 vector signed short vec_ld (int, const vector signed short *);
8814 vector signed short vec_ld (int, const short *);
8815 vector unsigned short vec_ld (int, const vector unsigned short *);
8816 vector unsigned short vec_ld (int, const unsigned short *);
8817 vector bool char vec_ld (int, const vector bool char *);
8818 vector signed char vec_ld (int, const vector signed char *);
8819 vector signed char vec_ld (int, const signed char *);
8820 vector unsigned char vec_ld (int, const vector unsigned char *);
8821 vector unsigned char vec_ld (int, const unsigned char *);
8823 vector signed char vec_lde (int, const signed char *);
8824 vector unsigned char vec_lde (int, const unsigned char *);
8825 vector signed short vec_lde (int, const short *);
8826 vector unsigned short vec_lde (int, const unsigned short *);
8827 vector float vec_lde (int, const float *);
8828 vector signed int vec_lde (int, const int *);
8829 vector unsigned int vec_lde (int, const unsigned int *);
8830 vector signed int vec_lde (int, const long *);
8831 vector unsigned int vec_lde (int, const unsigned long *);
8833 vector float vec_lvewx (int, float *);
8834 vector signed int vec_lvewx (int, int *);
8835 vector unsigned int vec_lvewx (int, unsigned int *);
8836 vector signed int vec_lvewx (int, long *);
8837 vector unsigned int vec_lvewx (int, unsigned long *);
8839 vector signed short vec_lvehx (int, short *);
8840 vector unsigned short vec_lvehx (int, unsigned short *);
8842 vector signed char vec_lvebx (int, char *);
8843 vector unsigned char vec_lvebx (int, unsigned char *);
8845 vector float vec_ldl (int, const vector float *);
8846 vector float vec_ldl (int, const float *);
8847 vector bool int vec_ldl (int, const vector bool int *);
8848 vector signed int vec_ldl (int, const vector signed int *);
8849 vector signed int vec_ldl (int, const int *);
8850 vector signed int vec_ldl (int, const long *);
8851 vector unsigned int vec_ldl (int, const vector unsigned int *);
8852 vector unsigned int vec_ldl (int, const unsigned int *);
8853 vector unsigned int vec_ldl (int, const unsigned long *);
8854 vector bool short vec_ldl (int, const vector bool short *);
8855 vector pixel vec_ldl (int, const vector pixel *);
8856 vector signed short vec_ldl (int, const vector signed short *);
8857 vector signed short vec_ldl (int, const short *);
8858 vector unsigned short vec_ldl (int, const vector unsigned short *);
8859 vector unsigned short vec_ldl (int, const unsigned short *);
8860 vector bool char vec_ldl (int, const vector bool char *);
8861 vector signed char vec_ldl (int, const vector signed char *);
8862 vector signed char vec_ldl (int, const signed char *);
8863 vector unsigned char vec_ldl (int, const vector unsigned char *);
8864 vector unsigned char vec_ldl (int, const unsigned char *);
8866 vector float vec_loge (vector float);
8868 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
8869 vector unsigned char vec_lvsl (int, const volatile signed char *);
8870 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
8871 vector unsigned char vec_lvsl (int, const volatile short *);
8872 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
8873 vector unsigned char vec_lvsl (int, const volatile int *);
8874 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
8875 vector unsigned char vec_lvsl (int, const volatile long *);
8876 vector unsigned char vec_lvsl (int, const volatile float *);
8878 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
8879 vector unsigned char vec_lvsr (int, const volatile signed char *);
8880 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
8881 vector unsigned char vec_lvsr (int, const volatile short *);
8882 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
8883 vector unsigned char vec_lvsr (int, const volatile int *);
8884 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
8885 vector unsigned char vec_lvsr (int, const volatile long *);
8886 vector unsigned char vec_lvsr (int, const volatile float *);
8888 vector float vec_madd (vector float, vector float, vector float);
8890 vector signed short vec_madds (vector signed short,
8891 vector signed short,
8892 vector signed short);
8894 vector unsigned char vec_max (vector bool char, vector unsigned char);
8895 vector unsigned char vec_max (vector unsigned char, vector bool char);
8896 vector unsigned char vec_max (vector unsigned char,
8897 vector unsigned char);
8898 vector signed char vec_max (vector bool char, vector signed char);
8899 vector signed char vec_max (vector signed char, vector bool char);
8900 vector signed char vec_max (vector signed char, vector signed char);
8901 vector unsigned short vec_max (vector bool short,
8902 vector unsigned short);
8903 vector unsigned short vec_max (vector unsigned short,
8905 vector unsigned short vec_max (vector unsigned short,
8906 vector unsigned short);
8907 vector signed short vec_max (vector bool short, vector signed short);
8908 vector signed short vec_max (vector signed short, vector bool short);
8909 vector signed short vec_max (vector signed short, vector signed short);
8910 vector unsigned int vec_max (vector bool int, vector unsigned int);
8911 vector unsigned int vec_max (vector unsigned int, vector bool int);
8912 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
8913 vector signed int vec_max (vector bool int, vector signed int);
8914 vector signed int vec_max (vector signed int, vector bool int);
8915 vector signed int vec_max (vector signed int, vector signed int);
8916 vector float vec_max (vector float, vector float);
8918 vector float vec_vmaxfp (vector float, vector float);
8920 vector signed int vec_vmaxsw (vector bool int, vector signed int);
8921 vector signed int vec_vmaxsw (vector signed int, vector bool int);
8922 vector signed int vec_vmaxsw (vector signed int, vector signed int);
8924 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
8925 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
8926 vector unsigned int vec_vmaxuw (vector unsigned int,
8927 vector unsigned int);
8929 vector signed short vec_vmaxsh (vector bool short, vector signed short);
8930 vector signed short vec_vmaxsh (vector signed short, vector bool short);
8931 vector signed short vec_vmaxsh (vector signed short,
8932 vector signed short);
8934 vector unsigned short vec_vmaxuh (vector bool short,
8935 vector unsigned short);
8936 vector unsigned short vec_vmaxuh (vector unsigned short,
8938 vector unsigned short vec_vmaxuh (vector unsigned short,
8939 vector unsigned short);
8941 vector signed char vec_vmaxsb (vector bool char, vector signed char);
8942 vector signed char vec_vmaxsb (vector signed char, vector bool char);
8943 vector signed char vec_vmaxsb (vector signed char, vector signed char);
8945 vector unsigned char vec_vmaxub (vector bool char,
8946 vector unsigned char);
8947 vector unsigned char vec_vmaxub (vector unsigned char,
8949 vector unsigned char vec_vmaxub (vector unsigned char,
8950 vector unsigned char);
8952 vector bool char vec_mergeh (vector bool char, vector bool char);
8953 vector signed char vec_mergeh (vector signed char, vector signed char);
8954 vector unsigned char vec_mergeh (vector unsigned char,
8955 vector unsigned char);
8956 vector bool short vec_mergeh (vector bool short, vector bool short);
8957 vector pixel vec_mergeh (vector pixel, vector pixel);
8958 vector signed short vec_mergeh (vector signed short,
8959 vector signed short);
8960 vector unsigned short vec_mergeh (vector unsigned short,
8961 vector unsigned short);
8962 vector float vec_mergeh (vector float, vector float);
8963 vector bool int vec_mergeh (vector bool int, vector bool int);
8964 vector signed int vec_mergeh (vector signed int, vector signed int);
8965 vector unsigned int vec_mergeh (vector unsigned int,
8966 vector unsigned int);
8968 vector float vec_vmrghw (vector float, vector float);
8969 vector bool int vec_vmrghw (vector bool int, vector bool int);
8970 vector signed int vec_vmrghw (vector signed int, vector signed int);
8971 vector unsigned int vec_vmrghw (vector unsigned int,
8972 vector unsigned int);
8974 vector bool short vec_vmrghh (vector bool short, vector bool short);
8975 vector signed short vec_vmrghh (vector signed short,
8976 vector signed short);
8977 vector unsigned short vec_vmrghh (vector unsigned short,
8978 vector unsigned short);
8979 vector pixel vec_vmrghh (vector pixel, vector pixel);
8981 vector bool char vec_vmrghb (vector bool char, vector bool char);
8982 vector signed char vec_vmrghb (vector signed char, vector signed char);
8983 vector unsigned char vec_vmrghb (vector unsigned char,
8984 vector unsigned char);
8986 vector bool char vec_mergel (vector bool char, vector bool char);
8987 vector signed char vec_mergel (vector signed char, vector signed char);
8988 vector unsigned char vec_mergel (vector unsigned char,
8989 vector unsigned char);
8990 vector bool short vec_mergel (vector bool short, vector bool short);
8991 vector pixel vec_mergel (vector pixel, vector pixel);
8992 vector signed short vec_mergel (vector signed short,
8993 vector signed short);
8994 vector unsigned short vec_mergel (vector unsigned short,
8995 vector unsigned short);
8996 vector float vec_mergel (vector float, vector float);
8997 vector bool int vec_mergel (vector bool int, vector bool int);
8998 vector signed int vec_mergel (vector signed int, vector signed int);
8999 vector unsigned int vec_mergel (vector unsigned int,
9000 vector unsigned int);
9002 vector float vec_vmrglw (vector float, vector float);
9003 vector signed int vec_vmrglw (vector signed int, vector signed int);
9004 vector unsigned int vec_vmrglw (vector unsigned int,
9005 vector unsigned int);
9006 vector bool int vec_vmrglw (vector bool int, vector bool int);
9008 vector bool short vec_vmrglh (vector bool short, vector bool short);
9009 vector signed short vec_vmrglh (vector signed short,
9010 vector signed short);
9011 vector unsigned short vec_vmrglh (vector unsigned short,
9012 vector unsigned short);
9013 vector pixel vec_vmrglh (vector pixel, vector pixel);
9015 vector bool char vec_vmrglb (vector bool char, vector bool char);
9016 vector signed char vec_vmrglb (vector signed char, vector signed char);
9017 vector unsigned char vec_vmrglb (vector unsigned char,
9018 vector unsigned char);
9020 vector unsigned short vec_mfvscr (void);
9022 vector unsigned char vec_min (vector bool char, vector unsigned char);
9023 vector unsigned char vec_min (vector unsigned char, vector bool char);
9024 vector unsigned char vec_min (vector unsigned char,
9025 vector unsigned char);
9026 vector signed char vec_min (vector bool char, vector signed char);
9027 vector signed char vec_min (vector signed char, vector bool char);
9028 vector signed char vec_min (vector signed char, vector signed char);
9029 vector unsigned short vec_min (vector bool short,
9030 vector unsigned short);
9031 vector unsigned short vec_min (vector unsigned short,
9033 vector unsigned short vec_min (vector unsigned short,
9034 vector unsigned short);
9035 vector signed short vec_min (vector bool short, vector signed short);
9036 vector signed short vec_min (vector signed short, vector bool short);
9037 vector signed short vec_min (vector signed short, vector signed short);
9038 vector unsigned int vec_min (vector bool int, vector unsigned int);
9039 vector unsigned int vec_min (vector unsigned int, vector bool int);
9040 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
9041 vector signed int vec_min (vector bool int, vector signed int);
9042 vector signed int vec_min (vector signed int, vector bool int);
9043 vector signed int vec_min (vector signed int, vector signed int);
9044 vector float vec_min (vector float, vector float);
9046 vector float vec_vminfp (vector float, vector float);
9048 vector signed int vec_vminsw (vector bool int, vector signed int);
9049 vector signed int vec_vminsw (vector signed int, vector bool int);
9050 vector signed int vec_vminsw (vector signed int, vector signed int);
9052 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
9053 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
9054 vector unsigned int vec_vminuw (vector unsigned int,
9055 vector unsigned int);
9057 vector signed short vec_vminsh (vector bool short, vector signed short);
9058 vector signed short vec_vminsh (vector signed short, vector bool short);
9059 vector signed short vec_vminsh (vector signed short,
9060 vector signed short);
9062 vector unsigned short vec_vminuh (vector bool short,
9063 vector unsigned short);
9064 vector unsigned short vec_vminuh (vector unsigned short,
9066 vector unsigned short vec_vminuh (vector unsigned short,
9067 vector unsigned short);
9069 vector signed char vec_vminsb (vector bool char, vector signed char);
9070 vector signed char vec_vminsb (vector signed char, vector bool char);
9071 vector signed char vec_vminsb (vector signed char, vector signed char);
9073 vector unsigned char vec_vminub (vector bool char,
9074 vector unsigned char);
9075 vector unsigned char vec_vminub (vector unsigned char,
9077 vector unsigned char vec_vminub (vector unsigned char,
9078 vector unsigned char);
9080 vector signed short vec_mladd (vector signed short,
9081 vector signed short,
9082 vector signed short);
9083 vector signed short vec_mladd (vector signed short,
9084 vector unsigned short,
9085 vector unsigned short);
9086 vector signed short vec_mladd (vector unsigned short,
9087 vector signed short,
9088 vector signed short);
9089 vector unsigned short vec_mladd (vector unsigned short,
9090 vector unsigned short,
9091 vector unsigned short);
9093 vector signed short vec_mradds (vector signed short,
9094 vector signed short,
9095 vector signed short);
9097 vector unsigned int vec_msum (vector unsigned char,
9098 vector unsigned char,
9099 vector unsigned int);
9100 vector signed int vec_msum (vector signed char,
9101 vector unsigned char,
9103 vector unsigned int vec_msum (vector unsigned short,
9104 vector unsigned short,
9105 vector unsigned int);
9106 vector signed int vec_msum (vector signed short,
9107 vector signed short,
9110 vector signed int vec_vmsumshm (vector signed short,
9111 vector signed short,
9114 vector unsigned int vec_vmsumuhm (vector unsigned short,
9115 vector unsigned short,
9116 vector unsigned int);
9118 vector signed int vec_vmsummbm (vector signed char,
9119 vector unsigned char,
9122 vector unsigned int vec_vmsumubm (vector unsigned char,
9123 vector unsigned char,
9124 vector unsigned int);
9126 vector unsigned int vec_msums (vector unsigned short,
9127 vector unsigned short,
9128 vector unsigned int);
9129 vector signed int vec_msums (vector signed short,
9130 vector signed short,
9133 vector signed int vec_vmsumshs (vector signed short,
9134 vector signed short,
9137 vector unsigned int vec_vmsumuhs (vector unsigned short,
9138 vector unsigned short,
9139 vector unsigned int);
9141 void vec_mtvscr (vector signed int);
9142 void vec_mtvscr (vector unsigned int);
9143 void vec_mtvscr (vector bool int);
9144 void vec_mtvscr (vector signed short);
9145 void vec_mtvscr (vector unsigned short);
9146 void vec_mtvscr (vector bool short);
9147 void vec_mtvscr (vector pixel);
9148 void vec_mtvscr (vector signed char);
9149 void vec_mtvscr (vector unsigned char);
9150 void vec_mtvscr (vector bool char);
9152 vector unsigned short vec_mule (vector unsigned char,
9153 vector unsigned char);
9154 vector signed short vec_mule (vector signed char,
9155 vector signed char);
9156 vector unsigned int vec_mule (vector unsigned short,
9157 vector unsigned short);
9158 vector signed int vec_mule (vector signed short, vector signed short);
9160 vector signed int vec_vmulesh (vector signed short,
9161 vector signed short);
9163 vector unsigned int vec_vmuleuh (vector unsigned short,
9164 vector unsigned short);
9166 vector signed short vec_vmulesb (vector signed char,
9167 vector signed char);
9169 vector unsigned short vec_vmuleub (vector unsigned char,
9170 vector unsigned char);
9172 vector unsigned short vec_mulo (vector unsigned char,
9173 vector unsigned char);
9174 vector signed short vec_mulo (vector signed char, vector signed char);
9175 vector unsigned int vec_mulo (vector unsigned short,
9176 vector unsigned short);
9177 vector signed int vec_mulo (vector signed short, vector signed short);
9179 vector signed int vec_vmulosh (vector signed short,
9180 vector signed short);
9182 vector unsigned int vec_vmulouh (vector unsigned short,
9183 vector unsigned short);
9185 vector signed short vec_vmulosb (vector signed char,
9186 vector signed char);
9188 vector unsigned short vec_vmuloub (vector unsigned char,
9189 vector unsigned char);
9191 vector float vec_nmsub (vector float, vector float, vector float);
9193 vector float vec_nor (vector float, vector float);
9194 vector signed int vec_nor (vector signed int, vector signed int);
9195 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
9196 vector bool int vec_nor (vector bool int, vector bool int);
9197 vector signed short vec_nor (vector signed short, vector signed short);
9198 vector unsigned short vec_nor (vector unsigned short,
9199 vector unsigned short);
9200 vector bool short vec_nor (vector bool short, vector bool short);
9201 vector signed char vec_nor (vector signed char, vector signed char);
9202 vector unsigned char vec_nor (vector unsigned char,
9203 vector unsigned char);
9204 vector bool char vec_nor (vector bool char, vector bool char);
9206 vector float vec_or (vector float, vector float);
9207 vector float vec_or (vector float, vector bool int);
9208 vector float vec_or (vector bool int, vector float);
9209 vector bool int vec_or (vector bool int, vector bool int);
9210 vector signed int vec_or (vector bool int, vector signed int);
9211 vector signed int vec_or (vector signed int, vector bool int);
9212 vector signed int vec_or (vector signed int, vector signed int);
9213 vector unsigned int vec_or (vector bool int, vector unsigned int);
9214 vector unsigned int vec_or (vector unsigned int, vector bool int);
9215 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
9216 vector bool short vec_or (vector bool short, vector bool short);
9217 vector signed short vec_or (vector bool short, vector signed short);
9218 vector signed short vec_or (vector signed short, vector bool short);
9219 vector signed short vec_or (vector signed short, vector signed short);
9220 vector unsigned short vec_or (vector bool short, vector unsigned short);
9221 vector unsigned short vec_or (vector unsigned short, vector bool short);
9222 vector unsigned short vec_or (vector unsigned short,
9223 vector unsigned short);
9224 vector signed char vec_or (vector bool char, vector signed char);
9225 vector bool char vec_or (vector bool char, vector bool char);
9226 vector signed char vec_or (vector signed char, vector bool char);
9227 vector signed char vec_or (vector signed char, vector signed char);
9228 vector unsigned char vec_or (vector bool char, vector unsigned char);
9229 vector unsigned char vec_or (vector unsigned char, vector bool char);
9230 vector unsigned char vec_or (vector unsigned char,
9231 vector unsigned char);
9233 vector signed char vec_pack (vector signed short, vector signed short);
9234 vector unsigned char vec_pack (vector unsigned short,
9235 vector unsigned short);
9236 vector bool char vec_pack (vector bool short, vector bool short);
9237 vector signed short vec_pack (vector signed int, vector signed int);
9238 vector unsigned short vec_pack (vector unsigned int,
9239 vector unsigned int);
9240 vector bool short vec_pack (vector bool int, vector bool int);
9242 vector bool short vec_vpkuwum (vector bool int, vector bool int);
9243 vector signed short vec_vpkuwum (vector signed int, vector signed int);
9244 vector unsigned short vec_vpkuwum (vector unsigned int,
9245 vector unsigned int);
9247 vector bool char vec_vpkuhum (vector bool short, vector bool short);
9248 vector signed char vec_vpkuhum (vector signed short,
9249 vector signed short);
9250 vector unsigned char vec_vpkuhum (vector unsigned short,
9251 vector unsigned short);
9253 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
9255 vector unsigned char vec_packs (vector unsigned short,
9256 vector unsigned short);
9257 vector signed char vec_packs (vector signed short, vector signed short);
9258 vector unsigned short vec_packs (vector unsigned int,
9259 vector unsigned int);
9260 vector signed short vec_packs (vector signed int, vector signed int);
9262 vector signed short vec_vpkswss (vector signed int, vector signed int);
9264 vector unsigned short vec_vpkuwus (vector unsigned int,
9265 vector unsigned int);
9267 vector signed char vec_vpkshss (vector signed short,
9268 vector signed short);
9270 vector unsigned char vec_vpkuhus (vector unsigned short,
9271 vector unsigned short);
9273 vector unsigned char vec_packsu (vector unsigned short,
9274 vector unsigned short);
9275 vector unsigned char vec_packsu (vector signed short,
9276 vector signed short);
9277 vector unsigned short vec_packsu (vector unsigned int,
9278 vector unsigned int);
9279 vector unsigned short vec_packsu (vector signed int, vector signed int);
9281 vector unsigned short vec_vpkswus (vector signed int,
9284 vector unsigned char vec_vpkshus (vector signed short,
9285 vector signed short);
9287 vector float vec_perm (vector float,
9289 vector unsigned char);
9290 vector signed int vec_perm (vector signed int,
9292 vector unsigned char);
9293 vector unsigned int vec_perm (vector unsigned int,
9294 vector unsigned int,
9295 vector unsigned char);
9296 vector bool int vec_perm (vector bool int,
9298 vector unsigned char);
9299 vector signed short vec_perm (vector signed short,
9300 vector signed short,
9301 vector unsigned char);
9302 vector unsigned short vec_perm (vector unsigned short,
9303 vector unsigned short,
9304 vector unsigned char);
9305 vector bool short vec_perm (vector bool short,
9307 vector unsigned char);
9308 vector pixel vec_perm (vector pixel,
9310 vector unsigned char);
9311 vector signed char vec_perm (vector signed char,
9313 vector unsigned char);
9314 vector unsigned char vec_perm (vector unsigned char,
9315 vector unsigned char,
9316 vector unsigned char);
9317 vector bool char vec_perm (vector bool char,
9319 vector unsigned char);
9321 vector float vec_re (vector float);
9323 vector signed char vec_rl (vector signed char,
9324 vector unsigned char);
9325 vector unsigned char vec_rl (vector unsigned char,
9326 vector unsigned char);
9327 vector signed short vec_rl (vector signed short, vector unsigned short);
9328 vector unsigned short vec_rl (vector unsigned short,
9329 vector unsigned short);
9330 vector signed int vec_rl (vector signed int, vector unsigned int);
9331 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
9333 vector signed int vec_vrlw (vector signed int, vector unsigned int);
9334 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
9336 vector signed short vec_vrlh (vector signed short,
9337 vector unsigned short);
9338 vector unsigned short vec_vrlh (vector unsigned short,
9339 vector unsigned short);
9341 vector signed char vec_vrlb (vector signed char, vector unsigned char);
9342 vector unsigned char vec_vrlb (vector unsigned char,
9343 vector unsigned char);
9345 vector float vec_round (vector float);
9347 vector float vec_rsqrte (vector float);
9349 vector float vec_sel (vector float, vector float, vector bool int);
9350 vector float vec_sel (vector float, vector float, vector unsigned int);
9351 vector signed int vec_sel (vector signed int,
9354 vector signed int vec_sel (vector signed int,
9356 vector unsigned int);
9357 vector unsigned int vec_sel (vector unsigned int,
9358 vector unsigned int,
9360 vector unsigned int vec_sel (vector unsigned int,
9361 vector unsigned int,
9362 vector unsigned int);
9363 vector bool int vec_sel (vector bool int,
9366 vector bool int vec_sel (vector bool int,
9368 vector unsigned int);
9369 vector signed short vec_sel (vector signed short,
9370 vector signed short,
9372 vector signed short vec_sel (vector signed short,
9373 vector signed short,
9374 vector unsigned short);
9375 vector unsigned short vec_sel (vector unsigned short,
9376 vector unsigned short,
9378 vector unsigned short vec_sel (vector unsigned short,
9379 vector unsigned short,
9380 vector unsigned short);
9381 vector bool short vec_sel (vector bool short,
9384 vector bool short vec_sel (vector bool short,
9386 vector unsigned short);
9387 vector signed char vec_sel (vector signed char,
9390 vector signed char vec_sel (vector signed char,
9392 vector unsigned char);
9393 vector unsigned char vec_sel (vector unsigned char,
9394 vector unsigned char,
9396 vector unsigned char vec_sel (vector unsigned char,
9397 vector unsigned char,
9398 vector unsigned char);
9399 vector bool char vec_sel (vector bool char,
9402 vector bool char vec_sel (vector bool char,
9404 vector unsigned char);
9406 vector signed char vec_sl (vector signed char,
9407 vector unsigned char);
9408 vector unsigned char vec_sl (vector unsigned char,
9409 vector unsigned char);
9410 vector signed short vec_sl (vector signed short, vector unsigned short);
9411 vector unsigned short vec_sl (vector unsigned short,
9412 vector unsigned short);
9413 vector signed int vec_sl (vector signed int, vector unsigned int);
9414 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
9416 vector signed int vec_vslw (vector signed int, vector unsigned int);
9417 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
9419 vector signed short vec_vslh (vector signed short,
9420 vector unsigned short);
9421 vector unsigned short vec_vslh (vector unsigned short,
9422 vector unsigned short);
9424 vector signed char vec_vslb (vector signed char, vector unsigned char);
9425 vector unsigned char vec_vslb (vector unsigned char,
9426 vector unsigned char);
9428 vector float vec_sld (vector float, vector float, const int);
9429 vector signed int vec_sld (vector signed int,
9432 vector unsigned int vec_sld (vector unsigned int,
9433 vector unsigned int,
9435 vector bool int vec_sld (vector bool int,
9438 vector signed short vec_sld (vector signed short,
9439 vector signed short,
9441 vector unsigned short vec_sld (vector unsigned short,
9442 vector unsigned short,
9444 vector bool short vec_sld (vector bool short,
9447 vector pixel vec_sld (vector pixel,
9450 vector signed char vec_sld (vector signed char,
9453 vector unsigned char vec_sld (vector unsigned char,
9454 vector unsigned char,
9456 vector bool char vec_sld (vector bool char,
9460 vector signed int vec_sll (vector signed int,
9461 vector unsigned int);
9462 vector signed int vec_sll (vector signed int,
9463 vector unsigned short);
9464 vector signed int vec_sll (vector signed int,
9465 vector unsigned char);
9466 vector unsigned int vec_sll (vector unsigned int,
9467 vector unsigned int);
9468 vector unsigned int vec_sll (vector unsigned int,
9469 vector unsigned short);
9470 vector unsigned int vec_sll (vector unsigned int,
9471 vector unsigned char);
9472 vector bool int vec_sll (vector bool int,
9473 vector unsigned int);
9474 vector bool int vec_sll (vector bool int,
9475 vector unsigned short);
9476 vector bool int vec_sll (vector bool int,
9477 vector unsigned char);
9478 vector signed short vec_sll (vector signed short,
9479 vector unsigned int);
9480 vector signed short vec_sll (vector signed short,
9481 vector unsigned short);
9482 vector signed short vec_sll (vector signed short,
9483 vector unsigned char);
9484 vector unsigned short vec_sll (vector unsigned short,
9485 vector unsigned int);
9486 vector unsigned short vec_sll (vector unsigned short,
9487 vector unsigned short);
9488 vector unsigned short vec_sll (vector unsigned short,
9489 vector unsigned char);
9490 vector bool short vec_sll (vector bool short, vector unsigned int);
9491 vector bool short vec_sll (vector bool short, vector unsigned short);
9492 vector bool short vec_sll (vector bool short, vector unsigned char);
9493 vector pixel vec_sll (vector pixel, vector unsigned int);
9494 vector pixel vec_sll (vector pixel, vector unsigned short);
9495 vector pixel vec_sll (vector pixel, vector unsigned char);
9496 vector signed char vec_sll (vector signed char, vector unsigned int);
9497 vector signed char vec_sll (vector signed char, vector unsigned short);
9498 vector signed char vec_sll (vector signed char, vector unsigned char);
9499 vector unsigned char vec_sll (vector unsigned char,
9500 vector unsigned int);
9501 vector unsigned char vec_sll (vector unsigned char,
9502 vector unsigned short);
9503 vector unsigned char vec_sll (vector unsigned char,
9504 vector unsigned char);
9505 vector bool char vec_sll (vector bool char, vector unsigned int);
9506 vector bool char vec_sll (vector bool char, vector unsigned short);
9507 vector bool char vec_sll (vector bool char, vector unsigned char);
9509 vector float vec_slo (vector float, vector signed char);
9510 vector float vec_slo (vector float, vector unsigned char);
9511 vector signed int vec_slo (vector signed int, vector signed char);
9512 vector signed int vec_slo (vector signed int, vector unsigned char);
9513 vector unsigned int vec_slo (vector unsigned int, vector signed char);
9514 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
9515 vector signed short vec_slo (vector signed short, vector signed char);
9516 vector signed short vec_slo (vector signed short, vector unsigned char);
9517 vector unsigned short vec_slo (vector unsigned short,
9518 vector signed char);
9519 vector unsigned short vec_slo (vector unsigned short,
9520 vector unsigned char);
9521 vector pixel vec_slo (vector pixel, vector signed char);
9522 vector pixel vec_slo (vector pixel, vector unsigned char);
9523 vector signed char vec_slo (vector signed char, vector signed char);
9524 vector signed char vec_slo (vector signed char, vector unsigned char);
9525 vector unsigned char vec_slo (vector unsigned char, vector signed char);
9526 vector unsigned char vec_slo (vector unsigned char,
9527 vector unsigned char);
9529 vector signed char vec_splat (vector signed char, const int);
9530 vector unsigned char vec_splat (vector unsigned char, const int);
9531 vector bool char vec_splat (vector bool char, const int);
9532 vector signed short vec_splat (vector signed short, const int);
9533 vector unsigned short vec_splat (vector unsigned short, const int);
9534 vector bool short vec_splat (vector bool short, const int);
9535 vector pixel vec_splat (vector pixel, const int);
9536 vector float vec_splat (vector float, const int);
9537 vector signed int vec_splat (vector signed int, const int);
9538 vector unsigned int vec_splat (vector unsigned int, const int);
9539 vector bool int vec_splat (vector bool int, const int);
9541 vector float vec_vspltw (vector float, const int);
9542 vector signed int vec_vspltw (vector signed int, const int);
9543 vector unsigned int vec_vspltw (vector unsigned int, const int);
9544 vector bool int vec_vspltw (vector bool int, const int);
9546 vector bool short vec_vsplth (vector bool short, const int);
9547 vector signed short vec_vsplth (vector signed short, const int);
9548 vector unsigned short vec_vsplth (vector unsigned short, const int);
9549 vector pixel vec_vsplth (vector pixel, const int);
9551 vector signed char vec_vspltb (vector signed char, const int);
9552 vector unsigned char vec_vspltb (vector unsigned char, const int);
9553 vector bool char vec_vspltb (vector bool char, const int);
9555 vector signed char vec_splat_s8 (const int);
9557 vector signed short vec_splat_s16 (const int);
9559 vector signed int vec_splat_s32 (const int);
9561 vector unsigned char vec_splat_u8 (const int);
9563 vector unsigned short vec_splat_u16 (const int);
9565 vector unsigned int vec_splat_u32 (const int);
9567 vector signed char vec_sr (vector signed char, vector unsigned char);
9568 vector unsigned char vec_sr (vector unsigned char,
9569 vector unsigned char);
9570 vector signed short vec_sr (vector signed short,
9571 vector unsigned short);
9572 vector unsigned short vec_sr (vector unsigned short,
9573 vector unsigned short);
9574 vector signed int vec_sr (vector signed int, vector unsigned int);
9575 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
9577 vector signed int vec_vsrw (vector signed int, vector unsigned int);
9578 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
9580 vector signed short vec_vsrh (vector signed short,
9581 vector unsigned short);
9582 vector unsigned short vec_vsrh (vector unsigned short,
9583 vector unsigned short);
9585 vector signed char vec_vsrb (vector signed char, vector unsigned char);
9586 vector unsigned char vec_vsrb (vector unsigned char,
9587 vector unsigned char);
9589 vector signed char vec_sra (vector signed char, vector unsigned char);
9590 vector unsigned char vec_sra (vector unsigned char,
9591 vector unsigned char);
9592 vector signed short vec_sra (vector signed short,
9593 vector unsigned short);
9594 vector unsigned short vec_sra (vector unsigned short,
9595 vector unsigned short);
9596 vector signed int vec_sra (vector signed int, vector unsigned int);
9597 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
9599 vector signed int vec_vsraw (vector signed int, vector unsigned int);
9600 vector unsigned int vec_vsraw (vector unsigned int,
9601 vector unsigned int);
9603 vector signed short vec_vsrah (vector signed short,
9604 vector unsigned short);
9605 vector unsigned short vec_vsrah (vector unsigned short,
9606 vector unsigned short);
9608 vector signed char vec_vsrab (vector signed char, vector unsigned char);
9609 vector unsigned char vec_vsrab (vector unsigned char,
9610 vector unsigned char);
9612 vector signed int vec_srl (vector signed int, vector unsigned int);
9613 vector signed int vec_srl (vector signed int, vector unsigned short);
9614 vector signed int vec_srl (vector signed int, vector unsigned char);
9615 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
9616 vector unsigned int vec_srl (vector unsigned int,
9617 vector unsigned short);
9618 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
9619 vector bool int vec_srl (vector bool int, vector unsigned int);
9620 vector bool int vec_srl (vector bool int, vector unsigned short);
9621 vector bool int vec_srl (vector bool int, vector unsigned char);
9622 vector signed short vec_srl (vector signed short, vector unsigned int);
9623 vector signed short vec_srl (vector signed short,
9624 vector unsigned short);
9625 vector signed short vec_srl (vector signed short, vector unsigned char);
9626 vector unsigned short vec_srl (vector unsigned short,
9627 vector unsigned int);
9628 vector unsigned short vec_srl (vector unsigned short,
9629 vector unsigned short);
9630 vector unsigned short vec_srl (vector unsigned short,
9631 vector unsigned char);
9632 vector bool short vec_srl (vector bool short, vector unsigned int);
9633 vector bool short vec_srl (vector bool short, vector unsigned short);
9634 vector bool short vec_srl (vector bool short, vector unsigned char);
9635 vector pixel vec_srl (vector pixel, vector unsigned int);
9636 vector pixel vec_srl (vector pixel, vector unsigned short);
9637 vector pixel vec_srl (vector pixel, vector unsigned char);
9638 vector signed char vec_srl (vector signed char, vector unsigned int);
9639 vector signed char vec_srl (vector signed char, vector unsigned short);
9640 vector signed char vec_srl (vector signed char, vector unsigned char);
9641 vector unsigned char vec_srl (vector unsigned char,
9642 vector unsigned int);
9643 vector unsigned char vec_srl (vector unsigned char,
9644 vector unsigned short);
9645 vector unsigned char vec_srl (vector unsigned char,
9646 vector unsigned char);
9647 vector bool char vec_srl (vector bool char, vector unsigned int);
9648 vector bool char vec_srl (vector bool char, vector unsigned short);
9649 vector bool char vec_srl (vector bool char, vector unsigned char);
9651 vector float vec_sro (vector float, vector signed char);
9652 vector float vec_sro (vector float, vector unsigned char);
9653 vector signed int vec_sro (vector signed int, vector signed char);
9654 vector signed int vec_sro (vector signed int, vector unsigned char);
9655 vector unsigned int vec_sro (vector unsigned int, vector signed char);
9656 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
9657 vector signed short vec_sro (vector signed short, vector signed char);
9658 vector signed short vec_sro (vector signed short, vector unsigned char);
9659 vector unsigned short vec_sro (vector unsigned short,
9660 vector signed char);
9661 vector unsigned short vec_sro (vector unsigned short,
9662 vector unsigned char);
9663 vector pixel vec_sro (vector pixel, vector signed char);
9664 vector pixel vec_sro (vector pixel, vector unsigned char);
9665 vector signed char vec_sro (vector signed char, vector signed char);
9666 vector signed char vec_sro (vector signed char, vector unsigned char);
9667 vector unsigned char vec_sro (vector unsigned char, vector signed char);
9668 vector unsigned char vec_sro (vector unsigned char,
9669 vector unsigned char);
9671 void vec_st (vector float, int, vector float *);
9672 void vec_st (vector float, int, float *);
9673 void vec_st (vector signed int, int, vector signed int *);
9674 void vec_st (vector signed int, int, int *);
9675 void vec_st (vector unsigned int, int, vector unsigned int *);
9676 void vec_st (vector unsigned int, int, unsigned int *);
9677 void vec_st (vector bool int, int, vector bool int *);
9678 void vec_st (vector bool int, int, unsigned int *);
9679 void vec_st (vector bool int, int, int *);
9680 void vec_st (vector signed short, int, vector signed short *);
9681 void vec_st (vector signed short, int, short *);
9682 void vec_st (vector unsigned short, int, vector unsigned short *);
9683 void vec_st (vector unsigned short, int, unsigned short *);
9684 void vec_st (vector bool short, int, vector bool short *);
9685 void vec_st (vector bool short, int, unsigned short *);
9686 void vec_st (vector pixel, int, vector pixel *);
9687 void vec_st (vector pixel, int, unsigned short *);
9688 void vec_st (vector pixel, int, short *);
9689 void vec_st (vector bool short, int, short *);
9690 void vec_st (vector signed char, int, vector signed char *);
9691 void vec_st (vector signed char, int, signed char *);
9692 void vec_st (vector unsigned char, int, vector unsigned char *);
9693 void vec_st (vector unsigned char, int, unsigned char *);
9694 void vec_st (vector bool char, int, vector bool char *);
9695 void vec_st (vector bool char, int, unsigned char *);
9696 void vec_st (vector bool char, int, signed char *);
9698 void vec_ste (vector signed char, int, signed char *);
9699 void vec_ste (vector unsigned char, int, unsigned char *);
9700 void vec_ste (vector bool char, int, signed char *);
9701 void vec_ste (vector bool char, int, unsigned char *);
9702 void vec_ste (vector signed short, int, short *);
9703 void vec_ste (vector unsigned short, int, unsigned short *);
9704 void vec_ste (vector bool short, int, short *);
9705 void vec_ste (vector bool short, int, unsigned short *);
9706 void vec_ste (vector pixel, int, short *);
9707 void vec_ste (vector pixel, int, unsigned short *);
9708 void vec_ste (vector float, int, float *);
9709 void vec_ste (vector signed int, int, int *);
9710 void vec_ste (vector unsigned int, int, unsigned int *);
9711 void vec_ste (vector bool int, int, int *);
9712 void vec_ste (vector bool int, int, unsigned int *);
9714 void vec_stvewx (vector float, int, float *);
9715 void vec_stvewx (vector signed int, int, int *);
9716 void vec_stvewx (vector unsigned int, int, unsigned int *);
9717 void vec_stvewx (vector bool int, int, int *);
9718 void vec_stvewx (vector bool int, int, unsigned int *);
9720 void vec_stvehx (vector signed short, int, short *);
9721 void vec_stvehx (vector unsigned short, int, unsigned short *);
9722 void vec_stvehx (vector bool short, int, short *);
9723 void vec_stvehx (vector bool short, int, unsigned short *);
9724 void vec_stvehx (vector pixel, int, short *);
9725 void vec_stvehx (vector pixel, int, unsigned short *);
9727 void vec_stvebx (vector signed char, int, signed char *);
9728 void vec_stvebx (vector unsigned char, int, unsigned char *);
9729 void vec_stvebx (vector bool char, int, signed char *);
9730 void vec_stvebx (vector bool char, int, unsigned char *);
9732 void vec_stl (vector float, int, vector float *);
9733 void vec_stl (vector float, int, float *);
9734 void vec_stl (vector signed int, int, vector signed int *);
9735 void vec_stl (vector signed int, int, int *);
9736 void vec_stl (vector unsigned int, int, vector unsigned int *);
9737 void vec_stl (vector unsigned int, int, unsigned int *);
9738 void vec_stl (vector bool int, int, vector bool int *);
9739 void vec_stl (vector bool int, int, unsigned int *);
9740 void vec_stl (vector bool int, int, int *);
9741 void vec_stl (vector signed short, int, vector signed short *);
9742 void vec_stl (vector signed short, int, short *);
9743 void vec_stl (vector unsigned short, int, vector unsigned short *);
9744 void vec_stl (vector unsigned short, int, unsigned short *);
9745 void vec_stl (vector bool short, int, vector bool short *);
9746 void vec_stl (vector bool short, int, unsigned short *);
9747 void vec_stl (vector bool short, int, short *);
9748 void vec_stl (vector pixel, int, vector pixel *);
9749 void vec_stl (vector pixel, int, unsigned short *);
9750 void vec_stl (vector pixel, int, short *);
9751 void vec_stl (vector signed char, int, vector signed char *);
9752 void vec_stl (vector signed char, int, signed char *);
9753 void vec_stl (vector unsigned char, int, vector unsigned char *);
9754 void vec_stl (vector unsigned char, int, unsigned char *);
9755 void vec_stl (vector bool char, int, vector bool char *);
9756 void vec_stl (vector bool char, int, unsigned char *);
9757 void vec_stl (vector bool char, int, signed char *);
9759 vector signed char vec_sub (vector bool char, vector signed char);
9760 vector signed char vec_sub (vector signed char, vector bool char);
9761 vector signed char vec_sub (vector signed char, vector signed char);
9762 vector unsigned char vec_sub (vector bool char, vector unsigned char);
9763 vector unsigned char vec_sub (vector unsigned char, vector bool char);
9764 vector unsigned char vec_sub (vector unsigned char,
9765 vector unsigned char);
9766 vector signed short vec_sub (vector bool short, vector signed short);
9767 vector signed short vec_sub (vector signed short, vector bool short);
9768 vector signed short vec_sub (vector signed short, vector signed short);
9769 vector unsigned short vec_sub (vector bool short,
9770 vector unsigned short);
9771 vector unsigned short vec_sub (vector unsigned short,
9773 vector unsigned short vec_sub (vector unsigned short,
9774 vector unsigned short);
9775 vector signed int vec_sub (vector bool int, vector signed int);
9776 vector signed int vec_sub (vector signed int, vector bool int);
9777 vector signed int vec_sub (vector signed int, vector signed int);
9778 vector unsigned int vec_sub (vector bool int, vector unsigned int);
9779 vector unsigned int vec_sub (vector unsigned int, vector bool int);
9780 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
9781 vector float vec_sub (vector float, vector float);
9783 vector float vec_vsubfp (vector float, vector float);
9785 vector signed int vec_vsubuwm (vector bool int, vector signed int);
9786 vector signed int vec_vsubuwm (vector signed int, vector bool int);
9787 vector signed int vec_vsubuwm (vector signed int, vector signed int);
9788 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
9789 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
9790 vector unsigned int vec_vsubuwm (vector unsigned int,
9791 vector unsigned int);
9793 vector signed short vec_vsubuhm (vector bool short,
9794 vector signed short);
9795 vector signed short vec_vsubuhm (vector signed short,
9797 vector signed short vec_vsubuhm (vector signed short,
9798 vector signed short);
9799 vector unsigned short vec_vsubuhm (vector bool short,
9800 vector unsigned short);
9801 vector unsigned short vec_vsubuhm (vector unsigned short,
9803 vector unsigned short vec_vsubuhm (vector unsigned short,
9804 vector unsigned short);
9806 vector signed char vec_vsububm (vector bool char, vector signed char);
9807 vector signed char vec_vsububm (vector signed char, vector bool char);
9808 vector signed char vec_vsububm (vector signed char, vector signed char);
9809 vector unsigned char vec_vsububm (vector bool char,
9810 vector unsigned char);
9811 vector unsigned char vec_vsububm (vector unsigned char,
9813 vector unsigned char vec_vsububm (vector unsigned char,
9814 vector unsigned char);
9816 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
9818 vector unsigned char vec_subs (vector bool char, vector unsigned char);
9819 vector unsigned char vec_subs (vector unsigned char, vector bool char);
9820 vector unsigned char vec_subs (vector unsigned char,
9821 vector unsigned char);
9822 vector signed char vec_subs (vector bool char, vector signed char);
9823 vector signed char vec_subs (vector signed char, vector bool char);
9824 vector signed char vec_subs (vector signed char, vector signed char);
9825 vector unsigned short vec_subs (vector bool short,
9826 vector unsigned short);
9827 vector unsigned short vec_subs (vector unsigned short,
9829 vector unsigned short vec_subs (vector unsigned short,
9830 vector unsigned short);
9831 vector signed short vec_subs (vector bool short, vector signed short);
9832 vector signed short vec_subs (vector signed short, vector bool short);
9833 vector signed short vec_subs (vector signed short, vector signed short);
9834 vector unsigned int vec_subs (vector bool int, vector unsigned int);
9835 vector unsigned int vec_subs (vector unsigned int, vector bool int);
9836 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
9837 vector signed int vec_subs (vector bool int, vector signed int);
9838 vector signed int vec_subs (vector signed int, vector bool int);
9839 vector signed int vec_subs (vector signed int, vector signed int);
9841 vector signed int vec_vsubsws (vector bool int, vector signed int);
9842 vector signed int vec_vsubsws (vector signed int, vector bool int);
9843 vector signed int vec_vsubsws (vector signed int, vector signed int);
9845 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
9846 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
9847 vector unsigned int vec_vsubuws (vector unsigned int,
9848 vector unsigned int);
9850 vector signed short vec_vsubshs (vector bool short,
9851 vector signed short);
9852 vector signed short vec_vsubshs (vector signed short,
9854 vector signed short vec_vsubshs (vector signed short,
9855 vector signed short);
9857 vector unsigned short vec_vsubuhs (vector bool short,
9858 vector unsigned short);
9859 vector unsigned short vec_vsubuhs (vector unsigned short,
9861 vector unsigned short vec_vsubuhs (vector unsigned short,
9862 vector unsigned short);
9864 vector signed char vec_vsubsbs (vector bool char, vector signed char);
9865 vector signed char vec_vsubsbs (vector signed char, vector bool char);
9866 vector signed char vec_vsubsbs (vector signed char, vector signed char);
9868 vector unsigned char vec_vsububs (vector bool char,
9869 vector unsigned char);
9870 vector unsigned char vec_vsububs (vector unsigned char,
9872 vector unsigned char vec_vsububs (vector unsigned char,
9873 vector unsigned char);
9875 vector unsigned int vec_sum4s (vector unsigned char,
9876 vector unsigned int);
9877 vector signed int vec_sum4s (vector signed char, vector signed int);
9878 vector signed int vec_sum4s (vector signed short, vector signed int);
9880 vector signed int vec_vsum4shs (vector signed short, vector signed int);
9882 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
9884 vector unsigned int vec_vsum4ubs (vector unsigned char,
9885 vector unsigned int);
9887 vector signed int vec_sum2s (vector signed int, vector signed int);
9889 vector signed int vec_sums (vector signed int, vector signed int);
9891 vector float vec_trunc (vector float);
9893 vector signed short vec_unpackh (vector signed char);
9894 vector bool short vec_unpackh (vector bool char);
9895 vector signed int vec_unpackh (vector signed short);
9896 vector bool int vec_unpackh (vector bool short);
9897 vector unsigned int vec_unpackh (vector pixel);
9899 vector bool int vec_vupkhsh (vector bool short);
9900 vector signed int vec_vupkhsh (vector signed short);
9902 vector unsigned int vec_vupkhpx (vector pixel);
9904 vector bool short vec_vupkhsb (vector bool char);
9905 vector signed short vec_vupkhsb (vector signed char);
9907 vector signed short vec_unpackl (vector signed char);
9908 vector bool short vec_unpackl (vector bool char);
9909 vector unsigned int vec_unpackl (vector pixel);
9910 vector signed int vec_unpackl (vector signed short);
9911 vector bool int vec_unpackl (vector bool short);
9913 vector unsigned int vec_vupklpx (vector pixel);
9915 vector bool int vec_vupklsh (vector bool short);
9916 vector signed int vec_vupklsh (vector signed short);
9918 vector bool short vec_vupklsb (vector bool char);
9919 vector signed short vec_vupklsb (vector signed char);
9921 vector float vec_xor (vector float, vector float);
9922 vector float vec_xor (vector float, vector bool int);
9923 vector float vec_xor (vector bool int, vector float);
9924 vector bool int vec_xor (vector bool int, vector bool int);
9925 vector signed int vec_xor (vector bool int, vector signed int);
9926 vector signed int vec_xor (vector signed int, vector bool int);
9927 vector signed int vec_xor (vector signed int, vector signed int);
9928 vector unsigned int vec_xor (vector bool int, vector unsigned int);
9929 vector unsigned int vec_xor (vector unsigned int, vector bool int);
9930 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
9931 vector bool short vec_xor (vector bool short, vector bool short);
9932 vector signed short vec_xor (vector bool short, vector signed short);
9933 vector signed short vec_xor (vector signed short, vector bool short);
9934 vector signed short vec_xor (vector signed short, vector signed short);
9935 vector unsigned short vec_xor (vector bool short,
9936 vector unsigned short);
9937 vector unsigned short vec_xor (vector unsigned short,
9939 vector unsigned short vec_xor (vector unsigned short,
9940 vector unsigned short);
9941 vector signed char vec_xor (vector bool char, vector signed char);
9942 vector bool char vec_xor (vector bool char, vector bool char);
9943 vector signed char vec_xor (vector signed char, vector bool char);
9944 vector signed char vec_xor (vector signed char, vector signed char);
9945 vector unsigned char vec_xor (vector bool char, vector unsigned char);
9946 vector unsigned char vec_xor (vector unsigned char, vector bool char);
9947 vector unsigned char vec_xor (vector unsigned char,
9948 vector unsigned char);
9950 int vec_all_eq (vector signed char, vector bool char);
9951 int vec_all_eq (vector signed char, vector signed char);
9952 int vec_all_eq (vector unsigned char, vector bool char);
9953 int vec_all_eq (vector unsigned char, vector unsigned char);
9954 int vec_all_eq (vector bool char, vector bool char);
9955 int vec_all_eq (vector bool char, vector unsigned char);
9956 int vec_all_eq (vector bool char, vector signed char);
9957 int vec_all_eq (vector signed short, vector bool short);
9958 int vec_all_eq (vector signed short, vector signed short);
9959 int vec_all_eq (vector unsigned short, vector bool short);
9960 int vec_all_eq (vector unsigned short, vector unsigned short);
9961 int vec_all_eq (vector bool short, vector bool short);
9962 int vec_all_eq (vector bool short, vector unsigned short);
9963 int vec_all_eq (vector bool short, vector signed short);
9964 int vec_all_eq (vector pixel, vector pixel);
9965 int vec_all_eq (vector signed int, vector bool int);
9966 int vec_all_eq (vector signed int, vector signed int);
9967 int vec_all_eq (vector unsigned int, vector bool int);
9968 int vec_all_eq (vector unsigned int, vector unsigned int);
9969 int vec_all_eq (vector bool int, vector bool int);
9970 int vec_all_eq (vector bool int, vector unsigned int);
9971 int vec_all_eq (vector bool int, vector signed int);
9972 int vec_all_eq (vector float, vector float);
9974 int vec_all_ge (vector bool char, vector unsigned char);
9975 int vec_all_ge (vector unsigned char, vector bool char);
9976 int vec_all_ge (vector unsigned char, vector unsigned char);
9977 int vec_all_ge (vector bool char, vector signed char);
9978 int vec_all_ge (vector signed char, vector bool char);
9979 int vec_all_ge (vector signed char, vector signed char);
9980 int vec_all_ge (vector bool short, vector unsigned short);
9981 int vec_all_ge (vector unsigned short, vector bool short);
9982 int vec_all_ge (vector unsigned short, vector unsigned short);
9983 int vec_all_ge (vector signed short, vector signed short);
9984 int vec_all_ge (vector bool short, vector signed short);
9985 int vec_all_ge (vector signed short, vector bool short);
9986 int vec_all_ge (vector bool int, vector unsigned int);
9987 int vec_all_ge (vector unsigned int, vector bool int);
9988 int vec_all_ge (vector unsigned int, vector unsigned int);
9989 int vec_all_ge (vector bool int, vector signed int);
9990 int vec_all_ge (vector signed int, vector bool int);
9991 int vec_all_ge (vector signed int, vector signed int);
9992 int vec_all_ge (vector float, vector float);
9994 int vec_all_gt (vector bool char, vector unsigned char);
9995 int vec_all_gt (vector unsigned char, vector bool char);
9996 int vec_all_gt (vector unsigned char, vector unsigned char);
9997 int vec_all_gt (vector bool char, vector signed char);
9998 int vec_all_gt (vector signed char, vector bool char);
9999 int vec_all_gt (vector signed char, vector signed char);
10000 int vec_all_gt (vector bool short, vector unsigned short);
10001 int vec_all_gt (vector unsigned short, vector bool short);
10002 int vec_all_gt (vector unsigned short, vector unsigned short);
10003 int vec_all_gt (vector bool short, vector signed short);
10004 int vec_all_gt (vector signed short, vector bool short);
10005 int vec_all_gt (vector signed short, vector signed short);
10006 int vec_all_gt (vector bool int, vector unsigned int);
10007 int vec_all_gt (vector unsigned int, vector bool int);
10008 int vec_all_gt (vector unsigned int, vector unsigned int);
10009 int vec_all_gt (vector bool int, vector signed int);
10010 int vec_all_gt (vector signed int, vector bool int);
10011 int vec_all_gt (vector signed int, vector signed int);
10012 int vec_all_gt (vector float, vector float);
10014 int vec_all_in (vector float, vector float);
10016 int vec_all_le (vector bool char, vector unsigned char);
10017 int vec_all_le (vector unsigned char, vector bool char);
10018 int vec_all_le (vector unsigned char, vector unsigned char);
10019 int vec_all_le (vector bool char, vector signed char);
10020 int vec_all_le (vector signed char, vector bool char);
10021 int vec_all_le (vector signed char, vector signed char);
10022 int vec_all_le (vector bool short, vector unsigned short);
10023 int vec_all_le (vector unsigned short, vector bool short);
10024 int vec_all_le (vector unsigned short, vector unsigned short);
10025 int vec_all_le (vector bool short, vector signed short);
10026 int vec_all_le (vector signed short, vector bool short);
10027 int vec_all_le (vector signed short, vector signed short);
10028 int vec_all_le (vector bool int, vector unsigned int);
10029 int vec_all_le (vector unsigned int, vector bool int);
10030 int vec_all_le (vector unsigned int, vector unsigned int);
10031 int vec_all_le (vector bool int, vector signed int);
10032 int vec_all_le (vector signed int, vector bool int);
10033 int vec_all_le (vector signed int, vector signed int);
10034 int vec_all_le (vector float, vector float);
10036 int vec_all_lt (vector bool char, vector unsigned char);
10037 int vec_all_lt (vector unsigned char, vector bool char);
10038 int vec_all_lt (vector unsigned char, vector unsigned char);
10039 int vec_all_lt (vector bool char, vector signed char);
10040 int vec_all_lt (vector signed char, vector bool char);
10041 int vec_all_lt (vector signed char, vector signed char);
10042 int vec_all_lt (vector bool short, vector unsigned short);
10043 int vec_all_lt (vector unsigned short, vector bool short);
10044 int vec_all_lt (vector unsigned short, vector unsigned short);
10045 int vec_all_lt (vector bool short, vector signed short);
10046 int vec_all_lt (vector signed short, vector bool short);
10047 int vec_all_lt (vector signed short, vector signed short);
10048 int vec_all_lt (vector bool int, vector unsigned int);
10049 int vec_all_lt (vector unsigned int, vector bool int);
10050 int vec_all_lt (vector unsigned int, vector unsigned int);
10051 int vec_all_lt (vector bool int, vector signed int);
10052 int vec_all_lt (vector signed int, vector bool int);
10053 int vec_all_lt (vector signed int, vector signed int);
10054 int vec_all_lt (vector float, vector float);
10056 int vec_all_nan (vector float);
10058 int vec_all_ne (vector signed char, vector bool char);
10059 int vec_all_ne (vector signed char, vector signed char);
10060 int vec_all_ne (vector unsigned char, vector bool char);
10061 int vec_all_ne (vector unsigned char, vector unsigned char);
10062 int vec_all_ne (vector bool char, vector bool char);
10063 int vec_all_ne (vector bool char, vector unsigned char);
10064 int vec_all_ne (vector bool char, vector signed char);
10065 int vec_all_ne (vector signed short, vector bool short);
10066 int vec_all_ne (vector signed short, vector signed short);
10067 int vec_all_ne (vector unsigned short, vector bool short);
10068 int vec_all_ne (vector unsigned short, vector unsigned short);
10069 int vec_all_ne (vector bool short, vector bool short);
10070 int vec_all_ne (vector bool short, vector unsigned short);
10071 int vec_all_ne (vector bool short, vector signed short);
10072 int vec_all_ne (vector pixel, vector pixel);
10073 int vec_all_ne (vector signed int, vector bool int);
10074 int vec_all_ne (vector signed int, vector signed int);
10075 int vec_all_ne (vector unsigned int, vector bool int);
10076 int vec_all_ne (vector unsigned int, vector unsigned int);
10077 int vec_all_ne (vector bool int, vector bool int);
10078 int vec_all_ne (vector bool int, vector unsigned int);
10079 int vec_all_ne (vector bool int, vector signed int);
10080 int vec_all_ne (vector float, vector float);
10082 int vec_all_nge (vector float, vector float);
10084 int vec_all_ngt (vector float, vector float);
10086 int vec_all_nle (vector float, vector float);
10088 int vec_all_nlt (vector float, vector float);
10090 int vec_all_numeric (vector float);
10092 int vec_any_eq (vector signed char, vector bool char);
10093 int vec_any_eq (vector signed char, vector signed char);
10094 int vec_any_eq (vector unsigned char, vector bool char);
10095 int vec_any_eq (vector unsigned char, vector unsigned char);
10096 int vec_any_eq (vector bool char, vector bool char);
10097 int vec_any_eq (vector bool char, vector unsigned char);
10098 int vec_any_eq (vector bool char, vector signed char);
10099 int vec_any_eq (vector signed short, vector bool short);
10100 int vec_any_eq (vector signed short, vector signed short);
10101 int vec_any_eq (vector unsigned short, vector bool short);
10102 int vec_any_eq (vector unsigned short, vector unsigned short);
10103 int vec_any_eq (vector bool short, vector bool short);
10104 int vec_any_eq (vector bool short, vector unsigned short);
10105 int vec_any_eq (vector bool short, vector signed short);
10106 int vec_any_eq (vector pixel, vector pixel);
10107 int vec_any_eq (vector signed int, vector bool int);
10108 int vec_any_eq (vector signed int, vector signed int);
10109 int vec_any_eq (vector unsigned int, vector bool int);
10110 int vec_any_eq (vector unsigned int, vector unsigned int);
10111 int vec_any_eq (vector bool int, vector bool int);
10112 int vec_any_eq (vector bool int, vector unsigned int);
10113 int vec_any_eq (vector bool int, vector signed int);
10114 int vec_any_eq (vector float, vector float);
10116 int vec_any_ge (vector signed char, vector bool char);
10117 int vec_any_ge (vector unsigned char, vector bool char);
10118 int vec_any_ge (vector unsigned char, vector unsigned char);
10119 int vec_any_ge (vector signed char, vector signed char);
10120 int vec_any_ge (vector bool char, vector unsigned char);
10121 int vec_any_ge (vector bool char, vector signed char);
10122 int vec_any_ge (vector unsigned short, vector bool short);
10123 int vec_any_ge (vector unsigned short, vector unsigned short);
10124 int vec_any_ge (vector signed short, vector signed short);
10125 int vec_any_ge (vector signed short, vector bool short);
10126 int vec_any_ge (vector bool short, vector unsigned short);
10127 int vec_any_ge (vector bool short, vector signed short);
10128 int vec_any_ge (vector signed int, vector bool int);
10129 int vec_any_ge (vector unsigned int, vector bool int);
10130 int vec_any_ge (vector unsigned int, vector unsigned int);
10131 int vec_any_ge (vector signed int, vector signed int);
10132 int vec_any_ge (vector bool int, vector unsigned int);
10133 int vec_any_ge (vector bool int, vector signed int);
10134 int vec_any_ge (vector float, vector float);
10136 int vec_any_gt (vector bool char, vector unsigned char);
10137 int vec_any_gt (vector unsigned char, vector bool char);
10138 int vec_any_gt (vector unsigned char, vector unsigned char);
10139 int vec_any_gt (vector bool char, vector signed char);
10140 int vec_any_gt (vector signed char, vector bool char);
10141 int vec_any_gt (vector signed char, vector signed char);
10142 int vec_any_gt (vector bool short, vector unsigned short);
10143 int vec_any_gt (vector unsigned short, vector bool short);
10144 int vec_any_gt (vector unsigned short, vector unsigned short);
10145 int vec_any_gt (vector bool short, vector signed short);
10146 int vec_any_gt (vector signed short, vector bool short);
10147 int vec_any_gt (vector signed short, vector signed short);
10148 int vec_any_gt (vector bool int, vector unsigned int);
10149 int vec_any_gt (vector unsigned int, vector bool int);
10150 int vec_any_gt (vector unsigned int, vector unsigned int);
10151 int vec_any_gt (vector bool int, vector signed int);
10152 int vec_any_gt (vector signed int, vector bool int);
10153 int vec_any_gt (vector signed int, vector signed int);
10154 int vec_any_gt (vector float, vector float);
10156 int vec_any_le (vector bool char, vector unsigned char);
10157 int vec_any_le (vector unsigned char, vector bool char);
10158 int vec_any_le (vector unsigned char, vector unsigned char);
10159 int vec_any_le (vector bool char, vector signed char);
10160 int vec_any_le (vector signed char, vector bool char);
10161 int vec_any_le (vector signed char, vector signed char);
10162 int vec_any_le (vector bool short, vector unsigned short);
10163 int vec_any_le (vector unsigned short, vector bool short);
10164 int vec_any_le (vector unsigned short, vector unsigned short);
10165 int vec_any_le (vector bool short, vector signed short);
10166 int vec_any_le (vector signed short, vector bool short);
10167 int vec_any_le (vector signed short, vector signed short);
10168 int vec_any_le (vector bool int, vector unsigned int);
10169 int vec_any_le (vector unsigned int, vector bool int);
10170 int vec_any_le (vector unsigned int, vector unsigned int);
10171 int vec_any_le (vector bool int, vector signed int);
10172 int vec_any_le (vector signed int, vector bool int);
10173 int vec_any_le (vector signed int, vector signed int);
10174 int vec_any_le (vector float, vector float);
10176 int vec_any_lt (vector bool char, vector unsigned char);
10177 int vec_any_lt (vector unsigned char, vector bool char);
10178 int vec_any_lt (vector unsigned char, vector unsigned char);
10179 int vec_any_lt (vector bool char, vector signed char);
10180 int vec_any_lt (vector signed char, vector bool char);
10181 int vec_any_lt (vector signed char, vector signed char);
10182 int vec_any_lt (vector bool short, vector unsigned short);
10183 int vec_any_lt (vector unsigned short, vector bool short);
10184 int vec_any_lt (vector unsigned short, vector unsigned short);
10185 int vec_any_lt (vector bool short, vector signed short);
10186 int vec_any_lt (vector signed short, vector bool short);
10187 int vec_any_lt (vector signed short, vector signed short);
10188 int vec_any_lt (vector bool int, vector unsigned int);
10189 int vec_any_lt (vector unsigned int, vector bool int);
10190 int vec_any_lt (vector unsigned int, vector unsigned int);
10191 int vec_any_lt (vector bool int, vector signed int);
10192 int vec_any_lt (vector signed int, vector bool int);
10193 int vec_any_lt (vector signed int, vector signed int);
10194 int vec_any_lt (vector float, vector float);
10196 int vec_any_nan (vector float);
10198 int vec_any_ne (vector signed char, vector bool char);
10199 int vec_any_ne (vector signed char, vector signed char);
10200 int vec_any_ne (vector unsigned char, vector bool char);
10201 int vec_any_ne (vector unsigned char, vector unsigned char);
10202 int vec_any_ne (vector bool char, vector bool char);
10203 int vec_any_ne (vector bool char, vector unsigned char);
10204 int vec_any_ne (vector bool char, vector signed char);
10205 int vec_any_ne (vector signed short, vector bool short);
10206 int vec_any_ne (vector signed short, vector signed short);
10207 int vec_any_ne (vector unsigned short, vector bool short);
10208 int vec_any_ne (vector unsigned short, vector unsigned short);
10209 int vec_any_ne (vector bool short, vector bool short);
10210 int vec_any_ne (vector bool short, vector unsigned short);
10211 int vec_any_ne (vector bool short, vector signed short);
10212 int vec_any_ne (vector pixel, vector pixel);
10213 int vec_any_ne (vector signed int, vector bool int);
10214 int vec_any_ne (vector signed int, vector signed int);
10215 int vec_any_ne (vector unsigned int, vector bool int);
10216 int vec_any_ne (vector unsigned int, vector unsigned int);
10217 int vec_any_ne (vector bool int, vector bool int);
10218 int vec_any_ne (vector bool int, vector unsigned int);
10219 int vec_any_ne (vector bool int, vector signed int);
10220 int vec_any_ne (vector float, vector float);
10222 int vec_any_nge (vector float, vector float);
10224 int vec_any_ngt (vector float, vector float);
10226 int vec_any_nle (vector float, vector float);
10228 int vec_any_nlt (vector float, vector float);
10230 int vec_any_numeric (vector float);
10232 int vec_any_out (vector float, vector float);
10235 @node SPARC VIS Built-in Functions
10236 @subsection SPARC VIS Built-in Functions
10238 GCC supports SIMD operations on the SPARC using both the generic vector
10239 extensions (@pxref{Vector Extensions}) as well as built-in functions for
10240 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
10241 switch, the VIS extension is exposed as the following built-in functions:
10244 typedef int v2si __attribute__ ((vector_size (8)));
10245 typedef short v4hi __attribute__ ((vector_size (8)));
10246 typedef short v2hi __attribute__ ((vector_size (4)));
10247 typedef char v8qi __attribute__ ((vector_size (8)));
10248 typedef char v4qi __attribute__ ((vector_size (4)));
10250 void * __builtin_vis_alignaddr (void *, long);
10251 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
10252 v2si __builtin_vis_faligndatav2si (v2si, v2si);
10253 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
10254 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
10256 v4hi __builtin_vis_fexpand (v4qi);
10258 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
10259 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
10260 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
10261 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
10262 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
10263 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
10264 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
10266 v4qi __builtin_vis_fpack16 (v4hi);
10267 v8qi __builtin_vis_fpack32 (v2si, v2si);
10268 v2hi __builtin_vis_fpackfix (v2si);
10269 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
10271 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
10274 @node SPU Built-in Functions
10275 @subsection SPU Built-in Functions
10277 GCC provides extensions for the SPU processor as described in the
10278 Sony/Toshiba/IBM SPU Language Extensions Specification, which can be
10279 found at @uref{http://cell.scei.co.jp/} or
10280 @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's
10281 implementation differs in several ways.
10286 The optional extension of specifying vector constants in parentheses is
10290 A vector initializer requires no cast if the vector constant is of the
10291 same type as the variable it is initializing.
10294 If @code{signed} or @code{unsigned} is omitted, the signedness of the
10295 vector type is the default signedness of the base type. The default
10296 varies depending on the operating system, so a portable program should
10297 always specify the signedness.
10300 By default, the keyword @code{__vector} is added. The macro
10301 @code{vector} is defined in @code{<spu_intrinsics.h>} and can be
10305 GCC allows using a @code{typedef} name as the type specifier for a
10309 For C, overloaded functions are implemented with macros so the following
10313 spu_add ((vector signed int)@{1, 2, 3, 4@}, foo);
10316 Since @code{spu_add} is a macro, the vector constant in the example
10317 is treated as four separate arguments. Wrap the entire argument in
10318 parentheses for this to work.
10321 The extended version of @code{__builtin_expect} is not supported.
10325 @emph{Note:} Only the interface described in the aforementioned
10326 specification is supported. Internally, GCC uses built-in functions to
10327 implement the required functionality, but these are not supported and
10328 are subject to change without notice.
10330 @node Target Format Checks
10331 @section Format Checks Specific to Particular Target Machines
10333 For some target machines, GCC supports additional options to the
10335 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
10338 * Solaris Format Checks::
10341 @node Solaris Format Checks
10342 @subsection Solaris Format Checks
10344 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
10345 check. @code{cmn_err} accepts a subset of the standard @code{printf}
10346 conversions, and the two-argument @code{%b} conversion for displaying
10347 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
10350 @section Pragmas Accepted by GCC
10354 GCC supports several types of pragmas, primarily in order to compile
10355 code originally written for other compilers. Note that in general
10356 we do not recommend the use of pragmas; @xref{Function Attributes},
10357 for further explanation.
10362 * RS/6000 and PowerPC Pragmas::
10364 * Solaris Pragmas::
10365 * Symbol-Renaming Pragmas::
10366 * Structure-Packing Pragmas::
10368 * Diagnostic Pragmas::
10369 * Visibility Pragmas::
10373 @subsection ARM Pragmas
10375 The ARM target defines pragmas for controlling the default addition of
10376 @code{long_call} and @code{short_call} attributes to functions.
10377 @xref{Function Attributes}, for information about the effects of these
10382 @cindex pragma, long_calls
10383 Set all subsequent functions to have the @code{long_call} attribute.
10385 @item no_long_calls
10386 @cindex pragma, no_long_calls
10387 Set all subsequent functions to have the @code{short_call} attribute.
10389 @item long_calls_off
10390 @cindex pragma, long_calls_off
10391 Do not affect the @code{long_call} or @code{short_call} attributes of
10392 subsequent functions.
10396 @subsection M32C Pragmas
10399 @item memregs @var{number}
10400 @cindex pragma, memregs
10401 Overrides the command line option @code{-memregs=} for the current
10402 file. Use with care! This pragma must be before any function in the
10403 file, and mixing different memregs values in different objects may
10404 make them incompatible. This pragma is useful when a
10405 performance-critical function uses a memreg for temporary values,
10406 as it may allow you to reduce the number of memregs used.
10410 @node RS/6000 and PowerPC Pragmas
10411 @subsection RS/6000 and PowerPC Pragmas
10413 The RS/6000 and PowerPC targets define one pragma for controlling
10414 whether or not the @code{longcall} attribute is added to function
10415 declarations by default. This pragma overrides the @option{-mlongcall}
10416 option, but not the @code{longcall} and @code{shortcall} attributes.
10417 @xref{RS/6000 and PowerPC Options}, for more information about when long
10418 calls are and are not necessary.
10422 @cindex pragma, longcall
10423 Apply the @code{longcall} attribute to all subsequent function
10427 Do not apply the @code{longcall} attribute to subsequent function
10431 @c Describe c4x pragmas here.
10432 @c Describe h8300 pragmas here.
10433 @c Describe sh pragmas here.
10434 @c Describe v850 pragmas here.
10436 @node Darwin Pragmas
10437 @subsection Darwin Pragmas
10439 The following pragmas are available for all architectures running the
10440 Darwin operating system. These are useful for compatibility with other
10444 @item mark @var{tokens}@dots{}
10445 @cindex pragma, mark
10446 This pragma is accepted, but has no effect.
10448 @item options align=@var{alignment}
10449 @cindex pragma, options align
10450 This pragma sets the alignment of fields in structures. The values of
10451 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
10452 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
10453 properly; to restore the previous setting, use @code{reset} for the
10456 @item segment @var{tokens}@dots{}
10457 @cindex pragma, segment
10458 This pragma is accepted, but has no effect.
10460 @item unused (@var{var} [, @var{var}]@dots{})
10461 @cindex pragma, unused
10462 This pragma declares variables to be possibly unused. GCC will not
10463 produce warnings for the listed variables. The effect is similar to
10464 that of the @code{unused} attribute, except that this pragma may appear
10465 anywhere within the variables' scopes.
10468 @node Solaris Pragmas
10469 @subsection Solaris Pragmas
10471 The Solaris target supports @code{#pragma redefine_extname}
10472 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
10473 @code{#pragma} directives for compatibility with the system compiler.
10476 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
10477 @cindex pragma, align
10479 Increase the minimum alignment of each @var{variable} to @var{alignment}.
10480 This is the same as GCC's @code{aligned} attribute @pxref{Variable
10481 Attributes}). Macro expansion occurs on the arguments to this pragma
10482 when compiling C and Objective-C. It does not currently occur when
10483 compiling C++, but this is a bug which may be fixed in a future
10486 @item fini (@var{function} [, @var{function}]...)
10487 @cindex pragma, fini
10489 This pragma causes each listed @var{function} to be called after
10490 main, or during shared module unloading, by adding a call to the
10491 @code{.fini} section.
10493 @item init (@var{function} [, @var{function}]...)
10494 @cindex pragma, init
10496 This pragma causes each listed @var{function} to be called during
10497 initialization (before @code{main}) or during shared module loading, by
10498 adding a call to the @code{.init} section.
10502 @node Symbol-Renaming Pragmas
10503 @subsection Symbol-Renaming Pragmas
10505 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
10506 supports two @code{#pragma} directives which change the name used in
10507 assembly for a given declaration. These pragmas are only available on
10508 platforms whose system headers need them. To get this effect on all
10509 platforms supported by GCC, use the asm labels extension (@pxref{Asm
10513 @item redefine_extname @var{oldname} @var{newname}
10514 @cindex pragma, redefine_extname
10516 This pragma gives the C function @var{oldname} the assembly symbol
10517 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
10518 will be defined if this pragma is available (currently only on
10521 @item extern_prefix @var{string}
10522 @cindex pragma, extern_prefix
10524 This pragma causes all subsequent external function and variable
10525 declarations to have @var{string} prepended to their assembly symbols.
10526 This effect may be terminated with another @code{extern_prefix} pragma
10527 whose argument is an empty string. The preprocessor macro
10528 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
10529 available (currently only on Tru64 UNIX)@.
10532 These pragmas and the asm labels extension interact in a complicated
10533 manner. Here are some corner cases you may want to be aware of.
10536 @item Both pragmas silently apply only to declarations with external
10537 linkage. Asm labels do not have this restriction.
10539 @item In C++, both pragmas silently apply only to declarations with
10540 ``C'' linkage. Again, asm labels do not have this restriction.
10542 @item If any of the three ways of changing the assembly name of a
10543 declaration is applied to a declaration whose assembly name has
10544 already been determined (either by a previous use of one of these
10545 features, or because the compiler needed the assembly name in order to
10546 generate code), and the new name is different, a warning issues and
10547 the name does not change.
10549 @item The @var{oldname} used by @code{#pragma redefine_extname} is
10550 always the C-language name.
10552 @item If @code{#pragma extern_prefix} is in effect, and a declaration
10553 occurs with an asm label attached, the prefix is silently ignored for
10556 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
10557 apply to the same declaration, whichever triggered first wins, and a
10558 warning issues if they contradict each other. (We would like to have
10559 @code{#pragma redefine_extname} always win, for consistency with asm
10560 labels, but if @code{#pragma extern_prefix} triggers first we have no
10561 way of knowing that that happened.)
10564 @node Structure-Packing Pragmas
10565 @subsection Structure-Packing Pragmas
10567 For compatibility with Win32, GCC supports a set of @code{#pragma}
10568 directives which change the maximum alignment of members of structures
10569 (other than zero-width bitfields), unions, and classes subsequently
10570 defined. The @var{n} value below always is required to be a small power
10571 of two and specifies the new alignment in bytes.
10574 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
10575 @item @code{#pragma pack()} sets the alignment to the one that was in
10576 effect when compilation started (see also command line option
10577 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
10578 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
10579 setting on an internal stack and then optionally sets the new alignment.
10580 @item @code{#pragma pack(pop)} restores the alignment setting to the one
10581 saved at the top of the internal stack (and removes that stack entry).
10582 Note that @code{#pragma pack([@var{n}])} does not influence this internal
10583 stack; thus it is possible to have @code{#pragma pack(push)} followed by
10584 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
10585 @code{#pragma pack(pop)}.
10588 Some targets, e.g. i386 and powerpc, support the @code{ms_struct}
10589 @code{#pragma} which lays out a structure as the documented
10590 @code{__attribute__ ((ms_struct))}.
10592 @item @code{#pragma ms_struct on} turns on the layout for structures
10594 @item @code{#pragma ms_struct off} turns off the layout for structures
10596 @item @code{#pragma ms_struct reset} goes back to the default layout.
10600 @subsection Weak Pragmas
10602 For compatibility with SVR4, GCC supports a set of @code{#pragma}
10603 directives for declaring symbols to be weak, and defining weak
10607 @item #pragma weak @var{symbol}
10608 @cindex pragma, weak
10609 This pragma declares @var{symbol} to be weak, as if the declaration
10610 had the attribute of the same name. The pragma may appear before
10611 or after the declaration of @var{symbol}, but must appear before
10612 either its first use or its definition. It is not an error for
10613 @var{symbol} to never be defined at all.
10615 @item #pragma weak @var{symbol1} = @var{symbol2}
10616 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
10617 It is an error if @var{symbol2} is not defined in the current
10621 @node Diagnostic Pragmas
10622 @subsection Diagnostic Pragmas
10624 GCC allows the user to selectively enable or disable certain types of
10625 diagnostics, and change the kind of the diagnostic. For example, a
10626 project's policy might require that all sources compile with
10627 @option{-Werror} but certain files might have exceptions allowing
10628 specific types of warnings. Or, a project might selectively enable
10629 diagnostics and treat them as errors depending on which preprocessor
10630 macros are defined.
10633 @item #pragma GCC diagnostic @var{kind} @var{option}
10634 @cindex pragma, diagnostic
10636 Modifies the disposition of a diagnostic. Note that not all
10637 diagnostics are modifiable; at the moment only warnings (normally
10638 controlled by @samp{-W...}) can be controlled, and not all of them.
10639 Use @option{-fdiagnostics-show-option} to determine which diagnostics
10640 are controllable and which option controls them.
10642 @var{kind} is @samp{error} to treat this diagnostic as an error,
10643 @samp{warning} to treat it like a warning (even if @option{-Werror} is
10644 in effect), or @samp{ignored} if the diagnostic is to be ignored.
10645 @var{option} is a double quoted string which matches the command line
10649 #pragma GCC diagnostic warning "-Wformat"
10650 #pragma GCC diagnostic error "-Wformat"
10651 #pragma GCC diagnostic ignored "-Wformat"
10654 Note that these pragmas override any command line options. Also,
10655 while it is syntactically valid to put these pragmas anywhere in your
10656 sources, the only supported location for them is before any data or
10657 functions are defined. Doing otherwise may result in unpredictable
10658 results depending on how the optimizer manages your sources. If the
10659 same option is listed multiple times, the last one specified is the
10660 one that is in effect. This pragma is not intended to be a general
10661 purpose replacement for command line options, but for implementing
10662 strict control over project policies.
10666 @node Visibility Pragmas
10667 @subsection Visibility Pragmas
10670 @item #pragma GCC visibility push(@var{visibility})
10671 @itemx #pragma GCC visibility pop
10672 @cindex pragma, visibility
10674 This pragma allows the user to set the visibility for multiple
10675 declarations without having to give each a visibility attribute
10676 @xref{Function Attributes}, for more information about visibility and
10677 the attribute syntax.
10679 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
10680 declarations. Class members and template specializations are not
10681 affected; if you want to override the visibility for a particular
10682 member or instantiation, you must use an attribute.
10686 @node Unnamed Fields
10687 @section Unnamed struct/union fields within structs/unions
10691 For compatibility with other compilers, GCC allows you to define
10692 a structure or union that contains, as fields, structures and unions
10693 without names. For example:
10706 In this example, the user would be able to access members of the unnamed
10707 union with code like @samp{foo.b}. Note that only unnamed structs and
10708 unions are allowed, you may not have, for example, an unnamed
10711 You must never create such structures that cause ambiguous field definitions.
10712 For example, this structure:
10723 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
10724 Such constructs are not supported and must be avoided. In the future,
10725 such constructs may be detected and treated as compilation errors.
10727 @opindex fms-extensions
10728 Unless @option{-fms-extensions} is used, the unnamed field must be a
10729 structure or union definition without a tag (for example, @samp{struct
10730 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
10731 also be a definition with a tag such as @samp{struct foo @{ int a;
10732 @};}, a reference to a previously defined structure or union such as
10733 @samp{struct foo;}, or a reference to a @code{typedef} name for a
10734 previously defined structure or union type.
10737 @section Thread-Local Storage
10738 @cindex Thread-Local Storage
10739 @cindex @acronym{TLS}
10742 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
10743 are allocated such that there is one instance of the variable per extant
10744 thread. The run-time model GCC uses to implement this originates
10745 in the IA-64 processor-specific ABI, but has since been migrated
10746 to other processors as well. It requires significant support from
10747 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
10748 system libraries (@file{libc.so} and @file{libpthread.so}), so it
10749 is not available everywhere.
10751 At the user level, the extension is visible with a new storage
10752 class keyword: @code{__thread}. For example:
10756 extern __thread struct state s;
10757 static __thread char *p;
10760 The @code{__thread} specifier may be used alone, with the @code{extern}
10761 or @code{static} specifiers, but with no other storage class specifier.
10762 When used with @code{extern} or @code{static}, @code{__thread} must appear
10763 immediately after the other storage class specifier.
10765 The @code{__thread} specifier may be applied to any global, file-scoped
10766 static, function-scoped static, or static data member of a class. It may
10767 not be applied to block-scoped automatic or non-static data member.
10769 When the address-of operator is applied to a thread-local variable, it is
10770 evaluated at run-time and returns the address of the current thread's
10771 instance of that variable. An address so obtained may be used by any
10772 thread. When a thread terminates, any pointers to thread-local variables
10773 in that thread become invalid.
10775 No static initialization may refer to the address of a thread-local variable.
10777 In C++, if an initializer is present for a thread-local variable, it must
10778 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
10781 See @uref{http://people.redhat.com/drepper/tls.pdf,
10782 ELF Handling For Thread-Local Storage} for a detailed explanation of
10783 the four thread-local storage addressing models, and how the run-time
10784 is expected to function.
10787 * C99 Thread-Local Edits::
10788 * C++98 Thread-Local Edits::
10791 @node C99 Thread-Local Edits
10792 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
10794 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
10795 that document the exact semantics of the language extension.
10799 @cite{5.1.2 Execution environments}
10801 Add new text after paragraph 1
10804 Within either execution environment, a @dfn{thread} is a flow of
10805 control within a program. It is implementation defined whether
10806 or not there may be more than one thread associated with a program.
10807 It is implementation defined how threads beyond the first are
10808 created, the name and type of the function called at thread
10809 startup, and how threads may be terminated. However, objects
10810 with thread storage duration shall be initialized before thread
10815 @cite{6.2.4 Storage durations of objects}
10817 Add new text before paragraph 3
10820 An object whose identifier is declared with the storage-class
10821 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
10822 Its lifetime is the entire execution of the thread, and its
10823 stored value is initialized only once, prior to thread startup.
10827 @cite{6.4.1 Keywords}
10829 Add @code{__thread}.
10832 @cite{6.7.1 Storage-class specifiers}
10834 Add @code{__thread} to the list of storage class specifiers in
10837 Change paragraph 2 to
10840 With the exception of @code{__thread}, at most one storage-class
10841 specifier may be given [@dots{}]. The @code{__thread} specifier may
10842 be used alone, or immediately following @code{extern} or
10846 Add new text after paragraph 6
10849 The declaration of an identifier for a variable that has
10850 block scope that specifies @code{__thread} shall also
10851 specify either @code{extern} or @code{static}.
10853 The @code{__thread} specifier shall be used only with
10858 @node C++98 Thread-Local Edits
10859 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
10861 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
10862 that document the exact semantics of the language extension.
10866 @b{[intro.execution]}
10868 New text after paragraph 4
10871 A @dfn{thread} is a flow of control within the abstract machine.
10872 It is implementation defined whether or not there may be more than
10876 New text after paragraph 7
10879 It is unspecified whether additional action must be taken to
10880 ensure when and whether side effects are visible to other threads.
10886 Add @code{__thread}.
10889 @b{[basic.start.main]}
10891 Add after paragraph 5
10894 The thread that begins execution at the @code{main} function is called
10895 the @dfn{main thread}. It is implementation defined how functions
10896 beginning threads other than the main thread are designated or typed.
10897 A function so designated, as well as the @code{main} function, is called
10898 a @dfn{thread startup function}. It is implementation defined what
10899 happens if a thread startup function returns. It is implementation
10900 defined what happens to other threads when any thread calls @code{exit}.
10904 @b{[basic.start.init]}
10906 Add after paragraph 4
10909 The storage for an object of thread storage duration shall be
10910 statically initialized before the first statement of the thread startup
10911 function. An object of thread storage duration shall not require
10912 dynamic initialization.
10916 @b{[basic.start.term]}
10918 Add after paragraph 3
10921 The type of an object with thread storage duration shall not have a
10922 non-trivial destructor, nor shall it be an array type whose elements
10923 (directly or indirectly) have non-trivial destructors.
10929 Add ``thread storage duration'' to the list in paragraph 1.
10934 Thread, static, and automatic storage durations are associated with
10935 objects introduced by declarations [@dots{}].
10938 Add @code{__thread} to the list of specifiers in paragraph 3.
10941 @b{[basic.stc.thread]}
10943 New section before @b{[basic.stc.static]}
10946 The keyword @code{__thread} applied to a non-local object gives the
10947 object thread storage duration.
10949 A local variable or class data member declared both @code{static}
10950 and @code{__thread} gives the variable or member thread storage
10955 @b{[basic.stc.static]}
10960 All objects which have neither thread storage duration, dynamic
10961 storage duration nor are local [@dots{}].
10967 Add @code{__thread} to the list in paragraph 1.
10972 With the exception of @code{__thread}, at most one
10973 @var{storage-class-specifier} shall appear in a given
10974 @var{decl-specifier-seq}. The @code{__thread} specifier may
10975 be used alone, or immediately following the @code{extern} or
10976 @code{static} specifiers. [@dots{}]
10979 Add after paragraph 5
10982 The @code{__thread} specifier can be applied only to the names of objects
10983 and to anonymous unions.
10989 Add after paragraph 6
10992 Non-@code{static} members shall not be @code{__thread}.
10996 @node Binary constants
10997 @section Binary constants using the @samp{0b} prefix
10998 @cindex Binary constants using the @samp{0b} prefix
11000 Integer constants can be written as binary constants, consisting of a
11001 sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or
11002 @samp{0B}. This is particularly useful in environments that operate a
11003 lot on the bit-level (like microcontrollers).
11005 The following statements are identical:
11014 The type of these constants follows the same rules as for octal or
11015 hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL}
11018 @node C++ Extensions
11019 @chapter Extensions to the C++ Language
11020 @cindex extensions, C++ language
11021 @cindex C++ language extensions
11023 The GNU compiler provides these extensions to the C++ language (and you
11024 can also use most of the C language extensions in your C++ programs). If you
11025 want to write code that checks whether these features are available, you can
11026 test for the GNU compiler the same way as for C programs: check for a
11027 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
11028 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
11029 Predefined Macros,cpp,The GNU C Preprocessor}).
11032 * Volatiles:: What constitutes an access to a volatile object.
11033 * Restricted Pointers:: C99 restricted pointers and references.
11034 * Vague Linkage:: Where G++ puts inlines, vtables and such.
11035 * C++ Interface:: You can use a single C++ header file for both
11036 declarations and definitions.
11037 * Template Instantiation:: Methods for ensuring that exactly one copy of
11038 each needed template instantiation is emitted.
11039 * Bound member functions:: You can extract a function pointer to the
11040 method denoted by a @samp{->*} or @samp{.*} expression.
11041 * C++ Attributes:: Variable, function, and type attributes for C++ only.
11042 * Namespace Association:: Strong using-directives for namespace association.
11043 * Type Traits:: Compiler support for type traits
11044 * Java Exceptions:: Tweaking exception handling to work with Java.
11045 * Deprecated Features:: Things will disappear from g++.
11046 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
11050 @section When is a Volatile Object Accessed?
11051 @cindex accessing volatiles
11052 @cindex volatile read
11053 @cindex volatile write
11054 @cindex volatile access
11056 Both the C and C++ standard have the concept of volatile objects. These
11057 are normally accessed by pointers and used for accessing hardware. The
11058 standards encourage compilers to refrain from optimizations concerning
11059 accesses to volatile objects. The C standard leaves it implementation
11060 defined as to what constitutes a volatile access. The C++ standard omits
11061 to specify this, except to say that C++ should behave in a similar manner
11062 to C with respect to volatiles, where possible. The minimum either
11063 standard specifies is that at a sequence point all previous accesses to
11064 volatile objects have stabilized and no subsequent accesses have
11065 occurred. Thus an implementation is free to reorder and combine
11066 volatile accesses which occur between sequence points, but cannot do so
11067 for accesses across a sequence point. The use of volatiles does not
11068 allow you to violate the restriction on updating objects multiple times
11069 within a sequence point.
11071 @xref{Qualifiers implementation, , Volatile qualifier and the C compiler}.
11073 The behavior differs slightly between C and C++ in the non-obvious cases:
11076 volatile int *src = @var{somevalue};
11080 With C, such expressions are rvalues, and GCC interprets this either as a
11081 read of the volatile object being pointed to or only as request to evaluate
11082 the side-effects. The C++ standard specifies that such expressions do not
11083 undergo lvalue to rvalue conversion, and that the type of the dereferenced
11084 object may be incomplete. The C++ standard does not specify explicitly
11085 that it is this lvalue to rvalue conversion which may be responsible for
11086 causing an access. However, there is reason to believe that it is,
11087 because otherwise certain simple expressions become undefined. However,
11088 because it would surprise most programmers, G++ treats dereferencing a
11089 pointer to volatile object of complete type when the value is unused as
11090 GCC would do for an equivalent type in C. When the object has incomplete
11091 type, G++ issues a warning; if you wish to force an error, you must
11092 force a conversion to rvalue with, for instance, a static cast.
11094 When using a reference to volatile, G++ does not treat equivalent
11095 expressions as accesses to volatiles, but instead issues a warning that
11096 no volatile is accessed. The rationale for this is that otherwise it
11097 becomes difficult to determine where volatile access occur, and not
11098 possible to ignore the return value from functions returning volatile
11099 references. Again, if you wish to force a read, cast the reference to
11102 @node Restricted Pointers
11103 @section Restricting Pointer Aliasing
11104 @cindex restricted pointers
11105 @cindex restricted references
11106 @cindex restricted this pointer
11108 As with the C front end, G++ understands the C99 feature of restricted pointers,
11109 specified with the @code{__restrict__}, or @code{__restrict} type
11110 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
11111 language flag, @code{restrict} is not a keyword in C++.
11113 In addition to allowing restricted pointers, you can specify restricted
11114 references, which indicate that the reference is not aliased in the local
11118 void fn (int *__restrict__ rptr, int &__restrict__ rref)
11125 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
11126 @var{rref} refers to a (different) unaliased integer.
11128 You may also specify whether a member function's @var{this} pointer is
11129 unaliased by using @code{__restrict__} as a member function qualifier.
11132 void T::fn () __restrict__
11139 Within the body of @code{T::fn}, @var{this} will have the effective
11140 definition @code{T *__restrict__ const this}. Notice that the
11141 interpretation of a @code{__restrict__} member function qualifier is
11142 different to that of @code{const} or @code{volatile} qualifier, in that it
11143 is applied to the pointer rather than the object. This is consistent with
11144 other compilers which implement restricted pointers.
11146 As with all outermost parameter qualifiers, @code{__restrict__} is
11147 ignored in function definition matching. This means you only need to
11148 specify @code{__restrict__} in a function definition, rather than
11149 in a function prototype as well.
11151 @node Vague Linkage
11152 @section Vague Linkage
11153 @cindex vague linkage
11155 There are several constructs in C++ which require space in the object
11156 file but are not clearly tied to a single translation unit. We say that
11157 these constructs have ``vague linkage''. Typically such constructs are
11158 emitted wherever they are needed, though sometimes we can be more
11162 @item Inline Functions
11163 Inline functions are typically defined in a header file which can be
11164 included in many different compilations. Hopefully they can usually be
11165 inlined, but sometimes an out-of-line copy is necessary, if the address
11166 of the function is taken or if inlining fails. In general, we emit an
11167 out-of-line copy in all translation units where one is needed. As an
11168 exception, we only emit inline virtual functions with the vtable, since
11169 it will always require a copy.
11171 Local static variables and string constants used in an inline function
11172 are also considered to have vague linkage, since they must be shared
11173 between all inlined and out-of-line instances of the function.
11177 C++ virtual functions are implemented in most compilers using a lookup
11178 table, known as a vtable. The vtable contains pointers to the virtual
11179 functions provided by a class, and each object of the class contains a
11180 pointer to its vtable (or vtables, in some multiple-inheritance
11181 situations). If the class declares any non-inline, non-pure virtual
11182 functions, the first one is chosen as the ``key method'' for the class,
11183 and the vtable is only emitted in the translation unit where the key
11186 @emph{Note:} If the chosen key method is later defined as inline, the
11187 vtable will still be emitted in every translation unit which defines it.
11188 Make sure that any inline virtuals are declared inline in the class
11189 body, even if they are not defined there.
11191 @item type_info objects
11194 C++ requires information about types to be written out in order to
11195 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
11196 For polymorphic classes (classes with virtual functions), the type_info
11197 object is written out along with the vtable so that @samp{dynamic_cast}
11198 can determine the dynamic type of a class object at runtime. For all
11199 other types, we write out the type_info object when it is used: when
11200 applying @samp{typeid} to an expression, throwing an object, or
11201 referring to a type in a catch clause or exception specification.
11203 @item Template Instantiations
11204 Most everything in this section also applies to template instantiations,
11205 but there are other options as well.
11206 @xref{Template Instantiation,,Where's the Template?}.
11210 When used with GNU ld version 2.8 or later on an ELF system such as
11211 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
11212 these constructs will be discarded at link time. This is known as
11215 On targets that don't support COMDAT, but do support weak symbols, GCC
11216 will use them. This way one copy will override all the others, but
11217 the unused copies will still take up space in the executable.
11219 For targets which do not support either COMDAT or weak symbols,
11220 most entities with vague linkage will be emitted as local symbols to
11221 avoid duplicate definition errors from the linker. This will not happen
11222 for local statics in inlines, however, as having multiple copies will
11223 almost certainly break things.
11225 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
11226 another way to control placement of these constructs.
11228 @node C++ Interface
11229 @section #pragma interface and implementation
11231 @cindex interface and implementation headers, C++
11232 @cindex C++ interface and implementation headers
11233 @cindex pragmas, interface and implementation
11235 @code{#pragma interface} and @code{#pragma implementation} provide the
11236 user with a way of explicitly directing the compiler to emit entities
11237 with vague linkage (and debugging information) in a particular
11240 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
11241 most cases, because of COMDAT support and the ``key method'' heuristic
11242 mentioned in @ref{Vague Linkage}. Using them can actually cause your
11243 program to grow due to unnecessary out-of-line copies of inline
11244 functions. Currently (3.4) the only benefit of these
11245 @code{#pragma}s is reduced duplication of debugging information, and
11246 that should be addressed soon on DWARF 2 targets with the use of
11250 @item #pragma interface
11251 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
11252 @kindex #pragma interface
11253 Use this directive in @emph{header files} that define object classes, to save
11254 space in most of the object files that use those classes. Normally,
11255 local copies of certain information (backup copies of inline member
11256 functions, debugging information, and the internal tables that implement
11257 virtual functions) must be kept in each object file that includes class
11258 definitions. You can use this pragma to avoid such duplication. When a
11259 header file containing @samp{#pragma interface} is included in a
11260 compilation, this auxiliary information will not be generated (unless
11261 the main input source file itself uses @samp{#pragma implementation}).
11262 Instead, the object files will contain references to be resolved at link
11265 The second form of this directive is useful for the case where you have
11266 multiple headers with the same name in different directories. If you
11267 use this form, you must specify the same string to @samp{#pragma
11270 @item #pragma implementation
11271 @itemx #pragma implementation "@var{objects}.h"
11272 @kindex #pragma implementation
11273 Use this pragma in a @emph{main input file}, when you want full output from
11274 included header files to be generated (and made globally visible). The
11275 included header file, in turn, should use @samp{#pragma interface}.
11276 Backup copies of inline member functions, debugging information, and the
11277 internal tables used to implement virtual functions are all generated in
11278 implementation files.
11280 @cindex implied @code{#pragma implementation}
11281 @cindex @code{#pragma implementation}, implied
11282 @cindex naming convention, implementation headers
11283 If you use @samp{#pragma implementation} with no argument, it applies to
11284 an include file with the same basename@footnote{A file's @dfn{basename}
11285 was the name stripped of all leading path information and of trailing
11286 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
11287 file. For example, in @file{allclass.cc}, giving just
11288 @samp{#pragma implementation}
11289 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
11291 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
11292 an implementation file whenever you would include it from
11293 @file{allclass.cc} even if you never specified @samp{#pragma
11294 implementation}. This was deemed to be more trouble than it was worth,
11295 however, and disabled.
11297 Use the string argument if you want a single implementation file to
11298 include code from multiple header files. (You must also use
11299 @samp{#include} to include the header file; @samp{#pragma
11300 implementation} only specifies how to use the file---it doesn't actually
11303 There is no way to split up the contents of a single header file into
11304 multiple implementation files.
11307 @cindex inlining and C++ pragmas
11308 @cindex C++ pragmas, effect on inlining
11309 @cindex pragmas in C++, effect on inlining
11310 @samp{#pragma implementation} and @samp{#pragma interface} also have an
11311 effect on function inlining.
11313 If you define a class in a header file marked with @samp{#pragma
11314 interface}, the effect on an inline function defined in that class is
11315 similar to an explicit @code{extern} declaration---the compiler emits
11316 no code at all to define an independent version of the function. Its
11317 definition is used only for inlining with its callers.
11319 @opindex fno-implement-inlines
11320 Conversely, when you include the same header file in a main source file
11321 that declares it as @samp{#pragma implementation}, the compiler emits
11322 code for the function itself; this defines a version of the function
11323 that can be found via pointers (or by callers compiled without
11324 inlining). If all calls to the function can be inlined, you can avoid
11325 emitting the function by compiling with @option{-fno-implement-inlines}.
11326 If any calls were not inlined, you will get linker errors.
11328 @node Template Instantiation
11329 @section Where's the Template?
11330 @cindex template instantiation
11332 C++ templates are the first language feature to require more
11333 intelligence from the environment than one usually finds on a UNIX
11334 system. Somehow the compiler and linker have to make sure that each
11335 template instance occurs exactly once in the executable if it is needed,
11336 and not at all otherwise. There are two basic approaches to this
11337 problem, which are referred to as the Borland model and the Cfront model.
11340 @item Borland model
11341 Borland C++ solved the template instantiation problem by adding the code
11342 equivalent of common blocks to their linker; the compiler emits template
11343 instances in each translation unit that uses them, and the linker
11344 collapses them together. The advantage of this model is that the linker
11345 only has to consider the object files themselves; there is no external
11346 complexity to worry about. This disadvantage is that compilation time
11347 is increased because the template code is being compiled repeatedly.
11348 Code written for this model tends to include definitions of all
11349 templates in the header file, since they must be seen to be
11353 The AT&T C++ translator, Cfront, solved the template instantiation
11354 problem by creating the notion of a template repository, an
11355 automatically maintained place where template instances are stored. A
11356 more modern version of the repository works as follows: As individual
11357 object files are built, the compiler places any template definitions and
11358 instantiations encountered in the repository. At link time, the link
11359 wrapper adds in the objects in the repository and compiles any needed
11360 instances that were not previously emitted. The advantages of this
11361 model are more optimal compilation speed and the ability to use the
11362 system linker; to implement the Borland model a compiler vendor also
11363 needs to replace the linker. The disadvantages are vastly increased
11364 complexity, and thus potential for error; for some code this can be
11365 just as transparent, but in practice it can been very difficult to build
11366 multiple programs in one directory and one program in multiple
11367 directories. Code written for this model tends to separate definitions
11368 of non-inline member templates into a separate file, which should be
11369 compiled separately.
11372 When used with GNU ld version 2.8 or later on an ELF system such as
11373 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
11374 Borland model. On other systems, G++ implements neither automatic
11377 A future version of G++ will support a hybrid model whereby the compiler
11378 will emit any instantiations for which the template definition is
11379 included in the compile, and store template definitions and
11380 instantiation context information into the object file for the rest.
11381 The link wrapper will extract that information as necessary and invoke
11382 the compiler to produce the remaining instantiations. The linker will
11383 then combine duplicate instantiations.
11385 In the mean time, you have the following options for dealing with
11386 template instantiations:
11391 Compile your template-using code with @option{-frepo}. The compiler will
11392 generate files with the extension @samp{.rpo} listing all of the
11393 template instantiations used in the corresponding object files which
11394 could be instantiated there; the link wrapper, @samp{collect2}, will
11395 then update the @samp{.rpo} files to tell the compiler where to place
11396 those instantiations and rebuild any affected object files. The
11397 link-time overhead is negligible after the first pass, as the compiler
11398 will continue to place the instantiations in the same files.
11400 This is your best option for application code written for the Borland
11401 model, as it will just work. Code written for the Cfront model will
11402 need to be modified so that the template definitions are available at
11403 one or more points of instantiation; usually this is as simple as adding
11404 @code{#include <tmethods.cc>} to the end of each template header.
11406 For library code, if you want the library to provide all of the template
11407 instantiations it needs, just try to link all of its object files
11408 together; the link will fail, but cause the instantiations to be
11409 generated as a side effect. Be warned, however, that this may cause
11410 conflicts if multiple libraries try to provide the same instantiations.
11411 For greater control, use explicit instantiation as described in the next
11415 @opindex fno-implicit-templates
11416 Compile your code with @option{-fno-implicit-templates} to disable the
11417 implicit generation of template instances, and explicitly instantiate
11418 all the ones you use. This approach requires more knowledge of exactly
11419 which instances you need than do the others, but it's less
11420 mysterious and allows greater control. You can scatter the explicit
11421 instantiations throughout your program, perhaps putting them in the
11422 translation units where the instances are used or the translation units
11423 that define the templates themselves; you can put all of the explicit
11424 instantiations you need into one big file; or you can create small files
11431 template class Foo<int>;
11432 template ostream& operator <<
11433 (ostream&, const Foo<int>&);
11436 for each of the instances you need, and create a template instantiation
11437 library from those.
11439 If you are using Cfront-model code, you can probably get away with not
11440 using @option{-fno-implicit-templates} when compiling files that don't
11441 @samp{#include} the member template definitions.
11443 If you use one big file to do the instantiations, you may want to
11444 compile it without @option{-fno-implicit-templates} so you get all of the
11445 instances required by your explicit instantiations (but not by any
11446 other files) without having to specify them as well.
11448 G++ has extended the template instantiation syntax given in the ISO
11449 standard to allow forward declaration of explicit instantiations
11450 (with @code{extern}), instantiation of the compiler support data for a
11451 template class (i.e.@: the vtable) without instantiating any of its
11452 members (with @code{inline}), and instantiation of only the static data
11453 members of a template class, without the support data or member
11454 functions (with (@code{static}):
11457 extern template int max (int, int);
11458 inline template class Foo<int>;
11459 static template class Foo<int>;
11463 Do nothing. Pretend G++ does implement automatic instantiation
11464 management. Code written for the Borland model will work fine, but
11465 each translation unit will contain instances of each of the templates it
11466 uses. In a large program, this can lead to an unacceptable amount of code
11470 @node Bound member functions
11471 @section Extracting the function pointer from a bound pointer to member function
11473 @cindex pointer to member function
11474 @cindex bound pointer to member function
11476 In C++, pointer to member functions (PMFs) are implemented using a wide
11477 pointer of sorts to handle all the possible call mechanisms; the PMF
11478 needs to store information about how to adjust the @samp{this} pointer,
11479 and if the function pointed to is virtual, where to find the vtable, and
11480 where in the vtable to look for the member function. If you are using
11481 PMFs in an inner loop, you should really reconsider that decision. If
11482 that is not an option, you can extract the pointer to the function that
11483 would be called for a given object/PMF pair and call it directly inside
11484 the inner loop, to save a bit of time.
11486 Note that you will still be paying the penalty for the call through a
11487 function pointer; on most modern architectures, such a call defeats the
11488 branch prediction features of the CPU@. This is also true of normal
11489 virtual function calls.
11491 The syntax for this extension is
11495 extern int (A::*fp)();
11496 typedef int (*fptr)(A *);
11498 fptr p = (fptr)(a.*fp);
11501 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
11502 no object is needed to obtain the address of the function. They can be
11503 converted to function pointers directly:
11506 fptr p1 = (fptr)(&A::foo);
11509 @opindex Wno-pmf-conversions
11510 You must specify @option{-Wno-pmf-conversions} to use this extension.
11512 @node C++ Attributes
11513 @section C++-Specific Variable, Function, and Type Attributes
11515 Some attributes only make sense for C++ programs.
11518 @item init_priority (@var{priority})
11519 @cindex init_priority attribute
11522 In Standard C++, objects defined at namespace scope are guaranteed to be
11523 initialized in an order in strict accordance with that of their definitions
11524 @emph{in a given translation unit}. No guarantee is made for initializations
11525 across translation units. However, GNU C++ allows users to control the
11526 order of initialization of objects defined at namespace scope with the
11527 @code{init_priority} attribute by specifying a relative @var{priority},
11528 a constant integral expression currently bounded between 101 and 65535
11529 inclusive. Lower numbers indicate a higher priority.
11531 In the following example, @code{A} would normally be created before
11532 @code{B}, but the @code{init_priority} attribute has reversed that order:
11535 Some_Class A __attribute__ ((init_priority (2000)));
11536 Some_Class B __attribute__ ((init_priority (543)));
11540 Note that the particular values of @var{priority} do not matter; only their
11543 @item java_interface
11544 @cindex java_interface attribute
11546 This type attribute informs C++ that the class is a Java interface. It may
11547 only be applied to classes declared within an @code{extern "Java"} block.
11548 Calls to methods declared in this interface will be dispatched using GCJ's
11549 interface table mechanism, instead of regular virtual table dispatch.
11553 See also @xref{Namespace Association}.
11555 @node Namespace Association
11556 @section Namespace Association
11558 @strong{Caution:} The semantics of this extension are not fully
11559 defined. Users should refrain from using this extension as its
11560 semantics may change subtly over time. It is possible that this
11561 extension will be removed in future versions of G++.
11563 A using-directive with @code{__attribute ((strong))} is stronger
11564 than a normal using-directive in two ways:
11568 Templates from the used namespace can be specialized and explicitly
11569 instantiated as though they were members of the using namespace.
11572 The using namespace is considered an associated namespace of all
11573 templates in the used namespace for purposes of argument-dependent
11577 The used namespace must be nested within the using namespace so that
11578 normal unqualified lookup works properly.
11580 This is useful for composing a namespace transparently from
11581 implementation namespaces. For example:
11586 template <class T> struct A @{ @};
11588 using namespace debug __attribute ((__strong__));
11589 template <> struct A<int> @{ @}; // @r{ok to specialize}
11591 template <class T> void f (A<T>);
11596 f (std::A<float>()); // @r{lookup finds} std::f
11602 @section Type Traits
11604 The C++ front-end implements syntactic extensions that allow to
11605 determine at compile time various characteristics of a type (or of a
11609 @item __has_nothrow_assign (type)
11610 If @code{type} is const qualified or is a reference type then the trait is
11611 false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait
11612 is true, else if @code{type} is a cv class or union type with copy assignment
11613 operators that are known not to throw an exception then the trait is true,
11614 else it is false. Requires: @code{type} shall be a complete type, an array
11615 type of unknown bound, or is a @code{void} type.
11617 @item __has_nothrow_copy (type)
11618 If @code{__has_trivial_copy (type)} is true then the trait is true, else if
11619 @code{type} is a cv class or union type with copy constructors that
11620 are known not to throw an exception then the trait is true, else it is false.
11621 Requires: @code{type} shall be a complete type, an array type of
11622 unknown bound, or is a @code{void} type.
11624 @item __has_nothrow_constructor (type)
11625 If @code{__has_trivial_constructor (type)} is true then the trait is
11626 true, else if @code{type} is a cv class or union type (or array
11627 thereof) with a default constructor that is known not to throw an
11628 exception then the trait is true, else it is false. Requires:
11629 @code{type} shall be a complete type, an array type of unknown bound,
11630 or is a @code{void} type.
11632 @item __has_trivial_assign (type)
11633 If @code{type} is const qualified or is a reference type then the trait is
11634 false. Otherwise if @code{__is_pod (type)} is true then the trait is
11635 true, else if @code{type} is a cv class or union type with a trivial
11636 copy assignment ([class.copy]) then the trait is true, else it is
11637 false. Requires: @code{type} shall be a complete type, an array type
11638 of unknown bound, or is a @code{void} type.
11640 @item __has_trivial_copy (type)
11641 If @code{__is_pod (type)} is true or @code{type} is a reference type
11642 then the trait is true, else if @code{type} is a cv class or union type
11643 with a trivial copy constructor ([class.copy]) then the trait
11644 is true, else it is false. Requires: @code{type} shall be a complete
11645 type, an array type of unknown bound, or is a @code{void} type.
11647 @item __has_trivial_constructor (type)
11648 If @code{__is_pod (type)} is true then the trait is true, else if
11649 @code{type} is a cv class or union type (or array thereof) with a
11650 trivial default constructor ([class.ctor]) then the trait is true,
11651 else it is false. Requires: @code{type} shall be a complete type, an
11652 array type of unknown bound, or is a @code{void} type.
11654 @item __has_trivial_destructor (type)
11655 If @code{__is_pod (type)} is true or @code{type} is a reference type then
11656 the trait is true, else if @code{type} is a cv class or union type (or
11657 array thereof) with a trivial destructor ([class.dtor]) then the trait
11658 is true, else it is false. Requires: @code{type} shall be a complete
11659 type, an array type of unknown bound, or is a @code{void} type.
11661 @item __has_virtual_destructor (type)
11662 If @code{type} is a class type with a virtual destructor
11663 ([class.dtor]) then the trait is true, else it is false. Requires:
11664 @code{type} shall be a complete type, an array type of unknown bound,
11665 or is a @code{void} type.
11667 @item __is_abstract (type)
11668 If @code{type} is an abstract class ([class.abstract]) then the trait
11669 is true, else it is false. Requires: @code{type} shall be a complete
11670 type, an array type of unknown bound, or is a @code{void} type.
11672 @item __is_base_of (base_type, derived_type)
11673 If @code{base_type} is a base class of @code{derived_type}
11674 ([class.derived]) then the trait is true, otherwise it is false.
11675 Top-level cv qualifications of @code{base_type} and
11676 @code{derived_type} are ignored. For the purposes of this trait, a
11677 class type is considered is own base. Requires: if @code{__is_class
11678 (base_type)} and @code{__is_class (derived_type)} are true and
11679 @code{base_type} and @code{derived_type} are not the same type
11680 (disregarding cv-qualifiers), @code{derived_type} shall be a complete
11681 type. Diagnostic is produced if this requirement is not met.
11683 @item __is_class (type)
11684 If @code{type} is a cv class type, and not a union type
11685 ([basic.compound]) the the trait is true, else it is false.
11687 @item __is_empty (type)
11688 If @code{__is_class (type)} is false then the trait is false.
11689 Otherwise @code{type} is considered empty if and only if: @code{type}
11690 has no non-static data members, or all non-static data members, if
11691 any, are bit-fields of lenght 0, and @code{type} has no virtual
11692 members, and @code{type} has no virtual base classes, and @code{type}
11693 has no base classes @code{base_type} for which
11694 @code{__is_empty (base_type)} is false. Requires: @code{type} shall
11695 be a complete type, an array type of unknown bound, or is a
11698 @item __is_enum (type)
11699 If @code{type} is a cv enumeration type ([basic.compound]) the the trait is
11700 true, else it is false.
11702 @item __is_pod (type)
11703 If @code{type} is a cv POD type ([basic.types]) then the trait is true,
11704 else it is false. Requires: @code{type} shall be a complete type,
11705 an array type of unknown bound, or is a @code{void} type.
11707 @item __is_polymorphic (type)
11708 If @code{type} is a polymorphic class ([class.virtual]) then the trait
11709 is true, else it is false. Requires: @code{type} shall be a complete
11710 type, an array type of unknown bound, or is a @code{void} type.
11712 @item __is_union (type)
11713 If @code{type} is a cv union type ([basic.compound]) the the trait is
11714 true, else it is false.
11718 @node Java Exceptions
11719 @section Java Exceptions
11721 The Java language uses a slightly different exception handling model
11722 from C++. Normally, GNU C++ will automatically detect when you are
11723 writing C++ code that uses Java exceptions, and handle them
11724 appropriately. However, if C++ code only needs to execute destructors
11725 when Java exceptions are thrown through it, GCC will guess incorrectly.
11726 Sample problematic code is:
11729 struct S @{ ~S(); @};
11730 extern void bar(); // @r{is written in Java, and may throw exceptions}
11739 The usual effect of an incorrect guess is a link failure, complaining of
11740 a missing routine called @samp{__gxx_personality_v0}.
11742 You can inform the compiler that Java exceptions are to be used in a
11743 translation unit, irrespective of what it might think, by writing
11744 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
11745 @samp{#pragma} must appear before any functions that throw or catch
11746 exceptions, or run destructors when exceptions are thrown through them.
11748 You cannot mix Java and C++ exceptions in the same translation unit. It
11749 is believed to be safe to throw a C++ exception from one file through
11750 another file compiled for the Java exception model, or vice versa, but
11751 there may be bugs in this area.
11753 @node Deprecated Features
11754 @section Deprecated Features
11756 In the past, the GNU C++ compiler was extended to experiment with new
11757 features, at a time when the C++ language was still evolving. Now that
11758 the C++ standard is complete, some of those features are superseded by
11759 superior alternatives. Using the old features might cause a warning in
11760 some cases that the feature will be dropped in the future. In other
11761 cases, the feature might be gone already.
11763 While the list below is not exhaustive, it documents some of the options
11764 that are now deprecated:
11767 @item -fexternal-templates
11768 @itemx -falt-external-templates
11769 These are two of the many ways for G++ to implement template
11770 instantiation. @xref{Template Instantiation}. The C++ standard clearly
11771 defines how template definitions have to be organized across
11772 implementation units. G++ has an implicit instantiation mechanism that
11773 should work just fine for standard-conforming code.
11775 @item -fstrict-prototype
11776 @itemx -fno-strict-prototype
11777 Previously it was possible to use an empty prototype parameter list to
11778 indicate an unspecified number of parameters (like C), rather than no
11779 parameters, as C++ demands. This feature has been removed, except where
11780 it is required for backwards compatibility @xref{Backwards Compatibility}.
11783 G++ allows a virtual function returning @samp{void *} to be overridden
11784 by one returning a different pointer type. This extension to the
11785 covariant return type rules is now deprecated and will be removed from a
11788 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
11789 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
11790 and will be removed in a future version. Code using these operators
11791 should be modified to use @code{std::min} and @code{std::max} instead.
11793 The named return value extension has been deprecated, and is now
11796 The use of initializer lists with new expressions has been deprecated,
11797 and is now removed from G++.
11799 Floating and complex non-type template parameters have been deprecated,
11800 and are now removed from G++.
11802 The implicit typename extension has been deprecated and is now
11805 The use of default arguments in function pointers, function typedefs
11806 and other places where they are not permitted by the standard is
11807 deprecated and will be removed from a future version of G++.
11809 G++ allows floating-point literals to appear in integral constant expressions,
11810 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
11811 This extension is deprecated and will be removed from a future version.
11813 G++ allows static data members of const floating-point type to be declared
11814 with an initializer in a class definition. The standard only allows
11815 initializers for static members of const integral types and const
11816 enumeration types so this extension has been deprecated and will be removed
11817 from a future version.
11819 @node Backwards Compatibility
11820 @section Backwards Compatibility
11821 @cindex Backwards Compatibility
11822 @cindex ARM [Annotated C++ Reference Manual]
11824 Now that there is a definitive ISO standard C++, G++ has a specification
11825 to adhere to. The C++ language evolved over time, and features that
11826 used to be acceptable in previous drafts of the standard, such as the ARM
11827 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
11828 compilation of C++ written to such drafts, G++ contains some backwards
11829 compatibilities. @emph{All such backwards compatibility features are
11830 liable to disappear in future versions of G++.} They should be considered
11831 deprecated @xref{Deprecated Features}.
11835 If a variable is declared at for scope, it used to remain in scope until
11836 the end of the scope which contained the for statement (rather than just
11837 within the for scope). G++ retains this, but issues a warning, if such a
11838 variable is accessed outside the for scope.
11840 @item Implicit C language
11841 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
11842 scope to set the language. On such systems, all header files are
11843 implicitly scoped inside a C language scope. Also, an empty prototype
11844 @code{()} will be treated as an unspecified number of arguments, rather
11845 than no arguments, as C++ demands.