1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1996, 1998, 1999, 2000,
2 @c 2001, 2002, 2003, 2004, 2005, 2006 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 * Decimal Float:: Decimal Floating Point.
37 * Hex Floats:: Hexadecimal floating-point constants.
38 * Zero Length:: Zero-length arrays.
39 * Variable Length:: Arrays whose length is computed at run time.
40 * Empty Structures:: Structures with no members.
41 * Variadic Macros:: Macros with a variable number of arguments.
42 * Escaped Newlines:: Slightly looser rules for escaped newlines.
43 * Subscripting:: Any array can be subscripted, even if not an lvalue.
44 * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers.
45 * Initializers:: Non-constant initializers.
46 * Compound Literals:: Compound literals give structures, unions
48 * Designated Inits:: Labeling elements of initializers.
49 * Cast to Union:: Casting to union type from any member of the union.
50 * Case Ranges:: `case 1 ... 9' and such.
51 * Mixed Declarations:: Mixing declarations and code.
52 * Function Attributes:: Declaring that functions have no side effects,
53 or that they can never return.
54 * Attribute Syntax:: Formal syntax for attributes.
55 * Function Prototypes:: Prototype declarations and old-style definitions.
56 * C++ Comments:: C++ comments are recognized.
57 * Dollar Signs:: Dollar sign is allowed in identifiers.
58 * Character Escapes:: @samp{\e} stands for the character @key{ESC}.
59 * Variable Attributes:: Specifying attributes of variables.
60 * Type Attributes:: Specifying attributes of types.
61 * Alignment:: Inquiring about the alignment of a type or variable.
62 * Inline:: Defining inline functions (as fast as macros).
63 * Extended Asm:: Assembler instructions with C expressions as operands.
64 (With them you can define ``built-in'' functions.)
65 * Constraints:: Constraints for asm operands
66 * Asm Labels:: Specifying the assembler name to use for a C symbol.
67 * Explicit Reg Vars:: Defining variables residing in specified registers.
68 * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files.
69 * Incomplete Enums:: @code{enum foo;}, with details to follow.
70 * Function Names:: Printable strings which are the name of the current
72 * Return Address:: Getting the return or frame address of a function.
73 * Vector Extensions:: Using vector instructions through built-in functions.
74 * Offsetof:: Special syntax for implementing @code{offsetof}.
75 * Atomic Builtins:: Built-in functions for atomic memory access.
76 * Object Size Checking:: Built-in functions for limited buffer overflow
78 * Other Builtins:: Other built-in functions.
79 * Target Builtins:: Built-in functions specific to particular targets.
80 * Target Format Checks:: Format checks specific to particular targets.
81 * Pragmas:: Pragmas accepted by GCC.
82 * Unnamed Fields:: Unnamed struct/union fields within structs/unions.
83 * Thread-Local:: Per-thread variables.
87 @section Statements and Declarations in Expressions
88 @cindex statements inside expressions
89 @cindex declarations inside expressions
90 @cindex expressions containing statements
91 @cindex macros, statements in expressions
93 @c the above section title wrapped and causes an underfull hbox.. i
94 @c changed it from "within" to "in". --mew 4feb93
95 A compound statement enclosed in parentheses may appear as an expression
96 in GNU C@. This allows you to use loops, switches, and local variables
99 Recall that a compound statement is a sequence of statements surrounded
100 by braces; in this construct, parentheses go around the braces. For
104 (@{ int y = foo (); int z;
111 is a valid (though slightly more complex than necessary) expression
112 for the absolute value of @code{foo ()}.
114 The last thing in the compound statement should be an expression
115 followed by a semicolon; the value of this subexpression serves as the
116 value of the entire construct. (If you use some other kind of statement
117 last within the braces, the construct has type @code{void}, and thus
118 effectively no value.)
120 This feature is especially useful in making macro definitions ``safe'' (so
121 that they evaluate each operand exactly once). For example, the
122 ``maximum'' function is commonly defined as a macro in standard C as
126 #define max(a,b) ((a) > (b) ? (a) : (b))
130 @cindex side effects, macro argument
131 But this definition computes either @var{a} or @var{b} twice, with bad
132 results if the operand has side effects. In GNU C, if you know the
133 type of the operands (here taken as @code{int}), you can define
134 the macro safely as follows:
137 #define maxint(a,b) \
138 (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @})
141 Embedded statements are not allowed in constant expressions, such as
142 the value of an enumeration constant, the width of a bit-field, or
143 the initial value of a static variable.
145 If you don't know the type of the operand, you can still do this, but you
146 must use @code{typeof} (@pxref{Typeof}).
148 In G++, the result value of a statement expression undergoes array and
149 function pointer decay, and is returned by value to the enclosing
150 expression. For instance, if @code{A} is a class, then
159 will construct a temporary @code{A} object to hold the result of the
160 statement expression, and that will be used to invoke @code{Foo}.
161 Therefore the @code{this} pointer observed by @code{Foo} will not be the
164 Any temporaries created within a statement within a statement expression
165 will be destroyed at the statement's end. This makes statement
166 expressions inside macros slightly different from function calls. In
167 the latter case temporaries introduced during argument evaluation will
168 be destroyed at the end of the statement that includes the function
169 call. In the statement expression case they will be destroyed during
170 the statement expression. For instance,
173 #define macro(a) (@{__typeof__(a) b = (a); b + 3; @})
174 template<typename T> T function(T a) @{ T b = a; return b + 3; @}
184 will have different places where temporaries are destroyed. For the
185 @code{macro} case, the temporary @code{X} will be destroyed just after
186 the initialization of @code{b}. In the @code{function} case that
187 temporary will be destroyed when the function returns.
189 These considerations mean that it is probably a bad idea to use
190 statement-expressions of this form in header files that are designed to
191 work with C++. (Note that some versions of the GNU C Library contained
192 header files using statement-expression that lead to precisely this
195 Jumping into a statement expression with @code{goto} or using a
196 @code{switch} statement outside the statement expression with a
197 @code{case} or @code{default} label inside the statement expression is
198 not permitted. Jumping into a statement expression with a computed
199 @code{goto} (@pxref{Labels as Values}) yields undefined behavior.
200 Jumping out of a statement expression is permitted, but if the
201 statement expression is part of a larger expression then it is
202 unspecified which other subexpressions of that expression have been
203 evaluated except where the language definition requires certain
204 subexpressions to be evaluated before or after the statement
205 expression. In any case, as with a function call the evaluation of a
206 statement expression is not interleaved with the evaluation of other
207 parts of the containing expression. For example,
210 foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz();
214 will call @code{foo} and @code{bar1} and will not call @code{baz} but
215 may or may not call @code{bar2}. If @code{bar2} is called, it will be
216 called after @code{foo} and before @code{bar1}
219 @section Locally Declared Labels
221 @cindex macros, local labels
223 GCC allows you to declare @dfn{local labels} in any nested block
224 scope. A local label is just like an ordinary label, but you can
225 only reference it (with a @code{goto} statement, or by taking its
226 address) within the block in which it was declared.
228 A local label declaration looks like this:
231 __label__ @var{label};
238 __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */;
241 Local label declarations must come at the beginning of the block,
242 before any ordinary declarations or statements.
244 The label declaration defines the label @emph{name}, but does not define
245 the label itself. You must do this in the usual way, with
246 @code{@var{label}:}, within the statements of the statement expression.
248 The local label feature is useful for complex macros. If a macro
249 contains nested loops, a @code{goto} can be useful for breaking out of
250 them. However, an ordinary label whose scope is the whole function
251 cannot be used: if the macro can be expanded several times in one
252 function, the label will be multiply defined in that function. A
253 local label avoids this problem. For example:
256 #define SEARCH(value, array, target) \
259 typeof (target) _SEARCH_target = (target); \
260 typeof (*(array)) *_SEARCH_array = (array); \
263 for (i = 0; i < max; i++) \
264 for (j = 0; j < max; j++) \
265 if (_SEARCH_array[i][j] == _SEARCH_target) \
266 @{ (value) = i; goto found; @} \
272 This could also be written using a statement-expression:
275 #define SEARCH(array, target) \
278 typeof (target) _SEARCH_target = (target); \
279 typeof (*(array)) *_SEARCH_array = (array); \
282 for (i = 0; i < max; i++) \
283 for (j = 0; j < max; j++) \
284 if (_SEARCH_array[i][j] == _SEARCH_target) \
285 @{ value = i; goto found; @} \
292 Local label declarations also make the labels they declare visible to
293 nested functions, if there are any. @xref{Nested Functions}, for details.
295 @node Labels as Values
296 @section Labels as Values
297 @cindex labels as values
298 @cindex computed gotos
299 @cindex goto with computed label
300 @cindex address of a label
302 You can get the address of a label defined in the current function
303 (or a containing function) with the unary operator @samp{&&}. The
304 value has type @code{void *}. This value is a constant and can be used
305 wherever a constant of that type is valid. For example:
313 To use these values, you need to be able to jump to one. This is done
314 with the computed goto statement@footnote{The analogous feature in
315 Fortran is called an assigned goto, but that name seems inappropriate in
316 C, where one can do more than simply store label addresses in label
317 variables.}, @code{goto *@var{exp};}. For example,
324 Any expression of type @code{void *} is allowed.
326 One way of using these constants is in initializing a static array that
327 will serve as a jump table:
330 static void *array[] = @{ &&foo, &&bar, &&hack @};
333 Then you can select a label with indexing, like this:
340 Note that this does not check whether the subscript is in bounds---array
341 indexing in C never does that.
343 Such an array of label values serves a purpose much like that of the
344 @code{switch} statement. The @code{switch} statement is cleaner, so
345 use that rather than an array unless the problem does not fit a
346 @code{switch} statement very well.
348 Another use of label values is in an interpreter for threaded code.
349 The labels within the interpreter function can be stored in the
350 threaded code for super-fast dispatching.
352 You may not use this mechanism to jump to code in a different function.
353 If you do that, totally unpredictable things will happen. The best way to
354 avoid this is to store the label address only in automatic variables and
355 never pass it as an argument.
357 An alternate way to write the above example is
360 static const int array[] = @{ &&foo - &&foo, &&bar - &&foo,
362 goto *(&&foo + array[i]);
366 This is more friendly to code living in shared libraries, as it reduces
367 the number of dynamic relocations that are needed, and by consequence,
368 allows the data to be read-only.
370 @node Nested Functions
371 @section Nested Functions
372 @cindex nested functions
373 @cindex downward funargs
376 A @dfn{nested function} is a function defined inside another function.
377 (Nested functions are not supported for GNU C++.) The nested function's
378 name is local to the block where it is defined. For example, here we
379 define a nested function named @code{square}, and call it twice:
383 foo (double a, double b)
385 double square (double z) @{ return z * z; @}
387 return square (a) + square (b);
392 The nested function can access all the variables of the containing
393 function that are visible at the point of its definition. This is
394 called @dfn{lexical scoping}. For example, here we show a nested
395 function which uses an inherited variable named @code{offset}:
399 bar (int *array, int offset, int size)
401 int access (int *array, int index)
402 @{ return array[index + offset]; @}
405 for (i = 0; i < size; i++)
406 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
411 Nested function definitions are permitted within functions in the places
412 where variable definitions are allowed; that is, in any block, mixed
413 with the other declarations and statements in the block.
415 It is possible to call the nested function from outside the scope of its
416 name by storing its address or passing the address to another function:
419 hack (int *array, int size)
421 void store (int index, int value)
422 @{ array[index] = value; @}
424 intermediate (store, size);
428 Here, the function @code{intermediate} receives the address of
429 @code{store} as an argument. If @code{intermediate} calls @code{store},
430 the arguments given to @code{store} are used to store into @code{array}.
431 But this technique works only so long as the containing function
432 (@code{hack}, in this example) does not exit.
434 If you try to call the nested function through its address after the
435 containing function has exited, all hell will break loose. If you try
436 to call it after a containing scope level has exited, and if it refers
437 to some of the variables that are no longer in scope, you may be lucky,
438 but it's not wise to take the risk. If, however, the nested function
439 does not refer to anything that has gone out of scope, you should be
442 GCC implements taking the address of a nested function using a technique
443 called @dfn{trampolines}. A paper describing them is available as
446 @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}.
448 A nested function can jump to a label inherited from a containing
449 function, provided the label was explicitly declared in the containing
450 function (@pxref{Local Labels}). Such a jump returns instantly to the
451 containing function, exiting the nested function which did the
452 @code{goto} and any intermediate functions as well. Here is an example:
456 bar (int *array, int offset, int size)
459 int access (int *array, int index)
463 return array[index + offset];
467 for (i = 0; i < size; i++)
468 /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */
472 /* @r{Control comes here from @code{access}
473 if it detects an error.} */
480 A nested function always has no linkage. Declaring one with
481 @code{extern} or @code{static} is erroneous. If you need to declare the nested function
482 before its definition, use @code{auto} (which is otherwise meaningless
483 for function declarations).
486 bar (int *array, int offset, int size)
489 auto int access (int *, int);
491 int access (int *array, int index)
495 return array[index + offset];
501 @node Constructing Calls
502 @section Constructing Function Calls
503 @cindex constructing calls
504 @cindex forwarding calls
506 Using the built-in functions described below, you can record
507 the arguments a function received, and call another function
508 with the same arguments, without knowing the number or types
511 You can also record the return value of that function call,
512 and later return that value, without knowing what data type
513 the function tried to return (as long as your caller expects
516 However, these built-in functions may interact badly with some
517 sophisticated features or other extensions of the language. It
518 is, therefore, not recommended to use them outside very simple
519 functions acting as mere forwarders for their arguments.
521 @deftypefn {Built-in Function} {void *} __builtin_apply_args ()
522 This built-in function returns a pointer to data
523 describing how to perform a call with the same arguments as were passed
524 to the current function.
526 The function saves the arg pointer register, structure value address,
527 and all registers that might be used to pass arguments to a function
528 into a block of memory allocated on the stack. Then it returns the
529 address of that block.
532 @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size})
533 This built-in function invokes @var{function}
534 with a copy of the parameters described by @var{arguments}
537 The value of @var{arguments} should be the value returned by
538 @code{__builtin_apply_args}. The argument @var{size} specifies the size
539 of the stack argument data, in bytes.
541 This function returns a pointer to data describing
542 how to return whatever value was returned by @var{function}. The data
543 is saved in a block of memory allocated on the stack.
545 It is not always simple to compute the proper value for @var{size}. The
546 value is used by @code{__builtin_apply} to compute the amount of data
547 that should be pushed on the stack and copied from the incoming argument
551 @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result})
552 This built-in function returns the value described by @var{result} from
553 the containing function. You should specify, for @var{result}, a value
554 returned by @code{__builtin_apply}.
558 @section Referring to a Type with @code{typeof}
561 @cindex macros, types of arguments
563 Another way to refer to the type of an expression is with @code{typeof}.
564 The syntax of using of this keyword looks like @code{sizeof}, but the
565 construct acts semantically like a type name defined with @code{typedef}.
567 There are two ways of writing the argument to @code{typeof}: with an
568 expression or with a type. Here is an example with an expression:
575 This assumes that @code{x} is an array of pointers to functions;
576 the type described is that of the values of the functions.
578 Here is an example with a typename as the argument:
585 Here the type described is that of pointers to @code{int}.
587 If you are writing a header file that must work when included in ISO C
588 programs, write @code{__typeof__} instead of @code{typeof}.
589 @xref{Alternate Keywords}.
591 A @code{typeof}-construct can be used anywhere a typedef name could be
592 used. For example, you can use it in a declaration, in a cast, or inside
593 of @code{sizeof} or @code{typeof}.
595 @code{typeof} is often useful in conjunction with the
596 statements-within-expressions feature. Here is how the two together can
597 be used to define a safe ``maximum'' macro that operates on any
598 arithmetic type and evaluates each of its arguments exactly once:
602 (@{ typeof (a) _a = (a); \
603 typeof (b) _b = (b); \
604 _a > _b ? _a : _b; @})
607 @cindex underscores in variables in macros
608 @cindex @samp{_} in variables in macros
609 @cindex local variables in macros
610 @cindex variables, local, in macros
611 @cindex macros, local variables in
613 The reason for using names that start with underscores for the local
614 variables is to avoid conflicts with variable names that occur within the
615 expressions that are substituted for @code{a} and @code{b}. Eventually we
616 hope to design a new form of declaration syntax that allows you to declare
617 variables whose scopes start only after their initializers; this will be a
618 more reliable way to prevent such conflicts.
621 Some more examples of the use of @code{typeof}:
625 This declares @code{y} with the type of what @code{x} points to.
632 This declares @code{y} as an array of such values.
639 This declares @code{y} as an array of pointers to characters:
642 typeof (typeof (char *)[4]) y;
646 It is equivalent to the following traditional C declaration:
652 To see the meaning of the declaration using @code{typeof}, and why it
653 might be a useful way to write, rewrite it with these macros:
656 #define pointer(T) typeof(T *)
657 #define array(T, N) typeof(T [N])
661 Now the declaration can be rewritten this way:
664 array (pointer (char), 4) y;
668 Thus, @code{array (pointer (char), 4)} is the type of arrays of 4
669 pointers to @code{char}.
672 @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported
673 a more limited extension which permitted one to write
676 typedef @var{T} = @var{expr};
680 with the effect of declaring @var{T} to have the type of the expression
681 @var{expr}. This extension does not work with GCC 3 (versions between
682 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which
683 relies on it should be rewritten to use @code{typeof}:
686 typedef typeof(@var{expr}) @var{T};
690 This will work with all versions of GCC@.
693 @section Conditionals with Omitted Operands
694 @cindex conditional expressions, extensions
695 @cindex omitted middle-operands
696 @cindex middle-operands, omitted
697 @cindex extensions, @code{?:}
698 @cindex @code{?:} extensions
700 The middle operand in a conditional expression may be omitted. Then
701 if the first operand is nonzero, its value is the value of the conditional
704 Therefore, the expression
711 has the value of @code{x} if that is nonzero; otherwise, the value of
714 This example is perfectly equivalent to
720 @cindex side effect in ?:
721 @cindex ?: side effect
723 In this simple case, the ability to omit the middle operand is not
724 especially useful. When it becomes useful is when the first operand does,
725 or may (if it is a macro argument), contain a side effect. Then repeating
726 the operand in the middle would perform the side effect twice. Omitting
727 the middle operand uses the value already computed without the undesirable
728 effects of recomputing it.
731 @section Double-Word Integers
732 @cindex @code{long long} data types
733 @cindex double-word arithmetic
734 @cindex multiprecision arithmetic
735 @cindex @code{LL} integer suffix
736 @cindex @code{ULL} integer suffix
738 ISO C99 supports data types for integers that are at least 64 bits wide,
739 and as an extension GCC supports them in C89 mode and in C++.
740 Simply write @code{long long int} for a signed integer, or
741 @code{unsigned long long int} for an unsigned integer. To make an
742 integer constant of type @code{long long int}, add the suffix @samp{LL}
743 to the integer. To make an integer constant of type @code{unsigned long
744 long int}, add the suffix @samp{ULL} to the integer.
746 You can use these types in arithmetic like any other integer types.
747 Addition, subtraction, and bitwise boolean operations on these types
748 are open-coded on all types of machines. Multiplication is open-coded
749 if the machine supports fullword-to-doubleword a widening multiply
750 instruction. Division and shifts are open-coded only on machines that
751 provide special support. The operations that are not open-coded use
752 special library routines that come with GCC@.
754 There may be pitfalls when you use @code{long long} types for function
755 arguments, unless you declare function prototypes. If a function
756 expects type @code{int} for its argument, and you pass a value of type
757 @code{long long int}, confusion will result because the caller and the
758 subroutine will disagree about the number of bytes for the argument.
759 Likewise, if the function expects @code{long long int} and you pass
760 @code{int}. The best way to avoid such problems is to use prototypes.
763 @section Complex Numbers
764 @cindex complex numbers
765 @cindex @code{_Complex} keyword
766 @cindex @code{__complex__} keyword
768 ISO C99 supports complex floating data types, and as an extension GCC
769 supports them in C89 mode and in C++, and supports complex integer data
770 types which are not part of ISO C99. You can declare complex types
771 using the keyword @code{_Complex}. As an extension, the older GNU
772 keyword @code{__complex__} is also supported.
774 For example, @samp{_Complex double x;} declares @code{x} as a
775 variable whose real part and imaginary part are both of type
776 @code{double}. @samp{_Complex short int y;} declares @code{y} to
777 have real and imaginary parts of type @code{short int}; this is not
778 likely to be useful, but it shows that the set of complex types is
781 To write a constant with a complex data type, use the suffix @samp{i} or
782 @samp{j} (either one; they are equivalent). For example, @code{2.5fi}
783 has type @code{_Complex float} and @code{3i} has type
784 @code{_Complex int}. Such a constant always has a pure imaginary
785 value, but you can form any complex value you like by adding one to a
786 real constant. This is a GNU extension; if you have an ISO C99
787 conforming C library (such as GNU libc), and want to construct complex
788 constants of floating type, you should include @code{<complex.h>} and
789 use the macros @code{I} or @code{_Complex_I} instead.
791 @cindex @code{__real__} keyword
792 @cindex @code{__imag__} keyword
793 To extract the real part of a complex-valued expression @var{exp}, write
794 @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to
795 extract the imaginary part. This is a GNU extension; for values of
796 floating type, you should use the ISO C99 functions @code{crealf},
797 @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and
798 @code{cimagl}, declared in @code{<complex.h>} and also provided as
799 built-in functions by GCC@.
801 @cindex complex conjugation
802 The operator @samp{~} performs complex conjugation when used on a value
803 with a complex type. This is a GNU extension; for values of
804 floating type, you should use the ISO C99 functions @code{conjf},
805 @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also
806 provided as built-in functions by GCC@.
808 GCC can allocate complex automatic variables in a noncontiguous
809 fashion; it's even possible for the real part to be in a register while
810 the imaginary part is on the stack (or vice-versa). Only the DWARF2
811 debug info format can represent this, so use of DWARF2 is recommended.
812 If you are using the stabs debug info format, GCC describes a noncontiguous
813 complex variable as if it were two separate variables of noncomplex type.
814 If the variable's actual name is @code{foo}, the two fictitious
815 variables are named @code{foo$real} and @code{foo$imag}. You can
816 examine and set these two fictitious variables with your debugger.
819 @section Decimal Floating Point
820 @cindex decimal floating point
821 @cindex @code{_Decimal32} data type
822 @cindex @code{_Decimal64} data type
823 @cindex @code{_Decimal128} data type
824 @cindex @code{df} integer suffix
825 @cindex @code{dd} integer suffix
826 @cindex @code{dl} integer suffix
827 @cindex @code{DF} integer suffix
828 @cindex @code{DD} integer suffix
829 @cindex @code{DL} integer suffix
831 GNU C supports decimal floating point types in addition to the
832 standard floating-point types. This extension supports decimal
833 floating-point arithmetic as defined in IEEE-754R, the proposed
834 revision of IEEE-754. The C language extension is defined in ISO/IEC
835 DTR 24732, Draft 5. Support for this functionality will change when
836 it is accepted into the C standard and might change for new drafts
837 of the proposal. Calling conventions for any target might also change.
838 Not all targets support decimal floating point.
840 Support for decimal floating point includes the arithmetic operators
841 add, subtract, multiply, divide; unary arithmetic operators;
842 relational operators; equality operators; and conversions to and from
843 integer and other floating-point types. Use a suffix @samp{df} or
844 @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd}
845 or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for
848 Passing a decimal floating-point value as an argument to a function
849 without a prototype is undefined.
851 Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128}
852 are supported by the DWARF2 debug information format.
858 ISO C99 supports floating-point numbers written not only in the usual
859 decimal notation, such as @code{1.55e1}, but also numbers such as
860 @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC
861 supports this in C89 mode (except in some cases when strictly
862 conforming) and in C++. In that format the
863 @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are
864 mandatory. The exponent is a decimal number that indicates the power of
865 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is
872 @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3}
873 is the same as @code{1.55e1}.
875 Unlike for floating-point numbers in the decimal notation the exponent
876 is always required in the hexadecimal notation. Otherwise the compiler
877 would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This
878 could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the
879 extension for floating-point constants of type @code{float}.
882 @section Arrays of Length Zero
883 @cindex arrays of length zero
884 @cindex zero-length arrays
885 @cindex length-zero arrays
886 @cindex flexible array members
888 Zero-length arrays are allowed in GNU C@. They are very useful as the
889 last element of a structure which is really a header for a variable-length
898 struct line *thisline = (struct line *)
899 malloc (sizeof (struct line) + this_length);
900 thisline->length = this_length;
903 In ISO C90, you would have to give @code{contents} a length of 1, which
904 means either you waste space or complicate the argument to @code{malloc}.
906 In ISO C99, you would use a @dfn{flexible array member}, which is
907 slightly different in syntax and semantics:
911 Flexible array members are written as @code{contents[]} without
915 Flexible array members have incomplete type, and so the @code{sizeof}
916 operator may not be applied. As a quirk of the original implementation
917 of zero-length arrays, @code{sizeof} evaluates to zero.
920 Flexible array members may only appear as the last member of a
921 @code{struct} that is otherwise non-empty.
924 A structure containing a flexible array member, or a union containing
925 such a structure (possibly recursively), may not be a member of a
926 structure or an element of an array. (However, these uses are
927 permitted by GCC as extensions.)
930 GCC versions before 3.0 allowed zero-length arrays to be statically
931 initialized, as if they were flexible arrays. In addition to those
932 cases that were useful, it also allowed initializations in situations
933 that would corrupt later data. Non-empty initialization of zero-length
934 arrays is now treated like any case where there are more initializer
935 elements than the array holds, in that a suitable warning about "excess
936 elements in array" is given, and the excess elements (all of them, in
937 this case) are ignored.
939 Instead GCC allows static initialization of flexible array members.
940 This is equivalent to defining a new structure containing the original
941 structure followed by an array of sufficient size to contain the data.
942 I.e.@: in the following, @code{f1} is constructed as if it were declared
948 @} f1 = @{ 1, @{ 2, 3, 4 @} @};
951 struct f1 f1; int data[3];
952 @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @};
956 The convenience of this extension is that @code{f1} has the desired
957 type, eliminating the need to consistently refer to @code{f2.f1}.
959 This has symmetry with normal static arrays, in that an array of
960 unknown size is also written with @code{[]}.
962 Of course, this extension only makes sense if the extra data comes at
963 the end of a top-level object, as otherwise we would be overwriting
964 data at subsequent offsets. To avoid undue complication and confusion
965 with initialization of deeply nested arrays, we simply disallow any
966 non-empty initialization except when the structure is the top-level
970 struct foo @{ int x; int y[]; @};
971 struct bar @{ struct foo z; @};
973 struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.}
974 struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
975 struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.}
976 struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.}
979 @node Empty Structures
980 @section Structures With No Members
981 @cindex empty structures
982 @cindex zero-size structures
984 GCC permits a C structure to have no members:
991 The structure will have size zero. In C++, empty structures are part
992 of the language. G++ treats empty structures as if they had a single
993 member of type @code{char}.
995 @node Variable Length
996 @section Arrays of Variable Length
997 @cindex variable-length arrays
998 @cindex arrays of variable length
1001 Variable-length automatic arrays are allowed in ISO C99, and as an
1002 extension GCC accepts them in C89 mode and in C++. (However, GCC's
1003 implementation of variable-length arrays does not yet conform in detail
1004 to the ISO C99 standard.) These arrays are
1005 declared like any other automatic arrays, but with a length that is not
1006 a constant expression. The storage is allocated at the point of
1007 declaration and deallocated when the brace-level is exited. For
1012 concat_fopen (char *s1, char *s2, char *mode)
1014 char str[strlen (s1) + strlen (s2) + 1];
1017 return fopen (str, mode);
1021 @cindex scope of a variable length array
1022 @cindex variable-length array scope
1023 @cindex deallocating variable length arrays
1024 Jumping or breaking out of the scope of the array name deallocates the
1025 storage. Jumping into the scope is not allowed; you get an error
1028 @cindex @code{alloca} vs variable-length arrays
1029 You can use the function @code{alloca} to get an effect much like
1030 variable-length arrays. The function @code{alloca} is available in
1031 many other C implementations (but not in all). On the other hand,
1032 variable-length arrays are more elegant.
1034 There are other differences between these two methods. Space allocated
1035 with @code{alloca} exists until the containing @emph{function} returns.
1036 The space for a variable-length array is deallocated as soon as the array
1037 name's scope ends. (If you use both variable-length arrays and
1038 @code{alloca} in the same function, deallocation of a variable-length array
1039 will also deallocate anything more recently allocated with @code{alloca}.)
1041 You can also use variable-length arrays as arguments to functions:
1045 tester (int len, char data[len][len])
1051 The length of an array is computed once when the storage is allocated
1052 and is remembered for the scope of the array in case you access it with
1055 If you want to pass the array first and the length afterward, you can
1056 use a forward declaration in the parameter list---another GNU extension.
1060 tester (int len; char data[len][len], int len)
1066 @cindex parameter forward declaration
1067 The @samp{int len} before the semicolon is a @dfn{parameter forward
1068 declaration}, and it serves the purpose of making the name @code{len}
1069 known when the declaration of @code{data} is parsed.
1071 You can write any number of such parameter forward declarations in the
1072 parameter list. They can be separated by commas or semicolons, but the
1073 last one must end with a semicolon, which is followed by the ``real''
1074 parameter declarations. Each forward declaration must match a ``real''
1075 declaration in parameter name and data type. ISO C99 does not support
1076 parameter forward declarations.
1078 @node Variadic Macros
1079 @section Macros with a Variable Number of Arguments.
1080 @cindex variable number of arguments
1081 @cindex macro with variable arguments
1082 @cindex rest argument (in macro)
1083 @cindex variadic macros
1085 In the ISO C standard of 1999, a macro can be declared to accept a
1086 variable number of arguments much as a function can. The syntax for
1087 defining the macro is similar to that of a function. Here is an
1091 #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__)
1094 Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of
1095 such a macro, it represents the zero or more tokens until the closing
1096 parenthesis that ends the invocation, including any commas. This set of
1097 tokens replaces the identifier @code{__VA_ARGS__} in the macro body
1098 wherever it appears. See the CPP manual for more information.
1100 GCC has long supported variadic macros, and used a different syntax that
1101 allowed you to give a name to the variable arguments just like any other
1102 argument. Here is an example:
1105 #define debug(format, args...) fprintf (stderr, format, args)
1108 This is in all ways equivalent to the ISO C example above, but arguably
1109 more readable and descriptive.
1111 GNU CPP has two further variadic macro extensions, and permits them to
1112 be used with either of the above forms of macro definition.
1114 In standard C, you are not allowed to leave the variable argument out
1115 entirely; but you are allowed to pass an empty argument. For example,
1116 this invocation is invalid in ISO C, because there is no comma after
1123 GNU CPP permits you to completely omit the variable arguments in this
1124 way. In the above examples, the compiler would complain, though since
1125 the expansion of the macro still has the extra comma after the format
1128 To help solve this problem, CPP behaves specially for variable arguments
1129 used with the token paste operator, @samp{##}. If instead you write
1132 #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__)
1135 and if the variable arguments are omitted or empty, the @samp{##}
1136 operator causes the preprocessor to remove the comma before it. If you
1137 do provide some variable arguments in your macro invocation, GNU CPP
1138 does not complain about the paste operation and instead places the
1139 variable arguments after the comma. Just like any other pasted macro
1140 argument, these arguments are not macro expanded.
1142 @node Escaped Newlines
1143 @section Slightly Looser Rules for Escaped Newlines
1144 @cindex escaped newlines
1145 @cindex newlines (escaped)
1147 Recently, the preprocessor has relaxed its treatment of escaped
1148 newlines. Previously, the newline had to immediately follow a
1149 backslash. The current implementation allows whitespace in the form
1150 of spaces, horizontal and vertical tabs, and form feeds between the
1151 backslash and the subsequent newline. The preprocessor issues a
1152 warning, but treats it as a valid escaped newline and combines the two
1153 lines to form a single logical line. This works within comments and
1154 tokens, as well as between tokens. Comments are @emph{not} treated as
1155 whitespace for the purposes of this relaxation, since they have not
1156 yet been replaced with spaces.
1159 @section Non-Lvalue Arrays May Have Subscripts
1160 @cindex subscripting
1161 @cindex arrays, non-lvalue
1163 @cindex subscripting and function values
1164 In ISO C99, arrays that are not lvalues still decay to pointers, and
1165 may be subscripted, although they may not be modified or used after
1166 the next sequence point and the unary @samp{&} operator may not be
1167 applied to them. As an extension, GCC allows such arrays to be
1168 subscripted in C89 mode, though otherwise they do not decay to
1169 pointers outside C99 mode. For example,
1170 this is valid in GNU C though not valid in C89:
1174 struct foo @{int a[4];@};
1180 return f().a[index];
1186 @section Arithmetic on @code{void}- and Function-Pointers
1187 @cindex void pointers, arithmetic
1188 @cindex void, size of pointer to
1189 @cindex function pointers, arithmetic
1190 @cindex function, size of pointer to
1192 In GNU C, addition and subtraction operations are supported on pointers to
1193 @code{void} and on pointers to functions. This is done by treating the
1194 size of a @code{void} or of a function as 1.
1196 A consequence of this is that @code{sizeof} is also allowed on @code{void}
1197 and on function types, and returns 1.
1199 @opindex Wpointer-arith
1200 The option @option{-Wpointer-arith} requests a warning if these extensions
1204 @section Non-Constant Initializers
1205 @cindex initializers, non-constant
1206 @cindex non-constant initializers
1208 As in standard C++ and ISO C99, the elements of an aggregate initializer for an
1209 automatic variable are not required to be constant expressions in GNU C@.
1210 Here is an example of an initializer with run-time varying elements:
1213 foo (float f, float g)
1215 float beat_freqs[2] = @{ f-g, f+g @};
1220 @node Compound Literals
1221 @section Compound Literals
1222 @cindex constructor expressions
1223 @cindex initializations in expressions
1224 @cindex structures, constructor expression
1225 @cindex expressions, constructor
1226 @cindex compound literals
1227 @c The GNU C name for what C99 calls compound literals was "constructor expressions".
1229 ISO C99 supports compound literals. A compound literal looks like
1230 a cast containing an initializer. Its value is an object of the
1231 type specified in the cast, containing the elements specified in
1232 the initializer; it is an lvalue. As an extension, GCC supports
1233 compound literals in C89 mode and in C++.
1235 Usually, the specified type is a structure. Assume that
1236 @code{struct foo} and @code{structure} are declared as shown:
1239 struct foo @{int a; char b[2];@} structure;
1243 Here is an example of constructing a @code{struct foo} with a compound literal:
1246 structure = ((struct foo) @{x + y, 'a', 0@});
1250 This is equivalent to writing the following:
1254 struct foo temp = @{x + y, 'a', 0@};
1259 You can also construct an array. If all the elements of the compound literal
1260 are (made up of) simple constant expressions, suitable for use in
1261 initializers of objects of static storage duration, then the compound
1262 literal can be coerced to a pointer to its first element and used in
1263 such an initializer, as shown here:
1266 char **foo = (char *[]) @{ "x", "y", "z" @};
1269 Compound literals for scalar types and union types are is
1270 also allowed, but then the compound literal is equivalent
1273 As a GNU extension, GCC allows initialization of objects with static storage
1274 duration by compound literals (which is not possible in ISO C99, because
1275 the initializer is not a constant).
1276 It is handled as if the object was initialized only with the bracket
1277 enclosed list if compound literal's and object types match.
1278 The initializer list of the compound literal must be constant.
1279 If the object being initialized has array type of unknown size, the size is
1280 determined by compound literal size.
1283 static struct foo x = (struct foo) @{1, 'a', 'b'@};
1284 static int y[] = (int []) @{1, 2, 3@};
1285 static int z[] = (int [3]) @{1@};
1289 The above lines are equivalent to the following:
1291 static struct foo x = @{1, 'a', 'b'@};
1292 static int y[] = @{1, 2, 3@};
1293 static int z[] = @{1, 0, 0@};
1296 @node Designated Inits
1297 @section Designated Initializers
1298 @cindex initializers with labeled elements
1299 @cindex labeled elements in initializers
1300 @cindex case labels in initializers
1301 @cindex designated initializers
1303 Standard C89 requires the elements of an initializer to appear in a fixed
1304 order, the same as the order of the elements in the array or structure
1307 In ISO C99 you can give the elements in any order, specifying the array
1308 indices or structure field names they apply to, and GNU C allows this as
1309 an extension in C89 mode as well. This extension is not
1310 implemented in GNU C++.
1312 To specify an array index, write
1313 @samp{[@var{index}] =} before the element value. For example,
1316 int a[6] = @{ [4] = 29, [2] = 15 @};
1323 int a[6] = @{ 0, 0, 15, 0, 29, 0 @};
1327 The index values must be constant expressions, even if the array being
1328 initialized is automatic.
1330 An alternative syntax for this which has been obsolete since GCC 2.5 but
1331 GCC still accepts is to write @samp{[@var{index}]} before the element
1332 value, with no @samp{=}.
1334 To initialize a range of elements to the same value, write
1335 @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU
1336 extension. For example,
1339 int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @};
1343 If the value in it has side-effects, the side-effects will happen only once,
1344 not for each initialized field by the range initializer.
1347 Note that the length of the array is the highest value specified
1350 In a structure initializer, specify the name of a field to initialize
1351 with @samp{.@var{fieldname} =} before the element value. For example,
1352 given the following structure,
1355 struct point @{ int x, y; @};
1359 the following initialization
1362 struct point p = @{ .y = yvalue, .x = xvalue @};
1369 struct point p = @{ xvalue, yvalue @};
1372 Another syntax which has the same meaning, obsolete since GCC 2.5, is
1373 @samp{@var{fieldname}:}, as shown here:
1376 struct point p = @{ y: yvalue, x: xvalue @};
1380 The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a
1381 @dfn{designator}. You can also use a designator (or the obsolete colon
1382 syntax) when initializing a union, to specify which element of the union
1383 should be used. For example,
1386 union foo @{ int i; double d; @};
1388 union foo f = @{ .d = 4 @};
1392 will convert 4 to a @code{double} to store it in the union using
1393 the second element. By contrast, casting 4 to type @code{union foo}
1394 would store it into the union as the integer @code{i}, since it is
1395 an integer. (@xref{Cast to Union}.)
1397 You can combine this technique of naming elements with ordinary C
1398 initialization of successive elements. Each initializer element that
1399 does not have a designator applies to the next consecutive element of the
1400 array or structure. For example,
1403 int a[6] = @{ [1] = v1, v2, [4] = v4 @};
1410 int a[6] = @{ 0, v1, v2, 0, v4, 0 @};
1413 Labeling the elements of an array initializer is especially useful
1414 when the indices are characters or belong to an @code{enum} type.
1419 = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1,
1420 ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @};
1423 @cindex designator lists
1424 You can also write a series of @samp{.@var{fieldname}} and
1425 @samp{[@var{index}]} designators before an @samp{=} to specify a
1426 nested subobject to initialize; the list is taken relative to the
1427 subobject corresponding to the closest surrounding brace pair. For
1428 example, with the @samp{struct point} declaration above:
1431 struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @};
1435 If the same field is initialized multiple times, it will have value from
1436 the last initialization. If any such overridden initialization has
1437 side-effect, it is unspecified whether the side-effect happens or not.
1438 Currently, GCC will discard them and issue a warning.
1441 @section Case Ranges
1443 @cindex ranges in case statements
1445 You can specify a range of consecutive values in a single @code{case} label,
1449 case @var{low} ... @var{high}:
1453 This has the same effect as the proper number of individual @code{case}
1454 labels, one for each integer value from @var{low} to @var{high}, inclusive.
1456 This feature is especially useful for ranges of ASCII character codes:
1462 @strong{Be careful:} Write spaces around the @code{...}, for otherwise
1463 it may be parsed wrong when you use it with integer values. For example,
1478 @section Cast to a Union Type
1479 @cindex cast to a union
1480 @cindex union, casting to a
1482 A cast to union type is similar to other casts, except that the type
1483 specified is a union type. You can specify the type either with
1484 @code{union @var{tag}} or with a typedef name. A cast to union is actually
1485 a constructor though, not a cast, and hence does not yield an lvalue like
1486 normal casts. (@xref{Compound Literals}.)
1488 The types that may be cast to the union type are those of the members
1489 of the union. Thus, given the following union and variables:
1492 union foo @{ int i; double d; @};
1498 both @code{x} and @code{y} can be cast to type @code{union foo}.
1500 Using the cast as the right-hand side of an assignment to a variable of
1501 union type is equivalent to storing in a member of the union:
1506 u = (union foo) x @equiv{} u.i = x
1507 u = (union foo) y @equiv{} u.d = y
1510 You can also use the union cast as a function argument:
1513 void hack (union foo);
1515 hack ((union foo) x);
1518 @node Mixed Declarations
1519 @section Mixed Declarations and Code
1520 @cindex mixed declarations and code
1521 @cindex declarations, mixed with code
1522 @cindex code, mixed with declarations
1524 ISO C99 and ISO C++ allow declarations and code to be freely mixed
1525 within compound statements. As an extension, GCC also allows this in
1526 C89 mode. For example, you could do:
1535 Each identifier is visible from where it is declared until the end of
1536 the enclosing block.
1538 @node Function Attributes
1539 @section Declaring Attributes of Functions
1540 @cindex function attributes
1541 @cindex declaring attributes of functions
1542 @cindex functions that never return
1543 @cindex functions that return more than once
1544 @cindex functions that have no side effects
1545 @cindex functions in arbitrary sections
1546 @cindex functions that behave like malloc
1547 @cindex @code{volatile} applied to function
1548 @cindex @code{const} applied to function
1549 @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments
1550 @cindex functions with non-null pointer arguments
1551 @cindex functions that are passed arguments in registers on the 386
1552 @cindex functions that pop the argument stack on the 386
1553 @cindex functions that do not pop the argument stack on the 386
1555 In GNU C, you declare certain things about functions called in your program
1556 which help the compiler optimize function calls and check your code more
1559 The keyword @code{__attribute__} allows you to specify special
1560 attributes when making a declaration. This keyword is followed by an
1561 attribute specification inside double parentheses. The following
1562 attributes are currently defined for functions on all targets:
1563 @code{noreturn}, @code{returns_twice}, @code{noinline}, @code{always_inline},
1564 @code{flatten}, @code{pure}, @code{const}, @code{nothrow}, @code{sentinel},
1565 @code{format}, @code{format_arg}, @code{no_instrument_function},
1566 @code{section}, @code{constructor}, @code{destructor}, @code{used},
1567 @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
1568 @code{alias}, @code{warn_unused_result}, @code{nonnull}
1569 and @code{externally_visible}. Several other
1570 attributes are defined for functions on particular target systems. Other
1571 attributes, including @code{section} are supported for variables declarations
1572 (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
1574 You may also specify attributes with @samp{__} preceding and following
1575 each keyword. This allows you to use them in header files without
1576 being concerned about a possible macro of the same name. For example,
1577 you may use @code{__noreturn__} instead of @code{noreturn}.
1579 @xref{Attribute Syntax}, for details of the exact syntax for using
1583 @c Keep this table alphabetized by attribute name. Treat _ as space.
1585 @item alias ("@var{target}")
1586 @cindex @code{alias} attribute
1587 The @code{alias} attribute causes the declaration to be emitted as an
1588 alias for another symbol, which must be specified. For instance,
1591 void __f () @{ /* @r{Do something.} */; @}
1592 void f () __attribute__ ((weak, alias ("__f")));
1595 defines @samp{f} to be a weak alias for @samp{__f}. In C++, the
1596 mangled name for the target must be used. It is an error if @samp{__f}
1597 is not defined in the same translation unit.
1599 Not all target machines support this attribute.
1602 @cindex @code{always_inline} function attribute
1603 Generally, functions are not inlined unless optimization is specified.
1604 For functions declared inline, this attribute inlines the function even
1605 if no optimization level was specified.
1607 @cindex @code{flatten} function attribute
1609 Generally, inlining into a function is limited. For a function marked with
1610 this attribute, every call inside this function will be inlined, if possible.
1611 Whether the function itself is considered for inlining depends on its size and
1612 the current inlining parameters. The @code{flatten} attribute only works
1613 reliably in unit-at-a-time mode.
1616 @cindex functions that do pop the argument stack on the 386
1618 On the Intel 386, the @code{cdecl} attribute causes the compiler to
1619 assume that the calling function will pop off the stack space used to
1620 pass arguments. This is
1621 useful to override the effects of the @option{-mrtd} switch.
1624 @cindex @code{const} function attribute
1625 Many functions do not examine any values except their arguments, and
1626 have no effects except the return value. Basically this is just slightly
1627 more strict class than the @code{pure} attribute below, since function is not
1628 allowed to read global memory.
1630 @cindex pointer arguments
1631 Note that a function that has pointer arguments and examines the data
1632 pointed to must @emph{not} be declared @code{const}. Likewise, a
1633 function that calls a non-@code{const} function usually must not be
1634 @code{const}. It does not make sense for a @code{const} function to
1637 The attribute @code{const} is not implemented in GCC versions earlier
1638 than 2.5. An alternative way to declare that a function has no side
1639 effects, which works in the current version and in some older versions,
1643 typedef int intfn ();
1645 extern const intfn square;
1648 This approach does not work in GNU C++ from 2.6.0 on, since the language
1649 specifies that the @samp{const} must be attached to the return value.
1653 @cindex @code{constructor} function attribute
1654 @cindex @code{destructor} function attribute
1655 The @code{constructor} attribute causes the function to be called
1656 automatically before execution enters @code{main ()}. Similarly, the
1657 @code{destructor} attribute causes the function to be called
1658 automatically after @code{main ()} has completed or @code{exit ()} has
1659 been called. Functions with these attributes are useful for
1660 initializing data that will be used implicitly during the execution of
1663 These attributes are not currently implemented for Objective-C@.
1666 @cindex @code{deprecated} attribute.
1667 The @code{deprecated} attribute results in a warning if the function
1668 is used anywhere in the source file. This is useful when identifying
1669 functions that are expected to be removed in a future version of a
1670 program. The warning also includes the location of the declaration
1671 of the deprecated function, to enable users to easily find further
1672 information about why the function is deprecated, or what they should
1673 do instead. Note that the warnings only occurs for uses:
1676 int old_fn () __attribute__ ((deprecated));
1678 int (*fn_ptr)() = old_fn;
1681 results in a warning on line 3 but not line 2.
1683 The @code{deprecated} attribute can also be used for variables and
1684 types (@pxref{Variable Attributes}, @pxref{Type Attributes}.)
1687 @cindex @code{__declspec(dllexport)}
1688 On Microsoft Windows targets and Symbian OS targets the
1689 @code{dllexport} attribute causes the compiler to provide a global
1690 pointer to a pointer in a DLL, so that it can be referenced with the
1691 @code{dllimport} attribute. On Microsoft Windows targets, the pointer
1692 name is formed by combining @code{_imp__} and the function or variable
1695 You can use @code{__declspec(dllexport)} as a synonym for
1696 @code{__attribute__ ((dllexport))} for compatibility with other
1699 On systems that support the @code{visibility} attribute, this
1700 attribute also implies ``default'' visibility, unless a
1701 @code{visibility} attribute is explicitly specified. You should avoid
1702 the use of @code{dllexport} with ``hidden'' or ``internal''
1703 visibility; in the future GCC may issue an error for those cases.
1705 Currently, the @code{dllexport} attribute is ignored for inlined
1706 functions, unless the @option{-fkeep-inline-functions} flag has been
1707 used. The attribute is also ignored for undefined symbols.
1709 When applied to C++ classes, the attribute marks defined non-inlined
1710 member functions and static data members as exports. Static consts
1711 initialized in-class are not marked unless they are also defined
1714 For Microsoft Windows targets there are alternative methods for
1715 including the symbol in the DLL's export table such as using a
1716 @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using
1717 the @option{--export-all} linker flag.
1720 @cindex @code{__declspec(dllimport)}
1721 On Microsoft Windows and Symbian OS targets, the @code{dllimport}
1722 attribute causes the compiler to reference a function or variable via
1723 a global pointer to a pointer that is set up by the DLL exporting the
1724 symbol. The attribute implies @code{extern} storage. On Microsoft
1725 Windows targets, the pointer name is formed by combining @code{_imp__}
1726 and the function or variable name.
1728 You can use @code{__declspec(dllimport)} as a synonym for
1729 @code{__attribute__ ((dllimport))} for compatibility with other
1732 Currently, the attribute is ignored for inlined functions. If the
1733 attribute is applied to a symbol @emph{definition}, an error is reported.
1734 If a symbol previously declared @code{dllimport} is later defined, the
1735 attribute is ignored in subsequent references, and a warning is emitted.
1736 The attribute is also overridden by a subsequent declaration as
1739 When applied to C++ classes, the attribute marks non-inlined
1740 member functions and static data members as imports. However, the
1741 attribute is ignored for virtual methods to allow creation of vtables
1744 On the SH Symbian OS target the @code{dllimport} attribute also has
1745 another affect---it can cause the vtable and run-time type information
1746 for a class to be exported. This happens when the class has a
1747 dllimport'ed constructor or a non-inline, non-pure virtual function
1748 and, for either of those two conditions, the class also has a inline
1749 constructor or destructor and has a key function that is defined in
1750 the current translation unit.
1752 For Microsoft Windows based targets the use of the @code{dllimport}
1753 attribute on functions is not necessary, but provides a small
1754 performance benefit by eliminating a thunk in the DLL@. The use of the
1755 @code{dllimport} attribute on imported variables was required on older
1756 versions of the GNU linker, but can now be avoided by passing the
1757 @option{--enable-auto-import} switch to the GNU linker. As with
1758 functions, using the attribute for a variable eliminates a thunk in
1761 One drawback to using this attribute is that a pointer to a function
1762 or variable marked as @code{dllimport} cannot be used as a constant
1763 address. On Microsoft Windows targets, the attribute can be disabled
1764 for functions by setting the @option{-mnop-fun-dllimport} flag.
1767 @cindex eight bit data on the H8/300, H8/300H, and H8S
1768 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1769 variable should be placed into the eight bit data section.
1770 The compiler will generate more efficient code for certain operations
1771 on data in the eight bit data area. Note the eight bit data area is limited to
1774 You must use GAS and GLD from GNU binutils version 2.7 or later for
1775 this attribute to work correctly.
1777 @item exception_handler
1778 @cindex exception handler functions on the Blackfin processor
1779 Use this attribute on the Blackfin to indicate that the specified function
1780 is an exception handler. The compiler will generate function entry and
1781 exit sequences suitable for use in an exception handler when this
1782 attribute is present.
1785 @cindex functions which handle memory bank switching
1786 On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to
1787 use a calling convention that takes care of switching memory banks when
1788 entering and leaving a function. This calling convention is also the
1789 default when using the @option{-mlong-calls} option.
1791 On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions
1792 to call and return from a function.
1794 On 68HC11 the compiler will generate a sequence of instructions
1795 to invoke a board-specific routine to switch the memory bank and call the
1796 real function. The board-specific routine simulates a @code{call}.
1797 At the end of a function, it will jump to a board-specific routine
1798 instead of using @code{rts}. The board-specific return routine simulates
1802 @cindex functions that pop the argument stack on the 386
1803 On the Intel 386, the @code{fastcall} attribute causes the compiler to
1804 pass the first argument (if of integral type) in the register ECX and
1805 the second argument (if of integral type) in the register EDX@. Subsequent
1806 and other typed arguments are passed on the stack. The called function will
1807 pop the arguments off the stack. If the number of arguments is variable all
1808 arguments are pushed on the stack.
1810 @item format (@var{archetype}, @var{string-index}, @var{first-to-check})
1811 @cindex @code{format} function attribute
1813 The @code{format} attribute specifies that a function takes @code{printf},
1814 @code{scanf}, @code{strftime} or @code{strfmon} style arguments which
1815 should be type-checked against a format string. For example, the
1820 my_printf (void *my_object, const char *my_format, ...)
1821 __attribute__ ((format (printf, 2, 3)));
1825 causes the compiler to check the arguments in calls to @code{my_printf}
1826 for consistency with the @code{printf} style format string argument
1829 The parameter @var{archetype} determines how the format string is
1830 interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}
1831 or @code{strfmon}. (You can also use @code{__printf__},
1832 @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The
1833 parameter @var{string-index} specifies which argument is the format
1834 string argument (starting from 1), while @var{first-to-check} is the
1835 number of the first argument to check against the format string. For
1836 functions where the arguments are not available to be checked (such as
1837 @code{vprintf}), specify the third parameter as zero. In this case the
1838 compiler only checks the format string for consistency. For
1839 @code{strftime} formats, the third parameter is required to be zero.
1840 Since non-static C++ methods have an implicit @code{this} argument, the
1841 arguments of such methods should be counted from two, not one, when
1842 giving values for @var{string-index} and @var{first-to-check}.
1844 In the example above, the format string (@code{my_format}) is the second
1845 argument of the function @code{my_print}, and the arguments to check
1846 start with the third argument, so the correct parameters for the format
1847 attribute are 2 and 3.
1849 @opindex ffreestanding
1850 @opindex fno-builtin
1851 The @code{format} attribute allows you to identify your own functions
1852 which take format strings as arguments, so that GCC can check the
1853 calls to these functions for errors. The compiler always (unless
1854 @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats
1855 for the standard library functions @code{printf}, @code{fprintf},
1856 @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime},
1857 @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such
1858 warnings are requested (using @option{-Wformat}), so there is no need to
1859 modify the header file @file{stdio.h}. In C99 mode, the functions
1860 @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and
1861 @code{vsscanf} are also checked. Except in strictly conforming C
1862 standard modes, the X/Open function @code{strfmon} is also checked as
1863 are @code{printf_unlocked} and @code{fprintf_unlocked}.
1864 @xref{C Dialect Options,,Options Controlling C Dialect}.
1866 The target may provide additional types of format checks.
1867 @xref{Target Format Checks,,Format Checks Specific to Particular
1870 @item format_arg (@var{string-index})
1871 @cindex @code{format_arg} function attribute
1872 @opindex Wformat-nonliteral
1873 The @code{format_arg} attribute specifies that a function takes a format
1874 string for a @code{printf}, @code{scanf}, @code{strftime} or
1875 @code{strfmon} style function and modifies it (for example, to translate
1876 it into another language), so the result can be passed to a
1877 @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style
1878 function (with the remaining arguments to the format function the same
1879 as they would have been for the unmodified string). For example, the
1884 my_dgettext (char *my_domain, const char *my_format)
1885 __attribute__ ((format_arg (2)));
1889 causes the compiler to check the arguments in calls to a @code{printf},
1890 @code{scanf}, @code{strftime} or @code{strfmon} type function, whose
1891 format string argument is a call to the @code{my_dgettext} function, for
1892 consistency with the format string argument @code{my_format}. If the
1893 @code{format_arg} attribute had not been specified, all the compiler
1894 could tell in such calls to format functions would be that the format
1895 string argument is not constant; this would generate a warning when
1896 @option{-Wformat-nonliteral} is used, but the calls could not be checked
1897 without the attribute.
1899 The parameter @var{string-index} specifies which argument is the format
1900 string argument (starting from one). Since non-static C++ methods have
1901 an implicit @code{this} argument, the arguments of such methods should
1902 be counted from two.
1904 The @code{format-arg} attribute allows you to identify your own
1905 functions which modify format strings, so that GCC can check the
1906 calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon}
1907 type function whose operands are a call to one of your own function.
1908 The compiler always treats @code{gettext}, @code{dgettext}, and
1909 @code{dcgettext} in this manner except when strict ISO C support is
1910 requested by @option{-ansi} or an appropriate @option{-std} option, or
1911 @option{-ffreestanding} or @option{-fno-builtin}
1912 is used. @xref{C Dialect Options,,Options
1913 Controlling C Dialect}.
1915 @item function_vector
1916 @cindex calling functions through the function vector on the H8/300 processors
1917 Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified
1918 function should be called through the function vector. Calling a
1919 function through the function vector will reduce code size, however;
1920 the function vector has a limited size (maximum 128 entries on the H8/300
1921 and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector.
1923 You must use GAS and GLD from GNU binutils version 2.7 or later for
1924 this attribute to work correctly.
1927 @cindex interrupt handler functions
1928 Use this attribute on the ARM, AVR, C4x, CRX, M32C, M32R/D, MS1, and Xstormy16
1929 ports to indicate that the specified function is an interrupt handler.
1930 The compiler will generate function entry and exit sequences suitable
1931 for use in an interrupt handler when this attribute is present.
1933 Note, interrupt handlers for the Blackfin, m68k, H8/300, H8/300H, H8S, and
1934 SH processors can be specified via the @code{interrupt_handler} attribute.
1936 Note, on the AVR, interrupts will be enabled inside the function.
1938 Note, for the ARM, you can specify the kind of interrupt to be handled by
1939 adding an optional parameter to the interrupt attribute like this:
1942 void f () __attribute__ ((interrupt ("IRQ")));
1945 Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@.
1947 @item interrupt_handler
1948 @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors
1949 Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to
1950 indicate that the specified function is an interrupt handler. The compiler
1951 will generate function entry and exit sequences suitable for use in an
1952 interrupt handler when this attribute is present.
1955 @cindex User stack pointer in interrupts on the Blackfin
1956 When used together with @code{interrupt_handler}, @code{exception_handler}
1957 or @code{nmi_handler}, code will be generated to load the stack pointer
1958 from the USP register in the function prologue.
1960 @item long_call/short_call
1961 @cindex indirect calls on ARM
1962 This attribute specifies how a particular function is called on
1963 ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options})
1964 command line switch and @code{#pragma long_calls} settings. The
1965 @code{long_call} attribute indicates that the function might be far
1966 away from the call site and require a different (more expensive)
1967 calling sequence. The @code{short_call} attribute always places
1968 the offset to the function from the call site into the @samp{BL}
1969 instruction directly.
1971 @item longcall/shortcall
1972 @cindex functions called via pointer on the RS/6000 and PowerPC
1973 On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute
1974 indicates that the function might be far away from the call site and
1975 require a different (more expensive) calling sequence. The
1976 @code{shortcall} attribute indicates that the function is always close
1977 enough for the shorter calling sequence to be used. These attributes
1978 override both the @option{-mlongcall} switch and, on the RS/6000 and
1979 PowerPC, the @code{#pragma longcall} setting.
1981 @xref{RS/6000 and PowerPC Options}, for more information on whether long
1982 calls are necessary.
1985 @cindex indirect calls on MIPS
1986 This attribute specifies how a particular function is called on MIPS@.
1987 The attribute overrides the @option{-mlong-calls} (@pxref{MIPS Options})
1988 command line switch. This attribute causes the compiler to always call
1989 the function by first loading its address into a register, and then using
1990 the contents of that register.
1993 @cindex @code{malloc} attribute
1994 The @code{malloc} attribute is used to tell the compiler that a function
1995 may be treated as if any non-@code{NULL} pointer it returns cannot
1996 alias any other pointer valid when the function returns.
1997 This will often improve optimization.
1998 Standard functions with this property include @code{malloc} and
1999 @code{calloc}. @code{realloc}-like functions have this property as
2000 long as the old pointer is never referred to (including comparing it
2001 to the new pointer) after the function returns a non-@code{NULL}
2004 @item model (@var{model-name})
2005 @cindex function addressability on the M32R/D
2006 @cindex variable addressability on the IA-64
2008 On the M32R/D, use this attribute to set the addressability of an
2009 object, and of the code generated for a function. The identifier
2010 @var{model-name} is one of @code{small}, @code{medium}, or
2011 @code{large}, representing each of the code models.
2013 Small model objects live in the lower 16MB of memory (so that their
2014 addresses can be loaded with the @code{ld24} instruction), and are
2015 callable with the @code{bl} instruction.
2017 Medium model objects may live anywhere in the 32-bit address space (the
2018 compiler will generate @code{seth/add3} instructions to load their addresses),
2019 and are callable with the @code{bl} instruction.
2021 Large model objects may live anywhere in the 32-bit address space (the
2022 compiler will generate @code{seth/add3} instructions to load their addresses),
2023 and may not be reachable with the @code{bl} instruction (the compiler will
2024 generate the much slower @code{seth/add3/jl} instruction sequence).
2026 On IA-64, use this attribute to set the addressability of an object.
2027 At present, the only supported identifier for @var{model-name} is
2028 @code{small}, indicating addressability via ``small'' (22-bit)
2029 addresses (so that their addresses can be loaded with the @code{addl}
2030 instruction). Caveat: such addressing is by definition not position
2031 independent and hence this attribute must not be used for objects
2032 defined by shared libraries.
2035 @cindex function without a prologue/epilogue code
2036 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
2037 specified function does not need prologue/epilogue sequences generated by
2038 the compiler. It is up to the programmer to provide these sequences.
2041 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2042 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2043 use the normal calling convention based on @code{jsr} and @code{rts}.
2044 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2048 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2049 Use this attribute together with @code{interrupt_handler},
2050 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2051 entry code should enable nested interrupts or exceptions.
2054 @cindex NMI handler functions on the Blackfin processor
2055 Use this attribute on the Blackfin to indicate that the specified function
2056 is an NMI handler. The compiler will generate function entry and
2057 exit sequences suitable for use in an NMI handler when this
2058 attribute is present.
2060 @item no_instrument_function
2061 @cindex @code{no_instrument_function} function attribute
2062 @opindex finstrument-functions
2063 If @option{-finstrument-functions} is given, profiling function calls will
2064 be generated at entry and exit of most user-compiled functions.
2065 Functions with this attribute will not be so instrumented.
2068 @cindex @code{noinline} function attribute
2069 This function attribute prevents a function from being considered for
2072 @item nonnull (@var{arg-index}, @dots{})
2073 @cindex @code{nonnull} function attribute
2074 The @code{nonnull} attribute specifies that some function parameters should
2075 be non-null pointers. For instance, the declaration:
2079 my_memcpy (void *dest, const void *src, size_t len)
2080 __attribute__((nonnull (1, 2)));
2084 causes the compiler to check that, in calls to @code{my_memcpy},
2085 arguments @var{dest} and @var{src} are non-null. If the compiler
2086 determines that a null pointer is passed in an argument slot marked
2087 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2088 is issued. The compiler may also choose to make optimizations based
2089 on the knowledge that certain function arguments will not be null.
2091 If no argument index list is given to the @code{nonnull} attribute,
2092 all pointer arguments are marked as non-null. To illustrate, the
2093 following declaration is equivalent to the previous example:
2097 my_memcpy (void *dest, const void *src, size_t len)
2098 __attribute__((nonnull));
2102 @cindex @code{noreturn} function attribute
2103 A few standard library functions, such as @code{abort} and @code{exit},
2104 cannot return. GCC knows this automatically. Some programs define
2105 their own functions that never return. You can declare them
2106 @code{noreturn} to tell the compiler this fact. For example,
2110 void fatal () __attribute__ ((noreturn));
2113 fatal (/* @r{@dots{}} */)
2115 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2121 The @code{noreturn} keyword tells the compiler to assume that
2122 @code{fatal} cannot return. It can then optimize without regard to what
2123 would happen if @code{fatal} ever did return. This makes slightly
2124 better code. More importantly, it helps avoid spurious warnings of
2125 uninitialized variables.
2127 The @code{noreturn} keyword does not affect the exceptional path when that
2128 applies: a @code{noreturn}-marked function may still return to the caller
2129 by throwing an exception or calling @code{longjmp}.
2131 Do not assume that registers saved by the calling function are
2132 restored before calling the @code{noreturn} function.
2134 It does not make sense for a @code{noreturn} function to have a return
2135 type other than @code{void}.
2137 The attribute @code{noreturn} is not implemented in GCC versions
2138 earlier than 2.5. An alternative way to declare that a function does
2139 not return, which works in the current version and in some older
2140 versions, is as follows:
2143 typedef void voidfn ();
2145 volatile voidfn fatal;
2148 This approach does not work in GNU C++.
2151 @cindex @code{nothrow} function attribute
2152 The @code{nothrow} attribute is used to inform the compiler that a
2153 function cannot throw an exception. For example, most functions in
2154 the standard C library can be guaranteed not to throw an exception
2155 with the notable exceptions of @code{qsort} and @code{bsearch} that
2156 take function pointer arguments. The @code{nothrow} attribute is not
2157 implemented in GCC versions earlier than 3.3.
2160 @cindex @code{pure} function attribute
2161 Many functions have no effects except the return value and their
2162 return value depends only on the parameters and/or global variables.
2163 Such a function can be subject
2164 to common subexpression elimination and loop optimization just as an
2165 arithmetic operator would be. These functions should be declared
2166 with the attribute @code{pure}. For example,
2169 int square (int) __attribute__ ((pure));
2173 says that the hypothetical function @code{square} is safe to call
2174 fewer times than the program says.
2176 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2177 Interesting non-pure functions are functions with infinite loops or those
2178 depending on volatile memory or other system resource, that may change between
2179 two consecutive calls (such as @code{feof} in a multithreading environment).
2181 The attribute @code{pure} is not implemented in GCC versions earlier
2184 @item regparm (@var{number})
2185 @cindex @code{regparm} attribute
2186 @cindex functions that are passed arguments in registers on the 386
2187 On the Intel 386, the @code{regparm} attribute causes the compiler to
2188 pass arguments number one to @var{number} if they are of integral type
2189 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2190 take a variable number of arguments will continue to be passed all of their
2191 arguments on the stack.
2193 Beware that on some ELF systems this attribute is unsuitable for
2194 global functions in shared libraries with lazy binding (which is the
2195 default). Lazy binding will send the first call via resolving code in
2196 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2197 per the standard calling conventions. Solaris 8 is affected by this.
2198 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2199 safe since the loaders there save all registers. (Lazy binding can be
2200 disabled with the linker or the loader if desired, to avoid the
2204 @cindex @code{sseregparm} attribute
2205 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2206 causes the compiler to pass up to 8 floating point arguments in
2207 SSE registers instead of on the stack. Functions that take a
2208 variable number of arguments will continue to pass all of their
2209 floating point arguments on the stack.
2211 @item force_align_arg_pointer
2212 @cindex @code{force_align_arg_pointer} attribute
2213 On the Intel x86, the @code{force_align_arg_pointer} attribute may be
2214 applied to individual function definitions, generating an alternate
2215 prologue and epilogue that realigns the runtime stack. This supports
2216 mixing legacy codes that run with a 4-byte aligned stack with modern
2217 codes that keep a 16-byte stack for SSE compatibility. The alternate
2218 prologue and epilogue are slower and bigger than the regular ones, and
2219 the alternate prologue requires a scratch register; this lowers the
2220 number of registers available if used in conjunction with the
2221 @code{regparm} attribute. The @code{force_align_arg_pointer}
2222 attribute is incompatible with nested functions; this is considered a
2226 @cindex @code{returns_twice} attribute
2227 The @code{returns_twice} attribute tells the compiler that a function may
2228 return more than one time. The compiler will ensure that all registers
2229 are dead before calling such a function and will emit a warning about
2230 the variables that may be clobbered after the second return from the
2231 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2232 The @code{longjmp}-like counterpart of such function, if any, might need
2233 to be marked with the @code{noreturn} attribute.
2236 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2237 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2238 all registers except the stack pointer should be saved in the prologue
2239 regardless of whether they are used or not.
2241 @item section ("@var{section-name}")
2242 @cindex @code{section} function attribute
2243 Normally, the compiler places the code it generates in the @code{text} section.
2244 Sometimes, however, you need additional sections, or you need certain
2245 particular functions to appear in special sections. The @code{section}
2246 attribute specifies that a function lives in a particular section.
2247 For example, the declaration:
2250 extern void foobar (void) __attribute__ ((section ("bar")));
2254 puts the function @code{foobar} in the @code{bar} section.
2256 Some file formats do not support arbitrary sections so the @code{section}
2257 attribute is not available on all platforms.
2258 If you need to map the entire contents of a module to a particular
2259 section, consider using the facilities of the linker instead.
2262 @cindex @code{sentinel} function attribute
2263 This function attribute ensures that a parameter in a function call is
2264 an explicit @code{NULL}. The attribute is only valid on variadic
2265 functions. By default, the sentinel is located at position zero, the
2266 last parameter of the function call. If an optional integer position
2267 argument P is supplied to the attribute, the sentinel must be located at
2268 position P counting backwards from the end of the argument list.
2271 __attribute__ ((sentinel))
2273 __attribute__ ((sentinel(0)))
2276 The attribute is automatically set with a position of 0 for the built-in
2277 functions @code{execl} and @code{execlp}. The built-in function
2278 @code{execle} has the attribute set with a position of 1.
2280 A valid @code{NULL} in this context is defined as zero with any pointer
2281 type. If your system defines the @code{NULL} macro with an integer type
2282 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2283 with a copy that redefines NULL appropriately.
2285 The warnings for missing or incorrect sentinels are enabled with
2289 See long_call/short_call.
2292 See longcall/shortcall.
2295 @cindex signal handler functions on the AVR processors
2296 Use this attribute on the AVR to indicate that the specified
2297 function is a signal handler. The compiler will generate function
2298 entry and exit sequences suitable for use in a signal handler when this
2299 attribute is present. Interrupts will be disabled inside the function.
2302 Use this attribute on the SH to indicate an @code{interrupt_handler}
2303 function should switch to an alternate stack. It expects a string
2304 argument that names a global variable holding the address of the
2309 void f () __attribute__ ((interrupt_handler,
2310 sp_switch ("alt_stack")));
2314 @cindex functions that pop the argument stack on the 386
2315 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2316 assume that the called function will pop off the stack space used to
2317 pass arguments, unless it takes a variable number of arguments.
2320 @cindex tiny data section on the H8/300H and H8S
2321 Use this attribute on the H8/300H and H8S to indicate that the specified
2322 variable should be placed into the tiny data section.
2323 The compiler will generate more efficient code for loads and stores
2324 on data in the tiny data section. Note the tiny data area is limited to
2325 slightly under 32kbytes of data.
2328 Use this attribute on the SH for an @code{interrupt_handler} to return using
2329 @code{trapa} instead of @code{rte}. This attribute expects an integer
2330 argument specifying the trap number to be used.
2333 @cindex @code{unused} attribute.
2334 This attribute, attached to a function, means that the function is meant
2335 to be possibly unused. GCC will not produce a warning for this
2339 @cindex @code{used} attribute.
2340 This attribute, attached to a function, means that code must be emitted
2341 for the function even if it appears that the function is not referenced.
2342 This is useful, for example, when the function is referenced only in
2345 @item visibility ("@var{visibility_type}")
2346 @cindex @code{visibility} attribute
2347 This attribute affects the linkage of the declaration to which it is attached.
2348 There are four supported @var{visibility_type} values: default,
2349 hidden, protected or internal visibility.
2352 void __attribute__ ((visibility ("protected")))
2353 f () @{ /* @r{Do something.} */; @}
2354 int i __attribute__ ((visibility ("hidden")));
2357 The possible values of @var{visibility_type} correspond to the
2358 visibility settings in the ELF gABI.
2361 @c keep this list of visibilities in alphabetical order.
2364 Default visibility is the normal case for the object file format.
2365 This value is available for the visibility attribute to override other
2366 options that may change the assumed visibility of entities.
2368 On ELF, default visibility means that the declaration is visible to other
2369 modules and, in shared libraries, means that the declared entity may be
2372 On Darwin, default visibility means that the declaration is visible to
2375 Default visibility corresponds to ``external linkage'' in the language.
2378 Hidden visibility indicates that the entity declared will have a new
2379 form of linkage, which we'll call ``hidden linkage''. Two
2380 declarations of an object with hidden linkage refer to the same object
2381 if they are in the same shared object.
2384 Internal visibility is like hidden visibility, but with additional
2385 processor specific semantics. Unless otherwise specified by the
2386 psABI, GCC defines internal visibility to mean that a function is
2387 @emph{never} called from another module. Compare this with hidden
2388 functions which, while they cannot be referenced directly by other
2389 modules, can be referenced indirectly via function pointers. By
2390 indicating that a function cannot be called from outside the module,
2391 GCC may for instance omit the load of a PIC register since it is known
2392 that the calling function loaded the correct value.
2395 Protected visibility is like default visibility except that it
2396 indicates that references within the defining module will bind to the
2397 definition in that module. That is, the declared entity cannot be
2398 overridden by another module.
2402 All visibilities are supported on many, but not all, ELF targets
2403 (supported when the assembler supports the @samp{.visibility}
2404 pseudo-op). Default visibility is supported everywhere. Hidden
2405 visibility is supported on Darwin targets.
2407 The visibility attribute should be applied only to declarations which
2408 would otherwise have external linkage. The attribute should be applied
2409 consistently, so that the same entity should not be declared with
2410 different settings of the attribute.
2412 In C++, the visibility attribute applies to types as well as functions
2413 and objects, because in C++ types have linkage. A class must not have
2414 greater visibility than its non-static data member types and bases,
2415 and class members default to the visibility of their class. Also, a
2416 declaration must not have greater visibility than its type.
2418 In C++, you can mark member functions and static member variables of a
2419 class with the visibility attribute. This is useful if if you know a
2420 particular method or static member variable should only be used from
2421 one shared object; then you can mark it hidden while the rest of the
2422 class has default visibility. Care must be taken to avoid breaking
2423 the One Definition Rule; for example, it is not useful to mark a
2424 method which is defined inside a class definition as hidden without
2425 marking the whole class as hidden.
2427 A C++ namespace declaration can also have the visibility attribute.
2428 This attribute applies only to the particular namespace body, not to
2429 other definitions of the same namespace; it is equivalent to using
2430 @samp{#pragma GCC visibility} before and after the namespace
2431 definition (@pxref{Visibility Pragmas}).
2433 In C++, if a template argument has limited visibility, this
2434 restriction is implicitly propagated to the template instantiation.
2435 Otherwise, template instantiations and specializations default to the
2436 visibility of their template.
2438 @item warn_unused_result
2439 @cindex @code{warn_unused_result} attribute
2440 The @code{warn_unused_result} attribute causes a warning to be emitted
2441 if a caller of the function with this attribute does not use its
2442 return value. This is useful for functions where not checking
2443 the result is either a security problem or always a bug, such as
2447 int fn () __attribute__ ((warn_unused_result));
2450 if (fn () < 0) return -1;
2456 results in warning on line 5.
2459 @cindex @code{weak} attribute
2460 The @code{weak} attribute causes the declaration to be emitted as a weak
2461 symbol rather than a global. This is primarily useful in defining
2462 library functions which can be overridden in user code, though it can
2463 also be used with non-function declarations. Weak symbols are supported
2464 for ELF targets, and also for a.out targets when using the GNU assembler
2468 @itemx weakref ("@var{target}")
2469 @cindex @code{weakref} attribute
2470 The @code{weakref} attribute marks a declaration as a weak reference.
2471 Without arguments, it should be accompanied by an @code{alias} attribute
2472 naming the target symbol. Optionally, the @var{target} may be given as
2473 an argument to @code{weakref} itself. In either case, @code{weakref}
2474 implicitly marks the declaration as @code{weak}. Without a
2475 @var{target}, given as an argument to @code{weakref} or to @code{alias},
2476 @code{weakref} is equivalent to @code{weak}.
2479 static int x() __attribute__ ((weakref ("y")));
2480 /* is equivalent to... */
2481 static int x() __attribute__ ((weak, weakref, alias ("y")));
2483 static int x() __attribute__ ((weakref));
2484 static int x() __attribute__ ((alias ("y")));
2487 A weak reference is an alias that does not by itself require a
2488 definition to be given for the target symbol. If the target symbol is
2489 only referenced through weak references, then the becomes a @code{weak}
2490 undefined symbol. If it is directly referenced, however, then such
2491 strong references prevail, and a definition will be required for the
2492 symbol, not necessarily in the same translation unit.
2494 The effect is equivalent to moving all references to the alias to a
2495 separate translation unit, renaming the alias to the aliased symbol,
2496 declaring it as weak, compiling the two separate translation units and
2497 performing a reloadable link on them.
2499 At present, a declaration to which @code{weakref} is attached can
2500 only be @code{static}.
2502 @item externally_visible
2503 @cindex @code{externally_visible} attribute.
2504 This attribute, attached to a global variable or function nullify
2505 effect of @option{-fwhole-program} command line option, so the object
2506 remain visible outside the current compilation unit
2510 You can specify multiple attributes in a declaration by separating them
2511 by commas within the double parentheses or by immediately following an
2512 attribute declaration with another attribute declaration.
2514 @cindex @code{#pragma}, reason for not using
2515 @cindex pragma, reason for not using
2516 Some people object to the @code{__attribute__} feature, suggesting that
2517 ISO C's @code{#pragma} should be used instead. At the time
2518 @code{__attribute__} was designed, there were two reasons for not doing
2523 It is impossible to generate @code{#pragma} commands from a macro.
2526 There is no telling what the same @code{#pragma} might mean in another
2530 These two reasons applied to almost any application that might have been
2531 proposed for @code{#pragma}. It was basically a mistake to use
2532 @code{#pragma} for @emph{anything}.
2534 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2535 to be generated from macros. In addition, a @code{#pragma GCC}
2536 namespace is now in use for GCC-specific pragmas. However, it has been
2537 found convenient to use @code{__attribute__} to achieve a natural
2538 attachment of attributes to their corresponding declarations, whereas
2539 @code{#pragma GCC} is of use for constructs that do not naturally form
2540 part of the grammar. @xref{Other Directives,,Miscellaneous
2541 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2543 @node Attribute Syntax
2544 @section Attribute Syntax
2545 @cindex attribute syntax
2547 This section describes the syntax with which @code{__attribute__} may be
2548 used, and the constructs to which attribute specifiers bind, for the C
2549 language. Some details may vary for C++ and Objective-C@. Because of
2550 infelicities in the grammar for attributes, some forms described here
2551 may not be successfully parsed in all cases.
2553 There are some problems with the semantics of attributes in C++. For
2554 example, there are no manglings for attributes, although they may affect
2555 code generation, so problems may arise when attributed types are used in
2556 conjunction with templates or overloading. Similarly, @code{typeid}
2557 does not distinguish between types with different attributes. Support
2558 for attributes in C++ may be restricted in future to attributes on
2559 declarations only, but not on nested declarators.
2561 @xref{Function Attributes}, for details of the semantics of attributes
2562 applying to functions. @xref{Variable Attributes}, for details of the
2563 semantics of attributes applying to variables. @xref{Type Attributes},
2564 for details of the semantics of attributes applying to structure, union
2565 and enumerated types.
2567 An @dfn{attribute specifier} is of the form
2568 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2569 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2570 each attribute is one of the following:
2574 Empty. Empty attributes are ignored.
2577 A word (which may be an identifier such as @code{unused}, or a reserved
2578 word such as @code{const}).
2581 A word, followed by, in parentheses, parameters for the attribute.
2582 These parameters take one of the following forms:
2586 An identifier. For example, @code{mode} attributes use this form.
2589 An identifier followed by a comma and a non-empty comma-separated list
2590 of expressions. For example, @code{format} attributes use this form.
2593 A possibly empty comma-separated list of expressions. For example,
2594 @code{format_arg} attributes use this form with the list being a single
2595 integer constant expression, and @code{alias} attributes use this form
2596 with the list being a single string constant.
2600 An @dfn{attribute specifier list} is a sequence of one or more attribute
2601 specifiers, not separated by any other tokens.
2603 In GNU C, an attribute specifier list may appear after the colon following a
2604 label, other than a @code{case} or @code{default} label. The only
2605 attribute it makes sense to use after a label is @code{unused}. This
2606 feature is intended for code generated by programs which contains labels
2607 that may be unused but which is compiled with @option{-Wall}. It would
2608 not normally be appropriate to use in it human-written code, though it
2609 could be useful in cases where the code that jumps to the label is
2610 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2611 such placement of attribute lists, as it is permissible for a
2612 declaration, which could begin with an attribute list, to be labelled in
2613 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2614 does not arise there.
2616 An attribute specifier list may appear as part of a @code{struct},
2617 @code{union} or @code{enum} specifier. It may go either immediately
2618 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2619 the closing brace. The former syntax is preferred.
2620 Where attribute specifiers follow the closing brace, they are considered
2621 to relate to the structure, union or enumerated type defined, not to any
2622 enclosing declaration the type specifier appears in, and the type
2623 defined is not complete until after the attribute specifiers.
2624 @c Otherwise, there would be the following problems: a shift/reduce
2625 @c conflict between attributes binding the struct/union/enum and
2626 @c binding to the list of specifiers/qualifiers; and "aligned"
2627 @c attributes could use sizeof for the structure, but the size could be
2628 @c changed later by "packed" attributes.
2630 Otherwise, an attribute specifier appears as part of a declaration,
2631 counting declarations of unnamed parameters and type names, and relates
2632 to that declaration (which may be nested in another declaration, for
2633 example in the case of a parameter declaration), or to a particular declarator
2634 within a declaration. Where an
2635 attribute specifier is applied to a parameter declared as a function or
2636 an array, it should apply to the function or array rather than the
2637 pointer to which the parameter is implicitly converted, but this is not
2638 yet correctly implemented.
2640 Any list of specifiers and qualifiers at the start of a declaration may
2641 contain attribute specifiers, whether or not such a list may in that
2642 context contain storage class specifiers. (Some attributes, however,
2643 are essentially in the nature of storage class specifiers, and only make
2644 sense where storage class specifiers may be used; for example,
2645 @code{section}.) There is one necessary limitation to this syntax: the
2646 first old-style parameter declaration in a function definition cannot
2647 begin with an attribute specifier, because such an attribute applies to
2648 the function instead by syntax described below (which, however, is not
2649 yet implemented in this case). In some other cases, attribute
2650 specifiers are permitted by this grammar but not yet supported by the
2651 compiler. All attribute specifiers in this place relate to the
2652 declaration as a whole. In the obsolescent usage where a type of
2653 @code{int} is implied by the absence of type specifiers, such a list of
2654 specifiers and qualifiers may be an attribute specifier list with no
2655 other specifiers or qualifiers.
2657 At present, the first parameter in a function prototype must have some
2658 type specifier which is not an attribute specifier; this resolves an
2659 ambiguity in the interpretation of @code{void f(int
2660 (__attribute__((foo)) x))}, but is subject to change. At present, if
2661 the parentheses of a function declarator contain only attributes then
2662 those attributes are ignored, rather than yielding an error or warning
2663 or implying a single parameter of type int, but this is subject to
2666 An attribute specifier list may appear immediately before a declarator
2667 (other than the first) in a comma-separated list of declarators in a
2668 declaration of more than one identifier using a single list of
2669 specifiers and qualifiers. Such attribute specifiers apply
2670 only to the identifier before whose declarator they appear. For
2674 __attribute__((noreturn)) void d0 (void),
2675 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2680 the @code{noreturn} attribute applies to all the functions
2681 declared; the @code{format} attribute only applies to @code{d1}.
2683 An attribute specifier list may appear immediately before the comma,
2684 @code{=} or semicolon terminating the declaration of an identifier other
2685 than a function definition. At present, such attribute specifiers apply
2686 to the declared object or function, but in future they may attach to the
2687 outermost adjacent declarator. In simple cases there is no difference,
2688 but, for example, in
2691 void (****f)(void) __attribute__((noreturn));
2695 at present the @code{noreturn} attribute applies to @code{f}, which
2696 causes a warning since @code{f} is not a function, but in future it may
2697 apply to the function @code{****f}. The precise semantics of what
2698 attributes in such cases will apply to are not yet specified. Where an
2699 assembler name for an object or function is specified (@pxref{Asm
2700 Labels}), at present the attribute must follow the @code{asm}
2701 specification; in future, attributes before the @code{asm} specification
2702 may apply to the adjacent declarator, and those after it to the declared
2705 An attribute specifier list may, in future, be permitted to appear after
2706 the declarator in a function definition (before any old-style parameter
2707 declarations or the function body).
2709 Attribute specifiers may be mixed with type qualifiers appearing inside
2710 the @code{[]} of a parameter array declarator, in the C99 construct by
2711 which such qualifiers are applied to the pointer to which the array is
2712 implicitly converted. Such attribute specifiers apply to the pointer,
2713 not to the array, but at present this is not implemented and they are
2716 An attribute specifier list may appear at the start of a nested
2717 declarator. At present, there are some limitations in this usage: the
2718 attributes correctly apply to the declarator, but for most individual
2719 attributes the semantics this implies are not implemented.
2720 When attribute specifiers follow the @code{*} of a pointer
2721 declarator, they may be mixed with any type qualifiers present.
2722 The following describes the formal semantics of this syntax. It will make the
2723 most sense if you are familiar with the formal specification of
2724 declarators in the ISO C standard.
2726 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2727 D1}, where @code{T} contains declaration specifiers that specify a type
2728 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2729 contains an identifier @var{ident}. The type specified for @var{ident}
2730 for derived declarators whose type does not include an attribute
2731 specifier is as in the ISO C standard.
2733 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2734 and the declaration @code{T D} specifies the type
2735 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2736 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2737 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2739 If @code{D1} has the form @code{*
2740 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2741 declaration @code{T D} specifies the type
2742 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2743 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2744 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2750 void (__attribute__((noreturn)) ****f) (void);
2754 specifies the type ``pointer to pointer to pointer to pointer to
2755 non-returning function returning @code{void}''. As another example,
2758 char *__attribute__((aligned(8))) *f;
2762 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2763 Note again that this does not work with most attributes; for example,
2764 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2765 is not yet supported.
2767 For compatibility with existing code written for compiler versions that
2768 did not implement attributes on nested declarators, some laxity is
2769 allowed in the placing of attributes. If an attribute that only applies
2770 to types is applied to a declaration, it will be treated as applying to
2771 the type of that declaration. If an attribute that only applies to
2772 declarations is applied to the type of a declaration, it will be treated
2773 as applying to that declaration; and, for compatibility with code
2774 placing the attributes immediately before the identifier declared, such
2775 an attribute applied to a function return type will be treated as
2776 applying to the function type, and such an attribute applied to an array
2777 element type will be treated as applying to the array type. If an
2778 attribute that only applies to function types is applied to a
2779 pointer-to-function type, it will be treated as applying to the pointer
2780 target type; if such an attribute is applied to a function return type
2781 that is not a pointer-to-function type, it will be treated as applying
2782 to the function type.
2784 @node Function Prototypes
2785 @section Prototypes and Old-Style Function Definitions
2786 @cindex function prototype declarations
2787 @cindex old-style function definitions
2788 @cindex promotion of formal parameters
2790 GNU C extends ISO C to allow a function prototype to override a later
2791 old-style non-prototype definition. Consider the following example:
2794 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2801 /* @r{Prototype function declaration.} */
2802 int isroot P((uid_t));
2804 /* @r{Old-style function definition.} */
2806 isroot (x) /* @r{??? lossage here ???} */
2813 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2814 not allow this example, because subword arguments in old-style
2815 non-prototype definitions are promoted. Therefore in this example the
2816 function definition's argument is really an @code{int}, which does not
2817 match the prototype argument type of @code{short}.
2819 This restriction of ISO C makes it hard to write code that is portable
2820 to traditional C compilers, because the programmer does not know
2821 whether the @code{uid_t} type is @code{short}, @code{int}, or
2822 @code{long}. Therefore, in cases like these GNU C allows a prototype
2823 to override a later old-style definition. More precisely, in GNU C, a
2824 function prototype argument type overrides the argument type specified
2825 by a later old-style definition if the former type is the same as the
2826 latter type before promotion. Thus in GNU C the above example is
2827 equivalent to the following:
2840 GNU C++ does not support old-style function definitions, so this
2841 extension is irrelevant.
2844 @section C++ Style Comments
2846 @cindex C++ comments
2847 @cindex comments, C++ style
2849 In GNU C, you may use C++ style comments, which start with @samp{//} and
2850 continue until the end of the line. Many other C implementations allow
2851 such comments, and they are included in the 1999 C standard. However,
2852 C++ style comments are not recognized if you specify an @option{-std}
2853 option specifying a version of ISO C before C99, or @option{-ansi}
2854 (equivalent to @option{-std=c89}).
2857 @section Dollar Signs in Identifier Names
2859 @cindex dollar signs in identifier names
2860 @cindex identifier names, dollar signs in
2862 In GNU C, you may normally use dollar signs in identifier names.
2863 This is because many traditional C implementations allow such identifiers.
2864 However, dollar signs in identifiers are not supported on a few target
2865 machines, typically because the target assembler does not allow them.
2867 @node Character Escapes
2868 @section The Character @key{ESC} in Constants
2870 You can use the sequence @samp{\e} in a string or character constant to
2871 stand for the ASCII character @key{ESC}.
2874 @section Inquiring on Alignment of Types or Variables
2876 @cindex type alignment
2877 @cindex variable alignment
2879 The keyword @code{__alignof__} allows you to inquire about how an object
2880 is aligned, or the minimum alignment usually required by a type. Its
2881 syntax is just like @code{sizeof}.
2883 For example, if the target machine requires a @code{double} value to be
2884 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2885 This is true on many RISC machines. On more traditional machine
2886 designs, @code{__alignof__ (double)} is 4 or even 2.
2888 Some machines never actually require alignment; they allow reference to any
2889 data type even at an odd address. For these machines, @code{__alignof__}
2890 reports the @emph{recommended} alignment of a type.
2892 If the operand of @code{__alignof__} is an lvalue rather than a type,
2893 its value is the required alignment for its type, taking into account
2894 any minimum alignment specified with GCC's @code{__attribute__}
2895 extension (@pxref{Variable Attributes}). For example, after this
2899 struct foo @{ int x; char y; @} foo1;
2903 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2904 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2906 It is an error to ask for the alignment of an incomplete type.
2908 @node Variable Attributes
2909 @section Specifying Attributes of Variables
2910 @cindex attribute of variables
2911 @cindex variable attributes
2913 The keyword @code{__attribute__} allows you to specify special
2914 attributes of variables or structure fields. This keyword is followed
2915 by an attribute specification inside double parentheses. Some
2916 attributes are currently defined generically for variables.
2917 Other attributes are defined for variables on particular target
2918 systems. Other attributes are available for functions
2919 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
2920 Other front ends might define more attributes
2921 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2923 You may also specify attributes with @samp{__} preceding and following
2924 each keyword. This allows you to use them in header files without
2925 being concerned about a possible macro of the same name. For example,
2926 you may use @code{__aligned__} instead of @code{aligned}.
2928 @xref{Attribute Syntax}, for details of the exact syntax for using
2932 @cindex @code{aligned} attribute
2933 @item aligned (@var{alignment})
2934 This attribute specifies a minimum alignment for the variable or
2935 structure field, measured in bytes. For example, the declaration:
2938 int x __attribute__ ((aligned (16))) = 0;
2942 causes the compiler to allocate the global variable @code{x} on a
2943 16-byte boundary. On a 68040, this could be used in conjunction with
2944 an @code{asm} expression to access the @code{move16} instruction which
2945 requires 16-byte aligned operands.
2947 You can also specify the alignment of structure fields. For example, to
2948 create a double-word aligned @code{int} pair, you could write:
2951 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2955 This is an alternative to creating a union with a @code{double} member
2956 that forces the union to be double-word aligned.
2958 As in the preceding examples, you can explicitly specify the alignment
2959 (in bytes) that you wish the compiler to use for a given variable or
2960 structure field. Alternatively, you can leave out the alignment factor
2961 and just ask the compiler to align a variable or field to the maximum
2962 useful alignment for the target machine you are compiling for. For
2963 example, you could write:
2966 short array[3] __attribute__ ((aligned));
2969 Whenever you leave out the alignment factor in an @code{aligned} attribute
2970 specification, the compiler automatically sets the alignment for the declared
2971 variable or field to the largest alignment which is ever used for any data
2972 type on the target machine you are compiling for. Doing this can often make
2973 copy operations more efficient, because the compiler can use whatever
2974 instructions copy the biggest chunks of memory when performing copies to
2975 or from the variables or fields that you have aligned this way.
2977 The @code{aligned} attribute can only increase the alignment; but you
2978 can decrease it by specifying @code{packed} as well. See below.
2980 Note that the effectiveness of @code{aligned} attributes may be limited
2981 by inherent limitations in your linker. On many systems, the linker is
2982 only able to arrange for variables to be aligned up to a certain maximum
2983 alignment. (For some linkers, the maximum supported alignment may
2984 be very very small.) If your linker is only able to align variables
2985 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
2986 in an @code{__attribute__} will still only provide you with 8 byte
2987 alignment. See your linker documentation for further information.
2989 @item cleanup (@var{cleanup_function})
2990 @cindex @code{cleanup} attribute
2991 The @code{cleanup} attribute runs a function when the variable goes
2992 out of scope. This attribute can only be applied to auto function
2993 scope variables; it may not be applied to parameters or variables
2994 with static storage duration. The function must take one parameter,
2995 a pointer to a type compatible with the variable. The return value
2996 of the function (if any) is ignored.
2998 If @option{-fexceptions} is enabled, then @var{cleanup_function}
2999 will be run during the stack unwinding that happens during the
3000 processing of the exception. Note that the @code{cleanup} attribute
3001 does not allow the exception to be caught, only to perform an action.
3002 It is undefined what happens if @var{cleanup_function} does not
3007 @cindex @code{common} attribute
3008 @cindex @code{nocommon} attribute
3011 The @code{common} attribute requests GCC to place a variable in
3012 ``common'' storage. The @code{nocommon} attribute requests the
3013 opposite---to allocate space for it directly.
3015 These attributes override the default chosen by the
3016 @option{-fno-common} and @option{-fcommon} flags respectively.
3019 @cindex @code{deprecated} attribute
3020 The @code{deprecated} attribute results in a warning if the variable
3021 is used anywhere in the source file. This is useful when identifying
3022 variables that are expected to be removed in a future version of a
3023 program. The warning also includes the location of the declaration
3024 of the deprecated variable, to enable users to easily find further
3025 information about why the variable is deprecated, or what they should
3026 do instead. Note that the warning only occurs for uses:
3029 extern int old_var __attribute__ ((deprecated));
3031 int new_fn () @{ return old_var; @}
3034 results in a warning on line 3 but not line 2.
3036 The @code{deprecated} attribute can also be used for functions and
3037 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3039 @item mode (@var{mode})
3040 @cindex @code{mode} attribute
3041 This attribute specifies the data type for the declaration---whichever
3042 type corresponds to the mode @var{mode}. This in effect lets you
3043 request an integer or floating point type according to its width.
3045 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3046 indicate the mode corresponding to a one-byte integer, @samp{word} or
3047 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3048 or @samp{__pointer__} for the mode used to represent pointers.
3051 @cindex @code{packed} attribute
3052 The @code{packed} attribute specifies that a variable or structure field
3053 should have the smallest possible alignment---one byte for a variable,
3054 and one bit for a field, unless you specify a larger value with the
3055 @code{aligned} attribute.
3057 Here is a structure in which the field @code{x} is packed, so that it
3058 immediately follows @code{a}:
3064 int x[2] __attribute__ ((packed));
3068 @item section ("@var{section-name}")
3069 @cindex @code{section} variable attribute
3070 Normally, the compiler places the objects it generates in sections like
3071 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3072 or you need certain particular variables to appear in special sections,
3073 for example to map to special hardware. The @code{section}
3074 attribute specifies that a variable (or function) lives in a particular
3075 section. For example, this small program uses several specific section names:
3078 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3079 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3080 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3081 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3085 /* @r{Initialize stack pointer} */
3086 init_sp (stack + sizeof (stack));
3088 /* @r{Initialize initialized data} */
3089 memcpy (&init_data, &data, &edata - &data);
3091 /* @r{Turn on the serial ports} */
3098 Use the @code{section} attribute with an @emph{initialized} definition
3099 of a @emph{global} variable, as shown in the example. GCC issues
3100 a warning and otherwise ignores the @code{section} attribute in
3101 uninitialized variable declarations.
3103 You may only use the @code{section} attribute with a fully initialized
3104 global definition because of the way linkers work. The linker requires
3105 each object be defined once, with the exception that uninitialized
3106 variables tentatively go in the @code{common} (or @code{bss}) section
3107 and can be multiply ``defined''. You can force a variable to be
3108 initialized with the @option{-fno-common} flag or the @code{nocommon}
3111 Some file formats do not support arbitrary sections so the @code{section}
3112 attribute is not available on all platforms.
3113 If you need to map the entire contents of a module to a particular
3114 section, consider using the facilities of the linker instead.
3117 @cindex @code{shared} variable attribute
3118 On Microsoft Windows, in addition to putting variable definitions in a named
3119 section, the section can also be shared among all running copies of an
3120 executable or DLL@. For example, this small program defines shared data
3121 by putting it in a named section @code{shared} and marking the section
3125 int foo __attribute__((section ("shared"), shared)) = 0;
3130 /* @r{Read and write foo. All running
3131 copies see the same value.} */
3137 You may only use the @code{shared} attribute along with @code{section}
3138 attribute with a fully initialized global definition because of the way
3139 linkers work. See @code{section} attribute for more information.
3141 The @code{shared} attribute is only available on Microsoft Windows@.
3143 @item tls_model ("@var{tls_model}")
3144 @cindex @code{tls_model} attribute
3145 The @code{tls_model} attribute sets thread-local storage model
3146 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3147 overriding @option{-ftls-model=} command line switch on a per-variable
3149 The @var{tls_model} argument should be one of @code{global-dynamic},
3150 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3152 Not all targets support this attribute.
3155 This attribute, attached to a variable, means that the variable is meant
3156 to be possibly unused. GCC will not produce a warning for this
3159 @item vector_size (@var{bytes})
3160 This attribute specifies the vector size for the variable, measured in
3161 bytes. For example, the declaration:
3164 int foo __attribute__ ((vector_size (16)));
3168 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3169 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3170 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3172 This attribute is only applicable to integral and float scalars,
3173 although arrays, pointers, and function return values are allowed in
3174 conjunction with this construct.
3176 Aggregates with this attribute are invalid, even if they are of the same
3177 size as a corresponding scalar. For example, the declaration:
3180 struct S @{ int a; @};
3181 struct S __attribute__ ((vector_size (16))) foo;
3185 is invalid even if the size of the structure is the same as the size of
3189 The @code{selectany} attribute causes an initialized global variable to
3190 have link-once semantics. When multiple definitions of the variable are
3191 encountered by the linker, the first is selected and the remainder are
3192 discarded. Following usage by the Microsoft compiler, the linker is told
3193 @emph{not} to warn about size or content differences of the multiple
3196 Although the primary usage of this attribute is for POD types, the
3197 attribute can also be applied to global C++ objects that are initialized
3198 by a constructor. In this case, the static initialization and destruction
3199 code for the object is emitted in each translation defining the object,
3200 but the calls to the constructor and destructor are protected by a
3201 link-once guard variable.
3203 The @code{selectany} attribute is only available on Microsoft Windows
3204 targets. You can use @code{__declspec (selectany)} as a synonym for
3205 @code{__attribute__ ((selectany))} for compatibility with other
3209 The @code{weak} attribute is described in @xref{Function Attributes}.
3212 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3215 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3219 @subsection M32R/D Variable Attributes
3221 One attribute is currently defined for the M32R/D@.
3224 @item model (@var{model-name})
3225 @cindex variable addressability on the M32R/D
3226 Use this attribute on the M32R/D to set the addressability of an object.
3227 The identifier @var{model-name} is one of @code{small}, @code{medium},
3228 or @code{large}, representing each of the code models.
3230 Small model objects live in the lower 16MB of memory (so that their
3231 addresses can be loaded with the @code{ld24} instruction).
3233 Medium and large model objects may live anywhere in the 32-bit address space
3234 (the compiler will generate @code{seth/add3} instructions to load their
3238 @subsection i386 Variable Attributes
3240 Two attributes are currently defined for i386 configurations:
3241 @code{ms_struct} and @code{gcc_struct}
3246 @cindex @code{ms_struct} attribute
3247 @cindex @code{gcc_struct} attribute
3249 If @code{packed} is used on a structure, or if bit-fields are used
3250 it may be that the Microsoft ABI packs them differently
3251 than GCC would normally pack them. Particularly when moving packed
3252 data between functions compiled with GCC and the native Microsoft compiler
3253 (either via function call or as data in a file), it may be necessary to access
3256 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3257 compilers to match the native Microsoft compiler.
3259 The Microsoft structure layout algorithm is fairly simple with the exception
3260 of the bitfield packing:
3262 The padding and alignment of members of structures and whether a bit field
3263 can straddle a storage-unit boundary
3266 @item Structure members are stored sequentially in the order in which they are
3267 declared: the first member has the lowest memory address and the last member
3270 @item Every data object has an alignment-requirement. The alignment-requirement
3271 for all data except structures, unions, and arrays is either the size of the
3272 object or the current packing size (specified with either the aligned attribute
3273 or the pack pragma), whichever is less. For structures, unions, and arrays,
3274 the alignment-requirement is the largest alignment-requirement of its members.
3275 Every object is allocated an offset so that:
3277 offset % alignment-requirement == 0
3279 @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation
3280 unit if the integral types are the same size and if the next bit field fits
3281 into the current allocation unit without crossing the boundary imposed by the
3282 common alignment requirements of the bit fields.
3285 Handling of zero-length bitfields:
3287 MSVC interprets zero-length bitfields in the following ways:
3290 @item If a zero-length bitfield is inserted between two bitfields that would
3291 normally be coalesced, the bitfields will not be coalesced.
3298 unsigned long bf_1 : 12;
3300 unsigned long bf_2 : 12;
3304 The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the
3305 zero-length bitfield were removed, @code{t1}'s size would be 4 bytes.
3307 @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the
3308 alignment of the zero-length bitfield is greater than the member that follows it,
3309 @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield.
3329 For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1.
3330 Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length
3331 bitfield will not affect the alignment of @code{bar} or, as a result, the size
3334 Taking this into account, it is important to note the following:
3337 @item If a zero-length bitfield follows a normal bitfield, the type of the
3338 zero-length bitfield may affect the alignment of the structure as whole. For
3339 example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a
3340 normal bitfield, and is of type short.
3342 @item Even if a zero-length bitfield is not followed by a normal bitfield, it may
3343 still affect the alignment of the structure:
3353 Here, @code{t4} will take up 4 bytes.
3356 @item Zero-length bitfields following non-bitfield members are ignored:
3367 Here, @code{t5} will take up 2 bytes.
3371 @subsection Xstormy16 Variable Attributes
3373 One attribute is currently defined for xstormy16 configurations:
3378 @cindex @code{below100} attribute
3380 If a variable has the @code{below100} attribute (@code{BELOW100} is
3381 allowed also), GCC will place the variable in the first 0x100 bytes of
3382 memory and use special opcodes to access it. Such variables will be
3383 placed in either the @code{.bss_below100} section or the
3384 @code{.data_below100} section.
3388 @node Type Attributes
3389 @section Specifying Attributes of Types
3390 @cindex attribute of types
3391 @cindex type attributes
3393 The keyword @code{__attribute__} allows you to specify special
3394 attributes of @code{struct} and @code{union} types when you define
3395 such types. This keyword is followed by an attribute specification
3396 inside double parentheses. Seven attributes are currently defined for
3397 types: @code{aligned}, @code{packed}, @code{transparent_union},
3398 @code{unused}, @code{deprecated}, @code{visibility}, and
3399 @code{may_alias}. Other attributes are defined for functions
3400 (@pxref{Function Attributes}) and for variables (@pxref{Variable
3403 You may also specify any one of these attributes with @samp{__}
3404 preceding and following its keyword. This allows you to use these
3405 attributes in header files without being concerned about a possible
3406 macro of the same name. For example, you may use @code{__aligned__}
3407 instead of @code{aligned}.
3409 You may specify type attributes either in a @code{typedef} declaration
3410 or in an enum, struct or union type declaration or definition.
3412 For an enum, struct or union type, you may specify attributes either
3413 between the enum, struct or union tag and the name of the type, or
3414 just past the closing curly brace of the @emph{definition}. The
3415 former syntax is preferred.
3417 @xref{Attribute Syntax}, for details of the exact syntax for using
3421 @cindex @code{aligned} attribute
3422 @item aligned (@var{alignment})
3423 This attribute specifies a minimum alignment (in bytes) for variables
3424 of the specified type. For example, the declarations:
3427 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3428 typedef int more_aligned_int __attribute__ ((aligned (8)));
3432 force the compiler to insure (as far as it can) that each variable whose
3433 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3434 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3435 variables of type @code{struct S} aligned to 8-byte boundaries allows
3436 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3437 store) instructions when copying one variable of type @code{struct S} to
3438 another, thus improving run-time efficiency.
3440 Note that the alignment of any given @code{struct} or @code{union} type
3441 is required by the ISO C standard to be at least a perfect multiple of
3442 the lowest common multiple of the alignments of all of the members of
3443 the @code{struct} or @code{union} in question. This means that you @emph{can}
3444 effectively adjust the alignment of a @code{struct} or @code{union}
3445 type by attaching an @code{aligned} attribute to any one of the members
3446 of such a type, but the notation illustrated in the example above is a
3447 more obvious, intuitive, and readable way to request the compiler to
3448 adjust the alignment of an entire @code{struct} or @code{union} type.
3450 As in the preceding example, you can explicitly specify the alignment
3451 (in bytes) that you wish the compiler to use for a given @code{struct}
3452 or @code{union} type. Alternatively, you can leave out the alignment factor
3453 and just ask the compiler to align a type to the maximum
3454 useful alignment for the target machine you are compiling for. For
3455 example, you could write:
3458 struct S @{ short f[3]; @} __attribute__ ((aligned));
3461 Whenever you leave out the alignment factor in an @code{aligned}
3462 attribute specification, the compiler automatically sets the alignment
3463 for the type to the largest alignment which is ever used for any data
3464 type on the target machine you are compiling for. Doing this can often
3465 make copy operations more efficient, because the compiler can use
3466 whatever instructions copy the biggest chunks of memory when performing
3467 copies to or from the variables which have types that you have aligned
3470 In the example above, if the size of each @code{short} is 2 bytes, then
3471 the size of the entire @code{struct S} type is 6 bytes. The smallest
3472 power of two which is greater than or equal to that is 8, so the
3473 compiler sets the alignment for the entire @code{struct S} type to 8
3476 Note that although you can ask the compiler to select a time-efficient
3477 alignment for a given type and then declare only individual stand-alone
3478 objects of that type, the compiler's ability to select a time-efficient
3479 alignment is primarily useful only when you plan to create arrays of
3480 variables having the relevant (efficiently aligned) type. If you
3481 declare or use arrays of variables of an efficiently-aligned type, then
3482 it is likely that your program will also be doing pointer arithmetic (or
3483 subscripting, which amounts to the same thing) on pointers to the
3484 relevant type, and the code that the compiler generates for these
3485 pointer arithmetic operations will often be more efficient for
3486 efficiently-aligned types than for other types.
3488 The @code{aligned} attribute can only increase the alignment; but you
3489 can decrease it by specifying @code{packed} as well. See below.
3491 Note that the effectiveness of @code{aligned} attributes may be limited
3492 by inherent limitations in your linker. On many systems, the linker is
3493 only able to arrange for variables to be aligned up to a certain maximum
3494 alignment. (For some linkers, the maximum supported alignment may
3495 be very very small.) If your linker is only able to align variables
3496 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3497 in an @code{__attribute__} will still only provide you with 8 byte
3498 alignment. See your linker documentation for further information.
3501 This attribute, attached to @code{struct} or @code{union} type
3502 definition, specifies that each member (other than zero-width bitfields)
3503 of the structure or union is placed to minimize the memory required. When
3504 attached to an @code{enum} definition, it indicates that the smallest
3505 integral type should be used.
3507 @opindex fshort-enums
3508 Specifying this attribute for @code{struct} and @code{union} types is
3509 equivalent to specifying the @code{packed} attribute on each of the
3510 structure or union members. Specifying the @option{-fshort-enums}
3511 flag on the line is equivalent to specifying the @code{packed}
3512 attribute on all @code{enum} definitions.
3514 In the following example @code{struct my_packed_struct}'s members are
3515 packed closely together, but the internal layout of its @code{s} member
3516 is not packed---to do that, @code{struct my_unpacked_struct} would need to
3520 struct my_unpacked_struct
3526 struct __attribute__ ((__packed__)) my_packed_struct
3530 struct my_unpacked_struct s;
3534 You may only specify this attribute on the definition of a @code{enum},
3535 @code{struct} or @code{union}, not on a @code{typedef} which does not
3536 also define the enumerated type, structure or union.
3538 @item transparent_union
3539 This attribute, attached to a @code{union} type definition, indicates
3540 that any function parameter having that union type causes calls to that
3541 function to be treated in a special way.
3543 First, the argument corresponding to a transparent union type can be of
3544 any type in the union; no cast is required. Also, if the union contains
3545 a pointer type, the corresponding argument can be a null pointer
3546 constant or a void pointer expression; and if the union contains a void
3547 pointer type, the corresponding argument can be any pointer expression.
3548 If the union member type is a pointer, qualifiers like @code{const} on
3549 the referenced type must be respected, just as with normal pointer
3552 Second, the argument is passed to the function using the calling
3553 conventions of the first member of the transparent union, not the calling
3554 conventions of the union itself. All members of the union must have the
3555 same machine representation; this is necessary for this argument passing
3558 Transparent unions are designed for library functions that have multiple
3559 interfaces for compatibility reasons. For example, suppose the
3560 @code{wait} function must accept either a value of type @code{int *} to
3561 comply with Posix, or a value of type @code{union wait *} to comply with
3562 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3563 @code{wait} would accept both kinds of arguments, but it would also
3564 accept any other pointer type and this would make argument type checking
3565 less useful. Instead, @code{<sys/wait.h>} might define the interface
3573 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3575 pid_t wait (wait_status_ptr_t);
3578 This interface allows either @code{int *} or @code{union wait *}
3579 arguments to be passed, using the @code{int *} calling convention.
3580 The program can call @code{wait} with arguments of either type:
3583 int w1 () @{ int w; return wait (&w); @}
3584 int w2 () @{ union wait w; return wait (&w); @}
3587 With this interface, @code{wait}'s implementation might look like this:
3590 pid_t wait (wait_status_ptr_t p)
3592 return waitpid (-1, p.__ip, 0);
3597 When attached to a type (including a @code{union} or a @code{struct}),
3598 this attribute means that variables of that type are meant to appear
3599 possibly unused. GCC will not produce a warning for any variables of
3600 that type, even if the variable appears to do nothing. This is often
3601 the case with lock or thread classes, which are usually defined and then
3602 not referenced, but contain constructors and destructors that have
3603 nontrivial bookkeeping functions.
3606 The @code{deprecated} attribute results in a warning if the type
3607 is used anywhere in the source file. This is useful when identifying
3608 types that are expected to be removed in a future version of a program.
3609 If possible, the warning also includes the location of the declaration
3610 of the deprecated type, to enable users to easily find further
3611 information about why the type is deprecated, or what they should do
3612 instead. Note that the warnings only occur for uses and then only
3613 if the type is being applied to an identifier that itself is not being
3614 declared as deprecated.
3617 typedef int T1 __attribute__ ((deprecated));
3621 typedef T1 T3 __attribute__ ((deprecated));
3622 T3 z __attribute__ ((deprecated));
3625 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3626 warning is issued for line 4 because T2 is not explicitly
3627 deprecated. Line 5 has no warning because T3 is explicitly
3628 deprecated. Similarly for line 6.
3630 The @code{deprecated} attribute can also be used for functions and
3631 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3634 Accesses to objects with types with this attribute are not subjected to
3635 type-based alias analysis, but are instead assumed to be able to alias
3636 any other type of objects, just like the @code{char} type. See
3637 @option{-fstrict-aliasing} for more information on aliasing issues.
3642 typedef short __attribute__((__may_alias__)) short_a;
3648 short_a *b = (short_a *) &a;
3652 if (a == 0x12345678)
3659 If you replaced @code{short_a} with @code{short} in the variable
3660 declaration, the above program would abort when compiled with
3661 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3662 above in recent GCC versions.
3666 In C++, attribute visibility (@pxref{Function Attributes}) can also be
3667 applied to class, struct, union and enum types. Unlike other type
3668 attributes, the attribute must appear between the initial keyword and
3669 the name of the type; it cannot appear after the body of the type.
3671 @subsection ARM Type Attributes
3673 On those ARM targets that support @code{dllimport} (such as Symbian
3674 OS), you can use the @code{notshared} attribute to indicate that the
3675 virtual table and other similar data for a class should not be
3676 exported from a DLL@. For example:
3679 class __declspec(notshared) C @{
3681 __declspec(dllimport) C();
3685 __declspec(dllexport)
3689 In this code, @code{C::C} is exported from the current DLL, but the
3690 virtual table for @code{C} is not exported. (You can use
3691 @code{__attribute__} instead of @code{__declspec} if you prefer, but
3692 most Symbian OS code uses @code{__declspec}.)
3694 @subsection i386 Type Attributes
3696 Two attributes are currently defined for i386 configurations:
3697 @code{ms_struct} and @code{gcc_struct}
3701 @cindex @code{ms_struct}
3702 @cindex @code{gcc_struct}
3704 If @code{packed} is used on a structure, or if bit-fields are used
3705 it may be that the Microsoft ABI packs them differently
3706 than GCC would normally pack them. Particularly when moving packed
3707 data between functions compiled with GCC and the native Microsoft compiler
3708 (either via function call or as data in a file), it may be necessary to access
3711 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3712 compilers to match the native Microsoft compiler.
3715 To specify multiple attributes, separate them by commas within the
3716 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3720 @section An Inline Function is As Fast As a Macro
3721 @cindex inline functions
3722 @cindex integrating function code
3724 @cindex macros, inline alternative
3726 By declaring a function @code{inline}, you can direct GCC to
3727 integrate that function's code into the code for its callers. This
3728 makes execution faster by eliminating the function-call overhead; in
3729 addition, if any of the actual argument values are constant, their known
3730 values may permit simplifications at compile time so that not all of the
3731 inline function's code needs to be included. The effect on code size is
3732 less predictable; object code may be larger or smaller with function
3733 inlining, depending on the particular case. Inlining of functions is an
3734 optimization and it really ``works'' only in optimizing compilation. If
3735 you don't use @option{-O}, no function is really inline.
3737 Inline functions are included in the ISO C99 standard, but there are
3738 currently substantial differences between what GCC implements and what
3739 the ISO C99 standard requires.
3741 To declare a function inline, use the @code{inline} keyword in its
3742 declaration, like this:
3752 (If you are writing a header file to be included in ISO C programs, write
3753 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3754 You can also make all ``simple enough'' functions inline with the option
3755 @option{-finline-functions}.
3758 Note that certain usages in a function definition can make it unsuitable
3759 for inline substitution. Among these usages are: use of varargs, use of
3760 alloca, use of variable sized data types (@pxref{Variable Length}),
3761 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3762 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3763 will warn when a function marked @code{inline} could not be substituted,
3764 and will give the reason for the failure.
3766 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3767 does not affect the linkage of the function.
3769 @cindex automatic @code{inline} for C++ member fns
3770 @cindex @code{inline} automatic for C++ member fns
3771 @cindex member fns, automatically @code{inline}
3772 @cindex C++ member fns, automatically @code{inline}
3773 @opindex fno-default-inline
3774 GCC automatically inlines member functions defined within the class
3775 body of C++ programs even if they are not explicitly declared
3776 @code{inline}. (You can override this with @option{-fno-default-inline};
3777 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3779 @cindex inline functions, omission of
3780 @opindex fkeep-inline-functions
3781 When a function is both inline and @code{static}, if all calls to the
3782 function are integrated into the caller, and the function's address is
3783 never used, then the function's own assembler code is never referenced.
3784 In this case, GCC does not actually output assembler code for the
3785 function, unless you specify the option @option{-fkeep-inline-functions}.
3786 Some calls cannot be integrated for various reasons (in particular,
3787 calls that precede the function's definition cannot be integrated, and
3788 neither can recursive calls within the definition). If there is a
3789 nonintegrated call, then the function is compiled to assembler code as
3790 usual. The function must also be compiled as usual if the program
3791 refers to its address, because that can't be inlined.
3793 @cindex non-static inline function
3794 When an inline function is not @code{static}, then the compiler must assume
3795 that there may be calls from other source files; since a global symbol can
3796 be defined only once in any program, the function must not be defined in
3797 the other source files, so the calls therein cannot be integrated.
3798 Therefore, a non-@code{static} inline function is always compiled on its
3799 own in the usual fashion.
3801 If you specify both @code{inline} and @code{extern} in the function
3802 definition, then the definition is used only for inlining. In no case
3803 is the function compiled on its own, not even if you refer to its
3804 address explicitly. Such an address becomes an external reference, as
3805 if you had only declared the function, and had not defined it.
3807 This combination of @code{inline} and @code{extern} has almost the
3808 effect of a macro. The way to use it is to put a function definition in
3809 a header file with these keywords, and put another copy of the
3810 definition (lacking @code{inline} and @code{extern}) in a library file.
3811 The definition in the header file will cause most calls to the function
3812 to be inlined. If any uses of the function remain, they will refer to
3813 the single copy in the library.
3815 Since GCC eventually will implement ISO C99 semantics for
3816 inline functions, it is best to use @code{static inline} only
3817 to guarantee compatibility. (The
3818 existing semantics will remain available when @option{-std=gnu89} is
3819 specified, but eventually the default will be @option{-std=gnu99} and
3820 that will implement the C99 semantics, though it does not do so yet.)
3822 GCC does not inline any functions when not optimizing unless you specify
3823 the @samp{always_inline} attribute for the function, like this:
3826 /* @r{Prototype.} */
3827 inline void foo (const char) __attribute__((always_inline));
3831 @section Assembler Instructions with C Expression Operands
3832 @cindex extended @code{asm}
3833 @cindex @code{asm} expressions
3834 @cindex assembler instructions
3837 In an assembler instruction using @code{asm}, you can specify the
3838 operands of the instruction using C expressions. This means you need not
3839 guess which registers or memory locations will contain the data you want
3842 You must specify an assembler instruction template much like what
3843 appears in a machine description, plus an operand constraint string for
3846 For example, here is how to use the 68881's @code{fsinx} instruction:
3849 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3853 Here @code{angle} is the C expression for the input operand while
3854 @code{result} is that of the output operand. Each has @samp{"f"} as its
3855 operand constraint, saying that a floating point register is required.
3856 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3857 output operands' constraints must use @samp{=}. The constraints use the
3858 same language used in the machine description (@pxref{Constraints}).
3860 Each operand is described by an operand-constraint string followed by
3861 the C expression in parentheses. A colon separates the assembler
3862 template from the first output operand and another separates the last
3863 output operand from the first input, if any. Commas separate the
3864 operands within each group. The total number of operands is currently
3865 limited to 30; this limitation may be lifted in some future version of
3868 If there are no output operands but there are input operands, you must
3869 place two consecutive colons surrounding the place where the output
3872 As of GCC version 3.1, it is also possible to specify input and output
3873 operands using symbolic names which can be referenced within the
3874 assembler code. These names are specified inside square brackets
3875 preceding the constraint string, and can be referenced inside the
3876 assembler code using @code{%[@var{name}]} instead of a percentage sign
3877 followed by the operand number. Using named operands the above example
3881 asm ("fsinx %[angle],%[output]"
3882 : [output] "=f" (result)
3883 : [angle] "f" (angle));
3887 Note that the symbolic operand names have no relation whatsoever to
3888 other C identifiers. You may use any name you like, even those of
3889 existing C symbols, but you must ensure that no two operands within the same
3890 assembler construct use the same symbolic name.
3892 Output operand expressions must be lvalues; the compiler can check this.
3893 The input operands need not be lvalues. The compiler cannot check
3894 whether the operands have data types that are reasonable for the
3895 instruction being executed. It does not parse the assembler instruction
3896 template and does not know what it means or even whether it is valid
3897 assembler input. The extended @code{asm} feature is most often used for
3898 machine instructions the compiler itself does not know exist. If
3899 the output expression cannot be directly addressed (for example, it is a
3900 bit-field), your constraint must allow a register. In that case, GCC
3901 will use the register as the output of the @code{asm}, and then store
3902 that register into the output.
3904 The ordinary output operands must be write-only; GCC will assume that
3905 the values in these operands before the instruction are dead and need
3906 not be generated. Extended asm supports input-output or read-write
3907 operands. Use the constraint character @samp{+} to indicate such an
3908 operand and list it with the output operands. You should only use
3909 read-write operands when the constraints for the operand (or the
3910 operand in which only some of the bits are to be changed) allow a
3913 You may, as an alternative, logically split its function into two
3914 separate operands, one input operand and one write-only output
3915 operand. The connection between them is expressed by constraints
3916 which say they need to be in the same location when the instruction
3917 executes. You can use the same C expression for both operands, or
3918 different expressions. For example, here we write the (fictitious)
3919 @samp{combine} instruction with @code{bar} as its read-only source
3920 operand and @code{foo} as its read-write destination:
3923 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3927 The constraint @samp{"0"} for operand 1 says that it must occupy the
3928 same location as operand 0. A number in constraint is allowed only in
3929 an input operand and it must refer to an output operand.
3931 Only a number in the constraint can guarantee that one operand will be in
3932 the same place as another. The mere fact that @code{foo} is the value
3933 of both operands is not enough to guarantee that they will be in the
3934 same place in the generated assembler code. The following would not
3938 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3941 Various optimizations or reloading could cause operands 0 and 1 to be in
3942 different registers; GCC knows no reason not to do so. For example, the
3943 compiler might find a copy of the value of @code{foo} in one register and
3944 use it for operand 1, but generate the output operand 0 in a different
3945 register (copying it afterward to @code{foo}'s own address). Of course,
3946 since the register for operand 1 is not even mentioned in the assembler
3947 code, the result will not work, but GCC can't tell that.
3949 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3950 the operand number for a matching constraint. For example:
3953 asm ("cmoveq %1,%2,%[result]"
3954 : [result] "=r"(result)
3955 : "r" (test), "r"(new), "[result]"(old));
3958 Sometimes you need to make an @code{asm} operand be a specific register,
3959 but there's no matching constraint letter for that register @emph{by
3960 itself}. To force the operand into that register, use a local variable
3961 for the operand and specify the register in the variable declaration.
3962 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
3963 register constraint letter that matches the register:
3966 register int *p1 asm ("r0") = @dots{};
3967 register int *p2 asm ("r1") = @dots{};
3968 register int *result asm ("r0");
3969 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3972 @anchor{Example of asm with clobbered asm reg}
3973 In the above example, beware that a register that is call-clobbered by
3974 the target ABI will be overwritten by any function call in the
3975 assignment, including library calls for arithmetic operators.
3976 Assuming it is a call-clobbered register, this may happen to @code{r0}
3977 above by the assignment to @code{p2}. If you have to use such a
3978 register, use temporary variables for expressions between the register
3983 register int *p1 asm ("r0") = @dots{};
3984 register int *p2 asm ("r1") = t1;
3985 register int *result asm ("r0");
3986 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3989 Some instructions clobber specific hard registers. To describe this,
3990 write a third colon after the input operands, followed by the names of
3991 the clobbered hard registers (given as strings). Here is a realistic
3992 example for the VAX:
3995 asm volatile ("movc3 %0,%1,%2"
3996 : /* @r{no outputs} */
3997 : "g" (from), "g" (to), "g" (count)
3998 : "r0", "r1", "r2", "r3", "r4", "r5");
4001 You may not write a clobber description in a way that overlaps with an
4002 input or output operand. For example, you may not have an operand
4003 describing a register class with one member if you mention that register
4004 in the clobber list. Variables declared to live in specific registers
4005 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
4006 have no part mentioned in the clobber description.
4007 There is no way for you to specify that an input
4008 operand is modified without also specifying it as an output
4009 operand. Note that if all the output operands you specify are for this
4010 purpose (and hence unused), you will then also need to specify
4011 @code{volatile} for the @code{asm} construct, as described below, to
4012 prevent GCC from deleting the @code{asm} statement as unused.
4014 If you refer to a particular hardware register from the assembler code,
4015 you will probably have to list the register after the third colon to
4016 tell the compiler the register's value is modified. In some assemblers,
4017 the register names begin with @samp{%}; to produce one @samp{%} in the
4018 assembler code, you must write @samp{%%} in the input.
4020 If your assembler instruction can alter the condition code register, add
4021 @samp{cc} to the list of clobbered registers. GCC on some machines
4022 represents the condition codes as a specific hardware register;
4023 @samp{cc} serves to name this register. On other machines, the
4024 condition code is handled differently, and specifying @samp{cc} has no
4025 effect. But it is valid no matter what the machine.
4027 If your assembler instructions access memory in an unpredictable
4028 fashion, add @samp{memory} to the list of clobbered registers. This
4029 will cause GCC to not keep memory values cached in registers across the
4030 assembler instruction and not optimize stores or loads to that memory.
4031 You will also want to add the @code{volatile} keyword if the memory
4032 affected is not listed in the inputs or outputs of the @code{asm}, as
4033 the @samp{memory} clobber does not count as a side-effect of the
4034 @code{asm}. If you know how large the accessed memory is, you can add
4035 it as input or output but if this is not known, you should add
4036 @samp{memory}. As an example, if you access ten bytes of a string, you
4037 can use a memory input like:
4040 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
4043 Note that in the following example the memory input is necessary,
4044 otherwise GCC might optimize the store to @code{x} away:
4051 asm ("magic stuff accessing an 'int' pointed to by '%1'"
4052 "=&d" (r) : "a" (y), "m" (*y));
4057 You can put multiple assembler instructions together in a single
4058 @code{asm} template, separated by the characters normally used in assembly
4059 code for the system. A combination that works in most places is a newline
4060 to break the line, plus a tab character to move to the instruction field
4061 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
4062 assembler allows semicolons as a line-breaking character. Note that some
4063 assembler dialects use semicolons to start a comment.
4064 The input operands are guaranteed not to use any of the clobbered
4065 registers, and neither will the output operands' addresses, so you can
4066 read and write the clobbered registers as many times as you like. Here
4067 is an example of multiple instructions in a template; it assumes the
4068 subroutine @code{_foo} accepts arguments in registers 9 and 10:
4071 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
4073 : "g" (from), "g" (to)
4077 Unless an output operand has the @samp{&} constraint modifier, GCC
4078 may allocate it in the same register as an unrelated input operand, on
4079 the assumption the inputs are consumed before the outputs are produced.
4080 This assumption may be false if the assembler code actually consists of
4081 more than one instruction. In such a case, use @samp{&} for each output
4082 operand that may not overlap an input. @xref{Modifiers}.
4084 If you want to test the condition code produced by an assembler
4085 instruction, you must include a branch and a label in the @code{asm}
4086 construct, as follows:
4089 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
4095 This assumes your assembler supports local labels, as the GNU assembler
4096 and most Unix assemblers do.
4098 Speaking of labels, jumps from one @code{asm} to another are not
4099 supported. The compiler's optimizers do not know about these jumps, and
4100 therefore they cannot take account of them when deciding how to
4103 @cindex macros containing @code{asm}
4104 Usually the most convenient way to use these @code{asm} instructions is to
4105 encapsulate them in macros that look like functions. For example,
4109 (@{ double __value, __arg = (x); \
4110 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
4115 Here the variable @code{__arg} is used to make sure that the instruction
4116 operates on a proper @code{double} value, and to accept only those
4117 arguments @code{x} which can convert automatically to a @code{double}.
4119 Another way to make sure the instruction operates on the correct data
4120 type is to use a cast in the @code{asm}. This is different from using a
4121 variable @code{__arg} in that it converts more different types. For
4122 example, if the desired type were @code{int}, casting the argument to
4123 @code{int} would accept a pointer with no complaint, while assigning the
4124 argument to an @code{int} variable named @code{__arg} would warn about
4125 using a pointer unless the caller explicitly casts it.
4127 If an @code{asm} has output operands, GCC assumes for optimization
4128 purposes the instruction has no side effects except to change the output
4129 operands. This does not mean instructions with a side effect cannot be
4130 used, but you must be careful, because the compiler may eliminate them
4131 if the output operands aren't used, or move them out of loops, or
4132 replace two with one if they constitute a common subexpression. Also,
4133 if your instruction does have a side effect on a variable that otherwise
4134 appears not to change, the old value of the variable may be reused later
4135 if it happens to be found in a register.
4137 You can prevent an @code{asm} instruction from being deleted
4138 by writing the keyword @code{volatile} after
4139 the @code{asm}. For example:
4142 #define get_and_set_priority(new) \
4144 asm volatile ("get_and_set_priority %0, %1" \
4145 : "=g" (__old) : "g" (new)); \
4150 The @code{volatile} keyword indicates that the instruction has
4151 important side-effects. GCC will not delete a volatile @code{asm} if
4152 it is reachable. (The instruction can still be deleted if GCC can
4153 prove that control-flow will never reach the location of the
4154 instruction.) Note that even a volatile @code{asm} instruction
4155 can be moved relative to other code, including across jump
4156 instructions. For example, on many targets there is a system
4157 register which can be set to control the rounding mode of
4158 floating point operations. You might try
4159 setting it with a volatile @code{asm}, like this PowerPC example:
4162 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
4167 This will not work reliably, as the compiler may move the addition back
4168 before the volatile @code{asm}. To make it work you need to add an
4169 artificial dependency to the @code{asm} referencing a variable in the code
4170 you don't want moved, for example:
4173 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
4177 Similarly, you can't expect a
4178 sequence of volatile @code{asm} instructions to remain perfectly
4179 consecutive. If you want consecutive output, use a single @code{asm}.
4180 Also, GCC will perform some optimizations across a volatile @code{asm}
4181 instruction; GCC does not ``forget everything'' when it encounters
4182 a volatile @code{asm} instruction the way some other compilers do.
4184 An @code{asm} instruction without any output operands will be treated
4185 identically to a volatile @code{asm} instruction.
4187 It is a natural idea to look for a way to give access to the condition
4188 code left by the assembler instruction. However, when we attempted to
4189 implement this, we found no way to make it work reliably. The problem
4190 is that output operands might need reloading, which would result in
4191 additional following ``store'' instructions. On most machines, these
4192 instructions would alter the condition code before there was time to
4193 test it. This problem doesn't arise for ordinary ``test'' and
4194 ``compare'' instructions because they don't have any output operands.
4196 For reasons similar to those described above, it is not possible to give
4197 an assembler instruction access to the condition code left by previous
4200 If you are writing a header file that should be includable in ISO C
4201 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
4204 @subsection Size of an @code{asm}
4206 Some targets require that GCC track the size of each instruction used in
4207 order to generate correct code. Because the final length of an
4208 @code{asm} is only known by the assembler, GCC must make an estimate as
4209 to how big it will be. The estimate is formed by counting the number of
4210 statements in the pattern of the @code{asm} and multiplying that by the
4211 length of the longest instruction on that processor. Statements in the
4212 @code{asm} are identified by newline characters and whatever statement
4213 separator characters are supported by the assembler; on most processors
4214 this is the `@code{;}' character.
4216 Normally, GCC's estimate is perfectly adequate to ensure that correct
4217 code is generated, but it is possible to confuse the compiler if you use
4218 pseudo instructions or assembler macros that expand into multiple real
4219 instructions or if you use assembler directives that expand to more
4220 space in the object file than would be needed for a single instruction.
4221 If this happens then the assembler will produce a diagnostic saying that
4222 a label is unreachable.
4224 @subsection i386 floating point asm operands
4226 There are several rules on the usage of stack-like regs in
4227 asm_operands insns. These rules apply only to the operands that are
4232 Given a set of input regs that die in an asm_operands, it is
4233 necessary to know which are implicitly popped by the asm, and
4234 which must be explicitly popped by gcc.
4236 An input reg that is implicitly popped by the asm must be
4237 explicitly clobbered, unless it is constrained to match an
4241 For any input reg that is implicitly popped by an asm, it is
4242 necessary to know how to adjust the stack to compensate for the pop.
4243 If any non-popped input is closer to the top of the reg-stack than
4244 the implicitly popped reg, it would not be possible to know what the
4245 stack looked like---it's not clear how the rest of the stack ``slides
4248 All implicitly popped input regs must be closer to the top of
4249 the reg-stack than any input that is not implicitly popped.
4251 It is possible that if an input dies in an insn, reload might
4252 use the input reg for an output reload. Consider this example:
4255 asm ("foo" : "=t" (a) : "f" (b));
4258 This asm says that input B is not popped by the asm, and that
4259 the asm pushes a result onto the reg-stack, i.e., the stack is one
4260 deeper after the asm than it was before. But, it is possible that
4261 reload will think that it can use the same reg for both the input and
4262 the output, if input B dies in this insn.
4264 If any input operand uses the @code{f} constraint, all output reg
4265 constraints must use the @code{&} earlyclobber.
4267 The asm above would be written as
4270 asm ("foo" : "=&t" (a) : "f" (b));
4274 Some operands need to be in particular places on the stack. All
4275 output operands fall in this category---there is no other way to
4276 know which regs the outputs appear in unless the user indicates
4277 this in the constraints.
4279 Output operands must specifically indicate which reg an output
4280 appears in after an asm. @code{=f} is not allowed: the operand
4281 constraints must select a class with a single reg.
4284 Output operands may not be ``inserted'' between existing stack regs.
4285 Since no 387 opcode uses a read/write operand, all output operands
4286 are dead before the asm_operands, and are pushed by the asm_operands.
4287 It makes no sense to push anywhere but the top of the reg-stack.
4289 Output operands must start at the top of the reg-stack: output
4290 operands may not ``skip'' a reg.
4293 Some asm statements may need extra stack space for internal
4294 calculations. This can be guaranteed by clobbering stack registers
4295 unrelated to the inputs and outputs.
4299 Here are a couple of reasonable asms to want to write. This asm
4300 takes one input, which is internally popped, and produces two outputs.
4303 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4306 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4307 and replaces them with one output. The user must code the @code{st(1)}
4308 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4311 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4317 @section Controlling Names Used in Assembler Code
4318 @cindex assembler names for identifiers
4319 @cindex names used in assembler code
4320 @cindex identifiers, names in assembler code
4322 You can specify the name to be used in the assembler code for a C
4323 function or variable by writing the @code{asm} (or @code{__asm__})
4324 keyword after the declarator as follows:
4327 int foo asm ("myfoo") = 2;
4331 This specifies that the name to be used for the variable @code{foo} in
4332 the assembler code should be @samp{myfoo} rather than the usual
4335 On systems where an underscore is normally prepended to the name of a C
4336 function or variable, this feature allows you to define names for the
4337 linker that do not start with an underscore.
4339 It does not make sense to use this feature with a non-static local
4340 variable since such variables do not have assembler names. If you are
4341 trying to put the variable in a particular register, see @ref{Explicit
4342 Reg Vars}. GCC presently accepts such code with a warning, but will
4343 probably be changed to issue an error, rather than a warning, in the
4346 You cannot use @code{asm} in this way in a function @emph{definition}; but
4347 you can get the same effect by writing a declaration for the function
4348 before its definition and putting @code{asm} there, like this:
4351 extern func () asm ("FUNC");
4358 It is up to you to make sure that the assembler names you choose do not
4359 conflict with any other assembler symbols. Also, you must not use a
4360 register name; that would produce completely invalid assembler code. GCC
4361 does not as yet have the ability to store static variables in registers.
4362 Perhaps that will be added.
4364 @node Explicit Reg Vars
4365 @section Variables in Specified Registers
4366 @cindex explicit register variables
4367 @cindex variables in specified registers
4368 @cindex specified registers
4369 @cindex registers, global allocation
4371 GNU C allows you to put a few global variables into specified hardware
4372 registers. You can also specify the register in which an ordinary
4373 register variable should be allocated.
4377 Global register variables reserve registers throughout the program.
4378 This may be useful in programs such as programming language
4379 interpreters which have a couple of global variables that are accessed
4383 Local register variables in specific registers do not reserve the
4384 registers, except at the point where they are used as input or output
4385 operands in an @code{asm} statement and the @code{asm} statement itself is
4386 not deleted. The compiler's data flow analysis is capable of determining
4387 where the specified registers contain live values, and where they are
4388 available for other uses. Stores into local register variables may be deleted
4389 when they appear to be dead according to dataflow analysis. References
4390 to local register variables may be deleted or moved or simplified.
4392 These local variables are sometimes convenient for use with the extended
4393 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4394 output of the assembler instruction directly into a particular register.
4395 (This will work provided the register you specify fits the constraints
4396 specified for that operand in the @code{asm}.)
4404 @node Global Reg Vars
4405 @subsection Defining Global Register Variables
4406 @cindex global register variables
4407 @cindex registers, global variables in
4409 You can define a global register variable in GNU C like this:
4412 register int *foo asm ("a5");
4416 Here @code{a5} is the name of the register which should be used. Choose a
4417 register which is normally saved and restored by function calls on your
4418 machine, so that library routines will not clobber it.
4420 Naturally the register name is cpu-dependent, so you would need to
4421 conditionalize your program according to cpu type. The register
4422 @code{a5} would be a good choice on a 68000 for a variable of pointer
4423 type. On machines with register windows, be sure to choose a ``global''
4424 register that is not affected magically by the function call mechanism.
4426 In addition, operating systems on one type of cpu may differ in how they
4427 name the registers; then you would need additional conditionals. For
4428 example, some 68000 operating systems call this register @code{%a5}.
4430 Eventually there may be a way of asking the compiler to choose a register
4431 automatically, but first we need to figure out how it should choose and
4432 how to enable you to guide the choice. No solution is evident.
4434 Defining a global register variable in a certain register reserves that
4435 register entirely for this use, at least within the current compilation.
4436 The register will not be allocated for any other purpose in the functions
4437 in the current compilation. The register will not be saved and restored by
4438 these functions. Stores into this register are never deleted even if they
4439 would appear to be dead, but references may be deleted or moved or
4442 It is not safe to access the global register variables from signal
4443 handlers, or from more than one thread of control, because the system
4444 library routines may temporarily use the register for other things (unless
4445 you recompile them specially for the task at hand).
4447 @cindex @code{qsort}, and global register variables
4448 It is not safe for one function that uses a global register variable to
4449 call another such function @code{foo} by way of a third function
4450 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4451 different source file in which the variable wasn't declared). This is
4452 because @code{lose} might save the register and put some other value there.
4453 For example, you can't expect a global register variable to be available in
4454 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4455 might have put something else in that register. (If you are prepared to
4456 recompile @code{qsort} with the same global register variable, you can
4457 solve this problem.)
4459 If you want to recompile @code{qsort} or other source files which do not
4460 actually use your global register variable, so that they will not use that
4461 register for any other purpose, then it suffices to specify the compiler
4462 option @option{-ffixed-@var{reg}}. You need not actually add a global
4463 register declaration to their source code.
4465 A function which can alter the value of a global register variable cannot
4466 safely be called from a function compiled without this variable, because it
4467 could clobber the value the caller expects to find there on return.
4468 Therefore, the function which is the entry point into the part of the
4469 program that uses the global register variable must explicitly save and
4470 restore the value which belongs to its caller.
4472 @cindex register variable after @code{longjmp}
4473 @cindex global register after @code{longjmp}
4474 @cindex value after @code{longjmp}
4477 On most machines, @code{longjmp} will restore to each global register
4478 variable the value it had at the time of the @code{setjmp}. On some
4479 machines, however, @code{longjmp} will not change the value of global
4480 register variables. To be portable, the function that called @code{setjmp}
4481 should make other arrangements to save the values of the global register
4482 variables, and to restore them in a @code{longjmp}. This way, the same
4483 thing will happen regardless of what @code{longjmp} does.
4485 All global register variable declarations must precede all function
4486 definitions. If such a declaration could appear after function
4487 definitions, the declaration would be too late to prevent the register from
4488 being used for other purposes in the preceding functions.
4490 Global register variables may not have initial values, because an
4491 executable file has no means to supply initial contents for a register.
4493 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4494 registers, but certain library functions, such as @code{getwd}, as well
4495 as the subroutines for division and remainder, modify g3 and g4. g1 and
4496 g2 are local temporaries.
4498 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4499 Of course, it will not do to use more than a few of those.
4501 @node Local Reg Vars
4502 @subsection Specifying Registers for Local Variables
4503 @cindex local variables, specifying registers
4504 @cindex specifying registers for local variables
4505 @cindex registers for local variables
4507 You can define a local register variable with a specified register
4511 register int *foo asm ("a5");
4515 Here @code{a5} is the name of the register which should be used. Note
4516 that this is the same syntax used for defining global register
4517 variables, but for a local variable it would appear within a function.
4519 Naturally the register name is cpu-dependent, but this is not a
4520 problem, since specific registers are most often useful with explicit
4521 assembler instructions (@pxref{Extended Asm}). Both of these things
4522 generally require that you conditionalize your program according to
4525 In addition, operating systems on one type of cpu may differ in how they
4526 name the registers; then you would need additional conditionals. For
4527 example, some 68000 operating systems call this register @code{%a5}.
4529 Defining such a register variable does not reserve the register; it
4530 remains available for other uses in places where flow control determines
4531 the variable's value is not live.
4533 This option does not guarantee that GCC will generate code that has
4534 this variable in the register you specify at all times. You may not
4535 code an explicit reference to this register in the @emph{assembler
4536 instruction template} part of an @code{asm} statement and assume it will
4537 always refer to this variable. However, using the variable as an
4538 @code{asm} @emph{operand} guarantees that the specified register is used
4541 Stores into local register variables may be deleted when they appear to be dead
4542 according to dataflow analysis. References to local register variables may
4543 be deleted or moved or simplified.
4545 As for global register variables, it's recommended that you choose a
4546 register which is normally saved and restored by function calls on
4547 your machine, so that library routines will not clobber it. A common
4548 pitfall is to initialize multiple call-clobbered registers with
4549 arbitrary expressions, where a function call or library call for an
4550 arithmetic operator will overwrite a register value from a previous
4551 assignment, for example @code{r0} below:
4553 register int *p1 asm ("r0") = @dots{};
4554 register int *p2 asm ("r1") = @dots{};
4556 In those cases, a solution is to use a temporary variable for
4557 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
4559 @node Alternate Keywords
4560 @section Alternate Keywords
4561 @cindex alternate keywords
4562 @cindex keywords, alternate
4564 @option{-ansi} and the various @option{-std} options disable certain
4565 keywords. This causes trouble when you want to use GNU C extensions, or
4566 a general-purpose header file that should be usable by all programs,
4567 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4568 @code{inline} are not available in programs compiled with
4569 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4570 program compiled with @option{-std=c99}). The ISO C99 keyword
4571 @code{restrict} is only available when @option{-std=gnu99} (which will
4572 eventually be the default) or @option{-std=c99} (or the equivalent
4573 @option{-std=iso9899:1999}) is used.
4575 The way to solve these problems is to put @samp{__} at the beginning and
4576 end of each problematical keyword. For example, use @code{__asm__}
4577 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4579 Other C compilers won't accept these alternative keywords; if you want to
4580 compile with another compiler, you can define the alternate keywords as
4581 macros to replace them with the customary keywords. It looks like this:
4589 @findex __extension__
4591 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4593 prevent such warnings within one expression by writing
4594 @code{__extension__} before the expression. @code{__extension__} has no
4595 effect aside from this.
4597 @node Incomplete Enums
4598 @section Incomplete @code{enum} Types
4600 You can define an @code{enum} tag without specifying its possible values.
4601 This results in an incomplete type, much like what you get if you write
4602 @code{struct foo} without describing the elements. A later declaration
4603 which does specify the possible values completes the type.
4605 You can't allocate variables or storage using the type while it is
4606 incomplete. However, you can work with pointers to that type.
4608 This extension may not be very useful, but it makes the handling of
4609 @code{enum} more consistent with the way @code{struct} and @code{union}
4612 This extension is not supported by GNU C++.
4614 @node Function Names
4615 @section Function Names as Strings
4616 @cindex @code{__func__} identifier
4617 @cindex @code{__FUNCTION__} identifier
4618 @cindex @code{__PRETTY_FUNCTION__} identifier
4620 GCC provides three magic variables which hold the name of the current
4621 function, as a string. The first of these is @code{__func__}, which
4622 is part of the C99 standard:
4625 The identifier @code{__func__} is implicitly declared by the translator
4626 as if, immediately following the opening brace of each function
4627 definition, the declaration
4630 static const char __func__[] = "function-name";
4633 appeared, where function-name is the name of the lexically-enclosing
4634 function. This name is the unadorned name of the function.
4637 @code{__FUNCTION__} is another name for @code{__func__}. Older
4638 versions of GCC recognize only this name. However, it is not
4639 standardized. For maximum portability, we recommend you use
4640 @code{__func__}, but provide a fallback definition with the
4644 #if __STDC_VERSION__ < 199901L
4646 # define __func__ __FUNCTION__
4648 # define __func__ "<unknown>"
4653 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4654 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4655 the type signature of the function as well as its bare name. For
4656 example, this program:
4660 extern int printf (char *, ...);
4667 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4668 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4686 __PRETTY_FUNCTION__ = void a::sub(int)
4689 These identifiers are not preprocessor macros. In GCC 3.3 and
4690 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4691 were treated as string literals; they could be used to initialize
4692 @code{char} arrays, and they could be concatenated with other string
4693 literals. GCC 3.4 and later treat them as variables, like
4694 @code{__func__}. In C++, @code{__FUNCTION__} and
4695 @code{__PRETTY_FUNCTION__} have always been variables.
4697 @node Return Address
4698 @section Getting the Return or Frame Address of a Function
4700 These functions may be used to get information about the callers of a
4703 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4704 This function returns the return address of the current function, or of
4705 one of its callers. The @var{level} argument is number of frames to
4706 scan up the call stack. A value of @code{0} yields the return address
4707 of the current function, a value of @code{1} yields the return address
4708 of the caller of the current function, and so forth. When inlining
4709 the expected behavior is that the function will return the address of
4710 the function that will be returned to. To work around this behavior use
4711 the @code{noinline} function attribute.
4713 The @var{level} argument must be a constant integer.
4715 On some machines it may be impossible to determine the return address of
4716 any function other than the current one; in such cases, or when the top
4717 of the stack has been reached, this function will return @code{0} or a
4718 random value. In addition, @code{__builtin_frame_address} may be used
4719 to determine if the top of the stack has been reached.
4721 This function should only be used with a nonzero argument for debugging
4725 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4726 This function is similar to @code{__builtin_return_address}, but it
4727 returns the address of the function frame rather than the return address
4728 of the function. Calling @code{__builtin_frame_address} with a value of
4729 @code{0} yields the frame address of the current function, a value of
4730 @code{1} yields the frame address of the caller of the current function,
4733 The frame is the area on the stack which holds local variables and saved
4734 registers. The frame address is normally the address of the first word
4735 pushed on to the stack by the function. However, the exact definition
4736 depends upon the processor and the calling convention. If the processor
4737 has a dedicated frame pointer register, and the function has a frame,
4738 then @code{__builtin_frame_address} will return the value of the frame
4741 On some machines it may be impossible to determine the frame address of
4742 any function other than the current one; in such cases, or when the top
4743 of the stack has been reached, this function will return @code{0} if
4744 the first frame pointer is properly initialized by the startup code.
4746 This function should only be used with a nonzero argument for debugging
4750 @node Vector Extensions
4751 @section Using vector instructions through built-in functions
4753 On some targets, the instruction set contains SIMD vector instructions that
4754 operate on multiple values contained in one large register at the same time.
4755 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4758 The first step in using these extensions is to provide the necessary data
4759 types. This should be done using an appropriate @code{typedef}:
4762 typedef int v4si __attribute__ ((vector_size (16)));
4765 The @code{int} type specifies the base type, while the attribute specifies
4766 the vector size for the variable, measured in bytes. For example, the
4767 declaration above causes the compiler to set the mode for the @code{v4si}
4768 type to be 16 bytes wide and divided into @code{int} sized units. For
4769 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4770 corresponding mode of @code{foo} will be @acronym{V4SI}.
4772 The @code{vector_size} attribute is only applicable to integral and
4773 float scalars, although arrays, pointers, and function return values
4774 are allowed in conjunction with this construct.
4776 All the basic integer types can be used as base types, both as signed
4777 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4778 @code{long long}. In addition, @code{float} and @code{double} can be
4779 used to build floating-point vector types.
4781 Specifying a combination that is not valid for the current architecture
4782 will cause GCC to synthesize the instructions using a narrower mode.
4783 For example, if you specify a variable of type @code{V4SI} and your
4784 architecture does not allow for this specific SIMD type, GCC will
4785 produce code that uses 4 @code{SIs}.
4787 The types defined in this manner can be used with a subset of normal C
4788 operations. Currently, GCC will allow using the following operators
4789 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4791 The operations behave like C++ @code{valarrays}. Addition is defined as
4792 the addition of the corresponding elements of the operands. For
4793 example, in the code below, each of the 4 elements in @var{a} will be
4794 added to the corresponding 4 elements in @var{b} and the resulting
4795 vector will be stored in @var{c}.
4798 typedef int v4si __attribute__ ((vector_size (16)));
4805 Subtraction, multiplication, division, and the logical operations
4806 operate in a similar manner. Likewise, the result of using the unary
4807 minus or complement operators on a vector type is a vector whose
4808 elements are the negative or complemented values of the corresponding
4809 elements in the operand.
4811 You can declare variables and use them in function calls and returns, as
4812 well as in assignments and some casts. You can specify a vector type as
4813 a return type for a function. Vector types can also be used as function
4814 arguments. It is possible to cast from one vector type to another,
4815 provided they are of the same size (in fact, you can also cast vectors
4816 to and from other datatypes of the same size).
4818 You cannot operate between vectors of different lengths or different
4819 signedness without a cast.
4821 A port that supports hardware vector operations, usually provides a set
4822 of built-in functions that can be used to operate on vectors. For
4823 example, a function to add two vectors and multiply the result by a
4824 third could look like this:
4827 v4si f (v4si a, v4si b, v4si c)
4829 v4si tmp = __builtin_addv4si (a, b);
4830 return __builtin_mulv4si (tmp, c);
4837 @findex __builtin_offsetof
4839 GCC implements for both C and C++ a syntactic extension to implement
4840 the @code{offsetof} macro.
4844 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
4846 offsetof_member_designator:
4848 | offsetof_member_designator "." @code{identifier}
4849 | offsetof_member_designator "[" @code{expr} "]"
4852 This extension is sufficient such that
4855 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
4858 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
4859 may be dependent. In either case, @var{member} may consist of a single
4860 identifier, or a sequence of member accesses and array references.
4862 @node Atomic Builtins
4863 @section Built-in functions for atomic memory access
4865 The following builtins are intended to be compatible with those described
4866 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
4867 section 7.4. As such, they depart from the normal GCC practice of using
4868 the ``__builtin_'' prefix, and further that they are overloaded such that
4869 they work on multiple types.
4871 The definition given in the Intel documentation allows only for the use of
4872 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
4873 counterparts. GCC will allow any integral scalar or pointer type that is
4874 1, 2, 4 or 8 bytes in length.
4876 Not all operations are supported by all target processors. If a particular
4877 operation cannot be implemented on the target processor, a warning will be
4878 generated and a call an external function will be generated. The external
4879 function will carry the same name as the builtin, with an additional suffix
4880 @samp{_@var{n}} where @var{n} is the size of the data type.
4882 @c ??? Should we have a mechanism to suppress this warning? This is almost
4883 @c useful for implementing the operation under the control of an external
4886 In most cases, these builtins are considered a @dfn{full barrier}. That is,
4887 no memory operand will be moved across the operation, either forward or
4888 backward. Further, instructions will be issued as necessary to prevent the
4889 processor from speculating loads across the operation and from queuing stores
4890 after the operation.
4892 All of the routines are are described in the Intel documentation to take
4893 ``an optional list of variables protected by the memory barrier''. It's
4894 not clear what is meant by that; it could mean that @emph{only} the
4895 following variables are protected, or it could mean that these variables
4896 should in addition be protected. At present GCC ignores this list and
4897 protects all variables which are globally accessible. If in the future
4898 we make some use of this list, an empty list will continue to mean all
4899 globally accessible variables.
4902 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
4903 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
4904 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
4905 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
4906 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
4907 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
4908 @findex __sync_fetch_and_add
4909 @findex __sync_fetch_and_sub
4910 @findex __sync_fetch_and_or
4911 @findex __sync_fetch_and_and
4912 @findex __sync_fetch_and_xor
4913 @findex __sync_fetch_and_nand
4914 These builtins perform the operation suggested by the name, and
4915 returns the value that had previously been in memory. That is,
4918 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
4919 @{ tmp = *ptr; *ptr = ~tmp & value; return tmp; @} // nand
4922 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
4923 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
4924 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
4925 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
4926 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
4927 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
4928 @findex __sync_add_and_fetch
4929 @findex __sync_sub_and_fetch
4930 @findex __sync_or_and_fetch
4931 @findex __sync_and_and_fetch
4932 @findex __sync_xor_and_fetch
4933 @findex __sync_nand_and_fetch
4934 These builtins perform the operation suggested by the name, and
4935 return the new value. That is,
4938 @{ *ptr @var{op}= value; return *ptr; @}
4939 @{ *ptr = ~*ptr & value; return *ptr; @} // nand
4942 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
4943 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
4944 @findex __sync_bool_compare_and_swap
4945 @findex __sync_val_compare_and_swap
4946 These builtins perform an atomic compare and swap. That is, if the current
4947 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
4950 The ``bool'' version returns true if the comparison is successful and
4951 @var{newval} was written. The ``val'' version returns the contents
4952 of @code{*@var{ptr}} before the operation.
4954 @item __sync_synchronize (...)
4955 @findex __sync_synchronize
4956 This builtin issues a full memory barrier.
4958 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
4959 @findex __sync_lock_test_and_set
4960 This builtin, as described by Intel, is not a traditional test-and-set
4961 operation, but rather an atomic exchange operation. It writes @var{value}
4962 into @code{*@var{ptr}}, and returns the previous contents of
4965 Many targets have only minimal support for such locks, and do not support
4966 a full exchange operation. In this case, a target may support reduced
4967 functionality here by which the @emph{only} valid value to store is the
4968 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
4969 is implementation defined.
4971 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
4972 This means that references after the builtin cannot move to (or be
4973 speculated to) before the builtin, but previous memory stores may not
4974 be globally visible yet, and previous memory loads may not yet be
4977 @item void __sync_lock_release (@var{type} *ptr, ...)
4978 @findex __sync_lock_release
4979 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
4980 Normally this means writing the constant 0 to @code{*@var{ptr}}.
4982 This builtin is not a full barrier, but rather a @dfn{release barrier}.
4983 This means that all previous memory stores are globally visible, and all
4984 previous memory loads have been satisfied, but following memory reads
4985 are not prevented from being speculated to before the barrier.
4988 @node Object Size Checking
4989 @section Object Size Checking Builtins
4990 @findex __builtin_object_size
4991 @findex __builtin___memcpy_chk
4992 @findex __builtin___mempcpy_chk
4993 @findex __builtin___memmove_chk
4994 @findex __builtin___memset_chk
4995 @findex __builtin___strcpy_chk
4996 @findex __builtin___stpcpy_chk
4997 @findex __builtin___strncpy_chk
4998 @findex __builtin___strcat_chk
4999 @findex __builtin___strncat_chk
5000 @findex __builtin___sprintf_chk
5001 @findex __builtin___snprintf_chk
5002 @findex __builtin___vsprintf_chk
5003 @findex __builtin___vsnprintf_chk
5004 @findex __builtin___printf_chk
5005 @findex __builtin___vprintf_chk
5006 @findex __builtin___fprintf_chk
5007 @findex __builtin___vfprintf_chk
5009 GCC implements a limited buffer overflow protection mechanism
5010 that can prevent some buffer overflow attacks.
5012 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
5013 is a built-in construct that returns a constant number of bytes from
5014 @var{ptr} to the end of the object @var{ptr} pointer points to
5015 (if known at compile time). @code{__builtin_object_size} never evaluates
5016 its arguments for side-effects. If there are any side-effects in them, it
5017 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5018 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
5019 point to and all of them are known at compile time, the returned number
5020 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
5021 0 and minimum if nonzero. If it is not possible to determine which objects
5022 @var{ptr} points to at compile time, @code{__builtin_object_size} should
5023 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
5024 for @var{type} 2 or 3.
5026 @var{type} is an integer constant from 0 to 3. If the least significant
5027 bit is clear, objects are whole variables, if it is set, a closest
5028 surrounding subobject is considered the object a pointer points to.
5029 The second bit determines if maximum or minimum of remaining bytes
5033 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
5034 char *p = &var.buf1[1], *q = &var.b;
5036 /* Here the object p points to is var. */
5037 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
5038 /* The subobject p points to is var.buf1. */
5039 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
5040 /* The object q points to is var. */
5041 assert (__builtin_object_size (q, 0)
5042 == (char *) (&var + 1) - (char *) &var.b);
5043 /* The subobject q points to is var.b. */
5044 assert (__builtin_object_size (q, 1) == sizeof (var.b));
5048 There are built-in functions added for many common string operation
5049 functions, e.g. for @code{memcpy} @code{__builtin___memcpy_chk}
5050 built-in is provided. This built-in has an additional last argument,
5051 which is the number of bytes remaining in object the @var{dest}
5052 argument points to or @code{(size_t) -1} if the size is not known.
5054 The built-in functions are optimized into the normal string functions
5055 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
5056 it is known at compile time that the destination object will not
5057 be overflown. If the compiler can determine at compile time the
5058 object will be always overflown, it issues a warning.
5060 The intended use can be e.g.
5064 #define bos0(dest) __builtin_object_size (dest, 0)
5065 #define memcpy(dest, src, n) \
5066 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
5070 /* It is unknown what object p points to, so this is optimized
5071 into plain memcpy - no checking is possible. */
5072 memcpy (p, "abcde", n);
5073 /* Destination is known and length too. It is known at compile
5074 time there will be no overflow. */
5075 memcpy (&buf[5], "abcde", 5);
5076 /* Destination is known, but the length is not known at compile time.
5077 This will result in __memcpy_chk call that can check for overflow
5079 memcpy (&buf[5], "abcde", n);
5080 /* Destination is known and it is known at compile time there will
5081 be overflow. There will be a warning and __memcpy_chk call that
5082 will abort the program at runtime. */
5083 memcpy (&buf[6], "abcde", 5);
5086 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
5087 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
5088 @code{strcat} and @code{strncat}.
5090 There are also checking built-in functions for formatted output functions.
5092 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
5093 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5094 const char *fmt, ...);
5095 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
5097 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
5098 const char *fmt, va_list ap);
5101 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
5102 etc. functions and can contain implementation specific flags on what
5103 additional security measures the checking function might take, such as
5104 handling @code{%n} differently.
5106 The @var{os} argument is the object size @var{s} points to, like in the
5107 other built-in functions. There is a small difference in the behavior
5108 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
5109 optimized into the non-checking functions only if @var{flag} is 0, otherwise
5110 the checking function is called with @var{os} argument set to
5113 In addition to this, there are checking built-in functions
5114 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
5115 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
5116 These have just one additional argument, @var{flag}, right before
5117 format string @var{fmt}. If the compiler is able to optimize them to
5118 @code{fputc} etc. functions, it will, otherwise the checking function
5119 should be called and the @var{flag} argument passed to it.
5121 @node Other Builtins
5122 @section Other built-in functions provided by GCC
5123 @cindex built-in functions
5124 @findex __builtin_isgreater
5125 @findex __builtin_isgreaterequal
5126 @findex __builtin_isless
5127 @findex __builtin_islessequal
5128 @findex __builtin_islessgreater
5129 @findex __builtin_isunordered
5130 @findex __builtin_powi
5131 @findex __builtin_powif
5132 @findex __builtin_powil
5290 @findex fprintf_unlocked
5292 @findex fputs_unlocked
5402 @findex printf_unlocked
5431 @findex significandf
5432 @findex significandl
5503 GCC provides a large number of built-in functions other than the ones
5504 mentioned above. Some of these are for internal use in the processing
5505 of exceptions or variable-length argument lists and will not be
5506 documented here because they may change from time to time; we do not
5507 recommend general use of these functions.
5509 The remaining functions are provided for optimization purposes.
5511 @opindex fno-builtin
5512 GCC includes built-in versions of many of the functions in the standard
5513 C library. The versions prefixed with @code{__builtin_} will always be
5514 treated as having the same meaning as the C library function even if you
5515 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
5516 Many of these functions are only optimized in certain cases; if they are
5517 not optimized in a particular case, a call to the library function will
5522 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
5523 @option{-std=c99}), the functions
5524 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
5525 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
5526 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
5527 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
5528 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
5529 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
5530 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
5531 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
5532 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
5533 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
5534 @code{significandf}, @code{significandl}, @code{significand},
5535 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
5536 @code{stpncpy}, @code{strcasecmp}, @code{strdup}, @code{strfmon},
5537 @code{strncasecmp}, @code{strndup}, @code{toascii}, @code{y0f},
5538 @code{y0l}, @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf},
5539 @code{ynl} and @code{yn}
5540 may be handled as built-in functions.
5541 All these functions have corresponding versions
5542 prefixed with @code{__builtin_}, which may be used even in strict C89
5545 The ISO C99 functions
5546 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
5547 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
5548 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
5549 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
5550 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
5551 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
5552 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
5553 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
5554 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
5555 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
5556 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
5557 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
5558 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
5559 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
5560 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
5561 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
5562 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
5563 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
5564 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
5565 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
5566 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
5567 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
5568 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
5569 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
5570 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
5571 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
5572 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
5573 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
5574 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
5575 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
5576 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
5577 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
5578 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
5579 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
5580 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
5581 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
5582 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
5583 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
5584 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
5585 are handled as built-in functions
5586 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5588 There are also built-in versions of the ISO C99 functions
5589 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5590 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5591 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5592 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5593 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5594 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5595 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5596 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5597 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5598 that are recognized in any mode since ISO C90 reserves these names for
5599 the purpose to which ISO C99 puts them. All these functions have
5600 corresponding versions prefixed with @code{__builtin_}.
5602 The ISO C94 functions
5603 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
5604 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
5605 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
5607 are handled as built-in functions
5608 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5610 The ISO C90 functions
5611 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5612 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5613 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5614 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
5615 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
5616 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
5617 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
5618 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
5619 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
5620 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
5621 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
5622 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
5623 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
5624 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
5625 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
5626 @code{vprintf} and @code{vsprintf}
5627 are all recognized as built-in functions unless
5628 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5629 is specified for an individual function). All of these functions have
5630 corresponding versions prefixed with @code{__builtin_}.
5632 GCC provides built-in versions of the ISO C99 floating point comparison
5633 macros that avoid raising exceptions for unordered operands. They have
5634 the same names as the standard macros ( @code{isgreater},
5635 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5636 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5637 prefixed. We intend for a library implementor to be able to simply
5638 @code{#define} each standard macro to its built-in equivalent.
5640 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
5642 You can use the built-in function @code{__builtin_types_compatible_p} to
5643 determine whether two types are the same.
5645 This built-in function returns 1 if the unqualified versions of the
5646 types @var{type1} and @var{type2} (which are types, not expressions) are
5647 compatible, 0 otherwise. The result of this built-in function can be
5648 used in integer constant expressions.
5650 This built-in function ignores top level qualifiers (e.g., @code{const},
5651 @code{volatile}). For example, @code{int} is equivalent to @code{const
5654 The type @code{int[]} and @code{int[5]} are compatible. On the other
5655 hand, @code{int} and @code{char *} are not compatible, even if the size
5656 of their types, on the particular architecture are the same. Also, the
5657 amount of pointer indirection is taken into account when determining
5658 similarity. Consequently, @code{short *} is not similar to
5659 @code{short **}. Furthermore, two types that are typedefed are
5660 considered compatible if their underlying types are compatible.
5662 An @code{enum} type is not considered to be compatible with another
5663 @code{enum} type even if both are compatible with the same integer
5664 type; this is what the C standard specifies.
5665 For example, @code{enum @{foo, bar@}} is not similar to
5666 @code{enum @{hot, dog@}}.
5668 You would typically use this function in code whose execution varies
5669 depending on the arguments' types. For example:
5675 if (__builtin_types_compatible_p (typeof (x), long double)) \
5676 tmp = foo_long_double (tmp); \
5677 else if (__builtin_types_compatible_p (typeof (x), double)) \
5678 tmp = foo_double (tmp); \
5679 else if (__builtin_types_compatible_p (typeof (x), float)) \
5680 tmp = foo_float (tmp); \
5687 @emph{Note:} This construct is only available for C@.
5691 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
5693 You can use the built-in function @code{__builtin_choose_expr} to
5694 evaluate code depending on the value of a constant expression. This
5695 built-in function returns @var{exp1} if @var{const_exp}, which is a
5696 constant expression that must be able to be determined at compile time,
5697 is nonzero. Otherwise it returns 0.
5699 This built-in function is analogous to the @samp{? :} operator in C,
5700 except that the expression returned has its type unaltered by promotion
5701 rules. Also, the built-in function does not evaluate the expression
5702 that was not chosen. For example, if @var{const_exp} evaluates to true,
5703 @var{exp2} is not evaluated even if it has side-effects.
5705 This built-in function can return an lvalue if the chosen argument is an
5708 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5709 type. Similarly, if @var{exp2} is returned, its return type is the same
5716 __builtin_choose_expr ( \
5717 __builtin_types_compatible_p (typeof (x), double), \
5719 __builtin_choose_expr ( \
5720 __builtin_types_compatible_p (typeof (x), float), \
5722 /* @r{The void expression results in a compile-time error} \
5723 @r{when assigning the result to something.} */ \
5727 @emph{Note:} This construct is only available for C@. Furthermore, the
5728 unused expression (@var{exp1} or @var{exp2} depending on the value of
5729 @var{const_exp}) may still generate syntax errors. This may change in
5734 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5735 You can use the built-in function @code{__builtin_constant_p} to
5736 determine if a value is known to be constant at compile-time and hence
5737 that GCC can perform constant-folding on expressions involving that
5738 value. The argument of the function is the value to test. The function
5739 returns the integer 1 if the argument is known to be a compile-time
5740 constant and 0 if it is not known to be a compile-time constant. A
5741 return of 0 does not indicate that the value is @emph{not} a constant,
5742 but merely that GCC cannot prove it is a constant with the specified
5743 value of the @option{-O} option.
5745 You would typically use this function in an embedded application where
5746 memory was a critical resource. If you have some complex calculation,
5747 you may want it to be folded if it involves constants, but need to call
5748 a function if it does not. For example:
5751 #define Scale_Value(X) \
5752 (__builtin_constant_p (X) \
5753 ? ((X) * SCALE + OFFSET) : Scale (X))
5756 You may use this built-in function in either a macro or an inline
5757 function. However, if you use it in an inlined function and pass an
5758 argument of the function as the argument to the built-in, GCC will
5759 never return 1 when you call the inline function with a string constant
5760 or compound literal (@pxref{Compound Literals}) and will not return 1
5761 when you pass a constant numeric value to the inline function unless you
5762 specify the @option{-O} option.
5764 You may also use @code{__builtin_constant_p} in initializers for static
5765 data. For instance, you can write
5768 static const int table[] = @{
5769 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5775 This is an acceptable initializer even if @var{EXPRESSION} is not a
5776 constant expression. GCC must be more conservative about evaluating the
5777 built-in in this case, because it has no opportunity to perform
5780 Previous versions of GCC did not accept this built-in in data
5781 initializers. The earliest version where it is completely safe is
5785 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5786 @opindex fprofile-arcs
5787 You may use @code{__builtin_expect} to provide the compiler with
5788 branch prediction information. In general, you should prefer to
5789 use actual profile feedback for this (@option{-fprofile-arcs}), as
5790 programmers are notoriously bad at predicting how their programs
5791 actually perform. However, there are applications in which this
5792 data is hard to collect.
5794 The return value is the value of @var{exp}, which should be an
5795 integral expression. The value of @var{c} must be a compile-time
5796 constant. The semantics of the built-in are that it is expected
5797 that @var{exp} == @var{c}. For example:
5800 if (__builtin_expect (x, 0))
5805 would indicate that we do not expect to call @code{foo}, since
5806 we expect @code{x} to be zero. Since you are limited to integral
5807 expressions for @var{exp}, you should use constructions such as
5810 if (__builtin_expect (ptr != NULL, 1))
5815 when testing pointer or floating-point values.
5818 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5819 This function is used to minimize cache-miss latency by moving data into
5820 a cache before it is accessed.
5821 You can insert calls to @code{__builtin_prefetch} into code for which
5822 you know addresses of data in memory that is likely to be accessed soon.
5823 If the target supports them, data prefetch instructions will be generated.
5824 If the prefetch is done early enough before the access then the data will
5825 be in the cache by the time it is accessed.
5827 The value of @var{addr} is the address of the memory to prefetch.
5828 There are two optional arguments, @var{rw} and @var{locality}.
5829 The value of @var{rw} is a compile-time constant one or zero; one
5830 means that the prefetch is preparing for a write to the memory address
5831 and zero, the default, means that the prefetch is preparing for a read.
5832 The value @var{locality} must be a compile-time constant integer between
5833 zero and three. A value of zero means that the data has no temporal
5834 locality, so it need not be left in the cache after the access. A value
5835 of three means that the data has a high degree of temporal locality and
5836 should be left in all levels of cache possible. Values of one and two
5837 mean, respectively, a low or moderate degree of temporal locality. The
5841 for (i = 0; i < n; i++)
5844 __builtin_prefetch (&a[i+j], 1, 1);
5845 __builtin_prefetch (&b[i+j], 0, 1);
5850 Data prefetch does not generate faults if @var{addr} is invalid, but
5851 the address expression itself must be valid. For example, a prefetch
5852 of @code{p->next} will not fault if @code{p->next} is not a valid
5853 address, but evaluation will fault if @code{p} is not a valid address.
5855 If the target does not support data prefetch, the address expression
5856 is evaluated if it includes side effects but no other code is generated
5857 and GCC does not issue a warning.
5860 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5861 Returns a positive infinity, if supported by the floating-point format,
5862 else @code{DBL_MAX}. This function is suitable for implementing the
5863 ISO C macro @code{HUGE_VAL}.
5866 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5867 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5870 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5871 Similar to @code{__builtin_huge_val}, except the return
5872 type is @code{long double}.
5875 @deftypefn {Built-in Function} double __builtin_inf (void)
5876 Similar to @code{__builtin_huge_val}, except a warning is generated
5877 if the target floating-point format does not support infinities.
5880 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
5881 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
5884 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
5885 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
5888 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
5889 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
5892 @deftypefn {Built-in Function} float __builtin_inff (void)
5893 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5894 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5897 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5898 Similar to @code{__builtin_inf}, except the return
5899 type is @code{long double}.
5902 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5903 This is an implementation of the ISO C99 function @code{nan}.
5905 Since ISO C99 defines this function in terms of @code{strtod}, which we
5906 do not implement, a description of the parsing is in order. The string
5907 is parsed as by @code{strtol}; that is, the base is recognized by
5908 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5909 in the significand such that the least significant bit of the number
5910 is at the least significant bit of the significand. The number is
5911 truncated to fit the significand field provided. The significand is
5912 forced to be a quiet NaN@.
5914 This function, if given a string literal all of which would have been
5915 consumed by strtol, is evaluated early enough that it is considered a
5916 compile-time constant.
5919 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
5920 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
5923 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
5924 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
5927 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
5928 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
5931 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
5932 Similar to @code{__builtin_nan}, except the return type is @code{float}.
5935 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
5936 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
5939 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
5940 Similar to @code{__builtin_nan}, except the significand is forced
5941 to be a signaling NaN@. The @code{nans} function is proposed by
5942 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
5945 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
5946 Similar to @code{__builtin_nans}, except the return type is @code{float}.
5949 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
5950 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
5953 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
5954 Returns one plus the index of the least significant 1-bit of @var{x}, or
5955 if @var{x} is zero, returns zero.
5958 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
5959 Returns the number of leading 0-bits in @var{x}, starting at the most
5960 significant bit position. If @var{x} is 0, the result is undefined.
5963 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
5964 Returns the number of trailing 0-bits in @var{x}, starting at the least
5965 significant bit position. If @var{x} is 0, the result is undefined.
5968 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
5969 Returns the number of 1-bits in @var{x}.
5972 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
5973 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
5977 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
5978 Similar to @code{__builtin_ffs}, except the argument type is
5979 @code{unsigned long}.
5982 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
5983 Similar to @code{__builtin_clz}, except the argument type is
5984 @code{unsigned long}.
5987 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
5988 Similar to @code{__builtin_ctz}, except the argument type is
5989 @code{unsigned long}.
5992 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
5993 Similar to @code{__builtin_popcount}, except the argument type is
5994 @code{unsigned long}.
5997 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
5998 Similar to @code{__builtin_parity}, except the argument type is
5999 @code{unsigned long}.
6002 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
6003 Similar to @code{__builtin_ffs}, except the argument type is
6004 @code{unsigned long long}.
6007 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
6008 Similar to @code{__builtin_clz}, except the argument type is
6009 @code{unsigned long long}.
6012 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
6013 Similar to @code{__builtin_ctz}, except the argument type is
6014 @code{unsigned long long}.
6017 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
6018 Similar to @code{__builtin_popcount}, except the argument type is
6019 @code{unsigned long long}.
6022 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
6023 Similar to @code{__builtin_parity}, except the argument type is
6024 @code{unsigned long long}.
6027 @deftypefn {Built-in Function} double __builtin_powi (double, int)
6028 Returns the first argument raised to the power of the second. Unlike the
6029 @code{pow} function no guarantees about precision and rounding are made.
6032 @deftypefn {Built-in Function} float __builtin_powif (float, int)
6033 Similar to @code{__builtin_powi}, except the argument and return types
6037 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
6038 Similar to @code{__builtin_powi}, except the argument and return types
6039 are @code{long double}.
6043 @node Target Builtins
6044 @section Built-in Functions Specific to Particular Target Machines
6046 On some target machines, GCC supports many built-in functions specific
6047 to those machines. Generally these generate calls to specific machine
6048 instructions, but allow the compiler to schedule those calls.
6051 * Alpha Built-in Functions::
6052 * ARM Built-in Functions::
6053 * Blackfin Built-in Functions::
6054 * FR-V Built-in Functions::
6055 * X86 Built-in Functions::
6056 * MIPS DSP Built-in Functions::
6057 * MIPS Paired-Single Support::
6058 * PowerPC AltiVec Built-in Functions::
6059 * SPARC VIS Built-in Functions::
6062 @node Alpha Built-in Functions
6063 @subsection Alpha Built-in Functions
6065 These built-in functions are available for the Alpha family of
6066 processors, depending on the command-line switches used.
6068 The following built-in functions are always available. They
6069 all generate the machine instruction that is part of the name.
6072 long __builtin_alpha_implver (void)
6073 long __builtin_alpha_rpcc (void)
6074 long __builtin_alpha_amask (long)
6075 long __builtin_alpha_cmpbge (long, long)
6076 long __builtin_alpha_extbl (long, long)
6077 long __builtin_alpha_extwl (long, long)
6078 long __builtin_alpha_extll (long, long)
6079 long __builtin_alpha_extql (long, long)
6080 long __builtin_alpha_extwh (long, long)
6081 long __builtin_alpha_extlh (long, long)
6082 long __builtin_alpha_extqh (long, long)
6083 long __builtin_alpha_insbl (long, long)
6084 long __builtin_alpha_inswl (long, long)
6085 long __builtin_alpha_insll (long, long)
6086 long __builtin_alpha_insql (long, long)
6087 long __builtin_alpha_inswh (long, long)
6088 long __builtin_alpha_inslh (long, long)
6089 long __builtin_alpha_insqh (long, long)
6090 long __builtin_alpha_mskbl (long, long)
6091 long __builtin_alpha_mskwl (long, long)
6092 long __builtin_alpha_mskll (long, long)
6093 long __builtin_alpha_mskql (long, long)
6094 long __builtin_alpha_mskwh (long, long)
6095 long __builtin_alpha_msklh (long, long)
6096 long __builtin_alpha_mskqh (long, long)
6097 long __builtin_alpha_umulh (long, long)
6098 long __builtin_alpha_zap (long, long)
6099 long __builtin_alpha_zapnot (long, long)
6102 The following built-in functions are always with @option{-mmax}
6103 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
6104 later. They all generate the machine instruction that is part
6108 long __builtin_alpha_pklb (long)
6109 long __builtin_alpha_pkwb (long)
6110 long __builtin_alpha_unpkbl (long)
6111 long __builtin_alpha_unpkbw (long)
6112 long __builtin_alpha_minub8 (long, long)
6113 long __builtin_alpha_minsb8 (long, long)
6114 long __builtin_alpha_minuw4 (long, long)
6115 long __builtin_alpha_minsw4 (long, long)
6116 long __builtin_alpha_maxub8 (long, long)
6117 long __builtin_alpha_maxsb8 (long, long)
6118 long __builtin_alpha_maxuw4 (long, long)
6119 long __builtin_alpha_maxsw4 (long, long)
6120 long __builtin_alpha_perr (long, long)
6123 The following built-in functions are always with @option{-mcix}
6124 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
6125 later. They all generate the machine instruction that is part
6129 long __builtin_alpha_cttz (long)
6130 long __builtin_alpha_ctlz (long)
6131 long __builtin_alpha_ctpop (long)
6134 The following builtins are available on systems that use the OSF/1
6135 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
6136 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
6137 @code{rdval} and @code{wrval}.
6140 void *__builtin_thread_pointer (void)
6141 void __builtin_set_thread_pointer (void *)
6144 @node ARM Built-in Functions
6145 @subsection ARM Built-in Functions
6147 These built-in functions are available for the ARM family of
6148 processors, when the @option{-mcpu=iwmmxt} switch is used:
6151 typedef int v2si __attribute__ ((vector_size (8)));
6152 typedef short v4hi __attribute__ ((vector_size (8)));
6153 typedef char v8qi __attribute__ ((vector_size (8)));
6155 int __builtin_arm_getwcx (int)
6156 void __builtin_arm_setwcx (int, int)
6157 int __builtin_arm_textrmsb (v8qi, int)
6158 int __builtin_arm_textrmsh (v4hi, int)
6159 int __builtin_arm_textrmsw (v2si, int)
6160 int __builtin_arm_textrmub (v8qi, int)
6161 int __builtin_arm_textrmuh (v4hi, int)
6162 int __builtin_arm_textrmuw (v2si, int)
6163 v8qi __builtin_arm_tinsrb (v8qi, int)
6164 v4hi __builtin_arm_tinsrh (v4hi, int)
6165 v2si __builtin_arm_tinsrw (v2si, int)
6166 long long __builtin_arm_tmia (long long, int, int)
6167 long long __builtin_arm_tmiabb (long long, int, int)
6168 long long __builtin_arm_tmiabt (long long, int, int)
6169 long long __builtin_arm_tmiaph (long long, int, int)
6170 long long __builtin_arm_tmiatb (long long, int, int)
6171 long long __builtin_arm_tmiatt (long long, int, int)
6172 int __builtin_arm_tmovmskb (v8qi)
6173 int __builtin_arm_tmovmskh (v4hi)
6174 int __builtin_arm_tmovmskw (v2si)
6175 long long __builtin_arm_waccb (v8qi)
6176 long long __builtin_arm_wacch (v4hi)
6177 long long __builtin_arm_waccw (v2si)
6178 v8qi __builtin_arm_waddb (v8qi, v8qi)
6179 v8qi __builtin_arm_waddbss (v8qi, v8qi)
6180 v8qi __builtin_arm_waddbus (v8qi, v8qi)
6181 v4hi __builtin_arm_waddh (v4hi, v4hi)
6182 v4hi __builtin_arm_waddhss (v4hi, v4hi)
6183 v4hi __builtin_arm_waddhus (v4hi, v4hi)
6184 v2si __builtin_arm_waddw (v2si, v2si)
6185 v2si __builtin_arm_waddwss (v2si, v2si)
6186 v2si __builtin_arm_waddwus (v2si, v2si)
6187 v8qi __builtin_arm_walign (v8qi, v8qi, int)
6188 long long __builtin_arm_wand(long long, long long)
6189 long long __builtin_arm_wandn (long long, long long)
6190 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
6191 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
6192 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
6193 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
6194 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
6195 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
6196 v2si __builtin_arm_wcmpeqw (v2si, v2si)
6197 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
6198 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
6199 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
6200 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
6201 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
6202 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
6203 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
6204 long long __builtin_arm_wmacsz (v4hi, v4hi)
6205 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
6206 long long __builtin_arm_wmacuz (v4hi, v4hi)
6207 v4hi __builtin_arm_wmadds (v4hi, v4hi)
6208 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
6209 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
6210 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
6211 v2si __builtin_arm_wmaxsw (v2si, v2si)
6212 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
6213 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
6214 v2si __builtin_arm_wmaxuw (v2si, v2si)
6215 v8qi __builtin_arm_wminsb (v8qi, v8qi)
6216 v4hi __builtin_arm_wminsh (v4hi, v4hi)
6217 v2si __builtin_arm_wminsw (v2si, v2si)
6218 v8qi __builtin_arm_wminub (v8qi, v8qi)
6219 v4hi __builtin_arm_wminuh (v4hi, v4hi)
6220 v2si __builtin_arm_wminuw (v2si, v2si)
6221 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
6222 v4hi __builtin_arm_wmulul (v4hi, v4hi)
6223 v4hi __builtin_arm_wmulum (v4hi, v4hi)
6224 long long __builtin_arm_wor (long long, long long)
6225 v2si __builtin_arm_wpackdss (long long, long long)
6226 v2si __builtin_arm_wpackdus (long long, long long)
6227 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
6228 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
6229 v4hi __builtin_arm_wpackwss (v2si, v2si)
6230 v4hi __builtin_arm_wpackwus (v2si, v2si)
6231 long long __builtin_arm_wrord (long long, long long)
6232 long long __builtin_arm_wrordi (long long, int)
6233 v4hi __builtin_arm_wrorh (v4hi, long long)
6234 v4hi __builtin_arm_wrorhi (v4hi, int)
6235 v2si __builtin_arm_wrorw (v2si, long long)
6236 v2si __builtin_arm_wrorwi (v2si, int)
6237 v2si __builtin_arm_wsadb (v8qi, v8qi)
6238 v2si __builtin_arm_wsadbz (v8qi, v8qi)
6239 v2si __builtin_arm_wsadh (v4hi, v4hi)
6240 v2si __builtin_arm_wsadhz (v4hi, v4hi)
6241 v4hi __builtin_arm_wshufh (v4hi, int)
6242 long long __builtin_arm_wslld (long long, long long)
6243 long long __builtin_arm_wslldi (long long, int)
6244 v4hi __builtin_arm_wsllh (v4hi, long long)
6245 v4hi __builtin_arm_wsllhi (v4hi, int)
6246 v2si __builtin_arm_wsllw (v2si, long long)
6247 v2si __builtin_arm_wsllwi (v2si, int)
6248 long long __builtin_arm_wsrad (long long, long long)
6249 long long __builtin_arm_wsradi (long long, int)
6250 v4hi __builtin_arm_wsrah (v4hi, long long)
6251 v4hi __builtin_arm_wsrahi (v4hi, int)
6252 v2si __builtin_arm_wsraw (v2si, long long)
6253 v2si __builtin_arm_wsrawi (v2si, int)
6254 long long __builtin_arm_wsrld (long long, long long)
6255 long long __builtin_arm_wsrldi (long long, int)
6256 v4hi __builtin_arm_wsrlh (v4hi, long long)
6257 v4hi __builtin_arm_wsrlhi (v4hi, int)
6258 v2si __builtin_arm_wsrlw (v2si, long long)
6259 v2si __builtin_arm_wsrlwi (v2si, int)
6260 v8qi __builtin_arm_wsubb (v8qi, v8qi)
6261 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
6262 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
6263 v4hi __builtin_arm_wsubh (v4hi, v4hi)
6264 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
6265 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
6266 v2si __builtin_arm_wsubw (v2si, v2si)
6267 v2si __builtin_arm_wsubwss (v2si, v2si)
6268 v2si __builtin_arm_wsubwus (v2si, v2si)
6269 v4hi __builtin_arm_wunpckehsb (v8qi)
6270 v2si __builtin_arm_wunpckehsh (v4hi)
6271 long long __builtin_arm_wunpckehsw (v2si)
6272 v4hi __builtin_arm_wunpckehub (v8qi)
6273 v2si __builtin_arm_wunpckehuh (v4hi)
6274 long long __builtin_arm_wunpckehuw (v2si)
6275 v4hi __builtin_arm_wunpckelsb (v8qi)
6276 v2si __builtin_arm_wunpckelsh (v4hi)
6277 long long __builtin_arm_wunpckelsw (v2si)
6278 v4hi __builtin_arm_wunpckelub (v8qi)
6279 v2si __builtin_arm_wunpckeluh (v4hi)
6280 long long __builtin_arm_wunpckeluw (v2si)
6281 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
6282 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
6283 v2si __builtin_arm_wunpckihw (v2si, v2si)
6284 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
6285 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
6286 v2si __builtin_arm_wunpckilw (v2si, v2si)
6287 long long __builtin_arm_wxor (long long, long long)
6288 long long __builtin_arm_wzero ()
6291 @node Blackfin Built-in Functions
6292 @subsection Blackfin Built-in Functions
6294 Currently, there are two Blackfin-specific built-in functions. These are
6295 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
6296 using inline assembly; by using these built-in functions the compiler can
6297 automatically add workarounds for hardware errata involving these
6298 instructions. These functions are named as follows:
6301 void __builtin_bfin_csync (void)
6302 void __builtin_bfin_ssync (void)
6305 @node FR-V Built-in Functions
6306 @subsection FR-V Built-in Functions
6308 GCC provides many FR-V-specific built-in functions. In general,
6309 these functions are intended to be compatible with those described
6310 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
6311 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
6312 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
6313 pointer rather than by value.
6315 Most of the functions are named after specific FR-V instructions.
6316 Such functions are said to be ``directly mapped'' and are summarized
6317 here in tabular form.
6321 * Directly-mapped Integer Functions::
6322 * Directly-mapped Media Functions::
6323 * Raw read/write Functions::
6324 * Other Built-in Functions::
6327 @node Argument Types
6328 @subsubsection Argument Types
6330 The arguments to the built-in functions can be divided into three groups:
6331 register numbers, compile-time constants and run-time values. In order
6332 to make this classification clear at a glance, the arguments and return
6333 values are given the following pseudo types:
6335 @multitable @columnfractions .20 .30 .15 .35
6336 @item Pseudo type @tab Real C type @tab Constant? @tab Description
6337 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
6338 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
6339 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
6340 @item @code{uw2} @tab @code{unsigned long long} @tab No
6341 @tab an unsigned doubleword
6342 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
6343 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
6344 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
6345 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
6348 These pseudo types are not defined by GCC, they are simply a notational
6349 convenience used in this manual.
6351 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
6352 and @code{sw2} are evaluated at run time. They correspond to
6353 register operands in the underlying FR-V instructions.
6355 @code{const} arguments represent immediate operands in the underlying
6356 FR-V instructions. They must be compile-time constants.
6358 @code{acc} arguments are evaluated at compile time and specify the number
6359 of an accumulator register. For example, an @code{acc} argument of 2
6360 will select the ACC2 register.
6362 @code{iacc} arguments are similar to @code{acc} arguments but specify the
6363 number of an IACC register. See @pxref{Other Built-in Functions}
6366 @node Directly-mapped Integer Functions
6367 @subsubsection Directly-mapped Integer Functions
6369 The functions listed below map directly to FR-V I-type instructions.
6371 @multitable @columnfractions .45 .32 .23
6372 @item Function prototype @tab Example usage @tab Assembly output
6373 @item @code{sw1 __ADDSS (sw1, sw1)}
6374 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
6375 @tab @code{ADDSS @var{a},@var{b},@var{c}}
6376 @item @code{sw1 __SCAN (sw1, sw1)}
6377 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
6378 @tab @code{SCAN @var{a},@var{b},@var{c}}
6379 @item @code{sw1 __SCUTSS (sw1)}
6380 @tab @code{@var{b} = __SCUTSS (@var{a})}
6381 @tab @code{SCUTSS @var{a},@var{b}}
6382 @item @code{sw1 __SLASS (sw1, sw1)}
6383 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
6384 @tab @code{SLASS @var{a},@var{b},@var{c}}
6385 @item @code{void __SMASS (sw1, sw1)}
6386 @tab @code{__SMASS (@var{a}, @var{b})}
6387 @tab @code{SMASS @var{a},@var{b}}
6388 @item @code{void __SMSSS (sw1, sw1)}
6389 @tab @code{__SMSSS (@var{a}, @var{b})}
6390 @tab @code{SMSSS @var{a},@var{b}}
6391 @item @code{void __SMU (sw1, sw1)}
6392 @tab @code{__SMU (@var{a}, @var{b})}
6393 @tab @code{SMU @var{a},@var{b}}
6394 @item @code{sw2 __SMUL (sw1, sw1)}
6395 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
6396 @tab @code{SMUL @var{a},@var{b},@var{c}}
6397 @item @code{sw1 __SUBSS (sw1, sw1)}
6398 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
6399 @tab @code{SUBSS @var{a},@var{b},@var{c}}
6400 @item @code{uw2 __UMUL (uw1, uw1)}
6401 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
6402 @tab @code{UMUL @var{a},@var{b},@var{c}}
6405 @node Directly-mapped Media Functions
6406 @subsubsection Directly-mapped Media Functions
6408 The functions listed below map directly to FR-V M-type instructions.
6410 @multitable @columnfractions .45 .32 .23
6411 @item Function prototype @tab Example usage @tab Assembly output
6412 @item @code{uw1 __MABSHS (sw1)}
6413 @tab @code{@var{b} = __MABSHS (@var{a})}
6414 @tab @code{MABSHS @var{a},@var{b}}
6415 @item @code{void __MADDACCS (acc, acc)}
6416 @tab @code{__MADDACCS (@var{b}, @var{a})}
6417 @tab @code{MADDACCS @var{a},@var{b}}
6418 @item @code{sw1 __MADDHSS (sw1, sw1)}
6419 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
6420 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
6421 @item @code{uw1 __MADDHUS (uw1, uw1)}
6422 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
6423 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
6424 @item @code{uw1 __MAND (uw1, uw1)}
6425 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
6426 @tab @code{MAND @var{a},@var{b},@var{c}}
6427 @item @code{void __MASACCS (acc, acc)}
6428 @tab @code{__MASACCS (@var{b}, @var{a})}
6429 @tab @code{MASACCS @var{a},@var{b}}
6430 @item @code{uw1 __MAVEH (uw1, uw1)}
6431 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
6432 @tab @code{MAVEH @var{a},@var{b},@var{c}}
6433 @item @code{uw2 __MBTOH (uw1)}
6434 @tab @code{@var{b} = __MBTOH (@var{a})}
6435 @tab @code{MBTOH @var{a},@var{b}}
6436 @item @code{void __MBTOHE (uw1 *, uw1)}
6437 @tab @code{__MBTOHE (&@var{b}, @var{a})}
6438 @tab @code{MBTOHE @var{a},@var{b}}
6439 @item @code{void __MCLRACC (acc)}
6440 @tab @code{__MCLRACC (@var{a})}
6441 @tab @code{MCLRACC @var{a}}
6442 @item @code{void __MCLRACCA (void)}
6443 @tab @code{__MCLRACCA ()}
6444 @tab @code{MCLRACCA}
6445 @item @code{uw1 __Mcop1 (uw1, uw1)}
6446 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
6447 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
6448 @item @code{uw1 __Mcop2 (uw1, uw1)}
6449 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
6450 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
6451 @item @code{uw1 __MCPLHI (uw2, const)}
6452 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
6453 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
6454 @item @code{uw1 __MCPLI (uw2, const)}
6455 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
6456 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
6457 @item @code{void __MCPXIS (acc, sw1, sw1)}
6458 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
6459 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
6460 @item @code{void __MCPXIU (acc, uw1, uw1)}
6461 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
6462 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
6463 @item @code{void __MCPXRS (acc, sw1, sw1)}
6464 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
6465 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
6466 @item @code{void __MCPXRU (acc, uw1, uw1)}
6467 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
6468 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
6469 @item @code{uw1 __MCUT (acc, uw1)}
6470 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
6471 @tab @code{MCUT @var{a},@var{b},@var{c}}
6472 @item @code{uw1 __MCUTSS (acc, sw1)}
6473 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
6474 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
6475 @item @code{void __MDADDACCS (acc, acc)}
6476 @tab @code{__MDADDACCS (@var{b}, @var{a})}
6477 @tab @code{MDADDACCS @var{a},@var{b}}
6478 @item @code{void __MDASACCS (acc, acc)}
6479 @tab @code{__MDASACCS (@var{b}, @var{a})}
6480 @tab @code{MDASACCS @var{a},@var{b}}
6481 @item @code{uw2 __MDCUTSSI (acc, const)}
6482 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
6483 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
6484 @item @code{uw2 __MDPACKH (uw2, uw2)}
6485 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
6486 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
6487 @item @code{uw2 __MDROTLI (uw2, const)}
6488 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
6489 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
6490 @item @code{void __MDSUBACCS (acc, acc)}
6491 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
6492 @tab @code{MDSUBACCS @var{a},@var{b}}
6493 @item @code{void __MDUNPACKH (uw1 *, uw2)}
6494 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
6495 @tab @code{MDUNPACKH @var{a},@var{b}}
6496 @item @code{uw2 __MEXPDHD (uw1, const)}
6497 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
6498 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
6499 @item @code{uw1 __MEXPDHW (uw1, const)}
6500 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
6501 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
6502 @item @code{uw1 __MHDSETH (uw1, const)}
6503 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
6504 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
6505 @item @code{sw1 __MHDSETS (const)}
6506 @tab @code{@var{b} = __MHDSETS (@var{a})}
6507 @tab @code{MHDSETS #@var{a},@var{b}}
6508 @item @code{uw1 __MHSETHIH (uw1, const)}
6509 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
6510 @tab @code{MHSETHIH #@var{a},@var{b}}
6511 @item @code{sw1 __MHSETHIS (sw1, const)}
6512 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
6513 @tab @code{MHSETHIS #@var{a},@var{b}}
6514 @item @code{uw1 __MHSETLOH (uw1, const)}
6515 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
6516 @tab @code{MHSETLOH #@var{a},@var{b}}
6517 @item @code{sw1 __MHSETLOS (sw1, const)}
6518 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
6519 @tab @code{MHSETLOS #@var{a},@var{b}}
6520 @item @code{uw1 __MHTOB (uw2)}
6521 @tab @code{@var{b} = __MHTOB (@var{a})}
6522 @tab @code{MHTOB @var{a},@var{b}}
6523 @item @code{void __MMACHS (acc, sw1, sw1)}
6524 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
6525 @tab @code{MMACHS @var{a},@var{b},@var{c}}
6526 @item @code{void __MMACHU (acc, uw1, uw1)}
6527 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
6528 @tab @code{MMACHU @var{a},@var{b},@var{c}}
6529 @item @code{void __MMRDHS (acc, sw1, sw1)}
6530 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
6531 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
6532 @item @code{void __MMRDHU (acc, uw1, uw1)}
6533 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
6534 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
6535 @item @code{void __MMULHS (acc, sw1, sw1)}
6536 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
6537 @tab @code{MMULHS @var{a},@var{b},@var{c}}
6538 @item @code{void __MMULHU (acc, uw1, uw1)}
6539 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
6540 @tab @code{MMULHU @var{a},@var{b},@var{c}}
6541 @item @code{void __MMULXHS (acc, sw1, sw1)}
6542 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
6543 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
6544 @item @code{void __MMULXHU (acc, uw1, uw1)}
6545 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
6546 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
6547 @item @code{uw1 __MNOT (uw1)}
6548 @tab @code{@var{b} = __MNOT (@var{a})}
6549 @tab @code{MNOT @var{a},@var{b}}
6550 @item @code{uw1 __MOR (uw1, uw1)}
6551 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
6552 @tab @code{MOR @var{a},@var{b},@var{c}}
6553 @item @code{uw1 __MPACKH (uh, uh)}
6554 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
6555 @tab @code{MPACKH @var{a},@var{b},@var{c}}
6556 @item @code{sw2 __MQADDHSS (sw2, sw2)}
6557 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
6558 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
6559 @item @code{uw2 __MQADDHUS (uw2, uw2)}
6560 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
6561 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
6562 @item @code{void __MQCPXIS (acc, sw2, sw2)}
6563 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
6564 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
6565 @item @code{void __MQCPXIU (acc, uw2, uw2)}
6566 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
6567 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
6568 @item @code{void __MQCPXRS (acc, sw2, sw2)}
6569 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
6570 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
6571 @item @code{void __MQCPXRU (acc, uw2, uw2)}
6572 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
6573 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
6574 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
6575 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
6576 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
6577 @item @code{sw2 __MQLMTHS (sw2, sw2)}
6578 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
6579 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
6580 @item @code{void __MQMACHS (acc, sw2, sw2)}
6581 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
6582 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
6583 @item @code{void __MQMACHU (acc, uw2, uw2)}
6584 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
6585 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
6586 @item @code{void __MQMACXHS (acc, sw2, sw2)}
6587 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
6588 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
6589 @item @code{void __MQMULHS (acc, sw2, sw2)}
6590 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
6591 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
6592 @item @code{void __MQMULHU (acc, uw2, uw2)}
6593 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
6594 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
6595 @item @code{void __MQMULXHS (acc, sw2, sw2)}
6596 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
6597 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
6598 @item @code{void __MQMULXHU (acc, uw2, uw2)}
6599 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
6600 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
6601 @item @code{sw2 __MQSATHS (sw2, sw2)}
6602 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
6603 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
6604 @item @code{uw2 __MQSLLHI (uw2, int)}
6605 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
6606 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
6607 @item @code{sw2 __MQSRAHI (sw2, int)}
6608 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
6609 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
6610 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
6611 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
6612 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
6613 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
6614 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
6615 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
6616 @item @code{void __MQXMACHS (acc, sw2, sw2)}
6617 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
6618 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
6619 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
6620 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
6621 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
6622 @item @code{uw1 __MRDACC (acc)}
6623 @tab @code{@var{b} = __MRDACC (@var{a})}
6624 @tab @code{MRDACC @var{a},@var{b}}
6625 @item @code{uw1 __MRDACCG (acc)}
6626 @tab @code{@var{b} = __MRDACCG (@var{a})}
6627 @tab @code{MRDACCG @var{a},@var{b}}
6628 @item @code{uw1 __MROTLI (uw1, const)}
6629 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
6630 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
6631 @item @code{uw1 __MROTRI (uw1, const)}
6632 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
6633 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
6634 @item @code{sw1 __MSATHS (sw1, sw1)}
6635 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
6636 @tab @code{MSATHS @var{a},@var{b},@var{c}}
6637 @item @code{uw1 __MSATHU (uw1, uw1)}
6638 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
6639 @tab @code{MSATHU @var{a},@var{b},@var{c}}
6640 @item @code{uw1 __MSLLHI (uw1, const)}
6641 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
6642 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
6643 @item @code{sw1 __MSRAHI (sw1, const)}
6644 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
6645 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
6646 @item @code{uw1 __MSRLHI (uw1, const)}
6647 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
6648 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
6649 @item @code{void __MSUBACCS (acc, acc)}
6650 @tab @code{__MSUBACCS (@var{b}, @var{a})}
6651 @tab @code{MSUBACCS @var{a},@var{b}}
6652 @item @code{sw1 __MSUBHSS (sw1, sw1)}
6653 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
6654 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
6655 @item @code{uw1 __MSUBHUS (uw1, uw1)}
6656 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
6657 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
6658 @item @code{void __MTRAP (void)}
6659 @tab @code{__MTRAP ()}
6661 @item @code{uw2 __MUNPACKH (uw1)}
6662 @tab @code{@var{b} = __MUNPACKH (@var{a})}
6663 @tab @code{MUNPACKH @var{a},@var{b}}
6664 @item @code{uw1 __MWCUT (uw2, uw1)}
6665 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
6666 @tab @code{MWCUT @var{a},@var{b},@var{c}}
6667 @item @code{void __MWTACC (acc, uw1)}
6668 @tab @code{__MWTACC (@var{b}, @var{a})}
6669 @tab @code{MWTACC @var{a},@var{b}}
6670 @item @code{void __MWTACCG (acc, uw1)}
6671 @tab @code{__MWTACCG (@var{b}, @var{a})}
6672 @tab @code{MWTACCG @var{a},@var{b}}
6673 @item @code{uw1 __MXOR (uw1, uw1)}
6674 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
6675 @tab @code{MXOR @var{a},@var{b},@var{c}}
6678 @node Raw read/write Functions
6679 @subsubsection Raw read/write Functions
6681 This sections describes built-in functions related to read and write
6682 instructions to access memory. These functions generate
6683 @code{membar} instructions to flush the I/O load and stores where
6684 appropriate, as described in Fujitsu's manual described above.
6688 @item unsigned char __builtin_read8 (void *@var{data})
6689 @item unsigned short __builtin_read16 (void *@var{data})
6690 @item unsigned long __builtin_read32 (void *@var{data})
6691 @item unsigned long long __builtin_read64 (void *@var{data})
6693 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
6694 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
6695 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
6696 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
6699 @node Other Built-in Functions
6700 @subsubsection Other Built-in Functions
6702 This section describes built-in functions that are not named after
6703 a specific FR-V instruction.
6706 @item sw2 __IACCreadll (iacc @var{reg})
6707 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
6708 for future expansion and must be 0.
6710 @item sw1 __IACCreadl (iacc @var{reg})
6711 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
6712 Other values of @var{reg} are rejected as invalid.
6714 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
6715 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
6716 is reserved for future expansion and must be 0.
6718 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
6719 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
6720 is 1. Other values of @var{reg} are rejected as invalid.
6722 @item void __data_prefetch0 (const void *@var{x})
6723 Use the @code{dcpl} instruction to load the contents of address @var{x}
6724 into the data cache.
6726 @item void __data_prefetch (const void *@var{x})
6727 Use the @code{nldub} instruction to load the contents of address @var{x}
6728 into the data cache. The instruction will be issued in slot I1@.
6731 @node X86 Built-in Functions
6732 @subsection X86 Built-in Functions
6734 These built-in functions are available for the i386 and x86-64 family
6735 of computers, depending on the command-line switches used.
6737 Note that, if you specify command-line switches such as @option{-msse},
6738 the compiler could use the extended instruction sets even if the built-ins
6739 are not used explicitly in the program. For this reason, applications
6740 which perform runtime CPU detection must compile separate files for each
6741 supported architecture, using the appropriate flags. In particular,
6742 the file containing the CPU detection code should be compiled without
6745 The following machine modes are available for use with MMX built-in functions
6746 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
6747 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
6748 vector of eight 8-bit integers. Some of the built-in functions operate on
6749 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
6751 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
6752 of two 32-bit floating point values.
6754 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
6755 floating point values. Some instructions use a vector of four 32-bit
6756 integers, these use @code{V4SI}. Finally, some instructions operate on an
6757 entire vector register, interpreting it as a 128-bit integer, these use mode
6760 The following built-in functions are made available by @option{-mmmx}.
6761 All of them generate the machine instruction that is part of the name.
6764 v8qi __builtin_ia32_paddb (v8qi, v8qi)
6765 v4hi __builtin_ia32_paddw (v4hi, v4hi)
6766 v2si __builtin_ia32_paddd (v2si, v2si)
6767 v8qi __builtin_ia32_psubb (v8qi, v8qi)
6768 v4hi __builtin_ia32_psubw (v4hi, v4hi)
6769 v2si __builtin_ia32_psubd (v2si, v2si)
6770 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
6771 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
6772 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
6773 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
6774 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
6775 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
6776 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
6777 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
6778 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
6779 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
6780 di __builtin_ia32_pand (di, di)
6781 di __builtin_ia32_pandn (di,di)
6782 di __builtin_ia32_por (di, di)
6783 di __builtin_ia32_pxor (di, di)
6784 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
6785 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
6786 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
6787 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
6788 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
6789 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
6790 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
6791 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
6792 v2si __builtin_ia32_punpckhdq (v2si, v2si)
6793 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
6794 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
6795 v2si __builtin_ia32_punpckldq (v2si, v2si)
6796 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
6797 v4hi __builtin_ia32_packssdw (v2si, v2si)
6798 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
6801 The following built-in functions are made available either with
6802 @option{-msse}, or with a combination of @option{-m3dnow} and
6803 @option{-march=athlon}. All of them generate the machine
6804 instruction that is part of the name.
6807 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
6808 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
6809 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
6810 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
6811 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
6812 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
6813 v8qi __builtin_ia32_pminub (v8qi, v8qi)
6814 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
6815 int __builtin_ia32_pextrw (v4hi, int)
6816 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
6817 int __builtin_ia32_pmovmskb (v8qi)
6818 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
6819 void __builtin_ia32_movntq (di *, di)
6820 void __builtin_ia32_sfence (void)
6823 The following built-in functions are available when @option{-msse} is used.
6824 All of them generate the machine instruction that is part of the name.
6827 int __builtin_ia32_comieq (v4sf, v4sf)
6828 int __builtin_ia32_comineq (v4sf, v4sf)
6829 int __builtin_ia32_comilt (v4sf, v4sf)
6830 int __builtin_ia32_comile (v4sf, v4sf)
6831 int __builtin_ia32_comigt (v4sf, v4sf)
6832 int __builtin_ia32_comige (v4sf, v4sf)
6833 int __builtin_ia32_ucomieq (v4sf, v4sf)
6834 int __builtin_ia32_ucomineq (v4sf, v4sf)
6835 int __builtin_ia32_ucomilt (v4sf, v4sf)
6836 int __builtin_ia32_ucomile (v4sf, v4sf)
6837 int __builtin_ia32_ucomigt (v4sf, v4sf)
6838 int __builtin_ia32_ucomige (v4sf, v4sf)
6839 v4sf __builtin_ia32_addps (v4sf, v4sf)
6840 v4sf __builtin_ia32_subps (v4sf, v4sf)
6841 v4sf __builtin_ia32_mulps (v4sf, v4sf)
6842 v4sf __builtin_ia32_divps (v4sf, v4sf)
6843 v4sf __builtin_ia32_addss (v4sf, v4sf)
6844 v4sf __builtin_ia32_subss (v4sf, v4sf)
6845 v4sf __builtin_ia32_mulss (v4sf, v4sf)
6846 v4sf __builtin_ia32_divss (v4sf, v4sf)
6847 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
6848 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
6849 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
6850 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
6851 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
6852 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
6853 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
6854 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
6855 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
6856 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
6857 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
6858 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
6859 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
6860 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
6861 v4si __builtin_ia32_cmpless (v4sf, v4sf)
6862 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
6863 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
6864 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
6865 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
6866 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
6867 v4sf __builtin_ia32_maxps (v4sf, v4sf)
6868 v4sf __builtin_ia32_maxss (v4sf, v4sf)
6869 v4sf __builtin_ia32_minps (v4sf, v4sf)
6870 v4sf __builtin_ia32_minss (v4sf, v4sf)
6871 v4sf __builtin_ia32_andps (v4sf, v4sf)
6872 v4sf __builtin_ia32_andnps (v4sf, v4sf)
6873 v4sf __builtin_ia32_orps (v4sf, v4sf)
6874 v4sf __builtin_ia32_xorps (v4sf, v4sf)
6875 v4sf __builtin_ia32_movss (v4sf, v4sf)
6876 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
6877 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
6878 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
6879 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
6880 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
6881 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
6882 v2si __builtin_ia32_cvtps2pi (v4sf)
6883 int __builtin_ia32_cvtss2si (v4sf)
6884 v2si __builtin_ia32_cvttps2pi (v4sf)
6885 int __builtin_ia32_cvttss2si (v4sf)
6886 v4sf __builtin_ia32_rcpps (v4sf)
6887 v4sf __builtin_ia32_rsqrtps (v4sf)
6888 v4sf __builtin_ia32_sqrtps (v4sf)
6889 v4sf __builtin_ia32_rcpss (v4sf)
6890 v4sf __builtin_ia32_rsqrtss (v4sf)
6891 v4sf __builtin_ia32_sqrtss (v4sf)
6892 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
6893 void __builtin_ia32_movntps (float *, v4sf)
6894 int __builtin_ia32_movmskps (v4sf)
6897 The following built-in functions are available when @option{-msse} is used.
6900 @item v4sf __builtin_ia32_loadaps (float *)
6901 Generates the @code{movaps} machine instruction as a load from memory.
6902 @item void __builtin_ia32_storeaps (float *, v4sf)
6903 Generates the @code{movaps} machine instruction as a store to memory.
6904 @item v4sf __builtin_ia32_loadups (float *)
6905 Generates the @code{movups} machine instruction as a load from memory.
6906 @item void __builtin_ia32_storeups (float *, v4sf)
6907 Generates the @code{movups} machine instruction as a store to memory.
6908 @item v4sf __builtin_ia32_loadsss (float *)
6909 Generates the @code{movss} machine instruction as a load from memory.
6910 @item void __builtin_ia32_storess (float *, v4sf)
6911 Generates the @code{movss} machine instruction as a store to memory.
6912 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
6913 Generates the @code{movhps} machine instruction as a load from memory.
6914 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
6915 Generates the @code{movlps} machine instruction as a load from memory
6916 @item void __builtin_ia32_storehps (v4sf, v2si *)
6917 Generates the @code{movhps} machine instruction as a store to memory.
6918 @item void __builtin_ia32_storelps (v4sf, v2si *)
6919 Generates the @code{movlps} machine instruction as a store to memory.
6922 The following built-in functions are available when @option{-msse3} is used.
6923 All of them generate the machine instruction that is part of the name.
6926 v2df __builtin_ia32_addsubpd (v2df, v2df)
6927 v2df __builtin_ia32_addsubps (v2df, v2df)
6928 v2df __builtin_ia32_haddpd (v2df, v2df)
6929 v2df __builtin_ia32_haddps (v2df, v2df)
6930 v2df __builtin_ia32_hsubpd (v2df, v2df)
6931 v2df __builtin_ia32_hsubps (v2df, v2df)
6932 v16qi __builtin_ia32_lddqu (char const *)
6933 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
6934 v2df __builtin_ia32_movddup (v2df)
6935 v4sf __builtin_ia32_movshdup (v4sf)
6936 v4sf __builtin_ia32_movsldup (v4sf)
6937 void __builtin_ia32_mwait (unsigned int, unsigned int)
6940 The following built-in functions are available when @option{-msse3} is used.
6943 @item v2df __builtin_ia32_loadddup (double const *)
6944 Generates the @code{movddup} machine instruction as a load from memory.
6947 The following built-in functions are available when @option{-m3dnow} is used.
6948 All of them generate the machine instruction that is part of the name.
6951 void __builtin_ia32_femms (void)
6952 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
6953 v2si __builtin_ia32_pf2id (v2sf)
6954 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
6955 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
6956 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
6957 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
6958 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
6959 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
6960 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
6961 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
6962 v2sf __builtin_ia32_pfrcp (v2sf)
6963 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
6964 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
6965 v2sf __builtin_ia32_pfrsqrt (v2sf)
6966 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
6967 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
6968 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
6969 v2sf __builtin_ia32_pi2fd (v2si)
6970 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
6973 The following built-in functions are available when both @option{-m3dnow}
6974 and @option{-march=athlon} are used. All of them generate the machine
6975 instruction that is part of the name.
6978 v2si __builtin_ia32_pf2iw (v2sf)
6979 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
6980 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
6981 v2sf __builtin_ia32_pi2fw (v2si)
6982 v2sf __builtin_ia32_pswapdsf (v2sf)
6983 v2si __builtin_ia32_pswapdsi (v2si)
6986 @node MIPS DSP Built-in Functions
6987 @subsection MIPS DSP Built-in Functions
6989 The MIPS DSP Application-Specific Extension (ASE) includes new
6990 instructions that are designed to improve the performance of DSP and
6991 media applications. It provides instructions that operate on packed
6992 8-bit integer data, Q15 fractional data and Q31 fractional data.
6994 GCC supports MIPS DSP operations using both the generic
6995 vector extensions (@pxref{Vector Extensions}) and a collection of
6996 MIPS-specific built-in functions. Both kinds of support are
6997 enabled by the @option{-mdsp} command-line option.
6999 At present, GCC only provides support for operations on 32-bit
7000 vectors. The vector type associated with 8-bit integer data is
7001 usually called @code{v4i8} and the vector type associated with Q15 is
7002 usually called @code{v2q15}. They can be defined in C as follows:
7005 typedef char v4i8 __attribute__ ((vector_size(4)));
7006 typedef short v2q15 __attribute__ ((vector_size(4)));
7009 @code{v4i8} and @code{v2q15} values are initialized in the same way as
7010 aggregates. For example:
7013 v4i8 a = @{1, 2, 3, 4@};
7015 b = (v4i8) @{5, 6, 7, 8@};
7017 v2q15 c = @{0x0fcb, 0x3a75@};
7019 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
7022 @emph{Note:} The CPU's endianness determines the order in which values
7023 are packed. On little-endian targets, the first value is the least
7024 significant and the last value is the most significant. The opposite
7025 order applies to big-endian targets. For example, the code above will
7026 set the lowest byte of @code{a} to @code{1} on little-endian targets
7027 and @code{4} on big-endian targets.
7029 @emph{Note:} Q15 and Q31 values must be initialized with their integer
7030 representation. As shown in this example, the integer representation
7031 of a Q15 value can be obtained by multiplying the fractional value by
7032 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
7035 The table below lists the @code{v4i8} and @code{v2q15} operations for which
7036 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
7037 and @code{c} and @code{d} are @code{v2q15} values.
7039 @multitable @columnfractions .50 .50
7040 @item C code @tab MIPS instruction
7041 @item @code{a + b} @tab @code{addu.qb}
7042 @item @code{c + d} @tab @code{addq.ph}
7043 @item @code{a - b} @tab @code{subu.qb}
7044 @item @code{c - d} @tab @code{subq.ph}
7047 It is easier to describe the DSP built-in functions if we first define
7048 the following types:
7053 typedef long long a64;
7056 @code{q31} and @code{i32} are actually the same as @code{int}, but we
7057 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
7058 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
7059 @code{long long}, but we use @code{a64} to indicate values that will
7060 be placed in one of the four DSP accumulators (@code{$ac0},
7061 @code{$ac1}, @code{$ac2} or @code{$ac3}).
7063 Also, some built-in functions prefer or require immediate numbers as
7064 parameters, because the corresponding DSP instructions accept both immediate
7065 numbers and register operands, or accept immediate numbers only. The
7066 immediate parameters are listed as follows.
7074 imm_n32_31: -32 to 31.
7075 imm_n512_511: -512 to 511.
7078 The following built-in functions map directly to a particular MIPS DSP
7079 instruction. Please refer to the architecture specification
7080 for details on what each instruction does.
7083 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
7084 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
7085 q31 __builtin_mips_addq_s_w (q31, q31)
7086 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
7087 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
7088 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
7089 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
7090 q31 __builtin_mips_subq_s_w (q31, q31)
7091 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
7092 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
7093 i32 __builtin_mips_addsc (i32, i32)
7094 i32 __builtin_mips_addwc (i32, i32)
7095 i32 __builtin_mips_modsub (i32, i32)
7096 i32 __builtin_mips_raddu_w_qb (v4i8)
7097 v2q15 __builtin_mips_absq_s_ph (v2q15)
7098 q31 __builtin_mips_absq_s_w (q31)
7099 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
7100 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
7101 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
7102 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
7103 q31 __builtin_mips_preceq_w_phl (v2q15)
7104 q31 __builtin_mips_preceq_w_phr (v2q15)
7105 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
7106 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
7107 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
7108 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
7109 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
7110 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
7111 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
7112 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
7113 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
7114 v4i8 __builtin_mips_shll_qb (v4i8, i32)
7115 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
7116 v2q15 __builtin_mips_shll_ph (v2q15, i32)
7117 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
7118 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
7119 q31 __builtin_mips_shll_s_w (q31, imm0_31)
7120 q31 __builtin_mips_shll_s_w (q31, i32)
7121 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
7122 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
7123 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
7124 v2q15 __builtin_mips_shra_ph (v2q15, i32)
7125 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
7126 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
7127 q31 __builtin_mips_shra_r_w (q31, imm0_31)
7128 q31 __builtin_mips_shra_r_w (q31, i32)
7129 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
7130 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
7131 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
7132 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
7133 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
7134 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
7135 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
7136 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
7137 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
7138 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
7139 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
7140 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
7141 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
7142 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
7143 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
7144 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
7145 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
7146 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
7147 i32 __builtin_mips_bitrev (i32)
7148 i32 __builtin_mips_insv (i32, i32)
7149 v4i8 __builtin_mips_repl_qb (imm0_255)
7150 v4i8 __builtin_mips_repl_qb (i32)
7151 v2q15 __builtin_mips_repl_ph (imm_n512_511)
7152 v2q15 __builtin_mips_repl_ph (i32)
7153 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
7154 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
7155 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
7156 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
7157 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
7158 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
7159 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
7160 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
7161 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
7162 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
7163 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
7164 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
7165 i32 __builtin_mips_extr_w (a64, imm0_31)
7166 i32 __builtin_mips_extr_w (a64, i32)
7167 i32 __builtin_mips_extr_r_w (a64, imm0_31)
7168 i32 __builtin_mips_extr_s_h (a64, i32)
7169 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
7170 i32 __builtin_mips_extr_rs_w (a64, i32)
7171 i32 __builtin_mips_extr_s_h (a64, imm0_31)
7172 i32 __builtin_mips_extr_r_w (a64, i32)
7173 i32 __builtin_mips_extp (a64, imm0_31)
7174 i32 __builtin_mips_extp (a64, i32)
7175 i32 __builtin_mips_extpdp (a64, imm0_31)
7176 i32 __builtin_mips_extpdp (a64, i32)
7177 a64 __builtin_mips_shilo (a64, imm_n32_31)
7178 a64 __builtin_mips_shilo (a64, i32)
7179 a64 __builtin_mips_mthlip (a64, i32)
7180 void __builtin_mips_wrdsp (i32, imm0_63)
7181 i32 __builtin_mips_rddsp (imm0_63)
7182 i32 __builtin_mips_lbux (void *, i32)
7183 i32 __builtin_mips_lhx (void *, i32)
7184 i32 __builtin_mips_lwx (void *, i32)
7185 i32 __builtin_mips_bposge32 (void)
7188 @node MIPS Paired-Single Support
7189 @subsection MIPS Paired-Single Support
7191 The MIPS64 architecture includes a number of instructions that
7192 operate on pairs of single-precision floating-point values.
7193 Each pair is packed into a 64-bit floating-point register,
7194 with one element being designated the ``upper half'' and
7195 the other being designated the ``lower half''.
7197 GCC supports paired-single operations using both the generic
7198 vector extensions (@pxref{Vector Extensions}) and a collection of
7199 MIPS-specific built-in functions. Both kinds of support are
7200 enabled by the @option{-mpaired-single} command-line option.
7202 The vector type associated with paired-single values is usually
7203 called @code{v2sf}. It can be defined in C as follows:
7206 typedef float v2sf __attribute__ ((vector_size (8)));
7209 @code{v2sf} values are initialized in the same way as aggregates.
7213 v2sf a = @{1.5, 9.1@};
7216 b = (v2sf) @{e, f@};
7219 @emph{Note:} The CPU's endianness determines which value is stored in
7220 the upper half of a register and which value is stored in the lower half.
7221 On little-endian targets, the first value is the lower one and the second
7222 value is the upper one. The opposite order applies to big-endian targets.
7223 For example, the code above will set the lower half of @code{a} to
7224 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
7227 * Paired-Single Arithmetic::
7228 * Paired-Single Built-in Functions::
7229 * MIPS-3D Built-in Functions::
7232 @node Paired-Single Arithmetic
7233 @subsubsection Paired-Single Arithmetic
7235 The table below lists the @code{v2sf} operations for which hardware
7236 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
7237 values and @code{x} is an integral value.
7239 @multitable @columnfractions .50 .50
7240 @item C code @tab MIPS instruction
7241 @item @code{a + b} @tab @code{add.ps}
7242 @item @code{a - b} @tab @code{sub.ps}
7243 @item @code{-a} @tab @code{neg.ps}
7244 @item @code{a * b} @tab @code{mul.ps}
7245 @item @code{a * b + c} @tab @code{madd.ps}
7246 @item @code{a * b - c} @tab @code{msub.ps}
7247 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
7248 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
7249 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
7252 Note that the multiply-accumulate instructions can be disabled
7253 using the command-line option @code{-mno-fused-madd}.
7255 @node Paired-Single Built-in Functions
7256 @subsubsection Paired-Single Built-in Functions
7258 The following paired-single functions map directly to a particular
7259 MIPS instruction. Please refer to the architecture specification
7260 for details on what each instruction does.
7263 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
7264 Pair lower lower (@code{pll.ps}).
7266 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
7267 Pair upper lower (@code{pul.ps}).
7269 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
7270 Pair lower upper (@code{plu.ps}).
7272 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
7273 Pair upper upper (@code{puu.ps}).
7275 @item v2sf __builtin_mips_cvt_ps_s (float, float)
7276 Convert pair to paired single (@code{cvt.ps.s}).
7278 @item float __builtin_mips_cvt_s_pl (v2sf)
7279 Convert pair lower to single (@code{cvt.s.pl}).
7281 @item float __builtin_mips_cvt_s_pu (v2sf)
7282 Convert pair upper to single (@code{cvt.s.pu}).
7284 @item v2sf __builtin_mips_abs_ps (v2sf)
7285 Absolute value (@code{abs.ps}).
7287 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
7288 Align variable (@code{alnv.ps}).
7290 @emph{Note:} The value of the third parameter must be 0 or 4
7291 modulo 8, otherwise the result will be unpredictable. Please read the
7292 instruction description for details.
7295 The following multi-instruction functions are also available.
7296 In each case, @var{cond} can be any of the 16 floating-point conditions:
7297 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
7298 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
7299 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
7302 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7303 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7304 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
7305 @code{movt.ps}/@code{movf.ps}).
7307 The @code{movt} functions return the value @var{x} computed by:
7310 c.@var{cond}.ps @var{cc},@var{a},@var{b}
7311 mov.ps @var{x},@var{c}
7312 movt.ps @var{x},@var{d},@var{cc}
7315 The @code{movf} functions are similar but use @code{movf.ps} instead
7318 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7319 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7320 Comparison of two paired-single values (@code{c.@var{cond}.ps},
7321 @code{bc1t}/@code{bc1f}).
7323 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
7324 and return either the upper or lower half of the result. For example:
7328 if (__builtin_mips_upper_c_eq_ps (a, b))
7329 upper_halves_are_equal ();
7331 upper_halves_are_unequal ();
7333 if (__builtin_mips_lower_c_eq_ps (a, b))
7334 lower_halves_are_equal ();
7336 lower_halves_are_unequal ();
7340 @node MIPS-3D Built-in Functions
7341 @subsubsection MIPS-3D Built-in Functions
7343 The MIPS-3D Application-Specific Extension (ASE) includes additional
7344 paired-single instructions that are designed to improve the performance
7345 of 3D graphics operations. Support for these instructions is controlled
7346 by the @option{-mips3d} command-line option.
7348 The functions listed below map directly to a particular MIPS-3D
7349 instruction. Please refer to the architecture specification for
7350 more details on what each instruction does.
7353 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
7354 Reduction add (@code{addr.ps}).
7356 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
7357 Reduction multiply (@code{mulr.ps}).
7359 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
7360 Convert paired single to paired word (@code{cvt.pw.ps}).
7362 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
7363 Convert paired word to paired single (@code{cvt.ps.pw}).
7365 @item float __builtin_mips_recip1_s (float)
7366 @itemx double __builtin_mips_recip1_d (double)
7367 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
7368 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
7370 @item float __builtin_mips_recip2_s (float, float)
7371 @itemx double __builtin_mips_recip2_d (double, double)
7372 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
7373 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
7375 @item float __builtin_mips_rsqrt1_s (float)
7376 @itemx double __builtin_mips_rsqrt1_d (double)
7377 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
7378 Reduced precision reciprocal square root (sequence step 1)
7379 (@code{rsqrt1.@var{fmt}}).
7381 @item float __builtin_mips_rsqrt2_s (float, float)
7382 @itemx double __builtin_mips_rsqrt2_d (double, double)
7383 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
7384 Reduced precision reciprocal square root (sequence step 2)
7385 (@code{rsqrt2.@var{fmt}}).
7388 The following multi-instruction functions are also available.
7389 In each case, @var{cond} can be any of the 16 floating-point conditions:
7390 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
7391 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
7392 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
7395 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
7396 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
7397 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
7398 @code{bc1t}/@code{bc1f}).
7400 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
7401 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
7406 if (__builtin_mips_cabs_eq_s (a, b))
7412 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7413 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7414 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
7415 @code{bc1t}/@code{bc1f}).
7417 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
7418 and return either the upper or lower half of the result. For example:
7422 if (__builtin_mips_upper_cabs_eq_ps (a, b))
7423 upper_halves_are_equal ();
7425 upper_halves_are_unequal ();
7427 if (__builtin_mips_lower_cabs_eq_ps (a, b))
7428 lower_halves_are_equal ();
7430 lower_halves_are_unequal ();
7433 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7434 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7435 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
7436 @code{movt.ps}/@code{movf.ps}).
7438 The @code{movt} functions return the value @var{x} computed by:
7441 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
7442 mov.ps @var{x},@var{c}
7443 movt.ps @var{x},@var{d},@var{cc}
7446 The @code{movf} functions are similar but use @code{movf.ps} instead
7449 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7450 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7451 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7452 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7453 Comparison of two paired-single values
7454 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
7455 @code{bc1any2t}/@code{bc1any2f}).
7457 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
7458 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
7459 result is true and the @code{all} forms return true if both results are true.
7464 if (__builtin_mips_any_c_eq_ps (a, b))
7469 if (__builtin_mips_all_c_eq_ps (a, b))
7475 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7476 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7477 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7478 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7479 Comparison of four paired-single values
7480 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
7481 @code{bc1any4t}/@code{bc1any4f}).
7483 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
7484 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
7485 The @code{any} forms return true if any of the four results are true
7486 and the @code{all} forms return true if all four results are true.
7491 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
7496 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
7503 @node PowerPC AltiVec Built-in Functions
7504 @subsection PowerPC AltiVec Built-in Functions
7506 GCC provides an interface for the PowerPC family of processors to access
7507 the AltiVec operations described in Motorola's AltiVec Programming
7508 Interface Manual. The interface is made available by including
7509 @code{<altivec.h>} and using @option{-maltivec} and
7510 @option{-mabi=altivec}. The interface supports the following vector
7514 vector unsigned char
7518 vector unsigned short
7529 GCC's implementation of the high-level language interface available from
7530 C and C++ code differs from Motorola's documentation in several ways.
7535 A vector constant is a list of constant expressions within curly braces.
7538 A vector initializer requires no cast if the vector constant is of the
7539 same type as the variable it is initializing.
7542 If @code{signed} or @code{unsigned} is omitted, the signedness of the
7543 vector type is the default signedness of the base type. The default
7544 varies depending on the operating system, so a portable program should
7545 always specify the signedness.
7548 Compiling with @option{-maltivec} adds keywords @code{__vector},
7549 @code{__pixel}, and @code{__bool}. Macros @option{vector},
7550 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
7554 GCC allows using a @code{typedef} name as the type specifier for a
7558 For C, overloaded functions are implemented with macros so the following
7562 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
7565 Since @code{vec_add} is a macro, the vector constant in the example
7566 is treated as four separate arguments. Wrap the entire argument in
7567 parentheses for this to work.
7570 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
7571 Internally, GCC uses built-in functions to achieve the functionality in
7572 the aforementioned header file, but they are not supported and are
7573 subject to change without notice.
7575 The following interfaces are supported for the generic and specific
7576 AltiVec operations and the AltiVec predicates. In cases where there
7577 is a direct mapping between generic and specific operations, only the
7578 generic names are shown here, although the specific operations can also
7581 Arguments that are documented as @code{const int} require literal
7582 integral values within the range required for that operation.
7585 vector signed char vec_abs (vector signed char);
7586 vector signed short vec_abs (vector signed short);
7587 vector signed int vec_abs (vector signed int);
7588 vector float vec_abs (vector float);
7590 vector signed char vec_abss (vector signed char);
7591 vector signed short vec_abss (vector signed short);
7592 vector signed int vec_abss (vector signed int);
7594 vector signed char vec_add (vector bool char, vector signed char);
7595 vector signed char vec_add (vector signed char, vector bool char);
7596 vector signed char vec_add (vector signed char, vector signed char);
7597 vector unsigned char vec_add (vector bool char, vector unsigned char);
7598 vector unsigned char vec_add (vector unsigned char, vector bool char);
7599 vector unsigned char vec_add (vector unsigned char,
7600 vector unsigned char);
7601 vector signed short vec_add (vector bool short, vector signed short);
7602 vector signed short vec_add (vector signed short, vector bool short);
7603 vector signed short vec_add (vector signed short, vector signed short);
7604 vector unsigned short vec_add (vector bool short,
7605 vector unsigned short);
7606 vector unsigned short vec_add (vector unsigned short,
7608 vector unsigned short vec_add (vector unsigned short,
7609 vector unsigned short);
7610 vector signed int vec_add (vector bool int, vector signed int);
7611 vector signed int vec_add (vector signed int, vector bool int);
7612 vector signed int vec_add (vector signed int, vector signed int);
7613 vector unsigned int vec_add (vector bool int, vector unsigned int);
7614 vector unsigned int vec_add (vector unsigned int, vector bool int);
7615 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
7616 vector float vec_add (vector float, vector float);
7618 vector float vec_vaddfp (vector float, vector float);
7620 vector signed int vec_vadduwm (vector bool int, vector signed int);
7621 vector signed int vec_vadduwm (vector signed int, vector bool int);
7622 vector signed int vec_vadduwm (vector signed int, vector signed int);
7623 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
7624 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
7625 vector unsigned int vec_vadduwm (vector unsigned int,
7626 vector unsigned int);
7628 vector signed short vec_vadduhm (vector bool short,
7629 vector signed short);
7630 vector signed short vec_vadduhm (vector signed short,
7632 vector signed short vec_vadduhm (vector signed short,
7633 vector signed short);
7634 vector unsigned short vec_vadduhm (vector bool short,
7635 vector unsigned short);
7636 vector unsigned short vec_vadduhm (vector unsigned short,
7638 vector unsigned short vec_vadduhm (vector unsigned short,
7639 vector unsigned short);
7641 vector signed char vec_vaddubm (vector bool char, vector signed char);
7642 vector signed char vec_vaddubm (vector signed char, vector bool char);
7643 vector signed char vec_vaddubm (vector signed char, vector signed char);
7644 vector unsigned char vec_vaddubm (vector bool char,
7645 vector unsigned char);
7646 vector unsigned char vec_vaddubm (vector unsigned char,
7648 vector unsigned char vec_vaddubm (vector unsigned char,
7649 vector unsigned char);
7651 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
7653 vector unsigned char vec_adds (vector bool char, vector unsigned char);
7654 vector unsigned char vec_adds (vector unsigned char, vector bool char);
7655 vector unsigned char vec_adds (vector unsigned char,
7656 vector unsigned char);
7657 vector signed char vec_adds (vector bool char, vector signed char);
7658 vector signed char vec_adds (vector signed char, vector bool char);
7659 vector signed char vec_adds (vector signed char, vector signed char);
7660 vector unsigned short vec_adds (vector bool short,
7661 vector unsigned short);
7662 vector unsigned short vec_adds (vector unsigned short,
7664 vector unsigned short vec_adds (vector unsigned short,
7665 vector unsigned short);
7666 vector signed short vec_adds (vector bool short, vector signed short);
7667 vector signed short vec_adds (vector signed short, vector bool short);
7668 vector signed short vec_adds (vector signed short, vector signed short);
7669 vector unsigned int vec_adds (vector bool int, vector unsigned int);
7670 vector unsigned int vec_adds (vector unsigned int, vector bool int);
7671 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
7672 vector signed int vec_adds (vector bool int, vector signed int);
7673 vector signed int vec_adds (vector signed int, vector bool int);
7674 vector signed int vec_adds (vector signed int, vector signed int);
7676 vector signed int vec_vaddsws (vector bool int, vector signed int);
7677 vector signed int vec_vaddsws (vector signed int, vector bool int);
7678 vector signed int vec_vaddsws (vector signed int, vector signed int);
7680 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
7681 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
7682 vector unsigned int vec_vadduws (vector unsigned int,
7683 vector unsigned int);
7685 vector signed short vec_vaddshs (vector bool short,
7686 vector signed short);
7687 vector signed short vec_vaddshs (vector signed short,
7689 vector signed short vec_vaddshs (vector signed short,
7690 vector signed short);
7692 vector unsigned short vec_vadduhs (vector bool short,
7693 vector unsigned short);
7694 vector unsigned short vec_vadduhs (vector unsigned short,
7696 vector unsigned short vec_vadduhs (vector unsigned short,
7697 vector unsigned short);
7699 vector signed char vec_vaddsbs (vector bool char, vector signed char);
7700 vector signed char vec_vaddsbs (vector signed char, vector bool char);
7701 vector signed char vec_vaddsbs (vector signed char, vector signed char);
7703 vector unsigned char vec_vaddubs (vector bool char,
7704 vector unsigned char);
7705 vector unsigned char vec_vaddubs (vector unsigned char,
7707 vector unsigned char vec_vaddubs (vector unsigned char,
7708 vector unsigned char);
7710 vector float vec_and (vector float, vector float);
7711 vector float vec_and (vector float, vector bool int);
7712 vector float vec_and (vector bool int, vector float);
7713 vector bool int vec_and (vector bool int, vector bool int);
7714 vector signed int vec_and (vector bool int, vector signed int);
7715 vector signed int vec_and (vector signed int, vector bool int);
7716 vector signed int vec_and (vector signed int, vector signed int);
7717 vector unsigned int vec_and (vector bool int, vector unsigned int);
7718 vector unsigned int vec_and (vector unsigned int, vector bool int);
7719 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
7720 vector bool short vec_and (vector bool short, vector bool short);
7721 vector signed short vec_and (vector bool short, vector signed short);
7722 vector signed short vec_and (vector signed short, vector bool short);
7723 vector signed short vec_and (vector signed short, vector signed short);
7724 vector unsigned short vec_and (vector bool short,
7725 vector unsigned short);
7726 vector unsigned short vec_and (vector unsigned short,
7728 vector unsigned short vec_and (vector unsigned short,
7729 vector unsigned short);
7730 vector signed char vec_and (vector bool char, vector signed char);
7731 vector bool char vec_and (vector bool char, vector bool char);
7732 vector signed char vec_and (vector signed char, vector bool char);
7733 vector signed char vec_and (vector signed char, vector signed char);
7734 vector unsigned char vec_and (vector bool char, vector unsigned char);
7735 vector unsigned char vec_and (vector unsigned char, vector bool char);
7736 vector unsigned char vec_and (vector unsigned char,
7737 vector unsigned char);
7739 vector float vec_andc (vector float, vector float);
7740 vector float vec_andc (vector float, vector bool int);
7741 vector float vec_andc (vector bool int, vector float);
7742 vector bool int vec_andc (vector bool int, vector bool int);
7743 vector signed int vec_andc (vector bool int, vector signed int);
7744 vector signed int vec_andc (vector signed int, vector bool int);
7745 vector signed int vec_andc (vector signed int, vector signed int);
7746 vector unsigned int vec_andc (vector bool int, vector unsigned int);
7747 vector unsigned int vec_andc (vector unsigned int, vector bool int);
7748 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
7749 vector bool short vec_andc (vector bool short, vector bool short);
7750 vector signed short vec_andc (vector bool short, vector signed short);
7751 vector signed short vec_andc (vector signed short, vector bool short);
7752 vector signed short vec_andc (vector signed short, vector signed short);
7753 vector unsigned short vec_andc (vector bool short,
7754 vector unsigned short);
7755 vector unsigned short vec_andc (vector unsigned short,
7757 vector unsigned short vec_andc (vector unsigned short,
7758 vector unsigned short);
7759 vector signed char vec_andc (vector bool char, vector signed char);
7760 vector bool char vec_andc (vector bool char, vector bool char);
7761 vector signed char vec_andc (vector signed char, vector bool char);
7762 vector signed char vec_andc (vector signed char, vector signed char);
7763 vector unsigned char vec_andc (vector bool char, vector unsigned char);
7764 vector unsigned char vec_andc (vector unsigned char, vector bool char);
7765 vector unsigned char vec_andc (vector unsigned char,
7766 vector unsigned char);
7768 vector unsigned char vec_avg (vector unsigned char,
7769 vector unsigned char);
7770 vector signed char vec_avg (vector signed char, vector signed char);
7771 vector unsigned short vec_avg (vector unsigned short,
7772 vector unsigned short);
7773 vector signed short vec_avg (vector signed short, vector signed short);
7774 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
7775 vector signed int vec_avg (vector signed int, vector signed int);
7777 vector signed int vec_vavgsw (vector signed int, vector signed int);
7779 vector unsigned int vec_vavguw (vector unsigned int,
7780 vector unsigned int);
7782 vector signed short vec_vavgsh (vector signed short,
7783 vector signed short);
7785 vector unsigned short vec_vavguh (vector unsigned short,
7786 vector unsigned short);
7788 vector signed char vec_vavgsb (vector signed char, vector signed char);
7790 vector unsigned char vec_vavgub (vector unsigned char,
7791 vector unsigned char);
7793 vector float vec_ceil (vector float);
7795 vector signed int vec_cmpb (vector float, vector float);
7797 vector bool char vec_cmpeq (vector signed char, vector signed char);
7798 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
7799 vector bool short vec_cmpeq (vector signed short, vector signed short);
7800 vector bool short vec_cmpeq (vector unsigned short,
7801 vector unsigned short);
7802 vector bool int vec_cmpeq (vector signed int, vector signed int);
7803 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
7804 vector bool int vec_cmpeq (vector float, vector float);
7806 vector bool int vec_vcmpeqfp (vector float, vector float);
7808 vector bool int vec_vcmpequw (vector signed int, vector signed int);
7809 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
7811 vector bool short vec_vcmpequh (vector signed short,
7812 vector signed short);
7813 vector bool short vec_vcmpequh (vector unsigned short,
7814 vector unsigned short);
7816 vector bool char vec_vcmpequb (vector signed char, vector signed char);
7817 vector bool char vec_vcmpequb (vector unsigned char,
7818 vector unsigned char);
7820 vector bool int vec_cmpge (vector float, vector float);
7822 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
7823 vector bool char vec_cmpgt (vector signed char, vector signed char);
7824 vector bool short vec_cmpgt (vector unsigned short,
7825 vector unsigned short);
7826 vector bool short vec_cmpgt (vector signed short, vector signed short);
7827 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
7828 vector bool int vec_cmpgt (vector signed int, vector signed int);
7829 vector bool int vec_cmpgt (vector float, vector float);
7831 vector bool int vec_vcmpgtfp (vector float, vector float);
7833 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
7835 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
7837 vector bool short vec_vcmpgtsh (vector signed short,
7838 vector signed short);
7840 vector bool short vec_vcmpgtuh (vector unsigned short,
7841 vector unsigned short);
7843 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
7845 vector bool char vec_vcmpgtub (vector unsigned char,
7846 vector unsigned char);
7848 vector bool int vec_cmple (vector float, vector float);
7850 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
7851 vector bool char vec_cmplt (vector signed char, vector signed char);
7852 vector bool short vec_cmplt (vector unsigned short,
7853 vector unsigned short);
7854 vector bool short vec_cmplt (vector signed short, vector signed short);
7855 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
7856 vector bool int vec_cmplt (vector signed int, vector signed int);
7857 vector bool int vec_cmplt (vector float, vector float);
7859 vector float vec_ctf (vector unsigned int, const int);
7860 vector float vec_ctf (vector signed int, const int);
7862 vector float vec_vcfsx (vector signed int, const int);
7864 vector float vec_vcfux (vector unsigned int, const int);
7866 vector signed int vec_cts (vector float, const int);
7868 vector unsigned int vec_ctu (vector float, const int);
7870 void vec_dss (const int);
7872 void vec_dssall (void);
7874 void vec_dst (const vector unsigned char *, int, const int);
7875 void vec_dst (const vector signed char *, int, const int);
7876 void vec_dst (const vector bool char *, int, const int);
7877 void vec_dst (const vector unsigned short *, int, const int);
7878 void vec_dst (const vector signed short *, int, const int);
7879 void vec_dst (const vector bool short *, int, const int);
7880 void vec_dst (const vector pixel *, int, const int);
7881 void vec_dst (const vector unsigned int *, int, const int);
7882 void vec_dst (const vector signed int *, int, const int);
7883 void vec_dst (const vector bool int *, int, const int);
7884 void vec_dst (const vector float *, int, const int);
7885 void vec_dst (const unsigned char *, int, const int);
7886 void vec_dst (const signed char *, int, const int);
7887 void vec_dst (const unsigned short *, int, const int);
7888 void vec_dst (const short *, int, const int);
7889 void vec_dst (const unsigned int *, int, const int);
7890 void vec_dst (const int *, int, const int);
7891 void vec_dst (const unsigned long *, int, const int);
7892 void vec_dst (const long *, int, const int);
7893 void vec_dst (const float *, int, const int);
7895 void vec_dstst (const vector unsigned char *, int, const int);
7896 void vec_dstst (const vector signed char *, int, const int);
7897 void vec_dstst (const vector bool char *, int, const int);
7898 void vec_dstst (const vector unsigned short *, int, const int);
7899 void vec_dstst (const vector signed short *, int, const int);
7900 void vec_dstst (const vector bool short *, int, const int);
7901 void vec_dstst (const vector pixel *, int, const int);
7902 void vec_dstst (const vector unsigned int *, int, const int);
7903 void vec_dstst (const vector signed int *, int, const int);
7904 void vec_dstst (const vector bool int *, int, const int);
7905 void vec_dstst (const vector float *, int, const int);
7906 void vec_dstst (const unsigned char *, int, const int);
7907 void vec_dstst (const signed char *, int, const int);
7908 void vec_dstst (const unsigned short *, int, const int);
7909 void vec_dstst (const short *, int, const int);
7910 void vec_dstst (const unsigned int *, int, const int);
7911 void vec_dstst (const int *, int, const int);
7912 void vec_dstst (const unsigned long *, int, const int);
7913 void vec_dstst (const long *, int, const int);
7914 void vec_dstst (const float *, int, const int);
7916 void vec_dststt (const vector unsigned char *, int, const int);
7917 void vec_dststt (const vector signed char *, int, const int);
7918 void vec_dststt (const vector bool char *, int, const int);
7919 void vec_dststt (const vector unsigned short *, int, const int);
7920 void vec_dststt (const vector signed short *, int, const int);
7921 void vec_dststt (const vector bool short *, int, const int);
7922 void vec_dststt (const vector pixel *, int, const int);
7923 void vec_dststt (const vector unsigned int *, int, const int);
7924 void vec_dststt (const vector signed int *, int, const int);
7925 void vec_dststt (const vector bool int *, int, const int);
7926 void vec_dststt (const vector float *, int, const int);
7927 void vec_dststt (const unsigned char *, int, const int);
7928 void vec_dststt (const signed char *, int, const int);
7929 void vec_dststt (const unsigned short *, int, const int);
7930 void vec_dststt (const short *, int, const int);
7931 void vec_dststt (const unsigned int *, int, const int);
7932 void vec_dststt (const int *, int, const int);
7933 void vec_dststt (const unsigned long *, int, const int);
7934 void vec_dststt (const long *, int, const int);
7935 void vec_dststt (const float *, int, const int);
7937 void vec_dstt (const vector unsigned char *, int, const int);
7938 void vec_dstt (const vector signed char *, int, const int);
7939 void vec_dstt (const vector bool char *, int, const int);
7940 void vec_dstt (const vector unsigned short *, int, const int);
7941 void vec_dstt (const vector signed short *, int, const int);
7942 void vec_dstt (const vector bool short *, int, const int);
7943 void vec_dstt (const vector pixel *, int, const int);
7944 void vec_dstt (const vector unsigned int *, int, const int);
7945 void vec_dstt (const vector signed int *, int, const int);
7946 void vec_dstt (const vector bool int *, int, const int);
7947 void vec_dstt (const vector float *, int, const int);
7948 void vec_dstt (const unsigned char *, int, const int);
7949 void vec_dstt (const signed char *, int, const int);
7950 void vec_dstt (const unsigned short *, int, const int);
7951 void vec_dstt (const short *, int, const int);
7952 void vec_dstt (const unsigned int *, int, const int);
7953 void vec_dstt (const int *, int, const int);
7954 void vec_dstt (const unsigned long *, int, const int);
7955 void vec_dstt (const long *, int, const int);
7956 void vec_dstt (const float *, int, const int);
7958 vector float vec_expte (vector float);
7960 vector float vec_floor (vector float);
7962 vector float vec_ld (int, const vector float *);
7963 vector float vec_ld (int, const float *);
7964 vector bool int vec_ld (int, const vector bool int *);
7965 vector signed int vec_ld (int, const vector signed int *);
7966 vector signed int vec_ld (int, const int *);
7967 vector signed int vec_ld (int, const long *);
7968 vector unsigned int vec_ld (int, const vector unsigned int *);
7969 vector unsigned int vec_ld (int, const unsigned int *);
7970 vector unsigned int vec_ld (int, const unsigned long *);
7971 vector bool short vec_ld (int, const vector bool short *);
7972 vector pixel vec_ld (int, const vector pixel *);
7973 vector signed short vec_ld (int, const vector signed short *);
7974 vector signed short vec_ld (int, const short *);
7975 vector unsigned short vec_ld (int, const vector unsigned short *);
7976 vector unsigned short vec_ld (int, const unsigned short *);
7977 vector bool char vec_ld (int, const vector bool char *);
7978 vector signed char vec_ld (int, const vector signed char *);
7979 vector signed char vec_ld (int, const signed char *);
7980 vector unsigned char vec_ld (int, const vector unsigned char *);
7981 vector unsigned char vec_ld (int, const unsigned char *);
7983 vector signed char vec_lde (int, const signed char *);
7984 vector unsigned char vec_lde (int, const unsigned char *);
7985 vector signed short vec_lde (int, const short *);
7986 vector unsigned short vec_lde (int, const unsigned short *);
7987 vector float vec_lde (int, const float *);
7988 vector signed int vec_lde (int, const int *);
7989 vector unsigned int vec_lde (int, const unsigned int *);
7990 vector signed int vec_lde (int, const long *);
7991 vector unsigned int vec_lde (int, const unsigned long *);
7993 vector float vec_lvewx (int, float *);
7994 vector signed int vec_lvewx (int, int *);
7995 vector unsigned int vec_lvewx (int, unsigned int *);
7996 vector signed int vec_lvewx (int, long *);
7997 vector unsigned int vec_lvewx (int, unsigned long *);
7999 vector signed short vec_lvehx (int, short *);
8000 vector unsigned short vec_lvehx (int, unsigned short *);
8002 vector signed char vec_lvebx (int, char *);
8003 vector unsigned char vec_lvebx (int, unsigned char *);
8005 vector float vec_ldl (int, const vector float *);
8006 vector float vec_ldl (int, const float *);
8007 vector bool int vec_ldl (int, const vector bool int *);
8008 vector signed int vec_ldl (int, const vector signed int *);
8009 vector signed int vec_ldl (int, const int *);
8010 vector signed int vec_ldl (int, const long *);
8011 vector unsigned int vec_ldl (int, const vector unsigned int *);
8012 vector unsigned int vec_ldl (int, const unsigned int *);
8013 vector unsigned int vec_ldl (int, const unsigned long *);
8014 vector bool short vec_ldl (int, const vector bool short *);
8015 vector pixel vec_ldl (int, const vector pixel *);
8016 vector signed short vec_ldl (int, const vector signed short *);
8017 vector signed short vec_ldl (int, const short *);
8018 vector unsigned short vec_ldl (int, const vector unsigned short *);
8019 vector unsigned short vec_ldl (int, const unsigned short *);
8020 vector bool char vec_ldl (int, const vector bool char *);
8021 vector signed char vec_ldl (int, const vector signed char *);
8022 vector signed char vec_ldl (int, const signed char *);
8023 vector unsigned char vec_ldl (int, const vector unsigned char *);
8024 vector unsigned char vec_ldl (int, const unsigned char *);
8026 vector float vec_loge (vector float);
8028 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
8029 vector unsigned char vec_lvsl (int, const volatile signed char *);
8030 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
8031 vector unsigned char vec_lvsl (int, const volatile short *);
8032 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
8033 vector unsigned char vec_lvsl (int, const volatile int *);
8034 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
8035 vector unsigned char vec_lvsl (int, const volatile long *);
8036 vector unsigned char vec_lvsl (int, const volatile float *);
8038 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
8039 vector unsigned char vec_lvsr (int, const volatile signed char *);
8040 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
8041 vector unsigned char vec_lvsr (int, const volatile short *);
8042 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
8043 vector unsigned char vec_lvsr (int, const volatile int *);
8044 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
8045 vector unsigned char vec_lvsr (int, const volatile long *);
8046 vector unsigned char vec_lvsr (int, const volatile float *);
8048 vector float vec_madd (vector float, vector float, vector float);
8050 vector signed short vec_madds (vector signed short,
8051 vector signed short,
8052 vector signed short);
8054 vector unsigned char vec_max (vector bool char, vector unsigned char);
8055 vector unsigned char vec_max (vector unsigned char, vector bool char);
8056 vector unsigned char vec_max (vector unsigned char,
8057 vector unsigned char);
8058 vector signed char vec_max (vector bool char, vector signed char);
8059 vector signed char vec_max (vector signed char, vector bool char);
8060 vector signed char vec_max (vector signed char, vector signed char);
8061 vector unsigned short vec_max (vector bool short,
8062 vector unsigned short);
8063 vector unsigned short vec_max (vector unsigned short,
8065 vector unsigned short vec_max (vector unsigned short,
8066 vector unsigned short);
8067 vector signed short vec_max (vector bool short, vector signed short);
8068 vector signed short vec_max (vector signed short, vector bool short);
8069 vector signed short vec_max (vector signed short, vector signed short);
8070 vector unsigned int vec_max (vector bool int, vector unsigned int);
8071 vector unsigned int vec_max (vector unsigned int, vector bool int);
8072 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
8073 vector signed int vec_max (vector bool int, vector signed int);
8074 vector signed int vec_max (vector signed int, vector bool int);
8075 vector signed int vec_max (vector signed int, vector signed int);
8076 vector float vec_max (vector float, vector float);
8078 vector float vec_vmaxfp (vector float, vector float);
8080 vector signed int vec_vmaxsw (vector bool int, vector signed int);
8081 vector signed int vec_vmaxsw (vector signed int, vector bool int);
8082 vector signed int vec_vmaxsw (vector signed int, vector signed int);
8084 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
8085 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
8086 vector unsigned int vec_vmaxuw (vector unsigned int,
8087 vector unsigned int);
8089 vector signed short vec_vmaxsh (vector bool short, vector signed short);
8090 vector signed short vec_vmaxsh (vector signed short, vector bool short);
8091 vector signed short vec_vmaxsh (vector signed short,
8092 vector signed short);
8094 vector unsigned short vec_vmaxuh (vector bool short,
8095 vector unsigned short);
8096 vector unsigned short vec_vmaxuh (vector unsigned short,
8098 vector unsigned short vec_vmaxuh (vector unsigned short,
8099 vector unsigned short);
8101 vector signed char vec_vmaxsb (vector bool char, vector signed char);
8102 vector signed char vec_vmaxsb (vector signed char, vector bool char);
8103 vector signed char vec_vmaxsb (vector signed char, vector signed char);
8105 vector unsigned char vec_vmaxub (vector bool char,
8106 vector unsigned char);
8107 vector unsigned char vec_vmaxub (vector unsigned char,
8109 vector unsigned char vec_vmaxub (vector unsigned char,
8110 vector unsigned char);
8112 vector bool char vec_mergeh (vector bool char, vector bool char);
8113 vector signed char vec_mergeh (vector signed char, vector signed char);
8114 vector unsigned char vec_mergeh (vector unsigned char,
8115 vector unsigned char);
8116 vector bool short vec_mergeh (vector bool short, vector bool short);
8117 vector pixel vec_mergeh (vector pixel, vector pixel);
8118 vector signed short vec_mergeh (vector signed short,
8119 vector signed short);
8120 vector unsigned short vec_mergeh (vector unsigned short,
8121 vector unsigned short);
8122 vector float vec_mergeh (vector float, vector float);
8123 vector bool int vec_mergeh (vector bool int, vector bool int);
8124 vector signed int vec_mergeh (vector signed int, vector signed int);
8125 vector unsigned int vec_mergeh (vector unsigned int,
8126 vector unsigned int);
8128 vector float vec_vmrghw (vector float, vector float);
8129 vector bool int vec_vmrghw (vector bool int, vector bool int);
8130 vector signed int vec_vmrghw (vector signed int, vector signed int);
8131 vector unsigned int vec_vmrghw (vector unsigned int,
8132 vector unsigned int);
8134 vector bool short vec_vmrghh (vector bool short, vector bool short);
8135 vector signed short vec_vmrghh (vector signed short,
8136 vector signed short);
8137 vector unsigned short vec_vmrghh (vector unsigned short,
8138 vector unsigned short);
8139 vector pixel vec_vmrghh (vector pixel, vector pixel);
8141 vector bool char vec_vmrghb (vector bool char, vector bool char);
8142 vector signed char vec_vmrghb (vector signed char, vector signed char);
8143 vector unsigned char vec_vmrghb (vector unsigned char,
8144 vector unsigned char);
8146 vector bool char vec_mergel (vector bool char, vector bool char);
8147 vector signed char vec_mergel (vector signed char, vector signed char);
8148 vector unsigned char vec_mergel (vector unsigned char,
8149 vector unsigned char);
8150 vector bool short vec_mergel (vector bool short, vector bool short);
8151 vector pixel vec_mergel (vector pixel, vector pixel);
8152 vector signed short vec_mergel (vector signed short,
8153 vector signed short);
8154 vector unsigned short vec_mergel (vector unsigned short,
8155 vector unsigned short);
8156 vector float vec_mergel (vector float, vector float);
8157 vector bool int vec_mergel (vector bool int, vector bool int);
8158 vector signed int vec_mergel (vector signed int, vector signed int);
8159 vector unsigned int vec_mergel (vector unsigned int,
8160 vector unsigned int);
8162 vector float vec_vmrglw (vector float, vector float);
8163 vector signed int vec_vmrglw (vector signed int, vector signed int);
8164 vector unsigned int vec_vmrglw (vector unsigned int,
8165 vector unsigned int);
8166 vector bool int vec_vmrglw (vector bool int, vector bool int);
8168 vector bool short vec_vmrglh (vector bool short, vector bool short);
8169 vector signed short vec_vmrglh (vector signed short,
8170 vector signed short);
8171 vector unsigned short vec_vmrglh (vector unsigned short,
8172 vector unsigned short);
8173 vector pixel vec_vmrglh (vector pixel, vector pixel);
8175 vector bool char vec_vmrglb (vector bool char, vector bool char);
8176 vector signed char vec_vmrglb (vector signed char, vector signed char);
8177 vector unsigned char vec_vmrglb (vector unsigned char,
8178 vector unsigned char);
8180 vector unsigned short vec_mfvscr (void);
8182 vector unsigned char vec_min (vector bool char, vector unsigned char);
8183 vector unsigned char vec_min (vector unsigned char, vector bool char);
8184 vector unsigned char vec_min (vector unsigned char,
8185 vector unsigned char);
8186 vector signed char vec_min (vector bool char, vector signed char);
8187 vector signed char vec_min (vector signed char, vector bool char);
8188 vector signed char vec_min (vector signed char, vector signed char);
8189 vector unsigned short vec_min (vector bool short,
8190 vector unsigned short);
8191 vector unsigned short vec_min (vector unsigned short,
8193 vector unsigned short vec_min (vector unsigned short,
8194 vector unsigned short);
8195 vector signed short vec_min (vector bool short, vector signed short);
8196 vector signed short vec_min (vector signed short, vector bool short);
8197 vector signed short vec_min (vector signed short, vector signed short);
8198 vector unsigned int vec_min (vector bool int, vector unsigned int);
8199 vector unsigned int vec_min (vector unsigned int, vector bool int);
8200 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
8201 vector signed int vec_min (vector bool int, vector signed int);
8202 vector signed int vec_min (vector signed int, vector bool int);
8203 vector signed int vec_min (vector signed int, vector signed int);
8204 vector float vec_min (vector float, vector float);
8206 vector float vec_vminfp (vector float, vector float);
8208 vector signed int vec_vminsw (vector bool int, vector signed int);
8209 vector signed int vec_vminsw (vector signed int, vector bool int);
8210 vector signed int vec_vminsw (vector signed int, vector signed int);
8212 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
8213 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
8214 vector unsigned int vec_vminuw (vector unsigned int,
8215 vector unsigned int);
8217 vector signed short vec_vminsh (vector bool short, vector signed short);
8218 vector signed short vec_vminsh (vector signed short, vector bool short);
8219 vector signed short vec_vminsh (vector signed short,
8220 vector signed short);
8222 vector unsigned short vec_vminuh (vector bool short,
8223 vector unsigned short);
8224 vector unsigned short vec_vminuh (vector unsigned short,
8226 vector unsigned short vec_vminuh (vector unsigned short,
8227 vector unsigned short);
8229 vector signed char vec_vminsb (vector bool char, vector signed char);
8230 vector signed char vec_vminsb (vector signed char, vector bool char);
8231 vector signed char vec_vminsb (vector signed char, vector signed char);
8233 vector unsigned char vec_vminub (vector bool char,
8234 vector unsigned char);
8235 vector unsigned char vec_vminub (vector unsigned char,
8237 vector unsigned char vec_vminub (vector unsigned char,
8238 vector unsigned char);
8240 vector signed short vec_mladd (vector signed short,
8241 vector signed short,
8242 vector signed short);
8243 vector signed short vec_mladd (vector signed short,
8244 vector unsigned short,
8245 vector unsigned short);
8246 vector signed short vec_mladd (vector unsigned short,
8247 vector signed short,
8248 vector signed short);
8249 vector unsigned short vec_mladd (vector unsigned short,
8250 vector unsigned short,
8251 vector unsigned short);
8253 vector signed short vec_mradds (vector signed short,
8254 vector signed short,
8255 vector signed short);
8257 vector unsigned int vec_msum (vector unsigned char,
8258 vector unsigned char,
8259 vector unsigned int);
8260 vector signed int vec_msum (vector signed char,
8261 vector unsigned char,
8263 vector unsigned int vec_msum (vector unsigned short,
8264 vector unsigned short,
8265 vector unsigned int);
8266 vector signed int vec_msum (vector signed short,
8267 vector signed short,
8270 vector signed int vec_vmsumshm (vector signed short,
8271 vector signed short,
8274 vector unsigned int vec_vmsumuhm (vector unsigned short,
8275 vector unsigned short,
8276 vector unsigned int);
8278 vector signed int vec_vmsummbm (vector signed char,
8279 vector unsigned char,
8282 vector unsigned int vec_vmsumubm (vector unsigned char,
8283 vector unsigned char,
8284 vector unsigned int);
8286 vector unsigned int vec_msums (vector unsigned short,
8287 vector unsigned short,
8288 vector unsigned int);
8289 vector signed int vec_msums (vector signed short,
8290 vector signed short,
8293 vector signed int vec_vmsumshs (vector signed short,
8294 vector signed short,
8297 vector unsigned int vec_vmsumuhs (vector unsigned short,
8298 vector unsigned short,
8299 vector unsigned int);
8301 void vec_mtvscr (vector signed int);
8302 void vec_mtvscr (vector unsigned int);
8303 void vec_mtvscr (vector bool int);
8304 void vec_mtvscr (vector signed short);
8305 void vec_mtvscr (vector unsigned short);
8306 void vec_mtvscr (vector bool short);
8307 void vec_mtvscr (vector pixel);
8308 void vec_mtvscr (vector signed char);
8309 void vec_mtvscr (vector unsigned char);
8310 void vec_mtvscr (vector bool char);
8312 vector unsigned short vec_mule (vector unsigned char,
8313 vector unsigned char);
8314 vector signed short vec_mule (vector signed char,
8315 vector signed char);
8316 vector unsigned int vec_mule (vector unsigned short,
8317 vector unsigned short);
8318 vector signed int vec_mule (vector signed short, vector signed short);
8320 vector signed int vec_vmulesh (vector signed short,
8321 vector signed short);
8323 vector unsigned int vec_vmuleuh (vector unsigned short,
8324 vector unsigned short);
8326 vector signed short vec_vmulesb (vector signed char,
8327 vector signed char);
8329 vector unsigned short vec_vmuleub (vector unsigned char,
8330 vector unsigned char);
8332 vector unsigned short vec_mulo (vector unsigned char,
8333 vector unsigned char);
8334 vector signed short vec_mulo (vector signed char, vector signed char);
8335 vector unsigned int vec_mulo (vector unsigned short,
8336 vector unsigned short);
8337 vector signed int vec_mulo (vector signed short, vector signed short);
8339 vector signed int vec_vmulosh (vector signed short,
8340 vector signed short);
8342 vector unsigned int vec_vmulouh (vector unsigned short,
8343 vector unsigned short);
8345 vector signed short vec_vmulosb (vector signed char,
8346 vector signed char);
8348 vector unsigned short vec_vmuloub (vector unsigned char,
8349 vector unsigned char);
8351 vector float vec_nmsub (vector float, vector float, vector float);
8353 vector float vec_nor (vector float, vector float);
8354 vector signed int vec_nor (vector signed int, vector signed int);
8355 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
8356 vector bool int vec_nor (vector bool int, vector bool int);
8357 vector signed short vec_nor (vector signed short, vector signed short);
8358 vector unsigned short vec_nor (vector unsigned short,
8359 vector unsigned short);
8360 vector bool short vec_nor (vector bool short, vector bool short);
8361 vector signed char vec_nor (vector signed char, vector signed char);
8362 vector unsigned char vec_nor (vector unsigned char,
8363 vector unsigned char);
8364 vector bool char vec_nor (vector bool char, vector bool char);
8366 vector float vec_or (vector float, vector float);
8367 vector float vec_or (vector float, vector bool int);
8368 vector float vec_or (vector bool int, vector float);
8369 vector bool int vec_or (vector bool int, vector bool int);
8370 vector signed int vec_or (vector bool int, vector signed int);
8371 vector signed int vec_or (vector signed int, vector bool int);
8372 vector signed int vec_or (vector signed int, vector signed int);
8373 vector unsigned int vec_or (vector bool int, vector unsigned int);
8374 vector unsigned int vec_or (vector unsigned int, vector bool int);
8375 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
8376 vector bool short vec_or (vector bool short, vector bool short);
8377 vector signed short vec_or (vector bool short, vector signed short);
8378 vector signed short vec_or (vector signed short, vector bool short);
8379 vector signed short vec_or (vector signed short, vector signed short);
8380 vector unsigned short vec_or (vector bool short, vector unsigned short);
8381 vector unsigned short vec_or (vector unsigned short, vector bool short);
8382 vector unsigned short vec_or (vector unsigned short,
8383 vector unsigned short);
8384 vector signed char vec_or (vector bool char, vector signed char);
8385 vector bool char vec_or (vector bool char, vector bool char);
8386 vector signed char vec_or (vector signed char, vector bool char);
8387 vector signed char vec_or (vector signed char, vector signed char);
8388 vector unsigned char vec_or (vector bool char, vector unsigned char);
8389 vector unsigned char vec_or (vector unsigned char, vector bool char);
8390 vector unsigned char vec_or (vector unsigned char,
8391 vector unsigned char);
8393 vector signed char vec_pack (vector signed short, vector signed short);
8394 vector unsigned char vec_pack (vector unsigned short,
8395 vector unsigned short);
8396 vector bool char vec_pack (vector bool short, vector bool short);
8397 vector signed short vec_pack (vector signed int, vector signed int);
8398 vector unsigned short vec_pack (vector unsigned int,
8399 vector unsigned int);
8400 vector bool short vec_pack (vector bool int, vector bool int);
8402 vector bool short vec_vpkuwum (vector bool int, vector bool int);
8403 vector signed short vec_vpkuwum (vector signed int, vector signed int);
8404 vector unsigned short vec_vpkuwum (vector unsigned int,
8405 vector unsigned int);
8407 vector bool char vec_vpkuhum (vector bool short, vector bool short);
8408 vector signed char vec_vpkuhum (vector signed short,
8409 vector signed short);
8410 vector unsigned char vec_vpkuhum (vector unsigned short,
8411 vector unsigned short);
8413 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
8415 vector unsigned char vec_packs (vector unsigned short,
8416 vector unsigned short);
8417 vector signed char vec_packs (vector signed short, vector signed short);
8418 vector unsigned short vec_packs (vector unsigned int,
8419 vector unsigned int);
8420 vector signed short vec_packs (vector signed int, vector signed int);
8422 vector signed short vec_vpkswss (vector signed int, vector signed int);
8424 vector unsigned short vec_vpkuwus (vector unsigned int,
8425 vector unsigned int);
8427 vector signed char vec_vpkshss (vector signed short,
8428 vector signed short);
8430 vector unsigned char vec_vpkuhus (vector unsigned short,
8431 vector unsigned short);
8433 vector unsigned char vec_packsu (vector unsigned short,
8434 vector unsigned short);
8435 vector unsigned char vec_packsu (vector signed short,
8436 vector signed short);
8437 vector unsigned short vec_packsu (vector unsigned int,
8438 vector unsigned int);
8439 vector unsigned short vec_packsu (vector signed int, vector signed int);
8441 vector unsigned short vec_vpkswus (vector signed int,
8444 vector unsigned char vec_vpkshus (vector signed short,
8445 vector signed short);
8447 vector float vec_perm (vector float,
8449 vector unsigned char);
8450 vector signed int vec_perm (vector signed int,
8452 vector unsigned char);
8453 vector unsigned int vec_perm (vector unsigned int,
8454 vector unsigned int,
8455 vector unsigned char);
8456 vector bool int vec_perm (vector bool int,
8458 vector unsigned char);
8459 vector signed short vec_perm (vector signed short,
8460 vector signed short,
8461 vector unsigned char);
8462 vector unsigned short vec_perm (vector unsigned short,
8463 vector unsigned short,
8464 vector unsigned char);
8465 vector bool short vec_perm (vector bool short,
8467 vector unsigned char);
8468 vector pixel vec_perm (vector pixel,
8470 vector unsigned char);
8471 vector signed char vec_perm (vector signed char,
8473 vector unsigned char);
8474 vector unsigned char vec_perm (vector unsigned char,
8475 vector unsigned char,
8476 vector unsigned char);
8477 vector bool char vec_perm (vector bool char,
8479 vector unsigned char);
8481 vector float vec_re (vector float);
8483 vector signed char vec_rl (vector signed char,
8484 vector unsigned char);
8485 vector unsigned char vec_rl (vector unsigned char,
8486 vector unsigned char);
8487 vector signed short vec_rl (vector signed short, vector unsigned short);
8488 vector unsigned short vec_rl (vector unsigned short,
8489 vector unsigned short);
8490 vector signed int vec_rl (vector signed int, vector unsigned int);
8491 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
8493 vector signed int vec_vrlw (vector signed int, vector unsigned int);
8494 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
8496 vector signed short vec_vrlh (vector signed short,
8497 vector unsigned short);
8498 vector unsigned short vec_vrlh (vector unsigned short,
8499 vector unsigned short);
8501 vector signed char vec_vrlb (vector signed char, vector unsigned char);
8502 vector unsigned char vec_vrlb (vector unsigned char,
8503 vector unsigned char);
8505 vector float vec_round (vector float);
8507 vector float vec_rsqrte (vector float);
8509 vector float vec_sel (vector float, vector float, vector bool int);
8510 vector float vec_sel (vector float, vector float, vector unsigned int);
8511 vector signed int vec_sel (vector signed int,
8514 vector signed int vec_sel (vector signed int,
8516 vector unsigned int);
8517 vector unsigned int vec_sel (vector unsigned int,
8518 vector unsigned int,
8520 vector unsigned int vec_sel (vector unsigned int,
8521 vector unsigned int,
8522 vector unsigned int);
8523 vector bool int vec_sel (vector bool int,
8526 vector bool int vec_sel (vector bool int,
8528 vector unsigned int);
8529 vector signed short vec_sel (vector signed short,
8530 vector signed short,
8532 vector signed short vec_sel (vector signed short,
8533 vector signed short,
8534 vector unsigned short);
8535 vector unsigned short vec_sel (vector unsigned short,
8536 vector unsigned short,
8538 vector unsigned short vec_sel (vector unsigned short,
8539 vector unsigned short,
8540 vector unsigned short);
8541 vector bool short vec_sel (vector bool short,
8544 vector bool short vec_sel (vector bool short,
8546 vector unsigned short);
8547 vector signed char vec_sel (vector signed char,
8550 vector signed char vec_sel (vector signed char,
8552 vector unsigned char);
8553 vector unsigned char vec_sel (vector unsigned char,
8554 vector unsigned char,
8556 vector unsigned char vec_sel (vector unsigned char,
8557 vector unsigned char,
8558 vector unsigned char);
8559 vector bool char vec_sel (vector bool char,
8562 vector bool char vec_sel (vector bool char,
8564 vector unsigned char);
8566 vector signed char vec_sl (vector signed char,
8567 vector unsigned char);
8568 vector unsigned char vec_sl (vector unsigned char,
8569 vector unsigned char);
8570 vector signed short vec_sl (vector signed short, vector unsigned short);
8571 vector unsigned short vec_sl (vector unsigned short,
8572 vector unsigned short);
8573 vector signed int vec_sl (vector signed int, vector unsigned int);
8574 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
8576 vector signed int vec_vslw (vector signed int, vector unsigned int);
8577 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
8579 vector signed short vec_vslh (vector signed short,
8580 vector unsigned short);
8581 vector unsigned short vec_vslh (vector unsigned short,
8582 vector unsigned short);
8584 vector signed char vec_vslb (vector signed char, vector unsigned char);
8585 vector unsigned char vec_vslb (vector unsigned char,
8586 vector unsigned char);
8588 vector float vec_sld (vector float, vector float, const int);
8589 vector signed int vec_sld (vector signed int,
8592 vector unsigned int vec_sld (vector unsigned int,
8593 vector unsigned int,
8595 vector bool int vec_sld (vector bool int,
8598 vector signed short vec_sld (vector signed short,
8599 vector signed short,
8601 vector unsigned short vec_sld (vector unsigned short,
8602 vector unsigned short,
8604 vector bool short vec_sld (vector bool short,
8607 vector pixel vec_sld (vector pixel,
8610 vector signed char vec_sld (vector signed char,
8613 vector unsigned char vec_sld (vector unsigned char,
8614 vector unsigned char,
8616 vector bool char vec_sld (vector bool char,
8620 vector signed int vec_sll (vector signed int,
8621 vector unsigned int);
8622 vector signed int vec_sll (vector signed int,
8623 vector unsigned short);
8624 vector signed int vec_sll (vector signed int,
8625 vector unsigned char);
8626 vector unsigned int vec_sll (vector unsigned int,
8627 vector unsigned int);
8628 vector unsigned int vec_sll (vector unsigned int,
8629 vector unsigned short);
8630 vector unsigned int vec_sll (vector unsigned int,
8631 vector unsigned char);
8632 vector bool int vec_sll (vector bool int,
8633 vector unsigned int);
8634 vector bool int vec_sll (vector bool int,
8635 vector unsigned short);
8636 vector bool int vec_sll (vector bool int,
8637 vector unsigned char);
8638 vector signed short vec_sll (vector signed short,
8639 vector unsigned int);
8640 vector signed short vec_sll (vector signed short,
8641 vector unsigned short);
8642 vector signed short vec_sll (vector signed short,
8643 vector unsigned char);
8644 vector unsigned short vec_sll (vector unsigned short,
8645 vector unsigned int);
8646 vector unsigned short vec_sll (vector unsigned short,
8647 vector unsigned short);
8648 vector unsigned short vec_sll (vector unsigned short,
8649 vector unsigned char);
8650 vector bool short vec_sll (vector bool short, vector unsigned int);
8651 vector bool short vec_sll (vector bool short, vector unsigned short);
8652 vector bool short vec_sll (vector bool short, vector unsigned char);
8653 vector pixel vec_sll (vector pixel, vector unsigned int);
8654 vector pixel vec_sll (vector pixel, vector unsigned short);
8655 vector pixel vec_sll (vector pixel, vector unsigned char);
8656 vector signed char vec_sll (vector signed char, vector unsigned int);
8657 vector signed char vec_sll (vector signed char, vector unsigned short);
8658 vector signed char vec_sll (vector signed char, vector unsigned char);
8659 vector unsigned char vec_sll (vector unsigned char,
8660 vector unsigned int);
8661 vector unsigned char vec_sll (vector unsigned char,
8662 vector unsigned short);
8663 vector unsigned char vec_sll (vector unsigned char,
8664 vector unsigned char);
8665 vector bool char vec_sll (vector bool char, vector unsigned int);
8666 vector bool char vec_sll (vector bool char, vector unsigned short);
8667 vector bool char vec_sll (vector bool char, vector unsigned char);
8669 vector float vec_slo (vector float, vector signed char);
8670 vector float vec_slo (vector float, vector unsigned char);
8671 vector signed int vec_slo (vector signed int, vector signed char);
8672 vector signed int vec_slo (vector signed int, vector unsigned char);
8673 vector unsigned int vec_slo (vector unsigned int, vector signed char);
8674 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
8675 vector signed short vec_slo (vector signed short, vector signed char);
8676 vector signed short vec_slo (vector signed short, vector unsigned char);
8677 vector unsigned short vec_slo (vector unsigned short,
8678 vector signed char);
8679 vector unsigned short vec_slo (vector unsigned short,
8680 vector unsigned char);
8681 vector pixel vec_slo (vector pixel, vector signed char);
8682 vector pixel vec_slo (vector pixel, vector unsigned char);
8683 vector signed char vec_slo (vector signed char, vector signed char);
8684 vector signed char vec_slo (vector signed char, vector unsigned char);
8685 vector unsigned char vec_slo (vector unsigned char, vector signed char);
8686 vector unsigned char vec_slo (vector unsigned char,
8687 vector unsigned char);
8689 vector signed char vec_splat (vector signed char, const int);
8690 vector unsigned char vec_splat (vector unsigned char, const int);
8691 vector bool char vec_splat (vector bool char, const int);
8692 vector signed short vec_splat (vector signed short, const int);
8693 vector unsigned short vec_splat (vector unsigned short, const int);
8694 vector bool short vec_splat (vector bool short, const int);
8695 vector pixel vec_splat (vector pixel, const int);
8696 vector float vec_splat (vector float, const int);
8697 vector signed int vec_splat (vector signed int, const int);
8698 vector unsigned int vec_splat (vector unsigned int, const int);
8699 vector bool int vec_splat (vector bool int, const int);
8701 vector float vec_vspltw (vector float, const int);
8702 vector signed int vec_vspltw (vector signed int, const int);
8703 vector unsigned int vec_vspltw (vector unsigned int, const int);
8704 vector bool int vec_vspltw (vector bool int, const int);
8706 vector bool short vec_vsplth (vector bool short, const int);
8707 vector signed short vec_vsplth (vector signed short, const int);
8708 vector unsigned short vec_vsplth (vector unsigned short, const int);
8709 vector pixel vec_vsplth (vector pixel, const int);
8711 vector signed char vec_vspltb (vector signed char, const int);
8712 vector unsigned char vec_vspltb (vector unsigned char, const int);
8713 vector bool char vec_vspltb (vector bool char, const int);
8715 vector signed char vec_splat_s8 (const int);
8717 vector signed short vec_splat_s16 (const int);
8719 vector signed int vec_splat_s32 (const int);
8721 vector unsigned char vec_splat_u8 (const int);
8723 vector unsigned short vec_splat_u16 (const int);
8725 vector unsigned int vec_splat_u32 (const int);
8727 vector signed char vec_sr (vector signed char, vector unsigned char);
8728 vector unsigned char vec_sr (vector unsigned char,
8729 vector unsigned char);
8730 vector signed short vec_sr (vector signed short,
8731 vector unsigned short);
8732 vector unsigned short vec_sr (vector unsigned short,
8733 vector unsigned short);
8734 vector signed int vec_sr (vector signed int, vector unsigned int);
8735 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
8737 vector signed int vec_vsrw (vector signed int, vector unsigned int);
8738 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
8740 vector signed short vec_vsrh (vector signed short,
8741 vector unsigned short);
8742 vector unsigned short vec_vsrh (vector unsigned short,
8743 vector unsigned short);
8745 vector signed char vec_vsrb (vector signed char, vector unsigned char);
8746 vector unsigned char vec_vsrb (vector unsigned char,
8747 vector unsigned char);
8749 vector signed char vec_sra (vector signed char, vector unsigned char);
8750 vector unsigned char vec_sra (vector unsigned char,
8751 vector unsigned char);
8752 vector signed short vec_sra (vector signed short,
8753 vector unsigned short);
8754 vector unsigned short vec_sra (vector unsigned short,
8755 vector unsigned short);
8756 vector signed int vec_sra (vector signed int, vector unsigned int);
8757 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
8759 vector signed int vec_vsraw (vector signed int, vector unsigned int);
8760 vector unsigned int vec_vsraw (vector unsigned int,
8761 vector unsigned int);
8763 vector signed short vec_vsrah (vector signed short,
8764 vector unsigned short);
8765 vector unsigned short vec_vsrah (vector unsigned short,
8766 vector unsigned short);
8768 vector signed char vec_vsrab (vector signed char, vector unsigned char);
8769 vector unsigned char vec_vsrab (vector unsigned char,
8770 vector unsigned char);
8772 vector signed int vec_srl (vector signed int, vector unsigned int);
8773 vector signed int vec_srl (vector signed int, vector unsigned short);
8774 vector signed int vec_srl (vector signed int, vector unsigned char);
8775 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
8776 vector unsigned int vec_srl (vector unsigned int,
8777 vector unsigned short);
8778 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
8779 vector bool int vec_srl (vector bool int, vector unsigned int);
8780 vector bool int vec_srl (vector bool int, vector unsigned short);
8781 vector bool int vec_srl (vector bool int, vector unsigned char);
8782 vector signed short vec_srl (vector signed short, vector unsigned int);
8783 vector signed short vec_srl (vector signed short,
8784 vector unsigned short);
8785 vector signed short vec_srl (vector signed short, vector unsigned char);
8786 vector unsigned short vec_srl (vector unsigned short,
8787 vector unsigned int);
8788 vector unsigned short vec_srl (vector unsigned short,
8789 vector unsigned short);
8790 vector unsigned short vec_srl (vector unsigned short,
8791 vector unsigned char);
8792 vector bool short vec_srl (vector bool short, vector unsigned int);
8793 vector bool short vec_srl (vector bool short, vector unsigned short);
8794 vector bool short vec_srl (vector bool short, vector unsigned char);
8795 vector pixel vec_srl (vector pixel, vector unsigned int);
8796 vector pixel vec_srl (vector pixel, vector unsigned short);
8797 vector pixel vec_srl (vector pixel, vector unsigned char);
8798 vector signed char vec_srl (vector signed char, vector unsigned int);
8799 vector signed char vec_srl (vector signed char, vector unsigned short);
8800 vector signed char vec_srl (vector signed char, vector unsigned char);
8801 vector unsigned char vec_srl (vector unsigned char,
8802 vector unsigned int);
8803 vector unsigned char vec_srl (vector unsigned char,
8804 vector unsigned short);
8805 vector unsigned char vec_srl (vector unsigned char,
8806 vector unsigned char);
8807 vector bool char vec_srl (vector bool char, vector unsigned int);
8808 vector bool char vec_srl (vector bool char, vector unsigned short);
8809 vector bool char vec_srl (vector bool char, vector unsigned char);
8811 vector float vec_sro (vector float, vector signed char);
8812 vector float vec_sro (vector float, vector unsigned char);
8813 vector signed int vec_sro (vector signed int, vector signed char);
8814 vector signed int vec_sro (vector signed int, vector unsigned char);
8815 vector unsigned int vec_sro (vector unsigned int, vector signed char);
8816 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
8817 vector signed short vec_sro (vector signed short, vector signed char);
8818 vector signed short vec_sro (vector signed short, vector unsigned char);
8819 vector unsigned short vec_sro (vector unsigned short,
8820 vector signed char);
8821 vector unsigned short vec_sro (vector unsigned short,
8822 vector unsigned char);
8823 vector pixel vec_sro (vector pixel, vector signed char);
8824 vector pixel vec_sro (vector pixel, vector unsigned char);
8825 vector signed char vec_sro (vector signed char, vector signed char);
8826 vector signed char vec_sro (vector signed char, vector unsigned char);
8827 vector unsigned char vec_sro (vector unsigned char, vector signed char);
8828 vector unsigned char vec_sro (vector unsigned char,
8829 vector unsigned char);
8831 void vec_st (vector float, int, vector float *);
8832 void vec_st (vector float, int, float *);
8833 void vec_st (vector signed int, int, vector signed int *);
8834 void vec_st (vector signed int, int, int *);
8835 void vec_st (vector unsigned int, int, vector unsigned int *);
8836 void vec_st (vector unsigned int, int, unsigned int *);
8837 void vec_st (vector bool int, int, vector bool int *);
8838 void vec_st (vector bool int, int, unsigned int *);
8839 void vec_st (vector bool int, int, int *);
8840 void vec_st (vector signed short, int, vector signed short *);
8841 void vec_st (vector signed short, int, short *);
8842 void vec_st (vector unsigned short, int, vector unsigned short *);
8843 void vec_st (vector unsigned short, int, unsigned short *);
8844 void vec_st (vector bool short, int, vector bool short *);
8845 void vec_st (vector bool short, int, unsigned short *);
8846 void vec_st (vector pixel, int, vector pixel *);
8847 void vec_st (vector pixel, int, unsigned short *);
8848 void vec_st (vector pixel, int, short *);
8849 void vec_st (vector bool short, int, short *);
8850 void vec_st (vector signed char, int, vector signed char *);
8851 void vec_st (vector signed char, int, signed char *);
8852 void vec_st (vector unsigned char, int, vector unsigned char *);
8853 void vec_st (vector unsigned char, int, unsigned char *);
8854 void vec_st (vector bool char, int, vector bool char *);
8855 void vec_st (vector bool char, int, unsigned char *);
8856 void vec_st (vector bool char, int, signed char *);
8858 void vec_ste (vector signed char, int, signed char *);
8859 void vec_ste (vector unsigned char, int, unsigned char *);
8860 void vec_ste (vector bool char, int, signed char *);
8861 void vec_ste (vector bool char, int, unsigned char *);
8862 void vec_ste (vector signed short, int, short *);
8863 void vec_ste (vector unsigned short, int, unsigned short *);
8864 void vec_ste (vector bool short, int, short *);
8865 void vec_ste (vector bool short, int, unsigned short *);
8866 void vec_ste (vector pixel, int, short *);
8867 void vec_ste (vector pixel, int, unsigned short *);
8868 void vec_ste (vector float, int, float *);
8869 void vec_ste (vector signed int, int, int *);
8870 void vec_ste (vector unsigned int, int, unsigned int *);
8871 void vec_ste (vector bool int, int, int *);
8872 void vec_ste (vector bool int, int, unsigned int *);
8874 void vec_stvewx (vector float, int, float *);
8875 void vec_stvewx (vector signed int, int, int *);
8876 void vec_stvewx (vector unsigned int, int, unsigned int *);
8877 void vec_stvewx (vector bool int, int, int *);
8878 void vec_stvewx (vector bool int, int, unsigned int *);
8880 void vec_stvehx (vector signed short, int, short *);
8881 void vec_stvehx (vector unsigned short, int, unsigned short *);
8882 void vec_stvehx (vector bool short, int, short *);
8883 void vec_stvehx (vector bool short, int, unsigned short *);
8884 void vec_stvehx (vector pixel, int, short *);
8885 void vec_stvehx (vector pixel, int, unsigned short *);
8887 void vec_stvebx (vector signed char, int, signed char *);
8888 void vec_stvebx (vector unsigned char, int, unsigned char *);
8889 void vec_stvebx (vector bool char, int, signed char *);
8890 void vec_stvebx (vector bool char, int, unsigned char *);
8892 void vec_stl (vector float, int, vector float *);
8893 void vec_stl (vector float, int, float *);
8894 void vec_stl (vector signed int, int, vector signed int *);
8895 void vec_stl (vector signed int, int, int *);
8896 void vec_stl (vector unsigned int, int, vector unsigned int *);
8897 void vec_stl (vector unsigned int, int, unsigned int *);
8898 void vec_stl (vector bool int, int, vector bool int *);
8899 void vec_stl (vector bool int, int, unsigned int *);
8900 void vec_stl (vector bool int, int, int *);
8901 void vec_stl (vector signed short, int, vector signed short *);
8902 void vec_stl (vector signed short, int, short *);
8903 void vec_stl (vector unsigned short, int, vector unsigned short *);
8904 void vec_stl (vector unsigned short, int, unsigned short *);
8905 void vec_stl (vector bool short, int, vector bool short *);
8906 void vec_stl (vector bool short, int, unsigned short *);
8907 void vec_stl (vector bool short, int, short *);
8908 void vec_stl (vector pixel, int, vector pixel *);
8909 void vec_stl (vector pixel, int, unsigned short *);
8910 void vec_stl (vector pixel, int, short *);
8911 void vec_stl (vector signed char, int, vector signed char *);
8912 void vec_stl (vector signed char, int, signed char *);
8913 void vec_stl (vector unsigned char, int, vector unsigned char *);
8914 void vec_stl (vector unsigned char, int, unsigned char *);
8915 void vec_stl (vector bool char, int, vector bool char *);
8916 void vec_stl (vector bool char, int, unsigned char *);
8917 void vec_stl (vector bool char, int, signed char *);
8919 vector signed char vec_sub (vector bool char, vector signed char);
8920 vector signed char vec_sub (vector signed char, vector bool char);
8921 vector signed char vec_sub (vector signed char, vector signed char);
8922 vector unsigned char vec_sub (vector bool char, vector unsigned char);
8923 vector unsigned char vec_sub (vector unsigned char, vector bool char);
8924 vector unsigned char vec_sub (vector unsigned char,
8925 vector unsigned char);
8926 vector signed short vec_sub (vector bool short, vector signed short);
8927 vector signed short vec_sub (vector signed short, vector bool short);
8928 vector signed short vec_sub (vector signed short, vector signed short);
8929 vector unsigned short vec_sub (vector bool short,
8930 vector unsigned short);
8931 vector unsigned short vec_sub (vector unsigned short,
8933 vector unsigned short vec_sub (vector unsigned short,
8934 vector unsigned short);
8935 vector signed int vec_sub (vector bool int, vector signed int);
8936 vector signed int vec_sub (vector signed int, vector bool int);
8937 vector signed int vec_sub (vector signed int, vector signed int);
8938 vector unsigned int vec_sub (vector bool int, vector unsigned int);
8939 vector unsigned int vec_sub (vector unsigned int, vector bool int);
8940 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
8941 vector float vec_sub (vector float, vector float);
8943 vector float vec_vsubfp (vector float, vector float);
8945 vector signed int vec_vsubuwm (vector bool int, vector signed int);
8946 vector signed int vec_vsubuwm (vector signed int, vector bool int);
8947 vector signed int vec_vsubuwm (vector signed int, vector signed int);
8948 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
8949 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
8950 vector unsigned int vec_vsubuwm (vector unsigned int,
8951 vector unsigned int);
8953 vector signed short vec_vsubuhm (vector bool short,
8954 vector signed short);
8955 vector signed short vec_vsubuhm (vector signed short,
8957 vector signed short vec_vsubuhm (vector signed short,
8958 vector signed short);
8959 vector unsigned short vec_vsubuhm (vector bool short,
8960 vector unsigned short);
8961 vector unsigned short vec_vsubuhm (vector unsigned short,
8963 vector unsigned short vec_vsubuhm (vector unsigned short,
8964 vector unsigned short);
8966 vector signed char vec_vsububm (vector bool char, vector signed char);
8967 vector signed char vec_vsububm (vector signed char, vector bool char);
8968 vector signed char vec_vsububm (vector signed char, vector signed char);
8969 vector unsigned char vec_vsububm (vector bool char,
8970 vector unsigned char);
8971 vector unsigned char vec_vsububm (vector unsigned char,
8973 vector unsigned char vec_vsububm (vector unsigned char,
8974 vector unsigned char);
8976 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
8978 vector unsigned char vec_subs (vector bool char, vector unsigned char);
8979 vector unsigned char vec_subs (vector unsigned char, vector bool char);
8980 vector unsigned char vec_subs (vector unsigned char,
8981 vector unsigned char);
8982 vector signed char vec_subs (vector bool char, vector signed char);
8983 vector signed char vec_subs (vector signed char, vector bool char);
8984 vector signed char vec_subs (vector signed char, vector signed char);
8985 vector unsigned short vec_subs (vector bool short,
8986 vector unsigned short);
8987 vector unsigned short vec_subs (vector unsigned short,
8989 vector unsigned short vec_subs (vector unsigned short,
8990 vector unsigned short);
8991 vector signed short vec_subs (vector bool short, vector signed short);
8992 vector signed short vec_subs (vector signed short, vector bool short);
8993 vector signed short vec_subs (vector signed short, vector signed short);
8994 vector unsigned int vec_subs (vector bool int, vector unsigned int);
8995 vector unsigned int vec_subs (vector unsigned int, vector bool int);
8996 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
8997 vector signed int vec_subs (vector bool int, vector signed int);
8998 vector signed int vec_subs (vector signed int, vector bool int);
8999 vector signed int vec_subs (vector signed int, vector signed int);
9001 vector signed int vec_vsubsws (vector bool int, vector signed int);
9002 vector signed int vec_vsubsws (vector signed int, vector bool int);
9003 vector signed int vec_vsubsws (vector signed int, vector signed int);
9005 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
9006 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
9007 vector unsigned int vec_vsubuws (vector unsigned int,
9008 vector unsigned int);
9010 vector signed short vec_vsubshs (vector bool short,
9011 vector signed short);
9012 vector signed short vec_vsubshs (vector signed short,
9014 vector signed short vec_vsubshs (vector signed short,
9015 vector signed short);
9017 vector unsigned short vec_vsubuhs (vector bool short,
9018 vector unsigned short);
9019 vector unsigned short vec_vsubuhs (vector unsigned short,
9021 vector unsigned short vec_vsubuhs (vector unsigned short,
9022 vector unsigned short);
9024 vector signed char vec_vsubsbs (vector bool char, vector signed char);
9025 vector signed char vec_vsubsbs (vector signed char, vector bool char);
9026 vector signed char vec_vsubsbs (vector signed char, vector signed char);
9028 vector unsigned char vec_vsububs (vector bool char,
9029 vector unsigned char);
9030 vector unsigned char vec_vsububs (vector unsigned char,
9032 vector unsigned char vec_vsububs (vector unsigned char,
9033 vector unsigned char);
9035 vector unsigned int vec_sum4s (vector unsigned char,
9036 vector unsigned int);
9037 vector signed int vec_sum4s (vector signed char, vector signed int);
9038 vector signed int vec_sum4s (vector signed short, vector signed int);
9040 vector signed int vec_vsum4shs (vector signed short, vector signed int);
9042 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
9044 vector unsigned int vec_vsum4ubs (vector unsigned char,
9045 vector unsigned int);
9047 vector signed int vec_sum2s (vector signed int, vector signed int);
9049 vector signed int vec_sums (vector signed int, vector signed int);
9051 vector float vec_trunc (vector float);
9053 vector signed short vec_unpackh (vector signed char);
9054 vector bool short vec_unpackh (vector bool char);
9055 vector signed int vec_unpackh (vector signed short);
9056 vector bool int vec_unpackh (vector bool short);
9057 vector unsigned int vec_unpackh (vector pixel);
9059 vector bool int vec_vupkhsh (vector bool short);
9060 vector signed int vec_vupkhsh (vector signed short);
9062 vector unsigned int vec_vupkhpx (vector pixel);
9064 vector bool short vec_vupkhsb (vector bool char);
9065 vector signed short vec_vupkhsb (vector signed char);
9067 vector signed short vec_unpackl (vector signed char);
9068 vector bool short vec_unpackl (vector bool char);
9069 vector unsigned int vec_unpackl (vector pixel);
9070 vector signed int vec_unpackl (vector signed short);
9071 vector bool int vec_unpackl (vector bool short);
9073 vector unsigned int vec_vupklpx (vector pixel);
9075 vector bool int vec_vupklsh (vector bool short);
9076 vector signed int vec_vupklsh (vector signed short);
9078 vector bool short vec_vupklsb (vector bool char);
9079 vector signed short vec_vupklsb (vector signed char);
9081 vector float vec_xor (vector float, vector float);
9082 vector float vec_xor (vector float, vector bool int);
9083 vector float vec_xor (vector bool int, vector float);
9084 vector bool int vec_xor (vector bool int, vector bool int);
9085 vector signed int vec_xor (vector bool int, vector signed int);
9086 vector signed int vec_xor (vector signed int, vector bool int);
9087 vector signed int vec_xor (vector signed int, vector signed int);
9088 vector unsigned int vec_xor (vector bool int, vector unsigned int);
9089 vector unsigned int vec_xor (vector unsigned int, vector bool int);
9090 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
9091 vector bool short vec_xor (vector bool short, vector bool short);
9092 vector signed short vec_xor (vector bool short, vector signed short);
9093 vector signed short vec_xor (vector signed short, vector bool short);
9094 vector signed short vec_xor (vector signed short, vector signed short);
9095 vector unsigned short vec_xor (vector bool short,
9096 vector unsigned short);
9097 vector unsigned short vec_xor (vector unsigned short,
9099 vector unsigned short vec_xor (vector unsigned short,
9100 vector unsigned short);
9101 vector signed char vec_xor (vector bool char, vector signed char);
9102 vector bool char vec_xor (vector bool char, vector bool char);
9103 vector signed char vec_xor (vector signed char, vector bool char);
9104 vector signed char vec_xor (vector signed char, vector signed char);
9105 vector unsigned char vec_xor (vector bool char, vector unsigned char);
9106 vector unsigned char vec_xor (vector unsigned char, vector bool char);
9107 vector unsigned char vec_xor (vector unsigned char,
9108 vector unsigned char);
9110 int vec_all_eq (vector signed char, vector bool char);
9111 int vec_all_eq (vector signed char, vector signed char);
9112 int vec_all_eq (vector unsigned char, vector bool char);
9113 int vec_all_eq (vector unsigned char, vector unsigned char);
9114 int vec_all_eq (vector bool char, vector bool char);
9115 int vec_all_eq (vector bool char, vector unsigned char);
9116 int vec_all_eq (vector bool char, vector signed char);
9117 int vec_all_eq (vector signed short, vector bool short);
9118 int vec_all_eq (vector signed short, vector signed short);
9119 int vec_all_eq (vector unsigned short, vector bool short);
9120 int vec_all_eq (vector unsigned short, vector unsigned short);
9121 int vec_all_eq (vector bool short, vector bool short);
9122 int vec_all_eq (vector bool short, vector unsigned short);
9123 int vec_all_eq (vector bool short, vector signed short);
9124 int vec_all_eq (vector pixel, vector pixel);
9125 int vec_all_eq (vector signed int, vector bool int);
9126 int vec_all_eq (vector signed int, vector signed int);
9127 int vec_all_eq (vector unsigned int, vector bool int);
9128 int vec_all_eq (vector unsigned int, vector unsigned int);
9129 int vec_all_eq (vector bool int, vector bool int);
9130 int vec_all_eq (vector bool int, vector unsigned int);
9131 int vec_all_eq (vector bool int, vector signed int);
9132 int vec_all_eq (vector float, vector float);
9134 int vec_all_ge (vector bool char, vector unsigned char);
9135 int vec_all_ge (vector unsigned char, vector bool char);
9136 int vec_all_ge (vector unsigned char, vector unsigned char);
9137 int vec_all_ge (vector bool char, vector signed char);
9138 int vec_all_ge (vector signed char, vector bool char);
9139 int vec_all_ge (vector signed char, vector signed char);
9140 int vec_all_ge (vector bool short, vector unsigned short);
9141 int vec_all_ge (vector unsigned short, vector bool short);
9142 int vec_all_ge (vector unsigned short, vector unsigned short);
9143 int vec_all_ge (vector signed short, vector signed short);
9144 int vec_all_ge (vector bool short, vector signed short);
9145 int vec_all_ge (vector signed short, vector bool short);
9146 int vec_all_ge (vector bool int, vector unsigned int);
9147 int vec_all_ge (vector unsigned int, vector bool int);
9148 int vec_all_ge (vector unsigned int, vector unsigned int);
9149 int vec_all_ge (vector bool int, vector signed int);
9150 int vec_all_ge (vector signed int, vector bool int);
9151 int vec_all_ge (vector signed int, vector signed int);
9152 int vec_all_ge (vector float, vector float);
9154 int vec_all_gt (vector bool char, vector unsigned char);
9155 int vec_all_gt (vector unsigned char, vector bool char);
9156 int vec_all_gt (vector unsigned char, vector unsigned char);
9157 int vec_all_gt (vector bool char, vector signed char);
9158 int vec_all_gt (vector signed char, vector bool char);
9159 int vec_all_gt (vector signed char, vector signed char);
9160 int vec_all_gt (vector bool short, vector unsigned short);
9161 int vec_all_gt (vector unsigned short, vector bool short);
9162 int vec_all_gt (vector unsigned short, vector unsigned short);
9163 int vec_all_gt (vector bool short, vector signed short);
9164 int vec_all_gt (vector signed short, vector bool short);
9165 int vec_all_gt (vector signed short, vector signed short);
9166 int vec_all_gt (vector bool int, vector unsigned int);
9167 int vec_all_gt (vector unsigned int, vector bool int);
9168 int vec_all_gt (vector unsigned int, vector unsigned int);
9169 int vec_all_gt (vector bool int, vector signed int);
9170 int vec_all_gt (vector signed int, vector bool int);
9171 int vec_all_gt (vector signed int, vector signed int);
9172 int vec_all_gt (vector float, vector float);
9174 int vec_all_in (vector float, vector float);
9176 int vec_all_le (vector bool char, vector unsigned char);
9177 int vec_all_le (vector unsigned char, vector bool char);
9178 int vec_all_le (vector unsigned char, vector unsigned char);
9179 int vec_all_le (vector bool char, vector signed char);
9180 int vec_all_le (vector signed char, vector bool char);
9181 int vec_all_le (vector signed char, vector signed char);
9182 int vec_all_le (vector bool short, vector unsigned short);
9183 int vec_all_le (vector unsigned short, vector bool short);
9184 int vec_all_le (vector unsigned short, vector unsigned short);
9185 int vec_all_le (vector bool short, vector signed short);
9186 int vec_all_le (vector signed short, vector bool short);
9187 int vec_all_le (vector signed short, vector signed short);
9188 int vec_all_le (vector bool int, vector unsigned int);
9189 int vec_all_le (vector unsigned int, vector bool int);
9190 int vec_all_le (vector unsigned int, vector unsigned int);
9191 int vec_all_le (vector bool int, vector signed int);
9192 int vec_all_le (vector signed int, vector bool int);
9193 int vec_all_le (vector signed int, vector signed int);
9194 int vec_all_le (vector float, vector float);
9196 int vec_all_lt (vector bool char, vector unsigned char);
9197 int vec_all_lt (vector unsigned char, vector bool char);
9198 int vec_all_lt (vector unsigned char, vector unsigned char);
9199 int vec_all_lt (vector bool char, vector signed char);
9200 int vec_all_lt (vector signed char, vector bool char);
9201 int vec_all_lt (vector signed char, vector signed char);
9202 int vec_all_lt (vector bool short, vector unsigned short);
9203 int vec_all_lt (vector unsigned short, vector bool short);
9204 int vec_all_lt (vector unsigned short, vector unsigned short);
9205 int vec_all_lt (vector bool short, vector signed short);
9206 int vec_all_lt (vector signed short, vector bool short);
9207 int vec_all_lt (vector signed short, vector signed short);
9208 int vec_all_lt (vector bool int, vector unsigned int);
9209 int vec_all_lt (vector unsigned int, vector bool int);
9210 int vec_all_lt (vector unsigned int, vector unsigned int);
9211 int vec_all_lt (vector bool int, vector signed int);
9212 int vec_all_lt (vector signed int, vector bool int);
9213 int vec_all_lt (vector signed int, vector signed int);
9214 int vec_all_lt (vector float, vector float);
9216 int vec_all_nan (vector float);
9218 int vec_all_ne (vector signed char, vector bool char);
9219 int vec_all_ne (vector signed char, vector signed char);
9220 int vec_all_ne (vector unsigned char, vector bool char);
9221 int vec_all_ne (vector unsigned char, vector unsigned char);
9222 int vec_all_ne (vector bool char, vector bool char);
9223 int vec_all_ne (vector bool char, vector unsigned char);
9224 int vec_all_ne (vector bool char, vector signed char);
9225 int vec_all_ne (vector signed short, vector bool short);
9226 int vec_all_ne (vector signed short, vector signed short);
9227 int vec_all_ne (vector unsigned short, vector bool short);
9228 int vec_all_ne (vector unsigned short, vector unsigned short);
9229 int vec_all_ne (vector bool short, vector bool short);
9230 int vec_all_ne (vector bool short, vector unsigned short);
9231 int vec_all_ne (vector bool short, vector signed short);
9232 int vec_all_ne (vector pixel, vector pixel);
9233 int vec_all_ne (vector signed int, vector bool int);
9234 int vec_all_ne (vector signed int, vector signed int);
9235 int vec_all_ne (vector unsigned int, vector bool int);
9236 int vec_all_ne (vector unsigned int, vector unsigned int);
9237 int vec_all_ne (vector bool int, vector bool int);
9238 int vec_all_ne (vector bool int, vector unsigned int);
9239 int vec_all_ne (vector bool int, vector signed int);
9240 int vec_all_ne (vector float, vector float);
9242 int vec_all_nge (vector float, vector float);
9244 int vec_all_ngt (vector float, vector float);
9246 int vec_all_nle (vector float, vector float);
9248 int vec_all_nlt (vector float, vector float);
9250 int vec_all_numeric (vector float);
9252 int vec_any_eq (vector signed char, vector bool char);
9253 int vec_any_eq (vector signed char, vector signed char);
9254 int vec_any_eq (vector unsigned char, vector bool char);
9255 int vec_any_eq (vector unsigned char, vector unsigned char);
9256 int vec_any_eq (vector bool char, vector bool char);
9257 int vec_any_eq (vector bool char, vector unsigned char);
9258 int vec_any_eq (vector bool char, vector signed char);
9259 int vec_any_eq (vector signed short, vector bool short);
9260 int vec_any_eq (vector signed short, vector signed short);
9261 int vec_any_eq (vector unsigned short, vector bool short);
9262 int vec_any_eq (vector unsigned short, vector unsigned short);
9263 int vec_any_eq (vector bool short, vector bool short);
9264 int vec_any_eq (vector bool short, vector unsigned short);
9265 int vec_any_eq (vector bool short, vector signed short);
9266 int vec_any_eq (vector pixel, vector pixel);
9267 int vec_any_eq (vector signed int, vector bool int);
9268 int vec_any_eq (vector signed int, vector signed int);
9269 int vec_any_eq (vector unsigned int, vector bool int);
9270 int vec_any_eq (vector unsigned int, vector unsigned int);
9271 int vec_any_eq (vector bool int, vector bool int);
9272 int vec_any_eq (vector bool int, vector unsigned int);
9273 int vec_any_eq (vector bool int, vector signed int);
9274 int vec_any_eq (vector float, vector float);
9276 int vec_any_ge (vector signed char, vector bool char);
9277 int vec_any_ge (vector unsigned char, vector bool char);
9278 int vec_any_ge (vector unsigned char, vector unsigned char);
9279 int vec_any_ge (vector signed char, vector signed char);
9280 int vec_any_ge (vector bool char, vector unsigned char);
9281 int vec_any_ge (vector bool char, vector signed char);
9282 int vec_any_ge (vector unsigned short, vector bool short);
9283 int vec_any_ge (vector unsigned short, vector unsigned short);
9284 int vec_any_ge (vector signed short, vector signed short);
9285 int vec_any_ge (vector signed short, vector bool short);
9286 int vec_any_ge (vector bool short, vector unsigned short);
9287 int vec_any_ge (vector bool short, vector signed short);
9288 int vec_any_ge (vector signed int, vector bool int);
9289 int vec_any_ge (vector unsigned int, vector bool int);
9290 int vec_any_ge (vector unsigned int, vector unsigned int);
9291 int vec_any_ge (vector signed int, vector signed int);
9292 int vec_any_ge (vector bool int, vector unsigned int);
9293 int vec_any_ge (vector bool int, vector signed int);
9294 int vec_any_ge (vector float, vector float);
9296 int vec_any_gt (vector bool char, vector unsigned char);
9297 int vec_any_gt (vector unsigned char, vector bool char);
9298 int vec_any_gt (vector unsigned char, vector unsigned char);
9299 int vec_any_gt (vector bool char, vector signed char);
9300 int vec_any_gt (vector signed char, vector bool char);
9301 int vec_any_gt (vector signed char, vector signed char);
9302 int vec_any_gt (vector bool short, vector unsigned short);
9303 int vec_any_gt (vector unsigned short, vector bool short);
9304 int vec_any_gt (vector unsigned short, vector unsigned short);
9305 int vec_any_gt (vector bool short, vector signed short);
9306 int vec_any_gt (vector signed short, vector bool short);
9307 int vec_any_gt (vector signed short, vector signed short);
9308 int vec_any_gt (vector bool int, vector unsigned int);
9309 int vec_any_gt (vector unsigned int, vector bool int);
9310 int vec_any_gt (vector unsigned int, vector unsigned int);
9311 int vec_any_gt (vector bool int, vector signed int);
9312 int vec_any_gt (vector signed int, vector bool int);
9313 int vec_any_gt (vector signed int, vector signed int);
9314 int vec_any_gt (vector float, vector float);
9316 int vec_any_le (vector bool char, vector unsigned char);
9317 int vec_any_le (vector unsigned char, vector bool char);
9318 int vec_any_le (vector unsigned char, vector unsigned char);
9319 int vec_any_le (vector bool char, vector signed char);
9320 int vec_any_le (vector signed char, vector bool char);
9321 int vec_any_le (vector signed char, vector signed char);
9322 int vec_any_le (vector bool short, vector unsigned short);
9323 int vec_any_le (vector unsigned short, vector bool short);
9324 int vec_any_le (vector unsigned short, vector unsigned short);
9325 int vec_any_le (vector bool short, vector signed short);
9326 int vec_any_le (vector signed short, vector bool short);
9327 int vec_any_le (vector signed short, vector signed short);
9328 int vec_any_le (vector bool int, vector unsigned int);
9329 int vec_any_le (vector unsigned int, vector bool int);
9330 int vec_any_le (vector unsigned int, vector unsigned int);
9331 int vec_any_le (vector bool int, vector signed int);
9332 int vec_any_le (vector signed int, vector bool int);
9333 int vec_any_le (vector signed int, vector signed int);
9334 int vec_any_le (vector float, vector float);
9336 int vec_any_lt (vector bool char, vector unsigned char);
9337 int vec_any_lt (vector unsigned char, vector bool char);
9338 int vec_any_lt (vector unsigned char, vector unsigned char);
9339 int vec_any_lt (vector bool char, vector signed char);
9340 int vec_any_lt (vector signed char, vector bool char);
9341 int vec_any_lt (vector signed char, vector signed char);
9342 int vec_any_lt (vector bool short, vector unsigned short);
9343 int vec_any_lt (vector unsigned short, vector bool short);
9344 int vec_any_lt (vector unsigned short, vector unsigned short);
9345 int vec_any_lt (vector bool short, vector signed short);
9346 int vec_any_lt (vector signed short, vector bool short);
9347 int vec_any_lt (vector signed short, vector signed short);
9348 int vec_any_lt (vector bool int, vector unsigned int);
9349 int vec_any_lt (vector unsigned int, vector bool int);
9350 int vec_any_lt (vector unsigned int, vector unsigned int);
9351 int vec_any_lt (vector bool int, vector signed int);
9352 int vec_any_lt (vector signed int, vector bool int);
9353 int vec_any_lt (vector signed int, vector signed int);
9354 int vec_any_lt (vector float, vector float);
9356 int vec_any_nan (vector float);
9358 int vec_any_ne (vector signed char, vector bool char);
9359 int vec_any_ne (vector signed char, vector signed char);
9360 int vec_any_ne (vector unsigned char, vector bool char);
9361 int vec_any_ne (vector unsigned char, vector unsigned char);
9362 int vec_any_ne (vector bool char, vector bool char);
9363 int vec_any_ne (vector bool char, vector unsigned char);
9364 int vec_any_ne (vector bool char, vector signed char);
9365 int vec_any_ne (vector signed short, vector bool short);
9366 int vec_any_ne (vector signed short, vector signed short);
9367 int vec_any_ne (vector unsigned short, vector bool short);
9368 int vec_any_ne (vector unsigned short, vector unsigned short);
9369 int vec_any_ne (vector bool short, vector bool short);
9370 int vec_any_ne (vector bool short, vector unsigned short);
9371 int vec_any_ne (vector bool short, vector signed short);
9372 int vec_any_ne (vector pixel, vector pixel);
9373 int vec_any_ne (vector signed int, vector bool int);
9374 int vec_any_ne (vector signed int, vector signed int);
9375 int vec_any_ne (vector unsigned int, vector bool int);
9376 int vec_any_ne (vector unsigned int, vector unsigned int);
9377 int vec_any_ne (vector bool int, vector bool int);
9378 int vec_any_ne (vector bool int, vector unsigned int);
9379 int vec_any_ne (vector bool int, vector signed int);
9380 int vec_any_ne (vector float, vector float);
9382 int vec_any_nge (vector float, vector float);
9384 int vec_any_ngt (vector float, vector float);
9386 int vec_any_nle (vector float, vector float);
9388 int vec_any_nlt (vector float, vector float);
9390 int vec_any_numeric (vector float);
9392 int vec_any_out (vector float, vector float);
9395 @node SPARC VIS Built-in Functions
9396 @subsection SPARC VIS Built-in Functions
9398 GCC supports SIMD operations on the SPARC using both the generic vector
9399 extensions (@pxref{Vector Extensions}) as well as built-in functions for
9400 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
9401 switch, the VIS extension is exposed as the following built-in functions:
9404 typedef int v2si __attribute__ ((vector_size (8)));
9405 typedef short v4hi __attribute__ ((vector_size (8)));
9406 typedef short v2hi __attribute__ ((vector_size (4)));
9407 typedef char v8qi __attribute__ ((vector_size (8)));
9408 typedef char v4qi __attribute__ ((vector_size (4)));
9410 void * __builtin_vis_alignaddr (void *, long);
9411 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
9412 v2si __builtin_vis_faligndatav2si (v2si, v2si);
9413 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
9414 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
9416 v4hi __builtin_vis_fexpand (v4qi);
9418 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
9419 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
9420 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
9421 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
9422 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
9423 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
9424 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
9426 v4qi __builtin_vis_fpack16 (v4hi);
9427 v8qi __builtin_vis_fpack32 (v2si, v2si);
9428 v2hi __builtin_vis_fpackfix (v2si);
9429 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
9431 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
9434 @node Target Format Checks
9435 @section Format Checks Specific to Particular Target Machines
9437 For some target machines, GCC supports additional options to the
9439 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
9442 * Solaris Format Checks::
9445 @node Solaris Format Checks
9446 @subsection Solaris Format Checks
9448 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
9449 check. @code{cmn_err} accepts a subset of the standard @code{printf}
9450 conversions, and the two-argument @code{%b} conversion for displaying
9451 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
9454 @section Pragmas Accepted by GCC
9458 GCC supports several types of pragmas, primarily in order to compile
9459 code originally written for other compilers. Note that in general
9460 we do not recommend the use of pragmas; @xref{Function Attributes},
9461 for further explanation.
9466 * RS/6000 and PowerPC Pragmas::
9469 * Symbol-Renaming Pragmas::
9470 * Structure-Packing Pragmas::
9472 * Diagnostic Pragmas::
9473 * Visibility Pragmas::
9477 @subsection ARM Pragmas
9479 The ARM target defines pragmas for controlling the default addition of
9480 @code{long_call} and @code{short_call} attributes to functions.
9481 @xref{Function Attributes}, for information about the effects of these
9486 @cindex pragma, long_calls
9487 Set all subsequent functions to have the @code{long_call} attribute.
9490 @cindex pragma, no_long_calls
9491 Set all subsequent functions to have the @code{short_call} attribute.
9493 @item long_calls_off
9494 @cindex pragma, long_calls_off
9495 Do not affect the @code{long_call} or @code{short_call} attributes of
9496 subsequent functions.
9500 @subsection M32C Pragmas
9503 @item memregs @var{number}
9504 @cindex pragma, memregs
9505 Overrides the command line option @code{-memregs=} for the current
9506 file. Use with care! This pragma must be before any function in the
9507 file, and mixing different memregs values in different objects may
9508 make them incompatible. This pragma is useful when a
9509 performance-critical function uses a memreg for temporary values,
9510 as it may allow you to reduce the number of memregs used.
9514 @node RS/6000 and PowerPC Pragmas
9515 @subsection RS/6000 and PowerPC Pragmas
9517 The RS/6000 and PowerPC targets define one pragma for controlling
9518 whether or not the @code{longcall} attribute is added to function
9519 declarations by default. This pragma overrides the @option{-mlongcall}
9520 option, but not the @code{longcall} and @code{shortcall} attributes.
9521 @xref{RS/6000 and PowerPC Options}, for more information about when long
9522 calls are and are not necessary.
9526 @cindex pragma, longcall
9527 Apply the @code{longcall} attribute to all subsequent function
9531 Do not apply the @code{longcall} attribute to subsequent function
9535 @c Describe c4x pragmas here.
9536 @c Describe h8300 pragmas here.
9537 @c Describe sh pragmas here.
9538 @c Describe v850 pragmas here.
9540 @node Darwin Pragmas
9541 @subsection Darwin Pragmas
9543 The following pragmas are available for all architectures running the
9544 Darwin operating system. These are useful for compatibility with other
9548 @item mark @var{tokens}@dots{}
9549 @cindex pragma, mark
9550 This pragma is accepted, but has no effect.
9552 @item options align=@var{alignment}
9553 @cindex pragma, options align
9554 This pragma sets the alignment of fields in structures. The values of
9555 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
9556 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
9557 properly; to restore the previous setting, use @code{reset} for the
9560 @item segment @var{tokens}@dots{}
9561 @cindex pragma, segment
9562 This pragma is accepted, but has no effect.
9564 @item unused (@var{var} [, @var{var}]@dots{})
9565 @cindex pragma, unused
9566 This pragma declares variables to be possibly unused. GCC will not
9567 produce warnings for the listed variables. The effect is similar to
9568 that of the @code{unused} attribute, except that this pragma may appear
9569 anywhere within the variables' scopes.
9572 @node Solaris Pragmas
9573 @subsection Solaris Pragmas
9575 The Solaris target supports @code{#pragma redefine_extname}
9576 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
9577 @code{#pragma} directives for compatibility with the system compiler.
9580 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
9581 @cindex pragma, align
9583 Increase the minimum alignment of each @var{variable} to @var{alignment}.
9584 This is the same as GCC's @code{aligned} attribute @pxref{Variable
9585 Attributes}). Macro expansion occurs on the arguments to this pragma
9586 when compiling C and Objective-C. It does not currently occur when
9587 compiling C++, but this is a bug which may be fixed in a future
9590 @item fini (@var{function} [, @var{function}]...)
9591 @cindex pragma, fini
9593 This pragma causes each listed @var{function} to be called after
9594 main, or during shared module unloading, by adding a call to the
9595 @code{.fini} section.
9597 @item init (@var{function} [, @var{function}]...)
9598 @cindex pragma, init
9600 This pragma causes each listed @var{function} to be called during
9601 initialization (before @code{main}) or during shared module loading, by
9602 adding a call to the @code{.init} section.
9606 @node Symbol-Renaming Pragmas
9607 @subsection Symbol-Renaming Pragmas
9609 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
9610 supports two @code{#pragma} directives which change the name used in
9611 assembly for a given declaration. These pragmas are only available on
9612 platforms whose system headers need them. To get this effect on all
9613 platforms supported by GCC, use the asm labels extension (@pxref{Asm
9617 @item redefine_extname @var{oldname} @var{newname}
9618 @cindex pragma, redefine_extname
9620 This pragma gives the C function @var{oldname} the assembly symbol
9621 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
9622 will be defined if this pragma is available (currently only on
9625 @item extern_prefix @var{string}
9626 @cindex pragma, extern_prefix
9628 This pragma causes all subsequent external function and variable
9629 declarations to have @var{string} prepended to their assembly symbols.
9630 This effect may be terminated with another @code{extern_prefix} pragma
9631 whose argument is an empty string. The preprocessor macro
9632 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
9633 available (currently only on Tru64 UNIX)@.
9636 These pragmas and the asm labels extension interact in a complicated
9637 manner. Here are some corner cases you may want to be aware of.
9640 @item Both pragmas silently apply only to declarations with external
9641 linkage. Asm labels do not have this restriction.
9643 @item In C++, both pragmas silently apply only to declarations with
9644 ``C'' linkage. Again, asm labels do not have this restriction.
9646 @item If any of the three ways of changing the assembly name of a
9647 declaration is applied to a declaration whose assembly name has
9648 already been determined (either by a previous use of one of these
9649 features, or because the compiler needed the assembly name in order to
9650 generate code), and the new name is different, a warning issues and
9651 the name does not change.
9653 @item The @var{oldname} used by @code{#pragma redefine_extname} is
9654 always the C-language name.
9656 @item If @code{#pragma extern_prefix} is in effect, and a declaration
9657 occurs with an asm label attached, the prefix is silently ignored for
9660 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
9661 apply to the same declaration, whichever triggered first wins, and a
9662 warning issues if they contradict each other. (We would like to have
9663 @code{#pragma redefine_extname} always win, for consistency with asm
9664 labels, but if @code{#pragma extern_prefix} triggers first we have no
9665 way of knowing that that happened.)
9668 @node Structure-Packing Pragmas
9669 @subsection Structure-Packing Pragmas
9671 For compatibility with Win32, GCC supports a set of @code{#pragma}
9672 directives which change the maximum alignment of members of structures
9673 (other than zero-width bitfields), unions, and classes subsequently
9674 defined. The @var{n} value below always is required to be a small power
9675 of two and specifies the new alignment in bytes.
9678 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
9679 @item @code{#pragma pack()} sets the alignment to the one that was in
9680 effect when compilation started (see also command line option
9681 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
9682 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
9683 setting on an internal stack and then optionally sets the new alignment.
9684 @item @code{#pragma pack(pop)} restores the alignment setting to the one
9685 saved at the top of the internal stack (and removes that stack entry).
9686 Note that @code{#pragma pack([@var{n}])} does not influence this internal
9687 stack; thus it is possible to have @code{#pragma pack(push)} followed by
9688 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
9689 @code{#pragma pack(pop)}.
9692 Some targets, e.g. i386 and powerpc, support the @code{ms_struct}
9693 @code{#pragma} which lays out a structure as the documented
9694 @code{__attribute__ ((ms_struct))}.
9696 @item @code{#pragma ms_struct on} turns on the layout for structures
9698 @item @code{#pragma ms_struct off} turns off the layout for structures
9700 @item @code{#pragma ms_struct reset} goes back to the default layout.
9704 @subsection Weak Pragmas
9706 For compatibility with SVR4, GCC supports a set of @code{#pragma}
9707 directives for declaring symbols to be weak, and defining weak
9711 @item #pragma weak @var{symbol}
9712 @cindex pragma, weak
9713 This pragma declares @var{symbol} to be weak, as if the declaration
9714 had the attribute of the same name. The pragma may appear before
9715 or after the declaration of @var{symbol}, but must appear before
9716 either its first use or its definition. It is not an error for
9717 @var{symbol} to never be defined at all.
9719 @item #pragma weak @var{symbol1} = @var{symbol2}
9720 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
9721 It is an error if @var{symbol2} is not defined in the current
9725 @node Diagnostic Pragmas
9726 @subsection Diagnostic Pragmas
9728 GCC allows the user to selectively enable or disable certain types of
9729 diagnostics, and change the kind of the diagnostic. For example, a
9730 project's policy might require that all sources compile with
9731 @option{-Werror} but certain files might have exceptions allowing
9732 specific types of warnings. Or, a project might selectively enable
9733 diagnostics and treat them as errors depending on which preprocessor
9737 @item #pragma GCC diagnostic @var{kind} @var{option}
9738 @cindex pragma, diagnostic
9740 Modifies the disposition of a diagnostic. Note that not all
9741 diagnostics are modifyiable; at the moment only warnings (normally
9742 controlled by @samp{-W...}) can be controlled, and not all of them.
9743 Use @option{-fdiagnostics-show-option} to determine which diagnostics
9744 are controllable and which option controls them.
9746 @var{kind} is @samp{error} to treat this diagnostic as an error,
9747 @samp{warning} to treat it like a warning (even if @option{-Werror} is
9748 in effect), or @samp{ignored} if the diagnostic is to be ignored.
9749 @var{option} is a double quoted string which matches the command line
9753 #pragma GCC diagnostic warning "-Wformat"
9754 #pragma GCC diagnostic error "-Walways-true"
9755 #pragma GCC diagnostic ignored "-Walways-true"
9758 Note that these pragmas override any command line options. Also,
9759 while it is syntactically valid to put these pragmas anywhere in your
9760 sources, the only supported location for them is before any data or
9761 functions are defined. Doing otherwise may result in unpredictable
9762 results depending on how the optimizer manages your sources. If the
9763 same option is listed multiple times, the last one specified is the
9764 one that is in effect. This pragma is not intended to be a general
9765 purpose replacement for command line options, but for implementing
9766 strict control over project policies.
9770 @node Visibility Pragmas
9771 @subsection Visibility Pragmas
9774 @item #pragma GCC visibility push(@var{visibility})
9775 @itemx #pragma GCC visibility pop
9776 @cindex pragma, visibility
9778 This pragma allows the user to set the visibility for multiple
9779 declarations without having to give each a visibility attribute
9780 @xref{Function Attributes}, for more information about visibility and
9781 the attribute syntax.
9783 In C++, @samp{#pragma GCC visibility} affects only namespace-scope
9784 declarations. Class members and template specializations are not
9785 affected; if you want to override the visibility for a particular
9786 member or instantiation, you must use an attribute.
9790 @node Unnamed Fields
9791 @section Unnamed struct/union fields within structs/unions
9795 For compatibility with other compilers, GCC allows you to define
9796 a structure or union that contains, as fields, structures and unions
9797 without names. For example:
9810 In this example, the user would be able to access members of the unnamed
9811 union with code like @samp{foo.b}. Note that only unnamed structs and
9812 unions are allowed, you may not have, for example, an unnamed
9815 You must never create such structures that cause ambiguous field definitions.
9816 For example, this structure:
9827 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
9828 Such constructs are not supported and must be avoided. In the future,
9829 such constructs may be detected and treated as compilation errors.
9831 @opindex fms-extensions
9832 Unless @option{-fms-extensions} is used, the unnamed field must be a
9833 structure or union definition without a tag (for example, @samp{struct
9834 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
9835 also be a definition with a tag such as @samp{struct foo @{ int a;
9836 @};}, a reference to a previously defined structure or union such as
9837 @samp{struct foo;}, or a reference to a @code{typedef} name for a
9838 previously defined structure or union type.
9841 @section Thread-Local Storage
9842 @cindex Thread-Local Storage
9843 @cindex @acronym{TLS}
9846 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
9847 are allocated such that there is one instance of the variable per extant
9848 thread. The run-time model GCC uses to implement this originates
9849 in the IA-64 processor-specific ABI, but has since been migrated
9850 to other processors as well. It requires significant support from
9851 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
9852 system libraries (@file{libc.so} and @file{libpthread.so}), so it
9853 is not available everywhere.
9855 At the user level, the extension is visible with a new storage
9856 class keyword: @code{__thread}. For example:
9860 extern __thread struct state s;
9861 static __thread char *p;
9864 The @code{__thread} specifier may be used alone, with the @code{extern}
9865 or @code{static} specifiers, but with no other storage class specifier.
9866 When used with @code{extern} or @code{static}, @code{__thread} must appear
9867 immediately after the other storage class specifier.
9869 The @code{__thread} specifier may be applied to any global, file-scoped
9870 static, function-scoped static, or static data member of a class. It may
9871 not be applied to block-scoped automatic or non-static data member.
9873 When the address-of operator is applied to a thread-local variable, it is
9874 evaluated at run-time and returns the address of the current thread's
9875 instance of that variable. An address so obtained may be used by any
9876 thread. When a thread terminates, any pointers to thread-local variables
9877 in that thread become invalid.
9879 No static initialization may refer to the address of a thread-local variable.
9881 In C++, if an initializer is present for a thread-local variable, it must
9882 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
9885 See @uref{http://people.redhat.com/drepper/tls.pdf,
9886 ELF Handling For Thread-Local Storage} for a detailed explanation of
9887 the four thread-local storage addressing models, and how the run-time
9888 is expected to function.
9891 * C99 Thread-Local Edits::
9892 * C++98 Thread-Local Edits::
9895 @node C99 Thread-Local Edits
9896 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
9898 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
9899 that document the exact semantics of the language extension.
9903 @cite{5.1.2 Execution environments}
9905 Add new text after paragraph 1
9908 Within either execution environment, a @dfn{thread} is a flow of
9909 control within a program. It is implementation defined whether
9910 or not there may be more than one thread associated with a program.
9911 It is implementation defined how threads beyond the first are
9912 created, the name and type of the function called at thread
9913 startup, and how threads may be terminated. However, objects
9914 with thread storage duration shall be initialized before thread
9919 @cite{6.2.4 Storage durations of objects}
9921 Add new text before paragraph 3
9924 An object whose identifier is declared with the storage-class
9925 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
9926 Its lifetime is the entire execution of the thread, and its
9927 stored value is initialized only once, prior to thread startup.
9931 @cite{6.4.1 Keywords}
9933 Add @code{__thread}.
9936 @cite{6.7.1 Storage-class specifiers}
9938 Add @code{__thread} to the list of storage class specifiers in
9941 Change paragraph 2 to
9944 With the exception of @code{__thread}, at most one storage-class
9945 specifier may be given [@dots{}]. The @code{__thread} specifier may
9946 be used alone, or immediately following @code{extern} or
9950 Add new text after paragraph 6
9953 The declaration of an identifier for a variable that has
9954 block scope that specifies @code{__thread} shall also
9955 specify either @code{extern} or @code{static}.
9957 The @code{__thread} specifier shall be used only with
9962 @node C++98 Thread-Local Edits
9963 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
9965 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
9966 that document the exact semantics of the language extension.
9970 @b{[intro.execution]}
9972 New text after paragraph 4
9975 A @dfn{thread} is a flow of control within the abstract machine.
9976 It is implementation defined whether or not there may be more than
9980 New text after paragraph 7
9983 It is unspecified whether additional action must be taken to
9984 ensure when and whether side effects are visible to other threads.
9990 Add @code{__thread}.
9993 @b{[basic.start.main]}
9995 Add after paragraph 5
9998 The thread that begins execution at the @code{main} function is called
9999 the @dfn{main thread}. It is implementation defined how functions
10000 beginning threads other than the main thread are designated or typed.
10001 A function so designated, as well as the @code{main} function, is called
10002 a @dfn{thread startup function}. It is implementation defined what
10003 happens if a thread startup function returns. It is implementation
10004 defined what happens to other threads when any thread calls @code{exit}.
10008 @b{[basic.start.init]}
10010 Add after paragraph 4
10013 The storage for an object of thread storage duration shall be
10014 statically initialized before the first statement of the thread startup
10015 function. An object of thread storage duration shall not require
10016 dynamic initialization.
10020 @b{[basic.start.term]}
10022 Add after paragraph 3
10025 The type of an object with thread storage duration shall not have a
10026 non-trivial destructor, nor shall it be an array type whose elements
10027 (directly or indirectly) have non-trivial destructors.
10033 Add ``thread storage duration'' to the list in paragraph 1.
10038 Thread, static, and automatic storage durations are associated with
10039 objects introduced by declarations [@dots{}].
10042 Add @code{__thread} to the list of specifiers in paragraph 3.
10045 @b{[basic.stc.thread]}
10047 New section before @b{[basic.stc.static]}
10050 The keyword @code{__thread} applied to a non-local object gives the
10051 object thread storage duration.
10053 A local variable or class data member declared both @code{static}
10054 and @code{__thread} gives the variable or member thread storage
10059 @b{[basic.stc.static]}
10064 All objects which have neither thread storage duration, dynamic
10065 storage duration nor are local [@dots{}].
10071 Add @code{__thread} to the list in paragraph 1.
10076 With the exception of @code{__thread}, at most one
10077 @var{storage-class-specifier} shall appear in a given
10078 @var{decl-specifier-seq}. The @code{__thread} specifier may
10079 be used alone, or immediately following the @code{extern} or
10080 @code{static} specifiers. [@dots{}]
10083 Add after paragraph 5
10086 The @code{__thread} specifier can be applied only to the names of objects
10087 and to anonymous unions.
10093 Add after paragraph 6
10096 Non-@code{static} members shall not be @code{__thread}.
10100 @node C++ Extensions
10101 @chapter Extensions to the C++ Language
10102 @cindex extensions, C++ language
10103 @cindex C++ language extensions
10105 The GNU compiler provides these extensions to the C++ language (and you
10106 can also use most of the C language extensions in your C++ programs). If you
10107 want to write code that checks whether these features are available, you can
10108 test for the GNU compiler the same way as for C programs: check for a
10109 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
10110 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
10111 Predefined Macros,cpp,The GNU C Preprocessor}).
10114 * Volatiles:: What constitutes an access to a volatile object.
10115 * Restricted Pointers:: C99 restricted pointers and references.
10116 * Vague Linkage:: Where G++ puts inlines, vtables and such.
10117 * C++ Interface:: You can use a single C++ header file for both
10118 declarations and definitions.
10119 * Template Instantiation:: Methods for ensuring that exactly one copy of
10120 each needed template instantiation is emitted.
10121 * Bound member functions:: You can extract a function pointer to the
10122 method denoted by a @samp{->*} or @samp{.*} expression.
10123 * C++ Attributes:: Variable, function, and type attributes for C++ only.
10124 * Namespace Association:: Strong using-directives for namespace association.
10125 * Java Exceptions:: Tweaking exception handling to work with Java.
10126 * Deprecated Features:: Things will disappear from g++.
10127 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
10131 @section When is a Volatile Object Accessed?
10132 @cindex accessing volatiles
10133 @cindex volatile read
10134 @cindex volatile write
10135 @cindex volatile access
10137 Both the C and C++ standard have the concept of volatile objects. These
10138 are normally accessed by pointers and used for accessing hardware. The
10139 standards encourage compilers to refrain from optimizations
10140 concerning accesses to volatile objects that it might perform on
10141 non-volatile objects. The C standard leaves it implementation defined
10142 as to what constitutes a volatile access. The C++ standard omits to
10143 specify this, except to say that C++ should behave in a similar manner
10144 to C with respect to volatiles, where possible. The minimum either
10145 standard specifies is that at a sequence point all previous accesses to
10146 volatile objects have stabilized and no subsequent accesses have
10147 occurred. Thus an implementation is free to reorder and combine
10148 volatile accesses which occur between sequence points, but cannot do so
10149 for accesses across a sequence point. The use of volatiles does not
10150 allow you to violate the restriction on updating objects multiple times
10151 within a sequence point.
10153 In most expressions, it is intuitively obvious what is a read and what is
10154 a write. For instance
10157 volatile int *dst = @var{somevalue};
10158 volatile int *src = @var{someothervalue};
10163 will cause a read of the volatile object pointed to by @var{src} and stores the
10164 value into the volatile object pointed to by @var{dst}. There is no
10165 guarantee that these reads and writes are atomic, especially for objects
10166 larger than @code{int}.
10168 Less obvious expressions are where something which looks like an access
10169 is used in a void context. An example would be,
10172 volatile int *src = @var{somevalue};
10176 With C, such expressions are rvalues, and as rvalues cause a read of
10177 the object, GCC interprets this as a read of the volatile being pointed
10178 to. The C++ standard specifies that such expressions do not undergo
10179 lvalue to rvalue conversion, and that the type of the dereferenced
10180 object may be incomplete. The C++ standard does not specify explicitly
10181 that it is this lvalue to rvalue conversion which is responsible for
10182 causing an access. However, there is reason to believe that it is,
10183 because otherwise certain simple expressions become undefined. However,
10184 because it would surprise most programmers, G++ treats dereferencing a
10185 pointer to volatile object of complete type in a void context as a read
10186 of the object. When the object has incomplete type, G++ issues a
10191 struct T @{int m;@};
10192 volatile S *ptr1 = @var{somevalue};
10193 volatile T *ptr2 = @var{somevalue};
10198 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
10199 causes a read of the object pointed to. If you wish to force an error on
10200 the first case, you must force a conversion to rvalue with, for instance
10201 a static cast, @code{static_cast<S>(*ptr1)}.
10203 When using a reference to volatile, G++ does not treat equivalent
10204 expressions as accesses to volatiles, but instead issues a warning that
10205 no volatile is accessed. The rationale for this is that otherwise it
10206 becomes difficult to determine where volatile access occur, and not
10207 possible to ignore the return value from functions returning volatile
10208 references. Again, if you wish to force a read, cast the reference to
10211 @node Restricted Pointers
10212 @section Restricting Pointer Aliasing
10213 @cindex restricted pointers
10214 @cindex restricted references
10215 @cindex restricted this pointer
10217 As with the C front end, G++ understands the C99 feature of restricted pointers,
10218 specified with the @code{__restrict__}, or @code{__restrict} type
10219 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
10220 language flag, @code{restrict} is not a keyword in C++.
10222 In addition to allowing restricted pointers, you can specify restricted
10223 references, which indicate that the reference is not aliased in the local
10227 void fn (int *__restrict__ rptr, int &__restrict__ rref)
10234 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
10235 @var{rref} refers to a (different) unaliased integer.
10237 You may also specify whether a member function's @var{this} pointer is
10238 unaliased by using @code{__restrict__} as a member function qualifier.
10241 void T::fn () __restrict__
10248 Within the body of @code{T::fn}, @var{this} will have the effective
10249 definition @code{T *__restrict__ const this}. Notice that the
10250 interpretation of a @code{__restrict__} member function qualifier is
10251 different to that of @code{const} or @code{volatile} qualifier, in that it
10252 is applied to the pointer rather than the object. This is consistent with
10253 other compilers which implement restricted pointers.
10255 As with all outermost parameter qualifiers, @code{__restrict__} is
10256 ignored in function definition matching. This means you only need to
10257 specify @code{__restrict__} in a function definition, rather than
10258 in a function prototype as well.
10260 @node Vague Linkage
10261 @section Vague Linkage
10262 @cindex vague linkage
10264 There are several constructs in C++ which require space in the object
10265 file but are not clearly tied to a single translation unit. We say that
10266 these constructs have ``vague linkage''. Typically such constructs are
10267 emitted wherever they are needed, though sometimes we can be more
10271 @item Inline Functions
10272 Inline functions are typically defined in a header file which can be
10273 included in many different compilations. Hopefully they can usually be
10274 inlined, but sometimes an out-of-line copy is necessary, if the address
10275 of the function is taken or if inlining fails. In general, we emit an
10276 out-of-line copy in all translation units where one is needed. As an
10277 exception, we only emit inline virtual functions with the vtable, since
10278 it will always require a copy.
10280 Local static variables and string constants used in an inline function
10281 are also considered to have vague linkage, since they must be shared
10282 between all inlined and out-of-line instances of the function.
10286 C++ virtual functions are implemented in most compilers using a lookup
10287 table, known as a vtable. The vtable contains pointers to the virtual
10288 functions provided by a class, and each object of the class contains a
10289 pointer to its vtable (or vtables, in some multiple-inheritance
10290 situations). If the class declares any non-inline, non-pure virtual
10291 functions, the first one is chosen as the ``key method'' for the class,
10292 and the vtable is only emitted in the translation unit where the key
10295 @emph{Note:} If the chosen key method is later defined as inline, the
10296 vtable will still be emitted in every translation unit which defines it.
10297 Make sure that any inline virtuals are declared inline in the class
10298 body, even if they are not defined there.
10300 @item type_info objects
10303 C++ requires information about types to be written out in order to
10304 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
10305 For polymorphic classes (classes with virtual functions), the type_info
10306 object is written out along with the vtable so that @samp{dynamic_cast}
10307 can determine the dynamic type of a class object at runtime. For all
10308 other types, we write out the type_info object when it is used: when
10309 applying @samp{typeid} to an expression, throwing an object, or
10310 referring to a type in a catch clause or exception specification.
10312 @item Template Instantiations
10313 Most everything in this section also applies to template instantiations,
10314 but there are other options as well.
10315 @xref{Template Instantiation,,Where's the Template?}.
10319 When used with GNU ld version 2.8 or later on an ELF system such as
10320 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
10321 these constructs will be discarded at link time. This is known as
10324 On targets that don't support COMDAT, but do support weak symbols, GCC
10325 will use them. This way one copy will override all the others, but
10326 the unused copies will still take up space in the executable.
10328 For targets which do not support either COMDAT or weak symbols,
10329 most entities with vague linkage will be emitted as local symbols to
10330 avoid duplicate definition errors from the linker. This will not happen
10331 for local statics in inlines, however, as having multiple copies will
10332 almost certainly break things.
10334 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
10335 another way to control placement of these constructs.
10337 @node C++ Interface
10338 @section #pragma interface and implementation
10340 @cindex interface and implementation headers, C++
10341 @cindex C++ interface and implementation headers
10342 @cindex pragmas, interface and implementation
10344 @code{#pragma interface} and @code{#pragma implementation} provide the
10345 user with a way of explicitly directing the compiler to emit entities
10346 with vague linkage (and debugging information) in a particular
10349 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
10350 most cases, because of COMDAT support and the ``key method'' heuristic
10351 mentioned in @ref{Vague Linkage}. Using them can actually cause your
10352 program to grow due to unnecessary out-of-line copies of inline
10353 functions. Currently (3.4) the only benefit of these
10354 @code{#pragma}s is reduced duplication of debugging information, and
10355 that should be addressed soon on DWARF 2 targets with the use of
10359 @item #pragma interface
10360 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
10361 @kindex #pragma interface
10362 Use this directive in @emph{header files} that define object classes, to save
10363 space in most of the object files that use those classes. Normally,
10364 local copies of certain information (backup copies of inline member
10365 functions, debugging information, and the internal tables that implement
10366 virtual functions) must be kept in each object file that includes class
10367 definitions. You can use this pragma to avoid such duplication. When a
10368 header file containing @samp{#pragma interface} is included in a
10369 compilation, this auxiliary information will not be generated (unless
10370 the main input source file itself uses @samp{#pragma implementation}).
10371 Instead, the object files will contain references to be resolved at link
10374 The second form of this directive is useful for the case where you have
10375 multiple headers with the same name in different directories. If you
10376 use this form, you must specify the same string to @samp{#pragma
10379 @item #pragma implementation
10380 @itemx #pragma implementation "@var{objects}.h"
10381 @kindex #pragma implementation
10382 Use this pragma in a @emph{main input file}, when you want full output from
10383 included header files to be generated (and made globally visible). The
10384 included header file, in turn, should use @samp{#pragma interface}.
10385 Backup copies of inline member functions, debugging information, and the
10386 internal tables used to implement virtual functions are all generated in
10387 implementation files.
10389 @cindex implied @code{#pragma implementation}
10390 @cindex @code{#pragma implementation}, implied
10391 @cindex naming convention, implementation headers
10392 If you use @samp{#pragma implementation} with no argument, it applies to
10393 an include file with the same basename@footnote{A file's @dfn{basename}
10394 was the name stripped of all leading path information and of trailing
10395 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
10396 file. For example, in @file{allclass.cc}, giving just
10397 @samp{#pragma implementation}
10398 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
10400 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
10401 an implementation file whenever you would include it from
10402 @file{allclass.cc} even if you never specified @samp{#pragma
10403 implementation}. This was deemed to be more trouble than it was worth,
10404 however, and disabled.
10406 Use the string argument if you want a single implementation file to
10407 include code from multiple header files. (You must also use
10408 @samp{#include} to include the header file; @samp{#pragma
10409 implementation} only specifies how to use the file---it doesn't actually
10412 There is no way to split up the contents of a single header file into
10413 multiple implementation files.
10416 @cindex inlining and C++ pragmas
10417 @cindex C++ pragmas, effect on inlining
10418 @cindex pragmas in C++, effect on inlining
10419 @samp{#pragma implementation} and @samp{#pragma interface} also have an
10420 effect on function inlining.
10422 If you define a class in a header file marked with @samp{#pragma
10423 interface}, the effect on an inline function defined in that class is
10424 similar to an explicit @code{extern} declaration---the compiler emits
10425 no code at all to define an independent version of the function. Its
10426 definition is used only for inlining with its callers.
10428 @opindex fno-implement-inlines
10429 Conversely, when you include the same header file in a main source file
10430 that declares it as @samp{#pragma implementation}, the compiler emits
10431 code for the function itself; this defines a version of the function
10432 that can be found via pointers (or by callers compiled without
10433 inlining). If all calls to the function can be inlined, you can avoid
10434 emitting the function by compiling with @option{-fno-implement-inlines}.
10435 If any calls were not inlined, you will get linker errors.
10437 @node Template Instantiation
10438 @section Where's the Template?
10439 @cindex template instantiation
10441 C++ templates are the first language feature to require more
10442 intelligence from the environment than one usually finds on a UNIX
10443 system. Somehow the compiler and linker have to make sure that each
10444 template instance occurs exactly once in the executable if it is needed,
10445 and not at all otherwise. There are two basic approaches to this
10446 problem, which are referred to as the Borland model and the Cfront model.
10449 @item Borland model
10450 Borland C++ solved the template instantiation problem by adding the code
10451 equivalent of common blocks to their linker; the compiler emits template
10452 instances in each translation unit that uses them, and the linker
10453 collapses them together. The advantage of this model is that the linker
10454 only has to consider the object files themselves; there is no external
10455 complexity to worry about. This disadvantage is that compilation time
10456 is increased because the template code is being compiled repeatedly.
10457 Code written for this model tends to include definitions of all
10458 templates in the header file, since they must be seen to be
10462 The AT&T C++ translator, Cfront, solved the template instantiation
10463 problem by creating the notion of a template repository, an
10464 automatically maintained place where template instances are stored. A
10465 more modern version of the repository works as follows: As individual
10466 object files are built, the compiler places any template definitions and
10467 instantiations encountered in the repository. At link time, the link
10468 wrapper adds in the objects in the repository and compiles any needed
10469 instances that were not previously emitted. The advantages of this
10470 model are more optimal compilation speed and the ability to use the
10471 system linker; to implement the Borland model a compiler vendor also
10472 needs to replace the linker. The disadvantages are vastly increased
10473 complexity, and thus potential for error; for some code this can be
10474 just as transparent, but in practice it can been very difficult to build
10475 multiple programs in one directory and one program in multiple
10476 directories. Code written for this model tends to separate definitions
10477 of non-inline member templates into a separate file, which should be
10478 compiled separately.
10481 When used with GNU ld version 2.8 or later on an ELF system such as
10482 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
10483 Borland model. On other systems, G++ implements neither automatic
10486 A future version of G++ will support a hybrid model whereby the compiler
10487 will emit any instantiations for which the template definition is
10488 included in the compile, and store template definitions and
10489 instantiation context information into the object file for the rest.
10490 The link wrapper will extract that information as necessary and invoke
10491 the compiler to produce the remaining instantiations. The linker will
10492 then combine duplicate instantiations.
10494 In the mean time, you have the following options for dealing with
10495 template instantiations:
10500 Compile your template-using code with @option{-frepo}. The compiler will
10501 generate files with the extension @samp{.rpo} listing all of the
10502 template instantiations used in the corresponding object files which
10503 could be instantiated there; the link wrapper, @samp{collect2}, will
10504 then update the @samp{.rpo} files to tell the compiler where to place
10505 those instantiations and rebuild any affected object files. The
10506 link-time overhead is negligible after the first pass, as the compiler
10507 will continue to place the instantiations in the same files.
10509 This is your best option for application code written for the Borland
10510 model, as it will just work. Code written for the Cfront model will
10511 need to be modified so that the template definitions are available at
10512 one or more points of instantiation; usually this is as simple as adding
10513 @code{#include <tmethods.cc>} to the end of each template header.
10515 For library code, if you want the library to provide all of the template
10516 instantiations it needs, just try to link all of its object files
10517 together; the link will fail, but cause the instantiations to be
10518 generated as a side effect. Be warned, however, that this may cause
10519 conflicts if multiple libraries try to provide the same instantiations.
10520 For greater control, use explicit instantiation as described in the next
10524 @opindex fno-implicit-templates
10525 Compile your code with @option{-fno-implicit-templates} to disable the
10526 implicit generation of template instances, and explicitly instantiate
10527 all the ones you use. This approach requires more knowledge of exactly
10528 which instances you need than do the others, but it's less
10529 mysterious and allows greater control. You can scatter the explicit
10530 instantiations throughout your program, perhaps putting them in the
10531 translation units where the instances are used or the translation units
10532 that define the templates themselves; you can put all of the explicit
10533 instantiations you need into one big file; or you can create small files
10540 template class Foo<int>;
10541 template ostream& operator <<
10542 (ostream&, const Foo<int>&);
10545 for each of the instances you need, and create a template instantiation
10546 library from those.
10548 If you are using Cfront-model code, you can probably get away with not
10549 using @option{-fno-implicit-templates} when compiling files that don't
10550 @samp{#include} the member template definitions.
10552 If you use one big file to do the instantiations, you may want to
10553 compile it without @option{-fno-implicit-templates} so you get all of the
10554 instances required by your explicit instantiations (but not by any
10555 other files) without having to specify them as well.
10557 G++ has extended the template instantiation syntax given in the ISO
10558 standard to allow forward declaration of explicit instantiations
10559 (with @code{extern}), instantiation of the compiler support data for a
10560 template class (i.e.@: the vtable) without instantiating any of its
10561 members (with @code{inline}), and instantiation of only the static data
10562 members of a template class, without the support data or member
10563 functions (with (@code{static}):
10566 extern template int max (int, int);
10567 inline template class Foo<int>;
10568 static template class Foo<int>;
10572 Do nothing. Pretend G++ does implement automatic instantiation
10573 management. Code written for the Borland model will work fine, but
10574 each translation unit will contain instances of each of the templates it
10575 uses. In a large program, this can lead to an unacceptable amount of code
10579 @node Bound member functions
10580 @section Extracting the function pointer from a bound pointer to member function
10582 @cindex pointer to member function
10583 @cindex bound pointer to member function
10585 In C++, pointer to member functions (PMFs) are implemented using a wide
10586 pointer of sorts to handle all the possible call mechanisms; the PMF
10587 needs to store information about how to adjust the @samp{this} pointer,
10588 and if the function pointed to is virtual, where to find the vtable, and
10589 where in the vtable to look for the member function. If you are using
10590 PMFs in an inner loop, you should really reconsider that decision. If
10591 that is not an option, you can extract the pointer to the function that
10592 would be called for a given object/PMF pair and call it directly inside
10593 the inner loop, to save a bit of time.
10595 Note that you will still be paying the penalty for the call through a
10596 function pointer; on most modern architectures, such a call defeats the
10597 branch prediction features of the CPU@. This is also true of normal
10598 virtual function calls.
10600 The syntax for this extension is
10604 extern int (A::*fp)();
10605 typedef int (*fptr)(A *);
10607 fptr p = (fptr)(a.*fp);
10610 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
10611 no object is needed to obtain the address of the function. They can be
10612 converted to function pointers directly:
10615 fptr p1 = (fptr)(&A::foo);
10618 @opindex Wno-pmf-conversions
10619 You must specify @option{-Wno-pmf-conversions} to use this extension.
10621 @node C++ Attributes
10622 @section C++-Specific Variable, Function, and Type Attributes
10624 Some attributes only make sense for C++ programs.
10627 @item init_priority (@var{priority})
10628 @cindex init_priority attribute
10631 In Standard C++, objects defined at namespace scope are guaranteed to be
10632 initialized in an order in strict accordance with that of their definitions
10633 @emph{in a given translation unit}. No guarantee is made for initializations
10634 across translation units. However, GNU C++ allows users to control the
10635 order of initialization of objects defined at namespace scope with the
10636 @code{init_priority} attribute by specifying a relative @var{priority},
10637 a constant integral expression currently bounded between 101 and 65535
10638 inclusive. Lower numbers indicate a higher priority.
10640 In the following example, @code{A} would normally be created before
10641 @code{B}, but the @code{init_priority} attribute has reversed that order:
10644 Some_Class A __attribute__ ((init_priority (2000)));
10645 Some_Class B __attribute__ ((init_priority (543)));
10649 Note that the particular values of @var{priority} do not matter; only their
10652 @item java_interface
10653 @cindex java_interface attribute
10655 This type attribute informs C++ that the class is a Java interface. It may
10656 only be applied to classes declared within an @code{extern "Java"} block.
10657 Calls to methods declared in this interface will be dispatched using GCJ's
10658 interface table mechanism, instead of regular virtual table dispatch.
10662 See also @xref{Namespace Association}.
10664 @node Namespace Association
10665 @section Namespace Association
10667 @strong{Caution:} The semantics of this extension are not fully
10668 defined. Users should refrain from using this extension as its
10669 semantics may change subtly over time. It is possible that this
10670 extension will be removed in future versions of G++.
10672 A using-directive with @code{__attribute ((strong))} is stronger
10673 than a normal using-directive in two ways:
10677 Templates from the used namespace can be specialized and explicitly
10678 instantiated as though they were members of the using namespace.
10681 The using namespace is considered an associated namespace of all
10682 templates in the used namespace for purposes of argument-dependent
10686 The used namespace must be nested within the using namespace so that
10687 normal unqualified lookup works properly.
10689 This is useful for composing a namespace transparently from
10690 implementation namespaces. For example:
10695 template <class T> struct A @{ @};
10697 using namespace debug __attribute ((__strong__));
10698 template <> struct A<int> @{ @}; // @r{ok to specialize}
10700 template <class T> void f (A<T>);
10705 f (std::A<float>()); // @r{lookup finds} std::f
10710 @node Java Exceptions
10711 @section Java Exceptions
10713 The Java language uses a slightly different exception handling model
10714 from C++. Normally, GNU C++ will automatically detect when you are
10715 writing C++ code that uses Java exceptions, and handle them
10716 appropriately. However, if C++ code only needs to execute destructors
10717 when Java exceptions are thrown through it, GCC will guess incorrectly.
10718 Sample problematic code is:
10721 struct S @{ ~S(); @};
10722 extern void bar(); // @r{is written in Java, and may throw exceptions}
10731 The usual effect of an incorrect guess is a link failure, complaining of
10732 a missing routine called @samp{__gxx_personality_v0}.
10734 You can inform the compiler that Java exceptions are to be used in a
10735 translation unit, irrespective of what it might think, by writing
10736 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
10737 @samp{#pragma} must appear before any functions that throw or catch
10738 exceptions, or run destructors when exceptions are thrown through them.
10740 You cannot mix Java and C++ exceptions in the same translation unit. It
10741 is believed to be safe to throw a C++ exception from one file through
10742 another file compiled for the Java exception model, or vice versa, but
10743 there may be bugs in this area.
10745 @node Deprecated Features
10746 @section Deprecated Features
10748 In the past, the GNU C++ compiler was extended to experiment with new
10749 features, at a time when the C++ language was still evolving. Now that
10750 the C++ standard is complete, some of those features are superseded by
10751 superior alternatives. Using the old features might cause a warning in
10752 some cases that the feature will be dropped in the future. In other
10753 cases, the feature might be gone already.
10755 While the list below is not exhaustive, it documents some of the options
10756 that are now deprecated:
10759 @item -fexternal-templates
10760 @itemx -falt-external-templates
10761 These are two of the many ways for G++ to implement template
10762 instantiation. @xref{Template Instantiation}. The C++ standard clearly
10763 defines how template definitions have to be organized across
10764 implementation units. G++ has an implicit instantiation mechanism that
10765 should work just fine for standard-conforming code.
10767 @item -fstrict-prototype
10768 @itemx -fno-strict-prototype
10769 Previously it was possible to use an empty prototype parameter list to
10770 indicate an unspecified number of parameters (like C), rather than no
10771 parameters, as C++ demands. This feature has been removed, except where
10772 it is required for backwards compatibility @xref{Backwards Compatibility}.
10775 G++ allows a virtual function returning @samp{void *} to be overridden
10776 by one returning a different pointer type. This extension to the
10777 covariant return type rules is now deprecated and will be removed from a
10780 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
10781 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
10782 and will be removed in a future version. Code using these operators
10783 should be modified to use @code{std::min} and @code{std::max} instead.
10785 The named return value extension has been deprecated, and is now
10788 The use of initializer lists with new expressions has been deprecated,
10789 and is now removed from G++.
10791 Floating and complex non-type template parameters have been deprecated,
10792 and are now removed from G++.
10794 The implicit typename extension has been deprecated and is now
10797 The use of default arguments in function pointers, function typedefs and
10798 and other places where they are not permitted by the standard is
10799 deprecated and will be removed from a future version of G++.
10801 G++ allows floating-point literals to appear in integral constant expressions,
10802 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
10803 This extension is deprecated and will be removed from a future version.
10805 G++ allows static data members of const floating-point type to be declared
10806 with an initializer in a class definition. The standard only allows
10807 initializers for static members of const integral types and const
10808 enumeration types so this extension has been deprecated and will be removed
10809 from a future version.
10811 @node Backwards Compatibility
10812 @section Backwards Compatibility
10813 @cindex Backwards Compatibility
10814 @cindex ARM [Annotated C++ Reference Manual]
10816 Now that there is a definitive ISO standard C++, G++ has a specification
10817 to adhere to. The C++ language evolved over time, and features that
10818 used to be acceptable in previous drafts of the standard, such as the ARM
10819 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
10820 compilation of C++ written to such drafts, G++ contains some backwards
10821 compatibilities. @emph{All such backwards compatibility features are
10822 liable to disappear in future versions of G++.} They should be considered
10823 deprecated @xref{Deprecated Features}.
10827 If a variable is declared at for scope, it used to remain in scope until
10828 the end of the scope which contained the for statement (rather than just
10829 within the for scope). G++ retains this, but issues a warning, if such a
10830 variable is accessed outside the for scope.
10832 @item Implicit C language
10833 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
10834 scope to set the language. On such systems, all header files are
10835 implicitly scoped inside a C language scope. Also, an empty prototype
10836 @code{()} will be treated as an unspecified number of arguments, rather
10837 than no arguments, as C++ demands.