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 causes the compiler to always call the
1966 function by first loading its address into a register and then using the
1967 contents of that register. 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 causes
1974 the compiler to always call this function via a pointer, just as it would if
1975 the @option{-mlongcall} option had been specified. The @code{shortcall}
1976 attribute causes the compiler not to do this. These attributes override
1977 both the @option{-mlongcall} switch and, on the RS/6000 and PowerPC, the
1978 @code{#pragma longcall} setting.
1980 @xref{RS/6000 and PowerPC Options}, for more information on whether long
1981 calls are necessary.
1984 @cindex indirect calls on MIPS
1985 This attribute specifies how a particular function is called on MIPS@.
1986 The attribute overrides the @option{-mlong-calls} (@pxref{MIPS Options})
1987 command line switch. This attribute causes the compiler to always call
1988 the function by first loading its address into a register, and then using
1989 the contents of that register.
1992 @cindex @code{malloc} attribute
1993 The @code{malloc} attribute is used to tell the compiler that a function
1994 may be treated as if any non-@code{NULL} pointer it returns cannot
1995 alias any other pointer valid when the function returns.
1996 This will often improve optimization.
1997 Standard functions with this property include @code{malloc} and
1998 @code{calloc}. @code{realloc}-like functions have this property as
1999 long as the old pointer is never referred to (including comparing it
2000 to the new pointer) after the function returns a non-@code{NULL}
2003 @item model (@var{model-name})
2004 @cindex function addressability on the M32R/D
2005 @cindex variable addressability on the IA-64
2007 On the M32R/D, use this attribute to set the addressability of an
2008 object, and of the code generated for a function. The identifier
2009 @var{model-name} is one of @code{small}, @code{medium}, or
2010 @code{large}, representing each of the code models.
2012 Small model objects live in the lower 16MB of memory (so that their
2013 addresses can be loaded with the @code{ld24} instruction), and are
2014 callable with the @code{bl} instruction.
2016 Medium model objects may live anywhere in the 32-bit address space (the
2017 compiler will generate @code{seth/add3} instructions to load their addresses),
2018 and are callable with the @code{bl} instruction.
2020 Large model objects may live anywhere in the 32-bit address space (the
2021 compiler will generate @code{seth/add3} instructions to load their addresses),
2022 and may not be reachable with the @code{bl} instruction (the compiler will
2023 generate the much slower @code{seth/add3/jl} instruction sequence).
2025 On IA-64, use this attribute to set the addressability of an object.
2026 At present, the only supported identifier for @var{model-name} is
2027 @code{small}, indicating addressability via ``small'' (22-bit)
2028 addresses (so that their addresses can be loaded with the @code{addl}
2029 instruction). Caveat: such addressing is by definition not position
2030 independent and hence this attribute must not be used for objects
2031 defined by shared libraries.
2034 @cindex function without a prologue/epilogue code
2035 Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the
2036 specified function does not need prologue/epilogue sequences generated by
2037 the compiler. It is up to the programmer to provide these sequences.
2040 @cindex functions which do not handle memory bank switching on 68HC11/68HC12
2041 On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to
2042 use the normal calling convention based on @code{jsr} and @code{rts}.
2043 This attribute can be used to cancel the effect of the @option{-mlong-calls}
2047 @cindex Allow nesting in an interrupt handler on the Blackfin processor.
2048 Use this attribute together with @code{interrupt_handler},
2049 @code{exception_handler} or @code{nmi_handler} to indicate that the function
2050 entry code should enable nested interrupts or exceptions.
2053 @cindex NMI handler functions on the Blackfin processor
2054 Use this attribute on the Blackfin to indicate that the specified function
2055 is an NMI handler. The compiler will generate function entry and
2056 exit sequences suitable for use in an NMI handler when this
2057 attribute is present.
2059 @item no_instrument_function
2060 @cindex @code{no_instrument_function} function attribute
2061 @opindex finstrument-functions
2062 If @option{-finstrument-functions} is given, profiling function calls will
2063 be generated at entry and exit of most user-compiled functions.
2064 Functions with this attribute will not be so instrumented.
2067 @cindex @code{noinline} function attribute
2068 This function attribute prevents a function from being considered for
2071 @item nonnull (@var{arg-index}, @dots{})
2072 @cindex @code{nonnull} function attribute
2073 The @code{nonnull} attribute specifies that some function parameters should
2074 be non-null pointers. For instance, the declaration:
2078 my_memcpy (void *dest, const void *src, size_t len)
2079 __attribute__((nonnull (1, 2)));
2083 causes the compiler to check that, in calls to @code{my_memcpy},
2084 arguments @var{dest} and @var{src} are non-null. If the compiler
2085 determines that a null pointer is passed in an argument slot marked
2086 as non-null, and the @option{-Wnonnull} option is enabled, a warning
2087 is issued. The compiler may also choose to make optimizations based
2088 on the knowledge that certain function arguments will not be null.
2090 If no argument index list is given to the @code{nonnull} attribute,
2091 all pointer arguments are marked as non-null. To illustrate, the
2092 following declaration is equivalent to the previous example:
2096 my_memcpy (void *dest, const void *src, size_t len)
2097 __attribute__((nonnull));
2101 @cindex @code{noreturn} function attribute
2102 A few standard library functions, such as @code{abort} and @code{exit},
2103 cannot return. GCC knows this automatically. Some programs define
2104 their own functions that never return. You can declare them
2105 @code{noreturn} to tell the compiler this fact. For example,
2109 void fatal () __attribute__ ((noreturn));
2112 fatal (/* @r{@dots{}} */)
2114 /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */
2120 The @code{noreturn} keyword tells the compiler to assume that
2121 @code{fatal} cannot return. It can then optimize without regard to what
2122 would happen if @code{fatal} ever did return. This makes slightly
2123 better code. More importantly, it helps avoid spurious warnings of
2124 uninitialized variables.
2126 The @code{noreturn} keyword does not affect the exceptional path when that
2127 applies: a @code{noreturn}-marked function may still return to the caller
2128 by throwing an exception or calling @code{longjmp}.
2130 Do not assume that registers saved by the calling function are
2131 restored before calling the @code{noreturn} function.
2133 It does not make sense for a @code{noreturn} function to have a return
2134 type other than @code{void}.
2136 The attribute @code{noreturn} is not implemented in GCC versions
2137 earlier than 2.5. An alternative way to declare that a function does
2138 not return, which works in the current version and in some older
2139 versions, is as follows:
2142 typedef void voidfn ();
2144 volatile voidfn fatal;
2147 This approach does not work in GNU C++.
2150 @cindex @code{nothrow} function attribute
2151 The @code{nothrow} attribute is used to inform the compiler that a
2152 function cannot throw an exception. For example, most functions in
2153 the standard C library can be guaranteed not to throw an exception
2154 with the notable exceptions of @code{qsort} and @code{bsearch} that
2155 take function pointer arguments. The @code{nothrow} attribute is not
2156 implemented in GCC versions earlier than 3.3.
2159 @cindex @code{pure} function attribute
2160 Many functions have no effects except the return value and their
2161 return value depends only on the parameters and/or global variables.
2162 Such a function can be subject
2163 to common subexpression elimination and loop optimization just as an
2164 arithmetic operator would be. These functions should be declared
2165 with the attribute @code{pure}. For example,
2168 int square (int) __attribute__ ((pure));
2172 says that the hypothetical function @code{square} is safe to call
2173 fewer times than the program says.
2175 Some of common examples of pure functions are @code{strlen} or @code{memcmp}.
2176 Interesting non-pure functions are functions with infinite loops or those
2177 depending on volatile memory or other system resource, that may change between
2178 two consecutive calls (such as @code{feof} in a multithreading environment).
2180 The attribute @code{pure} is not implemented in GCC versions earlier
2183 @item regparm (@var{number})
2184 @cindex @code{regparm} attribute
2185 @cindex functions that are passed arguments in registers on the 386
2186 On the Intel 386, the @code{regparm} attribute causes the compiler to
2187 pass arguments number one to @var{number} if they are of integral type
2188 in registers EAX, EDX, and ECX instead of on the stack. Functions that
2189 take a variable number of arguments will continue to be passed all of their
2190 arguments on the stack.
2192 Beware that on some ELF systems this attribute is unsuitable for
2193 global functions in shared libraries with lazy binding (which is the
2194 default). Lazy binding will send the first call via resolving code in
2195 the loader, which might assume EAX, EDX and ECX can be clobbered, as
2196 per the standard calling conventions. Solaris 8 is affected by this.
2197 GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be
2198 safe since the loaders there save all registers. (Lazy binding can be
2199 disabled with the linker or the loader if desired, to avoid the
2203 @cindex @code{sseregparm} attribute
2204 On the Intel 386 with SSE support, the @code{sseregparm} attribute
2205 causes the compiler to pass up to 8 floating point arguments in
2206 SSE registers instead of on the stack. Functions that take a
2207 variable number of arguments will continue to pass all of their
2208 floating point arguments on the stack.
2211 @cindex @code{returns_twice} attribute
2212 The @code{returns_twice} attribute tells the compiler that a function may
2213 return more than one time. The compiler will ensure that all registers
2214 are dead before calling such a function and will emit a warning about
2215 the variables that may be clobbered after the second return from the
2216 function. Examples of such functions are @code{setjmp} and @code{vfork}.
2217 The @code{longjmp}-like counterpart of such function, if any, might need
2218 to be marked with the @code{noreturn} attribute.
2221 @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S
2222 Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that
2223 all registers except the stack pointer should be saved in the prologue
2224 regardless of whether they are used or not.
2226 @item section ("@var{section-name}")
2227 @cindex @code{section} function attribute
2228 Normally, the compiler places the code it generates in the @code{text} section.
2229 Sometimes, however, you need additional sections, or you need certain
2230 particular functions to appear in special sections. The @code{section}
2231 attribute specifies that a function lives in a particular section.
2232 For example, the declaration:
2235 extern void foobar (void) __attribute__ ((section ("bar")));
2239 puts the function @code{foobar} in the @code{bar} section.
2241 Some file formats do not support arbitrary sections so the @code{section}
2242 attribute is not available on all platforms.
2243 If you need to map the entire contents of a module to a particular
2244 section, consider using the facilities of the linker instead.
2247 @cindex @code{sentinel} function attribute
2248 This function attribute ensures that a parameter in a function call is
2249 an explicit @code{NULL}. The attribute is only valid on variadic
2250 functions. By default, the sentinel is located at position zero, the
2251 last parameter of the function call. If an optional integer position
2252 argument P is supplied to the attribute, the sentinel must be located at
2253 position P counting backwards from the end of the argument list.
2256 __attribute__ ((sentinel))
2258 __attribute__ ((sentinel(0)))
2261 The attribute is automatically set with a position of 0 for the built-in
2262 functions @code{execl} and @code{execlp}. The built-in function
2263 @code{execle} has the attribute set with a position of 1.
2265 A valid @code{NULL} in this context is defined as zero with any pointer
2266 type. If your system defines the @code{NULL} macro with an integer type
2267 then you need to add an explicit cast. GCC replaces @code{stddef.h}
2268 with a copy that redefines NULL appropriately.
2270 The warnings for missing or incorrect sentinels are enabled with
2274 See long_call/short_call.
2277 See longcall/shortcall.
2280 @cindex signal handler functions on the AVR processors
2281 Use this attribute on the AVR to indicate that the specified
2282 function is a signal handler. The compiler will generate function
2283 entry and exit sequences suitable for use in a signal handler when this
2284 attribute is present. Interrupts will be disabled inside the function.
2287 Use this attribute on the SH to indicate an @code{interrupt_handler}
2288 function should switch to an alternate stack. It expects a string
2289 argument that names a global variable holding the address of the
2294 void f () __attribute__ ((interrupt_handler,
2295 sp_switch ("alt_stack")));
2299 @cindex functions that pop the argument stack on the 386
2300 On the Intel 386, the @code{stdcall} attribute causes the compiler to
2301 assume that the called function will pop off the stack space used to
2302 pass arguments, unless it takes a variable number of arguments.
2305 @cindex tiny data section on the H8/300H and H8S
2306 Use this attribute on the H8/300H and H8S to indicate that the specified
2307 variable should be placed into the tiny data section.
2308 The compiler will generate more efficient code for loads and stores
2309 on data in the tiny data section. Note the tiny data area is limited to
2310 slightly under 32kbytes of data.
2313 Use this attribute on the SH for an @code{interrupt_handler} to return using
2314 @code{trapa} instead of @code{rte}. This attribute expects an integer
2315 argument specifying the trap number to be used.
2318 @cindex @code{unused} attribute.
2319 This attribute, attached to a function, means that the function is meant
2320 to be possibly unused. GCC will not produce a warning for this
2324 @cindex @code{used} attribute.
2325 This attribute, attached to a function, means that code must be emitted
2326 for the function even if it appears that the function is not referenced.
2327 This is useful, for example, when the function is referenced only in
2330 @item visibility ("@var{visibility_type}")
2331 @cindex @code{visibility} attribute
2332 This attribute affects the linkage of the declaration to which it is attached.
2333 There are four supported @var{visibility_type} values: default,
2334 hidden, protected or internal visibility.
2337 void __attribute__ ((visibility ("protected")))
2338 f () @{ /* @r{Do something.} */; @}
2339 int i __attribute__ ((visibility ("hidden")));
2342 The possible values of @var{visibility_type} correspond to the
2343 visibility settings in the ELF gABI.
2346 @c keep this list of visibilities in alphabetical order.
2349 Default visibility is the normal case for the object file format.
2350 This value is available for the visibility attribute to override other
2351 options that may change the assumed visibility of entities.
2353 On ELF, default visibility means that the declaration is visible to other
2354 modules and, in shared libraries, means that the declared entity may be
2357 On Darwin, default visibility means that the declaration is visible to
2360 Default visibility corresponds to ``external linkage'' in the language.
2363 Hidden visibility indicates that the entity declared will have a new
2364 form of linkage, which we'll call ``hidden linkage''. Two
2365 declarations of an object with hidden linkage refer to the same object
2366 if they are in the same shared object.
2369 Internal visibility is like hidden visibility, but with additional
2370 processor specific semantics. Unless otherwise specified by the
2371 psABI, GCC defines internal visibility to mean that a function is
2372 @emph{never} called from another module. Compare this with hidden
2373 functions which, while they cannot be referenced directly by other
2374 modules, can be referenced indirectly via function pointers. By
2375 indicating that a function cannot be called from outside the module,
2376 GCC may for instance omit the load of a PIC register since it is known
2377 that the calling function loaded the correct value.
2380 Protected visibility is like default visibility except that it
2381 indicates that references within the defining module will bind to the
2382 definition in that module. That is, the declared entity cannot be
2383 overridden by another module.
2387 All visibilities are supported on many, but not all, ELF targets
2388 (supported when the assembler supports the @samp{.visibility}
2389 pseudo-op). Default visibility is supported everywhere. Hidden
2390 visibility is supported on Darwin targets.
2392 The visibility attribute should be applied only to declarations which
2393 would otherwise have external linkage. The attribute should be applied
2394 consistently, so that the same entity should not be declared with
2395 different settings of the attribute.
2397 In C++, the visibility attribute applies to types as well as functions
2398 and objects, because in C++ types have linkage. There are some bugs
2399 in the C++ support for this flag, for example a template which has a
2400 hidden type as a parameter is not properly hidden.
2403 In C++, you can mark member functions and static member variables of a
2404 class with the visibility attribute. This is useful if if you know a
2405 particular method or static member variable should only be used from
2406 one shared object; then you can mark it hidden while the rest of the
2407 class has default visibility. Care must be taken to avoid breaking
2408 the One Definition Rule; for example, it is not useful to mark a
2409 method which is defined inside a class definition as hidden without
2410 marking the whole class as hidden.
2412 @item warn_unused_result
2413 @cindex @code{warn_unused_result} attribute
2414 The @code{warn_unused_result} attribute causes a warning to be emitted
2415 if a caller of the function with this attribute does not use its
2416 return value. This is useful for functions where not checking
2417 the result is either a security problem or always a bug, such as
2421 int fn () __attribute__ ((warn_unused_result));
2424 if (fn () < 0) return -1;
2430 results in warning on line 5.
2433 @cindex @code{weak} attribute
2434 The @code{weak} attribute causes the declaration to be emitted as a weak
2435 symbol rather than a global. This is primarily useful in defining
2436 library functions which can be overridden in user code, though it can
2437 also be used with non-function declarations. Weak symbols are supported
2438 for ELF targets, and also for a.out targets when using the GNU assembler
2442 @itemx weakref ("@var{target}")
2443 @cindex @code{weakref} attribute
2444 The @code{weakref} attribute marks a declaration as a weak reference.
2445 Without arguments, it should be accompanied by an @code{alias} attribute
2446 naming the target symbol. Optionally, the @var{target} may be given as
2447 an argument to @code{weakref} itself. In either case, @code{weakref}
2448 implicitly marks the declaration as @code{weak}. Without a
2449 @var{target}, given as an argument to @code{weakref} or to @code{alias},
2450 @code{weakref} is equivalent to @code{weak}.
2453 static int x() __attribute__ ((weakref ("y")));
2454 /* is equivalent to... */
2455 static int x() __attribute__ ((weak, weakref, alias ("y")));
2457 static int x() __attribute__ ((weakref));
2458 static int x() __attribute__ ((alias ("y")));
2461 A weak reference is an alias that does not by itself require a
2462 definition to be given for the target symbol. If the target symbol is
2463 only referenced through weak references, then the becomes a @code{weak}
2464 undefined symbol. If it is directly referenced, however, then such
2465 strong references prevail, and a definition will be required for the
2466 symbol, not necessarily in the same translation unit.
2468 The effect is equivalent to moving all references to the alias to a
2469 separate translation unit, renaming the alias to the aliased symbol,
2470 declaring it as weak, compiling the two separate translation units and
2471 performing a reloadable link on them.
2473 At present, a declaration to which @code{weakref} is attached can
2474 only be @code{static}.
2476 @item externally_visible
2477 @cindex @code{externally_visible} attribute.
2478 This attribute, attached to a global variable or function nullify
2479 effect of @option{-fwhole-program} command line option, so the object
2480 remain visible outside the current compilation unit
2484 You can specify multiple attributes in a declaration by separating them
2485 by commas within the double parentheses or by immediately following an
2486 attribute declaration with another attribute declaration.
2488 @cindex @code{#pragma}, reason for not using
2489 @cindex pragma, reason for not using
2490 Some people object to the @code{__attribute__} feature, suggesting that
2491 ISO C's @code{#pragma} should be used instead. At the time
2492 @code{__attribute__} was designed, there were two reasons for not doing
2497 It is impossible to generate @code{#pragma} commands from a macro.
2500 There is no telling what the same @code{#pragma} might mean in another
2504 These two reasons applied to almost any application that might have been
2505 proposed for @code{#pragma}. It was basically a mistake to use
2506 @code{#pragma} for @emph{anything}.
2508 The ISO C99 standard includes @code{_Pragma}, which now allows pragmas
2509 to be generated from macros. In addition, a @code{#pragma GCC}
2510 namespace is now in use for GCC-specific pragmas. However, it has been
2511 found convenient to use @code{__attribute__} to achieve a natural
2512 attachment of attributes to their corresponding declarations, whereas
2513 @code{#pragma GCC} is of use for constructs that do not naturally form
2514 part of the grammar. @xref{Other Directives,,Miscellaneous
2515 Preprocessing Directives, cpp, The GNU C Preprocessor}.
2517 @node Attribute Syntax
2518 @section Attribute Syntax
2519 @cindex attribute syntax
2521 This section describes the syntax with which @code{__attribute__} may be
2522 used, and the constructs to which attribute specifiers bind, for the C
2523 language. Some details may vary for C++ and Objective-C@. Because of
2524 infelicities in the grammar for attributes, some forms described here
2525 may not be successfully parsed in all cases.
2527 There are some problems with the semantics of attributes in C++. For
2528 example, there are no manglings for attributes, although they may affect
2529 code generation, so problems may arise when attributed types are used in
2530 conjunction with templates or overloading. Similarly, @code{typeid}
2531 does not distinguish between types with different attributes. Support
2532 for attributes in C++ may be restricted in future to attributes on
2533 declarations only, but not on nested declarators.
2535 @xref{Function Attributes}, for details of the semantics of attributes
2536 applying to functions. @xref{Variable Attributes}, for details of the
2537 semantics of attributes applying to variables. @xref{Type Attributes},
2538 for details of the semantics of attributes applying to structure, union
2539 and enumerated types.
2541 An @dfn{attribute specifier} is of the form
2542 @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list}
2543 is a possibly empty comma-separated sequence of @dfn{attributes}, where
2544 each attribute is one of the following:
2548 Empty. Empty attributes are ignored.
2551 A word (which may be an identifier such as @code{unused}, or a reserved
2552 word such as @code{const}).
2555 A word, followed by, in parentheses, parameters for the attribute.
2556 These parameters take one of the following forms:
2560 An identifier. For example, @code{mode} attributes use this form.
2563 An identifier followed by a comma and a non-empty comma-separated list
2564 of expressions. For example, @code{format} attributes use this form.
2567 A possibly empty comma-separated list of expressions. For example,
2568 @code{format_arg} attributes use this form with the list being a single
2569 integer constant expression, and @code{alias} attributes use this form
2570 with the list being a single string constant.
2574 An @dfn{attribute specifier list} is a sequence of one or more attribute
2575 specifiers, not separated by any other tokens.
2577 In GNU C, an attribute specifier list may appear after the colon following a
2578 label, other than a @code{case} or @code{default} label. The only
2579 attribute it makes sense to use after a label is @code{unused}. This
2580 feature is intended for code generated by programs which contains labels
2581 that may be unused but which is compiled with @option{-Wall}. It would
2582 not normally be appropriate to use in it human-written code, though it
2583 could be useful in cases where the code that jumps to the label is
2584 contained within an @code{#ifdef} conditional. GNU C++ does not permit
2585 such placement of attribute lists, as it is permissible for a
2586 declaration, which could begin with an attribute list, to be labelled in
2587 C++. Declarations cannot be labelled in C90 or C99, so the ambiguity
2588 does not arise there.
2590 An attribute specifier list may appear as part of a @code{struct},
2591 @code{union} or @code{enum} specifier. It may go either immediately
2592 after the @code{struct}, @code{union} or @code{enum} keyword, or after
2593 the closing brace. It is ignored if the content of the structure, union
2594 or enumerated type is not defined in the specifier in which the
2595 attribute specifier list is used---that is, in usages such as
2596 @code{struct __attribute__((foo)) bar} with no following opening brace.
2597 Where attribute specifiers follow the closing brace, they are considered
2598 to relate to the structure, union or enumerated type defined, not to any
2599 enclosing declaration the type specifier appears in, and the type
2600 defined is not complete until after the attribute specifiers.
2601 @c Otherwise, there would be the following problems: a shift/reduce
2602 @c conflict between attributes binding the struct/union/enum and
2603 @c binding to the list of specifiers/qualifiers; and "aligned"
2604 @c attributes could use sizeof for the structure, but the size could be
2605 @c changed later by "packed" attributes.
2607 Otherwise, an attribute specifier appears as part of a declaration,
2608 counting declarations of unnamed parameters and type names, and relates
2609 to that declaration (which may be nested in another declaration, for
2610 example in the case of a parameter declaration), or to a particular declarator
2611 within a declaration. Where an
2612 attribute specifier is applied to a parameter declared as a function or
2613 an array, it should apply to the function or array rather than the
2614 pointer to which the parameter is implicitly converted, but this is not
2615 yet correctly implemented.
2617 Any list of specifiers and qualifiers at the start of a declaration may
2618 contain attribute specifiers, whether or not such a list may in that
2619 context contain storage class specifiers. (Some attributes, however,
2620 are essentially in the nature of storage class specifiers, and only make
2621 sense where storage class specifiers may be used; for example,
2622 @code{section}.) There is one necessary limitation to this syntax: the
2623 first old-style parameter declaration in a function definition cannot
2624 begin with an attribute specifier, because such an attribute applies to
2625 the function instead by syntax described below (which, however, is not
2626 yet implemented in this case). In some other cases, attribute
2627 specifiers are permitted by this grammar but not yet supported by the
2628 compiler. All attribute specifiers in this place relate to the
2629 declaration as a whole. In the obsolescent usage where a type of
2630 @code{int} is implied by the absence of type specifiers, such a list of
2631 specifiers and qualifiers may be an attribute specifier list with no
2632 other specifiers or qualifiers.
2634 At present, the first parameter in a function prototype must have some
2635 type specifier which is not an attribute specifier; this resolves an
2636 ambiguity in the interpretation of @code{void f(int
2637 (__attribute__((foo)) x))}, but is subject to change. At present, if
2638 the parentheses of a function declarator contain only attributes then
2639 those attributes are ignored, rather than yielding an error or warning
2640 or implying a single parameter of type int, but this is subject to
2643 An attribute specifier list may appear immediately before a declarator
2644 (other than the first) in a comma-separated list of declarators in a
2645 declaration of more than one identifier using a single list of
2646 specifiers and qualifiers. Such attribute specifiers apply
2647 only to the identifier before whose declarator they appear. For
2651 __attribute__((noreturn)) void d0 (void),
2652 __attribute__((format(printf, 1, 2))) d1 (const char *, ...),
2657 the @code{noreturn} attribute applies to all the functions
2658 declared; the @code{format} attribute only applies to @code{d1}.
2660 An attribute specifier list may appear immediately before the comma,
2661 @code{=} or semicolon terminating the declaration of an identifier other
2662 than a function definition. At present, such attribute specifiers apply
2663 to the declared object or function, but in future they may attach to the
2664 outermost adjacent declarator. In simple cases there is no difference,
2665 but, for example, in
2668 void (****f)(void) __attribute__((noreturn));
2672 at present the @code{noreturn} attribute applies to @code{f}, which
2673 causes a warning since @code{f} is not a function, but in future it may
2674 apply to the function @code{****f}. The precise semantics of what
2675 attributes in such cases will apply to are not yet specified. Where an
2676 assembler name for an object or function is specified (@pxref{Asm
2677 Labels}), at present the attribute must follow the @code{asm}
2678 specification; in future, attributes before the @code{asm} specification
2679 may apply to the adjacent declarator, and those after it to the declared
2682 An attribute specifier list may, in future, be permitted to appear after
2683 the declarator in a function definition (before any old-style parameter
2684 declarations or the function body).
2686 Attribute specifiers may be mixed with type qualifiers appearing inside
2687 the @code{[]} of a parameter array declarator, in the C99 construct by
2688 which such qualifiers are applied to the pointer to which the array is
2689 implicitly converted. Such attribute specifiers apply to the pointer,
2690 not to the array, but at present this is not implemented and they are
2693 An attribute specifier list may appear at the start of a nested
2694 declarator. At present, there are some limitations in this usage: the
2695 attributes correctly apply to the declarator, but for most individual
2696 attributes the semantics this implies are not implemented.
2697 When attribute specifiers follow the @code{*} of a pointer
2698 declarator, they may be mixed with any type qualifiers present.
2699 The following describes the formal semantics of this syntax. It will make the
2700 most sense if you are familiar with the formal specification of
2701 declarators in the ISO C standard.
2703 Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T
2704 D1}, where @code{T} contains declaration specifiers that specify a type
2705 @var{Type} (such as @code{int}) and @code{D1} is a declarator that
2706 contains an identifier @var{ident}. The type specified for @var{ident}
2707 for derived declarators whose type does not include an attribute
2708 specifier is as in the ISO C standard.
2710 If @code{D1} has the form @code{( @var{attribute-specifier-list} D )},
2711 and the declaration @code{T D} specifies the type
2712 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2713 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2714 @var{attribute-specifier-list} @var{Type}'' for @var{ident}.
2716 If @code{D1} has the form @code{*
2717 @var{type-qualifier-and-attribute-specifier-list} D}, and the
2718 declaration @code{T D} specifies the type
2719 ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then
2720 @code{T D1} specifies the type ``@var{derived-declarator-type-list}
2721 @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for
2727 void (__attribute__((noreturn)) ****f) (void);
2731 specifies the type ``pointer to pointer to pointer to pointer to
2732 non-returning function returning @code{void}''. As another example,
2735 char *__attribute__((aligned(8))) *f;
2739 specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''.
2740 Note again that this does not work with most attributes; for example,
2741 the usage of @samp{aligned} and @samp{noreturn} attributes given above
2742 is not yet supported.
2744 For compatibility with existing code written for compiler versions that
2745 did not implement attributes on nested declarators, some laxity is
2746 allowed in the placing of attributes. If an attribute that only applies
2747 to types is applied to a declaration, it will be treated as applying to
2748 the type of that declaration. If an attribute that only applies to
2749 declarations is applied to the type of a declaration, it will be treated
2750 as applying to that declaration; and, for compatibility with code
2751 placing the attributes immediately before the identifier declared, such
2752 an attribute applied to a function return type will be treated as
2753 applying to the function type, and such an attribute applied to an array
2754 element type will be treated as applying to the array type. If an
2755 attribute that only applies to function types is applied to a
2756 pointer-to-function type, it will be treated as applying to the pointer
2757 target type; if such an attribute is applied to a function return type
2758 that is not a pointer-to-function type, it will be treated as applying
2759 to the function type.
2761 @node Function Prototypes
2762 @section Prototypes and Old-Style Function Definitions
2763 @cindex function prototype declarations
2764 @cindex old-style function definitions
2765 @cindex promotion of formal parameters
2767 GNU C extends ISO C to allow a function prototype to override a later
2768 old-style non-prototype definition. Consider the following example:
2771 /* @r{Use prototypes unless the compiler is old-fashioned.} */
2778 /* @r{Prototype function declaration.} */
2779 int isroot P((uid_t));
2781 /* @r{Old-style function definition.} */
2783 isroot (x) /* @r{??? lossage here ???} */
2790 Suppose the type @code{uid_t} happens to be @code{short}. ISO C does
2791 not allow this example, because subword arguments in old-style
2792 non-prototype definitions are promoted. Therefore in this example the
2793 function definition's argument is really an @code{int}, which does not
2794 match the prototype argument type of @code{short}.
2796 This restriction of ISO C makes it hard to write code that is portable
2797 to traditional C compilers, because the programmer does not know
2798 whether the @code{uid_t} type is @code{short}, @code{int}, or
2799 @code{long}. Therefore, in cases like these GNU C allows a prototype
2800 to override a later old-style definition. More precisely, in GNU C, a
2801 function prototype argument type overrides the argument type specified
2802 by a later old-style definition if the former type is the same as the
2803 latter type before promotion. Thus in GNU C the above example is
2804 equivalent to the following:
2817 GNU C++ does not support old-style function definitions, so this
2818 extension is irrelevant.
2821 @section C++ Style Comments
2823 @cindex C++ comments
2824 @cindex comments, C++ style
2826 In GNU C, you may use C++ style comments, which start with @samp{//} and
2827 continue until the end of the line. Many other C implementations allow
2828 such comments, and they are included in the 1999 C standard. However,
2829 C++ style comments are not recognized if you specify an @option{-std}
2830 option specifying a version of ISO C before C99, or @option{-ansi}
2831 (equivalent to @option{-std=c89}).
2834 @section Dollar Signs in Identifier Names
2836 @cindex dollar signs in identifier names
2837 @cindex identifier names, dollar signs in
2839 In GNU C, you may normally use dollar signs in identifier names.
2840 This is because many traditional C implementations allow such identifiers.
2841 However, dollar signs in identifiers are not supported on a few target
2842 machines, typically because the target assembler does not allow them.
2844 @node Character Escapes
2845 @section The Character @key{ESC} in Constants
2847 You can use the sequence @samp{\e} in a string or character constant to
2848 stand for the ASCII character @key{ESC}.
2851 @section Inquiring on Alignment of Types or Variables
2853 @cindex type alignment
2854 @cindex variable alignment
2856 The keyword @code{__alignof__} allows you to inquire about how an object
2857 is aligned, or the minimum alignment usually required by a type. Its
2858 syntax is just like @code{sizeof}.
2860 For example, if the target machine requires a @code{double} value to be
2861 aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8.
2862 This is true on many RISC machines. On more traditional machine
2863 designs, @code{__alignof__ (double)} is 4 or even 2.
2865 Some machines never actually require alignment; they allow reference to any
2866 data type even at an odd address. For these machines, @code{__alignof__}
2867 reports the @emph{recommended} alignment of a type.
2869 If the operand of @code{__alignof__} is an lvalue rather than a type,
2870 its value is the required alignment for its type, taking into account
2871 any minimum alignment specified with GCC's @code{__attribute__}
2872 extension (@pxref{Variable Attributes}). For example, after this
2876 struct foo @{ int x; char y; @} foo1;
2880 the value of @code{__alignof__ (foo1.y)} is 1, even though its actual
2881 alignment is probably 2 or 4, the same as @code{__alignof__ (int)}.
2883 It is an error to ask for the alignment of an incomplete type.
2885 @node Variable Attributes
2886 @section Specifying Attributes of Variables
2887 @cindex attribute of variables
2888 @cindex variable attributes
2890 The keyword @code{__attribute__} allows you to specify special
2891 attributes of variables or structure fields. This keyword is followed
2892 by an attribute specification inside double parentheses. Some
2893 attributes are currently defined generically for variables.
2894 Other attributes are defined for variables on particular target
2895 systems. Other attributes are available for functions
2896 (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}).
2897 Other front ends might define more attributes
2898 (@pxref{C++ Extensions,,Extensions to the C++ Language}).
2900 You may also specify attributes with @samp{__} preceding and following
2901 each keyword. This allows you to use them in header files without
2902 being concerned about a possible macro of the same name. For example,
2903 you may use @code{__aligned__} instead of @code{aligned}.
2905 @xref{Attribute Syntax}, for details of the exact syntax for using
2909 @cindex @code{aligned} attribute
2910 @item aligned (@var{alignment})
2911 This attribute specifies a minimum alignment for the variable or
2912 structure field, measured in bytes. For example, the declaration:
2915 int x __attribute__ ((aligned (16))) = 0;
2919 causes the compiler to allocate the global variable @code{x} on a
2920 16-byte boundary. On a 68040, this could be used in conjunction with
2921 an @code{asm} expression to access the @code{move16} instruction which
2922 requires 16-byte aligned operands.
2924 You can also specify the alignment of structure fields. For example, to
2925 create a double-word aligned @code{int} pair, you could write:
2928 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
2932 This is an alternative to creating a union with a @code{double} member
2933 that forces the union to be double-word aligned.
2935 As in the preceding examples, you can explicitly specify the alignment
2936 (in bytes) that you wish the compiler to use for a given variable or
2937 structure field. Alternatively, you can leave out the alignment factor
2938 and just ask the compiler to align a variable or field to the maximum
2939 useful alignment for the target machine you are compiling for. For
2940 example, you could write:
2943 short array[3] __attribute__ ((aligned));
2946 Whenever you leave out the alignment factor in an @code{aligned} attribute
2947 specification, the compiler automatically sets the alignment for the declared
2948 variable or field to the largest alignment which is ever used for any data
2949 type on the target machine you are compiling for. Doing this can often make
2950 copy operations more efficient, because the compiler can use whatever
2951 instructions copy the biggest chunks of memory when performing copies to
2952 or from the variables or fields that you have aligned this way.
2954 The @code{aligned} attribute can only increase the alignment; but you
2955 can decrease it by specifying @code{packed} as well. See below.
2957 Note that the effectiveness of @code{aligned} attributes may be limited
2958 by inherent limitations in your linker. On many systems, the linker is
2959 only able to arrange for variables to be aligned up to a certain maximum
2960 alignment. (For some linkers, the maximum supported alignment may
2961 be very very small.) If your linker is only able to align variables
2962 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
2963 in an @code{__attribute__} will still only provide you with 8 byte
2964 alignment. See your linker documentation for further information.
2966 @item cleanup (@var{cleanup_function})
2967 @cindex @code{cleanup} attribute
2968 The @code{cleanup} attribute runs a function when the variable goes
2969 out of scope. This attribute can only be applied to auto function
2970 scope variables; it may not be applied to parameters or variables
2971 with static storage duration. The function must take one parameter,
2972 a pointer to a type compatible with the variable. The return value
2973 of the function (if any) is ignored.
2975 If @option{-fexceptions} is enabled, then @var{cleanup_function}
2976 will be run during the stack unwinding that happens during the
2977 processing of the exception. Note that the @code{cleanup} attribute
2978 does not allow the exception to be caught, only to perform an action.
2979 It is undefined what happens if @var{cleanup_function} does not
2984 @cindex @code{common} attribute
2985 @cindex @code{nocommon} attribute
2988 The @code{common} attribute requests GCC to place a variable in
2989 ``common'' storage. The @code{nocommon} attribute requests the
2990 opposite---to allocate space for it directly.
2992 These attributes override the default chosen by the
2993 @option{-fno-common} and @option{-fcommon} flags respectively.
2996 @cindex @code{deprecated} attribute
2997 The @code{deprecated} attribute results in a warning if the variable
2998 is used anywhere in the source file. This is useful when identifying
2999 variables that are expected to be removed in a future version of a
3000 program. The warning also includes the location of the declaration
3001 of the deprecated variable, to enable users to easily find further
3002 information about why the variable is deprecated, or what they should
3003 do instead. Note that the warning only occurs for uses:
3006 extern int old_var __attribute__ ((deprecated));
3008 int new_fn () @{ return old_var; @}
3011 results in a warning on line 3 but not line 2.
3013 The @code{deprecated} attribute can also be used for functions and
3014 types (@pxref{Function Attributes}, @pxref{Type Attributes}.)
3016 @item mode (@var{mode})
3017 @cindex @code{mode} attribute
3018 This attribute specifies the data type for the declaration---whichever
3019 type corresponds to the mode @var{mode}. This in effect lets you
3020 request an integer or floating point type according to its width.
3022 You may also specify a mode of @samp{byte} or @samp{__byte__} to
3023 indicate the mode corresponding to a one-byte integer, @samp{word} or
3024 @samp{__word__} for the mode of a one-word integer, and @samp{pointer}
3025 or @samp{__pointer__} for the mode used to represent pointers.
3028 @cindex @code{packed} attribute
3029 The @code{packed} attribute specifies that a variable or structure field
3030 should have the smallest possible alignment---one byte for a variable,
3031 and one bit for a field, unless you specify a larger value with the
3032 @code{aligned} attribute.
3034 Here is a structure in which the field @code{x} is packed, so that it
3035 immediately follows @code{a}:
3041 int x[2] __attribute__ ((packed));
3045 @item section ("@var{section-name}")
3046 @cindex @code{section} variable attribute
3047 Normally, the compiler places the objects it generates in sections like
3048 @code{data} and @code{bss}. Sometimes, however, you need additional sections,
3049 or you need certain particular variables to appear in special sections,
3050 for example to map to special hardware. The @code{section}
3051 attribute specifies that a variable (or function) lives in a particular
3052 section. For example, this small program uses several specific section names:
3055 struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @};
3056 struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @};
3057 char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @};
3058 int init_data __attribute__ ((section ("INITDATA"))) = 0;
3062 /* @r{Initialize stack pointer} */
3063 init_sp (stack + sizeof (stack));
3065 /* @r{Initialize initialized data} */
3066 memcpy (&init_data, &data, &edata - &data);
3068 /* @r{Turn on the serial ports} */
3075 Use the @code{section} attribute with an @emph{initialized} definition
3076 of a @emph{global} variable, as shown in the example. GCC issues
3077 a warning and otherwise ignores the @code{section} attribute in
3078 uninitialized variable declarations.
3080 You may only use the @code{section} attribute with a fully initialized
3081 global definition because of the way linkers work. The linker requires
3082 each object be defined once, with the exception that uninitialized
3083 variables tentatively go in the @code{common} (or @code{bss}) section
3084 and can be multiply ``defined''. You can force a variable to be
3085 initialized with the @option{-fno-common} flag or the @code{nocommon}
3088 Some file formats do not support arbitrary sections so the @code{section}
3089 attribute is not available on all platforms.
3090 If you need to map the entire contents of a module to a particular
3091 section, consider using the facilities of the linker instead.
3094 @cindex @code{shared} variable attribute
3095 On Microsoft Windows, in addition to putting variable definitions in a named
3096 section, the section can also be shared among all running copies of an
3097 executable or DLL@. For example, this small program defines shared data
3098 by putting it in a named section @code{shared} and marking the section
3102 int foo __attribute__((section ("shared"), shared)) = 0;
3107 /* @r{Read and write foo. All running
3108 copies see the same value.} */
3114 You may only use the @code{shared} attribute along with @code{section}
3115 attribute with a fully initialized global definition because of the way
3116 linkers work. See @code{section} attribute for more information.
3118 The @code{shared} attribute is only available on Microsoft Windows@.
3120 @item tls_model ("@var{tls_model}")
3121 @cindex @code{tls_model} attribute
3122 The @code{tls_model} attribute sets thread-local storage model
3123 (@pxref{Thread-Local}) of a particular @code{__thread} variable,
3124 overriding @option{-ftls-model=} command line switch on a per-variable
3126 The @var{tls_model} argument should be one of @code{global-dynamic},
3127 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
3129 Not all targets support this attribute.
3132 This attribute, attached to a variable, means that the variable is meant
3133 to be possibly unused. GCC will not produce a warning for this
3136 @item vector_size (@var{bytes})
3137 This attribute specifies the vector size for the variable, measured in
3138 bytes. For example, the declaration:
3141 int foo __attribute__ ((vector_size (16)));
3145 causes the compiler to set the mode for @code{foo}, to be 16 bytes,
3146 divided into @code{int} sized units. Assuming a 32-bit int (a vector of
3147 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@.
3149 This attribute is only applicable to integral and float scalars,
3150 although arrays, pointers, and function return values are allowed in
3151 conjunction with this construct.
3153 Aggregates with this attribute are invalid, even if they are of the same
3154 size as a corresponding scalar. For example, the declaration:
3157 struct S @{ int a; @};
3158 struct S __attribute__ ((vector_size (16))) foo;
3162 is invalid even if the size of the structure is the same as the size of
3166 The @code{selectany} attribute causes an initialized global variable to
3167 have link-once semantics. When multiple definitions of the variable are
3168 encountered by the linker, the first is selected and the remainder are
3169 discarded. Following usage by the Microsoft compiler, the linker is told
3170 @emph{not} to warn about size or content differences of the multiple
3173 Although the primary usage of this attribute is for POD types, the
3174 attribute can also be applied to global C++ objects that are initialized
3175 by a constructor. In this case, the static initialization and destruction
3176 code for the object is emitted in each translation defining the object,
3177 but the calls to the constructor and destructor are protected by a
3178 link-once guard variable.
3180 The @code{selectany} attribute is only available on Microsoft Windows
3181 targets. You can use @code{__declspec (selectany)} as a synonym for
3182 @code{__attribute__ ((selectany))} for compatibility with other
3186 The @code{weak} attribute is described in @xref{Function Attributes}.
3189 The @code{dllimport} attribute is described in @xref{Function Attributes}.
3192 The @code{dllexport} attribute is described in @xref{Function Attributes}.
3196 @subsection M32R/D Variable Attributes
3198 One attribute is currently defined for the M32R/D@.
3201 @item model (@var{model-name})
3202 @cindex variable addressability on the M32R/D
3203 Use this attribute on the M32R/D to set the addressability of an object.
3204 The identifier @var{model-name} is one of @code{small}, @code{medium},
3205 or @code{large}, representing each of the code models.
3207 Small model objects live in the lower 16MB of memory (so that their
3208 addresses can be loaded with the @code{ld24} instruction).
3210 Medium and large model objects may live anywhere in the 32-bit address space
3211 (the compiler will generate @code{seth/add3} instructions to load their
3215 @subsection i386 Variable Attributes
3217 Two attributes are currently defined for i386 configurations:
3218 @code{ms_struct} and @code{gcc_struct}
3223 @cindex @code{ms_struct} attribute
3224 @cindex @code{gcc_struct} attribute
3226 If @code{packed} is used on a structure, or if bit-fields are used
3227 it may be that the Microsoft ABI packs them differently
3228 than GCC would normally pack them. Particularly when moving packed
3229 data between functions compiled with GCC and the native Microsoft compiler
3230 (either via function call or as data in a file), it may be necessary to access
3233 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3234 compilers to match the native Microsoft compiler.
3237 @subsection Xstormy16 Variable Attributes
3239 One attribute is currently defined for xstormy16 configurations:
3244 @cindex @code{below100} attribute
3246 If a variable has the @code{below100} attribute (@code{BELOW100} is
3247 allowed also), GCC will place the variable in the first 0x100 bytes of
3248 memory and use special opcodes to access it. Such variables will be
3249 placed in either the @code{.bss_below100} section or the
3250 @code{.data_below100} section.
3254 @node Type Attributes
3255 @section Specifying Attributes of Types
3256 @cindex attribute of types
3257 @cindex type attributes
3259 The keyword @code{__attribute__} allows you to specify special
3260 attributes of @code{struct} and @code{union} types when you define such
3261 types. This keyword is followed by an attribute specification inside
3262 double parentheses. Six attributes are currently defined for types:
3263 @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused},
3264 @code{deprecated} and @code{may_alias}. Other attributes are defined for
3265 functions (@pxref{Function Attributes}) and for variables
3266 (@pxref{Variable Attributes}).
3268 You may also specify any one of these attributes with @samp{__}
3269 preceding and following its keyword. This allows you to use these
3270 attributes in header files without being concerned about a possible
3271 macro of the same name. For example, you may use @code{__aligned__}
3272 instead of @code{aligned}.
3274 You may specify the @code{aligned} and @code{transparent_union}
3275 attributes either in a @code{typedef} declaration or just past the
3276 closing curly brace of a complete enum, struct or union type
3277 @emph{definition} and the @code{packed} attribute only past the closing
3278 brace of a definition.
3280 You may also specify attributes between the enum, struct or union
3281 tag and the name of the type rather than after the closing brace.
3283 @xref{Attribute Syntax}, for details of the exact syntax for using
3287 @cindex @code{aligned} attribute
3288 @item aligned (@var{alignment})
3289 This attribute specifies a minimum alignment (in bytes) for variables
3290 of the specified type. For example, the declarations:
3293 struct S @{ short f[3]; @} __attribute__ ((aligned (8)));
3294 typedef int more_aligned_int __attribute__ ((aligned (8)));
3298 force the compiler to insure (as far as it can) that each variable whose
3299 type is @code{struct S} or @code{more_aligned_int} will be allocated and
3300 aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all
3301 variables of type @code{struct S} aligned to 8-byte boundaries allows
3302 the compiler to use the @code{ldd} and @code{std} (doubleword load and
3303 store) instructions when copying one variable of type @code{struct S} to
3304 another, thus improving run-time efficiency.
3306 Note that the alignment of any given @code{struct} or @code{union} type
3307 is required by the ISO C standard to be at least a perfect multiple of
3308 the lowest common multiple of the alignments of all of the members of
3309 the @code{struct} or @code{union} in question. This means that you @emph{can}
3310 effectively adjust the alignment of a @code{struct} or @code{union}
3311 type by attaching an @code{aligned} attribute to any one of the members
3312 of such a type, but the notation illustrated in the example above is a
3313 more obvious, intuitive, and readable way to request the compiler to
3314 adjust the alignment of an entire @code{struct} or @code{union} type.
3316 As in the preceding example, you can explicitly specify the alignment
3317 (in bytes) that you wish the compiler to use for a given @code{struct}
3318 or @code{union} type. Alternatively, you can leave out the alignment factor
3319 and just ask the compiler to align a type to the maximum
3320 useful alignment for the target machine you are compiling for. For
3321 example, you could write:
3324 struct S @{ short f[3]; @} __attribute__ ((aligned));
3327 Whenever you leave out the alignment factor in an @code{aligned}
3328 attribute specification, the compiler automatically sets the alignment
3329 for the type to the largest alignment which is ever used for any data
3330 type on the target machine you are compiling for. Doing this can often
3331 make copy operations more efficient, because the compiler can use
3332 whatever instructions copy the biggest chunks of memory when performing
3333 copies to or from the variables which have types that you have aligned
3336 In the example above, if the size of each @code{short} is 2 bytes, then
3337 the size of the entire @code{struct S} type is 6 bytes. The smallest
3338 power of two which is greater than or equal to that is 8, so the
3339 compiler sets the alignment for the entire @code{struct S} type to 8
3342 Note that although you can ask the compiler to select a time-efficient
3343 alignment for a given type and then declare only individual stand-alone
3344 objects of that type, the compiler's ability to select a time-efficient
3345 alignment is primarily useful only when you plan to create arrays of
3346 variables having the relevant (efficiently aligned) type. If you
3347 declare or use arrays of variables of an efficiently-aligned type, then
3348 it is likely that your program will also be doing pointer arithmetic (or
3349 subscripting, which amounts to the same thing) on pointers to the
3350 relevant type, and the code that the compiler generates for these
3351 pointer arithmetic operations will often be more efficient for
3352 efficiently-aligned types than for other types.
3354 The @code{aligned} attribute can only increase the alignment; but you
3355 can decrease it by specifying @code{packed} as well. See below.
3357 Note that the effectiveness of @code{aligned} attributes may be limited
3358 by inherent limitations in your linker. On many systems, the linker is
3359 only able to arrange for variables to be aligned up to a certain maximum
3360 alignment. (For some linkers, the maximum supported alignment may
3361 be very very small.) If your linker is only able to align variables
3362 up to a maximum of 8 byte alignment, then specifying @code{aligned(16)}
3363 in an @code{__attribute__} will still only provide you with 8 byte
3364 alignment. See your linker documentation for further information.
3367 This attribute, attached to @code{struct} or @code{union} type
3368 definition, specifies that each member (other than zero-width bitfields)
3369 of the structure or union is placed to minimize the memory required. When
3370 attached to an @code{enum} definition, it indicates that the smallest
3371 integral type should be used.
3373 @opindex fshort-enums
3374 Specifying this attribute for @code{struct} and @code{union} types is
3375 equivalent to specifying the @code{packed} attribute on each of the
3376 structure or union members. Specifying the @option{-fshort-enums}
3377 flag on the line is equivalent to specifying the @code{packed}
3378 attribute on all @code{enum} definitions.
3380 In the following example @code{struct my_packed_struct}'s members are
3381 packed closely together, but the internal layout of its @code{s} member
3382 is not packed---to do that, @code{struct my_unpacked_struct} would need to
3386 struct my_unpacked_struct
3392 struct __attribute__ ((__packed__)) my_packed_struct
3396 struct my_unpacked_struct s;
3400 You may only specify this attribute on the definition of a @code{enum},
3401 @code{struct} or @code{union}, not on a @code{typedef} which does not
3402 also define the enumerated type, structure or union.
3404 @item transparent_union
3405 This attribute, attached to a @code{union} type definition, indicates
3406 that any function parameter having that union type causes calls to that
3407 function to be treated in a special way.
3409 First, the argument corresponding to a transparent union type can be of
3410 any type in the union; no cast is required. Also, if the union contains
3411 a pointer type, the corresponding argument can be a null pointer
3412 constant or a void pointer expression; and if the union contains a void
3413 pointer type, the corresponding argument can be any pointer expression.
3414 If the union member type is a pointer, qualifiers like @code{const} on
3415 the referenced type must be respected, just as with normal pointer
3418 Second, the argument is passed to the function using the calling
3419 conventions of the first member of the transparent union, not the calling
3420 conventions of the union itself. All members of the union must have the
3421 same machine representation; this is necessary for this argument passing
3424 Transparent unions are designed for library functions that have multiple
3425 interfaces for compatibility reasons. For example, suppose the
3426 @code{wait} function must accept either a value of type @code{int *} to
3427 comply with Posix, or a value of type @code{union wait *} to comply with
3428 the 4.1BSD interface. If @code{wait}'s parameter were @code{void *},
3429 @code{wait} would accept both kinds of arguments, but it would also
3430 accept any other pointer type and this would make argument type checking
3431 less useful. Instead, @code{<sys/wait.h>} might define the interface
3439 @} wait_status_ptr_t __attribute__ ((__transparent_union__));
3441 pid_t wait (wait_status_ptr_t);
3444 This interface allows either @code{int *} or @code{union wait *}
3445 arguments to be passed, using the @code{int *} calling convention.
3446 The program can call @code{wait} with arguments of either type:
3449 int w1 () @{ int w; return wait (&w); @}
3450 int w2 () @{ union wait w; return wait (&w); @}
3453 With this interface, @code{wait}'s implementation might look like this:
3456 pid_t wait (wait_status_ptr_t p)
3458 return waitpid (-1, p.__ip, 0);
3463 When attached to a type (including a @code{union} or a @code{struct}),
3464 this attribute means that variables of that type are meant to appear
3465 possibly unused. GCC will not produce a warning for any variables of
3466 that type, even if the variable appears to do nothing. This is often
3467 the case with lock or thread classes, which are usually defined and then
3468 not referenced, but contain constructors and destructors that have
3469 nontrivial bookkeeping functions.
3472 The @code{deprecated} attribute results in a warning if the type
3473 is used anywhere in the source file. This is useful when identifying
3474 types that are expected to be removed in a future version of a program.
3475 If possible, the warning also includes the location of the declaration
3476 of the deprecated type, to enable users to easily find further
3477 information about why the type is deprecated, or what they should do
3478 instead. Note that the warnings only occur for uses and then only
3479 if the type is being applied to an identifier that itself is not being
3480 declared as deprecated.
3483 typedef int T1 __attribute__ ((deprecated));
3487 typedef T1 T3 __attribute__ ((deprecated));
3488 T3 z __attribute__ ((deprecated));
3491 results in a warning on line 2 and 3 but not lines 4, 5, or 6. No
3492 warning is issued for line 4 because T2 is not explicitly
3493 deprecated. Line 5 has no warning because T3 is explicitly
3494 deprecated. Similarly for line 6.
3496 The @code{deprecated} attribute can also be used for functions and
3497 variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.)
3500 Accesses to objects with types with this attribute are not subjected to
3501 type-based alias analysis, but are instead assumed to be able to alias
3502 any other type of objects, just like the @code{char} type. See
3503 @option{-fstrict-aliasing} for more information on aliasing issues.
3508 typedef short __attribute__((__may_alias__)) short_a;
3514 short_a *b = (short_a *) &a;
3518 if (a == 0x12345678)
3525 If you replaced @code{short_a} with @code{short} in the variable
3526 declaration, the above program would abort when compiled with
3527 @option{-fstrict-aliasing}, which is on by default at @option{-O2} or
3528 above in recent GCC versions.
3530 @subsection ARM Type Attributes
3532 On those ARM targets that support @code{dllimport} (such as Symbian
3533 OS), you can use the @code{notshared} attribute to indicate that the
3534 virtual table and other similar data for a class should not be
3535 exported from a DLL@. For example:
3538 class __declspec(notshared) C @{
3540 __declspec(dllimport) C();
3544 __declspec(dllexport)
3548 In this code, @code{C::C} is exported from the current DLL, but the
3549 virtual table for @code{C} is not exported. (You can use
3550 @code{__attribute__} instead of @code{__declspec} if you prefer, but
3551 most Symbian OS code uses @code{__declspec}.)
3553 @subsection i386 Type Attributes
3555 Two attributes are currently defined for i386 configurations:
3556 @code{ms_struct} and @code{gcc_struct}
3560 @cindex @code{ms_struct}
3561 @cindex @code{gcc_struct}
3563 If @code{packed} is used on a structure, or if bit-fields are used
3564 it may be that the Microsoft ABI packs them differently
3565 than GCC would normally pack them. Particularly when moving packed
3566 data between functions compiled with GCC and the native Microsoft compiler
3567 (either via function call or as data in a file), it may be necessary to access
3570 Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86
3571 compilers to match the native Microsoft compiler.
3574 To specify multiple attributes, separate them by commas within the
3575 double parentheses: for example, @samp{__attribute__ ((aligned (16),
3579 @section An Inline Function is As Fast As a Macro
3580 @cindex inline functions
3581 @cindex integrating function code
3583 @cindex macros, inline alternative
3585 By declaring a function @code{inline}, you can direct GCC to
3586 integrate that function's code into the code for its callers. This
3587 makes execution faster by eliminating the function-call overhead; in
3588 addition, if any of the actual argument values are constant, their known
3589 values may permit simplifications at compile time so that not all of the
3590 inline function's code needs to be included. The effect on code size is
3591 less predictable; object code may be larger or smaller with function
3592 inlining, depending on the particular case. Inlining of functions is an
3593 optimization and it really ``works'' only in optimizing compilation. If
3594 you don't use @option{-O}, no function is really inline.
3596 Inline functions are included in the ISO C99 standard, but there are
3597 currently substantial differences between what GCC implements and what
3598 the ISO C99 standard requires.
3600 To declare a function inline, use the @code{inline} keyword in its
3601 declaration, like this:
3611 (If you are writing a header file to be included in ISO C programs, write
3612 @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.)
3613 You can also make all ``simple enough'' functions inline with the option
3614 @option{-finline-functions}.
3617 Note that certain usages in a function definition can make it unsuitable
3618 for inline substitution. Among these usages are: use of varargs, use of
3619 alloca, use of variable sized data types (@pxref{Variable Length}),
3620 use of computed goto (@pxref{Labels as Values}), use of nonlocal goto,
3621 and nested functions (@pxref{Nested Functions}). Using @option{-Winline}
3622 will warn when a function marked @code{inline} could not be substituted,
3623 and will give the reason for the failure.
3625 Note that in C and Objective-C, unlike C++, the @code{inline} keyword
3626 does not affect the linkage of the function.
3628 @cindex automatic @code{inline} for C++ member fns
3629 @cindex @code{inline} automatic for C++ member fns
3630 @cindex member fns, automatically @code{inline}
3631 @cindex C++ member fns, automatically @code{inline}
3632 @opindex fno-default-inline
3633 GCC automatically inlines member functions defined within the class
3634 body of C++ programs even if they are not explicitly declared
3635 @code{inline}. (You can override this with @option{-fno-default-inline};
3636 @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.)
3638 @cindex inline functions, omission of
3639 @opindex fkeep-inline-functions
3640 When a function is both inline and @code{static}, if all calls to the
3641 function are integrated into the caller, and the function's address is
3642 never used, then the function's own assembler code is never referenced.
3643 In this case, GCC does not actually output assembler code for the
3644 function, unless you specify the option @option{-fkeep-inline-functions}.
3645 Some calls cannot be integrated for various reasons (in particular,
3646 calls that precede the function's definition cannot be integrated, and
3647 neither can recursive calls within the definition). If there is a
3648 nonintegrated call, then the function is compiled to assembler code as
3649 usual. The function must also be compiled as usual if the program
3650 refers to its address, because that can't be inlined.
3652 @cindex non-static inline function
3653 When an inline function is not @code{static}, then the compiler must assume
3654 that there may be calls from other source files; since a global symbol can
3655 be defined only once in any program, the function must not be defined in
3656 the other source files, so the calls therein cannot be integrated.
3657 Therefore, a non-@code{static} inline function is always compiled on its
3658 own in the usual fashion.
3660 If you specify both @code{inline} and @code{extern} in the function
3661 definition, then the definition is used only for inlining. In no case
3662 is the function compiled on its own, not even if you refer to its
3663 address explicitly. Such an address becomes an external reference, as
3664 if you had only declared the function, and had not defined it.
3666 This combination of @code{inline} and @code{extern} has almost the
3667 effect of a macro. The way to use it is to put a function definition in
3668 a header file with these keywords, and put another copy of the
3669 definition (lacking @code{inline} and @code{extern}) in a library file.
3670 The definition in the header file will cause most calls to the function
3671 to be inlined. If any uses of the function remain, they will refer to
3672 the single copy in the library.
3674 Since GCC eventually will implement ISO C99 semantics for
3675 inline functions, it is best to use @code{static inline} only
3676 to guarantee compatibility. (The
3677 existing semantics will remain available when @option{-std=gnu89} is
3678 specified, but eventually the default will be @option{-std=gnu99} and
3679 that will implement the C99 semantics, though it does not do so yet.)
3681 GCC does not inline any functions when not optimizing unless you specify
3682 the @samp{always_inline} attribute for the function, like this:
3685 /* @r{Prototype.} */
3686 inline void foo (const char) __attribute__((always_inline));
3690 @section Assembler Instructions with C Expression Operands
3691 @cindex extended @code{asm}
3692 @cindex @code{asm} expressions
3693 @cindex assembler instructions
3696 In an assembler instruction using @code{asm}, you can specify the
3697 operands of the instruction using C expressions. This means you need not
3698 guess which registers or memory locations will contain the data you want
3701 You must specify an assembler instruction template much like what
3702 appears in a machine description, plus an operand constraint string for
3705 For example, here is how to use the 68881's @code{fsinx} instruction:
3708 asm ("fsinx %1,%0" : "=f" (result) : "f" (angle));
3712 Here @code{angle} is the C expression for the input operand while
3713 @code{result} is that of the output operand. Each has @samp{"f"} as its
3714 operand constraint, saying that a floating point register is required.
3715 The @samp{=} in @samp{=f} indicates that the operand is an output; all
3716 output operands' constraints must use @samp{=}. The constraints use the
3717 same language used in the machine description (@pxref{Constraints}).
3719 Each operand is described by an operand-constraint string followed by
3720 the C expression in parentheses. A colon separates the assembler
3721 template from the first output operand and another separates the last
3722 output operand from the first input, if any. Commas separate the
3723 operands within each group. The total number of operands is currently
3724 limited to 30; this limitation may be lifted in some future version of
3727 If there are no output operands but there are input operands, you must
3728 place two consecutive colons surrounding the place where the output
3731 As of GCC version 3.1, it is also possible to specify input and output
3732 operands using symbolic names which can be referenced within the
3733 assembler code. These names are specified inside square brackets
3734 preceding the constraint string, and can be referenced inside the
3735 assembler code using @code{%[@var{name}]} instead of a percentage sign
3736 followed by the operand number. Using named operands the above example
3740 asm ("fsinx %[angle],%[output]"
3741 : [output] "=f" (result)
3742 : [angle] "f" (angle));
3746 Note that the symbolic operand names have no relation whatsoever to
3747 other C identifiers. You may use any name you like, even those of
3748 existing C symbols, but you must ensure that no two operands within the same
3749 assembler construct use the same symbolic name.
3751 Output operand expressions must be lvalues; the compiler can check this.
3752 The input operands need not be lvalues. The compiler cannot check
3753 whether the operands have data types that are reasonable for the
3754 instruction being executed. It does not parse the assembler instruction
3755 template and does not know what it means or even whether it is valid
3756 assembler input. The extended @code{asm} feature is most often used for
3757 machine instructions the compiler itself does not know exist. If
3758 the output expression cannot be directly addressed (for example, it is a
3759 bit-field), your constraint must allow a register. In that case, GCC
3760 will use the register as the output of the @code{asm}, and then store
3761 that register into the output.
3763 The ordinary output operands must be write-only; GCC will assume that
3764 the values in these operands before the instruction are dead and need
3765 not be generated. Extended asm supports input-output or read-write
3766 operands. Use the constraint character @samp{+} to indicate such an
3767 operand and list it with the output operands. You should only use
3768 read-write operands when the constraints for the operand (or the
3769 operand in which only some of the bits are to be changed) allow a
3772 You may, as an alternative, logically split its function into two
3773 separate operands, one input operand and one write-only output
3774 operand. The connection between them is expressed by constraints
3775 which say they need to be in the same location when the instruction
3776 executes. You can use the same C expression for both operands, or
3777 different expressions. For example, here we write the (fictitious)
3778 @samp{combine} instruction with @code{bar} as its read-only source
3779 operand and @code{foo} as its read-write destination:
3782 asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar));
3786 The constraint @samp{"0"} for operand 1 says that it must occupy the
3787 same location as operand 0. A number in constraint is allowed only in
3788 an input operand and it must refer to an output operand.
3790 Only a number in the constraint can guarantee that one operand will be in
3791 the same place as another. The mere fact that @code{foo} is the value
3792 of both operands is not enough to guarantee that they will be in the
3793 same place in the generated assembler code. The following would not
3797 asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar));
3800 Various optimizations or reloading could cause operands 0 and 1 to be in
3801 different registers; GCC knows no reason not to do so. For example, the
3802 compiler might find a copy of the value of @code{foo} in one register and
3803 use it for operand 1, but generate the output operand 0 in a different
3804 register (copying it afterward to @code{foo}'s own address). Of course,
3805 since the register for operand 1 is not even mentioned in the assembler
3806 code, the result will not work, but GCC can't tell that.
3808 As of GCC version 3.1, one may write @code{[@var{name}]} instead of
3809 the operand number for a matching constraint. For example:
3812 asm ("cmoveq %1,%2,%[result]"
3813 : [result] "=r"(result)
3814 : "r" (test), "r"(new), "[result]"(old));
3817 Sometimes you need to make an @code{asm} operand be a specific register,
3818 but there's no matching constraint letter for that register @emph{by
3819 itself}. To force the operand into that register, use a local variable
3820 for the operand and specify the register in the variable declaration.
3821 @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any
3822 register constraint letter that matches the register:
3825 register int *p1 asm ("r0") = @dots{};
3826 register int *p2 asm ("r1") = @dots{};
3827 register int *result asm ("r0");
3828 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3831 @anchor{Example of asm with clobbered asm reg}
3832 In the above example, beware that a register that is call-clobbered by
3833 the target ABI will be overwritten by any function call in the
3834 assignment, including library calls for arithmetic operators.
3835 Assuming it is a call-clobbered register, this may happen to @code{r0}
3836 above by the assignment to @code{p2}. If you have to use such a
3837 register, use temporary variables for expressions between the register
3842 register int *p1 asm ("r0") = @dots{};
3843 register int *p2 asm ("r1") = t1;
3844 register int *result asm ("r0");
3845 asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2));
3848 Some instructions clobber specific hard registers. To describe this,
3849 write a third colon after the input operands, followed by the names of
3850 the clobbered hard registers (given as strings). Here is a realistic
3851 example for the VAX:
3854 asm volatile ("movc3 %0,%1,%2"
3855 : /* @r{no outputs} */
3856 : "g" (from), "g" (to), "g" (count)
3857 : "r0", "r1", "r2", "r3", "r4", "r5");
3860 You may not write a clobber description in a way that overlaps with an
3861 input or output operand. For example, you may not have an operand
3862 describing a register class with one member if you mention that register
3863 in the clobber list. Variables declared to live in specific registers
3864 (@pxref{Explicit Reg Vars}), and used as asm input or output operands must
3865 have no part mentioned in the clobber description.
3866 There is no way for you to specify that an input
3867 operand is modified without also specifying it as an output
3868 operand. Note that if all the output operands you specify are for this
3869 purpose (and hence unused), you will then also need to specify
3870 @code{volatile} for the @code{asm} construct, as described below, to
3871 prevent GCC from deleting the @code{asm} statement as unused.
3873 If you refer to a particular hardware register from the assembler code,
3874 you will probably have to list the register after the third colon to
3875 tell the compiler the register's value is modified. In some assemblers,
3876 the register names begin with @samp{%}; to produce one @samp{%} in the
3877 assembler code, you must write @samp{%%} in the input.
3879 If your assembler instruction can alter the condition code register, add
3880 @samp{cc} to the list of clobbered registers. GCC on some machines
3881 represents the condition codes as a specific hardware register;
3882 @samp{cc} serves to name this register. On other machines, the
3883 condition code is handled differently, and specifying @samp{cc} has no
3884 effect. But it is valid no matter what the machine.
3886 If your assembler instructions access memory in an unpredictable
3887 fashion, add @samp{memory} to the list of clobbered registers. This
3888 will cause GCC to not keep memory values cached in registers across the
3889 assembler instruction and not optimize stores or loads to that memory.
3890 You will also want to add the @code{volatile} keyword if the memory
3891 affected is not listed in the inputs or outputs of the @code{asm}, as
3892 the @samp{memory} clobber does not count as a side-effect of the
3893 @code{asm}. If you know how large the accessed memory is, you can add
3894 it as input or output but if this is not known, you should add
3895 @samp{memory}. As an example, if you access ten bytes of a string, you
3896 can use a memory input like:
3899 @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}.
3902 Note that in the following example the memory input is necessary,
3903 otherwise GCC might optimize the store to @code{x} away:
3910 asm ("magic stuff accessing an 'int' pointed to by '%1'"
3911 "=&d" (r) : "a" (y), "m" (*y));
3916 You can put multiple assembler instructions together in a single
3917 @code{asm} template, separated by the characters normally used in assembly
3918 code for the system. A combination that works in most places is a newline
3919 to break the line, plus a tab character to move to the instruction field
3920 (written as @samp{\n\t}). Sometimes semicolons can be used, if the
3921 assembler allows semicolons as a line-breaking character. Note that some
3922 assembler dialects use semicolons to start a comment.
3923 The input operands are guaranteed not to use any of the clobbered
3924 registers, and neither will the output operands' addresses, so you can
3925 read and write the clobbered registers as many times as you like. Here
3926 is an example of multiple instructions in a template; it assumes the
3927 subroutine @code{_foo} accepts arguments in registers 9 and 10:
3930 asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo"
3932 : "g" (from), "g" (to)
3936 Unless an output operand has the @samp{&} constraint modifier, GCC
3937 may allocate it in the same register as an unrelated input operand, on
3938 the assumption the inputs are consumed before the outputs are produced.
3939 This assumption may be false if the assembler code actually consists of
3940 more than one instruction. In such a case, use @samp{&} for each output
3941 operand that may not overlap an input. @xref{Modifiers}.
3943 If you want to test the condition code produced by an assembler
3944 instruction, you must include a branch and a label in the @code{asm}
3945 construct, as follows:
3948 asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:"
3954 This assumes your assembler supports local labels, as the GNU assembler
3955 and most Unix assemblers do.
3957 Speaking of labels, jumps from one @code{asm} to another are not
3958 supported. The compiler's optimizers do not know about these jumps, and
3959 therefore they cannot take account of them when deciding how to
3962 @cindex macros containing @code{asm}
3963 Usually the most convenient way to use these @code{asm} instructions is to
3964 encapsulate them in macros that look like functions. For example,
3968 (@{ double __value, __arg = (x); \
3969 asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \
3974 Here the variable @code{__arg} is used to make sure that the instruction
3975 operates on a proper @code{double} value, and to accept only those
3976 arguments @code{x} which can convert automatically to a @code{double}.
3978 Another way to make sure the instruction operates on the correct data
3979 type is to use a cast in the @code{asm}. This is different from using a
3980 variable @code{__arg} in that it converts more different types. For
3981 example, if the desired type were @code{int}, casting the argument to
3982 @code{int} would accept a pointer with no complaint, while assigning the
3983 argument to an @code{int} variable named @code{__arg} would warn about
3984 using a pointer unless the caller explicitly casts it.
3986 If an @code{asm} has output operands, GCC assumes for optimization
3987 purposes the instruction has no side effects except to change the output
3988 operands. This does not mean instructions with a side effect cannot be
3989 used, but you must be careful, because the compiler may eliminate them
3990 if the output operands aren't used, or move them out of loops, or
3991 replace two with one if they constitute a common subexpression. Also,
3992 if your instruction does have a side effect on a variable that otherwise
3993 appears not to change, the old value of the variable may be reused later
3994 if it happens to be found in a register.
3996 You can prevent an @code{asm} instruction from being deleted
3997 by writing the keyword @code{volatile} after
3998 the @code{asm}. For example:
4001 #define get_and_set_priority(new) \
4003 asm volatile ("get_and_set_priority %0, %1" \
4004 : "=g" (__old) : "g" (new)); \
4009 The @code{volatile} keyword indicates that the instruction has
4010 important side-effects. GCC will not delete a volatile @code{asm} if
4011 it is reachable. (The instruction can still be deleted if GCC can
4012 prove that control-flow will never reach the location of the
4013 instruction.) Note that even a volatile @code{asm} instruction
4014 can be moved relative to other code, including across jump
4015 instructions. For example, on many targets there is a system
4016 register which can be set to control the rounding mode of
4017 floating point operations. You might try
4018 setting it with a volatile @code{asm}, like this PowerPC example:
4021 asm volatile("mtfsf 255,%0" : : "f" (fpenv));
4026 This will not work reliably, as the compiler may move the addition back
4027 before the volatile @code{asm}. To make it work you need to add an
4028 artificial dependency to the @code{asm} referencing a variable in the code
4029 you don't want moved, for example:
4032 asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv));
4036 Similarly, you can't expect a
4037 sequence of volatile @code{asm} instructions to remain perfectly
4038 consecutive. If you want consecutive output, use a single @code{asm}.
4039 Also, GCC will perform some optimizations across a volatile @code{asm}
4040 instruction; GCC does not ``forget everything'' when it encounters
4041 a volatile @code{asm} instruction the way some other compilers do.
4043 An @code{asm} instruction without any output operands will be treated
4044 identically to a volatile @code{asm} instruction.
4046 It is a natural idea to look for a way to give access to the condition
4047 code left by the assembler instruction. However, when we attempted to
4048 implement this, we found no way to make it work reliably. The problem
4049 is that output operands might need reloading, which would result in
4050 additional following ``store'' instructions. On most machines, these
4051 instructions would alter the condition code before there was time to
4052 test it. This problem doesn't arise for ordinary ``test'' and
4053 ``compare'' instructions because they don't have any output operands.
4055 For reasons similar to those described above, it is not possible to give
4056 an assembler instruction access to the condition code left by previous
4059 If you are writing a header file that should be includable in ISO C
4060 programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate
4063 @subsection Size of an @code{asm}
4065 Some targets require that GCC track the size of each instruction used in
4066 order to generate correct code. Because the final length of an
4067 @code{asm} is only known by the assembler, GCC must make an estimate as
4068 to how big it will be. The estimate is formed by counting the number of
4069 statements in the pattern of the @code{asm} and multiplying that by the
4070 length of the longest instruction on that processor. Statements in the
4071 @code{asm} are identified by newline characters and whatever statement
4072 separator characters are supported by the assembler; on most processors
4073 this is the `@code{;}' character.
4075 Normally, GCC's estimate is perfectly adequate to ensure that correct
4076 code is generated, but it is possible to confuse the compiler if you use
4077 pseudo instructions or assembler macros that expand into multiple real
4078 instructions or if you use assembler directives that expand to more
4079 space in the object file than would be needed for a single instruction.
4080 If this happens then the assembler will produce a diagnostic saying that
4081 a label is unreachable.
4083 @subsection i386 floating point asm operands
4085 There are several rules on the usage of stack-like regs in
4086 asm_operands insns. These rules apply only to the operands that are
4091 Given a set of input regs that die in an asm_operands, it is
4092 necessary to know which are implicitly popped by the asm, and
4093 which must be explicitly popped by gcc.
4095 An input reg that is implicitly popped by the asm must be
4096 explicitly clobbered, unless it is constrained to match an
4100 For any input reg that is implicitly popped by an asm, it is
4101 necessary to know how to adjust the stack to compensate for the pop.
4102 If any non-popped input is closer to the top of the reg-stack than
4103 the implicitly popped reg, it would not be possible to know what the
4104 stack looked like---it's not clear how the rest of the stack ``slides
4107 All implicitly popped input regs must be closer to the top of
4108 the reg-stack than any input that is not implicitly popped.
4110 It is possible that if an input dies in an insn, reload might
4111 use the input reg for an output reload. Consider this example:
4114 asm ("foo" : "=t" (a) : "f" (b));
4117 This asm says that input B is not popped by the asm, and that
4118 the asm pushes a result onto the reg-stack, i.e., the stack is one
4119 deeper after the asm than it was before. But, it is possible that
4120 reload will think that it can use the same reg for both the input and
4121 the output, if input B dies in this insn.
4123 If any input operand uses the @code{f} constraint, all output reg
4124 constraints must use the @code{&} earlyclobber.
4126 The asm above would be written as
4129 asm ("foo" : "=&t" (a) : "f" (b));
4133 Some operands need to be in particular places on the stack. All
4134 output operands fall in this category---there is no other way to
4135 know which regs the outputs appear in unless the user indicates
4136 this in the constraints.
4138 Output operands must specifically indicate which reg an output
4139 appears in after an asm. @code{=f} is not allowed: the operand
4140 constraints must select a class with a single reg.
4143 Output operands may not be ``inserted'' between existing stack regs.
4144 Since no 387 opcode uses a read/write operand, all output operands
4145 are dead before the asm_operands, and are pushed by the asm_operands.
4146 It makes no sense to push anywhere but the top of the reg-stack.
4148 Output operands must start at the top of the reg-stack: output
4149 operands may not ``skip'' a reg.
4152 Some asm statements may need extra stack space for internal
4153 calculations. This can be guaranteed by clobbering stack registers
4154 unrelated to the inputs and outputs.
4158 Here are a couple of reasonable asms to want to write. This asm
4159 takes one input, which is internally popped, and produces two outputs.
4162 asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp));
4165 This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode,
4166 and replaces them with one output. The user must code the @code{st(1)}
4167 clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs.
4170 asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)");
4176 @section Controlling Names Used in Assembler Code
4177 @cindex assembler names for identifiers
4178 @cindex names used in assembler code
4179 @cindex identifiers, names in assembler code
4181 You can specify the name to be used in the assembler code for a C
4182 function or variable by writing the @code{asm} (or @code{__asm__})
4183 keyword after the declarator as follows:
4186 int foo asm ("myfoo") = 2;
4190 This specifies that the name to be used for the variable @code{foo} in
4191 the assembler code should be @samp{myfoo} rather than the usual
4194 On systems where an underscore is normally prepended to the name of a C
4195 function or variable, this feature allows you to define names for the
4196 linker that do not start with an underscore.
4198 It does not make sense to use this feature with a non-static local
4199 variable since such variables do not have assembler names. If you are
4200 trying to put the variable in a particular register, see @ref{Explicit
4201 Reg Vars}. GCC presently accepts such code with a warning, but will
4202 probably be changed to issue an error, rather than a warning, in the
4205 You cannot use @code{asm} in this way in a function @emph{definition}; but
4206 you can get the same effect by writing a declaration for the function
4207 before its definition and putting @code{asm} there, like this:
4210 extern func () asm ("FUNC");
4217 It is up to you to make sure that the assembler names you choose do not
4218 conflict with any other assembler symbols. Also, you must not use a
4219 register name; that would produce completely invalid assembler code. GCC
4220 does not as yet have the ability to store static variables in registers.
4221 Perhaps that will be added.
4223 @node Explicit Reg Vars
4224 @section Variables in Specified Registers
4225 @cindex explicit register variables
4226 @cindex variables in specified registers
4227 @cindex specified registers
4228 @cindex registers, global allocation
4230 GNU C allows you to put a few global variables into specified hardware
4231 registers. You can also specify the register in which an ordinary
4232 register variable should be allocated.
4236 Global register variables reserve registers throughout the program.
4237 This may be useful in programs such as programming language
4238 interpreters which have a couple of global variables that are accessed
4242 Local register variables in specific registers do not reserve the
4243 registers, except at the point where they are used as input or output
4244 operands in an @code{asm} statement and the @code{asm} statement itself is
4245 not deleted. The compiler's data flow analysis is capable of determining
4246 where the specified registers contain live values, and where they are
4247 available for other uses. Stores into local register variables may be deleted
4248 when they appear to be dead according to dataflow analysis. References
4249 to local register variables may be deleted or moved or simplified.
4251 These local variables are sometimes convenient for use with the extended
4252 @code{asm} feature (@pxref{Extended Asm}), if you want to write one
4253 output of the assembler instruction directly into a particular register.
4254 (This will work provided the register you specify fits the constraints
4255 specified for that operand in the @code{asm}.)
4263 @node Global Reg Vars
4264 @subsection Defining Global Register Variables
4265 @cindex global register variables
4266 @cindex registers, global variables in
4268 You can define a global register variable in GNU C like this:
4271 register int *foo asm ("a5");
4275 Here @code{a5} is the name of the register which should be used. Choose a
4276 register which is normally saved and restored by function calls on your
4277 machine, so that library routines will not clobber it.
4279 Naturally the register name is cpu-dependent, so you would need to
4280 conditionalize your program according to cpu type. The register
4281 @code{a5} would be a good choice on a 68000 for a variable of pointer
4282 type. On machines with register windows, be sure to choose a ``global''
4283 register that is not affected magically by the function call mechanism.
4285 In addition, operating systems on one type of cpu may differ in how they
4286 name the registers; then you would need additional conditionals. For
4287 example, some 68000 operating systems call this register @code{%a5}.
4289 Eventually there may be a way of asking the compiler to choose a register
4290 automatically, but first we need to figure out how it should choose and
4291 how to enable you to guide the choice. No solution is evident.
4293 Defining a global register variable in a certain register reserves that
4294 register entirely for this use, at least within the current compilation.
4295 The register will not be allocated for any other purpose in the functions
4296 in the current compilation. The register will not be saved and restored by
4297 these functions. Stores into this register are never deleted even if they
4298 would appear to be dead, but references may be deleted or moved or
4301 It is not safe to access the global register variables from signal
4302 handlers, or from more than one thread of control, because the system
4303 library routines may temporarily use the register for other things (unless
4304 you recompile them specially for the task at hand).
4306 @cindex @code{qsort}, and global register variables
4307 It is not safe for one function that uses a global register variable to
4308 call another such function @code{foo} by way of a third function
4309 @code{lose} that was compiled without knowledge of this variable (i.e.@: in a
4310 different source file in which the variable wasn't declared). This is
4311 because @code{lose} might save the register and put some other value there.
4312 For example, you can't expect a global register variable to be available in
4313 the comparison-function that you pass to @code{qsort}, since @code{qsort}
4314 might have put something else in that register. (If you are prepared to
4315 recompile @code{qsort} with the same global register variable, you can
4316 solve this problem.)
4318 If you want to recompile @code{qsort} or other source files which do not
4319 actually use your global register variable, so that they will not use that
4320 register for any other purpose, then it suffices to specify the compiler
4321 option @option{-ffixed-@var{reg}}. You need not actually add a global
4322 register declaration to their source code.
4324 A function which can alter the value of a global register variable cannot
4325 safely be called from a function compiled without this variable, because it
4326 could clobber the value the caller expects to find there on return.
4327 Therefore, the function which is the entry point into the part of the
4328 program that uses the global register variable must explicitly save and
4329 restore the value which belongs to its caller.
4331 @cindex register variable after @code{longjmp}
4332 @cindex global register after @code{longjmp}
4333 @cindex value after @code{longjmp}
4336 On most machines, @code{longjmp} will restore to each global register
4337 variable the value it had at the time of the @code{setjmp}. On some
4338 machines, however, @code{longjmp} will not change the value of global
4339 register variables. To be portable, the function that called @code{setjmp}
4340 should make other arrangements to save the values of the global register
4341 variables, and to restore them in a @code{longjmp}. This way, the same
4342 thing will happen regardless of what @code{longjmp} does.
4344 All global register variable declarations must precede all function
4345 definitions. If such a declaration could appear after function
4346 definitions, the declaration would be too late to prevent the register from
4347 being used for other purposes in the preceding functions.
4349 Global register variables may not have initial values, because an
4350 executable file has no means to supply initial contents for a register.
4352 On the SPARC, there are reports that g3 @dots{} g7 are suitable
4353 registers, but certain library functions, such as @code{getwd}, as well
4354 as the subroutines for division and remainder, modify g3 and g4. g1 and
4355 g2 are local temporaries.
4357 On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7.
4358 Of course, it will not do to use more than a few of those.
4360 @node Local Reg Vars
4361 @subsection Specifying Registers for Local Variables
4362 @cindex local variables, specifying registers
4363 @cindex specifying registers for local variables
4364 @cindex registers for local variables
4366 You can define a local register variable with a specified register
4370 register int *foo asm ("a5");
4374 Here @code{a5} is the name of the register which should be used. Note
4375 that this is the same syntax used for defining global register
4376 variables, but for a local variable it would appear within a function.
4378 Naturally the register name is cpu-dependent, but this is not a
4379 problem, since specific registers are most often useful with explicit
4380 assembler instructions (@pxref{Extended Asm}). Both of these things
4381 generally require that you conditionalize your program according to
4384 In addition, operating systems on one type of cpu may differ in how they
4385 name the registers; then you would need additional conditionals. For
4386 example, some 68000 operating systems call this register @code{%a5}.
4388 Defining such a register variable does not reserve the register; it
4389 remains available for other uses in places where flow control determines
4390 the variable's value is not live.
4392 This option does not guarantee that GCC will generate code that has
4393 this variable in the register you specify at all times. You may not
4394 code an explicit reference to this register in the @emph{assembler
4395 instruction template} part of an @code{asm} statement and assume it will
4396 always refer to this variable. However, using the variable as an
4397 @code{asm} @emph{operand} guarantees that the specified register is used
4400 Stores into local register variables may be deleted when they appear to be dead
4401 according to dataflow analysis. References to local register variables may
4402 be deleted or moved or simplified.
4404 As for global register variables, it's recommended that you choose a
4405 register which is normally saved and restored by function calls on
4406 your machine, so that library routines will not clobber it. A common
4407 pitfall is to initialize multiple call-clobbered registers with
4408 arbitrary expressions, where a function call or library call for an
4409 arithmetic operator will overwrite a register value from a previous
4410 assignment, for example @code{r0} below:
4412 register int *p1 asm ("r0") = @dots{};
4413 register int *p2 asm ("r1") = @dots{};
4415 In those cases, a solution is to use a temporary variable for
4416 each arbitrary expression. @xref{Example of asm with clobbered asm reg}.
4418 @node Alternate Keywords
4419 @section Alternate Keywords
4420 @cindex alternate keywords
4421 @cindex keywords, alternate
4423 @option{-ansi} and the various @option{-std} options disable certain
4424 keywords. This causes trouble when you want to use GNU C extensions, or
4425 a general-purpose header file that should be usable by all programs,
4426 including ISO C programs. The keywords @code{asm}, @code{typeof} and
4427 @code{inline} are not available in programs compiled with
4428 @option{-ansi} or @option{-std} (although @code{inline} can be used in a
4429 program compiled with @option{-std=c99}). The ISO C99 keyword
4430 @code{restrict} is only available when @option{-std=gnu99} (which will
4431 eventually be the default) or @option{-std=c99} (or the equivalent
4432 @option{-std=iso9899:1999}) is used.
4434 The way to solve these problems is to put @samp{__} at the beginning and
4435 end of each problematical keyword. For example, use @code{__asm__}
4436 instead of @code{asm}, and @code{__inline__} instead of @code{inline}.
4438 Other C compilers won't accept these alternative keywords; if you want to
4439 compile with another compiler, you can define the alternate keywords as
4440 macros to replace them with the customary keywords. It looks like this:
4448 @findex __extension__
4450 @option{-pedantic} and other options cause warnings for many GNU C extensions.
4452 prevent such warnings within one expression by writing
4453 @code{__extension__} before the expression. @code{__extension__} has no
4454 effect aside from this.
4456 @node Incomplete Enums
4457 @section Incomplete @code{enum} Types
4459 You can define an @code{enum} tag without specifying its possible values.
4460 This results in an incomplete type, much like what you get if you write
4461 @code{struct foo} without describing the elements. A later declaration
4462 which does specify the possible values completes the type.
4464 You can't allocate variables or storage using the type while it is
4465 incomplete. However, you can work with pointers to that type.
4467 This extension may not be very useful, but it makes the handling of
4468 @code{enum} more consistent with the way @code{struct} and @code{union}
4471 This extension is not supported by GNU C++.
4473 @node Function Names
4474 @section Function Names as Strings
4475 @cindex @code{__func__} identifier
4476 @cindex @code{__FUNCTION__} identifier
4477 @cindex @code{__PRETTY_FUNCTION__} identifier
4479 GCC provides three magic variables which hold the name of the current
4480 function, as a string. The first of these is @code{__func__}, which
4481 is part of the C99 standard:
4484 The identifier @code{__func__} is implicitly declared by the translator
4485 as if, immediately following the opening brace of each function
4486 definition, the declaration
4489 static const char __func__[] = "function-name";
4492 appeared, where function-name is the name of the lexically-enclosing
4493 function. This name is the unadorned name of the function.
4496 @code{__FUNCTION__} is another name for @code{__func__}. Older
4497 versions of GCC recognize only this name. However, it is not
4498 standardized. For maximum portability, we recommend you use
4499 @code{__func__}, but provide a fallback definition with the
4503 #if __STDC_VERSION__ < 199901L
4505 # define __func__ __FUNCTION__
4507 # define __func__ "<unknown>"
4512 In C, @code{__PRETTY_FUNCTION__} is yet another name for
4513 @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains
4514 the type signature of the function as well as its bare name. For
4515 example, this program:
4519 extern int printf (char *, ...);
4526 printf ("__FUNCTION__ = %s\n", __FUNCTION__);
4527 printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__);
4545 __PRETTY_FUNCTION__ = void a::sub(int)
4548 These identifiers are not preprocessor macros. In GCC 3.3 and
4549 earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__}
4550 were treated as string literals; they could be used to initialize
4551 @code{char} arrays, and they could be concatenated with other string
4552 literals. GCC 3.4 and later treat them as variables, like
4553 @code{__func__}. In C++, @code{__FUNCTION__} and
4554 @code{__PRETTY_FUNCTION__} have always been variables.
4556 @node Return Address
4557 @section Getting the Return or Frame Address of a Function
4559 These functions may be used to get information about the callers of a
4562 @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level})
4563 This function returns the return address of the current function, or of
4564 one of its callers. The @var{level} argument is number of frames to
4565 scan up the call stack. A value of @code{0} yields the return address
4566 of the current function, a value of @code{1} yields the return address
4567 of the caller of the current function, and so forth. When inlining
4568 the expected behavior is that the function will return the address of
4569 the function that will be returned to. To work around this behavior use
4570 the @code{noinline} function attribute.
4572 The @var{level} argument must be a constant integer.
4574 On some machines it may be impossible to determine the return address of
4575 any function other than the current one; in such cases, or when the top
4576 of the stack has been reached, this function will return @code{0} or a
4577 random value. In addition, @code{__builtin_frame_address} may be used
4578 to determine if the top of the stack has been reached.
4580 This function should only be used with a nonzero argument for debugging
4584 @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level})
4585 This function is similar to @code{__builtin_return_address}, but it
4586 returns the address of the function frame rather than the return address
4587 of the function. Calling @code{__builtin_frame_address} with a value of
4588 @code{0} yields the frame address of the current function, a value of
4589 @code{1} yields the frame address of the caller of the current function,
4592 The frame is the area on the stack which holds local variables and saved
4593 registers. The frame address is normally the address of the first word
4594 pushed on to the stack by the function. However, the exact definition
4595 depends upon the processor and the calling convention. If the processor
4596 has a dedicated frame pointer register, and the function has a frame,
4597 then @code{__builtin_frame_address} will return the value of the frame
4600 On some machines it may be impossible to determine the frame address of
4601 any function other than the current one; in such cases, or when the top
4602 of the stack has been reached, this function will return @code{0} if
4603 the first frame pointer is properly initialized by the startup code.
4605 This function should only be used with a nonzero argument for debugging
4609 @node Vector Extensions
4610 @section Using vector instructions through built-in functions
4612 On some targets, the instruction set contains SIMD vector instructions that
4613 operate on multiple values contained in one large register at the same time.
4614 For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used
4617 The first step in using these extensions is to provide the necessary data
4618 types. This should be done using an appropriate @code{typedef}:
4621 typedef int v4si __attribute__ ((vector_size (16)));
4624 The @code{int} type specifies the base type, while the attribute specifies
4625 the vector size for the variable, measured in bytes. For example, the
4626 declaration above causes the compiler to set the mode for the @code{v4si}
4627 type to be 16 bytes wide and divided into @code{int} sized units. For
4628 a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the
4629 corresponding mode of @code{foo} will be @acronym{V4SI}.
4631 The @code{vector_size} attribute is only applicable to integral and
4632 float scalars, although arrays, pointers, and function return values
4633 are allowed in conjunction with this construct.
4635 All the basic integer types can be used as base types, both as signed
4636 and as unsigned: @code{char}, @code{short}, @code{int}, @code{long},
4637 @code{long long}. In addition, @code{float} and @code{double} can be
4638 used to build floating-point vector types.
4640 Specifying a combination that is not valid for the current architecture
4641 will cause GCC to synthesize the instructions using a narrower mode.
4642 For example, if you specify a variable of type @code{V4SI} and your
4643 architecture does not allow for this specific SIMD type, GCC will
4644 produce code that uses 4 @code{SIs}.
4646 The types defined in this manner can be used with a subset of normal C
4647 operations. Currently, GCC will allow using the following operators
4648 on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@.
4650 The operations behave like C++ @code{valarrays}. Addition is defined as
4651 the addition of the corresponding elements of the operands. For
4652 example, in the code below, each of the 4 elements in @var{a} will be
4653 added to the corresponding 4 elements in @var{b} and the resulting
4654 vector will be stored in @var{c}.
4657 typedef int v4si __attribute__ ((vector_size (16)));
4664 Subtraction, multiplication, division, and the logical operations
4665 operate in a similar manner. Likewise, the result of using the unary
4666 minus or complement operators on a vector type is a vector whose
4667 elements are the negative or complemented values of the corresponding
4668 elements in the operand.
4670 You can declare variables and use them in function calls and returns, as
4671 well as in assignments and some casts. You can specify a vector type as
4672 a return type for a function. Vector types can also be used as function
4673 arguments. It is possible to cast from one vector type to another,
4674 provided they are of the same size (in fact, you can also cast vectors
4675 to and from other datatypes of the same size).
4677 You cannot operate between vectors of different lengths or different
4678 signedness without a cast.
4680 A port that supports hardware vector operations, usually provides a set
4681 of built-in functions that can be used to operate on vectors. For
4682 example, a function to add two vectors and multiply the result by a
4683 third could look like this:
4686 v4si f (v4si a, v4si b, v4si c)
4688 v4si tmp = __builtin_addv4si (a, b);
4689 return __builtin_mulv4si (tmp, c);
4696 @findex __builtin_offsetof
4698 GCC implements for both C and C++ a syntactic extension to implement
4699 the @code{offsetof} macro.
4703 "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")"
4705 offsetof_member_designator:
4707 | offsetof_member_designator "." @code{identifier}
4708 | offsetof_member_designator "[" @code{expr} "]"
4711 This extension is sufficient such that
4714 #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member})
4717 is a suitable definition of the @code{offsetof} macro. In C++, @var{type}
4718 may be dependent. In either case, @var{member} may consist of a single
4719 identifier, or a sequence of member accesses and array references.
4721 @node Atomic Builtins
4722 @section Built-in functions for atomic memory access
4724 The following builtins are intended to be compatible with those described
4725 in the @cite{Intel Itanium Processor-specific Application Binary Interface},
4726 section 7.4. As such, they depart from the normal GCC practice of using
4727 the ``__builtin_'' prefix, and further that they are overloaded such that
4728 they work on multiple types.
4730 The definition given in the Intel documentation allows only for the use of
4731 the types @code{int}, @code{long}, @code{long long} as well as their unsigned
4732 counterparts. GCC will allow any integral scalar or pointer type that is
4733 1, 2, 4 or 8 bytes in length.
4735 Not all operations are supported by all target processors. If a particular
4736 operation cannot be implemented on the target processor, a warning will be
4737 generated and a call an external function will be generated. The external
4738 function will carry the same name as the builtin, with an additional suffix
4739 @samp{_@var{n}} where @var{n} is the size of the data type.
4741 @c ??? Should we have a mechanism to suppress this warning? This is almost
4742 @c useful for implementing the operation under the control of an external
4745 In most cases, these builtins are considered a @dfn{full barrier}. That is,
4746 no memory operand will be moved across the operation, either forward or
4747 backward. Further, instructions will be issued as necessary to prevent the
4748 processor from speculating loads across the operation and from queuing stores
4749 after the operation.
4751 All of the routines are are described in the Intel documentation to take
4752 ``an optional list of variables protected by the memory barrier''. It's
4753 not clear what is meant by that; it could mean that @emph{only} the
4754 following variables are protected, or it could mean that these variables
4755 should in addition be protected. At present GCC ignores this list and
4756 protects all variables which are globally accessible. If in the future
4757 we make some use of this list, an empty list will continue to mean all
4758 globally accessible variables.
4761 @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...)
4762 @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...)
4763 @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...)
4764 @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...)
4765 @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...)
4766 @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...)
4767 @findex __sync_fetch_and_add
4768 @findex __sync_fetch_and_sub
4769 @findex __sync_fetch_and_or
4770 @findex __sync_fetch_and_and
4771 @findex __sync_fetch_and_xor
4772 @findex __sync_fetch_and_nand
4773 These builtins perform the operation suggested by the name, and
4774 returns the value that had previously been in memory. That is,
4777 @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @}
4778 @{ tmp = *ptr; *ptr = ~tmp & value; return tmp; @} // nand
4781 @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...)
4782 @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...)
4783 @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...)
4784 @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...)
4785 @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...)
4786 @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...)
4787 @findex __sync_add_and_fetch
4788 @findex __sync_sub_and_fetch
4789 @findex __sync_or_and_fetch
4790 @findex __sync_and_and_fetch
4791 @findex __sync_xor_and_fetch
4792 @findex __sync_nand_and_fetch
4793 These builtins perform the operation suggested by the name, and
4794 return the new value. That is,
4797 @{ *ptr @var{op}= value; return *ptr; @}
4798 @{ *ptr = ~*ptr & value; return *ptr; @} // nand
4801 @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
4802 @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...)
4803 @findex __sync_bool_compare_and_swap
4804 @findex __sync_val_compare_and_swap
4805 These builtins perform an atomic compare and swap. That is, if the current
4806 value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into
4809 The ``bool'' version returns true if the comparison is successful and
4810 @var{newval} was written. The ``val'' version returns the contents
4811 of @code{*@var{ptr}} before the operation.
4813 @item __sync_synchronize (...)
4814 @findex __sync_synchronize
4815 This builtin issues a full memory barrier.
4817 @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...)
4818 @findex __sync_lock_test_and_set
4819 This builtin, as described by Intel, is not a traditional test-and-set
4820 operation, but rather an atomic exchange operation. It writes @var{value}
4821 into @code{*@var{ptr}}, and returns the previous contents of
4824 Many targets have only minimal support for such locks, and do not support
4825 a full exchange operation. In this case, a target may support reduced
4826 functionality here by which the @emph{only} valid value to store is the
4827 immediate constant 1. The exact value actually stored in @code{*@var{ptr}}
4828 is implementation defined.
4830 This builtin is not a full barrier, but rather an @dfn{acquire barrier}.
4831 This means that references after the builtin cannot move to (or be
4832 speculated to) before the builtin, but previous memory stores may not
4833 be globally visible yet, and previous memory loads may not yet be
4836 @item void __sync_lock_release (@var{type} *ptr, ...)
4837 @findex __sync_lock_release
4838 This builtin releases the lock acquired by @code{__sync_lock_test_and_set}.
4839 Normally this means writing the constant 0 to @code{*@var{ptr}}.
4841 This builtin is not a full barrier, but rather a @dfn{release barrier}.
4842 This means that all previous memory stores are globally visible, and all
4843 previous memory loads have been satisfied, but following memory reads
4844 are not prevented from being speculated to before the barrier.
4847 @node Object Size Checking
4848 @section Object Size Checking Builtins
4849 @findex __builtin_object_size
4850 @findex __builtin___memcpy_chk
4851 @findex __builtin___mempcpy_chk
4852 @findex __builtin___memmove_chk
4853 @findex __builtin___memset_chk
4854 @findex __builtin___strcpy_chk
4855 @findex __builtin___stpcpy_chk
4856 @findex __builtin___strncpy_chk
4857 @findex __builtin___strcat_chk
4858 @findex __builtin___strncat_chk
4859 @findex __builtin___sprintf_chk
4860 @findex __builtin___snprintf_chk
4861 @findex __builtin___vsprintf_chk
4862 @findex __builtin___vsnprintf_chk
4863 @findex __builtin___printf_chk
4864 @findex __builtin___vprintf_chk
4865 @findex __builtin___fprintf_chk
4866 @findex __builtin___vfprintf_chk
4868 GCC implements a limited buffer overflow protection mechanism
4869 that can prevent some buffer overflow attacks.
4871 @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type})
4872 is a built-in construct that returns a constant number of bytes from
4873 @var{ptr} to the end of the object @var{ptr} pointer points to
4874 (if known at compile time). @code{__builtin_object_size} never evaluates
4875 its arguments for side-effects. If there are any side-effects in them, it
4876 returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
4877 for @var{type} 2 or 3. If there are multiple objects @var{ptr} can
4878 point to and all of them are known at compile time, the returned number
4879 is the maximum of remaining byte counts in those objects if @var{type} & 2 is
4880 0 and minimum if nonzero. If it is not possible to determine which objects
4881 @var{ptr} points to at compile time, @code{__builtin_object_size} should
4882 return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0}
4883 for @var{type} 2 or 3.
4885 @var{type} is an integer constant from 0 to 3. If the least significant
4886 bit is clear, objects are whole variables, if it is set, a closest
4887 surrounding subobject is considered the object a pointer points to.
4888 The second bit determines if maximum or minimum of remaining bytes
4892 struct V @{ char buf1[10]; int b; char buf2[10]; @} var;
4893 char *p = &var.buf1[1], *q = &var.b;
4895 /* Here the object p points to is var. */
4896 assert (__builtin_object_size (p, 0) == sizeof (var) - 1);
4897 /* The subobject p points to is var.buf1. */
4898 assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1);
4899 /* The object q points to is var. */
4900 assert (__builtin_object_size (q, 0)
4901 == (char *) (&var + 1) - (char *) &var.b);
4902 /* The subobject q points to is var.b. */
4903 assert (__builtin_object_size (q, 1) == sizeof (var.b));
4907 There are built-in functions added for many common string operation
4908 functions, e.g. for @code{memcpy} @code{__builtin___memcpy_chk}
4909 built-in is provided. This built-in has an additional last argument,
4910 which is the number of bytes remaining in object the @var{dest}
4911 argument points to or @code{(size_t) -1} if the size is not known.
4913 The built-in functions are optimized into the normal string functions
4914 like @code{memcpy} if the last argument is @code{(size_t) -1} or if
4915 it is known at compile time that the destination object will not
4916 be overflown. If the compiler can determine at compile time the
4917 object will be always overflown, it issues a warning.
4919 The intended use can be e.g.
4923 #define bos0(dest) __builtin_object_size (dest, 0)
4924 #define memcpy(dest, src, n) \
4925 __builtin___memcpy_chk (dest, src, n, bos0 (dest))
4929 /* It is unknown what object p points to, so this is optimized
4930 into plain memcpy - no checking is possible. */
4931 memcpy (p, "abcde", n);
4932 /* Destination is known and length too. It is known at compile
4933 time there will be no overflow. */
4934 memcpy (&buf[5], "abcde", 5);
4935 /* Destination is known, but the length is not known at compile time.
4936 This will result in __memcpy_chk call that can check for overflow
4938 memcpy (&buf[5], "abcde", n);
4939 /* Destination is known and it is known at compile time there will
4940 be overflow. There will be a warning and __memcpy_chk call that
4941 will abort the program at runtime. */
4942 memcpy (&buf[6], "abcde", 5);
4945 Such built-in functions are provided for @code{memcpy}, @code{mempcpy},
4946 @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy},
4947 @code{strcat} and @code{strncat}.
4949 There are also checking built-in functions for formatted output functions.
4951 int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...);
4952 int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os,
4953 const char *fmt, ...);
4954 int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt,
4956 int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os,
4957 const char *fmt, va_list ap);
4960 The added @var{flag} argument is passed unchanged to @code{__sprintf_chk}
4961 etc. functions and can contain implementation specific flags on what
4962 additional security measures the checking function might take, such as
4963 handling @code{%n} differently.
4965 The @var{os} argument is the object size @var{s} points to, like in the
4966 other built-in functions. There is a small difference in the behavior
4967 though, if @var{os} is @code{(size_t) -1}, the built-in functions are
4968 optimized into the non-checking functions only if @var{flag} is 0, otherwise
4969 the checking function is called with @var{os} argument set to
4972 In addition to this, there are checking built-in functions
4973 @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk},
4974 @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}.
4975 These have just one additional argument, @var{flag}, right before
4976 format string @var{fmt}. If the compiler is able to optimize them to
4977 @code{fputc} etc. functions, it will, otherwise the checking function
4978 should be called and the @var{flag} argument passed to it.
4980 @node Other Builtins
4981 @section Other built-in functions provided by GCC
4982 @cindex built-in functions
4983 @findex __builtin_isgreater
4984 @findex __builtin_isgreaterequal
4985 @findex __builtin_isless
4986 @findex __builtin_islessequal
4987 @findex __builtin_islessgreater
4988 @findex __builtin_isunordered
4989 @findex __builtin_powi
4990 @findex __builtin_powif
4991 @findex __builtin_powil
5149 @findex fprintf_unlocked
5151 @findex fputs_unlocked
5261 @findex printf_unlocked
5290 @findex significandf
5291 @findex significandl
5362 GCC provides a large number of built-in functions other than the ones
5363 mentioned above. Some of these are for internal use in the processing
5364 of exceptions or variable-length argument lists and will not be
5365 documented here because they may change from time to time; we do not
5366 recommend general use of these functions.
5368 The remaining functions are provided for optimization purposes.
5370 @opindex fno-builtin
5371 GCC includes built-in versions of many of the functions in the standard
5372 C library. The versions prefixed with @code{__builtin_} will always be
5373 treated as having the same meaning as the C library function even if you
5374 specify the @option{-fno-builtin} option. (@pxref{C Dialect Options})
5375 Many of these functions are only optimized in certain cases; if they are
5376 not optimized in a particular case, a call to the library function will
5381 Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or
5382 @option{-std=c99}), the functions
5383 @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero},
5384 @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml},
5385 @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll},
5386 @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked},
5387 @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext},
5388 @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0},
5389 @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn},
5390 @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10},
5391 @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl},
5392 @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl},
5393 @code{significandf}, @code{significandl}, @code{significand},
5394 @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy},
5395 @code{stpncpy}, @code{strcasecmp}, @code{strdup}, @code{strfmon},
5396 @code{strncasecmp}, @code{strndup}, @code{toascii}, @code{y0f},
5397 @code{y0l}, @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf},
5398 @code{ynl} and @code{yn}
5399 may be handled as built-in functions.
5400 All these functions have corresponding versions
5401 prefixed with @code{__builtin_}, which may be used even in strict C89
5404 The ISO C99 functions
5405 @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf},
5406 @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh},
5407 @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf},
5408 @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos},
5409 @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf},
5410 @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin},
5411 @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh},
5412 @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt},
5413 @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl},
5414 @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf},
5415 @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog},
5416 @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl},
5417 @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf},
5418 @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal},
5419 @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl},
5420 @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf},
5421 @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan},
5422 @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl},
5423 @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f},
5424 @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim},
5425 @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax},
5426 @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf},
5427 @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb},
5428 @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf},
5429 @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl},
5430 @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround},
5431 @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l},
5432 @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf},
5433 @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl},
5434 @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint},
5435 @code{nextafterf}, @code{nextafterl}, @code{nextafter},
5436 @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward},
5437 @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof},
5438 @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint},
5439 @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf},
5440 @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl},
5441 @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal},
5442 @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc},
5443 @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf}
5444 are handled as built-in functions
5445 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5447 There are also built-in versions of the ISO C99 functions
5448 @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f},
5449 @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill},
5450 @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf},
5451 @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl},
5452 @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf},
5453 @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl},
5454 @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf},
5455 @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl},
5456 @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl}
5457 that are recognized in any mode since ISO C90 reserves these names for
5458 the purpose to which ISO C99 puts them. All these functions have
5459 corresponding versions prefixed with @code{__builtin_}.
5461 The ISO C94 functions
5462 @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit},
5463 @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct},
5464 @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and
5466 are handled as built-in functions
5467 except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}).
5469 The ISO C90 functions
5470 @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2},
5471 @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos},
5472 @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod},
5473 @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf},
5474 @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit},
5475 @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct},
5476 @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower},
5477 @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log},
5478 @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf},
5479 @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf},
5480 @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt},
5481 @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp},
5482 @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat},
5483 @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr},
5484 @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf},
5485 @code{vprintf} and @code{vsprintf}
5486 are all recognized as built-in functions unless
5487 @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}}
5488 is specified for an individual function). All of these functions have
5489 corresponding versions prefixed with @code{__builtin_}.
5491 GCC provides built-in versions of the ISO C99 floating point comparison
5492 macros that avoid raising exceptions for unordered operands. They have
5493 the same names as the standard macros ( @code{isgreater},
5494 @code{isgreaterequal}, @code{isless}, @code{islessequal},
5495 @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_}
5496 prefixed. We intend for a library implementor to be able to simply
5497 @code{#define} each standard macro to its built-in equivalent.
5499 @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2})
5501 You can use the built-in function @code{__builtin_types_compatible_p} to
5502 determine whether two types are the same.
5504 This built-in function returns 1 if the unqualified versions of the
5505 types @var{type1} and @var{type2} (which are types, not expressions) are
5506 compatible, 0 otherwise. The result of this built-in function can be
5507 used in integer constant expressions.
5509 This built-in function ignores top level qualifiers (e.g., @code{const},
5510 @code{volatile}). For example, @code{int} is equivalent to @code{const
5513 The type @code{int[]} and @code{int[5]} are compatible. On the other
5514 hand, @code{int} and @code{char *} are not compatible, even if the size
5515 of their types, on the particular architecture are the same. Also, the
5516 amount of pointer indirection is taken into account when determining
5517 similarity. Consequently, @code{short *} is not similar to
5518 @code{short **}. Furthermore, two types that are typedefed are
5519 considered compatible if their underlying types are compatible.
5521 An @code{enum} type is not considered to be compatible with another
5522 @code{enum} type even if both are compatible with the same integer
5523 type; this is what the C standard specifies.
5524 For example, @code{enum @{foo, bar@}} is not similar to
5525 @code{enum @{hot, dog@}}.
5527 You would typically use this function in code whose execution varies
5528 depending on the arguments' types. For example:
5534 if (__builtin_types_compatible_p (typeof (x), long double)) \
5535 tmp = foo_long_double (tmp); \
5536 else if (__builtin_types_compatible_p (typeof (x), double)) \
5537 tmp = foo_double (tmp); \
5538 else if (__builtin_types_compatible_p (typeof (x), float)) \
5539 tmp = foo_float (tmp); \
5546 @emph{Note:} This construct is only available for C@.
5550 @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2})
5552 You can use the built-in function @code{__builtin_choose_expr} to
5553 evaluate code depending on the value of a constant expression. This
5554 built-in function returns @var{exp1} if @var{const_exp}, which is a
5555 constant expression that must be able to be determined at compile time,
5556 is nonzero. Otherwise it returns 0.
5558 This built-in function is analogous to the @samp{? :} operator in C,
5559 except that the expression returned has its type unaltered by promotion
5560 rules. Also, the built-in function does not evaluate the expression
5561 that was not chosen. For example, if @var{const_exp} evaluates to true,
5562 @var{exp2} is not evaluated even if it has side-effects.
5564 This built-in function can return an lvalue if the chosen argument is an
5567 If @var{exp1} is returned, the return type is the same as @var{exp1}'s
5568 type. Similarly, if @var{exp2} is returned, its return type is the same
5575 __builtin_choose_expr ( \
5576 __builtin_types_compatible_p (typeof (x), double), \
5578 __builtin_choose_expr ( \
5579 __builtin_types_compatible_p (typeof (x), float), \
5581 /* @r{The void expression results in a compile-time error} \
5582 @r{when assigning the result to something.} */ \
5586 @emph{Note:} This construct is only available for C@. Furthermore, the
5587 unused expression (@var{exp1} or @var{exp2} depending on the value of
5588 @var{const_exp}) may still generate syntax errors. This may change in
5593 @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp})
5594 You can use the built-in function @code{__builtin_constant_p} to
5595 determine if a value is known to be constant at compile-time and hence
5596 that GCC can perform constant-folding on expressions involving that
5597 value. The argument of the function is the value to test. The function
5598 returns the integer 1 if the argument is known to be a compile-time
5599 constant and 0 if it is not known to be a compile-time constant. A
5600 return of 0 does not indicate that the value is @emph{not} a constant,
5601 but merely that GCC cannot prove it is a constant with the specified
5602 value of the @option{-O} option.
5604 You would typically use this function in an embedded application where
5605 memory was a critical resource. If you have some complex calculation,
5606 you may want it to be folded if it involves constants, but need to call
5607 a function if it does not. For example:
5610 #define Scale_Value(X) \
5611 (__builtin_constant_p (X) \
5612 ? ((X) * SCALE + OFFSET) : Scale (X))
5615 You may use this built-in function in either a macro or an inline
5616 function. However, if you use it in an inlined function and pass an
5617 argument of the function as the argument to the built-in, GCC will
5618 never return 1 when you call the inline function with a string constant
5619 or compound literal (@pxref{Compound Literals}) and will not return 1
5620 when you pass a constant numeric value to the inline function unless you
5621 specify the @option{-O} option.
5623 You may also use @code{__builtin_constant_p} in initializers for static
5624 data. For instance, you can write
5627 static const int table[] = @{
5628 __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1,
5634 This is an acceptable initializer even if @var{EXPRESSION} is not a
5635 constant expression. GCC must be more conservative about evaluating the
5636 built-in in this case, because it has no opportunity to perform
5639 Previous versions of GCC did not accept this built-in in data
5640 initializers. The earliest version where it is completely safe is
5644 @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c})
5645 @opindex fprofile-arcs
5646 You may use @code{__builtin_expect} to provide the compiler with
5647 branch prediction information. In general, you should prefer to
5648 use actual profile feedback for this (@option{-fprofile-arcs}), as
5649 programmers are notoriously bad at predicting how their programs
5650 actually perform. However, there are applications in which this
5651 data is hard to collect.
5653 The return value is the value of @var{exp}, which should be an
5654 integral expression. The value of @var{c} must be a compile-time
5655 constant. The semantics of the built-in are that it is expected
5656 that @var{exp} == @var{c}. For example:
5659 if (__builtin_expect (x, 0))
5664 would indicate that we do not expect to call @code{foo}, since
5665 we expect @code{x} to be zero. Since you are limited to integral
5666 expressions for @var{exp}, you should use constructions such as
5669 if (__builtin_expect (ptr != NULL, 1))
5674 when testing pointer or floating-point values.
5677 @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...)
5678 This function is used to minimize cache-miss latency by moving data into
5679 a cache before it is accessed.
5680 You can insert calls to @code{__builtin_prefetch} into code for which
5681 you know addresses of data in memory that is likely to be accessed soon.
5682 If the target supports them, data prefetch instructions will be generated.
5683 If the prefetch is done early enough before the access then the data will
5684 be in the cache by the time it is accessed.
5686 The value of @var{addr} is the address of the memory to prefetch.
5687 There are two optional arguments, @var{rw} and @var{locality}.
5688 The value of @var{rw} is a compile-time constant one or zero; one
5689 means that the prefetch is preparing for a write to the memory address
5690 and zero, the default, means that the prefetch is preparing for a read.
5691 The value @var{locality} must be a compile-time constant integer between
5692 zero and three. A value of zero means that the data has no temporal
5693 locality, so it need not be left in the cache after the access. A value
5694 of three means that the data has a high degree of temporal locality and
5695 should be left in all levels of cache possible. Values of one and two
5696 mean, respectively, a low or moderate degree of temporal locality. The
5700 for (i = 0; i < n; i++)
5703 __builtin_prefetch (&a[i+j], 1, 1);
5704 __builtin_prefetch (&b[i+j], 0, 1);
5709 Data prefetch does not generate faults if @var{addr} is invalid, but
5710 the address expression itself must be valid. For example, a prefetch
5711 of @code{p->next} will not fault if @code{p->next} is not a valid
5712 address, but evaluation will fault if @code{p} is not a valid address.
5714 If the target does not support data prefetch, the address expression
5715 is evaluated if it includes side effects but no other code is generated
5716 and GCC does not issue a warning.
5719 @deftypefn {Built-in Function} double __builtin_huge_val (void)
5720 Returns a positive infinity, if supported by the floating-point format,
5721 else @code{DBL_MAX}. This function is suitable for implementing the
5722 ISO C macro @code{HUGE_VAL}.
5725 @deftypefn {Built-in Function} float __builtin_huge_valf (void)
5726 Similar to @code{__builtin_huge_val}, except the return type is @code{float}.
5729 @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void)
5730 Similar to @code{__builtin_huge_val}, except the return
5731 type is @code{long double}.
5734 @deftypefn {Built-in Function} double __builtin_inf (void)
5735 Similar to @code{__builtin_huge_val}, except a warning is generated
5736 if the target floating-point format does not support infinities.
5739 @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void)
5740 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}.
5743 @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void)
5744 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}.
5747 @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void)
5748 Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}.
5751 @deftypefn {Built-in Function} float __builtin_inff (void)
5752 Similar to @code{__builtin_inf}, except the return type is @code{float}.
5753 This function is suitable for implementing the ISO C99 macro @code{INFINITY}.
5756 @deftypefn {Built-in Function} {long double} __builtin_infl (void)
5757 Similar to @code{__builtin_inf}, except the return
5758 type is @code{long double}.
5761 @deftypefn {Built-in Function} double __builtin_nan (const char *str)
5762 This is an implementation of the ISO C99 function @code{nan}.
5764 Since ISO C99 defines this function in terms of @code{strtod}, which we
5765 do not implement, a description of the parsing is in order. The string
5766 is parsed as by @code{strtol}; that is, the base is recognized by
5767 leading @samp{0} or @samp{0x} prefixes. The number parsed is placed
5768 in the significand such that the least significant bit of the number
5769 is at the least significant bit of the significand. The number is
5770 truncated to fit the significand field provided. The significand is
5771 forced to be a quiet NaN@.
5773 This function, if given a string literal all of which would have been
5774 consumed by strtol, is evaluated early enough that it is considered a
5775 compile-time constant.
5778 @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str)
5779 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}.
5782 @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str)
5783 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}.
5786 @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str)
5787 Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}.
5790 @deftypefn {Built-in Function} float __builtin_nanf (const char *str)
5791 Similar to @code{__builtin_nan}, except the return type is @code{float}.
5794 @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str)
5795 Similar to @code{__builtin_nan}, except the return type is @code{long double}.
5798 @deftypefn {Built-in Function} double __builtin_nans (const char *str)
5799 Similar to @code{__builtin_nan}, except the significand is forced
5800 to be a signaling NaN@. The @code{nans} function is proposed by
5801 @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}.
5804 @deftypefn {Built-in Function} float __builtin_nansf (const char *str)
5805 Similar to @code{__builtin_nans}, except the return type is @code{float}.
5808 @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str)
5809 Similar to @code{__builtin_nans}, except the return type is @code{long double}.
5812 @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x)
5813 Returns one plus the index of the least significant 1-bit of @var{x}, or
5814 if @var{x} is zero, returns zero.
5817 @deftypefn {Built-in Function} int __builtin_clz (unsigned int x)
5818 Returns the number of leading 0-bits in @var{x}, starting at the most
5819 significant bit position. If @var{x} is 0, the result is undefined.
5822 @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x)
5823 Returns the number of trailing 0-bits in @var{x}, starting at the least
5824 significant bit position. If @var{x} is 0, the result is undefined.
5827 @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x)
5828 Returns the number of 1-bits in @var{x}.
5831 @deftypefn {Built-in Function} int __builtin_parity (unsigned int x)
5832 Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x}
5836 @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long)
5837 Similar to @code{__builtin_ffs}, except the argument type is
5838 @code{unsigned long}.
5841 @deftypefn {Built-in Function} int __builtin_clzl (unsigned long)
5842 Similar to @code{__builtin_clz}, except the argument type is
5843 @code{unsigned long}.
5846 @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long)
5847 Similar to @code{__builtin_ctz}, except the argument type is
5848 @code{unsigned long}.
5851 @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long)
5852 Similar to @code{__builtin_popcount}, except the argument type is
5853 @code{unsigned long}.
5856 @deftypefn {Built-in Function} int __builtin_parityl (unsigned long)
5857 Similar to @code{__builtin_parity}, except the argument type is
5858 @code{unsigned long}.
5861 @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long)
5862 Similar to @code{__builtin_ffs}, except the argument type is
5863 @code{unsigned long long}.
5866 @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long)
5867 Similar to @code{__builtin_clz}, except the argument type is
5868 @code{unsigned long long}.
5871 @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long)
5872 Similar to @code{__builtin_ctz}, except the argument type is
5873 @code{unsigned long long}.
5876 @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long)
5877 Similar to @code{__builtin_popcount}, except the argument type is
5878 @code{unsigned long long}.
5881 @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long)
5882 Similar to @code{__builtin_parity}, except the argument type is
5883 @code{unsigned long long}.
5886 @deftypefn {Built-in Function} double __builtin_powi (double, int)
5887 Returns the first argument raised to the power of the second. Unlike the
5888 @code{pow} function no guarantees about precision and rounding are made.
5891 @deftypefn {Built-in Function} float __builtin_powif (float, int)
5892 Similar to @code{__builtin_powi}, except the argument and return types
5896 @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int)
5897 Similar to @code{__builtin_powi}, except the argument and return types
5898 are @code{long double}.
5902 @node Target Builtins
5903 @section Built-in Functions Specific to Particular Target Machines
5905 On some target machines, GCC supports many built-in functions specific
5906 to those machines. Generally these generate calls to specific machine
5907 instructions, but allow the compiler to schedule those calls.
5910 * Alpha Built-in Functions::
5911 * ARM Built-in Functions::
5912 * Blackfin Built-in Functions::
5913 * FR-V Built-in Functions::
5914 * X86 Built-in Functions::
5915 * MIPS DSP Built-in Functions::
5916 * MIPS Paired-Single Support::
5917 * PowerPC AltiVec Built-in Functions::
5918 * SPARC VIS Built-in Functions::
5921 @node Alpha Built-in Functions
5922 @subsection Alpha Built-in Functions
5924 These built-in functions are available for the Alpha family of
5925 processors, depending on the command-line switches used.
5927 The following built-in functions are always available. They
5928 all generate the machine instruction that is part of the name.
5931 long __builtin_alpha_implver (void)
5932 long __builtin_alpha_rpcc (void)
5933 long __builtin_alpha_amask (long)
5934 long __builtin_alpha_cmpbge (long, long)
5935 long __builtin_alpha_extbl (long, long)
5936 long __builtin_alpha_extwl (long, long)
5937 long __builtin_alpha_extll (long, long)
5938 long __builtin_alpha_extql (long, long)
5939 long __builtin_alpha_extwh (long, long)
5940 long __builtin_alpha_extlh (long, long)
5941 long __builtin_alpha_extqh (long, long)
5942 long __builtin_alpha_insbl (long, long)
5943 long __builtin_alpha_inswl (long, long)
5944 long __builtin_alpha_insll (long, long)
5945 long __builtin_alpha_insql (long, long)
5946 long __builtin_alpha_inswh (long, long)
5947 long __builtin_alpha_inslh (long, long)
5948 long __builtin_alpha_insqh (long, long)
5949 long __builtin_alpha_mskbl (long, long)
5950 long __builtin_alpha_mskwl (long, long)
5951 long __builtin_alpha_mskll (long, long)
5952 long __builtin_alpha_mskql (long, long)
5953 long __builtin_alpha_mskwh (long, long)
5954 long __builtin_alpha_msklh (long, long)
5955 long __builtin_alpha_mskqh (long, long)
5956 long __builtin_alpha_umulh (long, long)
5957 long __builtin_alpha_zap (long, long)
5958 long __builtin_alpha_zapnot (long, long)
5961 The following built-in functions are always with @option{-mmax}
5962 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or
5963 later. They all generate the machine instruction that is part
5967 long __builtin_alpha_pklb (long)
5968 long __builtin_alpha_pkwb (long)
5969 long __builtin_alpha_unpkbl (long)
5970 long __builtin_alpha_unpkbw (long)
5971 long __builtin_alpha_minub8 (long, long)
5972 long __builtin_alpha_minsb8 (long, long)
5973 long __builtin_alpha_minuw4 (long, long)
5974 long __builtin_alpha_minsw4 (long, long)
5975 long __builtin_alpha_maxub8 (long, long)
5976 long __builtin_alpha_maxsb8 (long, long)
5977 long __builtin_alpha_maxuw4 (long, long)
5978 long __builtin_alpha_maxsw4 (long, long)
5979 long __builtin_alpha_perr (long, long)
5982 The following built-in functions are always with @option{-mcix}
5983 or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or
5984 later. They all generate the machine instruction that is part
5988 long __builtin_alpha_cttz (long)
5989 long __builtin_alpha_ctlz (long)
5990 long __builtin_alpha_ctpop (long)
5993 The following builtins are available on systems that use the OSF/1
5994 PALcode. Normally they invoke the @code{rduniq} and @code{wruniq}
5995 PAL calls, but when invoked with @option{-mtls-kernel}, they invoke
5996 @code{rdval} and @code{wrval}.
5999 void *__builtin_thread_pointer (void)
6000 void __builtin_set_thread_pointer (void *)
6003 @node ARM Built-in Functions
6004 @subsection ARM Built-in Functions
6006 These built-in functions are available for the ARM family of
6007 processors, when the @option{-mcpu=iwmmxt} switch is used:
6010 typedef int v2si __attribute__ ((vector_size (8)));
6011 typedef short v4hi __attribute__ ((vector_size (8)));
6012 typedef char v8qi __attribute__ ((vector_size (8)));
6014 int __builtin_arm_getwcx (int)
6015 void __builtin_arm_setwcx (int, int)
6016 int __builtin_arm_textrmsb (v8qi, int)
6017 int __builtin_arm_textrmsh (v4hi, int)
6018 int __builtin_arm_textrmsw (v2si, int)
6019 int __builtin_arm_textrmub (v8qi, int)
6020 int __builtin_arm_textrmuh (v4hi, int)
6021 int __builtin_arm_textrmuw (v2si, int)
6022 v8qi __builtin_arm_tinsrb (v8qi, int)
6023 v4hi __builtin_arm_tinsrh (v4hi, int)
6024 v2si __builtin_arm_tinsrw (v2si, int)
6025 long long __builtin_arm_tmia (long long, int, int)
6026 long long __builtin_arm_tmiabb (long long, int, int)
6027 long long __builtin_arm_tmiabt (long long, int, int)
6028 long long __builtin_arm_tmiaph (long long, int, int)
6029 long long __builtin_arm_tmiatb (long long, int, int)
6030 long long __builtin_arm_tmiatt (long long, int, int)
6031 int __builtin_arm_tmovmskb (v8qi)
6032 int __builtin_arm_tmovmskh (v4hi)
6033 int __builtin_arm_tmovmskw (v2si)
6034 long long __builtin_arm_waccb (v8qi)
6035 long long __builtin_arm_wacch (v4hi)
6036 long long __builtin_arm_waccw (v2si)
6037 v8qi __builtin_arm_waddb (v8qi, v8qi)
6038 v8qi __builtin_arm_waddbss (v8qi, v8qi)
6039 v8qi __builtin_arm_waddbus (v8qi, v8qi)
6040 v4hi __builtin_arm_waddh (v4hi, v4hi)
6041 v4hi __builtin_arm_waddhss (v4hi, v4hi)
6042 v4hi __builtin_arm_waddhus (v4hi, v4hi)
6043 v2si __builtin_arm_waddw (v2si, v2si)
6044 v2si __builtin_arm_waddwss (v2si, v2si)
6045 v2si __builtin_arm_waddwus (v2si, v2si)
6046 v8qi __builtin_arm_walign (v8qi, v8qi, int)
6047 long long __builtin_arm_wand(long long, long long)
6048 long long __builtin_arm_wandn (long long, long long)
6049 v8qi __builtin_arm_wavg2b (v8qi, v8qi)
6050 v8qi __builtin_arm_wavg2br (v8qi, v8qi)
6051 v4hi __builtin_arm_wavg2h (v4hi, v4hi)
6052 v4hi __builtin_arm_wavg2hr (v4hi, v4hi)
6053 v8qi __builtin_arm_wcmpeqb (v8qi, v8qi)
6054 v4hi __builtin_arm_wcmpeqh (v4hi, v4hi)
6055 v2si __builtin_arm_wcmpeqw (v2si, v2si)
6056 v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi)
6057 v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi)
6058 v2si __builtin_arm_wcmpgtsw (v2si, v2si)
6059 v8qi __builtin_arm_wcmpgtub (v8qi, v8qi)
6060 v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi)
6061 v2si __builtin_arm_wcmpgtuw (v2si, v2si)
6062 long long __builtin_arm_wmacs (long long, v4hi, v4hi)
6063 long long __builtin_arm_wmacsz (v4hi, v4hi)
6064 long long __builtin_arm_wmacu (long long, v4hi, v4hi)
6065 long long __builtin_arm_wmacuz (v4hi, v4hi)
6066 v4hi __builtin_arm_wmadds (v4hi, v4hi)
6067 v4hi __builtin_arm_wmaddu (v4hi, v4hi)
6068 v8qi __builtin_arm_wmaxsb (v8qi, v8qi)
6069 v4hi __builtin_arm_wmaxsh (v4hi, v4hi)
6070 v2si __builtin_arm_wmaxsw (v2si, v2si)
6071 v8qi __builtin_arm_wmaxub (v8qi, v8qi)
6072 v4hi __builtin_arm_wmaxuh (v4hi, v4hi)
6073 v2si __builtin_arm_wmaxuw (v2si, v2si)
6074 v8qi __builtin_arm_wminsb (v8qi, v8qi)
6075 v4hi __builtin_arm_wminsh (v4hi, v4hi)
6076 v2si __builtin_arm_wminsw (v2si, v2si)
6077 v8qi __builtin_arm_wminub (v8qi, v8qi)
6078 v4hi __builtin_arm_wminuh (v4hi, v4hi)
6079 v2si __builtin_arm_wminuw (v2si, v2si)
6080 v4hi __builtin_arm_wmulsm (v4hi, v4hi)
6081 v4hi __builtin_arm_wmulul (v4hi, v4hi)
6082 v4hi __builtin_arm_wmulum (v4hi, v4hi)
6083 long long __builtin_arm_wor (long long, long long)
6084 v2si __builtin_arm_wpackdss (long long, long long)
6085 v2si __builtin_arm_wpackdus (long long, long long)
6086 v8qi __builtin_arm_wpackhss (v4hi, v4hi)
6087 v8qi __builtin_arm_wpackhus (v4hi, v4hi)
6088 v4hi __builtin_arm_wpackwss (v2si, v2si)
6089 v4hi __builtin_arm_wpackwus (v2si, v2si)
6090 long long __builtin_arm_wrord (long long, long long)
6091 long long __builtin_arm_wrordi (long long, int)
6092 v4hi __builtin_arm_wrorh (v4hi, long long)
6093 v4hi __builtin_arm_wrorhi (v4hi, int)
6094 v2si __builtin_arm_wrorw (v2si, long long)
6095 v2si __builtin_arm_wrorwi (v2si, int)
6096 v2si __builtin_arm_wsadb (v8qi, v8qi)
6097 v2si __builtin_arm_wsadbz (v8qi, v8qi)
6098 v2si __builtin_arm_wsadh (v4hi, v4hi)
6099 v2si __builtin_arm_wsadhz (v4hi, v4hi)
6100 v4hi __builtin_arm_wshufh (v4hi, int)
6101 long long __builtin_arm_wslld (long long, long long)
6102 long long __builtin_arm_wslldi (long long, int)
6103 v4hi __builtin_arm_wsllh (v4hi, long long)
6104 v4hi __builtin_arm_wsllhi (v4hi, int)
6105 v2si __builtin_arm_wsllw (v2si, long long)
6106 v2si __builtin_arm_wsllwi (v2si, int)
6107 long long __builtin_arm_wsrad (long long, long long)
6108 long long __builtin_arm_wsradi (long long, int)
6109 v4hi __builtin_arm_wsrah (v4hi, long long)
6110 v4hi __builtin_arm_wsrahi (v4hi, int)
6111 v2si __builtin_arm_wsraw (v2si, long long)
6112 v2si __builtin_arm_wsrawi (v2si, int)
6113 long long __builtin_arm_wsrld (long long, long long)
6114 long long __builtin_arm_wsrldi (long long, int)
6115 v4hi __builtin_arm_wsrlh (v4hi, long long)
6116 v4hi __builtin_arm_wsrlhi (v4hi, int)
6117 v2si __builtin_arm_wsrlw (v2si, long long)
6118 v2si __builtin_arm_wsrlwi (v2si, int)
6119 v8qi __builtin_arm_wsubb (v8qi, v8qi)
6120 v8qi __builtin_arm_wsubbss (v8qi, v8qi)
6121 v8qi __builtin_arm_wsubbus (v8qi, v8qi)
6122 v4hi __builtin_arm_wsubh (v4hi, v4hi)
6123 v4hi __builtin_arm_wsubhss (v4hi, v4hi)
6124 v4hi __builtin_arm_wsubhus (v4hi, v4hi)
6125 v2si __builtin_arm_wsubw (v2si, v2si)
6126 v2si __builtin_arm_wsubwss (v2si, v2si)
6127 v2si __builtin_arm_wsubwus (v2si, v2si)
6128 v4hi __builtin_arm_wunpckehsb (v8qi)
6129 v2si __builtin_arm_wunpckehsh (v4hi)
6130 long long __builtin_arm_wunpckehsw (v2si)
6131 v4hi __builtin_arm_wunpckehub (v8qi)
6132 v2si __builtin_arm_wunpckehuh (v4hi)
6133 long long __builtin_arm_wunpckehuw (v2si)
6134 v4hi __builtin_arm_wunpckelsb (v8qi)
6135 v2si __builtin_arm_wunpckelsh (v4hi)
6136 long long __builtin_arm_wunpckelsw (v2si)
6137 v4hi __builtin_arm_wunpckelub (v8qi)
6138 v2si __builtin_arm_wunpckeluh (v4hi)
6139 long long __builtin_arm_wunpckeluw (v2si)
6140 v8qi __builtin_arm_wunpckihb (v8qi, v8qi)
6141 v4hi __builtin_arm_wunpckihh (v4hi, v4hi)
6142 v2si __builtin_arm_wunpckihw (v2si, v2si)
6143 v8qi __builtin_arm_wunpckilb (v8qi, v8qi)
6144 v4hi __builtin_arm_wunpckilh (v4hi, v4hi)
6145 v2si __builtin_arm_wunpckilw (v2si, v2si)
6146 long long __builtin_arm_wxor (long long, long long)
6147 long long __builtin_arm_wzero ()
6150 @node Blackfin Built-in Functions
6151 @subsection Blackfin Built-in Functions
6153 Currently, there are two Blackfin-specific built-in functions. These are
6154 used for generating @code{CSYNC} and @code{SSYNC} machine insns without
6155 using inline assembly; by using these built-in functions the compiler can
6156 automatically add workarounds for hardware errata involving these
6157 instructions. These functions are named as follows:
6160 void __builtin_bfin_csync (void)
6161 void __builtin_bfin_ssync (void)
6164 @node FR-V Built-in Functions
6165 @subsection FR-V Built-in Functions
6167 GCC provides many FR-V-specific built-in functions. In general,
6168 these functions are intended to be compatible with those described
6169 by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu
6170 Semiconductor}. The two exceptions are @code{__MDUNPACKH} and
6171 @code{__MBTOHE}, the gcc forms of which pass 128-bit values by
6172 pointer rather than by value.
6174 Most of the functions are named after specific FR-V instructions.
6175 Such functions are said to be ``directly mapped'' and are summarized
6176 here in tabular form.
6180 * Directly-mapped Integer Functions::
6181 * Directly-mapped Media Functions::
6182 * Raw read/write Functions::
6183 * Other Built-in Functions::
6186 @node Argument Types
6187 @subsubsection Argument Types
6189 The arguments to the built-in functions can be divided into three groups:
6190 register numbers, compile-time constants and run-time values. In order
6191 to make this classification clear at a glance, the arguments and return
6192 values are given the following pseudo types:
6194 @multitable @columnfractions .20 .30 .15 .35
6195 @item Pseudo type @tab Real C type @tab Constant? @tab Description
6196 @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword
6197 @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word
6198 @item @code{sw1} @tab @code{int} @tab No @tab a signed word
6199 @item @code{uw2} @tab @code{unsigned long long} @tab No
6200 @tab an unsigned doubleword
6201 @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword
6202 @item @code{const} @tab @code{int} @tab Yes @tab an integer constant
6203 @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number
6204 @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number
6207 These pseudo types are not defined by GCC, they are simply a notational
6208 convenience used in this manual.
6210 Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2}
6211 and @code{sw2} are evaluated at run time. They correspond to
6212 register operands in the underlying FR-V instructions.
6214 @code{const} arguments represent immediate operands in the underlying
6215 FR-V instructions. They must be compile-time constants.
6217 @code{acc} arguments are evaluated at compile time and specify the number
6218 of an accumulator register. For example, an @code{acc} argument of 2
6219 will select the ACC2 register.
6221 @code{iacc} arguments are similar to @code{acc} arguments but specify the
6222 number of an IACC register. See @pxref{Other Built-in Functions}
6225 @node Directly-mapped Integer Functions
6226 @subsubsection Directly-mapped Integer Functions
6228 The functions listed below map directly to FR-V I-type instructions.
6230 @multitable @columnfractions .45 .32 .23
6231 @item Function prototype @tab Example usage @tab Assembly output
6232 @item @code{sw1 __ADDSS (sw1, sw1)}
6233 @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})}
6234 @tab @code{ADDSS @var{a},@var{b},@var{c}}
6235 @item @code{sw1 __SCAN (sw1, sw1)}
6236 @tab @code{@var{c} = __SCAN (@var{a}, @var{b})}
6237 @tab @code{SCAN @var{a},@var{b},@var{c}}
6238 @item @code{sw1 __SCUTSS (sw1)}
6239 @tab @code{@var{b} = __SCUTSS (@var{a})}
6240 @tab @code{SCUTSS @var{a},@var{b}}
6241 @item @code{sw1 __SLASS (sw1, sw1)}
6242 @tab @code{@var{c} = __SLASS (@var{a}, @var{b})}
6243 @tab @code{SLASS @var{a},@var{b},@var{c}}
6244 @item @code{void __SMASS (sw1, sw1)}
6245 @tab @code{__SMASS (@var{a}, @var{b})}
6246 @tab @code{SMASS @var{a},@var{b}}
6247 @item @code{void __SMSSS (sw1, sw1)}
6248 @tab @code{__SMSSS (@var{a}, @var{b})}
6249 @tab @code{SMSSS @var{a},@var{b}}
6250 @item @code{void __SMU (sw1, sw1)}
6251 @tab @code{__SMU (@var{a}, @var{b})}
6252 @tab @code{SMU @var{a},@var{b}}
6253 @item @code{sw2 __SMUL (sw1, sw1)}
6254 @tab @code{@var{c} = __SMUL (@var{a}, @var{b})}
6255 @tab @code{SMUL @var{a},@var{b},@var{c}}
6256 @item @code{sw1 __SUBSS (sw1, sw1)}
6257 @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})}
6258 @tab @code{SUBSS @var{a},@var{b},@var{c}}
6259 @item @code{uw2 __UMUL (uw1, uw1)}
6260 @tab @code{@var{c} = __UMUL (@var{a}, @var{b})}
6261 @tab @code{UMUL @var{a},@var{b},@var{c}}
6264 @node Directly-mapped Media Functions
6265 @subsubsection Directly-mapped Media Functions
6267 The functions listed below map directly to FR-V M-type instructions.
6269 @multitable @columnfractions .45 .32 .23
6270 @item Function prototype @tab Example usage @tab Assembly output
6271 @item @code{uw1 __MABSHS (sw1)}
6272 @tab @code{@var{b} = __MABSHS (@var{a})}
6273 @tab @code{MABSHS @var{a},@var{b}}
6274 @item @code{void __MADDACCS (acc, acc)}
6275 @tab @code{__MADDACCS (@var{b}, @var{a})}
6276 @tab @code{MADDACCS @var{a},@var{b}}
6277 @item @code{sw1 __MADDHSS (sw1, sw1)}
6278 @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})}
6279 @tab @code{MADDHSS @var{a},@var{b},@var{c}}
6280 @item @code{uw1 __MADDHUS (uw1, uw1)}
6281 @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})}
6282 @tab @code{MADDHUS @var{a},@var{b},@var{c}}
6283 @item @code{uw1 __MAND (uw1, uw1)}
6284 @tab @code{@var{c} = __MAND (@var{a}, @var{b})}
6285 @tab @code{MAND @var{a},@var{b},@var{c}}
6286 @item @code{void __MASACCS (acc, acc)}
6287 @tab @code{__MASACCS (@var{b}, @var{a})}
6288 @tab @code{MASACCS @var{a},@var{b}}
6289 @item @code{uw1 __MAVEH (uw1, uw1)}
6290 @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})}
6291 @tab @code{MAVEH @var{a},@var{b},@var{c}}
6292 @item @code{uw2 __MBTOH (uw1)}
6293 @tab @code{@var{b} = __MBTOH (@var{a})}
6294 @tab @code{MBTOH @var{a},@var{b}}
6295 @item @code{void __MBTOHE (uw1 *, uw1)}
6296 @tab @code{__MBTOHE (&@var{b}, @var{a})}
6297 @tab @code{MBTOHE @var{a},@var{b}}
6298 @item @code{void __MCLRACC (acc)}
6299 @tab @code{__MCLRACC (@var{a})}
6300 @tab @code{MCLRACC @var{a}}
6301 @item @code{void __MCLRACCA (void)}
6302 @tab @code{__MCLRACCA ()}
6303 @tab @code{MCLRACCA}
6304 @item @code{uw1 __Mcop1 (uw1, uw1)}
6305 @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})}
6306 @tab @code{Mcop1 @var{a},@var{b},@var{c}}
6307 @item @code{uw1 __Mcop2 (uw1, uw1)}
6308 @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})}
6309 @tab @code{Mcop2 @var{a},@var{b},@var{c}}
6310 @item @code{uw1 __MCPLHI (uw2, const)}
6311 @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})}
6312 @tab @code{MCPLHI @var{a},#@var{b},@var{c}}
6313 @item @code{uw1 __MCPLI (uw2, const)}
6314 @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})}
6315 @tab @code{MCPLI @var{a},#@var{b},@var{c}}
6316 @item @code{void __MCPXIS (acc, sw1, sw1)}
6317 @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})}
6318 @tab @code{MCPXIS @var{a},@var{b},@var{c}}
6319 @item @code{void __MCPXIU (acc, uw1, uw1)}
6320 @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})}
6321 @tab @code{MCPXIU @var{a},@var{b},@var{c}}
6322 @item @code{void __MCPXRS (acc, sw1, sw1)}
6323 @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})}
6324 @tab @code{MCPXRS @var{a},@var{b},@var{c}}
6325 @item @code{void __MCPXRU (acc, uw1, uw1)}
6326 @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})}
6327 @tab @code{MCPXRU @var{a},@var{b},@var{c}}
6328 @item @code{uw1 __MCUT (acc, uw1)}
6329 @tab @code{@var{c} = __MCUT (@var{a}, @var{b})}
6330 @tab @code{MCUT @var{a},@var{b},@var{c}}
6331 @item @code{uw1 __MCUTSS (acc, sw1)}
6332 @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})}
6333 @tab @code{MCUTSS @var{a},@var{b},@var{c}}
6334 @item @code{void __MDADDACCS (acc, acc)}
6335 @tab @code{__MDADDACCS (@var{b}, @var{a})}
6336 @tab @code{MDADDACCS @var{a},@var{b}}
6337 @item @code{void __MDASACCS (acc, acc)}
6338 @tab @code{__MDASACCS (@var{b}, @var{a})}
6339 @tab @code{MDASACCS @var{a},@var{b}}
6340 @item @code{uw2 __MDCUTSSI (acc, const)}
6341 @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})}
6342 @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}}
6343 @item @code{uw2 __MDPACKH (uw2, uw2)}
6344 @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})}
6345 @tab @code{MDPACKH @var{a},@var{b},@var{c}}
6346 @item @code{uw2 __MDROTLI (uw2, const)}
6347 @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})}
6348 @tab @code{MDROTLI @var{a},#@var{b},@var{c}}
6349 @item @code{void __MDSUBACCS (acc, acc)}
6350 @tab @code{__MDSUBACCS (@var{b}, @var{a})}
6351 @tab @code{MDSUBACCS @var{a},@var{b}}
6352 @item @code{void __MDUNPACKH (uw1 *, uw2)}
6353 @tab @code{__MDUNPACKH (&@var{b}, @var{a})}
6354 @tab @code{MDUNPACKH @var{a},@var{b}}
6355 @item @code{uw2 __MEXPDHD (uw1, const)}
6356 @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})}
6357 @tab @code{MEXPDHD @var{a},#@var{b},@var{c}}
6358 @item @code{uw1 __MEXPDHW (uw1, const)}
6359 @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})}
6360 @tab @code{MEXPDHW @var{a},#@var{b},@var{c}}
6361 @item @code{uw1 __MHDSETH (uw1, const)}
6362 @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})}
6363 @tab @code{MHDSETH @var{a},#@var{b},@var{c}}
6364 @item @code{sw1 __MHDSETS (const)}
6365 @tab @code{@var{b} = __MHDSETS (@var{a})}
6366 @tab @code{MHDSETS #@var{a},@var{b}}
6367 @item @code{uw1 __MHSETHIH (uw1, const)}
6368 @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})}
6369 @tab @code{MHSETHIH #@var{a},@var{b}}
6370 @item @code{sw1 __MHSETHIS (sw1, const)}
6371 @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})}
6372 @tab @code{MHSETHIS #@var{a},@var{b}}
6373 @item @code{uw1 __MHSETLOH (uw1, const)}
6374 @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})}
6375 @tab @code{MHSETLOH #@var{a},@var{b}}
6376 @item @code{sw1 __MHSETLOS (sw1, const)}
6377 @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})}
6378 @tab @code{MHSETLOS #@var{a},@var{b}}
6379 @item @code{uw1 __MHTOB (uw2)}
6380 @tab @code{@var{b} = __MHTOB (@var{a})}
6381 @tab @code{MHTOB @var{a},@var{b}}
6382 @item @code{void __MMACHS (acc, sw1, sw1)}
6383 @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})}
6384 @tab @code{MMACHS @var{a},@var{b},@var{c}}
6385 @item @code{void __MMACHU (acc, uw1, uw1)}
6386 @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})}
6387 @tab @code{MMACHU @var{a},@var{b},@var{c}}
6388 @item @code{void __MMRDHS (acc, sw1, sw1)}
6389 @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})}
6390 @tab @code{MMRDHS @var{a},@var{b},@var{c}}
6391 @item @code{void __MMRDHU (acc, uw1, uw1)}
6392 @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})}
6393 @tab @code{MMRDHU @var{a},@var{b},@var{c}}
6394 @item @code{void __MMULHS (acc, sw1, sw1)}
6395 @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})}
6396 @tab @code{MMULHS @var{a},@var{b},@var{c}}
6397 @item @code{void __MMULHU (acc, uw1, uw1)}
6398 @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})}
6399 @tab @code{MMULHU @var{a},@var{b},@var{c}}
6400 @item @code{void __MMULXHS (acc, sw1, sw1)}
6401 @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})}
6402 @tab @code{MMULXHS @var{a},@var{b},@var{c}}
6403 @item @code{void __MMULXHU (acc, uw1, uw1)}
6404 @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})}
6405 @tab @code{MMULXHU @var{a},@var{b},@var{c}}
6406 @item @code{uw1 __MNOT (uw1)}
6407 @tab @code{@var{b} = __MNOT (@var{a})}
6408 @tab @code{MNOT @var{a},@var{b}}
6409 @item @code{uw1 __MOR (uw1, uw1)}
6410 @tab @code{@var{c} = __MOR (@var{a}, @var{b})}
6411 @tab @code{MOR @var{a},@var{b},@var{c}}
6412 @item @code{uw1 __MPACKH (uh, uh)}
6413 @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})}
6414 @tab @code{MPACKH @var{a},@var{b},@var{c}}
6415 @item @code{sw2 __MQADDHSS (sw2, sw2)}
6416 @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})}
6417 @tab @code{MQADDHSS @var{a},@var{b},@var{c}}
6418 @item @code{uw2 __MQADDHUS (uw2, uw2)}
6419 @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})}
6420 @tab @code{MQADDHUS @var{a},@var{b},@var{c}}
6421 @item @code{void __MQCPXIS (acc, sw2, sw2)}
6422 @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})}
6423 @tab @code{MQCPXIS @var{a},@var{b},@var{c}}
6424 @item @code{void __MQCPXIU (acc, uw2, uw2)}
6425 @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})}
6426 @tab @code{MQCPXIU @var{a},@var{b},@var{c}}
6427 @item @code{void __MQCPXRS (acc, sw2, sw2)}
6428 @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})}
6429 @tab @code{MQCPXRS @var{a},@var{b},@var{c}}
6430 @item @code{void __MQCPXRU (acc, uw2, uw2)}
6431 @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})}
6432 @tab @code{MQCPXRU @var{a},@var{b},@var{c}}
6433 @item @code{sw2 __MQLCLRHS (sw2, sw2)}
6434 @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})}
6435 @tab @code{MQLCLRHS @var{a},@var{b},@var{c}}
6436 @item @code{sw2 __MQLMTHS (sw2, sw2)}
6437 @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})}
6438 @tab @code{MQLMTHS @var{a},@var{b},@var{c}}
6439 @item @code{void __MQMACHS (acc, sw2, sw2)}
6440 @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})}
6441 @tab @code{MQMACHS @var{a},@var{b},@var{c}}
6442 @item @code{void __MQMACHU (acc, uw2, uw2)}
6443 @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})}
6444 @tab @code{MQMACHU @var{a},@var{b},@var{c}}
6445 @item @code{void __MQMACXHS (acc, sw2, sw2)}
6446 @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})}
6447 @tab @code{MQMACXHS @var{a},@var{b},@var{c}}
6448 @item @code{void __MQMULHS (acc, sw2, sw2)}
6449 @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})}
6450 @tab @code{MQMULHS @var{a},@var{b},@var{c}}
6451 @item @code{void __MQMULHU (acc, uw2, uw2)}
6452 @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})}
6453 @tab @code{MQMULHU @var{a},@var{b},@var{c}}
6454 @item @code{void __MQMULXHS (acc, sw2, sw2)}
6455 @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})}
6456 @tab @code{MQMULXHS @var{a},@var{b},@var{c}}
6457 @item @code{void __MQMULXHU (acc, uw2, uw2)}
6458 @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})}
6459 @tab @code{MQMULXHU @var{a},@var{b},@var{c}}
6460 @item @code{sw2 __MQSATHS (sw2, sw2)}
6461 @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})}
6462 @tab @code{MQSATHS @var{a},@var{b},@var{c}}
6463 @item @code{uw2 __MQSLLHI (uw2, int)}
6464 @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})}
6465 @tab @code{MQSLLHI @var{a},@var{b},@var{c}}
6466 @item @code{sw2 __MQSRAHI (sw2, int)}
6467 @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})}
6468 @tab @code{MQSRAHI @var{a},@var{b},@var{c}}
6469 @item @code{sw2 __MQSUBHSS (sw2, sw2)}
6470 @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})}
6471 @tab @code{MQSUBHSS @var{a},@var{b},@var{c}}
6472 @item @code{uw2 __MQSUBHUS (uw2, uw2)}
6473 @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})}
6474 @tab @code{MQSUBHUS @var{a},@var{b},@var{c}}
6475 @item @code{void __MQXMACHS (acc, sw2, sw2)}
6476 @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})}
6477 @tab @code{MQXMACHS @var{a},@var{b},@var{c}}
6478 @item @code{void __MQXMACXHS (acc, sw2, sw2)}
6479 @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})}
6480 @tab @code{MQXMACXHS @var{a},@var{b},@var{c}}
6481 @item @code{uw1 __MRDACC (acc)}
6482 @tab @code{@var{b} = __MRDACC (@var{a})}
6483 @tab @code{MRDACC @var{a},@var{b}}
6484 @item @code{uw1 __MRDACCG (acc)}
6485 @tab @code{@var{b} = __MRDACCG (@var{a})}
6486 @tab @code{MRDACCG @var{a},@var{b}}
6487 @item @code{uw1 __MROTLI (uw1, const)}
6488 @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})}
6489 @tab @code{MROTLI @var{a},#@var{b},@var{c}}
6490 @item @code{uw1 __MROTRI (uw1, const)}
6491 @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})}
6492 @tab @code{MROTRI @var{a},#@var{b},@var{c}}
6493 @item @code{sw1 __MSATHS (sw1, sw1)}
6494 @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})}
6495 @tab @code{MSATHS @var{a},@var{b},@var{c}}
6496 @item @code{uw1 __MSATHU (uw1, uw1)}
6497 @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})}
6498 @tab @code{MSATHU @var{a},@var{b},@var{c}}
6499 @item @code{uw1 __MSLLHI (uw1, const)}
6500 @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})}
6501 @tab @code{MSLLHI @var{a},#@var{b},@var{c}}
6502 @item @code{sw1 __MSRAHI (sw1, const)}
6503 @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})}
6504 @tab @code{MSRAHI @var{a},#@var{b},@var{c}}
6505 @item @code{uw1 __MSRLHI (uw1, const)}
6506 @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})}
6507 @tab @code{MSRLHI @var{a},#@var{b},@var{c}}
6508 @item @code{void __MSUBACCS (acc, acc)}
6509 @tab @code{__MSUBACCS (@var{b}, @var{a})}
6510 @tab @code{MSUBACCS @var{a},@var{b}}
6511 @item @code{sw1 __MSUBHSS (sw1, sw1)}
6512 @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})}
6513 @tab @code{MSUBHSS @var{a},@var{b},@var{c}}
6514 @item @code{uw1 __MSUBHUS (uw1, uw1)}
6515 @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})}
6516 @tab @code{MSUBHUS @var{a},@var{b},@var{c}}
6517 @item @code{void __MTRAP (void)}
6518 @tab @code{__MTRAP ()}
6520 @item @code{uw2 __MUNPACKH (uw1)}
6521 @tab @code{@var{b} = __MUNPACKH (@var{a})}
6522 @tab @code{MUNPACKH @var{a},@var{b}}
6523 @item @code{uw1 __MWCUT (uw2, uw1)}
6524 @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})}
6525 @tab @code{MWCUT @var{a},@var{b},@var{c}}
6526 @item @code{void __MWTACC (acc, uw1)}
6527 @tab @code{__MWTACC (@var{b}, @var{a})}
6528 @tab @code{MWTACC @var{a},@var{b}}
6529 @item @code{void __MWTACCG (acc, uw1)}
6530 @tab @code{__MWTACCG (@var{b}, @var{a})}
6531 @tab @code{MWTACCG @var{a},@var{b}}
6532 @item @code{uw1 __MXOR (uw1, uw1)}
6533 @tab @code{@var{c} = __MXOR (@var{a}, @var{b})}
6534 @tab @code{MXOR @var{a},@var{b},@var{c}}
6537 @node Raw read/write Functions
6538 @subsubsection Raw read/write Functions
6540 This sections describes built-in functions related to read and write
6541 instructions to access memory. These functions generate
6542 @code{membar} instructions to flush the I/O load and stores where
6543 appropriate, as described in Fujitsu's manual described above.
6547 @item unsigned char __builtin_read8 (void *@var{data})
6548 @item unsigned short __builtin_read16 (void *@var{data})
6549 @item unsigned long __builtin_read32 (void *@var{data})
6550 @item unsigned long long __builtin_read64 (void *@var{data})
6552 @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum})
6553 @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum})
6554 @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum})
6555 @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum})
6558 @node Other Built-in Functions
6559 @subsubsection Other Built-in Functions
6561 This section describes built-in functions that are not named after
6562 a specific FR-V instruction.
6565 @item sw2 __IACCreadll (iacc @var{reg})
6566 Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved
6567 for future expansion and must be 0.
6569 @item sw1 __IACCreadl (iacc @var{reg})
6570 Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1.
6571 Other values of @var{reg} are rejected as invalid.
6573 @item void __IACCsetll (iacc @var{reg}, sw2 @var{x})
6574 Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument
6575 is reserved for future expansion and must be 0.
6577 @item void __IACCsetl (iacc @var{reg}, sw1 @var{x})
6578 Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg}
6579 is 1. Other values of @var{reg} are rejected as invalid.
6581 @item void __data_prefetch0 (const void *@var{x})
6582 Use the @code{dcpl} instruction to load the contents of address @var{x}
6583 into the data cache.
6585 @item void __data_prefetch (const void *@var{x})
6586 Use the @code{nldub} instruction to load the contents of address @var{x}
6587 into the data cache. The instruction will be issued in slot I1@.
6590 @node X86 Built-in Functions
6591 @subsection X86 Built-in Functions
6593 These built-in functions are available for the i386 and x86-64 family
6594 of computers, depending on the command-line switches used.
6596 Note that, if you specify command-line switches such as @option{-msse},
6597 the compiler could use the extended instruction sets even if the built-ins
6598 are not used explicitly in the program. For this reason, applications
6599 which perform runtime CPU detection must compile separate files for each
6600 supported architecture, using the appropriate flags. In particular,
6601 the file containing the CPU detection code should be compiled without
6604 The following machine modes are available for use with MMX built-in functions
6605 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers,
6606 @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a
6607 vector of eight 8-bit integers. Some of the built-in functions operate on
6608 MMX registers as a whole 64-bit entity, these use @code{DI} as their mode.
6610 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
6611 of two 32-bit floating point values.
6613 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit
6614 floating point values. Some instructions use a vector of four 32-bit
6615 integers, these use @code{V4SI}. Finally, some instructions operate on an
6616 entire vector register, interpreting it as a 128-bit integer, these use mode
6619 The following built-in functions are made available by @option{-mmmx}.
6620 All of them generate the machine instruction that is part of the name.
6623 v8qi __builtin_ia32_paddb (v8qi, v8qi)
6624 v4hi __builtin_ia32_paddw (v4hi, v4hi)
6625 v2si __builtin_ia32_paddd (v2si, v2si)
6626 v8qi __builtin_ia32_psubb (v8qi, v8qi)
6627 v4hi __builtin_ia32_psubw (v4hi, v4hi)
6628 v2si __builtin_ia32_psubd (v2si, v2si)
6629 v8qi __builtin_ia32_paddsb (v8qi, v8qi)
6630 v4hi __builtin_ia32_paddsw (v4hi, v4hi)
6631 v8qi __builtin_ia32_psubsb (v8qi, v8qi)
6632 v4hi __builtin_ia32_psubsw (v4hi, v4hi)
6633 v8qi __builtin_ia32_paddusb (v8qi, v8qi)
6634 v4hi __builtin_ia32_paddusw (v4hi, v4hi)
6635 v8qi __builtin_ia32_psubusb (v8qi, v8qi)
6636 v4hi __builtin_ia32_psubusw (v4hi, v4hi)
6637 v4hi __builtin_ia32_pmullw (v4hi, v4hi)
6638 v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
6639 di __builtin_ia32_pand (di, di)
6640 di __builtin_ia32_pandn (di,di)
6641 di __builtin_ia32_por (di, di)
6642 di __builtin_ia32_pxor (di, di)
6643 v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
6644 v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
6645 v2si __builtin_ia32_pcmpeqd (v2si, v2si)
6646 v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
6647 v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
6648 v2si __builtin_ia32_pcmpgtd (v2si, v2si)
6649 v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
6650 v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
6651 v2si __builtin_ia32_punpckhdq (v2si, v2si)
6652 v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
6653 v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
6654 v2si __builtin_ia32_punpckldq (v2si, v2si)
6655 v8qi __builtin_ia32_packsswb (v4hi, v4hi)
6656 v4hi __builtin_ia32_packssdw (v2si, v2si)
6657 v8qi __builtin_ia32_packuswb (v4hi, v4hi)
6660 The following built-in functions are made available either with
6661 @option{-msse}, or with a combination of @option{-m3dnow} and
6662 @option{-march=athlon}. All of them generate the machine
6663 instruction that is part of the name.
6666 v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
6667 v8qi __builtin_ia32_pavgb (v8qi, v8qi)
6668 v4hi __builtin_ia32_pavgw (v4hi, v4hi)
6669 v4hi __builtin_ia32_psadbw (v8qi, v8qi)
6670 v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
6671 v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
6672 v8qi __builtin_ia32_pminub (v8qi, v8qi)
6673 v4hi __builtin_ia32_pminsw (v4hi, v4hi)
6674 int __builtin_ia32_pextrw (v4hi, int)
6675 v4hi __builtin_ia32_pinsrw (v4hi, int, int)
6676 int __builtin_ia32_pmovmskb (v8qi)
6677 void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
6678 void __builtin_ia32_movntq (di *, di)
6679 void __builtin_ia32_sfence (void)
6682 The following built-in functions are available when @option{-msse} is used.
6683 All of them generate the machine instruction that is part of the name.
6686 int __builtin_ia32_comieq (v4sf, v4sf)
6687 int __builtin_ia32_comineq (v4sf, v4sf)
6688 int __builtin_ia32_comilt (v4sf, v4sf)
6689 int __builtin_ia32_comile (v4sf, v4sf)
6690 int __builtin_ia32_comigt (v4sf, v4sf)
6691 int __builtin_ia32_comige (v4sf, v4sf)
6692 int __builtin_ia32_ucomieq (v4sf, v4sf)
6693 int __builtin_ia32_ucomineq (v4sf, v4sf)
6694 int __builtin_ia32_ucomilt (v4sf, v4sf)
6695 int __builtin_ia32_ucomile (v4sf, v4sf)
6696 int __builtin_ia32_ucomigt (v4sf, v4sf)
6697 int __builtin_ia32_ucomige (v4sf, v4sf)
6698 v4sf __builtin_ia32_addps (v4sf, v4sf)
6699 v4sf __builtin_ia32_subps (v4sf, v4sf)
6700 v4sf __builtin_ia32_mulps (v4sf, v4sf)
6701 v4sf __builtin_ia32_divps (v4sf, v4sf)
6702 v4sf __builtin_ia32_addss (v4sf, v4sf)
6703 v4sf __builtin_ia32_subss (v4sf, v4sf)
6704 v4sf __builtin_ia32_mulss (v4sf, v4sf)
6705 v4sf __builtin_ia32_divss (v4sf, v4sf)
6706 v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
6707 v4si __builtin_ia32_cmpltps (v4sf, v4sf)
6708 v4si __builtin_ia32_cmpleps (v4sf, v4sf)
6709 v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
6710 v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
6711 v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
6712 v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
6713 v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
6714 v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
6715 v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
6716 v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
6717 v4si __builtin_ia32_cmpordps (v4sf, v4sf)
6718 v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
6719 v4si __builtin_ia32_cmpltss (v4sf, v4sf)
6720 v4si __builtin_ia32_cmpless (v4sf, v4sf)
6721 v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
6722 v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
6723 v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
6724 v4si __builtin_ia32_cmpnless (v4sf, v4sf)
6725 v4si __builtin_ia32_cmpordss (v4sf, v4sf)
6726 v4sf __builtin_ia32_maxps (v4sf, v4sf)
6727 v4sf __builtin_ia32_maxss (v4sf, v4sf)
6728 v4sf __builtin_ia32_minps (v4sf, v4sf)
6729 v4sf __builtin_ia32_minss (v4sf, v4sf)
6730 v4sf __builtin_ia32_andps (v4sf, v4sf)
6731 v4sf __builtin_ia32_andnps (v4sf, v4sf)
6732 v4sf __builtin_ia32_orps (v4sf, v4sf)
6733 v4sf __builtin_ia32_xorps (v4sf, v4sf)
6734 v4sf __builtin_ia32_movss (v4sf, v4sf)
6735 v4sf __builtin_ia32_movhlps (v4sf, v4sf)
6736 v4sf __builtin_ia32_movlhps (v4sf, v4sf)
6737 v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
6738 v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
6739 v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
6740 v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
6741 v2si __builtin_ia32_cvtps2pi (v4sf)
6742 int __builtin_ia32_cvtss2si (v4sf)
6743 v2si __builtin_ia32_cvttps2pi (v4sf)
6744 int __builtin_ia32_cvttss2si (v4sf)
6745 v4sf __builtin_ia32_rcpps (v4sf)
6746 v4sf __builtin_ia32_rsqrtps (v4sf)
6747 v4sf __builtin_ia32_sqrtps (v4sf)
6748 v4sf __builtin_ia32_rcpss (v4sf)
6749 v4sf __builtin_ia32_rsqrtss (v4sf)
6750 v4sf __builtin_ia32_sqrtss (v4sf)
6751 v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
6752 void __builtin_ia32_movntps (float *, v4sf)
6753 int __builtin_ia32_movmskps (v4sf)
6756 The following built-in functions are available when @option{-msse} is used.
6759 @item v4sf __builtin_ia32_loadaps (float *)
6760 Generates the @code{movaps} machine instruction as a load from memory.
6761 @item void __builtin_ia32_storeaps (float *, v4sf)
6762 Generates the @code{movaps} machine instruction as a store to memory.
6763 @item v4sf __builtin_ia32_loadups (float *)
6764 Generates the @code{movups} machine instruction as a load from memory.
6765 @item void __builtin_ia32_storeups (float *, v4sf)
6766 Generates the @code{movups} machine instruction as a store to memory.
6767 @item v4sf __builtin_ia32_loadsss (float *)
6768 Generates the @code{movss} machine instruction as a load from memory.
6769 @item void __builtin_ia32_storess (float *, v4sf)
6770 Generates the @code{movss} machine instruction as a store to memory.
6771 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
6772 Generates the @code{movhps} machine instruction as a load from memory.
6773 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
6774 Generates the @code{movlps} machine instruction as a load from memory
6775 @item void __builtin_ia32_storehps (v4sf, v2si *)
6776 Generates the @code{movhps} machine instruction as a store to memory.
6777 @item void __builtin_ia32_storelps (v4sf, v2si *)
6778 Generates the @code{movlps} machine instruction as a store to memory.
6781 The following built-in functions are available when @option{-msse3} is used.
6782 All of them generate the machine instruction that is part of the name.
6785 v2df __builtin_ia32_addsubpd (v2df, v2df)
6786 v2df __builtin_ia32_addsubps (v2df, v2df)
6787 v2df __builtin_ia32_haddpd (v2df, v2df)
6788 v2df __builtin_ia32_haddps (v2df, v2df)
6789 v2df __builtin_ia32_hsubpd (v2df, v2df)
6790 v2df __builtin_ia32_hsubps (v2df, v2df)
6791 v16qi __builtin_ia32_lddqu (char const *)
6792 void __builtin_ia32_monitor (void *, unsigned int, unsigned int)
6793 v2df __builtin_ia32_movddup (v2df)
6794 v4sf __builtin_ia32_movshdup (v4sf)
6795 v4sf __builtin_ia32_movsldup (v4sf)
6796 void __builtin_ia32_mwait (unsigned int, unsigned int)
6799 The following built-in functions are available when @option{-msse3} is used.
6802 @item v2df __builtin_ia32_loadddup (double const *)
6803 Generates the @code{movddup} machine instruction as a load from memory.
6806 The following built-in functions are available when @option{-m3dnow} is used.
6807 All of them generate the machine instruction that is part of the name.
6810 void __builtin_ia32_femms (void)
6811 v8qi __builtin_ia32_pavgusb (v8qi, v8qi)
6812 v2si __builtin_ia32_pf2id (v2sf)
6813 v2sf __builtin_ia32_pfacc (v2sf, v2sf)
6814 v2sf __builtin_ia32_pfadd (v2sf, v2sf)
6815 v2si __builtin_ia32_pfcmpeq (v2sf, v2sf)
6816 v2si __builtin_ia32_pfcmpge (v2sf, v2sf)
6817 v2si __builtin_ia32_pfcmpgt (v2sf, v2sf)
6818 v2sf __builtin_ia32_pfmax (v2sf, v2sf)
6819 v2sf __builtin_ia32_pfmin (v2sf, v2sf)
6820 v2sf __builtin_ia32_pfmul (v2sf, v2sf)
6821 v2sf __builtin_ia32_pfrcp (v2sf)
6822 v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf)
6823 v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf)
6824 v2sf __builtin_ia32_pfrsqrt (v2sf)
6825 v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf)
6826 v2sf __builtin_ia32_pfsub (v2sf, v2sf)
6827 v2sf __builtin_ia32_pfsubr (v2sf, v2sf)
6828 v2sf __builtin_ia32_pi2fd (v2si)
6829 v4hi __builtin_ia32_pmulhrw (v4hi, v4hi)
6832 The following built-in functions are available when both @option{-m3dnow}
6833 and @option{-march=athlon} are used. All of them generate the machine
6834 instruction that is part of the name.
6837 v2si __builtin_ia32_pf2iw (v2sf)
6838 v2sf __builtin_ia32_pfnacc (v2sf, v2sf)
6839 v2sf __builtin_ia32_pfpnacc (v2sf, v2sf)
6840 v2sf __builtin_ia32_pi2fw (v2si)
6841 v2sf __builtin_ia32_pswapdsf (v2sf)
6842 v2si __builtin_ia32_pswapdsi (v2si)
6845 The following built-in functions are available when @option{-msse2}
6846 is used. All of them generate calls to an SSE2 ABI IEEE754 math intrinsic
6847 that is part of the name. Rather than using these directly you may
6848 want them automatically substituted for calls to the regular intrinsics
6849 using the @option{-msselibm}.
6852 double __builtin_sse2_acos (double)
6853 float __builtin_sse2_acosf (float)
6854 double __builtin_sse2_asin (double)
6855 float __builtin_sse2_asinf (float)
6856 double __builtin_sse2_atan (double)
6857 float __builtin_sse2_atanf (float)
6858 double __builtin_sse2_atan2 (double, double)
6859 float __builtin_sse2_atan2f (float, float)
6860 double __builtin_sse2_cos (double)
6861 float __builtin_sse2_cosf (float)
6862 double __builtin_sse2_exp (double)
6863 float __builtin_sse2_expf (float)
6864 double __builtin_sse2_log10 (double)
6865 float __builtin_sse2_log10f (float)
6866 double __builtin_sse2_log (double)
6867 float __builtin_sse2_logf (float)
6868 double __builtin_sse2_sin (double)
6869 float __builtin_sse2_sinf (float)
6870 double __builtin_sse2_tan (double)
6871 float __builtin_sse2_tanf (float)
6874 @node MIPS DSP Built-in Functions
6875 @subsection MIPS DSP Built-in Functions
6877 The MIPS DSP Application-Specific Extension (ASE) includes new
6878 instructions that are designed to improve the performance of DSP and
6879 media applications. It provides instructions that operate on packed
6880 8-bit integer data, Q15 fractional data and Q31 fractional data.
6882 GCC supports MIPS DSP operations using both the generic
6883 vector extensions (@pxref{Vector Extensions}) and a collection of
6884 MIPS-specific built-in functions. Both kinds of support are
6885 enabled by the @option{-mdsp} command-line option.
6887 At present, GCC only provides support for operations on 32-bit
6888 vectors. The vector type associated with 8-bit integer data is
6889 usually called @code{v4i8} and the vector type associated with Q15 is
6890 usually called @code{v2q15}. They can be defined in C as follows:
6893 typedef char v4i8 __attribute__ ((vector_size(4)));
6894 typedef short v2q15 __attribute__ ((vector_size(4)));
6897 @code{v4i8} and @code{v2q15} values are initialized in the same way as
6898 aggregates. For example:
6901 v4i8 a = @{1, 2, 3, 4@};
6903 b = (v4i8) @{5, 6, 7, 8@};
6905 v2q15 c = @{0x0fcb, 0x3a75@};
6907 d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@};
6910 @emph{Note:} The CPU's endianness determines the order in which values
6911 are packed. On little-endian targets, the first value is the least
6912 significant and the last value is the most significant. The opposite
6913 order applies to big-endian targets. For example, the code above will
6914 set the lowest byte of @code{a} to @code{1} on little-endian targets
6915 and @code{4} on big-endian targets.
6917 @emph{Note:} Q15 and Q31 values must be initialized with their integer
6918 representation. As shown in this example, the integer representation
6919 of a Q15 value can be obtained by multiplying the fractional value by
6920 @code{0x1.0p15}. The equivalent for Q31 values is to multiply by
6923 The table below lists the @code{v4i8} and @code{v2q15} operations for which
6924 hardware support exists. @code{a} and @code{b} are @code{v4i8} values,
6925 and @code{c} and @code{d} are @code{v2q15} values.
6927 @multitable @columnfractions .50 .50
6928 @item C code @tab MIPS instruction
6929 @item @code{a + b} @tab @code{addu.qb}
6930 @item @code{c + d} @tab @code{addq.ph}
6931 @item @code{a - b} @tab @code{subu.qb}
6932 @item @code{c - d} @tab @code{subq.ph}
6935 It is easier to describe the DSP built-in functions if we first define
6936 the following types:
6941 typedef long long a64;
6944 @code{q31} and @code{i32} are actually the same as @code{int}, but we
6945 use @code{q31} to indicate a Q31 fractional value and @code{i32} to
6946 indicate a 32-bit integer value. Similarly, @code{a64} is the same as
6947 @code{long long}, but we use @code{a64} to indicate values that will
6948 be placed in one of the four DSP accumulators (@code{$ac0},
6949 @code{$ac1}, @code{$ac2} or @code{$ac3}).
6951 Also, some built-in functions prefer or require immediate numbers as
6952 parameters, because the corresponding DSP instructions accept both immediate
6953 numbers and register operands, or accept immediate numbers only. The
6954 immediate parameters are listed as follows.
6962 imm_n32_31: -32 to 31.
6963 imm_n512_511: -512 to 511.
6966 The following built-in functions map directly to a particular MIPS DSP
6967 instruction. Please refer to the architecture specification
6968 for details on what each instruction does.
6971 v2q15 __builtin_mips_addq_ph (v2q15, v2q15)
6972 v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15)
6973 q31 __builtin_mips_addq_s_w (q31, q31)
6974 v4i8 __builtin_mips_addu_qb (v4i8, v4i8)
6975 v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8)
6976 v2q15 __builtin_mips_subq_ph (v2q15, v2q15)
6977 v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15)
6978 q31 __builtin_mips_subq_s_w (q31, q31)
6979 v4i8 __builtin_mips_subu_qb (v4i8, v4i8)
6980 v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8)
6981 i32 __builtin_mips_addsc (i32, i32)
6982 i32 __builtin_mips_addwc (i32, i32)
6983 i32 __builtin_mips_modsub (i32, i32)
6984 i32 __builtin_mips_raddu_w_qb (v4i8)
6985 v2q15 __builtin_mips_absq_s_ph (v2q15)
6986 q31 __builtin_mips_absq_s_w (q31)
6987 v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15)
6988 v2q15 __builtin_mips_precrq_ph_w (q31, q31)
6989 v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31)
6990 v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15)
6991 q31 __builtin_mips_preceq_w_phl (v2q15)
6992 q31 __builtin_mips_preceq_w_phr (v2q15)
6993 v2q15 __builtin_mips_precequ_ph_qbl (v4i8)
6994 v2q15 __builtin_mips_precequ_ph_qbr (v4i8)
6995 v2q15 __builtin_mips_precequ_ph_qbla (v4i8)
6996 v2q15 __builtin_mips_precequ_ph_qbra (v4i8)
6997 v2q15 __builtin_mips_preceu_ph_qbl (v4i8)
6998 v2q15 __builtin_mips_preceu_ph_qbr (v4i8)
6999 v2q15 __builtin_mips_preceu_ph_qbla (v4i8)
7000 v2q15 __builtin_mips_preceu_ph_qbra (v4i8)
7001 v4i8 __builtin_mips_shll_qb (v4i8, imm0_7)
7002 v4i8 __builtin_mips_shll_qb (v4i8, i32)
7003 v2q15 __builtin_mips_shll_ph (v2q15, imm0_15)
7004 v2q15 __builtin_mips_shll_ph (v2q15, i32)
7005 v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15)
7006 v2q15 __builtin_mips_shll_s_ph (v2q15, i32)
7007 q31 __builtin_mips_shll_s_w (q31, imm0_31)
7008 q31 __builtin_mips_shll_s_w (q31, i32)
7009 v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7)
7010 v4i8 __builtin_mips_shrl_qb (v4i8, i32)
7011 v2q15 __builtin_mips_shra_ph (v2q15, imm0_15)
7012 v2q15 __builtin_mips_shra_ph (v2q15, i32)
7013 v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15)
7014 v2q15 __builtin_mips_shra_r_ph (v2q15, i32)
7015 q31 __builtin_mips_shra_r_w (q31, imm0_31)
7016 q31 __builtin_mips_shra_r_w (q31, i32)
7017 v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15)
7018 v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15)
7019 v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15)
7020 q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15)
7021 q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15)
7022 a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8)
7023 a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8)
7024 a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8)
7025 a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8)
7026 a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15)
7027 a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31)
7028 a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15)
7029 a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31)
7030 a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15)
7031 a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15)
7032 a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15)
7033 a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15)
7034 a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15)
7035 i32 __builtin_mips_bitrev (i32)
7036 i32 __builtin_mips_insv (i32, i32)
7037 v4i8 __builtin_mips_repl_qb (imm0_255)
7038 v4i8 __builtin_mips_repl_qb (i32)
7039 v2q15 __builtin_mips_repl_ph (imm_n512_511)
7040 v2q15 __builtin_mips_repl_ph (i32)
7041 void __builtin_mips_cmpu_eq_qb (v4i8, v4i8)
7042 void __builtin_mips_cmpu_lt_qb (v4i8, v4i8)
7043 void __builtin_mips_cmpu_le_qb (v4i8, v4i8)
7044 i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8)
7045 i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8)
7046 i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8)
7047 void __builtin_mips_cmp_eq_ph (v2q15, v2q15)
7048 void __builtin_mips_cmp_lt_ph (v2q15, v2q15)
7049 void __builtin_mips_cmp_le_ph (v2q15, v2q15)
7050 v4i8 __builtin_mips_pick_qb (v4i8, v4i8)
7051 v2q15 __builtin_mips_pick_ph (v2q15, v2q15)
7052 v2q15 __builtin_mips_packrl_ph (v2q15, v2q15)
7053 i32 __builtin_mips_extr_w (a64, imm0_31)
7054 i32 __builtin_mips_extr_w (a64, i32)
7055 i32 __builtin_mips_extr_r_w (a64, imm0_31)
7056 i32 __builtin_mips_extr_s_h (a64, i32)
7057 i32 __builtin_mips_extr_rs_w (a64, imm0_31)
7058 i32 __builtin_mips_extr_rs_w (a64, i32)
7059 i32 __builtin_mips_extr_s_h (a64, imm0_31)
7060 i32 __builtin_mips_extr_r_w (a64, i32)
7061 i32 __builtin_mips_extp (a64, imm0_31)
7062 i32 __builtin_mips_extp (a64, i32)
7063 i32 __builtin_mips_extpdp (a64, imm0_31)
7064 i32 __builtin_mips_extpdp (a64, i32)
7065 a64 __builtin_mips_shilo (a64, imm_n32_31)
7066 a64 __builtin_mips_shilo (a64, i32)
7067 a64 __builtin_mips_mthlip (a64, i32)
7068 void __builtin_mips_wrdsp (i32, imm0_63)
7069 i32 __builtin_mips_rddsp (imm0_63)
7070 i32 __builtin_mips_lbux (void *, i32)
7071 i32 __builtin_mips_lhx (void *, i32)
7072 i32 __builtin_mips_lwx (void *, i32)
7073 i32 __builtin_mips_bposge32 (void)
7076 @node MIPS Paired-Single Support
7077 @subsection MIPS Paired-Single Support
7079 The MIPS64 architecture includes a number of instructions that
7080 operate on pairs of single-precision floating-point values.
7081 Each pair is packed into a 64-bit floating-point register,
7082 with one element being designated the ``upper half'' and
7083 the other being designated the ``lower half''.
7085 GCC supports paired-single operations using both the generic
7086 vector extensions (@pxref{Vector Extensions}) and a collection of
7087 MIPS-specific built-in functions. Both kinds of support are
7088 enabled by the @option{-mpaired-single} command-line option.
7090 The vector type associated with paired-single values is usually
7091 called @code{v2sf}. It can be defined in C as follows:
7094 typedef float v2sf __attribute__ ((vector_size (8)));
7097 @code{v2sf} values are initialized in the same way as aggregates.
7101 v2sf a = @{1.5, 9.1@};
7104 b = (v2sf) @{e, f@};
7107 @emph{Note:} The CPU's endianness determines which value is stored in
7108 the upper half of a register and which value is stored in the lower half.
7109 On little-endian targets, the first value is the lower one and the second
7110 value is the upper one. The opposite order applies to big-endian targets.
7111 For example, the code above will set the lower half of @code{a} to
7112 @code{1.5} on little-endian targets and @code{9.1} on big-endian targets.
7115 * Paired-Single Arithmetic::
7116 * Paired-Single Built-in Functions::
7117 * MIPS-3D Built-in Functions::
7120 @node Paired-Single Arithmetic
7121 @subsubsection Paired-Single Arithmetic
7123 The table below lists the @code{v2sf} operations for which hardware
7124 support exists. @code{a}, @code{b} and @code{c} are @code{v2sf}
7125 values and @code{x} is an integral value.
7127 @multitable @columnfractions .50 .50
7128 @item C code @tab MIPS instruction
7129 @item @code{a + b} @tab @code{add.ps}
7130 @item @code{a - b} @tab @code{sub.ps}
7131 @item @code{-a} @tab @code{neg.ps}
7132 @item @code{a * b} @tab @code{mul.ps}
7133 @item @code{a * b + c} @tab @code{madd.ps}
7134 @item @code{a * b - c} @tab @code{msub.ps}
7135 @item @code{-(a * b + c)} @tab @code{nmadd.ps}
7136 @item @code{-(a * b - c)} @tab @code{nmsub.ps}
7137 @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps}
7140 Note that the multiply-accumulate instructions can be disabled
7141 using the command-line option @code{-mno-fused-madd}.
7143 @node Paired-Single Built-in Functions
7144 @subsubsection Paired-Single Built-in Functions
7146 The following paired-single functions map directly to a particular
7147 MIPS instruction. Please refer to the architecture specification
7148 for details on what each instruction does.
7151 @item v2sf __builtin_mips_pll_ps (v2sf, v2sf)
7152 Pair lower lower (@code{pll.ps}).
7154 @item v2sf __builtin_mips_pul_ps (v2sf, v2sf)
7155 Pair upper lower (@code{pul.ps}).
7157 @item v2sf __builtin_mips_plu_ps (v2sf, v2sf)
7158 Pair lower upper (@code{plu.ps}).
7160 @item v2sf __builtin_mips_puu_ps (v2sf, v2sf)
7161 Pair upper upper (@code{puu.ps}).
7163 @item v2sf __builtin_mips_cvt_ps_s (float, float)
7164 Convert pair to paired single (@code{cvt.ps.s}).
7166 @item float __builtin_mips_cvt_s_pl (v2sf)
7167 Convert pair lower to single (@code{cvt.s.pl}).
7169 @item float __builtin_mips_cvt_s_pu (v2sf)
7170 Convert pair upper to single (@code{cvt.s.pu}).
7172 @item v2sf __builtin_mips_abs_ps (v2sf)
7173 Absolute value (@code{abs.ps}).
7175 @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int)
7176 Align variable (@code{alnv.ps}).
7178 @emph{Note:} The value of the third parameter must be 0 or 4
7179 modulo 8, otherwise the result will be unpredictable. Please read the
7180 instruction description for details.
7183 The following multi-instruction functions are also available.
7184 In each case, @var{cond} can be any of the 16 floating-point conditions:
7185 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
7186 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl},
7187 @code{lt}, @code{nge}, @code{le} or @code{ngt}.
7190 @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7191 @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7192 Conditional move based on floating point comparison (@code{c.@var{cond}.ps},
7193 @code{movt.ps}/@code{movf.ps}).
7195 The @code{movt} functions return the value @var{x} computed by:
7198 c.@var{cond}.ps @var{cc},@var{a},@var{b}
7199 mov.ps @var{x},@var{c}
7200 movt.ps @var{x},@var{d},@var{cc}
7203 The @code{movf} functions are similar but use @code{movf.ps} instead
7206 @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7207 @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7208 Comparison of two paired-single values (@code{c.@var{cond}.ps},
7209 @code{bc1t}/@code{bc1f}).
7211 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
7212 and return either the upper or lower half of the result. For example:
7216 if (__builtin_mips_upper_c_eq_ps (a, b))
7217 upper_halves_are_equal ();
7219 upper_halves_are_unequal ();
7221 if (__builtin_mips_lower_c_eq_ps (a, b))
7222 lower_halves_are_equal ();
7224 lower_halves_are_unequal ();
7228 @node MIPS-3D Built-in Functions
7229 @subsubsection MIPS-3D Built-in Functions
7231 The MIPS-3D Application-Specific Extension (ASE) includes additional
7232 paired-single instructions that are designed to improve the performance
7233 of 3D graphics operations. Support for these instructions is controlled
7234 by the @option{-mips3d} command-line option.
7236 The functions listed below map directly to a particular MIPS-3D
7237 instruction. Please refer to the architecture specification for
7238 more details on what each instruction does.
7241 @item v2sf __builtin_mips_addr_ps (v2sf, v2sf)
7242 Reduction add (@code{addr.ps}).
7244 @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf)
7245 Reduction multiply (@code{mulr.ps}).
7247 @item v2sf __builtin_mips_cvt_pw_ps (v2sf)
7248 Convert paired single to paired word (@code{cvt.pw.ps}).
7250 @item v2sf __builtin_mips_cvt_ps_pw (v2sf)
7251 Convert paired word to paired single (@code{cvt.ps.pw}).
7253 @item float __builtin_mips_recip1_s (float)
7254 @itemx double __builtin_mips_recip1_d (double)
7255 @itemx v2sf __builtin_mips_recip1_ps (v2sf)
7256 Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}).
7258 @item float __builtin_mips_recip2_s (float, float)
7259 @itemx double __builtin_mips_recip2_d (double, double)
7260 @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf)
7261 Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}).
7263 @item float __builtin_mips_rsqrt1_s (float)
7264 @itemx double __builtin_mips_rsqrt1_d (double)
7265 @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf)
7266 Reduced precision reciprocal square root (sequence step 1)
7267 (@code{rsqrt1.@var{fmt}}).
7269 @item float __builtin_mips_rsqrt2_s (float, float)
7270 @itemx double __builtin_mips_rsqrt2_d (double, double)
7271 @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf)
7272 Reduced precision reciprocal square root (sequence step 2)
7273 (@code{rsqrt2.@var{fmt}}).
7276 The following multi-instruction functions are also available.
7277 In each case, @var{cond} can be any of the 16 floating-point conditions:
7278 @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult},
7279 @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq},
7280 @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}.
7283 @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b})
7284 @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b})
7285 Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}},
7286 @code{bc1t}/@code{bc1f}).
7288 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s}
7289 or @code{cabs.@var{cond}.d} and return the result as a boolean value.
7294 if (__builtin_mips_cabs_eq_s (a, b))
7300 @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7301 @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7302 Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps},
7303 @code{bc1t}/@code{bc1f}).
7305 These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps}
7306 and return either the upper or lower half of the result. For example:
7310 if (__builtin_mips_upper_cabs_eq_ps (a, b))
7311 upper_halves_are_equal ();
7313 upper_halves_are_unequal ();
7315 if (__builtin_mips_lower_cabs_eq_ps (a, b))
7316 lower_halves_are_equal ();
7318 lower_halves_are_unequal ();
7321 @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7322 @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7323 Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps},
7324 @code{movt.ps}/@code{movf.ps}).
7326 The @code{movt} functions return the value @var{x} computed by:
7329 cabs.@var{cond}.ps @var{cc},@var{a},@var{b}
7330 mov.ps @var{x},@var{c}
7331 movt.ps @var{x},@var{d},@var{cc}
7334 The @code{movf} functions are similar but use @code{movf.ps} instead
7337 @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7338 @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7339 @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7340 @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b})
7341 Comparison of two paired-single values
7342 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
7343 @code{bc1any2t}/@code{bc1any2f}).
7345 These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps}
7346 or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either
7347 result is true and the @code{all} forms return true if both results are true.
7352 if (__builtin_mips_any_c_eq_ps (a, b))
7357 if (__builtin_mips_all_c_eq_ps (a, b))
7363 @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7364 @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7365 @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7366 @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d})
7367 Comparison of four paired-single values
7368 (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps},
7369 @code{bc1any4t}/@code{bc1any4f}).
7371 These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps}
7372 to compare @var{a} with @var{b} and to compare @var{c} with @var{d}.
7373 The @code{any} forms return true if any of the four results are true
7374 and the @code{all} forms return true if all four results are true.
7379 if (__builtin_mips_any_c_eq_4s (a, b, c, d))
7384 if (__builtin_mips_all_c_eq_4s (a, b, c, d))
7391 @node PowerPC AltiVec Built-in Functions
7392 @subsection PowerPC AltiVec Built-in Functions
7394 GCC provides an interface for the PowerPC family of processors to access
7395 the AltiVec operations described in Motorola's AltiVec Programming
7396 Interface Manual. The interface is made available by including
7397 @code{<altivec.h>} and using @option{-maltivec} and
7398 @option{-mabi=altivec}. The interface supports the following vector
7402 vector unsigned char
7406 vector unsigned short
7417 GCC's implementation of the high-level language interface available from
7418 C and C++ code differs from Motorola's documentation in several ways.
7423 A vector constant is a list of constant expressions within curly braces.
7426 A vector initializer requires no cast if the vector constant is of the
7427 same type as the variable it is initializing.
7430 If @code{signed} or @code{unsigned} is omitted, the signedness of the
7431 vector type is the default signedness of the base type. The default
7432 varies depending on the operating system, so a portable program should
7433 always specify the signedness.
7436 Compiling with @option{-maltivec} adds keywords @code{__vector},
7437 @code{__pixel}, and @code{__bool}. Macros @option{vector},
7438 @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can
7442 GCC allows using a @code{typedef} name as the type specifier for a
7446 For C, overloaded functions are implemented with macros so the following
7450 vec_add ((vector signed int)@{1, 2, 3, 4@}, foo);
7453 Since @code{vec_add} is a macro, the vector constant in the example
7454 is treated as four separate arguments. Wrap the entire argument in
7455 parentheses for this to work.
7458 @emph{Note:} Only the @code{<altivec.h>} interface is supported.
7459 Internally, GCC uses built-in functions to achieve the functionality in
7460 the aforementioned header file, but they are not supported and are
7461 subject to change without notice.
7463 The following interfaces are supported for the generic and specific
7464 AltiVec operations and the AltiVec predicates. In cases where there
7465 is a direct mapping between generic and specific operations, only the
7466 generic names are shown here, although the specific operations can also
7469 Arguments that are documented as @code{const int} require literal
7470 integral values within the range required for that operation.
7473 vector signed char vec_abs (vector signed char);
7474 vector signed short vec_abs (vector signed short);
7475 vector signed int vec_abs (vector signed int);
7476 vector float vec_abs (vector float);
7478 vector signed char vec_abss (vector signed char);
7479 vector signed short vec_abss (vector signed short);
7480 vector signed int vec_abss (vector signed int);
7482 vector signed char vec_add (vector bool char, vector signed char);
7483 vector signed char vec_add (vector signed char, vector bool char);
7484 vector signed char vec_add (vector signed char, vector signed char);
7485 vector unsigned char vec_add (vector bool char, vector unsigned char);
7486 vector unsigned char vec_add (vector unsigned char, vector bool char);
7487 vector unsigned char vec_add (vector unsigned char,
7488 vector unsigned char);
7489 vector signed short vec_add (vector bool short, vector signed short);
7490 vector signed short vec_add (vector signed short, vector bool short);
7491 vector signed short vec_add (vector signed short, vector signed short);
7492 vector unsigned short vec_add (vector bool short,
7493 vector unsigned short);
7494 vector unsigned short vec_add (vector unsigned short,
7496 vector unsigned short vec_add (vector unsigned short,
7497 vector unsigned short);
7498 vector signed int vec_add (vector bool int, vector signed int);
7499 vector signed int vec_add (vector signed int, vector bool int);
7500 vector signed int vec_add (vector signed int, vector signed int);
7501 vector unsigned int vec_add (vector bool int, vector unsigned int);
7502 vector unsigned int vec_add (vector unsigned int, vector bool int);
7503 vector unsigned int vec_add (vector unsigned int, vector unsigned int);
7504 vector float vec_add (vector float, vector float);
7506 vector float vec_vaddfp (vector float, vector float);
7508 vector signed int vec_vadduwm (vector bool int, vector signed int);
7509 vector signed int vec_vadduwm (vector signed int, vector bool int);
7510 vector signed int vec_vadduwm (vector signed int, vector signed int);
7511 vector unsigned int vec_vadduwm (vector bool int, vector unsigned int);
7512 vector unsigned int vec_vadduwm (vector unsigned int, vector bool int);
7513 vector unsigned int vec_vadduwm (vector unsigned int,
7514 vector unsigned int);
7516 vector signed short vec_vadduhm (vector bool short,
7517 vector signed short);
7518 vector signed short vec_vadduhm (vector signed short,
7520 vector signed short vec_vadduhm (vector signed short,
7521 vector signed short);
7522 vector unsigned short vec_vadduhm (vector bool short,
7523 vector unsigned short);
7524 vector unsigned short vec_vadduhm (vector unsigned short,
7526 vector unsigned short vec_vadduhm (vector unsigned short,
7527 vector unsigned short);
7529 vector signed char vec_vaddubm (vector bool char, vector signed char);
7530 vector signed char vec_vaddubm (vector signed char, vector bool char);
7531 vector signed char vec_vaddubm (vector signed char, vector signed char);
7532 vector unsigned char vec_vaddubm (vector bool char,
7533 vector unsigned char);
7534 vector unsigned char vec_vaddubm (vector unsigned char,
7536 vector unsigned char vec_vaddubm (vector unsigned char,
7537 vector unsigned char);
7539 vector unsigned int vec_addc (vector unsigned int, vector unsigned int);
7541 vector unsigned char vec_adds (vector bool char, vector unsigned char);
7542 vector unsigned char vec_adds (vector unsigned char, vector bool char);
7543 vector unsigned char vec_adds (vector unsigned char,
7544 vector unsigned char);
7545 vector signed char vec_adds (vector bool char, vector signed char);
7546 vector signed char vec_adds (vector signed char, vector bool char);
7547 vector signed char vec_adds (vector signed char, vector signed char);
7548 vector unsigned short vec_adds (vector bool short,
7549 vector unsigned short);
7550 vector unsigned short vec_adds (vector unsigned short,
7552 vector unsigned short vec_adds (vector unsigned short,
7553 vector unsigned short);
7554 vector signed short vec_adds (vector bool short, vector signed short);
7555 vector signed short vec_adds (vector signed short, vector bool short);
7556 vector signed short vec_adds (vector signed short, vector signed short);
7557 vector unsigned int vec_adds (vector bool int, vector unsigned int);
7558 vector unsigned int vec_adds (vector unsigned int, vector bool int);
7559 vector unsigned int vec_adds (vector unsigned int, vector unsigned int);
7560 vector signed int vec_adds (vector bool int, vector signed int);
7561 vector signed int vec_adds (vector signed int, vector bool int);
7562 vector signed int vec_adds (vector signed int, vector signed int);
7564 vector signed int vec_vaddsws (vector bool int, vector signed int);
7565 vector signed int vec_vaddsws (vector signed int, vector bool int);
7566 vector signed int vec_vaddsws (vector signed int, vector signed int);
7568 vector unsigned int vec_vadduws (vector bool int, vector unsigned int);
7569 vector unsigned int vec_vadduws (vector unsigned int, vector bool int);
7570 vector unsigned int vec_vadduws (vector unsigned int,
7571 vector unsigned int);
7573 vector signed short vec_vaddshs (vector bool short,
7574 vector signed short);
7575 vector signed short vec_vaddshs (vector signed short,
7577 vector signed short vec_vaddshs (vector signed short,
7578 vector signed short);
7580 vector unsigned short vec_vadduhs (vector bool short,
7581 vector unsigned short);
7582 vector unsigned short vec_vadduhs (vector unsigned short,
7584 vector unsigned short vec_vadduhs (vector unsigned short,
7585 vector unsigned short);
7587 vector signed char vec_vaddsbs (vector bool char, vector signed char);
7588 vector signed char vec_vaddsbs (vector signed char, vector bool char);
7589 vector signed char vec_vaddsbs (vector signed char, vector signed char);
7591 vector unsigned char vec_vaddubs (vector bool char,
7592 vector unsigned char);
7593 vector unsigned char vec_vaddubs (vector unsigned char,
7595 vector unsigned char vec_vaddubs (vector unsigned char,
7596 vector unsigned char);
7598 vector float vec_and (vector float, vector float);
7599 vector float vec_and (vector float, vector bool int);
7600 vector float vec_and (vector bool int, vector float);
7601 vector bool int vec_and (vector bool int, vector bool int);
7602 vector signed int vec_and (vector bool int, vector signed int);
7603 vector signed int vec_and (vector signed int, vector bool int);
7604 vector signed int vec_and (vector signed int, vector signed int);
7605 vector unsigned int vec_and (vector bool int, vector unsigned int);
7606 vector unsigned int vec_and (vector unsigned int, vector bool int);
7607 vector unsigned int vec_and (vector unsigned int, vector unsigned int);
7608 vector bool short vec_and (vector bool short, vector bool short);
7609 vector signed short vec_and (vector bool short, vector signed short);
7610 vector signed short vec_and (vector signed short, vector bool short);
7611 vector signed short vec_and (vector signed short, vector signed short);
7612 vector unsigned short vec_and (vector bool short,
7613 vector unsigned short);
7614 vector unsigned short vec_and (vector unsigned short,
7616 vector unsigned short vec_and (vector unsigned short,
7617 vector unsigned short);
7618 vector signed char vec_and (vector bool char, vector signed char);
7619 vector bool char vec_and (vector bool char, vector bool char);
7620 vector signed char vec_and (vector signed char, vector bool char);
7621 vector signed char vec_and (vector signed char, vector signed char);
7622 vector unsigned char vec_and (vector bool char, vector unsigned char);
7623 vector unsigned char vec_and (vector unsigned char, vector bool char);
7624 vector unsigned char vec_and (vector unsigned char,
7625 vector unsigned char);
7627 vector float vec_andc (vector float, vector float);
7628 vector float vec_andc (vector float, vector bool int);
7629 vector float vec_andc (vector bool int, vector float);
7630 vector bool int vec_andc (vector bool int, vector bool int);
7631 vector signed int vec_andc (vector bool int, vector signed int);
7632 vector signed int vec_andc (vector signed int, vector bool int);
7633 vector signed int vec_andc (vector signed int, vector signed int);
7634 vector unsigned int vec_andc (vector bool int, vector unsigned int);
7635 vector unsigned int vec_andc (vector unsigned int, vector bool int);
7636 vector unsigned int vec_andc (vector unsigned int, vector unsigned int);
7637 vector bool short vec_andc (vector bool short, vector bool short);
7638 vector signed short vec_andc (vector bool short, vector signed short);
7639 vector signed short vec_andc (vector signed short, vector bool short);
7640 vector signed short vec_andc (vector signed short, vector signed short);
7641 vector unsigned short vec_andc (vector bool short,
7642 vector unsigned short);
7643 vector unsigned short vec_andc (vector unsigned short,
7645 vector unsigned short vec_andc (vector unsigned short,
7646 vector unsigned short);
7647 vector signed char vec_andc (vector bool char, vector signed char);
7648 vector bool char vec_andc (vector bool char, vector bool char);
7649 vector signed char vec_andc (vector signed char, vector bool char);
7650 vector signed char vec_andc (vector signed char, vector signed char);
7651 vector unsigned char vec_andc (vector bool char, vector unsigned char);
7652 vector unsigned char vec_andc (vector unsigned char, vector bool char);
7653 vector unsigned char vec_andc (vector unsigned char,
7654 vector unsigned char);
7656 vector unsigned char vec_avg (vector unsigned char,
7657 vector unsigned char);
7658 vector signed char vec_avg (vector signed char, vector signed char);
7659 vector unsigned short vec_avg (vector unsigned short,
7660 vector unsigned short);
7661 vector signed short vec_avg (vector signed short, vector signed short);
7662 vector unsigned int vec_avg (vector unsigned int, vector unsigned int);
7663 vector signed int vec_avg (vector signed int, vector signed int);
7665 vector signed int vec_vavgsw (vector signed int, vector signed int);
7667 vector unsigned int vec_vavguw (vector unsigned int,
7668 vector unsigned int);
7670 vector signed short vec_vavgsh (vector signed short,
7671 vector signed short);
7673 vector unsigned short vec_vavguh (vector unsigned short,
7674 vector unsigned short);
7676 vector signed char vec_vavgsb (vector signed char, vector signed char);
7678 vector unsigned char vec_vavgub (vector unsigned char,
7679 vector unsigned char);
7681 vector float vec_ceil (vector float);
7683 vector signed int vec_cmpb (vector float, vector float);
7685 vector bool char vec_cmpeq (vector signed char, vector signed char);
7686 vector bool char vec_cmpeq (vector unsigned char, vector unsigned char);
7687 vector bool short vec_cmpeq (vector signed short, vector signed short);
7688 vector bool short vec_cmpeq (vector unsigned short,
7689 vector unsigned short);
7690 vector bool int vec_cmpeq (vector signed int, vector signed int);
7691 vector bool int vec_cmpeq (vector unsigned int, vector unsigned int);
7692 vector bool int vec_cmpeq (vector float, vector float);
7694 vector bool int vec_vcmpeqfp (vector float, vector float);
7696 vector bool int vec_vcmpequw (vector signed int, vector signed int);
7697 vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int);
7699 vector bool short vec_vcmpequh (vector signed short,
7700 vector signed short);
7701 vector bool short vec_vcmpequh (vector unsigned short,
7702 vector unsigned short);
7704 vector bool char vec_vcmpequb (vector signed char, vector signed char);
7705 vector bool char vec_vcmpequb (vector unsigned char,
7706 vector unsigned char);
7708 vector bool int vec_cmpge (vector float, vector float);
7710 vector bool char vec_cmpgt (vector unsigned char, vector unsigned char);
7711 vector bool char vec_cmpgt (vector signed char, vector signed char);
7712 vector bool short vec_cmpgt (vector unsigned short,
7713 vector unsigned short);
7714 vector bool short vec_cmpgt (vector signed short, vector signed short);
7715 vector bool int vec_cmpgt (vector unsigned int, vector unsigned int);
7716 vector bool int vec_cmpgt (vector signed int, vector signed int);
7717 vector bool int vec_cmpgt (vector float, vector float);
7719 vector bool int vec_vcmpgtfp (vector float, vector float);
7721 vector bool int vec_vcmpgtsw (vector signed int, vector signed int);
7723 vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int);
7725 vector bool short vec_vcmpgtsh (vector signed short,
7726 vector signed short);
7728 vector bool short vec_vcmpgtuh (vector unsigned short,
7729 vector unsigned short);
7731 vector bool char vec_vcmpgtsb (vector signed char, vector signed char);
7733 vector bool char vec_vcmpgtub (vector unsigned char,
7734 vector unsigned char);
7736 vector bool int vec_cmple (vector float, vector float);
7738 vector bool char vec_cmplt (vector unsigned char, vector unsigned char);
7739 vector bool char vec_cmplt (vector signed char, vector signed char);
7740 vector bool short vec_cmplt (vector unsigned short,
7741 vector unsigned short);
7742 vector bool short vec_cmplt (vector signed short, vector signed short);
7743 vector bool int vec_cmplt (vector unsigned int, vector unsigned int);
7744 vector bool int vec_cmplt (vector signed int, vector signed int);
7745 vector bool int vec_cmplt (vector float, vector float);
7747 vector float vec_ctf (vector unsigned int, const int);
7748 vector float vec_ctf (vector signed int, const int);
7750 vector float vec_vcfsx (vector signed int, const int);
7752 vector float vec_vcfux (vector unsigned int, const int);
7754 vector signed int vec_cts (vector float, const int);
7756 vector unsigned int vec_ctu (vector float, const int);
7758 void vec_dss (const int);
7760 void vec_dssall (void);
7762 void vec_dst (const vector unsigned char *, int, const int);
7763 void vec_dst (const vector signed char *, int, const int);
7764 void vec_dst (const vector bool char *, int, const int);
7765 void vec_dst (const vector unsigned short *, int, const int);
7766 void vec_dst (const vector signed short *, int, const int);
7767 void vec_dst (const vector bool short *, int, const int);
7768 void vec_dst (const vector pixel *, int, const int);
7769 void vec_dst (const vector unsigned int *, int, const int);
7770 void vec_dst (const vector signed int *, int, const int);
7771 void vec_dst (const vector bool int *, int, const int);
7772 void vec_dst (const vector float *, int, const int);
7773 void vec_dst (const unsigned char *, int, const int);
7774 void vec_dst (const signed char *, int, const int);
7775 void vec_dst (const unsigned short *, int, const int);
7776 void vec_dst (const short *, int, const int);
7777 void vec_dst (const unsigned int *, int, const int);
7778 void vec_dst (const int *, int, const int);
7779 void vec_dst (const unsigned long *, int, const int);
7780 void vec_dst (const long *, int, const int);
7781 void vec_dst (const float *, int, const int);
7783 void vec_dstst (const vector unsigned char *, int, const int);
7784 void vec_dstst (const vector signed char *, int, const int);
7785 void vec_dstst (const vector bool char *, int, const int);
7786 void vec_dstst (const vector unsigned short *, int, const int);
7787 void vec_dstst (const vector signed short *, int, const int);
7788 void vec_dstst (const vector bool short *, int, const int);
7789 void vec_dstst (const vector pixel *, int, const int);
7790 void vec_dstst (const vector unsigned int *, int, const int);
7791 void vec_dstst (const vector signed int *, int, const int);
7792 void vec_dstst (const vector bool int *, int, const int);
7793 void vec_dstst (const vector float *, int, const int);
7794 void vec_dstst (const unsigned char *, int, const int);
7795 void vec_dstst (const signed char *, int, const int);
7796 void vec_dstst (const unsigned short *, int, const int);
7797 void vec_dstst (const short *, int, const int);
7798 void vec_dstst (const unsigned int *, int, const int);
7799 void vec_dstst (const int *, int, const int);
7800 void vec_dstst (const unsigned long *, int, const int);
7801 void vec_dstst (const long *, int, const int);
7802 void vec_dstst (const float *, int, const int);
7804 void vec_dststt (const vector unsigned char *, int, const int);
7805 void vec_dststt (const vector signed char *, int, const int);
7806 void vec_dststt (const vector bool char *, int, const int);
7807 void vec_dststt (const vector unsigned short *, int, const int);
7808 void vec_dststt (const vector signed short *, int, const int);
7809 void vec_dststt (const vector bool short *, int, const int);
7810 void vec_dststt (const vector pixel *, int, const int);
7811 void vec_dststt (const vector unsigned int *, int, const int);
7812 void vec_dststt (const vector signed int *, int, const int);
7813 void vec_dststt (const vector bool int *, int, const int);
7814 void vec_dststt (const vector float *, int, const int);
7815 void vec_dststt (const unsigned char *, int, const int);
7816 void vec_dststt (const signed char *, int, const int);
7817 void vec_dststt (const unsigned short *, int, const int);
7818 void vec_dststt (const short *, int, const int);
7819 void vec_dststt (const unsigned int *, int, const int);
7820 void vec_dststt (const int *, int, const int);
7821 void vec_dststt (const unsigned long *, int, const int);
7822 void vec_dststt (const long *, int, const int);
7823 void vec_dststt (const float *, int, const int);
7825 void vec_dstt (const vector unsigned char *, int, const int);
7826 void vec_dstt (const vector signed char *, int, const int);
7827 void vec_dstt (const vector bool char *, int, const int);
7828 void vec_dstt (const vector unsigned short *, int, const int);
7829 void vec_dstt (const vector signed short *, int, const int);
7830 void vec_dstt (const vector bool short *, int, const int);
7831 void vec_dstt (const vector pixel *, int, const int);
7832 void vec_dstt (const vector unsigned int *, int, const int);
7833 void vec_dstt (const vector signed int *, int, const int);
7834 void vec_dstt (const vector bool int *, int, const int);
7835 void vec_dstt (const vector float *, int, const int);
7836 void vec_dstt (const unsigned char *, int, const int);
7837 void vec_dstt (const signed char *, int, const int);
7838 void vec_dstt (const unsigned short *, int, const int);
7839 void vec_dstt (const short *, int, const int);
7840 void vec_dstt (const unsigned int *, int, const int);
7841 void vec_dstt (const int *, int, const int);
7842 void vec_dstt (const unsigned long *, int, const int);
7843 void vec_dstt (const long *, int, const int);
7844 void vec_dstt (const float *, int, const int);
7846 vector float vec_expte (vector float);
7848 vector float vec_floor (vector float);
7850 vector float vec_ld (int, const vector float *);
7851 vector float vec_ld (int, const float *);
7852 vector bool int vec_ld (int, const vector bool int *);
7853 vector signed int vec_ld (int, const vector signed int *);
7854 vector signed int vec_ld (int, const int *);
7855 vector signed int vec_ld (int, const long *);
7856 vector unsigned int vec_ld (int, const vector unsigned int *);
7857 vector unsigned int vec_ld (int, const unsigned int *);
7858 vector unsigned int vec_ld (int, const unsigned long *);
7859 vector bool short vec_ld (int, const vector bool short *);
7860 vector pixel vec_ld (int, const vector pixel *);
7861 vector signed short vec_ld (int, const vector signed short *);
7862 vector signed short vec_ld (int, const short *);
7863 vector unsigned short vec_ld (int, const vector unsigned short *);
7864 vector unsigned short vec_ld (int, const unsigned short *);
7865 vector bool char vec_ld (int, const vector bool char *);
7866 vector signed char vec_ld (int, const vector signed char *);
7867 vector signed char vec_ld (int, const signed char *);
7868 vector unsigned char vec_ld (int, const vector unsigned char *);
7869 vector unsigned char vec_ld (int, const unsigned char *);
7871 vector signed char vec_lde (int, const signed char *);
7872 vector unsigned char vec_lde (int, const unsigned char *);
7873 vector signed short vec_lde (int, const short *);
7874 vector unsigned short vec_lde (int, const unsigned short *);
7875 vector float vec_lde (int, const float *);
7876 vector signed int vec_lde (int, const int *);
7877 vector unsigned int vec_lde (int, const unsigned int *);
7878 vector signed int vec_lde (int, const long *);
7879 vector unsigned int vec_lde (int, const unsigned long *);
7881 vector float vec_lvewx (int, float *);
7882 vector signed int vec_lvewx (int, int *);
7883 vector unsigned int vec_lvewx (int, unsigned int *);
7884 vector signed int vec_lvewx (int, long *);
7885 vector unsigned int vec_lvewx (int, unsigned long *);
7887 vector signed short vec_lvehx (int, short *);
7888 vector unsigned short vec_lvehx (int, unsigned short *);
7890 vector signed char vec_lvebx (int, char *);
7891 vector unsigned char vec_lvebx (int, unsigned char *);
7893 vector float vec_ldl (int, const vector float *);
7894 vector float vec_ldl (int, const float *);
7895 vector bool int vec_ldl (int, const vector bool int *);
7896 vector signed int vec_ldl (int, const vector signed int *);
7897 vector signed int vec_ldl (int, const int *);
7898 vector signed int vec_ldl (int, const long *);
7899 vector unsigned int vec_ldl (int, const vector unsigned int *);
7900 vector unsigned int vec_ldl (int, const unsigned int *);
7901 vector unsigned int vec_ldl (int, const unsigned long *);
7902 vector bool short vec_ldl (int, const vector bool short *);
7903 vector pixel vec_ldl (int, const vector pixel *);
7904 vector signed short vec_ldl (int, const vector signed short *);
7905 vector signed short vec_ldl (int, const short *);
7906 vector unsigned short vec_ldl (int, const vector unsigned short *);
7907 vector unsigned short vec_ldl (int, const unsigned short *);
7908 vector bool char vec_ldl (int, const vector bool char *);
7909 vector signed char vec_ldl (int, const vector signed char *);
7910 vector signed char vec_ldl (int, const signed char *);
7911 vector unsigned char vec_ldl (int, const vector unsigned char *);
7912 vector unsigned char vec_ldl (int, const unsigned char *);
7914 vector float vec_loge (vector float);
7916 vector unsigned char vec_lvsl (int, const volatile unsigned char *);
7917 vector unsigned char vec_lvsl (int, const volatile signed char *);
7918 vector unsigned char vec_lvsl (int, const volatile unsigned short *);
7919 vector unsigned char vec_lvsl (int, const volatile short *);
7920 vector unsigned char vec_lvsl (int, const volatile unsigned int *);
7921 vector unsigned char vec_lvsl (int, const volatile int *);
7922 vector unsigned char vec_lvsl (int, const volatile unsigned long *);
7923 vector unsigned char vec_lvsl (int, const volatile long *);
7924 vector unsigned char vec_lvsl (int, const volatile float *);
7926 vector unsigned char vec_lvsr (int, const volatile unsigned char *);
7927 vector unsigned char vec_lvsr (int, const volatile signed char *);
7928 vector unsigned char vec_lvsr (int, const volatile unsigned short *);
7929 vector unsigned char vec_lvsr (int, const volatile short *);
7930 vector unsigned char vec_lvsr (int, const volatile unsigned int *);
7931 vector unsigned char vec_lvsr (int, const volatile int *);
7932 vector unsigned char vec_lvsr (int, const volatile unsigned long *);
7933 vector unsigned char vec_lvsr (int, const volatile long *);
7934 vector unsigned char vec_lvsr (int, const volatile float *);
7936 vector float vec_madd (vector float, vector float, vector float);
7938 vector signed short vec_madds (vector signed short,
7939 vector signed short,
7940 vector signed short);
7942 vector unsigned char vec_max (vector bool char, vector unsigned char);
7943 vector unsigned char vec_max (vector unsigned char, vector bool char);
7944 vector unsigned char vec_max (vector unsigned char,
7945 vector unsigned char);
7946 vector signed char vec_max (vector bool char, vector signed char);
7947 vector signed char vec_max (vector signed char, vector bool char);
7948 vector signed char vec_max (vector signed char, vector signed char);
7949 vector unsigned short vec_max (vector bool short,
7950 vector unsigned short);
7951 vector unsigned short vec_max (vector unsigned short,
7953 vector unsigned short vec_max (vector unsigned short,
7954 vector unsigned short);
7955 vector signed short vec_max (vector bool short, vector signed short);
7956 vector signed short vec_max (vector signed short, vector bool short);
7957 vector signed short vec_max (vector signed short, vector signed short);
7958 vector unsigned int vec_max (vector bool int, vector unsigned int);
7959 vector unsigned int vec_max (vector unsigned int, vector bool int);
7960 vector unsigned int vec_max (vector unsigned int, vector unsigned int);
7961 vector signed int vec_max (vector bool int, vector signed int);
7962 vector signed int vec_max (vector signed int, vector bool int);
7963 vector signed int vec_max (vector signed int, vector signed int);
7964 vector float vec_max (vector float, vector float);
7966 vector float vec_vmaxfp (vector float, vector float);
7968 vector signed int vec_vmaxsw (vector bool int, vector signed int);
7969 vector signed int vec_vmaxsw (vector signed int, vector bool int);
7970 vector signed int vec_vmaxsw (vector signed int, vector signed int);
7972 vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int);
7973 vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int);
7974 vector unsigned int vec_vmaxuw (vector unsigned int,
7975 vector unsigned int);
7977 vector signed short vec_vmaxsh (vector bool short, vector signed short);
7978 vector signed short vec_vmaxsh (vector signed short, vector bool short);
7979 vector signed short vec_vmaxsh (vector signed short,
7980 vector signed short);
7982 vector unsigned short vec_vmaxuh (vector bool short,
7983 vector unsigned short);
7984 vector unsigned short vec_vmaxuh (vector unsigned short,
7986 vector unsigned short vec_vmaxuh (vector unsigned short,
7987 vector unsigned short);
7989 vector signed char vec_vmaxsb (vector bool char, vector signed char);
7990 vector signed char vec_vmaxsb (vector signed char, vector bool char);
7991 vector signed char vec_vmaxsb (vector signed char, vector signed char);
7993 vector unsigned char vec_vmaxub (vector bool char,
7994 vector unsigned char);
7995 vector unsigned char vec_vmaxub (vector unsigned char,
7997 vector unsigned char vec_vmaxub (vector unsigned char,
7998 vector unsigned char);
8000 vector bool char vec_mergeh (vector bool char, vector bool char);
8001 vector signed char vec_mergeh (vector signed char, vector signed char);
8002 vector unsigned char vec_mergeh (vector unsigned char,
8003 vector unsigned char);
8004 vector bool short vec_mergeh (vector bool short, vector bool short);
8005 vector pixel vec_mergeh (vector pixel, vector pixel);
8006 vector signed short vec_mergeh (vector signed short,
8007 vector signed short);
8008 vector unsigned short vec_mergeh (vector unsigned short,
8009 vector unsigned short);
8010 vector float vec_mergeh (vector float, vector float);
8011 vector bool int vec_mergeh (vector bool int, vector bool int);
8012 vector signed int vec_mergeh (vector signed int, vector signed int);
8013 vector unsigned int vec_mergeh (vector unsigned int,
8014 vector unsigned int);
8016 vector float vec_vmrghw (vector float, vector float);
8017 vector bool int vec_vmrghw (vector bool int, vector bool int);
8018 vector signed int vec_vmrghw (vector signed int, vector signed int);
8019 vector unsigned int vec_vmrghw (vector unsigned int,
8020 vector unsigned int);
8022 vector bool short vec_vmrghh (vector bool short, vector bool short);
8023 vector signed short vec_vmrghh (vector signed short,
8024 vector signed short);
8025 vector unsigned short vec_vmrghh (vector unsigned short,
8026 vector unsigned short);
8027 vector pixel vec_vmrghh (vector pixel, vector pixel);
8029 vector bool char vec_vmrghb (vector bool char, vector bool char);
8030 vector signed char vec_vmrghb (vector signed char, vector signed char);
8031 vector unsigned char vec_vmrghb (vector unsigned char,
8032 vector unsigned char);
8034 vector bool char vec_mergel (vector bool char, vector bool char);
8035 vector signed char vec_mergel (vector signed char, vector signed char);
8036 vector unsigned char vec_mergel (vector unsigned char,
8037 vector unsigned char);
8038 vector bool short vec_mergel (vector bool short, vector bool short);
8039 vector pixel vec_mergel (vector pixel, vector pixel);
8040 vector signed short vec_mergel (vector signed short,
8041 vector signed short);
8042 vector unsigned short vec_mergel (vector unsigned short,
8043 vector unsigned short);
8044 vector float vec_mergel (vector float, vector float);
8045 vector bool int vec_mergel (vector bool int, vector bool int);
8046 vector signed int vec_mergel (vector signed int, vector signed int);
8047 vector unsigned int vec_mergel (vector unsigned int,
8048 vector unsigned int);
8050 vector float vec_vmrglw (vector float, vector float);
8051 vector signed int vec_vmrglw (vector signed int, vector signed int);
8052 vector unsigned int vec_vmrglw (vector unsigned int,
8053 vector unsigned int);
8054 vector bool int vec_vmrglw (vector bool int, vector bool int);
8056 vector bool short vec_vmrglh (vector bool short, vector bool short);
8057 vector signed short vec_vmrglh (vector signed short,
8058 vector signed short);
8059 vector unsigned short vec_vmrglh (vector unsigned short,
8060 vector unsigned short);
8061 vector pixel vec_vmrglh (vector pixel, vector pixel);
8063 vector bool char vec_vmrglb (vector bool char, vector bool char);
8064 vector signed char vec_vmrglb (vector signed char, vector signed char);
8065 vector unsigned char vec_vmrglb (vector unsigned char,
8066 vector unsigned char);
8068 vector unsigned short vec_mfvscr (void);
8070 vector unsigned char vec_min (vector bool char, vector unsigned char);
8071 vector unsigned char vec_min (vector unsigned char, vector bool char);
8072 vector unsigned char vec_min (vector unsigned char,
8073 vector unsigned char);
8074 vector signed char vec_min (vector bool char, vector signed char);
8075 vector signed char vec_min (vector signed char, vector bool char);
8076 vector signed char vec_min (vector signed char, vector signed char);
8077 vector unsigned short vec_min (vector bool short,
8078 vector unsigned short);
8079 vector unsigned short vec_min (vector unsigned short,
8081 vector unsigned short vec_min (vector unsigned short,
8082 vector unsigned short);
8083 vector signed short vec_min (vector bool short, vector signed short);
8084 vector signed short vec_min (vector signed short, vector bool short);
8085 vector signed short vec_min (vector signed short, vector signed short);
8086 vector unsigned int vec_min (vector bool int, vector unsigned int);
8087 vector unsigned int vec_min (vector unsigned int, vector bool int);
8088 vector unsigned int vec_min (vector unsigned int, vector unsigned int);
8089 vector signed int vec_min (vector bool int, vector signed int);
8090 vector signed int vec_min (vector signed int, vector bool int);
8091 vector signed int vec_min (vector signed int, vector signed int);
8092 vector float vec_min (vector float, vector float);
8094 vector float vec_vminfp (vector float, vector float);
8096 vector signed int vec_vminsw (vector bool int, vector signed int);
8097 vector signed int vec_vminsw (vector signed int, vector bool int);
8098 vector signed int vec_vminsw (vector signed int, vector signed int);
8100 vector unsigned int vec_vminuw (vector bool int, vector unsigned int);
8101 vector unsigned int vec_vminuw (vector unsigned int, vector bool int);
8102 vector unsigned int vec_vminuw (vector unsigned int,
8103 vector unsigned int);
8105 vector signed short vec_vminsh (vector bool short, vector signed short);
8106 vector signed short vec_vminsh (vector signed short, vector bool short);
8107 vector signed short vec_vminsh (vector signed short,
8108 vector signed short);
8110 vector unsigned short vec_vminuh (vector bool short,
8111 vector unsigned short);
8112 vector unsigned short vec_vminuh (vector unsigned short,
8114 vector unsigned short vec_vminuh (vector unsigned short,
8115 vector unsigned short);
8117 vector signed char vec_vminsb (vector bool char, vector signed char);
8118 vector signed char vec_vminsb (vector signed char, vector bool char);
8119 vector signed char vec_vminsb (vector signed char, vector signed char);
8121 vector unsigned char vec_vminub (vector bool char,
8122 vector unsigned char);
8123 vector unsigned char vec_vminub (vector unsigned char,
8125 vector unsigned char vec_vminub (vector unsigned char,
8126 vector unsigned char);
8128 vector signed short vec_mladd (vector signed short,
8129 vector signed short,
8130 vector signed short);
8131 vector signed short vec_mladd (vector signed short,
8132 vector unsigned short,
8133 vector unsigned short);
8134 vector signed short vec_mladd (vector unsigned short,
8135 vector signed short,
8136 vector signed short);
8137 vector unsigned short vec_mladd (vector unsigned short,
8138 vector unsigned short,
8139 vector unsigned short);
8141 vector signed short vec_mradds (vector signed short,
8142 vector signed short,
8143 vector signed short);
8145 vector unsigned int vec_msum (vector unsigned char,
8146 vector unsigned char,
8147 vector unsigned int);
8148 vector signed int vec_msum (vector signed char,
8149 vector unsigned char,
8151 vector unsigned int vec_msum (vector unsigned short,
8152 vector unsigned short,
8153 vector unsigned int);
8154 vector signed int vec_msum (vector signed short,
8155 vector signed short,
8158 vector signed int vec_vmsumshm (vector signed short,
8159 vector signed short,
8162 vector unsigned int vec_vmsumuhm (vector unsigned short,
8163 vector unsigned short,
8164 vector unsigned int);
8166 vector signed int vec_vmsummbm (vector signed char,
8167 vector unsigned char,
8170 vector unsigned int vec_vmsumubm (vector unsigned char,
8171 vector unsigned char,
8172 vector unsigned int);
8174 vector unsigned int vec_msums (vector unsigned short,
8175 vector unsigned short,
8176 vector unsigned int);
8177 vector signed int vec_msums (vector signed short,
8178 vector signed short,
8181 vector signed int vec_vmsumshs (vector signed short,
8182 vector signed short,
8185 vector unsigned int vec_vmsumuhs (vector unsigned short,
8186 vector unsigned short,
8187 vector unsigned int);
8189 void vec_mtvscr (vector signed int);
8190 void vec_mtvscr (vector unsigned int);
8191 void vec_mtvscr (vector bool int);
8192 void vec_mtvscr (vector signed short);
8193 void vec_mtvscr (vector unsigned short);
8194 void vec_mtvscr (vector bool short);
8195 void vec_mtvscr (vector pixel);
8196 void vec_mtvscr (vector signed char);
8197 void vec_mtvscr (vector unsigned char);
8198 void vec_mtvscr (vector bool char);
8200 vector unsigned short vec_mule (vector unsigned char,
8201 vector unsigned char);
8202 vector signed short vec_mule (vector signed char,
8203 vector signed char);
8204 vector unsigned int vec_mule (vector unsigned short,
8205 vector unsigned short);
8206 vector signed int vec_mule (vector signed short, vector signed short);
8208 vector signed int vec_vmulesh (vector signed short,
8209 vector signed short);
8211 vector unsigned int vec_vmuleuh (vector unsigned short,
8212 vector unsigned short);
8214 vector signed short vec_vmulesb (vector signed char,
8215 vector signed char);
8217 vector unsigned short vec_vmuleub (vector unsigned char,
8218 vector unsigned char);
8220 vector unsigned short vec_mulo (vector unsigned char,
8221 vector unsigned char);
8222 vector signed short vec_mulo (vector signed char, vector signed char);
8223 vector unsigned int vec_mulo (vector unsigned short,
8224 vector unsigned short);
8225 vector signed int vec_mulo (vector signed short, vector signed short);
8227 vector signed int vec_vmulosh (vector signed short,
8228 vector signed short);
8230 vector unsigned int vec_vmulouh (vector unsigned short,
8231 vector unsigned short);
8233 vector signed short vec_vmulosb (vector signed char,
8234 vector signed char);
8236 vector unsigned short vec_vmuloub (vector unsigned char,
8237 vector unsigned char);
8239 vector float vec_nmsub (vector float, vector float, vector float);
8241 vector float vec_nor (vector float, vector float);
8242 vector signed int vec_nor (vector signed int, vector signed int);
8243 vector unsigned int vec_nor (vector unsigned int, vector unsigned int);
8244 vector bool int vec_nor (vector bool int, vector bool int);
8245 vector signed short vec_nor (vector signed short, vector signed short);
8246 vector unsigned short vec_nor (vector unsigned short,
8247 vector unsigned short);
8248 vector bool short vec_nor (vector bool short, vector bool short);
8249 vector signed char vec_nor (vector signed char, vector signed char);
8250 vector unsigned char vec_nor (vector unsigned char,
8251 vector unsigned char);
8252 vector bool char vec_nor (vector bool char, vector bool char);
8254 vector float vec_or (vector float, vector float);
8255 vector float vec_or (vector float, vector bool int);
8256 vector float vec_or (vector bool int, vector float);
8257 vector bool int vec_or (vector bool int, vector bool int);
8258 vector signed int vec_or (vector bool int, vector signed int);
8259 vector signed int vec_or (vector signed int, vector bool int);
8260 vector signed int vec_or (vector signed int, vector signed int);
8261 vector unsigned int vec_or (vector bool int, vector unsigned int);
8262 vector unsigned int vec_or (vector unsigned int, vector bool int);
8263 vector unsigned int vec_or (vector unsigned int, vector unsigned int);
8264 vector bool short vec_or (vector bool short, vector bool short);
8265 vector signed short vec_or (vector bool short, vector signed short);
8266 vector signed short vec_or (vector signed short, vector bool short);
8267 vector signed short vec_or (vector signed short, vector signed short);
8268 vector unsigned short vec_or (vector bool short, vector unsigned short);
8269 vector unsigned short vec_or (vector unsigned short, vector bool short);
8270 vector unsigned short vec_or (vector unsigned short,
8271 vector unsigned short);
8272 vector signed char vec_or (vector bool char, vector signed char);
8273 vector bool char vec_or (vector bool char, vector bool char);
8274 vector signed char vec_or (vector signed char, vector bool char);
8275 vector signed char vec_or (vector signed char, vector signed char);
8276 vector unsigned char vec_or (vector bool char, vector unsigned char);
8277 vector unsigned char vec_or (vector unsigned char, vector bool char);
8278 vector unsigned char vec_or (vector unsigned char,
8279 vector unsigned char);
8281 vector signed char vec_pack (vector signed short, vector signed short);
8282 vector unsigned char vec_pack (vector unsigned short,
8283 vector unsigned short);
8284 vector bool char vec_pack (vector bool short, vector bool short);
8285 vector signed short vec_pack (vector signed int, vector signed int);
8286 vector unsigned short vec_pack (vector unsigned int,
8287 vector unsigned int);
8288 vector bool short vec_pack (vector bool int, vector bool int);
8290 vector bool short vec_vpkuwum (vector bool int, vector bool int);
8291 vector signed short vec_vpkuwum (vector signed int, vector signed int);
8292 vector unsigned short vec_vpkuwum (vector unsigned int,
8293 vector unsigned int);
8295 vector bool char vec_vpkuhum (vector bool short, vector bool short);
8296 vector signed char vec_vpkuhum (vector signed short,
8297 vector signed short);
8298 vector unsigned char vec_vpkuhum (vector unsigned short,
8299 vector unsigned short);
8301 vector pixel vec_packpx (vector unsigned int, vector unsigned int);
8303 vector unsigned char vec_packs (vector unsigned short,
8304 vector unsigned short);
8305 vector signed char vec_packs (vector signed short, vector signed short);
8306 vector unsigned short vec_packs (vector unsigned int,
8307 vector unsigned int);
8308 vector signed short vec_packs (vector signed int, vector signed int);
8310 vector signed short vec_vpkswss (vector signed int, vector signed int);
8312 vector unsigned short vec_vpkuwus (vector unsigned int,
8313 vector unsigned int);
8315 vector signed char vec_vpkshss (vector signed short,
8316 vector signed short);
8318 vector unsigned char vec_vpkuhus (vector unsigned short,
8319 vector unsigned short);
8321 vector unsigned char vec_packsu (vector unsigned short,
8322 vector unsigned short);
8323 vector unsigned char vec_packsu (vector signed short,
8324 vector signed short);
8325 vector unsigned short vec_packsu (vector unsigned int,
8326 vector unsigned int);
8327 vector unsigned short vec_packsu (vector signed int, vector signed int);
8329 vector unsigned short vec_vpkswus (vector signed int,
8332 vector unsigned char vec_vpkshus (vector signed short,
8333 vector signed short);
8335 vector float vec_perm (vector float,
8337 vector unsigned char);
8338 vector signed int vec_perm (vector signed int,
8340 vector unsigned char);
8341 vector unsigned int vec_perm (vector unsigned int,
8342 vector unsigned int,
8343 vector unsigned char);
8344 vector bool int vec_perm (vector bool int,
8346 vector unsigned char);
8347 vector signed short vec_perm (vector signed short,
8348 vector signed short,
8349 vector unsigned char);
8350 vector unsigned short vec_perm (vector unsigned short,
8351 vector unsigned short,
8352 vector unsigned char);
8353 vector bool short vec_perm (vector bool short,
8355 vector unsigned char);
8356 vector pixel vec_perm (vector pixel,
8358 vector unsigned char);
8359 vector signed char vec_perm (vector signed char,
8361 vector unsigned char);
8362 vector unsigned char vec_perm (vector unsigned char,
8363 vector unsigned char,
8364 vector unsigned char);
8365 vector bool char vec_perm (vector bool char,
8367 vector unsigned char);
8369 vector float vec_re (vector float);
8371 vector signed char vec_rl (vector signed char,
8372 vector unsigned char);
8373 vector unsigned char vec_rl (vector unsigned char,
8374 vector unsigned char);
8375 vector signed short vec_rl (vector signed short, vector unsigned short);
8376 vector unsigned short vec_rl (vector unsigned short,
8377 vector unsigned short);
8378 vector signed int vec_rl (vector signed int, vector unsigned int);
8379 vector unsigned int vec_rl (vector unsigned int, vector unsigned int);
8381 vector signed int vec_vrlw (vector signed int, vector unsigned int);
8382 vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int);
8384 vector signed short vec_vrlh (vector signed short,
8385 vector unsigned short);
8386 vector unsigned short vec_vrlh (vector unsigned short,
8387 vector unsigned short);
8389 vector signed char vec_vrlb (vector signed char, vector unsigned char);
8390 vector unsigned char vec_vrlb (vector unsigned char,
8391 vector unsigned char);
8393 vector float vec_round (vector float);
8395 vector float vec_rsqrte (vector float);
8397 vector float vec_sel (vector float, vector float, vector bool int);
8398 vector float vec_sel (vector float, vector float, vector unsigned int);
8399 vector signed int vec_sel (vector signed int,
8402 vector signed int vec_sel (vector signed int,
8404 vector unsigned int);
8405 vector unsigned int vec_sel (vector unsigned int,
8406 vector unsigned int,
8408 vector unsigned int vec_sel (vector unsigned int,
8409 vector unsigned int,
8410 vector unsigned int);
8411 vector bool int vec_sel (vector bool int,
8414 vector bool int vec_sel (vector bool int,
8416 vector unsigned int);
8417 vector signed short vec_sel (vector signed short,
8418 vector signed short,
8420 vector signed short vec_sel (vector signed short,
8421 vector signed short,
8422 vector unsigned short);
8423 vector unsigned short vec_sel (vector unsigned short,
8424 vector unsigned short,
8426 vector unsigned short vec_sel (vector unsigned short,
8427 vector unsigned short,
8428 vector unsigned short);
8429 vector bool short vec_sel (vector bool short,
8432 vector bool short vec_sel (vector bool short,
8434 vector unsigned short);
8435 vector signed char vec_sel (vector signed char,
8438 vector signed char vec_sel (vector signed char,
8440 vector unsigned char);
8441 vector unsigned char vec_sel (vector unsigned char,
8442 vector unsigned char,
8444 vector unsigned char vec_sel (vector unsigned char,
8445 vector unsigned char,
8446 vector unsigned char);
8447 vector bool char vec_sel (vector bool char,
8450 vector bool char vec_sel (vector bool char,
8452 vector unsigned char);
8454 vector signed char vec_sl (vector signed char,
8455 vector unsigned char);
8456 vector unsigned char vec_sl (vector unsigned char,
8457 vector unsigned char);
8458 vector signed short vec_sl (vector signed short, vector unsigned short);
8459 vector unsigned short vec_sl (vector unsigned short,
8460 vector unsigned short);
8461 vector signed int vec_sl (vector signed int, vector unsigned int);
8462 vector unsigned int vec_sl (vector unsigned int, vector unsigned int);
8464 vector signed int vec_vslw (vector signed int, vector unsigned int);
8465 vector unsigned int vec_vslw (vector unsigned int, vector unsigned int);
8467 vector signed short vec_vslh (vector signed short,
8468 vector unsigned short);
8469 vector unsigned short vec_vslh (vector unsigned short,
8470 vector unsigned short);
8472 vector signed char vec_vslb (vector signed char, vector unsigned char);
8473 vector unsigned char vec_vslb (vector unsigned char,
8474 vector unsigned char);
8476 vector float vec_sld (vector float, vector float, const int);
8477 vector signed int vec_sld (vector signed int,
8480 vector unsigned int vec_sld (vector unsigned int,
8481 vector unsigned int,
8483 vector bool int vec_sld (vector bool int,
8486 vector signed short vec_sld (vector signed short,
8487 vector signed short,
8489 vector unsigned short vec_sld (vector unsigned short,
8490 vector unsigned short,
8492 vector bool short vec_sld (vector bool short,
8495 vector pixel vec_sld (vector pixel,
8498 vector signed char vec_sld (vector signed char,
8501 vector unsigned char vec_sld (vector unsigned char,
8502 vector unsigned char,
8504 vector bool char vec_sld (vector bool char,
8508 vector signed int vec_sll (vector signed int,
8509 vector unsigned int);
8510 vector signed int vec_sll (vector signed int,
8511 vector unsigned short);
8512 vector signed int vec_sll (vector signed int,
8513 vector unsigned char);
8514 vector unsigned int vec_sll (vector unsigned int,
8515 vector unsigned int);
8516 vector unsigned int vec_sll (vector unsigned int,
8517 vector unsigned short);
8518 vector unsigned int vec_sll (vector unsigned int,
8519 vector unsigned char);
8520 vector bool int vec_sll (vector bool int,
8521 vector unsigned int);
8522 vector bool int vec_sll (vector bool int,
8523 vector unsigned short);
8524 vector bool int vec_sll (vector bool int,
8525 vector unsigned char);
8526 vector signed short vec_sll (vector signed short,
8527 vector unsigned int);
8528 vector signed short vec_sll (vector signed short,
8529 vector unsigned short);
8530 vector signed short vec_sll (vector signed short,
8531 vector unsigned char);
8532 vector unsigned short vec_sll (vector unsigned short,
8533 vector unsigned int);
8534 vector unsigned short vec_sll (vector unsigned short,
8535 vector unsigned short);
8536 vector unsigned short vec_sll (vector unsigned short,
8537 vector unsigned char);
8538 vector bool short vec_sll (vector bool short, vector unsigned int);
8539 vector bool short vec_sll (vector bool short, vector unsigned short);
8540 vector bool short vec_sll (vector bool short, vector unsigned char);
8541 vector pixel vec_sll (vector pixel, vector unsigned int);
8542 vector pixel vec_sll (vector pixel, vector unsigned short);
8543 vector pixel vec_sll (vector pixel, vector unsigned char);
8544 vector signed char vec_sll (vector signed char, vector unsigned int);
8545 vector signed char vec_sll (vector signed char, vector unsigned short);
8546 vector signed char vec_sll (vector signed char, vector unsigned char);
8547 vector unsigned char vec_sll (vector unsigned char,
8548 vector unsigned int);
8549 vector unsigned char vec_sll (vector unsigned char,
8550 vector unsigned short);
8551 vector unsigned char vec_sll (vector unsigned char,
8552 vector unsigned char);
8553 vector bool char vec_sll (vector bool char, vector unsigned int);
8554 vector bool char vec_sll (vector bool char, vector unsigned short);
8555 vector bool char vec_sll (vector bool char, vector unsigned char);
8557 vector float vec_slo (vector float, vector signed char);
8558 vector float vec_slo (vector float, vector unsigned char);
8559 vector signed int vec_slo (vector signed int, vector signed char);
8560 vector signed int vec_slo (vector signed int, vector unsigned char);
8561 vector unsigned int vec_slo (vector unsigned int, vector signed char);
8562 vector unsigned int vec_slo (vector unsigned int, vector unsigned char);
8563 vector signed short vec_slo (vector signed short, vector signed char);
8564 vector signed short vec_slo (vector signed short, vector unsigned char);
8565 vector unsigned short vec_slo (vector unsigned short,
8566 vector signed char);
8567 vector unsigned short vec_slo (vector unsigned short,
8568 vector unsigned char);
8569 vector pixel vec_slo (vector pixel, vector signed char);
8570 vector pixel vec_slo (vector pixel, vector unsigned char);
8571 vector signed char vec_slo (vector signed char, vector signed char);
8572 vector signed char vec_slo (vector signed char, vector unsigned char);
8573 vector unsigned char vec_slo (vector unsigned char, vector signed char);
8574 vector unsigned char vec_slo (vector unsigned char,
8575 vector unsigned char);
8577 vector signed char vec_splat (vector signed char, const int);
8578 vector unsigned char vec_splat (vector unsigned char, const int);
8579 vector bool char vec_splat (vector bool char, const int);
8580 vector signed short vec_splat (vector signed short, const int);
8581 vector unsigned short vec_splat (vector unsigned short, const int);
8582 vector bool short vec_splat (vector bool short, const int);
8583 vector pixel vec_splat (vector pixel, const int);
8584 vector float vec_splat (vector float, const int);
8585 vector signed int vec_splat (vector signed int, const int);
8586 vector unsigned int vec_splat (vector unsigned int, const int);
8587 vector bool int vec_splat (vector bool int, const int);
8589 vector float vec_vspltw (vector float, const int);
8590 vector signed int vec_vspltw (vector signed int, const int);
8591 vector unsigned int vec_vspltw (vector unsigned int, const int);
8592 vector bool int vec_vspltw (vector bool int, const int);
8594 vector bool short vec_vsplth (vector bool short, const int);
8595 vector signed short vec_vsplth (vector signed short, const int);
8596 vector unsigned short vec_vsplth (vector unsigned short, const int);
8597 vector pixel vec_vsplth (vector pixel, const int);
8599 vector signed char vec_vspltb (vector signed char, const int);
8600 vector unsigned char vec_vspltb (vector unsigned char, const int);
8601 vector bool char vec_vspltb (vector bool char, const int);
8603 vector signed char vec_splat_s8 (const int);
8605 vector signed short vec_splat_s16 (const int);
8607 vector signed int vec_splat_s32 (const int);
8609 vector unsigned char vec_splat_u8 (const int);
8611 vector unsigned short vec_splat_u16 (const int);
8613 vector unsigned int vec_splat_u32 (const int);
8615 vector signed char vec_sr (vector signed char, vector unsigned char);
8616 vector unsigned char vec_sr (vector unsigned char,
8617 vector unsigned char);
8618 vector signed short vec_sr (vector signed short,
8619 vector unsigned short);
8620 vector unsigned short vec_sr (vector unsigned short,
8621 vector unsigned short);
8622 vector signed int vec_sr (vector signed int, vector unsigned int);
8623 vector unsigned int vec_sr (vector unsigned int, vector unsigned int);
8625 vector signed int vec_vsrw (vector signed int, vector unsigned int);
8626 vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int);
8628 vector signed short vec_vsrh (vector signed short,
8629 vector unsigned short);
8630 vector unsigned short vec_vsrh (vector unsigned short,
8631 vector unsigned short);
8633 vector signed char vec_vsrb (vector signed char, vector unsigned char);
8634 vector unsigned char vec_vsrb (vector unsigned char,
8635 vector unsigned char);
8637 vector signed char vec_sra (vector signed char, vector unsigned char);
8638 vector unsigned char vec_sra (vector unsigned char,
8639 vector unsigned char);
8640 vector signed short vec_sra (vector signed short,
8641 vector unsigned short);
8642 vector unsigned short vec_sra (vector unsigned short,
8643 vector unsigned short);
8644 vector signed int vec_sra (vector signed int, vector unsigned int);
8645 vector unsigned int vec_sra (vector unsigned int, vector unsigned int);
8647 vector signed int vec_vsraw (vector signed int, vector unsigned int);
8648 vector unsigned int vec_vsraw (vector unsigned int,
8649 vector unsigned int);
8651 vector signed short vec_vsrah (vector signed short,
8652 vector unsigned short);
8653 vector unsigned short vec_vsrah (vector unsigned short,
8654 vector unsigned short);
8656 vector signed char vec_vsrab (vector signed char, vector unsigned char);
8657 vector unsigned char vec_vsrab (vector unsigned char,
8658 vector unsigned char);
8660 vector signed int vec_srl (vector signed int, vector unsigned int);
8661 vector signed int vec_srl (vector signed int, vector unsigned short);
8662 vector signed int vec_srl (vector signed int, vector unsigned char);
8663 vector unsigned int vec_srl (vector unsigned int, vector unsigned int);
8664 vector unsigned int vec_srl (vector unsigned int,
8665 vector unsigned short);
8666 vector unsigned int vec_srl (vector unsigned int, vector unsigned char);
8667 vector bool int vec_srl (vector bool int, vector unsigned int);
8668 vector bool int vec_srl (vector bool int, vector unsigned short);
8669 vector bool int vec_srl (vector bool int, vector unsigned char);
8670 vector signed short vec_srl (vector signed short, vector unsigned int);
8671 vector signed short vec_srl (vector signed short,
8672 vector unsigned short);
8673 vector signed short vec_srl (vector signed short, vector unsigned char);
8674 vector unsigned short vec_srl (vector unsigned short,
8675 vector unsigned int);
8676 vector unsigned short vec_srl (vector unsigned short,
8677 vector unsigned short);
8678 vector unsigned short vec_srl (vector unsigned short,
8679 vector unsigned char);
8680 vector bool short vec_srl (vector bool short, vector unsigned int);
8681 vector bool short vec_srl (vector bool short, vector unsigned short);
8682 vector bool short vec_srl (vector bool short, vector unsigned char);
8683 vector pixel vec_srl (vector pixel, vector unsigned int);
8684 vector pixel vec_srl (vector pixel, vector unsigned short);
8685 vector pixel vec_srl (vector pixel, vector unsigned char);
8686 vector signed char vec_srl (vector signed char, vector unsigned int);
8687 vector signed char vec_srl (vector signed char, vector unsigned short);
8688 vector signed char vec_srl (vector signed char, vector unsigned char);
8689 vector unsigned char vec_srl (vector unsigned char,
8690 vector unsigned int);
8691 vector unsigned char vec_srl (vector unsigned char,
8692 vector unsigned short);
8693 vector unsigned char vec_srl (vector unsigned char,
8694 vector unsigned char);
8695 vector bool char vec_srl (vector bool char, vector unsigned int);
8696 vector bool char vec_srl (vector bool char, vector unsigned short);
8697 vector bool char vec_srl (vector bool char, vector unsigned char);
8699 vector float vec_sro (vector float, vector signed char);
8700 vector float vec_sro (vector float, vector unsigned char);
8701 vector signed int vec_sro (vector signed int, vector signed char);
8702 vector signed int vec_sro (vector signed int, vector unsigned char);
8703 vector unsigned int vec_sro (vector unsigned int, vector signed char);
8704 vector unsigned int vec_sro (vector unsigned int, vector unsigned char);
8705 vector signed short vec_sro (vector signed short, vector signed char);
8706 vector signed short vec_sro (vector signed short, vector unsigned char);
8707 vector unsigned short vec_sro (vector unsigned short,
8708 vector signed char);
8709 vector unsigned short vec_sro (vector unsigned short,
8710 vector unsigned char);
8711 vector pixel vec_sro (vector pixel, vector signed char);
8712 vector pixel vec_sro (vector pixel, vector unsigned char);
8713 vector signed char vec_sro (vector signed char, vector signed char);
8714 vector signed char vec_sro (vector signed char, vector unsigned char);
8715 vector unsigned char vec_sro (vector unsigned char, vector signed char);
8716 vector unsigned char vec_sro (vector unsigned char,
8717 vector unsigned char);
8719 void vec_st (vector float, int, vector float *);
8720 void vec_st (vector float, int, float *);
8721 void vec_st (vector signed int, int, vector signed int *);
8722 void vec_st (vector signed int, int, int *);
8723 void vec_st (vector unsigned int, int, vector unsigned int *);
8724 void vec_st (vector unsigned int, int, unsigned int *);
8725 void vec_st (vector bool int, int, vector bool int *);
8726 void vec_st (vector bool int, int, unsigned int *);
8727 void vec_st (vector bool int, int, int *);
8728 void vec_st (vector signed short, int, vector signed short *);
8729 void vec_st (vector signed short, int, short *);
8730 void vec_st (vector unsigned short, int, vector unsigned short *);
8731 void vec_st (vector unsigned short, int, unsigned short *);
8732 void vec_st (vector bool short, int, vector bool short *);
8733 void vec_st (vector bool short, int, unsigned short *);
8734 void vec_st (vector pixel, int, vector pixel *);
8735 void vec_st (vector pixel, int, unsigned short *);
8736 void vec_st (vector pixel, int, short *);
8737 void vec_st (vector bool short, int, short *);
8738 void vec_st (vector signed char, int, vector signed char *);
8739 void vec_st (vector signed char, int, signed char *);
8740 void vec_st (vector unsigned char, int, vector unsigned char *);
8741 void vec_st (vector unsigned char, int, unsigned char *);
8742 void vec_st (vector bool char, int, vector bool char *);
8743 void vec_st (vector bool char, int, unsigned char *);
8744 void vec_st (vector bool char, int, signed char *);
8746 void vec_ste (vector signed char, int, signed char *);
8747 void vec_ste (vector unsigned char, int, unsigned char *);
8748 void vec_ste (vector bool char, int, signed char *);
8749 void vec_ste (vector bool char, int, unsigned char *);
8750 void vec_ste (vector signed short, int, short *);
8751 void vec_ste (vector unsigned short, int, unsigned short *);
8752 void vec_ste (vector bool short, int, short *);
8753 void vec_ste (vector bool short, int, unsigned short *);
8754 void vec_ste (vector pixel, int, short *);
8755 void vec_ste (vector pixel, int, unsigned short *);
8756 void vec_ste (vector float, int, float *);
8757 void vec_ste (vector signed int, int, int *);
8758 void vec_ste (vector unsigned int, int, unsigned int *);
8759 void vec_ste (vector bool int, int, int *);
8760 void vec_ste (vector bool int, int, unsigned int *);
8762 void vec_stvewx (vector float, int, float *);
8763 void vec_stvewx (vector signed int, int, int *);
8764 void vec_stvewx (vector unsigned int, int, unsigned int *);
8765 void vec_stvewx (vector bool int, int, int *);
8766 void vec_stvewx (vector bool int, int, unsigned int *);
8768 void vec_stvehx (vector signed short, int, short *);
8769 void vec_stvehx (vector unsigned short, int, unsigned short *);
8770 void vec_stvehx (vector bool short, int, short *);
8771 void vec_stvehx (vector bool short, int, unsigned short *);
8772 void vec_stvehx (vector pixel, int, short *);
8773 void vec_stvehx (vector pixel, int, unsigned short *);
8775 void vec_stvebx (vector signed char, int, signed char *);
8776 void vec_stvebx (vector unsigned char, int, unsigned char *);
8777 void vec_stvebx (vector bool char, int, signed char *);
8778 void vec_stvebx (vector bool char, int, unsigned char *);
8780 void vec_stl (vector float, int, vector float *);
8781 void vec_stl (vector float, int, float *);
8782 void vec_stl (vector signed int, int, vector signed int *);
8783 void vec_stl (vector signed int, int, int *);
8784 void vec_stl (vector unsigned int, int, vector unsigned int *);
8785 void vec_stl (vector unsigned int, int, unsigned int *);
8786 void vec_stl (vector bool int, int, vector bool int *);
8787 void vec_stl (vector bool int, int, unsigned int *);
8788 void vec_stl (vector bool int, int, int *);
8789 void vec_stl (vector signed short, int, vector signed short *);
8790 void vec_stl (vector signed short, int, short *);
8791 void vec_stl (vector unsigned short, int, vector unsigned short *);
8792 void vec_stl (vector unsigned short, int, unsigned short *);
8793 void vec_stl (vector bool short, int, vector bool short *);
8794 void vec_stl (vector bool short, int, unsigned short *);
8795 void vec_stl (vector bool short, int, short *);
8796 void vec_stl (vector pixel, int, vector pixel *);
8797 void vec_stl (vector pixel, int, unsigned short *);
8798 void vec_stl (vector pixel, int, short *);
8799 void vec_stl (vector signed char, int, vector signed char *);
8800 void vec_stl (vector signed char, int, signed char *);
8801 void vec_stl (vector unsigned char, int, vector unsigned char *);
8802 void vec_stl (vector unsigned char, int, unsigned char *);
8803 void vec_stl (vector bool char, int, vector bool char *);
8804 void vec_stl (vector bool char, int, unsigned char *);
8805 void vec_stl (vector bool char, int, signed char *);
8807 vector signed char vec_sub (vector bool char, vector signed char);
8808 vector signed char vec_sub (vector signed char, vector bool char);
8809 vector signed char vec_sub (vector signed char, vector signed char);
8810 vector unsigned char vec_sub (vector bool char, vector unsigned char);
8811 vector unsigned char vec_sub (vector unsigned char, vector bool char);
8812 vector unsigned char vec_sub (vector unsigned char,
8813 vector unsigned char);
8814 vector signed short vec_sub (vector bool short, vector signed short);
8815 vector signed short vec_sub (vector signed short, vector bool short);
8816 vector signed short vec_sub (vector signed short, vector signed short);
8817 vector unsigned short vec_sub (vector bool short,
8818 vector unsigned short);
8819 vector unsigned short vec_sub (vector unsigned short,
8821 vector unsigned short vec_sub (vector unsigned short,
8822 vector unsigned short);
8823 vector signed int vec_sub (vector bool int, vector signed int);
8824 vector signed int vec_sub (vector signed int, vector bool int);
8825 vector signed int vec_sub (vector signed int, vector signed int);
8826 vector unsigned int vec_sub (vector bool int, vector unsigned int);
8827 vector unsigned int vec_sub (vector unsigned int, vector bool int);
8828 vector unsigned int vec_sub (vector unsigned int, vector unsigned int);
8829 vector float vec_sub (vector float, vector float);
8831 vector float vec_vsubfp (vector float, vector float);
8833 vector signed int vec_vsubuwm (vector bool int, vector signed int);
8834 vector signed int vec_vsubuwm (vector signed int, vector bool int);
8835 vector signed int vec_vsubuwm (vector signed int, vector signed int);
8836 vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int);
8837 vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int);
8838 vector unsigned int vec_vsubuwm (vector unsigned int,
8839 vector unsigned int);
8841 vector signed short vec_vsubuhm (vector bool short,
8842 vector signed short);
8843 vector signed short vec_vsubuhm (vector signed short,
8845 vector signed short vec_vsubuhm (vector signed short,
8846 vector signed short);
8847 vector unsigned short vec_vsubuhm (vector bool short,
8848 vector unsigned short);
8849 vector unsigned short vec_vsubuhm (vector unsigned short,
8851 vector unsigned short vec_vsubuhm (vector unsigned short,
8852 vector unsigned short);
8854 vector signed char vec_vsububm (vector bool char, vector signed char);
8855 vector signed char vec_vsububm (vector signed char, vector bool char);
8856 vector signed char vec_vsububm (vector signed char, vector signed char);
8857 vector unsigned char vec_vsububm (vector bool char,
8858 vector unsigned char);
8859 vector unsigned char vec_vsububm (vector unsigned char,
8861 vector unsigned char vec_vsububm (vector unsigned char,
8862 vector unsigned char);
8864 vector unsigned int vec_subc (vector unsigned int, vector unsigned int);
8866 vector unsigned char vec_subs (vector bool char, vector unsigned char);
8867 vector unsigned char vec_subs (vector unsigned char, vector bool char);
8868 vector unsigned char vec_subs (vector unsigned char,
8869 vector unsigned char);
8870 vector signed char vec_subs (vector bool char, vector signed char);
8871 vector signed char vec_subs (vector signed char, vector bool char);
8872 vector signed char vec_subs (vector signed char, vector signed char);
8873 vector unsigned short vec_subs (vector bool short,
8874 vector unsigned short);
8875 vector unsigned short vec_subs (vector unsigned short,
8877 vector unsigned short vec_subs (vector unsigned short,
8878 vector unsigned short);
8879 vector signed short vec_subs (vector bool short, vector signed short);
8880 vector signed short vec_subs (vector signed short, vector bool short);
8881 vector signed short vec_subs (vector signed short, vector signed short);
8882 vector unsigned int vec_subs (vector bool int, vector unsigned int);
8883 vector unsigned int vec_subs (vector unsigned int, vector bool int);
8884 vector unsigned int vec_subs (vector unsigned int, vector unsigned int);
8885 vector signed int vec_subs (vector bool int, vector signed int);
8886 vector signed int vec_subs (vector signed int, vector bool int);
8887 vector signed int vec_subs (vector signed int, vector signed int);
8889 vector signed int vec_vsubsws (vector bool int, vector signed int);
8890 vector signed int vec_vsubsws (vector signed int, vector bool int);
8891 vector signed int vec_vsubsws (vector signed int, vector signed int);
8893 vector unsigned int vec_vsubuws (vector bool int, vector unsigned int);
8894 vector unsigned int vec_vsubuws (vector unsigned int, vector bool int);
8895 vector unsigned int vec_vsubuws (vector unsigned int,
8896 vector unsigned int);
8898 vector signed short vec_vsubshs (vector bool short,
8899 vector signed short);
8900 vector signed short vec_vsubshs (vector signed short,
8902 vector signed short vec_vsubshs (vector signed short,
8903 vector signed short);
8905 vector unsigned short vec_vsubuhs (vector bool short,
8906 vector unsigned short);
8907 vector unsigned short vec_vsubuhs (vector unsigned short,
8909 vector unsigned short vec_vsubuhs (vector unsigned short,
8910 vector unsigned short);
8912 vector signed char vec_vsubsbs (vector bool char, vector signed char);
8913 vector signed char vec_vsubsbs (vector signed char, vector bool char);
8914 vector signed char vec_vsubsbs (vector signed char, vector signed char);
8916 vector unsigned char vec_vsububs (vector bool char,
8917 vector unsigned char);
8918 vector unsigned char vec_vsububs (vector unsigned char,
8920 vector unsigned char vec_vsububs (vector unsigned char,
8921 vector unsigned char);
8923 vector unsigned int vec_sum4s (vector unsigned char,
8924 vector unsigned int);
8925 vector signed int vec_sum4s (vector signed char, vector signed int);
8926 vector signed int vec_sum4s (vector signed short, vector signed int);
8928 vector signed int vec_vsum4shs (vector signed short, vector signed int);
8930 vector signed int vec_vsum4sbs (vector signed char, vector signed int);
8932 vector unsigned int vec_vsum4ubs (vector unsigned char,
8933 vector unsigned int);
8935 vector signed int vec_sum2s (vector signed int, vector signed int);
8937 vector signed int vec_sums (vector signed int, vector signed int);
8939 vector float vec_trunc (vector float);
8941 vector signed short vec_unpackh (vector signed char);
8942 vector bool short vec_unpackh (vector bool char);
8943 vector signed int vec_unpackh (vector signed short);
8944 vector bool int vec_unpackh (vector bool short);
8945 vector unsigned int vec_unpackh (vector pixel);
8947 vector bool int vec_vupkhsh (vector bool short);
8948 vector signed int vec_vupkhsh (vector signed short);
8950 vector unsigned int vec_vupkhpx (vector pixel);
8952 vector bool short vec_vupkhsb (vector bool char);
8953 vector signed short vec_vupkhsb (vector signed char);
8955 vector signed short vec_unpackl (vector signed char);
8956 vector bool short vec_unpackl (vector bool char);
8957 vector unsigned int vec_unpackl (vector pixel);
8958 vector signed int vec_unpackl (vector signed short);
8959 vector bool int vec_unpackl (vector bool short);
8961 vector unsigned int vec_vupklpx (vector pixel);
8963 vector bool int vec_vupklsh (vector bool short);
8964 vector signed int vec_vupklsh (vector signed short);
8966 vector bool short vec_vupklsb (vector bool char);
8967 vector signed short vec_vupklsb (vector signed char);
8969 vector float vec_xor (vector float, vector float);
8970 vector float vec_xor (vector float, vector bool int);
8971 vector float vec_xor (vector bool int, vector float);
8972 vector bool int vec_xor (vector bool int, vector bool int);
8973 vector signed int vec_xor (vector bool int, vector signed int);
8974 vector signed int vec_xor (vector signed int, vector bool int);
8975 vector signed int vec_xor (vector signed int, vector signed int);
8976 vector unsigned int vec_xor (vector bool int, vector unsigned int);
8977 vector unsigned int vec_xor (vector unsigned int, vector bool int);
8978 vector unsigned int vec_xor (vector unsigned int, vector unsigned int);
8979 vector bool short vec_xor (vector bool short, vector bool short);
8980 vector signed short vec_xor (vector bool short, vector signed short);
8981 vector signed short vec_xor (vector signed short, vector bool short);
8982 vector signed short vec_xor (vector signed short, vector signed short);
8983 vector unsigned short vec_xor (vector bool short,
8984 vector unsigned short);
8985 vector unsigned short vec_xor (vector unsigned short,
8987 vector unsigned short vec_xor (vector unsigned short,
8988 vector unsigned short);
8989 vector signed char vec_xor (vector bool char, vector signed char);
8990 vector bool char vec_xor (vector bool char, vector bool char);
8991 vector signed char vec_xor (vector signed char, vector bool char);
8992 vector signed char vec_xor (vector signed char, vector signed char);
8993 vector unsigned char vec_xor (vector bool char, vector unsigned char);
8994 vector unsigned char vec_xor (vector unsigned char, vector bool char);
8995 vector unsigned char vec_xor (vector unsigned char,
8996 vector unsigned char);
8998 int vec_all_eq (vector signed char, vector bool char);
8999 int vec_all_eq (vector signed char, vector signed char);
9000 int vec_all_eq (vector unsigned char, vector bool char);
9001 int vec_all_eq (vector unsigned char, vector unsigned char);
9002 int vec_all_eq (vector bool char, vector bool char);
9003 int vec_all_eq (vector bool char, vector unsigned char);
9004 int vec_all_eq (vector bool char, vector signed char);
9005 int vec_all_eq (vector signed short, vector bool short);
9006 int vec_all_eq (vector signed short, vector signed short);
9007 int vec_all_eq (vector unsigned short, vector bool short);
9008 int vec_all_eq (vector unsigned short, vector unsigned short);
9009 int vec_all_eq (vector bool short, vector bool short);
9010 int vec_all_eq (vector bool short, vector unsigned short);
9011 int vec_all_eq (vector bool short, vector signed short);
9012 int vec_all_eq (vector pixel, vector pixel);
9013 int vec_all_eq (vector signed int, vector bool int);
9014 int vec_all_eq (vector signed int, vector signed int);
9015 int vec_all_eq (vector unsigned int, vector bool int);
9016 int vec_all_eq (vector unsigned int, vector unsigned int);
9017 int vec_all_eq (vector bool int, vector bool int);
9018 int vec_all_eq (vector bool int, vector unsigned int);
9019 int vec_all_eq (vector bool int, vector signed int);
9020 int vec_all_eq (vector float, vector float);
9022 int vec_all_ge (vector bool char, vector unsigned char);
9023 int vec_all_ge (vector unsigned char, vector bool char);
9024 int vec_all_ge (vector unsigned char, vector unsigned char);
9025 int vec_all_ge (vector bool char, vector signed char);
9026 int vec_all_ge (vector signed char, vector bool char);
9027 int vec_all_ge (vector signed char, vector signed char);
9028 int vec_all_ge (vector bool short, vector unsigned short);
9029 int vec_all_ge (vector unsigned short, vector bool short);
9030 int vec_all_ge (vector unsigned short, vector unsigned short);
9031 int vec_all_ge (vector signed short, vector signed short);
9032 int vec_all_ge (vector bool short, vector signed short);
9033 int vec_all_ge (vector signed short, vector bool short);
9034 int vec_all_ge (vector bool int, vector unsigned int);
9035 int vec_all_ge (vector unsigned int, vector bool int);
9036 int vec_all_ge (vector unsigned int, vector unsigned int);
9037 int vec_all_ge (vector bool int, vector signed int);
9038 int vec_all_ge (vector signed int, vector bool int);
9039 int vec_all_ge (vector signed int, vector signed int);
9040 int vec_all_ge (vector float, vector float);
9042 int vec_all_gt (vector bool char, vector unsigned char);
9043 int vec_all_gt (vector unsigned char, vector bool char);
9044 int vec_all_gt (vector unsigned char, vector unsigned char);
9045 int vec_all_gt (vector bool char, vector signed char);
9046 int vec_all_gt (vector signed char, vector bool char);
9047 int vec_all_gt (vector signed char, vector signed char);
9048 int vec_all_gt (vector bool short, vector unsigned short);
9049 int vec_all_gt (vector unsigned short, vector bool short);
9050 int vec_all_gt (vector unsigned short, vector unsigned short);
9051 int vec_all_gt (vector bool short, vector signed short);
9052 int vec_all_gt (vector signed short, vector bool short);
9053 int vec_all_gt (vector signed short, vector signed short);
9054 int vec_all_gt (vector bool int, vector unsigned int);
9055 int vec_all_gt (vector unsigned int, vector bool int);
9056 int vec_all_gt (vector unsigned int, vector unsigned int);
9057 int vec_all_gt (vector bool int, vector signed int);
9058 int vec_all_gt (vector signed int, vector bool int);
9059 int vec_all_gt (vector signed int, vector signed int);
9060 int vec_all_gt (vector float, vector float);
9062 int vec_all_in (vector float, vector float);
9064 int vec_all_le (vector bool char, vector unsigned char);
9065 int vec_all_le (vector unsigned char, vector bool char);
9066 int vec_all_le (vector unsigned char, vector unsigned char);
9067 int vec_all_le (vector bool char, vector signed char);
9068 int vec_all_le (vector signed char, vector bool char);
9069 int vec_all_le (vector signed char, vector signed char);
9070 int vec_all_le (vector bool short, vector unsigned short);
9071 int vec_all_le (vector unsigned short, vector bool short);
9072 int vec_all_le (vector unsigned short, vector unsigned short);
9073 int vec_all_le (vector bool short, vector signed short);
9074 int vec_all_le (vector signed short, vector bool short);
9075 int vec_all_le (vector signed short, vector signed short);
9076 int vec_all_le (vector bool int, vector unsigned int);
9077 int vec_all_le (vector unsigned int, vector bool int);
9078 int vec_all_le (vector unsigned int, vector unsigned int);
9079 int vec_all_le (vector bool int, vector signed int);
9080 int vec_all_le (vector signed int, vector bool int);
9081 int vec_all_le (vector signed int, vector signed int);
9082 int vec_all_le (vector float, vector float);
9084 int vec_all_lt (vector bool char, vector unsigned char);
9085 int vec_all_lt (vector unsigned char, vector bool char);
9086 int vec_all_lt (vector unsigned char, vector unsigned char);
9087 int vec_all_lt (vector bool char, vector signed char);
9088 int vec_all_lt (vector signed char, vector bool char);
9089 int vec_all_lt (vector signed char, vector signed char);
9090 int vec_all_lt (vector bool short, vector unsigned short);
9091 int vec_all_lt (vector unsigned short, vector bool short);
9092 int vec_all_lt (vector unsigned short, vector unsigned short);
9093 int vec_all_lt (vector bool short, vector signed short);
9094 int vec_all_lt (vector signed short, vector bool short);
9095 int vec_all_lt (vector signed short, vector signed short);
9096 int vec_all_lt (vector bool int, vector unsigned int);
9097 int vec_all_lt (vector unsigned int, vector bool int);
9098 int vec_all_lt (vector unsigned int, vector unsigned int);
9099 int vec_all_lt (vector bool int, vector signed int);
9100 int vec_all_lt (vector signed int, vector bool int);
9101 int vec_all_lt (vector signed int, vector signed int);
9102 int vec_all_lt (vector float, vector float);
9104 int vec_all_nan (vector float);
9106 int vec_all_ne (vector signed char, vector bool char);
9107 int vec_all_ne (vector signed char, vector signed char);
9108 int vec_all_ne (vector unsigned char, vector bool char);
9109 int vec_all_ne (vector unsigned char, vector unsigned char);
9110 int vec_all_ne (vector bool char, vector bool char);
9111 int vec_all_ne (vector bool char, vector unsigned char);
9112 int vec_all_ne (vector bool char, vector signed char);
9113 int vec_all_ne (vector signed short, vector bool short);
9114 int vec_all_ne (vector signed short, vector signed short);
9115 int vec_all_ne (vector unsigned short, vector bool short);
9116 int vec_all_ne (vector unsigned short, vector unsigned short);
9117 int vec_all_ne (vector bool short, vector bool short);
9118 int vec_all_ne (vector bool short, vector unsigned short);
9119 int vec_all_ne (vector bool short, vector signed short);
9120 int vec_all_ne (vector pixel, vector pixel);
9121 int vec_all_ne (vector signed int, vector bool int);
9122 int vec_all_ne (vector signed int, vector signed int);
9123 int vec_all_ne (vector unsigned int, vector bool int);
9124 int vec_all_ne (vector unsigned int, vector unsigned int);
9125 int vec_all_ne (vector bool int, vector bool int);
9126 int vec_all_ne (vector bool int, vector unsigned int);
9127 int vec_all_ne (vector bool int, vector signed int);
9128 int vec_all_ne (vector float, vector float);
9130 int vec_all_nge (vector float, vector float);
9132 int vec_all_ngt (vector float, vector float);
9134 int vec_all_nle (vector float, vector float);
9136 int vec_all_nlt (vector float, vector float);
9138 int vec_all_numeric (vector float);
9140 int vec_any_eq (vector signed char, vector bool char);
9141 int vec_any_eq (vector signed char, vector signed char);
9142 int vec_any_eq (vector unsigned char, vector bool char);
9143 int vec_any_eq (vector unsigned char, vector unsigned char);
9144 int vec_any_eq (vector bool char, vector bool char);
9145 int vec_any_eq (vector bool char, vector unsigned char);
9146 int vec_any_eq (vector bool char, vector signed char);
9147 int vec_any_eq (vector signed short, vector bool short);
9148 int vec_any_eq (vector signed short, vector signed short);
9149 int vec_any_eq (vector unsigned short, vector bool short);
9150 int vec_any_eq (vector unsigned short, vector unsigned short);
9151 int vec_any_eq (vector bool short, vector bool short);
9152 int vec_any_eq (vector bool short, vector unsigned short);
9153 int vec_any_eq (vector bool short, vector signed short);
9154 int vec_any_eq (vector pixel, vector pixel);
9155 int vec_any_eq (vector signed int, vector bool int);
9156 int vec_any_eq (vector signed int, vector signed int);
9157 int vec_any_eq (vector unsigned int, vector bool int);
9158 int vec_any_eq (vector unsigned int, vector unsigned int);
9159 int vec_any_eq (vector bool int, vector bool int);
9160 int vec_any_eq (vector bool int, vector unsigned int);
9161 int vec_any_eq (vector bool int, vector signed int);
9162 int vec_any_eq (vector float, vector float);
9164 int vec_any_ge (vector signed char, vector bool char);
9165 int vec_any_ge (vector unsigned char, vector bool char);
9166 int vec_any_ge (vector unsigned char, vector unsigned char);
9167 int vec_any_ge (vector signed char, vector signed char);
9168 int vec_any_ge (vector bool char, vector unsigned char);
9169 int vec_any_ge (vector bool char, vector signed char);
9170 int vec_any_ge (vector unsigned short, vector bool short);
9171 int vec_any_ge (vector unsigned short, vector unsigned short);
9172 int vec_any_ge (vector signed short, vector signed short);
9173 int vec_any_ge (vector signed short, vector bool short);
9174 int vec_any_ge (vector bool short, vector unsigned short);
9175 int vec_any_ge (vector bool short, vector signed short);
9176 int vec_any_ge (vector signed int, vector bool int);
9177 int vec_any_ge (vector unsigned int, vector bool int);
9178 int vec_any_ge (vector unsigned int, vector unsigned int);
9179 int vec_any_ge (vector signed int, vector signed int);
9180 int vec_any_ge (vector bool int, vector unsigned int);
9181 int vec_any_ge (vector bool int, vector signed int);
9182 int vec_any_ge (vector float, vector float);
9184 int vec_any_gt (vector bool char, vector unsigned char);
9185 int vec_any_gt (vector unsigned char, vector bool char);
9186 int vec_any_gt (vector unsigned char, vector unsigned char);
9187 int vec_any_gt (vector bool char, vector signed char);
9188 int vec_any_gt (vector signed char, vector bool char);
9189 int vec_any_gt (vector signed char, vector signed char);
9190 int vec_any_gt (vector bool short, vector unsigned short);
9191 int vec_any_gt (vector unsigned short, vector bool short);
9192 int vec_any_gt (vector unsigned short, vector unsigned short);
9193 int vec_any_gt (vector bool short, vector signed short);
9194 int vec_any_gt (vector signed short, vector bool short);
9195 int vec_any_gt (vector signed short, vector signed short);
9196 int vec_any_gt (vector bool int, vector unsigned int);
9197 int vec_any_gt (vector unsigned int, vector bool int);
9198 int vec_any_gt (vector unsigned int, vector unsigned int);
9199 int vec_any_gt (vector bool int, vector signed int);
9200 int vec_any_gt (vector signed int, vector bool int);
9201 int vec_any_gt (vector signed int, vector signed int);
9202 int vec_any_gt (vector float, vector float);
9204 int vec_any_le (vector bool char, vector unsigned char);
9205 int vec_any_le (vector unsigned char, vector bool char);
9206 int vec_any_le (vector unsigned char, vector unsigned char);
9207 int vec_any_le (vector bool char, vector signed char);
9208 int vec_any_le (vector signed char, vector bool char);
9209 int vec_any_le (vector signed char, vector signed char);
9210 int vec_any_le (vector bool short, vector unsigned short);
9211 int vec_any_le (vector unsigned short, vector bool short);
9212 int vec_any_le (vector unsigned short, vector unsigned short);
9213 int vec_any_le (vector bool short, vector signed short);
9214 int vec_any_le (vector signed short, vector bool short);
9215 int vec_any_le (vector signed short, vector signed short);
9216 int vec_any_le (vector bool int, vector unsigned int);
9217 int vec_any_le (vector unsigned int, vector bool int);
9218 int vec_any_le (vector unsigned int, vector unsigned int);
9219 int vec_any_le (vector bool int, vector signed int);
9220 int vec_any_le (vector signed int, vector bool int);
9221 int vec_any_le (vector signed int, vector signed int);
9222 int vec_any_le (vector float, vector float);
9224 int vec_any_lt (vector bool char, vector unsigned char);
9225 int vec_any_lt (vector unsigned char, vector bool char);
9226 int vec_any_lt (vector unsigned char, vector unsigned char);
9227 int vec_any_lt (vector bool char, vector signed char);
9228 int vec_any_lt (vector signed char, vector bool char);
9229 int vec_any_lt (vector signed char, vector signed char);
9230 int vec_any_lt (vector bool short, vector unsigned short);
9231 int vec_any_lt (vector unsigned short, vector bool short);
9232 int vec_any_lt (vector unsigned short, vector unsigned short);
9233 int vec_any_lt (vector bool short, vector signed short);
9234 int vec_any_lt (vector signed short, vector bool short);
9235 int vec_any_lt (vector signed short, vector signed short);
9236 int vec_any_lt (vector bool int, vector unsigned int);
9237 int vec_any_lt (vector unsigned int, vector bool int);
9238 int vec_any_lt (vector unsigned int, vector unsigned int);
9239 int vec_any_lt (vector bool int, vector signed int);
9240 int vec_any_lt (vector signed int, vector bool int);
9241 int vec_any_lt (vector signed int, vector signed int);
9242 int vec_any_lt (vector float, vector float);
9244 int vec_any_nan (vector float);
9246 int vec_any_ne (vector signed char, vector bool char);
9247 int vec_any_ne (vector signed char, vector signed char);
9248 int vec_any_ne (vector unsigned char, vector bool char);
9249 int vec_any_ne (vector unsigned char, vector unsigned char);
9250 int vec_any_ne (vector bool char, vector bool char);
9251 int vec_any_ne (vector bool char, vector unsigned char);
9252 int vec_any_ne (vector bool char, vector signed char);
9253 int vec_any_ne (vector signed short, vector bool short);
9254 int vec_any_ne (vector signed short, vector signed short);
9255 int vec_any_ne (vector unsigned short, vector bool short);
9256 int vec_any_ne (vector unsigned short, vector unsigned short);
9257 int vec_any_ne (vector bool short, vector bool short);
9258 int vec_any_ne (vector bool short, vector unsigned short);
9259 int vec_any_ne (vector bool short, vector signed short);
9260 int vec_any_ne (vector pixel, vector pixel);
9261 int vec_any_ne (vector signed int, vector bool int);
9262 int vec_any_ne (vector signed int, vector signed int);
9263 int vec_any_ne (vector unsigned int, vector bool int);
9264 int vec_any_ne (vector unsigned int, vector unsigned int);
9265 int vec_any_ne (vector bool int, vector bool int);
9266 int vec_any_ne (vector bool int, vector unsigned int);
9267 int vec_any_ne (vector bool int, vector signed int);
9268 int vec_any_ne (vector float, vector float);
9270 int vec_any_nge (vector float, vector float);
9272 int vec_any_ngt (vector float, vector float);
9274 int vec_any_nle (vector float, vector float);
9276 int vec_any_nlt (vector float, vector float);
9278 int vec_any_numeric (vector float);
9280 int vec_any_out (vector float, vector float);
9283 @node SPARC VIS Built-in Functions
9284 @subsection SPARC VIS Built-in Functions
9286 GCC supports SIMD operations on the SPARC using both the generic vector
9287 extensions (@pxref{Vector Extensions}) as well as built-in functions for
9288 the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis}
9289 switch, the VIS extension is exposed as the following built-in functions:
9292 typedef int v2si __attribute__ ((vector_size (8)));
9293 typedef short v4hi __attribute__ ((vector_size (8)));
9294 typedef short v2hi __attribute__ ((vector_size (4)));
9295 typedef char v8qi __attribute__ ((vector_size (8)));
9296 typedef char v4qi __attribute__ ((vector_size (4)));
9298 void * __builtin_vis_alignaddr (void *, long);
9299 int64_t __builtin_vis_faligndatadi (int64_t, int64_t);
9300 v2si __builtin_vis_faligndatav2si (v2si, v2si);
9301 v4hi __builtin_vis_faligndatav4hi (v4si, v4si);
9302 v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi);
9304 v4hi __builtin_vis_fexpand (v4qi);
9306 v4hi __builtin_vis_fmul8x16 (v4qi, v4hi);
9307 v4hi __builtin_vis_fmul8x16au (v4qi, v4hi);
9308 v4hi __builtin_vis_fmul8x16al (v4qi, v4hi);
9309 v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi);
9310 v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi);
9311 v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi);
9312 v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi);
9314 v4qi __builtin_vis_fpack16 (v4hi);
9315 v8qi __builtin_vis_fpack32 (v2si, v2si);
9316 v2hi __builtin_vis_fpackfix (v2si);
9317 v8qi __builtin_vis_fpmerge (v4qi, v4qi);
9319 int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t);
9322 @node Target Format Checks
9323 @section Format Checks Specific to Particular Target Machines
9325 For some target machines, GCC supports additional options to the
9327 (@pxref{Function Attributes,,Declaring Attributes of Functions}).
9330 * Solaris Format Checks::
9333 @node Solaris Format Checks
9334 @subsection Solaris Format Checks
9336 Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format
9337 check. @code{cmn_err} accepts a subset of the standard @code{printf}
9338 conversions, and the two-argument @code{%b} conversion for displaying
9339 bit-fields. See the Solaris man page for @code{cmn_err} for more information.
9342 @section Pragmas Accepted by GCC
9346 GCC supports several types of pragmas, primarily in order to compile
9347 code originally written for other compilers. Note that in general
9348 we do not recommend the use of pragmas; @xref{Function Attributes},
9349 for further explanation.
9354 * RS/6000 and PowerPC Pragmas::
9357 * Symbol-Renaming Pragmas::
9358 * Structure-Packing Pragmas::
9360 * Diagnostic Pragmas::
9364 @subsection ARM Pragmas
9366 The ARM target defines pragmas for controlling the default addition of
9367 @code{long_call} and @code{short_call} attributes to functions.
9368 @xref{Function Attributes}, for information about the effects of these
9373 @cindex pragma, long_calls
9374 Set all subsequent functions to have the @code{long_call} attribute.
9377 @cindex pragma, no_long_calls
9378 Set all subsequent functions to have the @code{short_call} attribute.
9380 @item long_calls_off
9381 @cindex pragma, long_calls_off
9382 Do not affect the @code{long_call} or @code{short_call} attributes of
9383 subsequent functions.
9387 @subsection M32C Pragmas
9390 @item memregs @var{number}
9391 @cindex pragma, memregs
9392 Overrides the command line option @code{-memregs=} for the current
9393 file. Use with care! This pragma must be before any function in the
9394 file, and mixing different memregs values in different objects may
9395 make them incompatible. This pragma is useful when a
9396 performance-critical function uses a memreg for temporary values,
9397 as it may allow you to reduce the number of memregs used.
9401 @node RS/6000 and PowerPC Pragmas
9402 @subsection RS/6000 and PowerPC Pragmas
9404 The RS/6000 and PowerPC targets define one pragma for controlling
9405 whether or not the @code{longcall} attribute is added to function
9406 declarations by default. This pragma overrides the @option{-mlongcall}
9407 option, but not the @code{longcall} and @code{shortcall} attributes.
9408 @xref{RS/6000 and PowerPC Options}, for more information about when long
9409 calls are and are not necessary.
9413 @cindex pragma, longcall
9414 Apply the @code{longcall} attribute to all subsequent function
9418 Do not apply the @code{longcall} attribute to subsequent function
9422 @c Describe c4x pragmas here.
9423 @c Describe h8300 pragmas here.
9424 @c Describe sh pragmas here.
9425 @c Describe v850 pragmas here.
9427 @node Darwin Pragmas
9428 @subsection Darwin Pragmas
9430 The following pragmas are available for all architectures running the
9431 Darwin operating system. These are useful for compatibility with other
9435 @item mark @var{tokens}@dots{}
9436 @cindex pragma, mark
9437 This pragma is accepted, but has no effect.
9439 @item options align=@var{alignment}
9440 @cindex pragma, options align
9441 This pragma sets the alignment of fields in structures. The values of
9442 @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or
9443 @code{power}, to emulate PowerPC alignment. Uses of this pragma nest
9444 properly; to restore the previous setting, use @code{reset} for the
9447 @item segment @var{tokens}@dots{}
9448 @cindex pragma, segment
9449 This pragma is accepted, but has no effect.
9451 @item unused (@var{var} [, @var{var}]@dots{})
9452 @cindex pragma, unused
9453 This pragma declares variables to be possibly unused. GCC will not
9454 produce warnings for the listed variables. The effect is similar to
9455 that of the @code{unused} attribute, except that this pragma may appear
9456 anywhere within the variables' scopes.
9459 @node Solaris Pragmas
9460 @subsection Solaris Pragmas
9462 The Solaris target supports @code{#pragma redefine_extname}
9463 (@pxref{Symbol-Renaming Pragmas}). It also supports additional
9464 @code{#pragma} directives for compatibility with the system compiler.
9467 @item align @var{alignment} (@var{variable} [, @var{variable}]...)
9468 @cindex pragma, align
9470 Increase the minimum alignment of each @var{variable} to @var{alignment}.
9471 This is the same as GCC's @code{aligned} attribute @pxref{Variable
9472 Attributes}). Macro expansion occurs on the arguments to this pragma
9473 when compiling C and Objective-C. It does not currently occur when
9474 compiling C++, but this is a bug which may be fixed in a future
9477 @item fini (@var{function} [, @var{function}]...)
9478 @cindex pragma, fini
9480 This pragma causes each listed @var{function} to be called after
9481 main, or during shared module unloading, by adding a call to the
9482 @code{.fini} section.
9484 @item init (@var{function} [, @var{function}]...)
9485 @cindex pragma, init
9487 This pragma causes each listed @var{function} to be called during
9488 initialization (before @code{main}) or during shared module loading, by
9489 adding a call to the @code{.init} section.
9493 @node Symbol-Renaming Pragmas
9494 @subsection Symbol-Renaming Pragmas
9496 For compatibility with the Solaris and Tru64 UNIX system headers, GCC
9497 supports two @code{#pragma} directives which change the name used in
9498 assembly for a given declaration. These pragmas are only available on
9499 platforms whose system headers need them. To get this effect on all
9500 platforms supported by GCC, use the asm labels extension (@pxref{Asm
9504 @item redefine_extname @var{oldname} @var{newname}
9505 @cindex pragma, redefine_extname
9507 This pragma gives the C function @var{oldname} the assembly symbol
9508 @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME}
9509 will be defined if this pragma is available (currently only on
9512 @item extern_prefix @var{string}
9513 @cindex pragma, extern_prefix
9515 This pragma causes all subsequent external function and variable
9516 declarations to have @var{string} prepended to their assembly symbols.
9517 This effect may be terminated with another @code{extern_prefix} pragma
9518 whose argument is an empty string. The preprocessor macro
9519 @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is
9520 available (currently only on Tru64 UNIX)@.
9523 These pragmas and the asm labels extension interact in a complicated
9524 manner. Here are some corner cases you may want to be aware of.
9527 @item Both pragmas silently apply only to declarations with external
9528 linkage. Asm labels do not have this restriction.
9530 @item In C++, both pragmas silently apply only to declarations with
9531 ``C'' linkage. Again, asm labels do not have this restriction.
9533 @item If any of the three ways of changing the assembly name of a
9534 declaration is applied to a declaration whose assembly name has
9535 already been determined (either by a previous use of one of these
9536 features, or because the compiler needed the assembly name in order to
9537 generate code), and the new name is different, a warning issues and
9538 the name does not change.
9540 @item The @var{oldname} used by @code{#pragma redefine_extname} is
9541 always the C-language name.
9543 @item If @code{#pragma extern_prefix} is in effect, and a declaration
9544 occurs with an asm label attached, the prefix is silently ignored for
9547 @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname}
9548 apply to the same declaration, whichever triggered first wins, and a
9549 warning issues if they contradict each other. (We would like to have
9550 @code{#pragma redefine_extname} always win, for consistency with asm
9551 labels, but if @code{#pragma extern_prefix} triggers first we have no
9552 way of knowing that that happened.)
9555 @node Structure-Packing Pragmas
9556 @subsection Structure-Packing Pragmas
9558 For compatibility with Win32, GCC supports a set of @code{#pragma}
9559 directives which change the maximum alignment of members of structures
9560 (other than zero-width bitfields), unions, and classes subsequently
9561 defined. The @var{n} value below always is required to be a small power
9562 of two and specifies the new alignment in bytes.
9565 @item @code{#pragma pack(@var{n})} simply sets the new alignment.
9566 @item @code{#pragma pack()} sets the alignment to the one that was in
9567 effect when compilation started (see also command line option
9568 @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}).
9569 @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment
9570 setting on an internal stack and then optionally sets the new alignment.
9571 @item @code{#pragma pack(pop)} restores the alignment setting to the one
9572 saved at the top of the internal stack (and removes that stack entry).
9573 Note that @code{#pragma pack([@var{n}])} does not influence this internal
9574 stack; thus it is possible to have @code{#pragma pack(push)} followed by
9575 multiple @code{#pragma pack(@var{n})} instances and finalized by a single
9576 @code{#pragma pack(pop)}.
9580 @subsection Weak Pragmas
9582 For compatibility with SVR4, GCC supports a set of @code{#pragma}
9583 directives for declaring symbols to be weak, and defining weak
9587 @item #pragma weak @var{symbol}
9588 @cindex pragma, weak
9589 This pragma declares @var{symbol} to be weak, as if the declaration
9590 had the attribute of the same name. The pragma may appear before
9591 or after the declaration of @var{symbol}, but must appear before
9592 either its first use or its definition. It is not an error for
9593 @var{symbol} to never be defined at all.
9595 @item #pragma weak @var{symbol1} = @var{symbol2}
9596 This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}.
9597 It is an error if @var{symbol2} is not defined in the current
9601 @node Diagnostic Pragmas
9602 @subsection Diagnostic Pragmas
9604 GCC allows the user to selectively enable or disable certain types of
9605 diagnostics, and change the kind of the diagnostic. For example, a
9606 project's policy might require that all sources compile with
9607 @option{-Werror} but certain files might have exceptions allowing
9608 specific types of warnings. Or, a project might selectively enable
9609 diagnostics and treat them as errors depending on which preprocessor
9613 @item #pragma GCC diagnostic @var{kind} @var{option}
9614 @cindex pragma, diagnostic
9616 Modifies the disposition of a diagnostic. Note that not all
9617 diagnostics are modifyiable; at the moment only warnings (normally
9618 controlled by @samp{-W...}) can be controlled, and not all of them.
9619 Use @option{-fdiagnostics-show-option} to determine which diagnostics
9620 are controllable and which option controls them.
9622 @var{kind} is @samp{error} to treat this diagnostic as an error,
9623 @samp{warning} to treat it like a warning (even if @option{-Werror} is
9624 in effect), or @samp{ignored} if the diagnostic is to be ignored.
9625 @var{option} is a double quoted string which matches the command line
9629 #pragma GCC diagnostic warning "-Wformat"
9630 #pragma GCC diagnostic error "-Walways-true"
9631 #pragma GCC diagnostic ignored "-Walways-true"
9634 Note that these pragmas override any command line options. Also,
9635 while it is syntactically valid to put these pragmas anywhere in your
9636 sources, the only supported location for them is before any data or
9637 functions are defined. Doing otherwise may result in unpredictable
9638 results depending on how the optimizer manages your sources. If the
9639 same option is listed multiple times, the last one specified is the
9640 one that is in effect. This pragma is not intended to be a general
9641 purpose replacement for command line options, but for implementing
9642 strict control over project policies.
9646 @node Unnamed Fields
9647 @section Unnamed struct/union fields within structs/unions
9651 For compatibility with other compilers, GCC allows you to define
9652 a structure or union that contains, as fields, structures and unions
9653 without names. For example:
9666 In this example, the user would be able to access members of the unnamed
9667 union with code like @samp{foo.b}. Note that only unnamed structs and
9668 unions are allowed, you may not have, for example, an unnamed
9671 You must never create such structures that cause ambiguous field definitions.
9672 For example, this structure:
9683 It is ambiguous which @code{a} is being referred to with @samp{foo.a}.
9684 Such constructs are not supported and must be avoided. In the future,
9685 such constructs may be detected and treated as compilation errors.
9687 @opindex fms-extensions
9688 Unless @option{-fms-extensions} is used, the unnamed field must be a
9689 structure or union definition without a tag (for example, @samp{struct
9690 @{ int a; @};}). If @option{-fms-extensions} is used, the field may
9691 also be a definition with a tag such as @samp{struct foo @{ int a;
9692 @};}, a reference to a previously defined structure or union such as
9693 @samp{struct foo;}, or a reference to a @code{typedef} name for a
9694 previously defined structure or union type.
9697 @section Thread-Local Storage
9698 @cindex Thread-Local Storage
9699 @cindex @acronym{TLS}
9702 Thread-local storage (@acronym{TLS}) is a mechanism by which variables
9703 are allocated such that there is one instance of the variable per extant
9704 thread. The run-time model GCC uses to implement this originates
9705 in the IA-64 processor-specific ABI, but has since been migrated
9706 to other processors as well. It requires significant support from
9707 the linker (@command{ld}), dynamic linker (@command{ld.so}), and
9708 system libraries (@file{libc.so} and @file{libpthread.so}), so it
9709 is not available everywhere.
9711 At the user level, the extension is visible with a new storage
9712 class keyword: @code{__thread}. For example:
9716 extern __thread struct state s;
9717 static __thread char *p;
9720 The @code{__thread} specifier may be used alone, with the @code{extern}
9721 or @code{static} specifiers, but with no other storage class specifier.
9722 When used with @code{extern} or @code{static}, @code{__thread} must appear
9723 immediately after the other storage class specifier.
9725 The @code{__thread} specifier may be applied to any global, file-scoped
9726 static, function-scoped static, or static data member of a class. It may
9727 not be applied to block-scoped automatic or non-static data member.
9729 When the address-of operator is applied to a thread-local variable, it is
9730 evaluated at run-time and returns the address of the current thread's
9731 instance of that variable. An address so obtained may be used by any
9732 thread. When a thread terminates, any pointers to thread-local variables
9733 in that thread become invalid.
9735 No static initialization may refer to the address of a thread-local variable.
9737 In C++, if an initializer is present for a thread-local variable, it must
9738 be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++
9741 See @uref{http://people.redhat.com/drepper/tls.pdf,
9742 ELF Handling For Thread-Local Storage} for a detailed explanation of
9743 the four thread-local storage addressing models, and how the run-time
9744 is expected to function.
9747 * C99 Thread-Local Edits::
9748 * C++98 Thread-Local Edits::
9751 @node C99 Thread-Local Edits
9752 @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage
9754 The following are a set of changes to ISO/IEC 9899:1999 (aka C99)
9755 that document the exact semantics of the language extension.
9759 @cite{5.1.2 Execution environments}
9761 Add new text after paragraph 1
9764 Within either execution environment, a @dfn{thread} is a flow of
9765 control within a program. It is implementation defined whether
9766 or not there may be more than one thread associated with a program.
9767 It is implementation defined how threads beyond the first are
9768 created, the name and type of the function called at thread
9769 startup, and how threads may be terminated. However, objects
9770 with thread storage duration shall be initialized before thread
9775 @cite{6.2.4 Storage durations of objects}
9777 Add new text before paragraph 3
9780 An object whose identifier is declared with the storage-class
9781 specifier @w{@code{__thread}} has @dfn{thread storage duration}.
9782 Its lifetime is the entire execution of the thread, and its
9783 stored value is initialized only once, prior to thread startup.
9787 @cite{6.4.1 Keywords}
9789 Add @code{__thread}.
9792 @cite{6.7.1 Storage-class specifiers}
9794 Add @code{__thread} to the list of storage class specifiers in
9797 Change paragraph 2 to
9800 With the exception of @code{__thread}, at most one storage-class
9801 specifier may be given [@dots{}]. The @code{__thread} specifier may
9802 be used alone, or immediately following @code{extern} or
9806 Add new text after paragraph 6
9809 The declaration of an identifier for a variable that has
9810 block scope that specifies @code{__thread} shall also
9811 specify either @code{extern} or @code{static}.
9813 The @code{__thread} specifier shall be used only with
9818 @node C++98 Thread-Local Edits
9819 @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage
9821 The following are a set of changes to ISO/IEC 14882:1998 (aka C++98)
9822 that document the exact semantics of the language extension.
9826 @b{[intro.execution]}
9828 New text after paragraph 4
9831 A @dfn{thread} is a flow of control within the abstract machine.
9832 It is implementation defined whether or not there may be more than
9836 New text after paragraph 7
9839 It is unspecified whether additional action must be taken to
9840 ensure when and whether side effects are visible to other threads.
9846 Add @code{__thread}.
9849 @b{[basic.start.main]}
9851 Add after paragraph 5
9854 The thread that begins execution at the @code{main} function is called
9855 the @dfn{main thread}. It is implementation defined how functions
9856 beginning threads other than the main thread are designated or typed.
9857 A function so designated, as well as the @code{main} function, is called
9858 a @dfn{thread startup function}. It is implementation defined what
9859 happens if a thread startup function returns. It is implementation
9860 defined what happens to other threads when any thread calls @code{exit}.
9864 @b{[basic.start.init]}
9866 Add after paragraph 4
9869 The storage for an object of thread storage duration shall be
9870 statically initialized before the first statement of the thread startup
9871 function. An object of thread storage duration shall not require
9872 dynamic initialization.
9876 @b{[basic.start.term]}
9878 Add after paragraph 3
9881 The type of an object with thread storage duration shall not have a
9882 non-trivial destructor, nor shall it be an array type whose elements
9883 (directly or indirectly) have non-trivial destructors.
9889 Add ``thread storage duration'' to the list in paragraph 1.
9894 Thread, static, and automatic storage durations are associated with
9895 objects introduced by declarations [@dots{}].
9898 Add @code{__thread} to the list of specifiers in paragraph 3.
9901 @b{[basic.stc.thread]}
9903 New section before @b{[basic.stc.static]}
9906 The keyword @code{__thread} applied to a non-local object gives the
9907 object thread storage duration.
9909 A local variable or class data member declared both @code{static}
9910 and @code{__thread} gives the variable or member thread storage
9915 @b{[basic.stc.static]}
9920 All objects which have neither thread storage duration, dynamic
9921 storage duration nor are local [@dots{}].
9927 Add @code{__thread} to the list in paragraph 1.
9932 With the exception of @code{__thread}, at most one
9933 @var{storage-class-specifier} shall appear in a given
9934 @var{decl-specifier-seq}. The @code{__thread} specifier may
9935 be used alone, or immediately following the @code{extern} or
9936 @code{static} specifiers. [@dots{}]
9939 Add after paragraph 5
9942 The @code{__thread} specifier can be applied only to the names of objects
9943 and to anonymous unions.
9949 Add after paragraph 6
9952 Non-@code{static} members shall not be @code{__thread}.
9956 @node C++ Extensions
9957 @chapter Extensions to the C++ Language
9958 @cindex extensions, C++ language
9959 @cindex C++ language extensions
9961 The GNU compiler provides these extensions to the C++ language (and you
9962 can also use most of the C language extensions in your C++ programs). If you
9963 want to write code that checks whether these features are available, you can
9964 test for the GNU compiler the same way as for C programs: check for a
9965 predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to
9966 test specifically for GNU C++ (@pxref{Common Predefined Macros,,
9967 Predefined Macros,cpp,The GNU C Preprocessor}).
9970 * Volatiles:: What constitutes an access to a volatile object.
9971 * Restricted Pointers:: C99 restricted pointers and references.
9972 * Vague Linkage:: Where G++ puts inlines, vtables and such.
9973 * C++ Interface:: You can use a single C++ header file for both
9974 declarations and definitions.
9975 * Template Instantiation:: Methods for ensuring that exactly one copy of
9976 each needed template instantiation is emitted.
9977 * Bound member functions:: You can extract a function pointer to the
9978 method denoted by a @samp{->*} or @samp{.*} expression.
9979 * C++ Attributes:: Variable, function, and type attributes for C++ only.
9980 * Namespace Association:: Strong using-directives for namespace association.
9981 * Java Exceptions:: Tweaking exception handling to work with Java.
9982 * Deprecated Features:: Things will disappear from g++.
9983 * Backwards Compatibility:: Compatibilities with earlier definitions of C++.
9987 @section When is a Volatile Object Accessed?
9988 @cindex accessing volatiles
9989 @cindex volatile read
9990 @cindex volatile write
9991 @cindex volatile access
9993 Both the C and C++ standard have the concept of volatile objects. These
9994 are normally accessed by pointers and used for accessing hardware. The
9995 standards encourage compilers to refrain from optimizations
9996 concerning accesses to volatile objects that it might perform on
9997 non-volatile objects. The C standard leaves it implementation defined
9998 as to what constitutes a volatile access. The C++ standard omits to
9999 specify this, except to say that C++ should behave in a similar manner
10000 to C with respect to volatiles, where possible. The minimum either
10001 standard specifies is that at a sequence point all previous accesses to
10002 volatile objects have stabilized and no subsequent accesses have
10003 occurred. Thus an implementation is free to reorder and combine
10004 volatile accesses which occur between sequence points, but cannot do so
10005 for accesses across a sequence point. The use of volatiles does not
10006 allow you to violate the restriction on updating objects multiple times
10007 within a sequence point.
10009 In most expressions, it is intuitively obvious what is a read and what is
10010 a write. For instance
10013 volatile int *dst = @var{somevalue};
10014 volatile int *src = @var{someothervalue};
10019 will cause a read of the volatile object pointed to by @var{src} and stores the
10020 value into the volatile object pointed to by @var{dst}. There is no
10021 guarantee that these reads and writes are atomic, especially for objects
10022 larger than @code{int}.
10024 Less obvious expressions are where something which looks like an access
10025 is used in a void context. An example would be,
10028 volatile int *src = @var{somevalue};
10032 With C, such expressions are rvalues, and as rvalues cause a read of
10033 the object, GCC interprets this as a read of the volatile being pointed
10034 to. The C++ standard specifies that such expressions do not undergo
10035 lvalue to rvalue conversion, and that the type of the dereferenced
10036 object may be incomplete. The C++ standard does not specify explicitly
10037 that it is this lvalue to rvalue conversion which is responsible for
10038 causing an access. However, there is reason to believe that it is,
10039 because otherwise certain simple expressions become undefined. However,
10040 because it would surprise most programmers, G++ treats dereferencing a
10041 pointer to volatile object of complete type in a void context as a read
10042 of the object. When the object has incomplete type, G++ issues a
10047 struct T @{int m;@};
10048 volatile S *ptr1 = @var{somevalue};
10049 volatile T *ptr2 = @var{somevalue};
10054 In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2}
10055 causes a read of the object pointed to. If you wish to force an error on
10056 the first case, you must force a conversion to rvalue with, for instance
10057 a static cast, @code{static_cast<S>(*ptr1)}.
10059 When using a reference to volatile, G++ does not treat equivalent
10060 expressions as accesses to volatiles, but instead issues a warning that
10061 no volatile is accessed. The rationale for this is that otherwise it
10062 becomes difficult to determine where volatile access occur, and not
10063 possible to ignore the return value from functions returning volatile
10064 references. Again, if you wish to force a read, cast the reference to
10067 @node Restricted Pointers
10068 @section Restricting Pointer Aliasing
10069 @cindex restricted pointers
10070 @cindex restricted references
10071 @cindex restricted this pointer
10073 As with the C front end, G++ understands the C99 feature of restricted pointers,
10074 specified with the @code{__restrict__}, or @code{__restrict} type
10075 qualifier. Because you cannot compile C++ by specifying the @option{-std=c99}
10076 language flag, @code{restrict} is not a keyword in C++.
10078 In addition to allowing restricted pointers, you can specify restricted
10079 references, which indicate that the reference is not aliased in the local
10083 void fn (int *__restrict__ rptr, int &__restrict__ rref)
10090 In the body of @code{fn}, @var{rptr} points to an unaliased integer and
10091 @var{rref} refers to a (different) unaliased integer.
10093 You may also specify whether a member function's @var{this} pointer is
10094 unaliased by using @code{__restrict__} as a member function qualifier.
10097 void T::fn () __restrict__
10104 Within the body of @code{T::fn}, @var{this} will have the effective
10105 definition @code{T *__restrict__ const this}. Notice that the
10106 interpretation of a @code{__restrict__} member function qualifier is
10107 different to that of @code{const} or @code{volatile} qualifier, in that it
10108 is applied to the pointer rather than the object. This is consistent with
10109 other compilers which implement restricted pointers.
10111 As with all outermost parameter qualifiers, @code{__restrict__} is
10112 ignored in function definition matching. This means you only need to
10113 specify @code{__restrict__} in a function definition, rather than
10114 in a function prototype as well.
10116 @node Vague Linkage
10117 @section Vague Linkage
10118 @cindex vague linkage
10120 There are several constructs in C++ which require space in the object
10121 file but are not clearly tied to a single translation unit. We say that
10122 these constructs have ``vague linkage''. Typically such constructs are
10123 emitted wherever they are needed, though sometimes we can be more
10127 @item Inline Functions
10128 Inline functions are typically defined in a header file which can be
10129 included in many different compilations. Hopefully they can usually be
10130 inlined, but sometimes an out-of-line copy is necessary, if the address
10131 of the function is taken or if inlining fails. In general, we emit an
10132 out-of-line copy in all translation units where one is needed. As an
10133 exception, we only emit inline virtual functions with the vtable, since
10134 it will always require a copy.
10136 Local static variables and string constants used in an inline function
10137 are also considered to have vague linkage, since they must be shared
10138 between all inlined and out-of-line instances of the function.
10142 C++ virtual functions are implemented in most compilers using a lookup
10143 table, known as a vtable. The vtable contains pointers to the virtual
10144 functions provided by a class, and each object of the class contains a
10145 pointer to its vtable (or vtables, in some multiple-inheritance
10146 situations). If the class declares any non-inline, non-pure virtual
10147 functions, the first one is chosen as the ``key method'' for the class,
10148 and the vtable is only emitted in the translation unit where the key
10151 @emph{Note:} If the chosen key method is later defined as inline, the
10152 vtable will still be emitted in every translation unit which defines it.
10153 Make sure that any inline virtuals are declared inline in the class
10154 body, even if they are not defined there.
10156 @item type_info objects
10159 C++ requires information about types to be written out in order to
10160 implement @samp{dynamic_cast}, @samp{typeid} and exception handling.
10161 For polymorphic classes (classes with virtual functions), the type_info
10162 object is written out along with the vtable so that @samp{dynamic_cast}
10163 can determine the dynamic type of a class object at runtime. For all
10164 other types, we write out the type_info object when it is used: when
10165 applying @samp{typeid} to an expression, throwing an object, or
10166 referring to a type in a catch clause or exception specification.
10168 @item Template Instantiations
10169 Most everything in this section also applies to template instantiations,
10170 but there are other options as well.
10171 @xref{Template Instantiation,,Where's the Template?}.
10175 When used with GNU ld version 2.8 or later on an ELF system such as
10176 GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of
10177 these constructs will be discarded at link time. This is known as
10180 On targets that don't support COMDAT, but do support weak symbols, GCC
10181 will use them. This way one copy will override all the others, but
10182 the unused copies will still take up space in the executable.
10184 For targets which do not support either COMDAT or weak symbols,
10185 most entities with vague linkage will be emitted as local symbols to
10186 avoid duplicate definition errors from the linker. This will not happen
10187 for local statics in inlines, however, as having multiple copies will
10188 almost certainly break things.
10190 @xref{C++ Interface,,Declarations and Definitions in One Header}, for
10191 another way to control placement of these constructs.
10193 @node C++ Interface
10194 @section #pragma interface and implementation
10196 @cindex interface and implementation headers, C++
10197 @cindex C++ interface and implementation headers
10198 @cindex pragmas, interface and implementation
10200 @code{#pragma interface} and @code{#pragma implementation} provide the
10201 user with a way of explicitly directing the compiler to emit entities
10202 with vague linkage (and debugging information) in a particular
10205 @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in
10206 most cases, because of COMDAT support and the ``key method'' heuristic
10207 mentioned in @ref{Vague Linkage}. Using them can actually cause your
10208 program to grow due to unnecessary out-of-line copies of inline
10209 functions. Currently (3.4) the only benefit of these
10210 @code{#pragma}s is reduced duplication of debugging information, and
10211 that should be addressed soon on DWARF 2 targets with the use of
10215 @item #pragma interface
10216 @itemx #pragma interface "@var{subdir}/@var{objects}.h"
10217 @kindex #pragma interface
10218 Use this directive in @emph{header files} that define object classes, to save
10219 space in most of the object files that use those classes. Normally,
10220 local copies of certain information (backup copies of inline member
10221 functions, debugging information, and the internal tables that implement
10222 virtual functions) must be kept in each object file that includes class
10223 definitions. You can use this pragma to avoid such duplication. When a
10224 header file containing @samp{#pragma interface} is included in a
10225 compilation, this auxiliary information will not be generated (unless
10226 the main input source file itself uses @samp{#pragma implementation}).
10227 Instead, the object files will contain references to be resolved at link
10230 The second form of this directive is useful for the case where you have
10231 multiple headers with the same name in different directories. If you
10232 use this form, you must specify the same string to @samp{#pragma
10235 @item #pragma implementation
10236 @itemx #pragma implementation "@var{objects}.h"
10237 @kindex #pragma implementation
10238 Use this pragma in a @emph{main input file}, when you want full output from
10239 included header files to be generated (and made globally visible). The
10240 included header file, in turn, should use @samp{#pragma interface}.
10241 Backup copies of inline member functions, debugging information, and the
10242 internal tables used to implement virtual functions are all generated in
10243 implementation files.
10245 @cindex implied @code{#pragma implementation}
10246 @cindex @code{#pragma implementation}, implied
10247 @cindex naming convention, implementation headers
10248 If you use @samp{#pragma implementation} with no argument, it applies to
10249 an include file with the same basename@footnote{A file's @dfn{basename}
10250 was the name stripped of all leading path information and of trailing
10251 suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source
10252 file. For example, in @file{allclass.cc}, giving just
10253 @samp{#pragma implementation}
10254 by itself is equivalent to @samp{#pragma implementation "allclass.h"}.
10256 In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as
10257 an implementation file whenever you would include it from
10258 @file{allclass.cc} even if you never specified @samp{#pragma
10259 implementation}. This was deemed to be more trouble than it was worth,
10260 however, and disabled.
10262 Use the string argument if you want a single implementation file to
10263 include code from multiple header files. (You must also use
10264 @samp{#include} to include the header file; @samp{#pragma
10265 implementation} only specifies how to use the file---it doesn't actually
10268 There is no way to split up the contents of a single header file into
10269 multiple implementation files.
10272 @cindex inlining and C++ pragmas
10273 @cindex C++ pragmas, effect on inlining
10274 @cindex pragmas in C++, effect on inlining
10275 @samp{#pragma implementation} and @samp{#pragma interface} also have an
10276 effect on function inlining.
10278 If you define a class in a header file marked with @samp{#pragma
10279 interface}, the effect on an inline function defined in that class is
10280 similar to an explicit @code{extern} declaration---the compiler emits
10281 no code at all to define an independent version of the function. Its
10282 definition is used only for inlining with its callers.
10284 @opindex fno-implement-inlines
10285 Conversely, when you include the same header file in a main source file
10286 that declares it as @samp{#pragma implementation}, the compiler emits
10287 code for the function itself; this defines a version of the function
10288 that can be found via pointers (or by callers compiled without
10289 inlining). If all calls to the function can be inlined, you can avoid
10290 emitting the function by compiling with @option{-fno-implement-inlines}.
10291 If any calls were not inlined, you will get linker errors.
10293 @node Template Instantiation
10294 @section Where's the Template?
10295 @cindex template instantiation
10297 C++ templates are the first language feature to require more
10298 intelligence from the environment than one usually finds on a UNIX
10299 system. Somehow the compiler and linker have to make sure that each
10300 template instance occurs exactly once in the executable if it is needed,
10301 and not at all otherwise. There are two basic approaches to this
10302 problem, which are referred to as the Borland model and the Cfront model.
10305 @item Borland model
10306 Borland C++ solved the template instantiation problem by adding the code
10307 equivalent of common blocks to their linker; the compiler emits template
10308 instances in each translation unit that uses them, and the linker
10309 collapses them together. The advantage of this model is that the linker
10310 only has to consider the object files themselves; there is no external
10311 complexity to worry about. This disadvantage is that compilation time
10312 is increased because the template code is being compiled repeatedly.
10313 Code written for this model tends to include definitions of all
10314 templates in the header file, since they must be seen to be
10318 The AT&T C++ translator, Cfront, solved the template instantiation
10319 problem by creating the notion of a template repository, an
10320 automatically maintained place where template instances are stored. A
10321 more modern version of the repository works as follows: As individual
10322 object files are built, the compiler places any template definitions and
10323 instantiations encountered in the repository. At link time, the link
10324 wrapper adds in the objects in the repository and compiles any needed
10325 instances that were not previously emitted. The advantages of this
10326 model are more optimal compilation speed and the ability to use the
10327 system linker; to implement the Borland model a compiler vendor also
10328 needs to replace the linker. The disadvantages are vastly increased
10329 complexity, and thus potential for error; for some code this can be
10330 just as transparent, but in practice it can been very difficult to build
10331 multiple programs in one directory and one program in multiple
10332 directories. Code written for this model tends to separate definitions
10333 of non-inline member templates into a separate file, which should be
10334 compiled separately.
10337 When used with GNU ld version 2.8 or later on an ELF system such as
10338 GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the
10339 Borland model. On other systems, G++ implements neither automatic
10342 A future version of G++ will support a hybrid model whereby the compiler
10343 will emit any instantiations for which the template definition is
10344 included in the compile, and store template definitions and
10345 instantiation context information into the object file for the rest.
10346 The link wrapper will extract that information as necessary and invoke
10347 the compiler to produce the remaining instantiations. The linker will
10348 then combine duplicate instantiations.
10350 In the mean time, you have the following options for dealing with
10351 template instantiations:
10356 Compile your template-using code with @option{-frepo}. The compiler will
10357 generate files with the extension @samp{.rpo} listing all of the
10358 template instantiations used in the corresponding object files which
10359 could be instantiated there; the link wrapper, @samp{collect2}, will
10360 then update the @samp{.rpo} files to tell the compiler where to place
10361 those instantiations and rebuild any affected object files. The
10362 link-time overhead is negligible after the first pass, as the compiler
10363 will continue to place the instantiations in the same files.
10365 This is your best option for application code written for the Borland
10366 model, as it will just work. Code written for the Cfront model will
10367 need to be modified so that the template definitions are available at
10368 one or more points of instantiation; usually this is as simple as adding
10369 @code{#include <tmethods.cc>} to the end of each template header.
10371 For library code, if you want the library to provide all of the template
10372 instantiations it needs, just try to link all of its object files
10373 together; the link will fail, but cause the instantiations to be
10374 generated as a side effect. Be warned, however, that this may cause
10375 conflicts if multiple libraries try to provide the same instantiations.
10376 For greater control, use explicit instantiation as described in the next
10380 @opindex fno-implicit-templates
10381 Compile your code with @option{-fno-implicit-templates} to disable the
10382 implicit generation of template instances, and explicitly instantiate
10383 all the ones you use. This approach requires more knowledge of exactly
10384 which instances you need than do the others, but it's less
10385 mysterious and allows greater control. You can scatter the explicit
10386 instantiations throughout your program, perhaps putting them in the
10387 translation units where the instances are used or the translation units
10388 that define the templates themselves; you can put all of the explicit
10389 instantiations you need into one big file; or you can create small files
10396 template class Foo<int>;
10397 template ostream& operator <<
10398 (ostream&, const Foo<int>&);
10401 for each of the instances you need, and create a template instantiation
10402 library from those.
10404 If you are using Cfront-model code, you can probably get away with not
10405 using @option{-fno-implicit-templates} when compiling files that don't
10406 @samp{#include} the member template definitions.
10408 If you use one big file to do the instantiations, you may want to
10409 compile it without @option{-fno-implicit-templates} so you get all of the
10410 instances required by your explicit instantiations (but not by any
10411 other files) without having to specify them as well.
10413 G++ has extended the template instantiation syntax given in the ISO
10414 standard to allow forward declaration of explicit instantiations
10415 (with @code{extern}), instantiation of the compiler support data for a
10416 template class (i.e.@: the vtable) without instantiating any of its
10417 members (with @code{inline}), and instantiation of only the static data
10418 members of a template class, without the support data or member
10419 functions (with (@code{static}):
10422 extern template int max (int, int);
10423 inline template class Foo<int>;
10424 static template class Foo<int>;
10428 Do nothing. Pretend G++ does implement automatic instantiation
10429 management. Code written for the Borland model will work fine, but
10430 each translation unit will contain instances of each of the templates it
10431 uses. In a large program, this can lead to an unacceptable amount of code
10435 @node Bound member functions
10436 @section Extracting the function pointer from a bound pointer to member function
10438 @cindex pointer to member function
10439 @cindex bound pointer to member function
10441 In C++, pointer to member functions (PMFs) are implemented using a wide
10442 pointer of sorts to handle all the possible call mechanisms; the PMF
10443 needs to store information about how to adjust the @samp{this} pointer,
10444 and if the function pointed to is virtual, where to find the vtable, and
10445 where in the vtable to look for the member function. If you are using
10446 PMFs in an inner loop, you should really reconsider that decision. If
10447 that is not an option, you can extract the pointer to the function that
10448 would be called for a given object/PMF pair and call it directly inside
10449 the inner loop, to save a bit of time.
10451 Note that you will still be paying the penalty for the call through a
10452 function pointer; on most modern architectures, such a call defeats the
10453 branch prediction features of the CPU@. This is also true of normal
10454 virtual function calls.
10456 The syntax for this extension is
10460 extern int (A::*fp)();
10461 typedef int (*fptr)(A *);
10463 fptr p = (fptr)(a.*fp);
10466 For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}),
10467 no object is needed to obtain the address of the function. They can be
10468 converted to function pointers directly:
10471 fptr p1 = (fptr)(&A::foo);
10474 @opindex Wno-pmf-conversions
10475 You must specify @option{-Wno-pmf-conversions} to use this extension.
10477 @node C++ Attributes
10478 @section C++-Specific Variable, Function, and Type Attributes
10480 Some attributes only make sense for C++ programs.
10483 @item init_priority (@var{priority})
10484 @cindex init_priority attribute
10487 In Standard C++, objects defined at namespace scope are guaranteed to be
10488 initialized in an order in strict accordance with that of their definitions
10489 @emph{in a given translation unit}. No guarantee is made for initializations
10490 across translation units. However, GNU C++ allows users to control the
10491 order of initialization of objects defined at namespace scope with the
10492 @code{init_priority} attribute by specifying a relative @var{priority},
10493 a constant integral expression currently bounded between 101 and 65535
10494 inclusive. Lower numbers indicate a higher priority.
10496 In the following example, @code{A} would normally be created before
10497 @code{B}, but the @code{init_priority} attribute has reversed that order:
10500 Some_Class A __attribute__ ((init_priority (2000)));
10501 Some_Class B __attribute__ ((init_priority (543)));
10505 Note that the particular values of @var{priority} do not matter; only their
10508 @item java_interface
10509 @cindex java_interface attribute
10511 This type attribute informs C++ that the class is a Java interface. It may
10512 only be applied to classes declared within an @code{extern "Java"} block.
10513 Calls to methods declared in this interface will be dispatched using GCJ's
10514 interface table mechanism, instead of regular virtual table dispatch.
10518 See also @xref{Namespace Association}.
10520 @node Namespace Association
10521 @section Namespace Association
10523 @strong{Caution:} The semantics of this extension are not fully
10524 defined. Users should refrain from using this extension as its
10525 semantics may change subtly over time. It is possible that this
10526 extension will be removed in future versions of G++.
10528 A using-directive with @code{__attribute ((strong))} is stronger
10529 than a normal using-directive in two ways:
10533 Templates from the used namespace can be specialized and explicitly
10534 instantiated as though they were members of the using namespace.
10537 The using namespace is considered an associated namespace of all
10538 templates in the used namespace for purposes of argument-dependent
10542 The used namespace must be nested within the using namespace so that
10543 normal unqualified lookup works properly.
10545 This is useful for composing a namespace transparently from
10546 implementation namespaces. For example:
10551 template <class T> struct A @{ @};
10553 using namespace debug __attribute ((__strong__));
10554 template <> struct A<int> @{ @}; // @r{ok to specialize}
10556 template <class T> void f (A<T>);
10561 f (std::A<float>()); // @r{lookup finds} std::f
10566 @node Java Exceptions
10567 @section Java Exceptions
10569 The Java language uses a slightly different exception handling model
10570 from C++. Normally, GNU C++ will automatically detect when you are
10571 writing C++ code that uses Java exceptions, and handle them
10572 appropriately. However, if C++ code only needs to execute destructors
10573 when Java exceptions are thrown through it, GCC will guess incorrectly.
10574 Sample problematic code is:
10577 struct S @{ ~S(); @};
10578 extern void bar(); // @r{is written in Java, and may throw exceptions}
10587 The usual effect of an incorrect guess is a link failure, complaining of
10588 a missing routine called @samp{__gxx_personality_v0}.
10590 You can inform the compiler that Java exceptions are to be used in a
10591 translation unit, irrespective of what it might think, by writing
10592 @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This
10593 @samp{#pragma} must appear before any functions that throw or catch
10594 exceptions, or run destructors when exceptions are thrown through them.
10596 You cannot mix Java and C++ exceptions in the same translation unit. It
10597 is believed to be safe to throw a C++ exception from one file through
10598 another file compiled for the Java exception model, or vice versa, but
10599 there may be bugs in this area.
10601 @node Deprecated Features
10602 @section Deprecated Features
10604 In the past, the GNU C++ compiler was extended to experiment with new
10605 features, at a time when the C++ language was still evolving. Now that
10606 the C++ standard is complete, some of those features are superseded by
10607 superior alternatives. Using the old features might cause a warning in
10608 some cases that the feature will be dropped in the future. In other
10609 cases, the feature might be gone already.
10611 While the list below is not exhaustive, it documents some of the options
10612 that are now deprecated:
10615 @item -fexternal-templates
10616 @itemx -falt-external-templates
10617 These are two of the many ways for G++ to implement template
10618 instantiation. @xref{Template Instantiation}. The C++ standard clearly
10619 defines how template definitions have to be organized across
10620 implementation units. G++ has an implicit instantiation mechanism that
10621 should work just fine for standard-conforming code.
10623 @item -fstrict-prototype
10624 @itemx -fno-strict-prototype
10625 Previously it was possible to use an empty prototype parameter list to
10626 indicate an unspecified number of parameters (like C), rather than no
10627 parameters, as C++ demands. This feature has been removed, except where
10628 it is required for backwards compatibility @xref{Backwards Compatibility}.
10631 G++ allows a virtual function returning @samp{void *} to be overridden
10632 by one returning a different pointer type. This extension to the
10633 covariant return type rules is now deprecated and will be removed from a
10636 The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and
10637 their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated
10638 and will be removed in a future version. Code using these operators
10639 should be modified to use @code{std::min} and @code{std::max} instead.
10641 The named return value extension has been deprecated, and is now
10644 The use of initializer lists with new expressions has been deprecated,
10645 and is now removed from G++.
10647 Floating and complex non-type template parameters have been deprecated,
10648 and are now removed from G++.
10650 The implicit typename extension has been deprecated and is now
10653 The use of default arguments in function pointers, function typedefs and
10654 and other places where they are not permitted by the standard is
10655 deprecated and will be removed from a future version of G++.
10657 G++ allows floating-point literals to appear in integral constant expressions,
10658 e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} }
10659 This extension is deprecated and will be removed from a future version.
10661 G++ allows static data members of const floating-point type to be declared
10662 with an initializer in a class definition. The standard only allows
10663 initializers for static members of const integral types and const
10664 enumeration types so this extension has been deprecated and will be removed
10665 from a future version.
10667 @node Backwards Compatibility
10668 @section Backwards Compatibility
10669 @cindex Backwards Compatibility
10670 @cindex ARM [Annotated C++ Reference Manual]
10672 Now that there is a definitive ISO standard C++, G++ has a specification
10673 to adhere to. The C++ language evolved over time, and features that
10674 used to be acceptable in previous drafts of the standard, such as the ARM
10675 [Annotated C++ Reference Manual], are no longer accepted. In order to allow
10676 compilation of C++ written to such drafts, G++ contains some backwards
10677 compatibilities. @emph{All such backwards compatibility features are
10678 liable to disappear in future versions of G++.} They should be considered
10679 deprecated @xref{Deprecated Features}.
10683 If a variable is declared at for scope, it used to remain in scope until
10684 the end of the scope which contained the for statement (rather than just
10685 within the for scope). G++ retains this, but issues a warning, if such a
10686 variable is accessed outside the for scope.
10688 @item Implicit C language
10689 Old C system header files did not contain an @code{extern "C" @{@dots{}@}}
10690 scope to set the language. On such systems, all header files are
10691 implicitly scoped inside a C language scope. Also, an empty prototype
10692 @code{()} will be treated as an unspecified number of arguments, rather
10693 than no arguments, as C++ demands.